Functional and expression analysis of Androglobin in  different branches of the metazoan tree          Dissertation  Zur Erlangung des Grades  „Doktor der Naturwissenschaften“    am Fachbereich Biologie  der Johannes‐Gutenberg‐Universität Mainz        Carina Osterhof  geb. am 30.07.1992 in Dülmen      Mainz, den 22.09.2022      Dean: Prof. Dr. Eckhard Thines  First examiner:  Second examiner:  Date of defense: 09.11.2022 ii                        „Es  ist ein Gedankengebäude“, seufzte Kolibril. „Meine ewige Baustelle. Halbgare Theorien,  Ideenruinen.  Ich  bezweifle,  dass  ich mit  dieser Doktorarbeit  zu  Lebzeiten  jemals  zu  Rande  kommen werde“.   aus Walter Moers: Rumo und die Wunder im Dunkeln                          iii        iv    Table of content Table of figures ......................................................................................................................... II Table of abbreviations ............................................................................................................. III 1 Introduction ...................................................................................................................... 1 1.1 Globins: a diverse gene family ................................................................................... 1 1.2 Globin phylogeny: ancestral and vertebrate-specific lineages ................................... 4 1.3 Functions of conventional globins in health and disease ........................................... 9 1.4 Androglobin – an ancient globin with a peculiar structure ...................................... 12 1.5 Aim of this study ...................................................................................................... 16 2 Results ............................................................................................................................. 19 2.1 Androglobin gene expression patterns and FOXJ1-dependent regulation indicate its functional association with ciliogenesis ...................................................................... 19 2.2 Androglobin, a chimeric mammalian globin, is required for male fertility ............... 45 2.3 A role of Androglobin in cancer? – insights from an overexpression system and transcriptome data mining ......................................................................................... 73 2.4 Androglobin expression pattern in basal metazoans confirms its conserved functional association with cilia ................................................................................................. 113 3 Discussion and Outlook ................................................................................................. 145 3.1 Androglobin as part of the ciliary proteome .......................................................... 147 3.2 Loss of Androglobin is an indicator of ciliary or flagellar specialization in metazoans ……………………………………………………………………………………………………………………………149 3.3 Adgb beyond metazoans ........................................................................................ 151 3.4 Androglobin: the maverick of the protein family? ................................................. 154 4 Summary ....................................................................................................................... 159 5 Zusammenfassung ........................................................................................................ 161 6 References .................................................................................................................... 163 7 Appendix ....................................................................................................................... 177 7.1 Supplementary data ............................................................................................... 177 7.2 Index of electronic supplement ............................................................................. 178 7.3 Acknowledgements ................................................................................................ 179 7.4 Eidesstattliche Erklärung ........................................................................................ 180 7.5 Curriculum vitae ..................................................................................................... 181 I Table of figures Figure 1-1: Structural models of the globin fold with its heme group (black). ......................... 2 Figure 1-2: Phylogeny of vertebrate globins. ............................................................................ 4 Figure 1-3: Model of the evolution of GbX in vertebrate species. ............................................ 6 Figure 1-4: Occurrence of globins among vertebrates. ............................................................ 8 Figure 1-5: Structure of Androglobin ...................................................................................... 13 Figure 1-6: mRNA expression analysis of Adgb in mice. ......................................................... 14 Figure 1-7: Phylogenetic tree of Androglobin homologues .................................................... 15 Figure 3-1: Occurrence of putative Adgb sequences across the eucaryotic tree. ................. 153 Figure 3-2: Expression data on Sept10 and Adgb from the Human protein atlas. ................ 157 II     Table of abbreviations  AA  amino acid  acc.  accession  Adgb  Androglobin  BLAST  Basic Local Alignment Search Tool  CaM  Calmodulin  CLS  ciliary localisation signal  CPC  ciliary pore complex  Cygb  Cytoglobin  DHC5  dynein heavy chain 5  DHC6  dynein heavy chain 6  EZH2  Enhancer of zeste homolog 2  GbE  Globin E  GbX  Globin X  GbY  Globin Y  glob2  globin 2  glob3  globin 3  Hb  Hemoglobin  KD  knockdown  kDa  kilo Dalton  KIF17  Kinesin Family Member 17  KO  knockout  Mb  Myoglobin  mRNA  messenger ribonucleic acid  MTT  3‐(4,5‐dimethylthiazol‐2‐yl)‐2,5‐diphenyltetrazolium bromide  N. vectensis  Nematostella vectensis  NCBI  National Center for Biotechnology Information  Ngb  Neuroglobin  NLS  nuclear localisation signal  NO  Nitric oxide  NPC  nuclear pore complex  III    O2  oxygen  PC1  polycystin‐1  qPCR  quantitative polymerase chain reaction  qRT‐PCR  quantitative reverse transcription polymerase chain reaction  RNA  Ribonucleic acid  RNAi  Ribonucleic acid interference  RNA‐Seq  RNA‐Sequencing  RNS  reactive nitrogen species  ROS  reactive oxygen species  S. purpuratus  Strongylocentrotus purpuratus  S. rosetta  Salpingocea rosetta  shRNA  short hairpin ribonucleic acid  STXBP5‐AS1  Syntaxin Binding Protein 5 Antisense RNA 1  IV        1 Introduction  1.1 Globins: a diverse gene family  Globins  are  a  family  of  small,  globular  proteins  which  are  widely  distributed  over  the  phylogenetic tree of  life. Their most  famous members  in vertebrates, Myoglobin  (Mb) and  Hemoglobin  (Hb),  are  known  for  their  vital  role  in  oxygen  storage  and  transport. Globin  proteins are characterized by a highly conserved three‐dimensional structure, were eight a‐ Helices  (A‐H)  fold  in  a  so‐called  3‐over‐3  sandwich,  the  “globin”‐fold  (Figure  1‐1,  A).  The  resulting  hydrophobic  pocket  allows  binding  of  a  heme  prosthetic  group  at  the  first  phenylalanine between helices C and D. At this heme, a variety of gaseous ligands such as O2,  CO2 and NO may bind reversibly. The iron atom at the centre of the heme group is coordinated  at  four positions via nitrogen  from  the protoporphyrin and by  the histidine at position F8  (proximal histidine) of  the protein. Depending on  the  state of  the  sixth  coordination  site,  globins may  be  distinguished  into  two  groups:  in  penta‐coordinated  globins,  such  as  the  classical Mb, the distal binding site is free, whereas in hexa‐coordinated globins, the iron atom  is additionally coordinated by the side chain of histidine (or glutamine) at position 7 of the E‐ Helix (distal histidine E7) (Figure 1‐1, B and C). This side chain competes with ligand binding to  the iron atom, resulting in a structural shift which enables globins to act as sensors or signalling  proteins (Reeder et al., 2011).  For nearly a century, Hb and Mb were regarded as the only members of this group of proteins.  However, with the discovery of Neuroglobin  (Ngb)  (Burmester et al., 2000) and Cytoglobin  (Cygb)  (Kawada  et  al.,  2001;  Burmester  et  al.,  2002;  Trent & Hargrove,  2002),  it  became  evident that vertebrates harbour a surprisingly diverse set of globins:   Neuroglobin was initially described in brain tissue from humans and rodents, hence its name  (Burmester et al., 2000). It is highly expressed in the hypothalamus (Fabrizius et al., 2016), but  lower  amounts  can  also be  found  in  several other  central and peripheral nervous  system  regions. A few non‐neuronal, endocrine tissues such as the testis, adrenal and pituitary gland  do also express Ngb (Reuss et al., 2002; Fabrizius et al., 2016).     1    Figure 1‐1: Structural models of the globin fold with its heme group (black). (A) 3‐over3‐ helix structure  of Mb. (B) Penta‐coordinated sperm whale Mb. The distal binding site is accessible for ligand binding.  (C) Hexa‐coordinated Neuroglobin (Ngb). The distal binding site is occupied by HisE7, which competes  with the external ligand binding. Helices of the globin fold are named A‐H starting from the N‐terminus.  Adapted from Storz, 2018.  Cytoglobin  is mainly expressed  in  fibroblasts and related cell types and therefore  found  in  many organs (Kawada et al., 2001; Burmester et al., 2002; Nakatani et al., 2004).  In addition,  it  is expressed  in  the  smooth muscle cells of  the vascular  system, were  it  regulates blood  pressure via the consumption of NO (Liu et al., 2017). Melanocytes and some cell lines derived  from melanomas show elevated levels of Cygb (Fujita et al., 2014), as well as adipocytes and  adipose tissues (Doğan et al., 2017; Thuy et al., 2020; Elena Porto, unpublished data).  Globin X (GbX) was first identified via sequence analysis in frogs and fish (Rösner et al., 2005)  and is the only membrane bound member of the vertebrate globin family (Blank, Wollberg, et  al., 2011). GbX expression could be observed in a variety of tissues, such as brain, eye, heart,  intestine, liver and spleen and differs between taxa (Blank, Wollberg, et al., 2011; Schwarze et  al., 2015; Gallagher & Macqueen, 2017). In vitro studies suggest that it could protect cells from  ROS (Koch & Burmester, 2016) and  it was unexpectedly reported  in red blood cells of fish,  were it could act as a fast nitrite reductase (Corti et al., 2016).  Globin E (GbE) demonstrates an even higher oxygen affinity than the specialized Mb isoforms  in diving mammals, making it an ideal candidate for oxygen supply (Blank, Kiger, et al., 2011;  Wright & Davis, 2015) . Its expression pattern is very distinct: in birds and turtles, GbE mRNA  was detected only in the eye (Kugelstadt et al., 2004; Schwarze et al., 2015). In lungfish, were  2        the GbE gene underwent duplications, all gene copies are exclusively expressed in the ovary  (Lüdemann et al., 2019). The exact function of this globin, however, is still not clear.    Of all vertebrate globins, Globin Y (GbY) has been studied the least. It was initially discovered  in the genome of Xenopus tropicalis (Fuchs et al., 2006) and later also identified in several fish  species and turtles. Like Cygb, it is widely expressed and the cell‐ and organ‐specific expression  sites vary depending on the species analysed. In fish, GbY is expressed in gills and the spleen  whereas in turtles it is mostly found in kidney, lung and brain (Schwarze et al., 2015; Gallagher  & Macqueen, 2017).    Androglobin  (Adgb), discovered  in  2012,  is  the  focus of  this  thesis,  and will  therefore be  discussed in more detail below (Chapter 1.4).      3    1.2 Globin phylogeny: ancestral and vertebrate‐specific lineages  Globins have been described in bacteria and archaea (Vinogradov et al., 2013), indicating that  they were already present  in the  last universal common ancestor. Horizontal gene transfer  during endosymbiotic events may have led to the incorporation of the first globin sequence  into a eucaryotic genome (Vinogradov et al., 2007).  Their rather short sequences of about 150  amino  acids  with  high  amounts  of  primary  sequence  divergence  pose  difficulties  for  phylogenetic reconstructions, however, the highly conserved folding pattern allows for robust  identification  of  family  members  (Herman  et  al.,  2014).  In  addition,  globin  genes  carry  characteristic  introns  at  fixed positions, namely one  in  the 12th  amino  acid of  the B‐Helix  (B12.2) as well as one in the G‐Helix (G7.0). These introns may have already been present in  the last common ancestor of plants and animals (as reviewed in: Hardison, 1996) and also do  facilitate the identification of potential globin gene family members. In Figure 1‐2, an overview  over  the  globin  lineages  is  given,  as well  the  current hypothesis  about  their phylogenetic  relationship:       Figure 1‐2: Phylogeny of vertebrate globins. Adgb, Ngb and GbX are the ancestral metazoan globin  lineages, whereas  the vertebrate  specific globins developed during  the whole genome duplication  events which preceded the emergence of the vertebrate subphlyum. Adapted from Storz, 2018.    4        The emergence of Ngb, one of the “oldest” globins, dates back prior to the split of protostomes  and deuterostomes (Burmester et al., 2000). In phylogenetic analyses, vertebrate Ngb clusters  with globin sequences from sea urchins and cephalochordates as well as annelid nerve globins.  A possible Ngb ortholog was found in the placozoan Trichoplax, thus predating the emergence  of a real nervous system (Dröge et al., 2012; Hankeln and Ebner, unpublished).  It has been  proposed  that  Ngbs  represent  the  most  ancient  clade  of  the  classically  arranged,  contemporary  globins  (Hoffmann  et  al.,  2012),  but was  independently  lost  several  times  during evolution, as it is the case in hemichordates and tunicates, some cartilaginous fishes  (Schwarze et al., 2014; Opazo et al., 2015) and also in arthropods (Prothmann et al., 2020).   The evolution of GbX is special: it is, on the one hand, an ancient globin such as Ngb with a  very wide distribution over protostomes as well as deuterostomes  (Hoffmann et al., 2012;  Prothmann et al., 2020). On the other hand, the GbX repertoire has also been expanded on  several occasions during globin evolution. Phylogenetic analysis of the globin repertoires of  the acorn worm and branchiostoma suggest that there are two main GbX lineages, GbX and  GbX‐like (or Clade 2 and 3), which reflect a duplication early  in evolution (Hoffmann et al.,  2012; Blank & Burmester, 2012). In line with this, members of both lineages were identified  in  arthropod  genomes  and  phylogenetic  reconstruction  clusters  both  at  the  base  of  the  phylogenetic tree (Prothmann et al., 2020). Many vertebrate taxa only harbour one variant of  GbX, whereas the GbX‐L lineage was lost. However, during comparative phylogenetic analyses  it became clear that this perceived orthology was in fact hidden paralogy: The GbX locus was  duplicated during one of the two WGD events which precede the emergence of vertebrates  (Figure 1‐3). One of these proto‐GbX variants was then respectively lost in cyclostomes and  gnathostomes and the second duplicated, probably due to additional WGD or polyploidization  events  (Opazo  et  al.,  2015;  Gallagher  &  Macqueen,  2017;  Hoffmann  et  al.,  2021).  In  gnathostome evolution, both copies were differentially retained and duplicated several times,  and sometimes copies of closely related taxa such as Xenopus and the salamanders are only  paralogous (Figure 1‐3). In contrast, Androglobin, which is the third ancient metazoan globin  lineage is almost exclusively present as single copy. In fact, Adgb originated already at the very  base  of  the  metazoan  tree  and  can  be  traced  back  to  the  last  common  ancestor  of  choanoflagellates and animals (Hoogewijs et al., 2012).     5      Figure 1‐3: Model of the evolution of GbX in vertebrate species. The GbX locus was duplicated during  one of the two WGD events which precede the emergence of vertebrates. One of these proto‐GbX  variants was  then  independently  lost  in cyclostomes and gnathostomes and  the second duplicated  (probably  due  to  additional  WGD  events).  Note  that  this  model  reflects  the  hypothesis  that  gnathostomes and cyclostomes only shared one of  the  two  rounds of WGD of vertebrates. During  gnathostome evolution, additional duplications and  loss events occurred and although many of the  analysed genomes only contain one GbX copy, they are not necessarily orthologous to each other.  Adapted from Hoffmann et al. 2021.       6        The  lineage  of  the  globin  family,  which  consist  of  vertebrate‐specific  members,  can  be  considered a textbook example of how whole genome and gene duplication events may drive  diversification of gene families (Storz et al., 2013; Hoffmann et al., 2021). All contemporary  vertebrate‐specific  globin  genes  are  probably  derived  from  a  single  common  ancestor.  Through  subsequent  whole  genome  duplication  events  during  vertebrate  evolution,  4  paralogous genomic loci arose: Cygb, Mb, Hb and a fourth globin locus which later on lost its  globin  gene  copy. Consecutively,  a  tandem  gene duplication  in  the Mb  locus  in  the  stem  lineage of jawed vertebrates created the GbE gene. GbE was secondarily lost again multiple  times  and  now  only  remains  in  birds,  turtles  (Schwarze  et  al.,  2015),  and  the  coelacanth  (Schwarze & Burmester, 2013). GbX, GbE and GbY were all secondarily lost in the mammalian  lineage. Currently, the coelacanth and turtles are the only taxa known to harbour the full set  of globin diversity (Schwarze et al., 2015). A summary of the distribution of all globins among  the vertebrate  lineages can be found  in Figure 1‐4. The retention of both the proto‐Hb and  Mb  in the ancestor of  jawed vertebrates allowed  for the specialisation towards O2 storage  (Mb) or transport function (Hb). Additional duplication of the proto‐Hb locus resulted in the  two subunits α and β, which permitted the formation of a multimeric and heterogenous Hb  and laid the foundation of development of sophisticated regulatory transport mechanism such  as allostery. Intriguingly, the specialized O2 transport in blood via Hb seems to have evolved  convergently in cyclostomes and jawed vertebrates (Hoffmann et al., 2010; Schwarze et al.,  2014). Also in cyclostomes, transport of oxygen is realised by a tetrameric and storage by a  monomeric globin variant, but the underlying molecular cooperativity mechanisms as well as  the oxygen affinity show significant differences (Fago & Weber, 1995; Qiu et al., 2000; Fago et  al., 2001).   7      Figure 1‐4: Occurrence of globins among vertebrates. On the branch of the tree, the globin variants  are  listed which could be  identified  in  the corresponding clade. Events of  losses or species specific  duplications  are  described  on  the  right.  Note  that  of  the  extant  lineages,  only  turtles  and  the  coelacanth harbour the full set of vertebrate globins. A more detailed depiction of the evolutionary  events is found in the text. Adapted from Keppner et al., 2020.       8        1.3 Functions of conventional globins in health and disease  Monomeric  globins  are  famously  known  for  their  role  in  oxygen  storage  and  transport.  However,  the precursor  globin, which  all our  contemporary  representatives were derived  from, probably served a completely different function, since oxygen was rather a cytotoxin  than an enabler of aerobic energy production in early evolution more than 1 billion years ago  (Hardison, 1998; Vinogradov et al., 2007). Conclusively, research on the properties of globins  has shown  that  they do not only bind and provide oxygen but  fulfil a variety of additional  functions (Burmester & Hankeln, 2014; Keppner et al., 2020). Cygb, for example, regulates NO  bioactivity  (as  reviewed  in: Mathai  et  al.,  2020)  and  thus  influences  blood  pressure  and  vascular tone in mammals in vivo (Liu et al., 2017). Mouse work has shown that Mb can act as  an NO scavenger to protect cardiac function (Gödecke 2003). Depending on its oxygenation  state, it may also produce NO, a signalling molecule, under low oxygen environments (Cossins  2008, Hendgen‐Cotta 2008). Cygb and Mb also serve as protectors against oxidative stress:  Mb  exhibits  catalase  and  peroxidase  activity  which  neutralizes  toxic  H2O2  in  mammals  (Mannino et al., 2019) as well as fish (Helbo et al., 2012). In vitro studies have shown that Cygb  expression  is beneficial under oxidative stress  (e.g.: Fordel et al., 2006; Nishi et al., 2011).  Furthermore, work  in additional mouse models confirms  its antioxidant role, as aged Cygb  knockout mice  showed  signs  of  an  imbalanced  antioxidant  defence  system with  severely  damaged organs (Thuy et al., 2016).  In  line with this, Cygb overexpressing mice were more  resistant to induced oxidative stress and less prone to hepatic fibrosis (Thi Thanh Hai et al.,  2018). Also  for Ngb,  a  cytoprotective  function  towards  ROS  and  RNS  has  been  proposed  (Burmester & Hankeln, 2009; Ascenzi et al., 2016). Ectopic overexpression of Ngb enhanced  H202  tolerance  in  cell  culture models  (Fordel et al., 2006;  Li et al., 2007) and ameliorated  recovery after ischemic injury in mouse and rat models (as reviewed in: Van Acker et al., 2018).  In vitro data suggests that Ngb may act as a NO scavenger in its oxygenated form (Brunori et  al., 2005), but up to now the reduction system that would regenerate the heme has not been  found  (Smagghe et al., 2008).  In  its deoxygenated  form, Ngb might  reduce Nitrate  to NO,  which could then inhibit mitochondrial respiration and thus have cytoprotective effects under  hypoxia (Petersen et al., 2008; Tiso et al., 2011).     9    The hypoxic environment and prevalence of ROS/RNS are widely known hallmarks of cancer  and may serve as basis for therapeutical approaches (Trachootham et al., 2009; Sosa et al.,  2013). Whereas high amounts of ROS lead to extensive cell damage, low concentrations could  promote cell growth and proliferation and ultimately enhance malignancy (Sosa et al., 2013).  Since globins are able to produce or scavenge ROS depending on their oxygenation state, they  can both promote and  inhibit  tumour growth and are  therefore promising  candidates  for  cancer research. For some family members, a deregulation in cancer entities has already been  described. Mb  is upregulated  in a variety of epithelial tumours, such as breast,  lung, ovary,  and  colon  carcinomas  (Flonta  et  al.,  2009;  Kristiansen  et  al.,  2010),  and  can  even  have  prognostic value (as reviewed in: Elkholi et al., 2022). In brain tumours, for example, ectopic  Mb expression correlates with a more aggressive tumour phenotype (Elsherbiny et al., 2021),  whereas in breast and neck cancer, expression of Mb is considered advantageous and leads  to a better outcome for the patient (Kristiansen et al., 2010, 2011; Meller et al., 2016). Cygb,  on the other hand, is mostly downregulated in cancer entities (as reviewed in: Oleksiewicz et  al.,  2011).  In  gliomas  and  pancreatic  ductal  adenocarcinomas,  reduced  endogenous  expression of Cygb was correlated with a more fatal tumour phenotype and higher rates of  recurrence (Xu et al., 2013; Kono et al., 2021). These findings could be confirmed  in  in vivo  Cygb KO models. Two  independent mouse  lines  (Thi Thanh Thuy et al., 2011; Yassin et al.,  2018) showed higher tumour incidence upon Cygb ablation in liver, lung and colon tissue. Ngb,  in contrast, seems  to have a dualistic  function  in cancer:  it has been described both as an  oncogene as well as a  tumour  suppressor  in different entities.  In breast  cancer as well as  glioma cell culture models, Ngb might promote proliferation and suppress apoptosis (Zhang,  B. et al., 2018; Fiocchetti et al., 2018). In line with this potential oncogene function, antibody  stainings  showed a potential upregulation  in  several  tumour  samples  (Emara et al., 2009,  2010). However,  it has been contested whether  the cell culture models  that were used  to  study Ngb’s oncogenic attributes were suitable: a comprehensive reanalysis of Ngb expression  sites  found no evidence of endogenous Ngb mRNA expression  in a variety of cell  lines and  most  of  these  cancer  studies  relied  on  antibodies  which  were  not  thoroughly  validated  (Fabrizius et al., 2016).  In addition, antibody‐free  in  situ hybridisation analysis of Ngb  in a  cancer tissue array showed no significant expression of Ngb in a variety of malignancies (Gorr  et al., 2011). This  is also  reflected  in  the  transcriptomic analysis performed by  the Cancer  10        Genome  Atlas  project1  (e.g.:  Blum  et  al.,  2018;  Hutter  &  Zenklusen,  2018):  Ngb  is  downregulated  in  all  cancer  entities  derived  from  endogenous  expression  sites  (Carina  Osterhof  and  Thomas Hankeln, unpublished), which  strongly  argues  against  an oncogenic  function.  It  has  also  been  proposed  that  Ngb  could  act  as  a  tumour  suppressor  in  hepatocellular carcinoma (Zhang, J. et al., 2013), but also here, the expression analysis shows  no evidence of Ngb mRNA expression in healthy tissues. It is therefore unknown whether Ngb  plays any role in malignancies in vivo.         1 https://www.cancer.gov/about‐nci/organization/ccg/research/structural‐genomics/tcga  11    1.4 Androglobin – an ancient globin with a peculiar structure  Androglobin  (Adgb)  is  the most  recent  addition  to  the  diverse  family  of  globins.  It  was  discovered  in  2012  in  parallel  in  the  echinoderm  Strongylocentrotus  purpuratus  and  the  cephalopod Branchiostoma floridae (Hoogewijs et al., 2012). Through an extensive search of  public datasets, the Adgb gene could additionally be identified in most taxa of the metazoans,  including mammals, as well as  in some unicellular species of choanoflagellates (Figure 1‐7).  Unfortunately, it was lost in several genetical model species, such as Caenorhabditis elegans  and Drosophila melanogaster, which has complicated functional studies so far.   Adgb was discovered rather late given its widespread existence, which is probably due to its  unique globin domain structure (cf. Figure 1‐5): instead of the typical A‐H sequence of α‐helical  segments, it starts with helices C‐H followed by a putative IQ calmodulin binding site. Helices  A and B form the C‐terminal end of the domain, including the characteristic globin intron at  B12.2. This rearranged domain presumably resulted from multiple mutation events, starting  with a tandem duplication of the globin domain followed by truncation of the now redundant  additional helices. The IQ domain could have been acquired secondarily: the motif is enclosed  by two introns of the same phase, meaning that both introns are inserted at the same codon  position. This indicates that Adgb gained this motif via an exon‐shuffling event. Regardless of  these profound structural differences, alignments of mammalian globins and subsequent 3D  modelling of helixes C‐H suggest that Adgb forms a functional globin fold (Hoogewijs et al.,  2012), as it was already described for a comparably truncated form of Mb (Ribeiro & Ramos,  2005).  Intriguingly,  the  recombinantly  expressed  truncated  human  Adgb  globin  domain  exhibited absorption spectra characteristic of a typical hexacoordinated globin – despite the  substitution of HisE7 by glutamine.   In contrast to all other globins, which are usually comprised of about 150 amino acids with  only small N‐ or C‐terminal extensions, Adgb is more than ten times larger (~1600 amino acids)  and  contains  additional  protein  domains  (as  shown  exemplarily  for  the  human  variant  in  Figure 1‐5). At the N‐terminus, the protein sequence exhibits high similarity to the catalytic  domain of human calpain‐7. However, only one of the three important active site residues is  conserved, which  indicates  that  Adgb  is  possibly  not  proteolytically  active.  In  addition,  it  carries  two  large domains with currently unknown  function: one between  the calpain and  globin domain (~300 aa) and a second, larger one at the C‐terminus (~750 aa).   12            Figure  1‐5:  Structure  of  Androglobin  and  comparison  of  its  globin  domain with  other  vertebrate  globins. A) Schematic overview of Adgb’s protein domains and intron positions. The protein starts with  a putative calpain protease domain. The globin domain is situated in the centre, but helices A and B  are shifted towards the end. The domain is also interrupted by a calmodulin binding IQ‐motif. AT the  C‐terminus, a coiled coil domain and a nuclear localisation signal were predicted bioinformatically. B)  multiple sequence alignment of human MB, CYGB, NGB and the manually rearranged globin domain  of ADGB. Adapted from (Hoogewijs et al., 2012)  qRT‐PCR analysis on different tissues of mice showed that Adgb is mainly expressed in testis  and, at a tenfold lower level, in lungs. Minor levels of Adgb mRNA could also be detected in  brain and heart (Figure 1‐6, A). A more detailed analysis of testis development revealed that  Adgb mRNA expression surges at postnatal day 25, which concurs with the postmeiotic stage  of sperm differentiation (Figure 1‐6, B). In situ hybridization experiments on cryo‐sections of  mouse  testis  tissue  displayed  stronger  signal  intensities  towards  the  lumen  of  the  seminiferous tubules, confirming this association with late stages of spermatogenesis (Figure  1‐6, C (Hoogewijs et al., 2012)). Reanalysis of public microarray data comparing expression  levels of fertile and infertile males even suggested a 4‐fold downregulation of ADGB in infertile  human individuals, hinting at a crucial function of ADGB in reproduction.   13      Figure 1‐6: mRNA expression analysis of Adgb in mice. Adgb mRNA expression was measured by qRT‐ PCR in tissues (A) and different stages of testis development (B). C) mRNA ISH experiments performed  on sections of testis from mice. The signal intensifies towards the lumen of the seminiferous tubules.  a) smooth muscle cells; b) spermatogonia; c) Sertoli cells. Adapted from (Hoogewijs et al., 2012)  Analogous  to  its  other  family  members  (Chapter  1.3),  the  structurally  complicated  and  functionally elusive Adgb has been studied in the context of cancer as well. Adgb expression  was downregulated experimentally in a glioblastoma cell culture model via lentiviral mediated  shRNA. Colony  formation and MTT assays  showed  that Adgb‐knockdown cells proliferated  more slowly and displayed apoptotic attributes (Huang, B. et al., 2014). These results suggest  that Adgb has oncogenic  function  in  glioma  cell  lines. A  second  study  (Chen et  al., 2020)  analysed Adgb in pancreatic cancer. The authors focused initially on STXBP5‐AS1, a long non‐ coding RNA that was already described as a tumour suppressor before (Huang, J. et al., 2018;  Cen et al., 2019). They found that downregulation of STXBP5‐AS1 was correlated with a poor  prognosis in pancreatic cancer and that upregulation compromised stemness and metastasis  in several pancreas‐derived cell lines. Mechanistically, they proposed that the lnc‐RNA recruits  EZH2, which then epigenetically silences the neighbouring ADGB gene. Immunoprecipitation  analysis of EZH2 confirmed binding of the transcription factor to the ADGB promotor region.  In line with this, ectopic co‐expression of ADGB and STXBP5‐AS1 reversed the positive effects  of the latter, supporting ADGBs role as an oncogene in this pancreatic cancer model (Chen et  al., 2020).  It  remained unclear, however,  if ADGB  is expressed  in  real  tumour entities and  whether it has any prognostic value there.   14          Figure 1‐7: Phylogenetic tree of Androglobin homologues identified through bioinformatic analysis of  genome databases. On the right: schematic depiction of conserved elements of Adgb domain structure  and  globin  domain  confidence  scores.  Note  the  absence  of  Adgb  orthologs  in  Nematodes  and  Crustaceans and possibly truncated version in sponges. Adapted from Hoogewijs et al. 2012.      15    1.5 Aim of this study  Although 10 years have passed since its initial description, research on Adgb is still in its early  stages. We will therefore take a basic approach to characterize Adgbs role by focusing on its  expression patterns across a variety of taxa and tissues.   The following questions will be considered:  1) When and where is Androglobin expressed?  Early studies showed high amounts of Adgb mRNA in testis tissue und lower amounts  in both  lungs and brain. However,  it  is not clear whether  this expression pattern  is  driven by a common cell type (which would suggest a similar function in all tissues) or  whether  these  tissues  should be addressed  separately  in  the  functional analysis of  Adgb. Since its initial description in 2012, a vast amount of additional expression data  was generated by omics techniques which allows a more detailed view on Adgb mRNA  expression across tissues. We will therefore integrate bulk as well as single cell mRNA  sequencing  data  from  a  variety  of  sequencing  efforts  to  elucidate  the  underlying  expressing cell types. Bioinformatical findings will then be validated in the wet lab via  qPCR and immunohistological stainings (Chapter 2.1).   2) Is the expression pattern of Adgb conserved?  Some  globins  show  very  restricted  expression  patterns  (e.g.  Ngb,  which  is  predominantly expressed  in neuronal  tissues) while others display  a diverse  set of  expression sites in different cell types and organs, which may even change in different  taxa  (e.g. GbX, as described above). Due  to  its evolutionary antiquity, Androglobin  probably has a conserved function across the metazoan tree. Unfortunately, genetic  model organisms such as Drosophila melanogaster or Caenorhabditis elegans have lost  the Adgb gene and are consequently not useful for functional studies here. We will  therefore analyse bulk and single cell mRNA sequencing data  from a diverse set of  metazoan taxa to elucidate the expression pattern of Adgb across the phylogenetic  tree and to identify suitable model organisms for further functional analysis of Adgb  (Chapter 2.4).      16        3) Which impact does the modulation of endogenous Adgb levels have?  In  humans  and mice,  Adgb  is  expressed  at  the  highest  level  in  testis  tissue  and  comparative expression studies on fertile and infertile men suggest an association of  ADGB with fertility.  It is thus not unlikely that Adgb exerts a function in a basal process  such as spermatogenesis. We will therefore analyse the transcriptomic changes upon  ablation of Adgb in testis tissue of mice (Chapter 2.2). Together with the phenotypic  analysis  of  these  Adgb  KO mice  performed  by  our  cooperation  partners, we will  elucidate Adgb’s potential function in sperm formation.     4) Is Androglobin expressed in cancer entities, and does this have prognostic value?  In vitro studies have suggested that Adgb acts as oncogene, but no studies have been  performed on real patient samples. We will therefore analyse the expression pattern  of Adgb in published RNA‐Seq datasets of tumours (e.g. from the TCGA1 consortium)  to detect changes in expression during tumour progression. In addition, we will stably  overexpress Adgb in a lung cancer cell model. Subsequent RNA‐Seq analysis may add  further evidence of Adgb’s oncogenicity and provide inside into the mode of action of  Adgb in cancer entities (Chapter 2.3).             17        18        2 Results  The results presented in this thesis were already published in scientific journals (Chapters 2.1  & 2.2) or composed in a way which would allow publishing (Chapters 2.3 &2.4).  2.1 Androglobin gene expression patterns and FOXJ1‐dependent regulation indicate  its functional association with ciliogenesis    Teng Wei Koay*, Carina Osterhof*, Ilaria M.C. Orlando, Anna Keppner, Daniel Andre, Schayan  Yousefian,  María  Suárez  Alonso,  Miguel  Correia,  Robert  Markworth,  Johannes  Schödel,  Thomas Hankeln#, David Hoogewijs#(2021) Androglobin gene expression patterns and FOXJ1‐ dependent regulation indicate its functional association with ciliogenesis. Journal of Biological  Chemistry 296:100291  */#   Authors contributed equally  Supplementary files: https://doi.org/10.1016/j.jbc.2021.100291    Own contributions to this publication:  ‐ Mapping and quantification of published bulk RNA‐Seq data on tissues from human  and cattle, revealing a new expression site for Androglobin in the female reproductive  tract  ‐ Differential gene expression and correlation analysis on bulk RNA‐Seq data to identify  genes  strongly  associated with  the  presence  of Adgb mRNA;  subsequent GO‐term  analysis and interpretation of gene lists  ‐ Experimental  validation  of  the  new  expression  site  via  qRT‐PCR  and  correlation  analysis with known ciliary marker genes  ‐ Immunohistochemical verification of Adgb protein expression  in testis and  fallopian  tube tissue from cattle  ‐ Clustering analysis of publicly available single cell sequencing data on murine lung and  testis  Planning  of  experiments,  analysis,  and  interpretation  of  data  as  well  as  drafting  of  the  corresponding part of the manuscript were realized together with Prof. Dr. T. Hankeln. The  project was managed by Prof. Dr. David Hoogewijs and Prof. Dr. T. Hankeln.   19        20    RESEARCH ARTICLE Androglobin gene expression patterns and FOXJ1-dependent regulation indicate its functional association with ciliogenesis Received for publication, August 26, 2020, and in revised form, December 17, 2020 Published, Papers in Press, January 13, 2021, https://doi.org/10.1016/j.jbc.2021.100291 Teng Wei Koay1,‡, Carina Osterhof2,‡ , Ilaria M.C. Orlando1, Anna Keppner1, Daniel Andre2, Schayan Yousefian2, María Suárez Alonso1, Miguel Correia1, Robert Markworth3, Johannes Schödel4, Thomas Hankeln2,#, and David Hoogewijs1,*,# From the 1Section of Medicine, Department of Endocrinology, Metabolism and Cardiovascular System, University of Fribourg, Fribourg, Switzerland; 2Faculty of Biology, Institute of Organismic and Molecular Evolution, Molecular Genetics & Genome Analysis, Johannes Gutenberg University Mainz, Mainz, Germany; 3Institute of Physiology, University of Duisburg-Essen, Duisburg, Germany; 4Department of Nephrology and Hypertension, Universitätsklinikum Erlangen and Friedrich-Alexander-Universität Erlangen-Nürnberg, Erlangen, Germany Edited by John Denu Androglobin (ADGB) represents the latest addition to the Globins are small globular metallo-proteins consisting of globin superfamily in metazoans. The chimeric protein com- about 150 amino acids, which comprise eight α-helical seg- prises a calpain domain and a unique circularly permutated ments in a characteristic 3-over-3 α-helical sandwich struc- globin domain. ADGB expression levels are most abundant in ture. This conserved “globin fold” identifies them as members mammalian testis, but its cell-type-specific expression, regula- of a large protein superfamily. Globins contain a heme pros- tion, and function have remained unexplored. Analyzing bulk thetic group, by which they can reversibly bind gaseous ligands and single-cell mRNA-Seq data from mammalian tissues, we such as O2, CO, and NO. Historically, the familiar vertebrate found that—in addition to the testes—ADGB is prominently O2-binding hemoglobin (HB), a tetramer of α- and β-globins, expressed in the female reproductive tract, lungs, and brain, and the monomeric myoglobin (MB) were among the first specifically being associated with cell types forming motile proteins whose sequences and structures were determined cilia. Correlation analysis suggested coregulation of ADGB already over 50 years ago. Genomic analyses have considerably with FOXJ1, a crucial transcription factor of ciliogenesis. altered and extended our view of the globin family in mam- Investigating the transcriptional regulation of the ADGB gene, mals, leading to the discovery of novel globin types such as we characterized its promoter using epigenomic datasets, neuroglobin (NGB) and cytoglobin (CYGB), which are exogenous promoter-dependent luciferase assays, and CRISPR/ expressed in nerve and fibroblast-like cells, respectively (1, 2). dCas9-VPR-mediated activation approaches. Reporter gene Both globin types perform yet-to-be-illuminated functions, assays revealed that FOXJ1 indeed substantially enhanced which possibly reside in antioxidant defense, reactive oxygen luciferase activity driven by the ADGB promoter. ChIP assays species signaling, or even lipid metabolism (3, 4). confirmed binding of FOXJ1 to the endogenous ADGB pro- Recently, a novel family of large, chimeric proteins con- moter region. We dissected the minimal sequence required for taining a globin-like domain was discovered and termed FOXJ1-dependent regulation and fine mapped the FOXJ1 androglobin (ADGB) based on its preferential expression in binding site to two evolutionarily conserved regions within the mammalian testis tissue (5). ADGB is a chimeric protein of ADGB promoter. FOXJ1 overexpression significantly increased about 1500 amino acids, which contains an embedded globin endogenous ADGB mRNA levels in HEK293 and MCF-7 cells. domain. This globin domain is permutated with respect to its Similar results were observed upon RFX2 overexpression, characteristic alpha helices and interrupted by a calmodulin- another key transcription factor in ciliogenesis. The complex binding motif. Nevertheless, the globin domain appears to be transcriptional regulation of the ADGB locus was illustrated by able to bind oxygen in vitro (5). The N-terminal domain of identifying a distal enhancer, responsible for synergistic regu- ADGB shows high sequence similarity to the human protease lation by RFX2 and FOXJ1. Finally, cell culture studies indi- calpain 7, although functionally important amino acid residues cated an ADGB-dependent increase in the number of ciliated are mutated. ADGB was shown to be highly conserved cells upon overexpression of the full-length protein, confirming throughout the metazoan tree of life, and orthologous copies a ciliogenesis-associated role of ADGB in mammals. of the ADGB gene could be found from humans and other vertebrates down to very basal taxa such as the cnidarian Nematostella vectensis, the placozoan Trichoplax adherens, and even the choanoflagellate Monosiga brevicollis (5), which suggests an elementary and possibly conserved function in This article contains supporting information. ‡ Contributed equally. metazoans. ADGB is predominantly expressed in later stages # Joint senior authors. of spermatogenesis in mammalian testes and, to a much lower * For correspondence: David Hoogewijs, david.hoogewijs@unifr.ch. J. Biol. Chem. (2021) 296 100291 1 © 2021 THE AUTHORS. Published by Elsevier Inc on behalf of American Society for Biochemistry and Molecular Biology. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). ADGB is involved in ciliogenesis and a target of FOXJ1 extent, in the lung and brain tissue (5). An important role of Both human and cattle RNA-Seq data revealed a high ADGB in spermatogenesis was supported by analysis of pub- interindividual variability in expression intensity, which sug- lished microarray data revealing that endogenous levels of gested a temporal and/or spatial restriction of Adgb expression human ADGB mRNA were lower in the testes of infertile men in the female reproductive tract. To study the gene expression versus their healthy counterparts. An in vitro cell culture study footprint of the hypothetical Adgb-expressing cell type suggested that ADGB could act as an oncogene in glioma involved, we subdivided the bovine endometrial samples into 2 formation and an ADGB knockdown could inhibit the growth groups, “Adgb-high” (TPM >20, n = 4) and “Adgb-low” (TPM of glioma cell lines (6). Overall, studies on ADGB expression <5, n = 4), and performed differential gene expression analysis patterns and gene regulation were scarce, and the functional to infer genes associated with either high or low levels of Adgb. role of ADGB has remained elusive. Subsequent overrepresentation analysis (Table 1, Since the expressional profile of a gene, specifically Supplemental File 1) revealed that genes associated with high addressing the organs and their cell types, can provide a amounts of Adgb were connected to GO-terms such as “cilium valuable hint at its possible function (as illustrated e.g., by the and axoneme assembly”, “dynein-dependent microtubular specific presence of HB in erythrocytes), we revisited the transport”, “microtubular movement” and, interestingly, the expression patterns of ADGB using an integrative approach “sperm flagellum.” An independent clustering approach to of bioinformatical data mining. In particular, novel RNA-Seq identify genes with an Adgb-type expression pattern using an datasets from bulk and single-cell experiments were additional data set of human fallopian tube samples (9) analyzed with the aim to recognize common patterns with generated a smaller subset of genes, which were even more functional implications. The data yielded valuable insight strongly associated with cilia-related processes such as “cilium into the properties of ADGB-expressing cell types, which led movement”, “determination of left-right-symmetry” and the us to characterize in detail the gene-regulatory landscape “differentiation of lung epithelial cells” (Table 2, Supplemental determining ADGB expression. We comprehensively mined File 2). Among these approximately 100 Adgb-associated epigenomic databases for accessible chromatin and genes, we found Foxj1, the master transcription factor of cil- promoter/enhancer-associated histone marks, identified iogenesis (10), and Dnah5, a protein known for its specific transcription factors binding to the ADGB locus using re- localization to motile cilia of the respiratory tract (11). The porter gene assays and chromatin immunoprecipitation tissue with the highest amount of Adgb expression, however, (ChIP) experiments, and further characterized several func- was inconsistent between the samples of the two species tional distal enhancers in the ADGB locus. Finally, we per- (Fig. 1, A and B). A possible explanation could be that the formed ADGB overexpression in vitro to elicit a cellular samples were at different stages of the menstrual cycle, given phenotype. These different lines of experimental evidence that ciliogenesis is estrogen-dependent (12, 13). Additionally, converged and convincingly pointed out that the cellular depending on the part of the oviduct that was dissected, the function of ADGB is associated with the presence of motile ratio of epithelial cells to connective tissue and thus the overall cilia. number of ciliated cells may vary between samples (14). Reproductive aging (i.e., menopause), which decreases the number of ciliated cells (15), may also have contributed to the Results observed Adgb expression differences. ADGB expression in female reproductive tract, lung, and brain Experimentally, we confirmed the fallopian tube and the suggests functional association with ciliary structures endometrium as novel expression sites via RT-qPCR analysis The wealth of gene expression data, which have been pro- in cattle. In addition, we also determined the amount of Dnah5 duced since the initial description of ADGB in 2012, enabled and Foxj1 mRNA in these samples (Fig. 1C). Foxj1 transcript us to define a much more detailed expression profile of the levels showed a positive correlation with Adgb expression in gene in mammalian tissues and cell types. As such, the bulk the endometria and oviducts (R2 = 0.73 and R2 = 0.72, RNA-Seq data of the Human Protein Atlas (7) revealed the respectively). The association between Dnah5 and Adgb in the fallopian tube of the female reproductive tract as a novel endometrium was even stronger (R2 = 0.93). Though not as expression site of ADGB mRNA (Fig. 1A). Transcript levels prominent, there was also a significant positive correlation were even higher than in the lung, which was initially between expression of Adgb and Dnah5 in the oviducts (R2 = described as the second highest ADGB-expressing human 0.74). Immunohistochemistry analysis further confirmed the organ (5). To study this further, and noticing a shortage of data localization of Adgb protein in the epithelia in the bovine from healthy human samples, we evaluated bulk RNA-Seq data endometrium and specifically in multiciliated cells in the from the female reproductive tract of cattle. The bovine data oviduct (Fig. S2). sets confirmed Adgb expression in the oviducts, showing the We previously reported the lung to show the second highest highest amount of Adgb expression of all cattle organs Adgb mRNA expression, after the testes (5). Bulk RNA-Seq analyzed, and in endometrial tissue (Fig. 1B). Human endo- data from this tissue, however, led to inconclusive results metrial data appeared largely devoid of ADGB RNA, but with high interindividual variability and overall low levels of sequencing data of separate stromal and epithelial fractions (8) expression, or, as in some human samples, no expression at all revealed restriction of ADGB expression to the epithelial (Fig. 1). Therefore, we considered analyzing available single- fraction only (Fig. S1). cell RNA-Seq data obtained from the murine lung (16). To 2 J. Biol. Chem. (2021) 296 100291 ADGB is involved in ciliogenesis and a target of FOXJ1 Figure 1. Novel Adgb expression sites correlate with cilia-associated genes. A and B, expression levels of Adgb mRNA in human and bovine tissues as determined by bulk RNA-Seq. A, high levels of expression are found in human testis, but also in fallopian tubes of females. B, transcript levels of Adgb in the oviducts of cattle exceed expression in bovine testis. C–F, correlation analysis of Adgb mRNA expression and mRNA levels of cilia-associated genes Dnah5 (left) and Foxj1 (right). Expression was measured by RT-qPCR analysis in the endometrium (n = 11, upper panels) and oviduct (n = 12, lower panels) of cattle. Adgb shows very strong correlation with Dnah5 (R2 = 0.93) in the endometria (A), strong correlation (R2 = 0.73) with Dnah5 in the oviducts (C), and strong correlations with Foxj1 in the endometria (R2 = 0.73, B) as well as oviducts (R2 = 0.72, D). *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. J. Biol. Chem. (2021) 296 100291 3 ADGB is involved in ciliogenesis and a target of FOXJ1 Table 1 GO term enrichment analysis of genes coregulated with Adgb in differential gene expression analysis of bovine endometrial samples Gene set Description Enrichment ratio FDR Biological process GO:0000226 Microtubule cytoskeleton organization 8.37 0 GO:0060271 Cilium assembly 17.66 0 GO:0007018 Microtubule-based movement 16.96 0 GO:0035082 Axoneme assembly 52.00 0 GO:0003341 Cilium movement 45.50 0 GO:0070286 Axonemal dynein complex assembly 58.50 0 GO:0007368 Determination of left/right symmetry 19.31 2.56E-09 GO:0030317 Flagellated sperm motility 15.36 6.96E-05 GO:0001539 Cilium or flagellum-dependent cell motility 29.25 1.11E-04 GO:0042073 Intraciliary transport 24.38 2.97E-04 GO:0044458 Motile cilium assembly 27.00 0.0022 Cellular component GO:0015630 Microtubule cytoskeleton 6.16 0 GO:0005929 Cilium 14.60 0 GO:0031514 Motile cilium 27.16 0 GO:0005930 Axoneme 43.15 0 GO:0036126 Sperm flagellum 26.81 0 GO:0097729 9+2 Motile cilium 25.74 6.18E-15 GO:0097223 Sperm part 17.10 3.49E-14 GO:0005815 Microtubule organizing center 6.19 5.86E-12 GO:0044447 Axoneme part 56.57 8.32E-12 GO:0005858 Axonemal dynein complex 68.95 1.69E-09 GO:0005875 Microtubule-associated complex 14.45 7.82E-09 Molecular function GO:0003774 Motor activity 15.80 5.35E-06 GO:0003777 Microtubule motor activity 21.23 1.81E-05 GO:0015631 Tubulin binding 7.11 4.41E-05 GO:1990939 ATP-dependent microtubule motor activity 30.79 0.0021 GO:0008017 Microtubule binding 6.73 0.0061 GO:0008092 Cytoskeletal protein binding 3.23 0.0087 GO:0045504 Dynein heavy chain binding 37.53 0.0087 GO:0045503 Dynein light chain binding 37.53 0.0087 All GO terms show a strong connection to the motile cilium. The associated gene list and the full list of enriched terms are provided in Supplemental File 1. prove that this method was sensitive enough to detect Adgb restricted to later stages of spermatogenesis, where round mRNA expression, we first reanalyzed single-cell RNA-Seq spermatids differentiate into elongating spermatids and form data from murine testis (17). We could show that, in accor- the flagellum, a motile microtubular structure very similar to a dance with Hoogewijs et al. (5), Adgb mRNA expression was motile cilium. Fully differentiated condensed spermatids, Table 2 Enriched GO terms in genes showing correlation with ADGB expression in human fallopian tube samples Gene set Description Enrichment ratio FDR Biological process GO:0007017 Microtubule-based process 6.02 1.28E-04 GO:0035082 Axoneme assembly 28.76 1.96E-04 GO:0003341 Cilium movement 28.30 1.96E-04 GO:0007018 Microtubule-based movement 9.71 5.76E-04 GO:0001578 Microtubule bundle formation 19.71 8.64E-04 GO:0060271 Cilium assembly 6.70 0.0088 GO:0000226 Microtubule cytoskeleton organization 5.41 0.0146 GO:0007368 Determination of left/right symmetry 12.71 0.0439 GO:0060487 Lung epithelial cell differentiation 41.77 0.0439 GO:0009855 Determination of bilateral symmetry 11.89 0.0452 Cellular component GO:0005929 Cilium 12.87 0 GO:0005930 Axoneme 30.26 3.82E-10 GO:0097014 Ciliary plasm 30 3.82E-10 GO:0031514 Motile cilium 19.38 2.28E-08 GO:0005858 Axonemal dynein complex 76.67 2.73E-05 GO:0005875 Microtubule-associated complex 16.32 2.86E-05 GO:0005874 Microtubule 6.01 0.0109 GO:0097729 9+2 Motile cilium 14.68 0.0109 GO:0036157 Outer dynein arm 76.67 0.0191 GO:0070160 Tight junction 10.95 0.0284 Molecular function GO:1990939 ATP-dependent microtubule motor activity 45.47 6.49E-06 GO:0045503 Dynein light chain binding 66.68 9.42E-06 GO:0003777 Microtubule motor activity 24.10 1.06E-04 GO:0003774 Motor activity 14.82 0.0014 GO:0051959 Dynein light intermediate chain binding 34.49 0.0228 GO:0045505 Dynein intermediate chain binding 33.34 0.0228 All GO terms reveal a strong connection to the motile cilium. The associated gene list and the full list of enriched terms are provided in Supplemental File 2. 4 J. Biol. Chem. (2021) 296 100291 ADGB is involved in ciliogenesis and a target of FOXJ1 however, did not express Adgb mRNA anymore (Fig. S3). occupancy, all indicating chromatin accessibility and suggest- Following this proof of principle, we performed clustering ing putative promoter activity (Fig. S5). Furthermore, chro- analysis on single-cell RNA-Seq data sets from epithelial matin segmentation states coupled to HMM motifs suggest fractions of murine lungs (dataset from Montoro et al. (16)). promoter activity of this region in six different cell lines. This revealed a distinct entity of lung cells expressing Adgb. Additional analysis of epigenetic modifications typical of active Using known cell-type markers from literature, and in accor- chromatin regions showed that H3K4me3 was also enriched at dance with our original report, we identified these cells as this region in multiple additional cell lines. (Fig. S5). This being multiciliated (Fig. 2). As we had observed in the corre- epigenetic profile reflecting open chromatin is in striking lation analysis on fallopian tube samples, Adgb expression contrast to the rather limited, cell-type-specific expression of correlated well with Dnah5 and Foxj1, although the overall ADGB (see Discussion). number of Adgb-positive cells was lower. An additional round To experimentally explore the basal activity of the putative of clustering of these ciliated cells revealed no subtypes with human ADGB promoter, several potential promoter fragments noticeable differences in Adgb expression, so that we assume (431 bp, 1031 bp, and 1981 bp long and starting at −33 bp that Adgb-negative ciliated cells are due to dropout artifacts upstream of the transcriptional start site—TSS) were cloned in because of rather low endogenous levels of Adgb mRNA a pGL3-luciferase basic vector (Fig. 4A). Reporter gene assays (Fig. S4). No Adgb expression was observed in progenitors of were performed in three cell lines able to form cilia (21–23) multiciliated cells, such as basal cells (Fig. 2). Cell subcluster 4 and displaying reasonable mRNA expression levels of FOXJ1 (Fig. S4) showed slightly lower levels of both, Adgb and Foxj1, and RFX2. Following transfection in HeLa and MCF-7 cells, but a higher amount of expression of the basal cell marker moderate but consistent basal promoter activity could be Aqp3 (16). This could indicate that Adgb expression rises observed (Fig. 4A). No substantial changes were seen in during differentiation and is rather associated with later stages HEK293 cells. Based on screening of ENCODE-integrated of ciliogenesis or with a maintenance function in cells with ChIP-sequencing data for candidate promoter regulating fac- already established cilia. tors, these vectors were cotransfected in HeLa cells and Cells with multiple motile cilia are not only found in the consistently increased ADGB promoter-dependent luciferase airways and the reproductive tract, but also in the ventricles of activity (Fig. S6A). Additional cotransfection experiments with the brain, where they maintain proper circulation of cerebro- increasing amounts of GATA-3 encoding plasmids indicated spinal fluid (reviewed in (18)). To obtain further evidence for a GATA-3-dependent regulation of the ADGB promoter in a functional association of Adgb and motile cilia, and looking to dose-dependent way (Fig. S6B). Next, we employed CRISPR explain the previously reported low expression in brain tissue activation (CRISPRa) technology to activate transcription at (5), we reanalyzed single-cell RNA-Seq data from mouse the ADGB promoter. CRISPRa is based on a fusion of cata- brains enriched for ependymal cells and their neuronal pro- lytically inactive Cas9 (dCas9) with the activation domains of genitors (19). As expected, we could specifically detect Adgb three potent transcription factors, VP64, p65, and Rta (dCas9- mRNA expression in fully mature ependymal cells, although VPR), which is targeted to a specific genomic region with only in a small proportion of cells (Fig. 3). In addition, a single guide RNA (sgRNA) to trigger locus-specific transcrip- subpopulation of tanycytes (designated as “2”) showed a tional activation (24). Several gRNAs, designed to bind up- moderate amount of Adgb positive cells. GO term analysis of stream of the ADGB TSS region, were tested for their capacity genes overrepresented in ependymocytes and tanycytes “2” to induce ADGB promoter-driven luciferase activity and again showed a high amount of cilia-associated genes (Fig. 3D). endogenous ADGB expression. Using two gRNA sequences Further analysis revealed that Adgb-positive tanycytes belong (termed gRNA AP-1 and gRNA AP-2), the CRISPR-based to the α-subtype, whereas β-tanycytes were Adgb-negative. system was able to substantially induce ADGB promoter- Although not multiciliated such as lung epithelial cells or driven luciferase activity in HEK293 and MCF-7 cells, vali- ependymocytes, α-tanycytes can be biciliated with the motile dating functionality of these gRNAs (Fig. 4B). Similarly, the 9 + 2 microtubule conformation, whereas β-tanycytes only CRISPR-based system also robustly activated endogenous form 9 + 0 immotile cilia, if any (20). Altogether, these data ADGB gene expression on mRNA level in both cell lines point at an association of Adgb with cilia formation and/or (Fig. 4C). Interestingly, combined transfection of gRNA AP-1 function and a possible regulation by Foxj1. and gRNA AP-2 additively facilitated expression of the ADGB gene. On the protein level, a band could be observed of The upstream sequence of the ADGB gene displays promoter slightly lower molecular weight compared with predicted activity and is inducible by CRISPRa endogenous full-length ADGB in HEK293 (Fig. 4D). Immu- Gene expression is determined to a great extent by epige- noblotting experiments displayed similar results in MCF-7 netics and regulatory elements at promoters. As information (Fig. 4D). Taken together these results confirm that the up- on this for ADGB is scarce, we first inspected data derived stream ADGB gene region possesses promoter activity. from the ENCODE consortium. ENCODE data illustrate that the upstream region surrounding the ADGB first exon displays The ADGB locus contains functional enhancers strong DNase hypersensitivity, enrichment of the promoter The cell-type-specific regulation of ADGB expression is histone mark H3K4me3, and substantial transcription factor likely to be under the control of multiple cis-regulator J. Biol. Chem. (2021) 296 100291 5 ADGB is involved in ciliogenesis and a target of FOXJ1 elements apart from the promoter alone. To further delineate mRNA levels accordingly. Collectively, these data indicate a the ADGB regulatory landscape, we mined ENCODE and complex transcriptional regulation of the ADGB locus. ReMap-based data (25, 26) within the large ADGB locus. Multiple regions with strong transcription factor occupancy FOXJ1 activates the ADGB promoter via direct binding and DNase hypersensitivity are detectable within the ADGB The ADGB expression data described above suggested a locus (Fig. S7). Furthermore, GeneHancer-derived data suggest regulation of the gene by FOXJ1, an essential transcriptional that the ADGB promoter is regulated by distal enhancer regulator of motile cilia formation. To investigate the potential elements that come in close proximity with the promoter activation of the ADGB promoter by FOXJ1, we employed by long-range chromatin looping. More precisely seven reporter gene assays on cloned ADGB promoter of varying different potential enhancers (GH06J146620, GH06J146700, lengths. Overexpression of FOXJ1 significantly increased GH06J146770, GH06J146808, GH06J146812, GH06J146815, ADGB promoter-driven luciferase activity in MCF-7, HeLa, GH06J146819) display looping to the ADGB promoter, based and HEK293 cells, substantiating that FOXJ1 represents an on correlations between epigenetic marks and the gene- ADGB promoter-targeting transcription factor (Fig. 6A). As enhancer distance algorithm implemented by the Gene- FOXJ1-mediated activation of the ADGB promoter was Hancer database (27). Five of these potential ADGB enhancer observed in promoter segments of different but overlapping elements are situated within different introns of the ADGB lengths, the binding site of FOXJ1 might be situated in the gene and two are located immediately downstream of the last smallest −33 to −464 bp region, present in all three of the ADGB exon (exon 36) (Fig. 5A). All of them coincide with cloned ADGB promoter constructs, while the presence of strong DNase hypersensitivity and substantial transcription multiple interaction sites along the longest −33 to −2014 bp factor occupancy, as well as frequent or occasional (depending fragment cannot be excluded. To analyze FOXJ1-DNA binding on the enhancer) enrichment of enhancer histone marks to the endogenous ADGB promoter region, ChIP assays were (H3K4me1, H3K4me2, H3K27ac, and H3K9ac) in multiple performed using anti-FLAG and anti-FOXJ1 antibodies in mammalian cell lines (Fig. S8). Moreover, chromatin seg- HEK293 cells transiently transfected with FLAG-tagged FOXJ1 mentation state tools suggest activity of all enhancers in constructs. To control for FOXJ1 overexpression, we analyzed several cell lines (Fig. S8). In order to experimentally investi- endogenous mRNA levels of four established FOXJ1 target gate their functionality, we first analyzed their ability to drive genes (Fig. 6B). Whereas FOXJ1 expression levels were SV40 promoter-dependent luciferase activity and cloned all strongly upregulated, also mRNA levels of its target genes seven potential enhancer elements (for convenience renamed DNAAF1, TEKT1, CCDC151, and DNAL1 were robustly as ADGB enhancers (AE) based on intronic or 3’ position in: induced following transient transfection of FOXJ1. Similarly, Int1-AE, Int12-AE, Int29-AE, Int35-AE1, Int35-AE2, 3’-AE1, immunoblotting confirmed expression of the chimeric protein and 3’-AE2) in a pGL3Prom system (Fig. 5B). Reporter gene using anti-FLAG and anti-FOXJ1 antibodies (Fig. 6B). Quan- assays in MCF-7 cells displayed enhancing effects on the SV40 titative ChIP analysis revealed more than tenfold FOXJ1 promoter in the presence of Int35-AE1, 3’-AE1, and Int12-AE enrichment at the endogenous promoter region compared (Fig. 5B), indicating that these DNA segments possess with the IgG control using two primer pairs spanning the promoter-enhancing capability. Subsequently, all potential upstream proximal ADGB region (Fig. 6C), but not at two enhancer elements were cloned in the presence of the more distal upstream and downstream regions, neither at an endogenous ADGB promoter (−1 to −464 bp upstream of the independent region on chromosome 7 (Fig. 6C). Consistently, ADGB TSS). Corresponding with the SV40 promoter-driven no binding was observed in nontransfected cells (Fig. S9). luciferase assays, Int35-AE1 and 3’-AE1, but not Int12-AE, These results confirm that FOXJ1 also binds to the endoge- increased ADGB promoter-driven luciferase activity (Fig. 5C). nous ADGB promoter. 3’-AE1 displayed a more profound enhancing effect than Int35-AE1, and the effect of Int35-AE1 diminished to basal levels when the experiment was carried out in HeLa and Evolutionary conserved nucleotides within −71 ± 30 bp HEK293 cells (Fig. 5C). Intriguingly, Int35-AE1 and 3’-AE1 upstream of the ADGB gene are required for FOXJ1 binding. display substantial sequence similarity (63% identity), with To narrow down the search for the FOXJ1-binding site the entire 3’-AE1 sequence found within Int35-AE1, with some within the ADGB promoter, we further dissected the longest differences indicative of insertional or substitutional mutations ADGB promoter segment (−33 to −2014 bp) into three (data not shown). Finally, we employed CRISPRa technology to nonoverlapping segments (Fig. 7A) indicating the absence of activate transcription at the 3’-AE1 enhancer and validated FOXJ1-mediated activation in more distal ADGB promoter enhancer capacities of 3’-AE1 in an endogenous context. segments. In contrast, FOXJ1 overexpression significantly Whereas gRNA-3’-AE1 could modestly induce 3’-AE1- increased the promoter activity of segment −1 to −464 bp, dependent ADGB promoter-driven luciferase activity suggesting that the FOXJ1-binding site is limited to this (Fig. 5D), gRNA-3’-AE1 also robustly enhanced endogenous segment closest to the ADGB TSS (Fig. 7A). Further refine- ADGB mRNA levels (Fig. 5E), albeit to a considerably lower ment of the FOXJ1 responsive region by dividing the −1 extent as compared with those targeting the ADGB promoter. to −464 bp segment into three nonoverlapping segments Importantly, dose-dependent overexpression of the 3’-AE1 indicated that only the −1 to 140 bp segment closest to the enhancer targeting gRNA increased endogenous ADGB ADGB TSS was highly activated by FOXJ1 (Fig. 7A). Next, we 6 J. Biol. Chem. (2021) 296 100291 ADGB is involved in ciliogenesis and a target of FOXJ1 further trimmed down the length of the −1 to −140 bp different Cons2-overlapping gRNAs as well as a gRNA 8 bp segment from both the 5’- and 3’- ends by 10 bp, 20 bp, 30 bp, upstream of Cons2 all significantly reduced FOXJ1-mediated and 40 bp (Fig. 7B). The incremental reduction of −1 increase of ADGB transcription in both HEK293 and to −140 bp segment from both ends at 10 bp intervals did not MCF-7 cells (Fig. 8C). Taken together, these data strongly abolish the FOXJ1-mediated increase in ADGB promoter- indicate that the FOXJ1 interaction site is located within −71 ± driven luciferase activity, although a drop in the luciferase 30 bp upstream of the ADGB TSS and involves two evolu- signals could be observed in the smaller ADGB promoter tionary conserved regions. fragments of −21 to −120 bp/−31 to −110 bp/−41 to −100 bp (Fig. 7B). These results suggest that the FOXJ1 interaction site Overexpression of FOXJ1 and RFX2 induces endogenous remains in all of these segments. When the −1 to −140 bp ADGB mRNA levels fragment was divided into two equally long, nonoverlapping parts, FOXJ1-mediated increase in promoter activity was To investigate the effect of FOXJ1 in the regulation of abolished (Fig. 7C). This indicated that the mid-region ADGB expression, we overexpressed this transcription factor of −70 bp might be important for FOXJ1 interaction, or in HEK293 cells, expressing no endogenous ADGB as well as the −1 to −70 bp and −71 to −140 bp divided segments each very little endogenous FOXJ1, and subsequently measured contain part of the FOXJ1 interaction site. Multiple sequence endogenous ADGB mRNA expression levels. Overexpression alignments of several vertebrate species based on the MULTIZ of FOXJ1 in HEK293 cells profoundly increased expression algorithm within −71 ± 30 bp upstream of the ADGB TSS levels of endogenous ADGB (Fig. 9A), further confirming a indicated the presence of evolutionary conserved nucleotides FOXJ1-dependent regulation of ADGB expression. Given the within this segment of the promoter (Fig. 8A). To further ne- cooperative functional association between FOXJ1 and RFX2,fi map the FOXJ1 binding site, we separately mutated three re- another essential transcriptional regulator of ciliogenesis (28), gions within the −71 ± 30 bp ADGB promoter region, one we also assessed a potential RFX2-dependent regulation of containing a single conserved nucleotide (termed Cons1), one ADGB. Consistent with the RNA-Seq results of Rfx2-deficient region displaying evidence of evolutionary constraint as re- mice suggesting a RFX2-dependent regulation of ADGB flected by phyloP and phastCons scores (termed Cons2), and transcription (29), also overexpression of RFX2 in one within the mid-point at −71 bp, as the separation of this HEK293 cells increased endogenous ADGB expression levels region resulted in abolished FOXJ1-mediated activation (Fig. 9A), albeit to a lower extent than FOXJ1. FLAG-tag-based (Fig. 7C). It was suggested that Fox TF-binding sites are immunoblotting experiments excluded that this discrepancy approximately 8 to 10 bp in length (28). Therefore, we mutated arose from differences in plasmid expression (Fig. 9B). As the these regions by substitution of 5 to 6 residues with tandem A 36-exon containing ADGB pre-mRNA might be alternatively and/or T, which are likely to be suf cient to disrupt potential spliced to produce different variants of the protein, wefi FOXJ1 binding (Fig. 8A). For Cons2, two separate mutants of confirmed these results by employing multiple exon–exon proximal and distal parts were constructed due to the 12 bp primer pairs for RT-qPCR across the whole ADGB transcript size of the conserved region. Interestingly, mutations on Cons1 (Fig. S11). Similar results were obtained in MCF-7 cells and Cons2 abolished the FOXJ1-mediated increase in ADGB (Fig. 9A). Collectively, these findings indicate that ADGB is a promoter-driven luciferase activity, whereas the mid-point downstream effector of the two master regulators of cilio- mutation did not. (Fig. 8B). This result suggests that both genesis FOXJ1 and RFX2, further suggesting a potential role of conserved regions (Cons1 and Cons2) within the promoter ADGB in the formation and/or function of cilia. might be important for FOXJ1 interaction. Moreover, these conserved regions are probably mutually dependent on each FOXJ1 and RFX2 synergistically activate the ADGB promoter other to mediate FOXJ1 interaction as the absence of either in the presence of the 3’-AE1 enhancer in reporter assays part disrupted FOXJ1-mediated activation on the ADGB pro- In order to understand the role of RFX2 in regulating the moter. This might also explain the abolished FOXJ1 activation cis-regulatory elements of the ADGB gene, we employed re- in pGL3B-AP70-1 and pGL3B-AP140-71 (Fig. 7C), as both of porter gene assays to elucidate the interaction of RFX2 on these interdependent FOXJ1 interaction sites were separated these regulatory elements. Subsequently, we postulated that in these constructs. Similar findings were obtained with an RFX2 might be binding to the enhancer elements that are in ADGB promoter fragment of reduced size (Fig. S10). Finally, to close proximity with the promoter. In order to exclude the independently validate the FOXJ1-binding site endogenously, possible influence of the endogenous ADGB promoter on the we employed the CRISPR/dCas9 approach with ADGB pro- readout, we co-overexpressed RFX2 with heterologous SV40 moter gRNAs to block the genomic-binding site in the pres- promoter-driven luciferase reporter constructs coupled with ence of exogenous FOXJ1. The docking of a dCas9 variant ADGB enhancers. Our results displayed no RFX2-mediated uncoupled from VPR onto the putative FOXJ1-binding site activation of promoter activity despite the presence of hinders the interaction of FOXJ1 with the ADGB promoter. enhancer elements (Fig. S12), suggesting that RFX2 might not Similar to the negative control gRNA cotransfection of two interact with any of these enhancer elements in a direct more remotely located gRNAs (−590 and −119 bp upstream of manner. Next, we examined the role of RFX2-mediated acti- the ADGB TSS) had no effect on the FOXJ1-dependent acti- vation of ADGB regulatory elements in the presence of FOXJ1, vation of ADGB expression. In contrast, transfection of three ADGB promoter and enhancer elements. Reporter assays J. Biol. Chem. (2021) 296 100291 7 ADGB is involved in ciliogenesis and a target of FOXJ1 illustrated no difference in ADGB promoter activity across all Specifically, substantial ADGB expression was also observed in coupled enhancers with RFX2 overexpression alone (Fig. S12). the female reproductive tract, which obviously adds a However, under FOXJ1 overexpression conditions, FOXJ1- completely new perspective on its original designation sug- mediated activation in each of the ADGB promoter-driven gesting a predominant role in males. Prominent ADGB mRNA reporter constructs could be observed. Interestingly, co- expression was also observed in the lung epithelial cells and in overexpression of FOXJ1 and RFX2 displayed additive activa- the brain ependymocytes and α-tanycytes, altogether pos- tion of ADGB promoter activity only in the presence of 3’-AE1 sessing motile cilia. A particularly interesting result was the enhancer element (Fig. 9C), but none of the other ADGB en- bioinformatically inferred correlation of ADGB expression hancers (Fig. S12), indicating that the presence of FOXJ1 and with the master transcription factor of ciliogenesis, Foxj1 (see the 3’-AE1 enhancer is a prerequisite for RFX2-mediated below). Additional independent support for a ciliogenesis- activation of the ADGB promoter. Similar experiments in associated role of ADGB came from the CiliaCarta database, two independent cell lines, HEK293 and HeLa, validated the a multiomics-based comprehensive ciliary compendium sug- 3’-AE1 enhancer-dependent regulation. Whereas this regula- gesting that ADGB is a human ciliome component with a high tion was additive in HEK293, a synergistic regulation could be rank and probability score (31). A recent evolutionary prote- observed for HeLa cells (Fig. 9C). As this FOXJ1-RFX2 synergy omics approach (32) indicated that the association of ADGB was not observed in the sole ADGB promoter-driven reporter with ciliary structures may be phylogenetically ancient and can constructs, the 3’-AE1 enhancer is crucial in mediating the even be detected down to the flagellum-containing choano- synergistic effect. Hence, these experiments strongly suggest flagellates, which is in agreement with our own phylogenetic that RFX2 supports the FOXJ1-mediated regulation of ADGB reconstruction of ADGB ancestry (5). In fact, such a phylo- expression with the presence of the essential 3’-AE1 enhancer. genetic perspective lends additional weight to the proposed functional association of ADGB and cilia: despite the presence Ectopic ADGB overexpression promotes ciliogenesis of numerous globin genes in their genomes (33–35), As FOXJ1 and RFX2 represent both critical regulators of Drosophila melanogaster and Caenorhabditis elegans both are ciliogenesis, a potential role of ADGB in the formation and/or missing orthologues of ADGB (5), and both organisms are function of cilia is plausible. To explore a putative role for devoid of motile cilia on their somatic cells (36). Movement of ADGB in ciliogenesis, we examined ADGB requirements in spermatozoa in C. elegans is achieved via pseudopods, which cilia formation in cellular models. Due to the lack of cellular carry no resemblance to classical flagella (37). Drosophila, on models with robust endogenous ADGB expression levels, we the other hand, still develops motile flagella during sper- performed ADGB overexpression. Immunostaining with anti- matogenesis (38). However, the fly harbors two testis-specific acetylated tubulin revealed that the number of cilia was sub- globins, which are not phylogenetically related to ADGB stantially increased following ADGB overexpression in human (39). In expression analyses, these Drosophila testis globins HeLa cells (Fig. 10A). Cilia formation was similarly increased were correlated with genes characterized by GO-terms such as in ciliated mouse cortical collecting duct cells following sperm axoneme assembly and motility (40). It is therefore overexpression of ADGB, almost comparable with serum- tempting to speculate that these globins at least partially starved induction of ciliogenesis (Fig. 10B). These findings compensate for the loss of ADGB in the fruit fly. In addition, are in perfect agreement with the expression analyses pre- ADGB orthologues appear to be missing in the phylum of sented above and collectively suggest that ADGB is associated crustaceans (5), which form specialized, mostly immobile with ciliogenesis and could play an evolutionarily conserved spermatozoa (41). role in the formation and/or maintenance of cilia. The observed strictly cell-specific expression pattern of ADGB prompted us to comprehensively investigate its tran- scriptional regulation. In accordance with a suggested role in Discussion ciliogenesis, our experimental data provided direct evidence ADGB, the fifth member of the mammalian globin family that the ADGB gene is indeed regulated by FOXJ1. For further (5), is a chimeric protein with an unusual, embedded globin confirmation, we inspected transcriptome screens of FOXJ1 domain that is circularly permutated and exhibits hallmarks of knockout and overexpression models in mouse, zebrafish, and a hexacoordinated heme-binding scheme (30). Intriguingly, frogs (42–45) and detected consistent evidence for FOXJ1- abundant expression of ADGB in various species seemed to be dependent expression levels of ADGB in those data sets. restricted to the testis tissue (hence its name) and, more spe- Furthermore, a recent in silico study of FOXJ1-mediated cifically, to postmeiotic stages of spermatogenesis. The func- regulatory and signaling networks predicted ADGB as one tion of ADGB, however, has remained unclear. Since the gene’s of the direct FOXJ1-regulated genes (46). During spermato- initial description (5), a wealth of transcriptome data has been genesis, FOXJ1 expression coincides with the timely stages of produced by the scientific community, facilitating a re- flagella formation where it probably orchestrates the expres- evaluation of ADGB’s expression profile. Our extended sion of genes essential for flagella biogenesis (47, 48). The expression analysis of ADGB in mammalian tissues based on broader role of FOXJ1 as the master regulator of motile cil- bulk and single-cell RNA-Seq data, including confirmation by iogenesis has been reported as well (10). In addition to RT-qPCR and immunohistochemistry, revealed that ADGB is FOXJ1-dependent expression, our reporter gene assays and consistently detected in cells carrying motile cilia or flagella. overexpression experiments revealed that the ADGB gene is 8 J. Biol. Chem. (2021) 296 100291 ADGB is involved in ciliogenesis and a target of FOXJ1 Figure 2. Clustering analysis of single cell RNA-Seq data from murine lungs. A, tSNE representation of cell clusters (named in accordance with (16)). B, visualization of all clusters expressing mRNA of Adgb and ciliary marker genes Dnah5, Foxj1 and Cdhr3. Adgb expression is restricted to ciliated cells in murine lung epithelia. J. Biol. Chem. (2021) 296 100291 9 ADGB is involved in ciliogenesis and a target of FOXJ1 Figure 3. Analysis of single cell RNA-Seq data from mouse hypothalamus. A, tSNE representation of brain cells clustered by levels of expression similarity. Cell types were named in accordance with the initial publication (19). B, mRNA expression levels of Adgb, Dnah5 and Foxj1. Adgb expression is most prominent in ependymocytes, but also in subpopulation “2” of tanycytes. Foxj1 expression is also found in these two clusters and absent in tanycytes “1”. Dnah5 expression is restricted to ependymocytes. C, percentage of Adgb-positive cells in ependymocytes and tanycytes subtypes. D, gene ontology analysis of genes overrepresented in ependymocytes and tanycytes “2”. Adgb positivity correlates with terms connected to cilia. 10 J. Biol. Chem. (2021) 296 100291 ADGB is involved in ciliogenesis and a target of FOXJ1 A MCF-7 HeLa HEK293T Luc -464 Prom. -33 Luc * * -1064 Prom. -33 Luc ** * -2014 Prom. -33 Luc * ** 0 2 4 6 8 0 4 8 12 0.0 0.5 1.0 1.5 2.0 RLU (firefly/Renilla) RLU (firefly/Renilla) RLU (firefly/Renilla) B HEK293T MCF-7 **** **** 120 ** 1200 *** pGL3B **** *** 90 900 pGL3B-AP464 60 600 30 300 3 2 2 1 1 0 0 control gRNA - + - - - control gRNA - + - - - gRNA AP-1 - - + - + gRNA AP-1 - - + - + gRNA AP-2 - - - + + gRNA AP-2 - - - + + C 180 **** ***1000 **** 135 **750 90 500 45 250 5.0 5.0 2.5 2.5 0.0 0.0 control gRNA - + - - - control gRNA - + - - - gRNA AP-1 - - + - + gRNA AP-1 - - + - + gRNA AP-2 - - - + + gRNA AP-2 - - - + + D gRNA (- ctrl) - + - gRNA (- ctrl) - + - gRNA AP-1 - - + gRNA AP-1 - - + gRNA AP-2 - - + gRNA AP-2 - - + 250 kDa 250 kDa 150 kDa ADGB 150 kDa ADGB 100 kDa 100 kDa 75 kDa 75 kDa Figure 4. The upstream region of the ADGB transcriptional start site contains promoter activity and is inducible by CRISPRa. A, luciferase reporter assays of ADGB promoter (AP) elements of three different lengths from −33 bp to −2014 bp, −1064 bp, or −464 bp, respectively, upstream of the ADGB TSS in MCF-7 cells, HeLa cells, and HEK293T cells, showing consistent increase in ADGB promoter-driven luciferase activity in MCF-7 and HeLa (n = 3 inde- pendent experiments). Results are displayed as ratios of firefly to Renilla luciferase activities in relative light units (RLU) and normalized to results from pGL3- Basic control transfected cells. Schematic representation of cloned fragments upstream of the ADGB gene in a pGL3-Basic vector is shown with numbers representing positions corresponding to the first nucleotide of the TSS. B, HEK293T and MCF-7 cells were transfected with dCas9-VPR along with ADGB promoter (AP)-targeting gRNAs (gRNA AP-1 and/or gRNA AP-2) and ADGB promoter (pGL3B-AP464)-driven luciferase constructs. The gRNA used as negative control contains a nonspecific sequence as present in the pSPgRNA plasmid. Cas9-VPR-based activation of ADGB promoter (pGL3B-AP464)-driven luciferase constructs results in activation of the ADGB promoter construct (n = 3 independent experiments). Results are displayed as ratios of firefly to Renilla luciferase activities in RLU. C, HEK293T and MCF-7 cells were transfected with dCas9-VPR along with ADGB promoter-targeting gRNA AP-1 and/or gRNA AP-2, and relative ADGB transcript levels were quantified by RT-qPCR using a negative control gRNA as reference. Single-guide activation of the ADGB promoter with gRNA AP-1 and gRNA AP-2 results in substantial increment in ADGB transcript levels (n = 4 independent experiments). Simultaneous expression of gRNA AP- 1 and gRNA AP-2 leads to synergistic activation of endogenous ADGB expression (n = 4 independent experiments). D, immunoblotting of immunopre- cipitated ADGB from HEK293T and MCF-7 cells after gRNAs-dCas9-VPR-activation for 72 h detects endogenous ADGB expression. Data represent mean ± S.E.M (error bars); *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. also robustly regulated by RFX2, a transcriptional activator of Rfx2-dependent expression of Adgb. The latter study also spermiogenesis (29). Again, this is in strong agreement with listed Rfx2 binding to the mouse Adgb locus in ChIP- transcriptome data from Rfx2-deficient mice (29) showing sequencing experiments. J. Biol. Chem. (2021) 296 100291 11 Relative mRNA level RLU (firefly/Renilla) Relative mRNA level RLU (firefly/Renilla) ADGB is involved in ciliogenesis and a target of FOXJ1 A Scale 100 kb hg38 chr6: 146,610,000 146,640,000 146,670,000 146,700,000 146,730,000 146,760,000 146,790,000 146,820,000 GENCODE v32 Comprehensive Transcript Set ADGB Clustered interactions of GeneHancer regulatory elements and genes (Double Elite) GH06J146598 Enhancers and promoters from GeneHancer GH06J146620 GH06J146700 Int1-AE Int12-AE GH06J146770 Int29-AE Int35-AE1 GH06J146808 GH06J146812 Int35-AE2 GH06J146815 3’-AE1 GH06J146819 Interactions between GeneHancer regulatory elements and genes 3’-AE2 B D E SV Luc pGL3B 125 ng 250 ng pGL3B-AP-3'-AE1 none 375 ng *** Int35-AE1 ** 7.5 6 ****** * Int35-AE2 3’-AE1 ** 5.0 4 3’-AE2 Int29-AE 2.5 2 Int1-AE Int12-AE 0.0 0 0 1 2 3 4 5 control gRNA - - + - control gRNA + - - - RLU (firefly/Renilla) gRNA 3'-AE1 - - - + gRNA 3'-AE1 - + + + C -464 Prom. -1 Luc MCF-7 HeLa HEK293T none Int35-AE1 ** Int35-AE2 3’-AE1 **** **** 3’-AE2 Int29-AE Int1-AE Int12-AE 0 1 2 3 4 0.0 0.5 1.0 1.5 2.0 2.5 0.0 0.5 1.0 1.5 2.0 RLU (firefly/Renilla) RLU (firefly/Renilla) RLU (firefly/Renilla) Figure 5. Distal enhancers regulate ADGB promoter-driven gene transcription. A, potential ADGB regulatory enhancer elements interacting via long- range looping with the ADGB promoter region, derived from the GeneHancer database, are displayed in the UCSC Genome Browser. B, seven potential enhancer elements predicted to be in close proximity with the ADGB promoter region were cloned at −27 bp upstream of the SV40 promoter in a luciferase vector. These constructs were cotransfected into MCF-7 cells together with a Renilla control plasmid, and the effect of these enhancer elements was assessed. Results are displayed as ratios of firefly to Renilla luciferase activities in relative light units (RLU). The presence of Int35-AE1, 3’-AE1 and Int12-AE enhancer candidates increased SV40 promoter-driven luciferase activity (n = 3 independent experiments). C, the seven potential enhancer elements were cloned at 268 bp downstream of the 464 bp ADGB promoter-driven luciferase reporter gene. Consistent with (C), Int35-AE1 and 3’-AE1 sequences, but not the Int12-AE sequence, increased ADGB promoter-driven luciferase activity in MCF-7 cells. When tested in HeLa and HEK293T cells, 3’-AE1 sequence also 12 J. Biol. Chem. (2021) 296 100291 RLU (firefly/Renilla) Relative mRNA level ADGB is involved in ciliogenesis and a target of FOXJ1 A recent study reported that the cooperation of FOXJ1 and FOXJ1 in this RFX-motif was reported, probably due to RFX2 has a prominent role at promoters of ciliary genes cobinding of FOXJ1 and RFX factors (28). Strikingly, this compared with other established cilia transcription factors. binding motif displays very strong sequence similarity to the Both transcription factors were found to be positioned at the Cons2 region in the ADGB promoter whose mutation abol- anchor end point of chromatin loops, where RFX2 was sug- ished FOXJ1-mediated activation. Therefore, our gested to act as a scaffolding factor to stabilize the distal conservation-based analysis of FOXJ1 binding on the ADGB enhancer element with the proximal promoter, thus bringing promoter is independently validated by the in silico analysis of the enhancer-binding FOXJ1 closer to the promoter (28). FOXJ1-enriched motifs. Consistent with these findings, our study describes that the The transcription factor p73 plays a major role in cilio- remote enhancer 3’-AE1, located downstream of the ADGB genesis and acts upstream of FOXJ1 and RFX2 (49). Nem- gene, is important for RFX2 to cooperate in synergy with ajerova et al. (50) reported that TP73 deficiency broadly FOXJ1 in activating ADGB promoter-driven luciferase activity. attenuates ciliary gene expression by transcriptome analysis It is thus likely that RFX2 acts as the mediator that enables the of mouse tracheal epithelial cells (mTEC) derived from WT connection between enhancer 3’-AE1 and the ADGB pro- and TAp73-deficient mice. In line with a role of ADGB in moter, which could explain the lack of RFX2-mediated ciliogenesis, mTEC air–liquid interfaces (ALI) cultured for 0, transactivation of 3’-AE1-dependent SV40-driven luciferase 4, 7, and 14 days of differentiation displayed increasingly activity. Correspondingly, RFX2 also shows little to no acti- abundant ADGB expression. Moreover, RNA-Seq-based vation on enhancer 3’-AE1 coupled with the ADGB promoter, transcriptome analysis of ALI cultured mTECs derived from suggesting a possible scaffolding promoter–enhancer mediator TAp73-deficient mice showed significantly reduced ADGB role with no transcriptional activity. It remains to be deter- levels, further substantiating a Tp73-dependent regulation, mined if, in an endogenous genomic context, RFX2 is crucial either directly or more likely via its downstream targets Rfx2/ to establish the connection between the ADGB promoter and Foxj1, which both displayed downregulated mRNA levels in enhancer 3’-AE1, explaining the modest upregulation of TAp73 knockout mice. Simultaneously, ChIP-Seq experi- endogenous ADGB upon RFX2 overexpression. In slight ments (50) linked p73 directly to FOXJ1/RFX2 and, most contrast to the study of Quigley and Kintner (28), our findings interestingly, revealed p73 binding to the distal ADGB from reporter assays indicate a functional interaction of FOXJ1 enhancer 3’-AE1 of the ADGB locus. This exquisite de- at the ADGB promoter rather than at its distal enhancers. In pendency of ADGB expression on ciliogenesis-associated our reporter gene studies, RFX2 synergistically activates the transcription factors is accompanied by open chromatin ADGB promoter with FOXJ1 only in the presence of the distal marks at the ADGB promoter. Surprisingly, this epigenetic enhancer 3’-AE1, which is in line with Quigley and Kintner feature was also observed in a variety of transcriptionally si- (28) and indicates that FOXJ1 is stabilized at promoters of cilia lent cell types, which—for unknown reasons—may thus genes through cooperative interactions with RFX2. contain poised promoters. Our study provided extensive efforts in refining the inter- In conclusion, our study provides first-time evidence that action site(s) of FOXJ1 on the ADGB promoter, which enabled ADGB is specifically expressed in cell types with motile cilia, the identification of evolutionarily conserved nucleotides that that its cellular role is most probably associated with cilia are crucial for FOXJ1-mediated activation of the ADGB pro- biogenesis and function, and that it is a direct regulatory target moter. The FOXJ1-binding motif has not been fully annotated of FOXJ1 in a complex regulatory landscape. The exact role of so far. A single computational study deduced the preferential ADGB in ciliogenesis remains to be established. Future in- binding of FOXJ1 to the consensus sequence NNN[G/A] vestigations involving the generation of new animal models TAAACAAANNN, with N representing any nucleotide (46). with conditional knockout of Adgb in ciliated tissues will However, only a sole motif with this consensus can be found hopefully reveal the intriguing physiological role of ADGB in within the −2014 to −1065 bp upstream ADGB promoter cilia formation and the contribution of FOXJ1- and RFX2- sequence, whereas motifs with less stringent sequence simi- dependent gene regulation. larity can be found within −465 to −2014 bp upstream of the ADGB TSS. From our experimental data, this part of the Experimental procedures ADGB promoter shows no FOXJ1-mediated activation. In another study employing Xenopus laevis, an RFX-based anal- Analysis of bulk RNA sequencing, single-cell RNA sequencing, ysis for binding motifs in the promoters of multi-cilia-related and microarray data genes has proposed a consensus binding motif in human Publicly available transcriptome raw data (Table S1) were orthologs (TTCCTGGAAAC). Although this motif was sug- downloaded from either NCBI or ENA web servers (https:// gested to be the binding site for RFX TFs, also enrichment of www.ncbi.nlm.nih.gov/sra; https://www.ebi.ac.uk/ena). We displays an enhancing effect on ADGB promoter-driven luciferase activity (n = 3 independent experiments). D, dCas9-VPR-based activation of ADGB enhancer-dependent ADGB promoter (464 bp)-driven luciferase constructs using a 3’-AE1 ADGB enhancer-targeting gRNA (gRNA-3’-AE1) results in increased luciferase activity. E, HEK293T cells were transfected with dCas9-VPR along with different amounts of the 3’-AE1 ADGB enhancer-targeting gRNA (gRNA-3’- AE1) and relative ADGB transcript levels were quantified by RT-qPCR using the negative control as reference (n = 3 independent experiments). Data represent mean ± S.E.M (error bars); *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. J. Biol. Chem. (2021) 296 100291 13 ADGB is involved in ciliogenesis and a target of FOXJ1 only included data from Illumina machines with a minimal BBDuk (https://sourceforge.net/projects/bbmap/). We did not read length of 50 nt. For organism-wide gene expression apply the same trimming parameters to all the data sets to (PRJEB6971 (Hsa; https://www.proteinatlas.org/) and account for differences in quality and sequencing length; PRJNA263600 (Bta)), we focused on data from large however, differential expression analysis was only performed sequencing consortia, to ensure comparability. Trimming pa- on data sets from the same study with the same trimming rameters were assessed for each data set via inspection with mode. After processing, the reads were mapped against the FastQC (https://www.bioinformatics.babraham.ac.uk/projects/ corresponding reference genomes of either Homo sapiens fastqc/). Adapter and quality trimming were performed with (GRCh 38) or Bos taurus (Bta UMD3.1) with HISAT2 (51). A MCF-7 HeLa HEK293 Luc ** -464 Prom. -33 Luc * * *** -1064 Prom. -33 Luc * * ** - FOXJ1 -2014 Prom. -33 Luc ** * ** + FOXJ1 0 5 10 15 0 20 40 60 0 5 10 15 20 RLU (firefly/Renilla) RLU (firefly/Renilla) RLU (firefly/Renilla) B - FOXJ1 + FOXJ1 FOXJ1 TEKT1 DNAAF1 DNAL1 CCD151 50000 *** 10 60 20 8 * ** *** * 40000 8 15 6 30000 6 40 10 4 20000 4 20 10000 2 5 2 0 0 0 0 0 Empty plasmid + - Empty plasmid + - FLAG -FOXJ1 - + FLAG -FOXJ1 - + 75 kDa 75 kDa FLAG FOXJ1 50 kDa 50 kDa C 0.04 IgG ** FOXJ1 0.03 ** * Flag 0.02 ** 0.01 0.00 -184/+21 -309/-184 3' ADGB 5' ADGB Chr. 7 Figure 6. FOXJ1 activates the ADGB promoter via direct binding. A, reporter gene assays of ADGB promoter elements in MCF-7, HeLa, and HEK293T cells with or without co-overexpression of FOXJ1 display a FOXJ1-induced activation of ADGB promoter-driven luciferase activity. This increase in promoter activity is seen in all promoter elements of different lengths (−33 bp to −464, −1064, and −2014 bp upstream of the ADGB TSS). Results are displayed as ratios of firefly to Renilla luciferase activities in relative light units (RLU) and normalized to the pGL3-basic vector (n = 3 independent experiments). B, mRNA and protein experiments of HEK293 cells transiently transfected with a FLAG-FOXJ1 vector. FOXJ1 target gene mRNA levels were measured by RT-qPCR and normalized to β-actin mRNA levels. Immunoblotting using anti-FLAG and anti-FOXJ1 antibodies also confirmed FOXJ1 overexpression. C, the amount of coprecipitated chromatin derived from the proximal ADGB promoter region using two primer pairs (covering +21 to −184 and −184 to −309 upstream of the ADGB TSS, selected based on the reporter gene assays), its upstream (5’) and downstream (3’) regions as well as an independent region on chromosome 7 in the EPO locus (64), was determined by qPCR. Data represent mean ± S.E.M (error bars); *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. 14 J. Biol. Chem. (2021) 296 100291 Relative mRNA level % INPUT ADGB is involved in ciliogenesis and a target of FOXJ1 A -2014 Prom. -1 Luc - FOXJ1 + FOXJ1 pGL3B -2014 -1065 pGL3B-AP2014-1065 -1064 -465 pGL3B-AP1064-465 -464 -1 pGL3B-AP464-1 ** -464 -271 pGL3B-AP464-271 -270 -141 pGL3B-AP270-141 -140 -1 pGL3B-AP140-1 *** 0 10 20 30 RLU (firefly/Renilla) B -140 Prom. -1 Luc - FOXJ1 + FOXJ1 pGL3B -140 -1 pGL3B-AP140-1 * -130 -11 pGL3B-AP130-11 ** -120 -21 pGL3B-AP120-21 * -110 -31 pGL3B-AP110-31 * -100 -41 pGL3B-AP100-41 ** 0 5 10 15 RLU (firefly/Renilla) C -140 Prom. -1 Luc - FOXJ1 + FOXJ1 pGL3B -140 -1 pGL3B-AP140-1 **** -140 -71 pGL3B-AP140-71 -70 -1 pGL3B-AP70-1 0.0 2.5 5.0 7.5 10.0 RLU (firefly/Renilla) Figure 7. The proximal ADGB promoter confers FOXJ1-mediated increase in luciferase activity. A, the longest ADGB promoter (AP) element, AP2014 (−33 to −2014 bp upstream of TSS), displaying FOXJ1-induced promoter activity was first divided into three nonoverlapping segments (−1065 to −2014 bp, −465 to −1064 bp, and −1 to −464 bp upstream of the ADGB TSS) and cloned individually upstream of the firefly luciferase gene. Only the segment −1 to −464 bp upstream of the ADGB TSS displays FOXJ1-induced promoter activity, and no induction is observed with inserts covering the −465 to −2014 bp upstream of the ADGB TSS (n = 4 independent experiments). Schematic representation of different ADGB promoter segments is shown with numbers representing positions corresponding to the first nucleotide of the TSS. The −1 to −464 bp ADGB promoter construct was further subdivided into three nonoverlapping segments (−271 to −464 bp, −141 to −270 bp, and −1 to −140 bp upstream of the ADGB TSS) and cloned individually upstream of the firefly luciferase gene. From those only insert −1 to −140 bp displays FOXJ1-induced promoter activity (n = 4 independent experiments). B, the −1 to −140 bp ADGB promoter construct was sequentially reduced at both the 5’- and 3’- ends by 10 bp, 20 bp, 30 bp, and 40 bp, to generate smaller segments of −11 to −130 bp/−21 to −120 bp/−31 to −110 bp/−41 to −100 bp upstream of the ADGB TSS. All segments display FOXJ1-induced promoter activity including the smallest 60 bp segment of −41 to −100 bp (n = 3 independent experiments). C, −1 to −140 bp ADGB promoter element was subdivided into two segments of 70 bp, −1 to −70 bp and −71 to −140 bp. In both constructs, FOXJ1-induced promoter activity was abolished (n = 3 independent experiments) suggesting a disruption to the FOXJ1 binding site or missing binding sites for proper FOXJ1 interaction. Calculation of transcripts-per-million values (TPM) was per- Mapped single-cell data (Table S1) were downloaded from formed with StringTie (52) in guided mode. Differential gene NCBI as UMI count tables. In the lung data set, we only expression analysis (Bta) as well as hierarchical clustering of included mice #2 and #3, which share the same genetic genes (Hsa) was done in R (https://www.r-project.org/) with background, retaining 5010 out of 7193 cells from the original the help of Bioconductor’s DESeq2 package (53). Gene analysis (16). Graph-based clustering of cells was performed ontology analysis was performed with WebGestalt 2019 (54). using Seurat 3 (55). Our clusters were mostly in accordance J. Biol. Chem. (2021) 296 100291 15 ADGB is involved in ciliogenesis and a target of FOXJ1 A Scale 20 bases hg19 chr6: 146,920,060 146,920,080 146,920,100 C A G G G C C C C G C C C C C G C T C T C C G G G C G C T G G A C G C G G G A C G C C G T C T C C T G G C A A C G C A G 4.5 _ 100 vertebrates Basewise Conservation by PhyloP 0 - -4.5 _ 1 _ 100 vertebrates conservation by PhastCons 0 _ Multiz Alignments of 100 Vertebrates Human C A G G G C C C C G C C C C C G C T C T C C G G G C G C T G G A C G C G G G A C G C C G T C T C C T G G C A A C G C A G Chimp C A G G G C C C C G C C C C C G C T C T C C G G G C G C T G G A C G C G G G A C G C C G T C T C C T G G C A A C G C A G Gorilla C A G G G C C C C G C C C C C G C T C T C C G G G C G C T G G A C G C G A G A C G C C G T C T C C T G G C A A C C C A G Orangutan C A G G G A C C C G C C C C C G C C C T C C G G G C C C T G G A C G C G G G A C G C C G T C T C C T G G C A A C G C A G Gibbon C A G G G A C C C G C C C C C G C T C T C C G G G C G C T A G A C G C A G G A C G C C G T C T C C T G G C A A C G C A G Rhesus C A G G G A C C C G C C C C G G C T C T C A G G G C G C T A G A C G C C G G A C G C C G T C T C C T G G C A A C G C A G Crab-eating macaque C A G G G A C C C G C C C C G G C T C T C A G G G C G C T A G A C G C C G G A C G C C G T C T C C T G G C A A C G C A G Marmoset C A A G G A C C C G C C C C C G C T C T C C G G G C G C G G G A C G C T G C A C G C C G T C T C C T G G C A A C G C A G Bushbaby - - - - G C C C C G C C C C G C C C C G C C C G A C G C C C G A C C C C G C C C G A C G T C T C C T G G C A A C G C A G Chinese tree shrew C A G - - - - - C G C C C T C G C G C T C T G G A - - - - - - - C G C T G T C C G A G G T C T C C C G G C A A A G C A G Squirrel G C A G G A C C C - G C C C C G C G C T C A G G A C G - - - - - - - - C A - G C G C T G T C T C C T G G C A A C G C C A Lesser Egyptian jerboa C T G C G A G C C - T T C G G G C G C T C - - - G C T C G G G A - - - C G C G C G C G G T T G C C C G G C A A = G C A G Prairie vole G C T G G A G C C - T T C C G G T T C T C A G A A C T T G G T C - - A C G G T C T C C G T C T C C T G G C A A C G C G G Chinese hamster G C A G G A T T C - T T C C G G T T C T C A G A A C T - - - - - - - - C C T T C A C T G T C G C C T G G C A A G G C A G Golden hamster G C A G G A G C C - T T C C G G T T C T C A G A A C T - - - - - - - - C C T T C T C T G T C G C T T G G C A A G G C A G Mutations: A A A A A A A A T T T A A A A A A A A A T T T (AP140-1_Mut1) (AP140-1_Mut2) (AP140-1_Mut3) (AP140-1_Mut4) Cons_1 Mid-region Cons_2 B -140 Prom. -1 Luc - FOXJ1 + FOXJ1 pGL3B -100 -41 -140 -1 pGL3B-AP140-1 *** -100 -41 #### -140 -1 pGL3B-AP140-1_Mut1 * -100 -41 -140 -1 pGL3B-AP140-1_Mut2 **** -100 -41 -140 -1 pGL3B-AP140-1_Mut3 -100 -41 -140 -1 pGL3B-AP140-1_Mut4 0 5 10 15 20 RLU (firefly/Renilla) C HEK293T MCF-7 600 50 400 40 dCas9 - + 200 30 * 250 kDa 60 *** *** 20 ****** ** 150 kDa dCas9 30 **** 10 *** 2 1 100 kDa 0 1 2 0 75 kDa gRNA-AP - Ctrl A B C D E F - Ctrl A B C D E F Empty vector FOXJ1 + dCas9 + gRNA Cons_2 -57 -45 A B ADGBC D E F -1 -590 -119 -65 -37 -51 -36 Figure 8. Conserved nucleotides within −71 ± 30 bp upstream of ADGB TSS are crucial for FOXJ1 binding. A, UCSC Genome Browser output (hg19) of evolutionary conserved nucleotides within −41 bp to −100 bp upstream of the ADGB TSS based on a subset of vertebrate sequences extracted from the 100-MULTIZ whole-genome multiple sequence alignment algorithm. Basewise conservation scored by PhyloP indicates conserved and variable nucleotides in blue and red bars, respectively. Highly conserved nucleotides, also supported by the PhastCons track within this region of the ADGB promoter are boxed in red. Mutation strategy of conserved nucleotides is illustrated at the bottom. B, substitution-based mutation at −96 to −92 bp (Mut1), −57 to −52 bp (Mut3), and −51 to −46 bp (Mut4) results in loss of FOXJ1-dependent increase in ADGB promoter activity. Whereas the mutation at −73 to −68 bp (Mut2) did not abolish the FOXJ1-mediated activation (n = 3 independent experiments). C, sgRNA- mediated docking of dCas9 onto the Cons2 region in the ADGB promoter results in reduced FOXJ1-dependent activation of ADGB expression while 16 J. Biol. Chem. (2021) 296 100291 Relative mRNA level ADGB is involved in ciliogenesis and a target of FOXJ1 with the ones published before, although we did not prefilter SYBR FAST qPCR reagent kit (Sigma-Aldrich) in a CFX96 for contaminating cell types such as macrophages to ensure C1000 Thermal Cycler (BioRad). Transcript levels were impartial analysis. calculated as described before (57) and displayed as relative expression levels. Animals Cattle tissue was obtained from young females (heifers) Expression plasmid constructs immediately after slaughter in a regional commercial slaugh- pENTR233-FOXJ1 entry clone obtained from the DNASU terhouse. One uterine horn was opened lengthwise and the plasmid repository (58) and pcDNA3.1/nV5-DEST mamma- endometrial portion of the uterus was dissected. Tissue was lian expression vector were Gateway-recombined according to either flash-frozen on dry ice (RNA extraction) or fixed in 4% the manufacturer’s instruction (Invitrogen) to generate a para-formaldehyde (immunofluorescence). pcDNA3.1-nV5-FOXJ1 expression vector. The pcDNA3.1- HA-RFX2 plasmid was a generous gift from Prof. Zijie Sun RNA extraction and reverse-transcription quantitative PCR (Stanford). N-terminally FLAG-tagged FOXJ1 and RFX2 were (RT-qPCR) cloned by amplifying FOXJ1 and RFX2 genes using primer RNA extraction of cattle tissues was performed from snap- pairs with SalI/KpnI and BglII/SalI overhangs, respectively as frozen samples with the RNeasy Plus Universal Mini kit listed in Table S2. Amplicons were subsequently digested with (Qiagen) according to the instructions of the manufacturer. their respective restriction enzymes as designed on the primers Approximately 50 mg of tissues was grinded and homogenized and ligated into linearized pFLAG-CMV-6a vector to generate with a MiniLys (Precellys) system using mixed ceramic beads pFLAG-FOXJ1 and pFLAG-RFX2. (Precellys Lysing CKMix). Difficult tissues such as endometria were pregrinded manually on dry ice with a cool scalpel. RNA Luciferase constructs was eluted in nuclease-free water. RNA quality was assessed ADGB promoter elements spanning from −33 bp with a Bioanalyzer (Agilent), and only samples with RIN >7 to −464 bp, −1064 bp, and −2014 bp upstream of the ADGB were used for further analysis. RNA was quantified via Qubit transcriptional start site (TSS) were cloned into a pGL3- measurement using the Broad Range RNA Assay Kit (Thermo Basic (Promega) vector at −67 bp upstream of the firefly Fisher) and was stored at −80 C until further use. To confirm luciferase reporter gene. Promoter elements were amplified the bioinformatical findings, we performed reverse- from a pool of genomic DNA extracted from three human transcription quantitative PCR (RT-qPCR) on tissues from cell lines, MCF-7, HEK293T, and Hep3B cells, by PCR using the female reproductive tract of cattle. 1000 ng of total cattle the primer pairs described in Table S2. PCR amplicons were RNA per sample was used for reverse transcription with the digested with their respective restriction enzymes and ligated SuperScript III enzyme (10,000 units per assay; Invitrogen) into linearized pGL3-Basic (Promega) vector digested with using an Oligo-dT primer. In the absence of validated refer- KpnI and NheI. Evolutionarily conserved nucleotides ence genes, the amount of mRNA expression was normalized within −71 ± 30 bp upstream of the ADGB TSS were on the adjusted total amount of carefully quantified total RNA. mutated into a tandem of 5X or 6X A or T (or both) using To additionally control for differences in cDNA synthesis, oligonucleotide-based cloning of mutant promoter frag- 100 ng of Drosophila total RNA was added to the reaction as a ments into pGL3-Basic (Promega) vector at −67 bp upstream spike-in control. RT-qPCR was carried out using GoTaq qPCR of a firefly luciferase gene. The wild-type sequences were Master Mix (Promega) on the ABI Prism 7500 Fast Detection mutated as indicated in Table S2. Prior to cloning, phos- System (SDS, Applied Biosystems) and interpreted using 7500 phorylated oligo duplexes with designed 5’-KpnI overhang Software Version 2.3. Quantification of ADGB-cDNA mole- and 3’-Nhel overhang were generated by incubating 0.5 μM cules was done in absolute numbers applying a calibration of synthesized complementary oligo strands in T4 DNA standard curve with known amounts of target PCR product, ligase buffer (ThermoScientific) (40 mM Tris-HCl, 10 mM previously cloned into the pGEM T-easy vector system MgCl2, 10 mM DTT, 500 μM ATP) with T4 polynucleotide (Promega). Foxj1 and Dnah5 expressions were measured as kinase (ThermoScientific) at 37 C for 1 h, followed by relative values only and the sample with the highest expression heating to 95 C for 5 min and slow cooling at the rate of −5 was set to 100%. Copies of the Drosophila Globin 1 (Glob1) C min−1 to 10 C. Oligo duplexes were subsequently ligated cDNA of the internal control were measured in parallel to into KpnI and NheI digested pGL3-Basic vector backbone. identify samples with substandard reverse transcription. All Potential ADGB intronic and 3’ enhancer elements from the primers used are listed in Table S2. For HEK293 cells, total GeneHancer database (27) were cloned in a pGL3-SV40 RNA was extracted as previously described (56). Total RNA vector (Promega), at −27 bp upstream of the SV40 promoter. (2 μg) was reverse transcribed (RT) using the Prime Script RT These putative enhancer elements were amplified from reagent kit (Takara Bio USA) and cDNA levels were estimated genomic DNA, by PCR using the primer pairs as described by qPCR using the primers listed in Table S2 and a KAPA in Table S2. PCR amplicons were digested with their more remote control sgRNAs display no effect in HEK293 and MCF-7 cells (n = 3 independent experiments). gRNA positions are schematically represented. Immunoblotting analysis using a Cas9 antibody controlled for dCas9 overexpression. Data represent mean ± S.E.M (error bars); *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. J. Biol. Chem. (2021) 296 100291 17 ADGB is involved in ciliogenesis and a target of FOXJ1 respective restriction enzymes and ligated into linearized according to manufacturer’s instruction. Prior to trans- pGL3-SV40 (Promega) vector digested with KpnI and NheI fection, cells were seeded on 24-well plates at a density of or into the linearized ADGB promoter containing pGL3B- 5.2 × 104 cells (for MCF-7 and HEK293T) or 2.6 × 104 cells AP464 vector, at 269 bp downstream of the firefly lucif- (for HeLa) per cm2 of the dish surface area. For calcium erase gene, digested with BamHI. phosphate precipitation method of DNA delivery, overnight medium was aspirated and replaced with DMEM/FBS/PS Mammalian cell culture and DNA transfection containing 25 uM chloroquine. In each well, 10% (v/v) of MCF-7, HEK293T, and HeLa cells were cultured and transfection mixture was introduced to the cells, consisting maintained in DMEM (Thermo Fisher) medium supple- of 50% (v/v) 1.2 to 1.5 μg plasmid DNA/250 mM CaCl2 and mented with 10% fetal bovine serum (Chemie Brunschwig) 50% (v/v) HBS buffer (50 mM HEPES, 280 mM NaCl, and 100 μg/ml Pen/Strep Glutamine (Thermo Fisher), 10 mM KCl, 1.5 mM Na2HPO4, 12 mM glucose). Trans- DMEM/FBS/PS for simplification, and incubated at 37 C in fected cells were incubated at 37 C/5% CO2 for 24 h before a humidified incubator with 5% CO2. mCCDcl1 cells (59) the medium was replaced with fresh DMEM/FBS/PS and were maintained at 37 C and 5% CO2 in DMEM/F12 continued to incubate at 37 C/5% CO2 for another 24 h. (Thermo Fisher) supplemented with 5 μg/ml insulin Cells were harvested 48 h posttransfection. mCCDcl1 cells (Sigma), 50 nM dexamethasone (Sigma), 60 nM selenium were transfected using Polyplus JetPrime according to (Sigma), 5 μg/ml transferrin (Sigma), 1 nM triiodothyronine manufacturer’s instructions. For immunocytochemistry ex- (Sigma), 5 ng/ml mouse EGF (Sigma), 100 μg/ml Pen/Strep periments, HeLa and mCCD cells were seeded in 6-well Glutamine (Thermo Fisher), and 2% decomplemented fetal plates on glass coverslips at 0.2 x 106 cells per well. After bovine serum (Thermo Fisher). MCF-7, HEK293T, and 24 h, the cells were transfected and either left untreated or HeLa cells were transfected using Roti-Fect (ROTH), serum-starved for another 24 h. A HEK293T MCF-7 B Empty plasmid + - - 1000 * 1000 * FLAG-FOXJ1 - + - 800 FLAG-RFX2 - - + 600 800 250 kDa 400 600 200 150 kDa 20 *** 400 100 kDa 15 75 kDa RFX2 6 10 4 ** FOXJ1 2 1 2 50 kDa 0 0 Empty plasmid + - - Empty plasmid + - - FLAG-FOXJ1 - + - FLAG-FOXJ1 - + - 50 kDa TUB FLAG-RFX2 - - + FLAG-RFX2 - - + C MCF-7 HEK293T HeLa + RFX2 + FOXJ1 + FOXJ1 and RFX2 + RFX2 #### + FOXJ1 ### #### * ** + FOXJ1 and RFX2 *#### #### #### 0 4 8 12 0 5 10 15 20 25 0.0 7.5 15.0 22.5 30.0 RLU (firefly/Renilla) RLU (firefly/Renilla) RLU (firefly/Renilla) pGL3B pGL3B-AP-3’-AE1 Figure 9. FOXJ1 and RFX2 transcription factors induce endogenous ADGB transcription and cooperate to regulate ADGB expression in the presence of the 3’-AE1 enhancer element. A, HEK293T and MCF-7 cells were transiently transfected with a FLAG-tagged FOXJ1 or RFX2 expressing plasmid and ADGB mRNA levels were measured with RT-qPCR. ADGB expression levels were normalized to β-actin and displayed as relative values to cDNA of HEK293T or MCF-7 cells transfected with equal amount of empty vector. B, FLAG-tagged FOXJ1 and RFX2 plasmid-derived expression levels were controlled by immunoblotting using an anti-FLAG antibody. Tubulin (TUB) was used as loading control. C, 3’-AE1 ADGB enhancer-dependent ADGB promoter-driven reporter gene assays. Following overexpression of FOXJ1, FOXJ1-mediated activation of the ADGB promoter activity is observed as ex- pected from previous experiments. Co-overexpression of FOXJ1 and RFX2 shows synergistic activation of the ADGB promoter construct in the presence of 3’-AE1 enhancer element (n = 3 independent experiments). Similar experiments were performed in HEK293T and HeLa (n = 3 independent experiments). Data represent mean ± S.E.M (error bars); *p < 0.05; **p < 0.01. ###p < 0.001; ####p < 0.0001. 18 J. Biol. Chem. (2021) 296 100291 Relative mRNA level Relative mRNA level ADGB is involved in ciliogenesis and a target of FOXJ1 Luciferase reporter gene assays dCas9-mediated interference of FOXJ1 binding Fifty nanograms of promoter- and/or enhancer-containing Nuclease-null-Cas9 was cloned by amplifying the dCas9 firefly luciferase plasmid was cotransfected along with 1 ng gene using primer pairs with KpnI and NotI overhangs listed in of pRL-SV40 Renilla luciferase to control for differences in Table S2. The amplicon was subsequently digested with re- transfection efficiency and extract preparation. For the study striction enzymes as designed on the primers and ligated into a of FOXJ1 and RFX2 activity on ADGB promoters, 300 ng of linearized pcDNA3 vector to generate pcDNA3-dCas9. pcDNA3.1-nV5-FOXJ1 or pcDNA3.1-HA-RFX2 was Candidate gRNAs targeting the Cons2 region within the ADGB cotransfected with promoter plasmid, the total amounts of promoter and control gRNAs targeting more distal regions plasmid DNA used were normalized with pcDNA3.1-nV5- (Table S2) were cloned into and expressed from a pSPgRNA HisA empty vector. Luciferase activities were determined plasmid (Addgene #47108) as described above. For dCas9- using the Dual Luciferase Reporter Assay System (Promega) mediated interference of FOXJ1 binding, 800 ng of pFLAG- as described before (60). Reporter activities were expressed FOXJ1 was cotransfected with 1000 ng of pcDNA3-dCas9 as relative firefly/Renilla luciferase activities. All reporter and 600 ng of gRNA into HEK293T of MCF-7 cells with cal- gene assays were performed at least three times cium phosphate precipitation and Roti-Fect (ROTH), respec- independently. tively. Cells were incubated for 24 h before the transfection medium was replaced with fresh DMEM/FBS/PS, and allowed dCas9-VPR-mediated activation of endogenous ADGB to grow for another 24 h. promoter Chromatin immunoprecipitation Nuclease-null-Cas9 with tandem fusion of VP64-p65-Rta tripartite activator (dCas9-VPR, Addgene #63798) (24) was ChIP was carried out as described before with some modifi- delivered along with gRNAs as described before (61) to acti- cations (63, 64). Briefly, cells were cross-linked by adding 1% vate ADGB promoter activity. gRNAs candidates targeting (w/v) formaldehyde and incubated for 20 min at RT with gentle between −1 and −1700 bp upstream of ADGB TSS were cloned shaking. Cell fixation was interrupted by adding 110 mM glycine. into, and expressed from, pSPgRNA plasmid (Addgene Cells were scraped off and resuspended in lysis buffer following #47108), which was a generous gift from Prof. Charles Gers- the iDeal ChIP-qPCR kit protocol (Diagenode, Liège, Belgium). bach (62) (Table S2). Prior to cloning, phosphorylated oligo To obtain genomic DNA fragments between 500 and 100 bp, duplexes were generated by incubating 0.5 μM of synthesized cell lysates were sonicated for four rounds of ten cycles (30 s complementary oligo strands (5’–3’) in T4 DNA ligase buffer ON/30 s OFF) using the Bioruptor Pico (Diagenode) at high (ThermoScienti c) (40 mM Tris-HCl, 10 mM MgCl , 10 mM power setting. For immunoprecipitations, the following anti-fi 2 DTT, 500 μM ATP) with T4 polynucleotide kinase (Ther- bodies were used: 1 μg of rabbit polyclonal anti-IgG (C15410206, moScienti c) at 37 C for 1 h, followed by heating to 95 C for Diagenode) as negative control IP; 1 μg of mouse monoclonalfi 5 min and slow cooling at the rate of −5 C min−1 to 10 C. anti-FOXJ1 (14-9965-82, Thermo Fisher Scientific) and 3.8 μg of Oligo duplexes with sticky ends complementary to BbsI- mouse monoclonal anti-FLAG (F1804, Sigma Aldrich). digested pSPgRNA vector were then ligated into the vector. Chromatin–antibody complexes were immunoprecipitated by dCas9-VPR and gRNAs were delivered to HEK293T cells with DiaMag Protein A-coated magnetic beads (Diagenode). DNA Roti-Fect (ROTH) according to the manufacturer’s instruc- isolation and de-cross-linking was carried out as described by the tion. Cells were transfected in 24-well plates, seeded with 5.2 x iDeal ChIP-qPCR kit protocol (Diagenode). Coprecipitated DNA 104 cells per cm2 surface area, 24 h prior to transfection. In was quantified by real-time qPCR using the primers listed in each well, 375 ng of dCas9-VPR was delivered together with Table S2. 125 ng of gRNA(s) in antibiotic-free DMEM medium. For the transfection of gRNAs in combinations, the total amount of Immunocytochemistry and cilia counting gRNAs was equally distributed to a total of 125 ng. Cells were The cells were washed with phosphate-buffered saline incubated for 24 h before the transfection medium was (PBS) and fixed for 10 min in 4% para-formaldehyde (PFA), replaced with fresh DMEM/FBS/PS and allowed to grow for followed by 3 × 5 min PBS washes, permeabilization in PBS/ another 24 h before RNA extraction for analysis. For dCas9- Triton X-100 0.2% for 10 min, 3 × 5 min PBS washes, and VPR activation of endogenous ADGB and subsequent immu- blocking in PBS/BSA 1% for 1 h. The cells were then incu- noblot analysis, HEK293T cells were seeded on 6-well plates at bated overnight with mouse anti-acetylated tubulin (Santa a density of 5.2 × 104 cells per cm2 of the dish surface area and Cruz) diluted 1/500 in PBS/BSA 0.1%. On the following day, were allowed to grow at 37 C/5% CO2 for 24 h. For trans- cells were washed 3 x 5 min with PBS and incubated for 1 h fection, 2250 ng of dCas9-VPR and 750 ng of gRNAs mix with secondary Alexa Fluor goat anti-mouse IgG (Invi- (equal amount of gRNA AP-1 and gRNA AP-2) were delivered trogen) diluted 1/300 in PBS. The slides were mounted with to HEK293T cells with Roti-Fect (ROTH), and cells were Fluoromount mounting medium containing DAPI (Southern incubated for 24 h before the transfection medium was Biotech) and visualized on a Nikon Eclipse fluorescent mi- replaced with fresh DMEM/FBS/PS and allowed to grow for croscope (Nikon Corporation). Cilia counting was per- another 48 h. formed using ImageJ software. J. Biol. Chem. (2021) 296 100291 19 ADGB is involved in ciliogenesis and a target of FOXJ1 A Empty vector ADGB Starved HeLa 45 ** empty vector + - - 36 * ADGB - + - starved - - + 27 250 kDa 18 ADGB150 kDa 9 50 kDa TUB 0 tor GB d ec D rv e v A ty s ta p em B Empty vector ADGB Starved mCCD **** 60 **** empty vector + - - ADGB - + - 40 starved - - + 250 kDa 150 kDa ADGB 20 50 kDa TUB 0 cto r GB rve d ve ADy st a t em p Figure 10. ADGB overexpression promotes ciliogenesis in two independent cell lines. A, representative pictures of acetylated α-tubulin (red fluores- cence) and DAPI (blue fluorescence) in HeLa cells under basal conditions transfected with empty vector, following transfection with 2 μg ADGB or following 24 h serum starvation (starved), and corresponding cilia quantifications (expressed as percentage of ciliated cells) (n = 3–4 independent experiments, 3–5 pictures were counted for each condition and per experiment). Scale bar represents 100 μm. Overexpression of ADGB was verified by immunoblotting. Tubulin (TUB) was used as loading control. B, Similar representative immunofluorescence pictures of acetylated α-tubulin, corresponding cilia quantifi- cations and immunoblotting following transfection with 2 μg ADGB in mCCDcl1 cells. Scale bar represents 100 μm. Data represent mean ± S.E.M (error bars); *p < 0.05; ***p < 0.001; ****p < 0.0001. Immunohistochemistry of bovine tissue desiccation, the tissue was enveloped in parafilm and stored  After 20 h of fixation, the tissue was washed twice in PBS at −80 C until further use. The samples were embedded in and cryoprotected in 20% saccharose solution. To avoid Neg-50 Frozen Section Medium (Thermo Scientific) and 20 J. Biol. Chem. (2021) 296 100291 % ciliated cells % ciliated cells ADGB is involved in ciliogenesis and a target of FOXJ1 sectioned in a cryostat at −20 C. Before immunostaining, heat-induced epitope retrieval was performed for 30 min at pH Acknowledgments—We thank Darko Maric (University of Fribourg) 6. Slides were permeabilized (PBS with 0.1% Triton X), blocked for helpful advice and discussions. CO and TH would like to thank in blocking buffer (10% Horse serum in PBS with 0.1% Triton Klaus-Dieter Fuchs and Dr Heinrich Dahmen (Fuchs GmbH EU X), and probed with primary rabbit anti-ADGB antibody (1/ Zerlege-und Schlachtbetrieb, Prüm, Germany) for generous 100) (Sigma-Aldrich, HPA036340) in blocking buffer over- donation of tissue and assistance in sample acquisition and Timo night. Incubation with secondary goat anti-rabbit CF 488 Fredi Kopp and Julian Thomas Mohr (JGU Mainz) for assistance antibody (1/250) (Sigma-Aldrich, SAB4600036) or goat anti- in laboratory work. rabbit Alkaline Phosphatase (1/500) (Sigma-Aldrich, A3687) Author contributions—T. W. K., C. O., I. M. C. O., A. K., D. A., S. Y., was performed for 1 h. Fluorescent samples were counter- M. S. A., M. C., R. M., and D. H. data curation, formal analysis, and stained with DAPI (Roche) and embedded in RotiMount investigation; T. W. K., C. O., I. M. C. O., A. K., J. S., T. H., and D. H. Fluorcare antifading solution (Roth). For colorimetric staining, methodology; T. W. K., C. O., and D. H. visualization; T. W. K., C. slides were incubated in NBT-BCIP (Roche) under exclusion O., T. H., and D. H. writing-original draft; T. H. and D. H. funding of oxygen. Image acquisition was done either with a Leica SP5 acquisition, conceptualization, and supervision. or a fluorescence microscope BX61 (Olympus). Funding and additional information—This work was supported by Protein extraction and immunoblotting the Swiss National Science Foundation to D. H. (grant 31003A_173000) and the German Research Foundation to D. H. Protein extraction and immunoblotting were performed as (HO 5837/1-1), T. H. (HA 2103/9-1) and J. S. (Projektnummer before with some modifications (65). Cells were lysed in NP-40 387509280; SFB 1350). lysis buffer (10 mM Tris, pH 8.0, 1 mM EDTA, 400 mM NaCl, 0.1% NP-40, 2 μg ml−1 leupeptin, 2 μg ml−1 pepstatin, 2 μg ml−1 Conflict of interest—The authors declare no conflicts of interest in aprotinin, 1 mM PMSF) or in Triton X-100 buffer (50 mM Tris- regard to this article. HCl pH 7.4, 150 mM NaCl, 1% Triton X-100, 2 μg/ml aprotinin, Abbreviations—The abbreviations used are: ADGB, androglobin; 4 μg/ml leupeptin, 2 μg/ml pepstatin and 1 mM PMSF), and ALI, air–liquid interface; ANOVA, analysis of variance; ChIP, centrifuged for 10 min at 14,000 rpm at 4 C. For the detection chromatin immunoprecipitation; CYGB, cytoglobin; HB, hemoglo- of CRISPRa-induced ADGB, total cell lysate was first immuno- bin; MB, myoglobin; mTEC, mouse tracheal epithelial cells; NGB, precipitated with 0.2 μg of ADGB antibody (Sigma-Aldrich) and neuroglobin; PFA, para-formaldehyde; RT, reverse transcription; 25 μl bed volume of Protein-G-Sepharose beads (Sigma- sgRNA, single guide RNA; TSS, transcriptional start site. Aldrich). Protein quantification was performed by Bradford assay. In total, 50 to 100 μg proteins was separated on 10% SDS acrylamide gels and electrically transferred onto nitrocellulose References membranes (Amersham Protran Western blotting membranes). 1. Burmester, T., Weich, B., Reinhardt, S., and Hankeln, T. (2000) Following transfer membranes were blocked for 1 h with 5% A vertebrate globin expressed in the brain. Nature 407, 520–523 dried milk/TBS-tween at room temperature and incubated 2. Burmester, T., Ebner, B., Weich, B., and Hankeln, T. (2002) Cytoglobin: A overnight in 1% dried milk/TBS-tween with primary rabbit anti- novel globin type ubiquitously expressed in vertebrate tissues. Mol. Biol. Evol. 19, 416–421 ADGB (1/500) (Sigma-Aldrich, HPA036340), mouse anti-FOXJ1 3. Burmester, T., and Hankeln, T. (2014) Function and evolution of verte- (1/300) (Invitrogen, 14-9965-82), mouse anti-HA (1/100) (Santa brate globins. Acta Physiol. (Oxf.) 211, 501–514 Cruz, sc-57592), mouse anti FLAG (1/1000) (Sigma-Aldrich, 4. Keppner, A., Maric, D., Correia, M., Koay, T. W., Orlando, I. M. C., F3165), or rabbit anti-CAS9 (1/2500) (ABclonal, A14997) anti- Vinogradov, S. N., and Hoogewijs, D. (2020) Lessons from the post- bodies. Membranes were then incubated for 1 h with HRP- genomic era: Globin diversity beyond oxygen binding and transport. Redox Biol. 37, 101687 conjugated goat anti-rabbit secondary antibody (GE Health- 5. Hoogewijs, D., Ebner, B., Germani, F., Hoffmann, F. G., Fabrizius, A., care), and the signal was revealed using ECL Prime (Amersham) Moens, L., Burmester, T., Dewilde, S., Storz, J. F., Vinogradov, S. N., and on a C-DiGit western blot scanner (LI-COR Biosciences) or Hankeln, T. (2012) Androglobin: A chimeric globin in metazoans that is fluorescent signals were visualized via a Li-COR Odyssey preferentially expressed in mammalian testes. Mol. Biol. Evol. 29, 1105– infrared imaging system. 1114 6. Huang, B., Lu, Y. S., Li, X., Zhu, Z. C., Li, K., Liu, J. W., Zheng, J., and Hu, Z. L. (2014) Androglobin knockdown inhibits growth of glioma cell lines. Data analysis Int. J. Clin. Exp. Pathol. 7, 2179–2184 Results are shown as mean values ± SEM of at least three 7. Uhlen, M., Fagerberg, L., Hallstrom, B. M., Lindskog, C., Oksvold, P., independent experiments. Statistical analysis was performed Mardinoglu, A., Sivertsson, A., Kampf, C., Sjostedt, E., Asplund, A., Olsson, I., Edlund, K., Lundberg, E., Navani, S., Szigyarto, C. A., et al. applying two-tailed or paired Student’s t-test or one-way (2015) Proteomics. Tissue-based map of the human proteome. Science analysis of variance (ANOVA) for multiple comparisons, us- 347, 1260419 ing GraphPad Prism Version 7.0 (GraphPad Software). 8. Altmae, S., Koel, M., Vosa, U., Adler, P., Suhorutsenko, M., Laisk-Podar, T., Kukushkina, V., Saare, M., Velthut-Meikas, A., Krjutskov, K., Aghajanova, L., Data availability Lalitkumar, P. G., Gemzell-Danielsson, K., Giudice, L., Simon, C., et al. (2017) Meta-signature of human endometrial receptivity: A meta-analysis All data are contained within the article. and validation study of transcriptomic biomarkers. Sci. Rep. 7, 10077 J. Biol. Chem. (2021) 296 100291 21 ADGB is involved in ciliogenesis and a target of FOXJ1 9. Schuster, H., Peper, J. K., Bosmuller, H. C., Rohle, K., Backert, L., Bilich, 27. Fishilevich, S., Nudel, R., Rappaport, N., Hadar, R., Plaschkes, I., Iny T., Ney, B., Loffler, M. W., Kowalewski, D. J., Trautwein, N., Rabsteyn, A., Stein, T., Rosen, N., Kohn, A., Twik, M., Safran, M., Lancet, D., and Engler, T., Braun, S., Haen, S. P., Walz, J. S., et al. (2017) The immu- Cohen, D. (2017) GeneHancer: Genome-wide integration of enhancers nopeptidomic landscape of ovarian carcinomas. Proc. Natl. Acad. Sci. U. and target genes in GeneCards. Database (Oxford) 2017 S. A. 114, E9942–E9951 28. Quigley, I. K., and Kintner, C. (2017) Rfx2 stabilizes Foxj1 binding at 10. Yu, X., Ng, C. P., Habacher, H., and Roy, S. (2008) Foxj1 transcription chromatin loops to enable multiciliated cell gene expression. PLoS Genet. factors are master regulators of the motile ciliogenic program. Nat. Genet. 13, e1006538 40, 1445–1453 29. Kistler, W. S., Baas, D., Lemeille, S., Paschaki, M., Seguin-Estevez, Q., 11. Whitfield, M., Thomas, L., Bequignon, E., Schmitt, A., Stouvenel, L., Barras, E., Ma, W., Duteyrat, J. L., Morle, L., Durand, B., and Reith, W. Montantin, G., Tissier, S., Duquesnoy, P., Copin, B., Chantot, S., Dastot, (2015) RFX2 is a major transcriptional regulator of spermiogenesis. PLoS F., Faucon, C., Barbotin, A. L., Loyens, A., Siffroi, J. P., et al. (2019) Genet. 11, e1005368 Mutations in DNAH17, encoding a sperm-specific axonemal outer 30. Bracke, A., Hoogewijs, D., and Dewilde, S. (2018) Exploring three dynein arm heavy chain, cause isolated male infertility due to astheno- different expression systems for recombinant expression of globins: zoospermia. Am. J. Hum. Genet. 105, 198–212 Escherichia coli, Pichia pastoris and Spodoptera frugiperda. Anal. Bio- 12. Comer, M. T., Leese, H. J., and Southgate, J. (1998) Induction of a chem. 543, 62–70 differentiated ciliated cell phenotype in primary cultures of Fallopian tube 31. van Dam, T. J. P., Kennedy, J., van der Lee, R., de Vrieze, E., Wunderlich, epithelium. Hum. Reprod. 13, 3114–3120 K. A., Rix, S., Dougherty, G. W., Lambacher, N. J., Li, C., Jensen, V. L., 13. Haider, S., Gamperl, M., Burkard, T. R., Kunihs, V., Kaindl, U., Junttila, S., Leroux, M. R., Hjeij, R., Horn, N., Texier, Y., Wissinger, Y., et al. (2019) Fiala, C., Schmidt, K., Mendjan, S., Knofler, M., and Latos, P. A. (2019) CiliaCarta: An integrated and validated compendium of ciliary genes. Estrogen signaling drives ciliogenesis in human endometrial organoids. PLoS One 14, e0216705 Endocrinology 160, 2282–2297 32. Sigg, M. A., Menchen, T., Lee, C., Johnson, J., Jungnickel, M. K., Choksi, 14. Barton, B. E., Herrera, G. G., Anamthathmakula, P., Rock, J. K., Willie, A., S. P., Garcia, G., 3rd, Busengdal, H., Dougherty, G. W., Pennekamp, P., Harris, E. A., Takemaru, K. I., and Winuthayanon, W. (2020) Roles of Werner, C., Rentzsch, F., Florman, H. M., Krogan, N., Wallingford, J. B., steroid hormones in oviductal function. Reproduction 159, R125–R137 et al. (2017) Evolutionary proteomics uncovers ancient associations of 15. Ferenczy, A., Richart, R. M., Agate, F. J., Jr., Purkerson, M. L., and cilia with signaling pathways. Dev. Cell 43, 744–762.e711 Dempsey, E. W. (1972) Scanning electron microscopy of the human 33. Burmester, T., Storf, J., Hasenjager, A., Klawitter, S., and Hankeln, T. fallopian tube. Science 175, 783–784 (2006) The hemoglobin genes of Drosophila. FEBS J. 273, 468–480 16. Montoro, D. T., Haber, A. L., Biton, M., Vinarsky, V., Lin, B., Birket, S. E., 34. Hoogewijs, D., De Henau, S., Dewilde, S., Moens, L., Couvreur, M., Yuan, F., Chen, S., Leung, H. M., Villoria, J., Rogel, N., Burgin, G., Borgonie, G., Vinogradov, S. N., Roy, S. W., and Vanfleteren, J. R. (2008) Tsankov, A. M., Waghray, A., Slyper, M., et al. (2018) A revised airway The Caenorhabditis globin gene family reveals extensive nematode- epithelial hierarchy includes CFTR-expressing ionocytes. Nature 560, specific radiation and diversification. BMC Evol. Biol. 8, 279 319–324 35. Hoogewijs, D., Geuens, E., Dewilde, S., Moens, L., Vierstraete, A., 17. Lukassen, S., Bosch, E., Ekici, A. B., and Winterpacht, A. (2018) Char- Vinogradov, S., and Vanfleteren, J. (2004) Genome-wide analysis of the acterization of germ cell differentiation in the male mouse through single- globin gene family of C. elegans. IUBMB Life 56, 697–702 cell RNA sequencing. Sci. Rep. 8, 6521 36. Choksi, S. P., Lauter, G., Swoboda, P., and Roy, S. (2014) Switching on 18. Spassky, N., and Meunier, A. (2017) The development and functions of cilia: Transcriptional networks regulating ciliogenesis. Development 141, multiciliated epithelia. Nat. Rev. Mol. Cell Biol. 18, 423–436 1427–1441 19. Campbell, J. N., Macosko, E. Z., Fenselau, H., Pers, T. H., Lyubetskaya, A., 37. Nelson, G. A., Roberts, T. M., and Ward, S. (1982) Caenorhabditis elegans Tenen, D., Goldman, M., Verstegen, A. M., Resch, J. M., McCarroll, S. A., spermatozoan locomotion: Amoeboid movement with almost no actin. J. Rosen, E. D., Lowell, B. B., and Tsai, L. T. (2017) A molecular census of Cell Biol. 92, 121–131 arcuate hypothalamus and median eminence cell types. Nat. Neurosci. 20, 38. Acton, A. B. (1966) An unusual ciliumlike process. J. Cell Biol. 29, 366– 484–496 369 20. Mirzadeh, Z., Kusne, Y., Duran-Moreno, M., Cabrales, E., Gil-Perotin, S., 39. Gleixner, E., Herlyn, H., Zimmerling, S., Burmester, T., and Hankeln, T. Ortiz, C., Chen, B., Garcia-Verdugo, J. M., Sanai, N., and Alvarez-Buylla, (2012) Testes-specific hemoglobins in Drosophila evolved by a combi- A. (2017) Bi- and uniciliated ependymal cells define continuous floor- nation of sub- and neofunctionalization after gene duplication. BMC Evol. plate-derived tanycytic territories. Nat. Commun. 8, 13759 Biol. 12, 34 21. Gerdes, J. M., Liu, Y., Zaghloul, N. A., Leitch, C. C., Lawson, S. S., Kato, 40. Burmester, T., Wawrowski, A., Diepenbruck, I., Schrick, K., Seiwert, N., M., Beachy, P. A., Beales, P. L., DeMartino, G. N., Fisher, S., Badano, J. L., Ripp, F., Prothmann, A., and Hankeln, T. (2018) Divergent roles of the and Katsanis, N. (2007) Disruption of the basal body compromises pro- Drosophila melanogaster globins. J. Insect Physiol. 106, 224–231 teasomal function and perturbs intracellular Wnt response. Nat. Genet. 41. Jamieson, B. (1991) Ultrastructure and phylogeny of Crustacean sper- 39, 1350–1360 matozoa. Mem. Queensland Mus. 31, 109–142 22. Kowal, T. J., and Falk, M. M. (2015) Primary cilia found on HeLa and 42. Choksi, S. P., Babu, D., Lau, D., Yu, X., and Roy, S. (2014) Systematic other cancer cells. Cell Biol. Int. 39, 1341–1347 discovery of novel ciliary genes through functional genomics in the 23. Tang, Z., Lin, M. G., Stowe, T. R., Chen, S., Zhu, M., Stearns, T., Franco, zebrafish. Development 141, 3410–3419 B., and Zhong, Q. (2013) Autophagy promotes primary ciliogenesis by 43. Stauber, M., Weidemann, M., Dittrich-Breiholz, O., Lobschat, K., removing OFD1 from centriolar satellites. Nature 502, 254–257 Alten, L., Mai, M., Beckers, A., Kracht, M., and Gossler, A. (2017) 24. Chavez, A., Scheiman, J., Vora, S., Pruitt, B. W., Tuttle, M., E, P. R. I., Lin, Identification of FOXJ1 effectors during ciliogenesis in the foetal res- S., Kiani, S., Guzman, C. D., Wiegand, D. J., Ter-Ovanesyan, D., Braff, J. piratory epithelium and embryonic left-right organiser of the mouse. L., Davidsohn, N., Housden, B. E., Perrimon, N., et al. (2015) Highly Dev. Biol. 423, 170–188 efficient Cas9-mediated transcriptional programming. Nat. Methods 12, 44. Jacquet, B. V., Salinas-Mondragon, R., Liang, H., Therit, B., Buie, J. D., 326–328 Dykstra, M., Campbell, K., Ostrowski, L. E., Brody, S. L., and Ghashghaei, 25. Consortium, E. P. (2012) An integrated encyclopedia of DNA elements in H. T. (2009) FoxJ1-dependent gene expression is required for differenti- the human genome. Nature 489, 57–74 ation of radial glia into ependymal cells and a subset of astrocytes in the 26. Cheneby, J., Gheorghe, M., Artufel, M., Mathelier, A., and Ballester, B. postnatal brain. Development 136, 4021–4031 (2018) ReMap 2018: An updated atlas of regulatory regions from an 45. Campbell, E. P., Quigley, I. K., and Kintner, C. (2016) Foxn4 promotes integrative analysis of DNA-binding ChIP-seq experiments. Nucleic Acids gene expression required for the formation of multiple motile cilia. Res. 46, D267–D275 Development 143, 4654–4664 22 J. Biol. Chem. (2021) 296 100291 ADGB is involved in ciliogenesis and a target of FOXJ1 46. Mukherjee, I., Roy, S., and Chakrabarti, S. (2019) Identification of 58. Seiler, C. Y., Park, J. G., Sharma, A., Hunter, P., Surapaneni, P., Sedillo, C., important effector proteins in the FOXJ1 transcriptional network asso- Field, J., Algar, R., Price, A., Steel, J., Throop, A., Fiacco, M., and LaBaer, J. ciated with ciliogenesis and ciliary function. Front. Genet. 10, 23 (2014) DNASU plasmid and PSI:biology-materials repositories: Resources 47. Blatt, E. N., Yan, X. H., Wuerffel, M. K., Hamilos, D. L., and Brody, S. L. to accelerate biological research. Nucleic Acids Res. 42, D1253–D1260 (1999) Forkhead transcription factor HFH-4 expression is temporally 59. Gaeggeler, H. P., Gonzalez-Rodriguez, E., Jaeger, N. F., Loffing-Cueni, D., related to ciliogenesis. Am. J. Respir. Cell Mol. Biol. 21, 168–176 Norregaard, R., Loffing, J., Horisberger, J. D., and Rossier, B. C. (2005) 48. Weidemann, M., Schuster-Gossler, K., Stauber, M., Wrede, C., Heger- Mineralocorticoid versus glucocorticoid receptor occupancy mediating mann, J., Ott, T., Boldt, K., Beyer, T., Serth, K., Kremmer, E., Blum, M., aldosterone-stimulated sodium transport in a novel renal cell line. J. Am. Ueffing, M., and Gossler, A. (2016) CFAP157 is a murine downstream Soc. Nephrol. 16, 878–891 effector of FOXJ1 that is specifically required for flagellum morphogen- 60. Storti, F., Santambrogio, S., Crowther, L. M., Otto, T., Abreu-Rodriguez, esis and sperm motility. Development 143, 4736–4748 I., Kaufmann, M., Hu, C. J., Dame, C., Fandrey, J., Wenger, R. H., and 49. Lewis, M., and Stracker, T. H. (2020) Transcriptional regulation of Hoogewijs, D. (2014) A novel distal upstream hypoxia response element multiciliated cell differentiation. Semin. Cell Dev. Biol. https://doi.org/10. regulating oxygen-dependent erythropoietin gene expression. Haemato- 1016/j.semcdb.2020.04.007 logica 99, e45–e48 50. Nemajerova, A., Kramer, D., Siller, S. S., Herr, C., Shomroni, O., Pena, T., 61. Lauer, V., Grampp, S., Platt, J., Lafleur, V., Lombardi, O., Choudhry, H., Gallinas Suazo, C., Glaser, K., Wildung, M., Steffen, H., Sriraman, A., Kranz, F., Hartmann, A., Wullich, B., Yamamoto, A., Coleman, M. L., Oberle, F., Wienken, M., Hennion, M., Vidal, R., et al. (2016) TAp73 is a Ratcliffe, P. J., Mole, D. R., and Schodel, J. (2020) Hypoxia drives glucose central transcriptional regulator of airway multiciliogenesis. Genes Dev. transporter 3 expression through hypoxia-inducible transcription factor 30, 1300–1312 (HIF)-mediated induction of the long noncoding RNA NICI. J. Biol. 51. Kim, D., Paggi, J. M., Park, C., Bennett, C., and Salzberg, S. L. (2019) Chem. 295, 4065–4078 Graph-based genome alignment and genotyping with HISAT2 and 62. Perez-Pinera, P., Kocak, D. D., Vockley, C. M., Adler, A. F., Kabadi, A. M., HISAT-genotype. Nat. Biotechnol. 37, 907–915 Polstein, L. R., Thakore, P. I., Glass, K. A., Ousterout, D. G., Leong, K. W., 52. Pertea, M., Pertea, G. M., Antonescu, C. M., Chang, T. C., Mendell, J. T., Guilak, F., Crawford, G. E., Reddy, T. E., and Gersbach, C. A. (2013) and Salzberg, S. L. (2015) StringTie enables improved reconstruction of a RNA-guided gene activation by CRISPR-Cas9-based transcription factors. transcriptome from RNA-seq reads. Nat. Biotechnol. 33, 290–295 Nat. Methods 10, 973–976 53. Love, M. I., Huber, W., and Anders, S. (2014) Moderated estimation of 63. Schörg, A., Santambrogio, S., Platt, J. L., Schödel, J., Lindenmeyer, M. T., fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. Cohen, C. D., Schrödter, K., Mole, D. R., Wenger, R. H., and Hoogewijs, 15, 550 D. (2015) Destruction of a distal hypoxia response element abolishes 54. Liao, Y., Wang, J., Jaehnig, E. J., Shi, Z., and Zhang, B. (2019) WebGestalt trans-activation of the PAG1 gene mediated by HIF-independent chro- 2019: Gene set analysis toolkit with revamped UIs and APIs. Nucleic matin looping. Nucleic Acids Res. 43, 5810–5823 Acids Res. 47, W199–W205 64. Orlando, I. M. C., Lafleur, V. N., Storti, F., Spielmann, P., Crowther, L., 55. Stuart, T., Butler, A., Hoffman, P., Hafemeister, C., Papalexi, E., Mauck, Santambrogio, S., Schodel, J., Hoogewijs, D., Mole, D. R., and Wenger, R. W. M., 3rd, Hao, Y., Stoeckius, M., Smibert, P., and Satija, R. (2019) H. (2020) Distal and proximal hypoxia response elements cooperate to Comprehensive integration of single-cell data. Cell 177, 1888–1902.e1821 regulate organ-specific erythropoietin gene expression. Haematologica 56. Hoogewijs, D., Vogler, M., Zwenger, E., Krull, S., and Zieseniss, A. (2016) 105, 2774–2784 Oxygen-dependent regulation of aquaporin-3 expression. Hypoxia 65. Randi, E. B., Vervaet, B., Tsachaki, M., Porto, E., Vermeylen, S., (Auckl) 4, 91–97 Lindenmeyer, M. T., Thuy, L. T. T., Cohen, C. D., Devuyst, O., 57. Hoogewijs, D., Houthoofd, K., Matthijssens, F., Vandesompele, J., and Kistler, A. D., Szabo, C., Kawada, N., Hankeln, T., Odermatt, A., Vanfleteren, J. R. (2008) Selection and validation of a set of reliable Dewilde, S., et al. (2020) The antioxidative role of cytoglobin in reference genes for quantitative sod gene expression analysis in C. elegans. podocytes: Implications for a role in chronic kidney disease. Antioxid. BMC Mol. Biol. 9, 9 Redox Signal. 32, 1155–1171 J. Biol. Chem. (2021) 296 100291 23     44        2.2 Androglobin, a chimeric mammalian globin, is required for male fertility    Anna  Keppner, Miguel  Correia,  Sara  Santambrogio,  Teng Wei  Koay,  Darko Maric,  Carina  Osterhof, Denise V Winter, Angèle Clerc, Michael Stumpe, Frédéric Chalmel, Sylvia Dewilde,  Alex Odermatt, Dieter Kressler, Thomas Hankeln, Roland H Wenger, David Hoogewijs (2022)  Androglobin, a chimeric mammalian globin, is required for male fertility. eLife 11:e72374  Supplementary files: https://doi.org/10.7554/eLife.72374    Own contributions to this Publication:  ‐ mRNA and miRNA isolation of testis issue from WT and Adgb‐KO mice following quality  control via Nanodrop, Qubit and Bioanalyzer.  ‐ Quality trimming and back mapping of mRNA‐Seq reads of testis issue from WT and  Adgb‐KO mice.   ‐ Quantification  of  expression  levels  and  differential  expression  analysis  with  subsequent Gene Ontology analysis  ‐ Pathway  analysis  of  differentially  expressed  genes  as well  as  analysis  of  potential  transcription factors involved    Planning  of  experiments,  analysis,  and  interpretation  of  data  as  well  as  drafting  of  the  corresponding part of the manuscript were realized together with Prof. Dr. T. Hankeln. The  project was managed by Prof Dr. David Hoogewijs.       45      46    RESEARCH ARTICLE Androglobin, a chimeric mammalian globin, is required for male fertility Anna Keppner1, Miguel Correia1, Sara Santambrogio2, Teng Wei Koay1, Darko Maric1, Carina Osterhof3, Denise V Winter4, Angèle Clerc1, Michael Stumpe5, Frédéric Chalmel6, Sylvia Dewilde7, Alex Odermatt4, Dieter Kressler5, Thomas Hankeln3, Roland H Wenger2, David Hoogewijs1* 1Department of Endocrinology, Metabolism and Cardiovascular system, University of Fribourg, Fribourg, Switzerland; 2Institute of Physiology, University of Zurich, Zurich, Switzerland; 3Institute for Organismic and Molecular Evolutionary Biology, University of Mainz, Mainz, Germany; 4Department of Pharmaceutical Sciences, University of Basel, Basel, Switzerland; 5Department of Biology, University of Fribourg, Fribourg, Switzerland; 6University of Rennes, Inserm, UMR_S 1085, Rennes, France; 7Department of Biomedical Sciences, University of Antwerp, Antwerp, Belgium Abstract Spermatogenesis is a highly specialized differentiation process driven by a dynamic gene expression program and ending with the production of mature spermatozoa. Whereas hundreds of genes are known to be essential for male germline proliferation and differentiation, the contribution of several genes remains uncharacterized. The predominant expression of the latest globin family member, androglobin (Adgb), in mammalian testis tissue prompted us to assess its physiological function in spermatogenesis. Adgb knockout mice display male infertility, reduced testis weight, impaired maturation of elongating spermatids, abnormal sperm shape, and ultrastruc- tural defects in microtubule and mitochondrial organization. Epididymal sperm from Adgb knockout animals display multiple flagellar malformations including coiled, bifid or shortened flagella, and erratic acrosomal development. Following immunoprecipitation and mass spectrometry, we could *For correspondence: identify septin 10 (Sept10) as interactor of Adgb. The Sept10-A dgb interaction was confirmed both david.hoogewijs@unifr.ch in vivo using testis lysates and in vitro by reciprocal co- immunoprecipitation experiments. Further- Competing interest: The authors more, the absence of Adgb leads to mislocalization of Sept10 in sperm, indicating defective manch- declare that no competing ette and sperm annulus formation. Finally, in vitro data suggest that Adgb contributes to Sept10 interests exist. proteolysis in a calmodulin-d ependent manner. Collectively, our results provide evidence that Adgb Funding: See page 22 is essential for murine spermatogenesis and further suggest that Adgb is required for sperm head shaping via the manchette and proper flagellum formation. Received: 21 July 2021 Preprinted: 18 September 2021 Accepted: 13 June 2022 Published: 14 June 2022 Editor's evaluation This manuscript demonstrates that male mice lacking androglobin, a poorly understood heme- Reviewing Editor: T Rajendra containing protein, are infertile and have defects in late stage spermatogenesis. The revisions are Kumar, University of Colorado, United States thorough and the inclusion of additional data makes the manuscript solid. The lack of defects at the hypothalamus- pituitary level makes the phenotype more striking as a direct effect of the mutation at Copyright Keppner et al. This the testis level. Overall, this revised version is significantly improved. article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the Introduction original author and source are Spermatogenesis is a complex and dynamic differentiation process ending by the production of mature credited. haploid spermatozoa (Hermo et al., 2010a; Hermo et al., 2010b; Hermo et al., 2010c; Hermo et al., Keppner et al. eLife 2022;11:e72374. DOI: https://doi.org/10.7554/eLife.72374 1 of 26 Research article Cell Biology | Developmental Biology 2010d; Hermo et al., 2010e). While type A spermatogonia undergo mitosis, one of the daughter cells serves to replenish the stem cell population, and the other daughter cell further divides mitotically, differentiates, and eventually forms type B spermatogonia. Following another mitotic division, type B spermatogonia engender primary spermatocytes, which complete their first meiotic division and form two secondary spermatocytes. Each secondary spermatocyte completes a second meiotic division, leading to the production of two haploid round spermatids (Mecklenburg and Hermann, 2016). During the final differentiation phase of spermatogenesis, known as spermiogenesis, the spermatids undergo profound morphological changes to differentiate and elongate into spermatozoa. These changes include condensation of the nucleus and compaction of the genetic material, the formation of the acrosome (the sperm head) from the Golgi apparatus before migration of the latter to the cytoplasmic droplet (Khawar et al., 2019), the formation of the sperm flagellum from the centriole, around which mitochondria will migrate to form the midpiece, the sperm annulus, and the mobile tail, and cytoplasmic reduction, whereby all unnecessary cytoplasmic remnants are eliminated. At the end of the elongation process, the spermatozoa are released into the lumen of the seminiferous tubule to migrate into the rete testis and epididymis for final maturation. Dysfunctions in any of these tightly orchestrated steps could lead to impaired spermatogenesis, meiotic arrest, or abnormal sperm forma- tion with direct consequences on male fertility (Neto et al., 2016). It is currently estimated that sperm defects and abnormalities remain idiopathic in about 30% of cases (Fainberg and Kashanian, 2019; Tüttelmann et al., 2018), and unraveling their molecular basis appears challenging since it is believed that over 4000 genes are possibly implicated in spermatogenesis (Jan et al., 2017). Globins are small globular metallo- proteins, which have the capacity to reversibly bind gaseous ligands via a typical 8 alpha-h elical structure in which a heme prosthetic group can be embedded. In mammals, five globin types exist: the well- established hemoglobin and myoglobin, neuroglobin in neuronal cells, cytoglobin ubiquitously expressed in fibroblasts, and the more recently identified androglobin (Adgb), predominantly expressed in mammalian testis tissue (Keppner et  al., 2020). Adgb is a chimeric protein containing an N-t erminal calpain- like cysteine protease domain, followed by an uncharacterized 300 amino acid long region, a central permuted functional globin domain (Bracke et al., 2018), interrupted by a potential calmodulin (CaM)- binding IQ motif, and a large 700 amino acid long C- terminal tail of unknown identity (Hoogewijs et al., 2012). Intriguingly, the chimeric nature of this globin resembles the domain structure of globin-c oupled sensors found in prokaryotes (Hou et al., 2001; Thijs et al., 2007). CaM is one of the most conserved proteins in eukaryotes and serves as a secondary messenger following intracellular binding of Ca2+ and interaction with one of the more than 300 identified downstream target proteins (Andrews et al., 2020). CaM is the main intracellular receptor for Ca2+, thereby participating in almost every biological process, including sper- matogenesis and sperm maturation and function (Darszon et al., 2011). Decreased mRNA expression levels in infertile vs. fertile men (Platts et al., 2007) suggest a potential role of Adgb in spermatogen- esis. Gene regulation and expression studies further suggest an association of Adgb with ciliogenesis including flagellum formation (Koay et al., 2021). However, the in vivo function of Adgb remains unexplored. In this study, we investigated the physiological function of Adgb during murine spermato- genesis by generating and analyzing Adgb knockout mice. We show that Adgb is mainly expressed in late steps of spermiogenesis, that it locates to the sperm flagellum, the annulus, and the midpiece, and that it is crucial for male fertility and sperm formation. Furthermore, we demonstrate that Adgb interacts with septin 10 (Sept10) and that co-l ocalization is detectable within the sperm neck in stage 12 and stage 15 spermatids and within the annulus of stage 15 spermatids and mature sperm. Finally, in vitro data suggest that Adgb contributes to Sept10 proteolysis in a CaM- dependent manner. Results Adgb knockout mice display male infertility A gene- trap strategy, provided by the Knockout Mouse Project (KOMP) (Skarnes et al., 2011), was applied to target exons 13 and 14 of the Adgb gene (Figure 1—figure supplement 1A). The correct targeting of ESCs was verified first by long-r ange PCR (Figure 1—figure supplement 1B) and second by Southern blotting (Figure 1—figure supplement 1C). The targeted Adgbtm1a(KOMP)Wtsi allele (Adgb tm1a mice), generated on a C57BL/6N background, displays a gene-t rap DNA cassette, which was inserted into the 12th intron of the Adgb gene. The gene trap consists of a splice acceptor site, Keppner et al. eLife 2022;11:e72374. DOI: https://doi.org/10.7554/eLife.72374 2 of 26 Research article Cell Biology | Developmental Biology an internal ribosome entry site, a β-galactosidase reporter sequence, and a neomycin resistance sequence. Breeding of Adgb tm1a mice with ubiquitously expressed CMV Cre-d eleter mice allowed generation of mice deficient for exons 13 and 14 but still expressing the β-galactosidase reporter (Adgb tm1b mice) (Figure 1—figure supplement 1A). Furthermore, mating of Adgb tm1a mice with Flp- deleter mice enabled the generation of conditional floxed mice (Adgb tm1c) (Figure 1—figure supplement 1A). These mice were further crossed with CMV Cre-d eleter mice to generate the full knockout animals (Adgb tm1d) (Figure 1—figure supplement 1A) that were used for all downstream applications if not otherwise stated. Genotyping was performed by regular PCR (Figure 1—figure supplement 1D) and revealed no significant differences in the Mendelian distribution of offspring for Adgb tm1d animals following interbreeding of heterozygous parents (380 pups: +/+, n=101; +/-, n=168; -/-, n=111; Χ2=5.62, p>0.05, ns). The genetic ablation of Adgb expression was further verified by reverse transcription (RT)-q uantitative (q)PCR (Figure 1A) and immunoblotting (Figure 1B and C). While female knockout mice displayed no fertility issues, male knockout mice never generated offspring, indicative for infertility. Full penetrance male infertility was also observed in homozygous tm1a and tm1b male mice, whereas homozygous tm1c animals showed normal fertility (data not shown). Accordingly, the testis weight was significantly reduced in knockouts (Figure 1D). Intrates- ticular testosterone (Figure  1E) as well as serum luteinizing hormone (LH) and follicle- stimulating hormone (FSH) levels (Figure 1—figure supplement 2) remained comparable between wild type and knockouts. Stage- specific histological examination of seminiferous tubules of control animals revealed normal architecture, normal spermatogenic maturation steps, and the presence of mature sperm with flagella extending into the lumen (Figure 1F). In contrast, in knockout animals, despite the presence of meiotic events (stages X–XII, Figure 1F), no flagella could be observed during the spermiation stage (stages VII–VIII, Figure 1F). The absence of mature sperm was accompanied by abnormally shaped heads, trapped stage 16 spermatids within the epithelium, and the presence of cytoplasmic material filling the lumen of the tubules (Figure 1F). Within the cauda epididymis, knockout animals displayed accumulations of residual bodies, cytoplasmic material, shed germ cells, and occasional abnormally shaped sperm heads but an overall absence of mature sperm as compared to wild- type animals (Figure 1G). No differences were detected in knockout testes at mRNA levels of nitric oxide synthases 1–3 (Nos1-3 ) or superoxide dismutases 1–3 (Sod1- 3), the ratio of Bax/Bcl2 was unchanged compared to wild- type testes, and terminal deoxynucleotidyl transferase dUTP nick- end labeling (TUNEL) assays on testis sections revealed no differences, suggesting no increase in oxidative stress or apoptotic events (Figure 1—figure supplement 3). Absence of Adgb interferes with the maturation of elongating spermatids To determine the temporal expression pattern of Adgb during spermatogenesis, wild-t ype embryos and pups at different post- natal ages, corresponding to the stages of the first wave of spermato- genesis during puberty, were dissected and analyzed by RT- qPCR. Whereas the expression of Adgb remained nearly undetectable until post- natal day 21 (corresponding to the stage of round sper- matids), Adgb mRNA levels drastically increased to reach a peak at post- natal day 25, coinciding with the first elongating spermatids (Figure 2A). Bulk and single-c ell RNA sequencing (scRNAseq) analysis in mouse and human datasets available at the ReproGenomics Viewer resource (Darde et al., 2019; Darde et al., 2015) as well as more recent murine scRNAseq data (Kwak and Jung, 2019) confirmed the conserved expression pattern of ADGB in spermatids (Green et al., 2018; Jégou et al., 2017; Lukassen et al., 2018; Wang et al., 2018; Figure 2—figure supplement 1; Figure 2—figure supplement 2). Accordingly, Adgb protein expression equally reached its peak at post-n atal days 26–28, suggesting slightly delayed translation (Figure 2B and C). This finding was further confirmed by propidium iodide staining and FACS sorting on testis lysates of the different genotypes. While no variations could be detected for phases 2C (spermatogonia, secondary spermatocytes, and testic- ular somatic cells), S-p hase (pre-m eiotic spermatogonia), 4C (primary spermatocytes), and 1C (round spermatids), an abnormal accumulation of elongating and elongated spermatids (phase H) could be detected in knockout animals, suggesting a blockade in the elongation process (Figure 2D). Addition- ally, immunofluorescence (Figure 2E), mRNA in situ hybridization (Figure 2F), and X- gal (Figure 2G) stainings of testis sections from wild- type and knockout mice confirmed the presence of Adgb within layers containing post- meiotic cells and further intensifying toward the lumen and mature sperm Keppner et al. eLife 2022;11:e72374. DOI: https://doi.org/10.7554/eLife.72374 3 of 26 Research article Cell Biology | Developmental Biology Figure 1. Validation of the knockout model and testicular phenotype. (A) Relative mRNA expression levels of Adgb in testes of wild-t ype (+/+) and knockout mice (-/-) (n=6 per genotype; p=0.000008). (B) Representative immunoblot for Adgb in testis lysates from wild-t ype (+/+) and knockout mice (-/-) (n=6 per genotype) and (C) corresponding protein quantification. Tubulin was used as loading control. p=0.0007. (D) Testis weight (g) in Adgb wild- type (+/+), heterozygous (+/-), and knockout (-/-) mice (n=8–13 per genotype). p=0.000003. (E) Intratesticular testosterone levels (pg/mg) in Adgb wild- type (+/+) and knockout (-/-) mice (n=6 per genotype). (F) Representative periodic acid Schiff (PAS)-h ematoxylin-s tained sections of testes from Adgb wild-t ype (+/+) and knockout mice (-/-) at the different stages of spermatogenesis. Heads (H), flagella (F), residual bodies (R), cytoplasmic debris Figure 1 continued on next page Keppner et al. eLife 2022;11:e72374. DOI: https://doi.org/10.7554/eLife.72374 4 of 26 Research article Cell Biology | Developmental Biology Figure 1 continued (CD), meiosis (M), elongating spermatids (ES), round spermatids (RS), stage 16 spermatids (S16), and abnormal heads (A) are indicated. Note the full absence of flagella in knockout sections. Scale bar represents 50 µm. (G) Representative H&E stained sections of epididymides from Adgb wild-t ype (+/+) and knockout mice (-/-). H, tails (T), cytoplasmic bodies (C), R, germ cells (G), and A are shown. Note the empty lumen (EL) in knockout mice. Scale bar represents 50 µm. ** p<0.01, *** p<0.001. The online version of this article includes the following source data and figure supplement(s) for figure 1: Source data 1. Original uncropped immunoblots of Figure 1B with indication of the cropped areas. Figure supplement 1. Generation of Adgb knockout mice. Figure supplement 1—source data 1. Original uncropped gels of Figure 1—figure supplement 1B,C,D with indication of the cropped areas. Figure supplement 2. Normal serum gonadotropin levels in Adgb knockout mice. Figure supplement 3. Adgb knockout mice do not display changes in Nos, Sod, and apoptotic gene expression. (Figure 2E–G). Moreover, in mature sperm, Adgb expression could be visualized within the midpiece and along the whole flagellum by both X-g al (Figure  2G) and immunofluorescence (Figure  2H) stainings. Adgb is required for proper sperm flagellum formation To gain additional insights into the origin of male infertility, cauda epididymal sperm was collected from both wild- type and knockout mice, and visualized under a microscope. While wild-t ype sperm appeared normal, very few knockout spermatozoa were found and displayed various defects of the head and/or flagellum structure, including shortened or bifid tails, loopings of the flagellum, and immature acrosomal structures (Figure 3A). Knockout sperm acrosome structure appeared partially conserved, with sperm displaying either normal acrosomes, or various defects in shape or staining intensity, or total absence (Figure 3B). Stage- specific transmission electron microscope (TEM) ultra- structural analysis of testis sections revealed various defects associated with sperm flagellum forma- tion. Whereas numerous axonemes displaying the regular 9+2 structure from step 9 spermatids could be found in wild types, none were found in knockouts at the same stage. In later stages (step 12 spermatids and onward), rare and abnormal axonemes could be observed in knockouts, displaying disorganized microtubular structures and forming microtubular clusters without defined fibrous or mitochondrial sheath (Figure 3C). We could further observe misshaped heads with nuclear inclusions, defective manchette elongation, and abnormal acrosomes in knockout sections (Figure 3C–F). The Adgb-dependent transcriptome reveals dysregulation of multiple spermiogenesis genes To understand the molecular consequences of loss of Adgb in the testis, we performed RNAseq experiments on total testis RNA from wild-t ype and knockout mice at post-n atal day 25. An elab- orate set of significantly differentially expressed genes (74 genes upregulated and 204 downregu- lated) was identified, underscoring the crucial function of Adgb in spermatogenesis (Figure 3—figure supplement 1A, Figure 3—source data 1). Functional analysis based on gene ontology term enrich- ments confirmed that many of these genes are related to sperm head, acrosome reaction, acrosomal membrane, sperm motility, spermatid development, and spermatid differentiation, in line with the pronounced structural changes in spermatids during spermiogenesis (Figure 3—figure supplement 1B). Adgb interacts and co-localizes with Sept10 To obtain more insights into the physiological function of Adgb, we explored the Adgb- dependent interactome. Total protein extracts from wild- type testes were immunoprecipitated (IP) with anti- Adgb or control IgG antibodies and subsequently submitted to mass spectrometry (MS) analysis to reveal potential interacting proteins of Adgb. Among the specifically enriched proteins, there were various members of the septin family, such as Sept10, Sept11, Sept2, and Sept7 (Figure  4A and Figure 4—source data 4). Particular focus was put on Sept10 for further downstream experiments given its strong enrichment (4.509 log2- fold) combined with high abundance (3.864 log2[mean]) in the immunoprecipitation as well as substantial sequence coverage (40.5%), all reflected by its close position to Adgb among all septins in Figure  4A. To confirm the interaction between Adgb and Keppner et al. eLife 2022;11:e72374. DOI: https://doi.org/10.7554/eLife.72374 5 of 26 Research article Cell Biology | Developmental Biology Figure 2. Testicular Adgb expression pattern and localization. (A) Relative mRNA expression levels of Adgb in testes of wild-t ype mice during embryonic development (E) and early post- natal (P) life (n=3–4 per condition). (B) Representative immunoblot for Adgb in testis lysates from wild-t ype mice at different P ages (n=3 per condition) and (C) corresponding protein quantification. Tubulin was used as loading control. (D) Flow cytometric analysis of spermatogenic cell populations following propidium iodide staining in Adgb wild-t ype (+/+, white circles, n=4), heterozygous (+/-, gray Figure 2 continued on next page Keppner et al. eLife 2022;11:e72374. DOI: https://doi.org/10.7554/eLife.72374 6 of 26 Research article Cell Biology | Developmental Biology Figure 2 continued circles, n=5), and knockout (-/-, black circles, n=3) testes. H: elongating and elongated spermatids; 1 C: round spermatids; 2 C: spermatogonia, secondary spermatocytes, and testicular somatic cells; S: spermatogonia synthesizing DNA; 4 C: primary spermatocytes. p=0.00024. (E) Representative pictures of Adgb protein (red fluorescence) detection in testes of wild- type (+/+) and knockout (-/-) animals. Left panels 20× magnification, right panels 40× magnification, and scale bars represent 100 µm and 50 µm, respectively. Nuclei were stained with DAPI. (F) Representative pictures of Adgb mRNA in situ hybridization in testes from wild- type (+/+) and knockout (-/-) animals. Left panels 20× magnification; right panels, 40× magnification; scale bars represent 100 µm and 50 µm, respectively. Positive (ctrl +, PPIB) and negative (ctrl −, DapB) control sections are shown. (G) Representative pictures of β-galactosidase activity (X- gal staining) in testes from Tm1b wild- type (+/+) and Tm1b heterozygous (tg/+) mice and isolated spermatozoa from Tm1b heterozygous (tg/+) mice. Left panels, 20× magnification; middle panels, 40× magnification; right panel, 60× magnification; scale bars represent 100 µm, 50 µm, and 20 µm, respectively. Spermatozoa were counterstained with nuclear fast red. (H) Representative picture of Adgb protein (green fluorescence) in a single spermatozoon from wild- type (+/+) mice (left panel) and negative control (secondary antibody only, right panel). Scale bar represents 20 µm and nuclei were stained with DAPI. ** p<0.001. The online version of this article includes the following source data and figure supplement(s) for figure 2: Source data 1. Original uncropped immunoblots of Figure 2B with indication of the cropped areas. Figure supplement 1. Temporal Adgb and Sept10 expression profiles based on single- cell and bulk RNAseq datasets. Figure supplement 2. Adgb and Sept10 gene expression profiles based on single-c ell RNAseq of mouse testes along the temporospatial axis of spermatid maturation. Sept10, reciprocal co- immunoprecipitation (co- IP) experiments were performed (Figure 4B and C) on tissue extracts from wild-t ype and knockout testes (Figure 4B) and in HEK293 cells overexpressing full- length ADGB (Figure 4C) and SEPT10. The results demonstrate that Adgb and Sept10 interact both in vivo (Figure 4B) and in vitro (Figure 4C), whereas in testis lysates of Adgb-d eficient mice, no Sept10 co- precipitation was observed (Figure  4B). Endogenous Sept10 protein levels were equal in testis lysates of Adgb- deficient and wild- type mice, as were endogenous levels of Sept11, Sept7, and Sept2, as well as other septins that are crucial for spermatogenesis, including Sept8, Sept9, and Sept14 (Figure  4—figure supplement 1). Reciprocal co- IP experiments between Adgb and other enriched septins (Sept2, Sept7, and Sept11) revealed no interaction, neither in vivo nor in vitro (Figure 4—figure supplement 2, Figure 4—figure supplement 3), whereas SEPT7 and SEPT10 inter- acted in vitro (Figure 4—figure supplement 3). We next investigated whether the interaction with SEPT10 occurs at the N-t erminal or the C-t erminal portion of ADGB (Figure 4D and E). Following co- overexpression of ADGB deletion constructs (Figure 4—figure supplement 4) with SEPT10 and subsequent co- IP, immunoblotting revealed that both parts of ADGB interact with SEPT10 (Figure 4D and E) and that this interaction remained intact also upon deletion of the coiled-c oil domain of ADGB (Figure 4—figure supplement 5). Consistent with a functional interaction, the temporal expression profiles of SEPT10 and ADGB substantially overlap as illustrated by analysis of bulk and scRNAseq datasets of mouse and human RNA (Figure 2—figure supplement 1; Figure 2—figure supplement 2) as well as by RT-q PCR and immunoblotting of mouse tissue samples (Figure 4—figure supplement 6). The localization of Adgb and Sept10 was assessed in intact testis sections, microdissected tubules, and in epididymal sperm by immunofluorescence. Co-l ocalization of Adgb and Sept10 was visible during spermatid matu- ration and in flagella in testis sections (Figure  5A) and at the level of the sperm annulus in S12 and S15 spermatids and mature wild-t ype sperm (Figure 5B, Figure 5—figure supplement 1A). A moderate Sept10 staining was also observed in the neck region of S15 spermatids and in mature sperm (Figure 5B, Figure 5—figure supplement 1A). The localization of Sept10 in knockout testis sections appeared overall fainter but within the same locations as in wild types, with the exception of sperm flagella (Figure 5A). However, in knockout S12 and S15 spermatids, as well as in epididymal sperm, only a single signal, likely corresponding to the annulus as verified by Sept7 staining of epidid- ymal sperm (Figure 5—figure supplement 1A), was observed and displayed abnormal migration, indicating defective manchette or microtubule formation (Figure 5B, Figure 5—figure supplement 1A). The migration of the annulus drives mitochondrial placement along the forming mid-p iece (Toure et al., 2011). Since knockout animals displayed abnormal ultrastructural mitochondria organization (Figure  3), CoxIV staining was performed on wild-t ype and knockout microdissected tubules and epididymal sperm. As expected, a robust staining was observed along the whole midpiece in wild- type spermatids and sperm, whereas mitochondria were barely visible and formed a cloudy structure around the neck region of knockout spermatids and sperm (Figure 5—figure supplement 1B). Keppner et al. eLife 2022;11:e72374. DOI: https://doi.org/10.7554/eLife.72374 7 of 26 Research article Cell Biology | Developmental Biology Figure 3. Defective spermatogenesis is associated with flagellar malformation in Adgb knockout mice. (A) Representative pictures of cauda epididymis sperm from wild- type (+/+) and Adgb knockout animals (-/-). Scale bar represents 20 µm. (B) Representative pictures of peanut agglutinin- stained cauda epididymis sperm from wild- type (+/+) and Adgb knockout animals (-/-). Nuclei were stained with DAPI. Scale bar represents 20 µm. (C–F) Representative transmission electron microscope (TEM) pictures from wild- type (+/+, left panels) and knockout (-/-, right panels) testes at various Figure 3 continued on next page Keppner et al. eLife 2022;11:e72374. DOI: https://doi.org/10.7554/eLife.72374 8 of 26 Research article Cell Biology | Developmental Biology Figure 3 continued stages of the first wave, tubular stages are indicated. (C) 9+2 microtubular structure (asterisks), forming sperm flagella (crosses), and (impaired) outer dense fibers (arrows) are shown. (D) Misshaped sperm heads with nuclear inclusions (ampersand), (E) defective manchette elongation (dollar), and (F) abnormal acrosomes (arrowheads) are shown. Scale bar lengths are indicated on each picture. The online version of this article includes the following source data and figure supplement(s) for figure 3: Source data 1. Differentially regulated genes in wild- type vs. Adgb knockout testes. Figure supplement 1. Volcano plot and gene ontology (GO) term analysis of differentially expressed genes in Adgb knockout mice testis samples. ADGB contributes to SEPT10 proteolytic cleavage in vitro To analyze the functional consequences of the SEPT10- ADGB interaction, we transiently co- over- expressed both proteins in HEK293 cells. Intriguingly, apart from the intact form of overexpressed SEPT10 at 60 kDa, increased levels of a lower band of 37 kDa were consistently detected in the pres- ence of co- overexpressed ADGB in a dose-d ependent manner (Figure 6A). Immunoblotting with a V5- antibody upon co- overexpression of a C-t erminal V5-t agged SEPT10 with ADGB (Figure 6B) as well as the presence of this band upon SEPT10/ADGB co- IP (Figure 6B, Figure 6—figure supple- ment 1) further supports its origin as proteolytic SEPT10 product. To investigate a potential oxygen- dependent influence of the globin domain, this experiment was repeated under normoxic and hypoxic conditions (0.2% O2) but no differences were observed upon exposure to hypoxic conditions (Figure 6C). To determine a potential role of CaM, we constructed a deletion mutant lacking the IQ domain. Notably, transient overexpression of IQ- mutant ADGB resulted in considerably reduced appearance of the 37 kDa SEPT10 band relative to wild- type ADGB (Figure 6D), and the same was observed with an ADGB protease domain deletion mutant, further supporting a CaM- dependent proteolytic cleavage (Figure  6D). These findings prompted us to experimentally validate the suspected CaM- ADGB interaction. Whereas co-I P experiments following overexpression of full-l ength ADGB did not interact with CaM under basal experimental conditions in HEK293 cells (Figure 6— figure supplement 2A), a truncated construct covering the globin and IQ domains displayed robust ADGB- CaM interaction (Figure 6E). Consistently, MS analysis of proteins that were present in the IP of the overexpressed, isolated globin domain revealed a prominent enrichment of endogenous CaM (Figure 6—figure supplement 3 and Figure 6—source data 5). Importantly, individual or double mutation of the proximal histidine (HisF8) or distal glutamine (GlnE7), critical residues in the globin domain for heme coordination, did not alter the interaction, suggesting that the ADGB-C aM inter- action occurs independently of heme incorporation (Figure 6F, Figure 6—figure supplement 4). To fully exclude that the mutant globin domain might still be hemylated, we repeated these experiments under medium heme depletion and heme synthesis blocking conditions. Consistently, upon heme deprivation, robust interaction was observed between CaM and the isolated intact as well as double mutated (HisF8/GlnE7) globin domain. (Figure 6—figure supplement 5). As an additional layer of support for the ADGB- CaM interaction, chimeric Gal4 DNA-b inding domain and VP16 transactivation domain- based fusion constructs were generated for mammalian 2- hybrid (M2H) luciferase reporter gene assays (Figure 6G and H). These M2H assays were performed in two different cell lines, HEK293 and A375, and revealed up to 7.5- fold increase in luciferase activity upon co- transfection of both chimeric proteins compared to single construct transfections, providing independent evidence that ADGB interacts with CaM. However, when full- length ADGB was employed in M2H assays with inter- changed Gal4 and VP16 domains, no interaction with CaM was observed, corroborating the co-I P data (Figure 6—figure supplement 2B). To further assess the potential O2- dependency of the CaM- ADGB interaction, we repeated the M2H assays in A375 cells under hypoxic conditions. Exposure to hypoxic conditions (0.2% O2) did not alter luciferase activity (Figure 6H, left panel), while lucif- erase activity of a 5’/3’-hypoxia response element- dependent EPO promoter- driven reporter gene increased (Figure 6H, right panel), suggesting that the ADGB-C aM interaction is O2 independent. Of note, while maintaining the O2- independent interaction between Gal4-C aM and VP16- globin, the single Gal4-C aM construct control was increased under hypoxic conditions, albeit to a much lesser extent than the EPO control, possibly related to increases of cytoplasmic Ca2+ mobilized from intracel- lular stores or by extracellular influx under hypoxic conditions and according activation of CaM (Yuan et al., 2005). This result is consistent with the maintained ADGB- CaM interaction following muta- tion of critical residues in the globin domain in co-I P experiments and unchanged ADGB-d ependent Keppner et al. eLife 2022;11:e72374. DOI: https://doi.org/10.7554/eLife.72374 9 of 26 Research article Cell Biology | Developmental Biology Figure 4. Adgb and Sept10 interact in vivo and in vitro. (A) Proteins of the septin family are specifically enriched in the Adgb immunoprecipitation (IP). The iBAQ (intensity-b ased absolute quantification) values of each Adgb IP (triplicate) and IgG control IP (duplicate) were log2 transformed and normalized against the median value. Missing values were imputed before the mean values of the Adgb and IgG control IPs were calculated. The normalized abundance of each protein detected in the Adgb IP (log2 Adgb [mean]) is plotted against its specific enrichment compared to the IgG Figure 4 continued on next page Keppner et al. eLife 2022;11:e72374. DOI: https://doi.org/10.7554/eLife.72374 10 of 26 Research article Cell Biology | Developmental Biology Figure 4 continued control IP log2 (Adgb [mean]/IgG [mean]). Adgb and septins are highlighted as blue and red dots, respectively, in the christmas tree plot representation. (B) Representative immunoblot of Adgb and Sept10 in testis lysates from wild- type (+/+) and knockout (-/-) mice following co- IP of Adgb and Sept10. (C–E) Representative immunoblots of ADGB and SEPT10 in protein lysates of HEK293 cells (co-) transfected with full- length ADGB (C), N- ter ADGB (D) and C-t er ADGB (E), and SEPT10 following co-I P of ADGB and SEPT10. Schematic representation of deletion constructs is provided in Figure 4— figure supplement 4. The online version of this article includes the following source data and figure supplement(s) for figure 4: Source data 1. Original uncropped immunoblots of Figure 4B and C with indication of the cropped areas. Source data 2. Original uncropped immunoblots of Figure 4D with indication of the cropped areas. Source data 3. Original uncropped immunoblots of Figure 4E with indication of the cropped areas. Source data 4. Raw mass spectrometry (MS) data of the Adgb immunoprecipitation (IP) vs. IgG control IP. Figure supplement 1. The protein expression levels of Sept2, Sept7, Sept8, Sept9, Sept10, Sept11, and Sept14 are unaffected in Adgb knockout testis. Figure supplement 1—source data 1. Original uncropped immunoblots of Figure 4—figure supplement 1A-C with indication of the cropped areas. Figure supplement 1—source data 2. Original uncropped immunoblots of Figure 4—figure supplement 1D-F with indication of the cropped areas. Figure supplement 1—source data 3. Original uncropped immunoblots of Figure 4—figure supplement 1G with indication of the cropped areas. Figure supplement 2. Adgb does not interact with other septin family members in vivo. Figure supplement 2—source data 1. Original uncropped immunoblots of Figure 4—figure supplement 2A,B,C with indication of the cropped areas. Figure supplement 3. ADGB does not interact with other septin family members in vitro. Figure supplement 3—source data 1. Original uncropped immunoblots of Figure 4—figure supplement 3A with indication of the cropped areas. Figure supplement 3—source data 2. Original uncropped immunoblots of Figure 4—figure supplement 3B with indication of the cropped areas. Figure supplement 3—source data 3. Original uncropped immunoblots of Figure 4—figure supplement 3C with indication of the cropped areas. Figure supplement 3—source data 4. Original uncropped immunoblots of Figure 4—figure supplement 3D with indication of the cropped areas. Figure supplement 3—source data 5. Original uncropped immunoblots of Figure 4—figure supplement 3E with indication of the cropped areas. Figure supplement 4. ADGB constructs used throughout the study. Figure supplement 5. The interaction between ADGB and SEPT10 is maintained despite mutation of the coiled-c oil domains. Figure supplement 5—source data 1. Original uncropped immunoblots of Figure 4—figure supplement 5 with indication of the cropped areas. Figure supplement 6. Sept10 temporal expression profile on mRNA and protein levels. Figure supplement 6—source data 1. Original uncropped immunoblots of Figure 4—figure supplement 6 with indication of the cropped areas. cleavage of SEPT10 under hypoxic conditions. Collectively, these in vitro data suggest a scenario in which overexpressed ADGB proteolytically contributes to cleavage of overexpressed SEPT10 in an O2-i ndependent but CaM- dependent manner. Discussion In the present study, we explored the testicular function of Adgb by generating and analyzing Adgb constitutive knockout mice. Our results demonstrate that Adgb is indispensable for proper sperm formation and male fertility. The absence of Adgb has profound deleterious effects on the elongation of spermatids, leading to multiple malformations of the sperm flagellum and head, as evidenced by the presence of various ultrastructural defects, including aberrant microtubule formation, misshaped heads with nuclear inclusions, defective manchette formation, and abnormal acrosomes. Notably, Adgb could be localized not only to the midpiece and sperm flagellum but also in the neck region (likely the centriole) and within the sperm annulus. In principle, the observed male infertility could be the result of defects in the hypothalamic- pituitary- gonadal (HPG) axis. Unchanged LSH and LH levels indicate an intact HPG signaling and suggest the phenotypes represent direct (primary) effects of Adgb deficiency in testis. However, indi- rect effects cannot be entirely excluded despite normal serum levels of gonadotropins. Unchanged testicular testosterone levels are consistent with the total absence of Adgb expression in Leydig cells of wild- type mice. Strikingly, upon LC-M S/MS interactome analysis after Adgb immunoprecipitation in testicular lysates, several members of the septin family of proteins ranked among the top hits. Septins are Keppner et al. eLife 2022;11:e72374. DOI: https://doi.org/10.7554/eLife.72374 11 of 26 Research article Cell Biology | Developmental Biology Figure 5. Adgb and Sept10 co-l ocalize in the sperm neck and annulus. (A) Representative pictures of Adgb protein (green fluorescence) and Sept10 (red fluorescence) in testis sections from wild- type (+/+, left panels) and knockout (-/-, right panels) mice at various stages of the first wave, tubular stages are indicated. Sections were counterstained with DAPI. Negative control (secondary antibodies only) is shown (lower panels). Scale bar represents 50 µm. (B) Representative pictures of Adgb (green fluorescence) and Sept10 (red fluorescence) in elongating spermatids (stage 12 [S12] upper panels Figure 5 continued on next page Keppner et al. eLife 2022;11:e72374. DOI: https://doi.org/10.7554/eLife.72374 12 of 26 Research article Cell Biology | Developmental Biology Figure 5 continued and stage 15 [S15] middle panels) after stage- specific tubule dissection of wild-t ype (+/+) and knockout (-/-) testes. Nuclei were stained with DAPI. Negative controls (secondary antibodies only) are shown (lower panels). Scale bar represents 10 µm. Sperm neck (arrowhead) and annulus (arrow) are highlighted. The online version of this article includes the following figure supplement(s) for figure 5: Figure supplement 1. Absence of Adgb leads to abnormal annulus migration and mitochondrial disorganization. conserved GTPases that have the ability to form large oligomers and filamentous polymers and which associate with cell membranes and with the cytoskeleton. They serve as scaffolds for the proper local- ization of intracellular proteins via their diffusion barrier-f orming characteristics (Dolat et al., 2014). In sperm, various septins (including Sept1, Sept4, Sept6, Sept7, and Sept12) have been localized to the sperm annulus, where they polymerize to a filamentous structure called the septin ring, forming a barrier between the midpiece and the principal piece of the spermatozoon (Toure et al., 2011). Interestingly, despite several members of the septin family being present in the interactome, only Sept10 was found to interact with Adgb in co-I P experiments. Furthermore, the interaction of the two proteins could also be localized within the connecting piece and within the annulus in wild- type sperm, whereas in knockout sperm, only one signal was detected, suggesting the absence of the annulus. Indeed, when staining against Sept7 which interacts with Sept10, the annulus was either missing or failed to migrate properly in knockout sperm, supporting a defect in the annulus forma- tion and migration, thereby also leading to misalignment of the mitochondria in knockout sperm. In accordance with this observation, two main functions have been proposed for the annulus: (1) a diffusion barrier function to compartmentalize different proteins to various locations in the sperm tail and (2) a growth guide function for the sperm flagellum and for aligning the mitochondria along the axoneme (Avidor-R eiss et al., 2020; Avidor- Reiss et al., 2017; Toure et al., 2011). Supportive of this function, Sept4-/- and Sept12+/- mice are infertile and display disorganized sperm mitochon- dria (Ihara et al., 2005; Kissel et al., 2005; Lin et al., 2009). Moreover, Sept4-/- mice display a bent sperm tail and absence of annulus (Ihara et al., 2005; Kissel et al., 2005), whereas Sept12+/- mice exhibit broken acrosomes, misshaped nuclei, and increased apoptosis of germ cells (Lin et al., 2009). Correspondingly, SEPT12 mutations have been described in infertile men displaying abnormal sperm including defective annulus with a bent tail (Kuo et al., 2012). Additionally, defective sperm head morphology and DNA integrity have recently been reported for two different SEPT14 missense muta- tions (Lin et al., 2020; Wang et al., 2019). Interestingly, another ring- like septin structure was recently described at the sperm neck, composed of Sept12 which complexes together with Sept1, Sept2, Sept10, and Sept11. Two mutations of Sept12 identified in patients disrupted the complex, leading to unstable head-t ail junctions and defective connecting piece formation. Strikingly, the mutation of Sept12 and the subsequent disruption of the complex led to loss of Sept10 signal in the annulus (Shen et al., 2020) as also observed in Adgb-d eficient mice. Accordingly, our data suggest an inter- dependence between Adgb and Sept10, which is required for the maintenance of the annulus, head shaping, and proper mitochondrial localization. Sperm flagella and motile cilia, which are hair- like microtubular protrusions at the surface of various cell types, share numerous characteristics, not only in their structure but also in their regulation and growth, whereby septins have also been identified as components of cilia. Sept2 forms a diffusion barrier at the base of the cilium, impeding ciliary formation through loss of Sonic Hedgehog signaling when depleted (Hu et al., 2010). Sept2/7/9 form a complex that associates with the ciliary axoneme, thereby regulating ciliary length (Ghossoub et al., 2013). Accordingly, septin association with the cytoskeleton and particularly with microtubular structures has been extensively studied (Spiliotis and Nakos, 2021), and numerous other cilia-r elated proteins participate in sperm flagellum formation. Furthermore, most ciliopathies include male infertility and immotile sperm due to defective axonemal organization (Brown and Witman, 2014). Likewise, Adgb knockout mice display aberrant microtu- bule arrangements, thus it cannot be excluded that a scaffolding and simultaneous regulatory action between Adgb and Sept10 is necessary to support microtubular structure. Moreover, a recent study reported the consistent presence of Adgb in the ciliomes of three distinct evolutionary ancestral taxa, further suggesting a conserved function related to microtubular organization and likely flagella forma- tion (Sigg et al., 2017). In line with this study, our recent in vitro investigations have demonstrated Keppner et al. eLife 2022;11:e72374. DOI: https://doi.org/10.7554/eLife.72374 13 of 26 Research article Cell Biology | Developmental Biology Figure 6. ADGB contributes to in vitro CaM- dependent SEPT10 cleavage. (A) Representative immunoblots of ADGB and SEPT10 in protein lysates of HEK293 cells co- transfected with plasmids encoding SEPT10 and two dose-d ependent amounts of full-l ength ADGB. (B) Representative immunoblot of ADGB and V5 in protein lysates of HEK293 cells (co- )transfected with full-l ength ADGB and a C- terminally V5-t agged SEPT10 construct following co- immunoprecipitation (co- IP) of ADGB and V5- SEPT10. (C) Representative immunoblots of ADGB, V5, HIF- 1α, and PHD2 in protein lysates of HEK293 Figure 6 continued on next page Keppner et al. eLife 2022;11:e72374. DOI: https://doi.org/10.7554/eLife.72374 14 of 26 Research article Cell Biology | Developmental Biology Figure 6 continued cells (co- )transfected with full- length ADGB and V5- SEPT10 following exposure to normoxic (20% O2) and hypoxic conditions (0.2% O2) for 24 hr. HIF- 1α and PHD2 were used as positive controls for hypoxia. (D) Representative immunoblot of ADGB and V5 in protein lysates of HEK293 cells (co-) transfected with full-l ength ADGB, V5- SEPT10, an ADGB- IQ deletion mutant, and an ADGB- calpain protease domain deletion mutant and corresponding protein quantification of cleaved/full- size SEPT10 ratio (n=3–6 independent experiments). Ponceau S protein staining was used as loading control. Schematic representation of deletion constructs is provided in Figure 4—figure supplement 4. (E) Representative immunoblot of GFP and V5 in protein lysates of HEK293 cells (co- )transfected with a truncated construct of the globin domain of ADGB (spanning the IQ domain) (GFP-G lobin) and a V5- tagged CaM (V5-C almodulin) following IP of GFP. (F) Representative immunoblot of GFP and V5 in protein lysates of HEK293 cells (co-) transfected with GFP- Globin, a GFP- Globin construct with mutation of the proximal heme- binding histidine (GFP- Globin-F 8[Gly]), a GFP- Globin construct with mutation of the distal glutamine (GFP- Globin-E 7[Gly]), and V5-C aM following IP of GFP. (G, H) Mammalian 2- hybrid assays in HEK293 cells under normoxic conditions (G) and A375 cells under normoxic and hypoxic (0.2% O2) conditions (H) (n=3–5 independent experiments). HEK293 and A375 cells were transiently transfected with fusion protein vectors based on a Gal4 DNA- binding domain fused to calmodulin (Gal4-C aM) and a VP16 activation domain fused to the ADGB globin domain comprising the IQ domain (VP16- Globin), a Gal4 response element- driven firefly luciferase reporter, and a Renilla luciferase control vector. Increasing transfection amounts for the Gal4- CaM fusion protein were employed. Following transfection, A375 cells were incubated under normoxic (20% O2) or hypoxic (0.2% O2) conditions, and luciferase reporter gene activities were determined 24 hr later. Single construct transfections served as negative controls, whereby hypoxic regulation of CaM has been described previously (Yuan et al., 2005). A 5’/3’-hypoxia response element- dependent EPO promoter-d riven firefly luciferase construct served as hypoxic control. * p<0.05, ** p<0.01, and *** p<0.001. The online version of this article includes the following source data and figure supplement(s) for figure 6: Source data 1. Original uncropped immunoblots of Figure 6A and B with indication of the cropped areas. Source data 2. Original uncropped immunoblots of Figure 6C with indication of the cropped areas. Source data 3. Original uncropped immunoblots of Figure 6E with indication of the cropped areas. Source data 4. Original uncropped immunoblots of Figure 6F with indication of the cropped areas. Source data 5. Raw mass spectrometry (MS) data of the ADGB globin IP vs. GFP control IP. Figure supplement 1. Reciprocal co-i mmunoprecipitation (co- IP) of ADGB and V5- SEPT10 from Figure 6B. Figure supplement 1—source data 1. Original uncropped immunoblots of Figure 6—figure supplement 1 with indication of the cropped areas. Figure supplement 2. Full-l ength ADGB does not interact with CaM. Figure supplement 2—source data 1. Original uncropped immunoblots of Figure 6—figure supplement 2 with indication of the cropped areas. Figure supplement 3. ADGB globin immunoprecipitation (IP) vs. GFP control IP. Figure supplement 4. Double mutation of the key heme- binding residues does not abrogate interaction of ADGB and CaM. Figure supplement 4—source data 1. Original uncropped immunoblots of Figure 6—figure supplement 4 with indication of the cropped areas. Figure supplement 5. Heme depletion does not impact the interaction between GFP- Globin and CaM. Figure supplement 5—source data 1. Original uncropped immunoblots of Figure 6—figure supplement 5D with indication of the cropped areas. Figure supplement 5—source data 2. Original uncropped immunoblots of Figure 6—figure supplement 5E with indication of the cropped areas. that Adgb is transcriptionally regulated by FoxJ1, a master regulator of ciliogenesis (Koay et  al., 2021). A robust ADGB- SEPT10 interaction was also detected by co- IP experiments in transiently co-t rans- fected HEK293 cells. The interaction persisted despite various mutations of putative binding sites within ADGB, suggesting that the interaction occurs at different locations along the two proteins and that the proteins may entangle around each other. Indeed, it is not uncommon that protein- protein interactions involve multiple interaction surfaces, as referred to in a recent study (Egri et al., 2022). Since ADGB contains multiple larger domains of uncharacterized function, it is conceivable that various sites could contribute to interaction with proteins. The close ADGB-S EPT10 interaction may serve as targeting mechanism for the proteolytic processing of SEPT10 that we could observe upon ectopic expression of both proteins. The unique chimeric domain structure of Adgb with an N- terminal calpain-l ike protease domain and the presence of an IQ motif suggests a CaM-m ediated regulation. CaM- bound calcium represents a crucial activator of proteolytic activity of Ca2+- dependent calpain proteases (Bähler and Rhoads, 2002; Villalobo et  al., 2019). An interesting observation in our study is the interaction of CaM and ADGB upon isolation of the globin domain but not with the full-l ength ADGB protein. The lack of interaction upon overexpression of a full-l ength ADGB might be due to interference by the larger protein (particularly the 700 amino acid long C- terminal region of unidentified origin). As such the globin domain might be liberated possibly by proteolytic cleavage. This hypothesis is consistent with our previous observations of time- dependent truncation Keppner et al. eLife 2022;11:e72374. DOI: https://doi.org/10.7554/eLife.72374 15 of 26 Research article Cell Biology | Developmental Biology of full-l ength ADGB from baculovirally infected Spodoptera frugiperda (Sf9) insect cells (Bracke et al., 2018). Similar truncations are detected upon overexpression of full- length ADGB in HEK293 cells in the current study (e.g. Figure  6A). Furthermore, this observation is consistent with the nature of Ca2+- dependent calpains, displaying frequent autolytic activation by auto- proteolysis (Ermolova et al., 2015; Osako et al., 2010). Additional experiments would be required to prove these hypoth- eses. Importantly, the interaction between CaM and the ADGB IQ-b inding motif, as well as presence of the calpain protease domain, seems pivotal in the observed proteolytic cleavage of SEPT10 as the cleavage product completely disappeared upon overexpression of either IQ or protease deletion constructs. Calpain- dependent proteolytic cleavage of septins is not unprecedented; it was shown that Sept5 is a substrate of both calpain-1 and calpain- 2 in platelets, where the cleavage triggers secretion of chemokine-c ontaining granules (Randriamboavonjy et al., 2012). Septins are sensitive to molecular modifications, such as SUMOylation which impacts Sept6, 7, and 11- dependent filament formation (Ribet et al., 2017), or to PKA- dependent phosphorylation as reported for Sept12, leading to its dissociation from the septin complex and disruption of filament formation (Shen et al., 2017). It is therefore conceivable that CaM-d ependent Adgb- mediated proteolytic cleavage of Sept10 may be a prerequisite for proper Sept10 function or localization within the sperm neck or annulus. In conclusion, our study is the first to demonstrate a functional role for Adgb, the fifth mammalian globin. We present convincing in vivo evidence that Adgb is required for murine spermatogenesis. Adgb is necessary for sperm head shaping and for proper microtubule and flagellum formation. Our in vitro data illustrate CaM binding to ADGB and suggest that ADGB contributes to proteolytical cleavage of SEPT10 in a CaM-d ependent manner. Our work provides a crucial contribution to the characterization of the physiological role of this novel enigmatic chimeric globin type. Materials and methods Animals, ethics statement, and genotyping All experimental procedures and animal maintenance followed Swiss federal guidelines, and the study was revised and approved by the ‘Service de la sécurité alimentaire et des affaires vétérinaires’ (SAAV) of the canton of Fribourg, Switzerland (license number 2017_16_FR). Animals were housed in rooms with a 12 hr/12 hr light/dark cycle, controlled temperature and humidity levels and had free access to food and water. Interbreeding of heterozygous animals was performed to obtain wild-t ype (+/+), heterozygous (Tg/+ for tm1a and tm1b or +/- for tm1d), and homozygous/knockout (Tg/Tg for tm1a and tm1b or -/- for tm1d) littermates that were experimentally used, if not otherwise stated, between 3 and 9 months of age. Genotyping of tm1a and tm1b animals was performed using the following primers (Figure 1—figure supplement 1): F1 5’- CCGT GCCC AGCTA TAT GAGT -3’; R1 5’- CACA ACGG GTTC TTCTG TTA GTCC- 3’; R2 5’- CCAG CGGTG TTCC TTT CTTA -3’. Primers for tm1d genotyping were the following (Figure 1—figure supplement 1): F1, R2, and R4 5’-A CTG ATGG CGAG CTCAG ACC -3’. PCR amplification was performed for 36 cycles of 1 min at 95°C, 1 min at 56°C, and 1 min at 72°C. The PCR products were separated by electrophoresis on 2% agarose gels and visualized by ethidium bromide staining. Gene targeting and knockout mouse generation The Adgbtm1a(KOMP)Wtsi (tm1a) strain was generated by blastocyst microinjection of ESC clone EPD0707_3_ H06, provided by the KOMP (Skarnes et al., 2011). Correct targeting of the Adgb locus was verified prior to microinjection by long-r ange PCR using primers 5’S 5’-C TGT ACAC TGGT TGTA CACT GGTA CAAC TG-3’; 5’AS 5’- GGAC TAAC AGAA GAAC CCGT TGTG -3’; 3’S 5’- CACAC CTCC CCC TGAA CCTG AAAC- 3’; 3’AS 5’- GTAC TTGA TTGG ACGA TGAT CCAA G-3’ (Figure 1—figure supplement 1), gener- ating a band of 6.7 kb for 5’ primers and 5.1 kb for 3’ primers (Figure 1—figure supplement 1). Targeted clones were confirmed by Southern blot analysis using a hybridization probe that targets exon 13 (Figure 1—figure supplement 1) revealing a band of 4.1 kb (wild type) or 3.2 kb (tm1a allele) following digestion of genomic DNA with PvuII and a band of 2.8 kb (wild type) or 2.4 kb (tm1a allele) following digestion of genomic DNA with PstI (Figure 1—figure supplement 1). Chimeric mice were bred with C57BL/6-T yrc- Brd mice, and germline transmission in the F1 offspring was verified by PCR using primers F1, R1, and R2 (Figure 1—figure supplement 1). The mice were further bred to C57BL/6N- HprtTg(CMV- cre)Brd/Wtsi transgenic mice expressing the Cre allele to delete exons 13 and 14 and Keppner et al. eLife 2022;11:e72374. DOI: https://doi.org/10.7554/eLife.72374 16 of 26 Research article Cell Biology | Developmental Biology the neo cassette to generate the Adgbtm1b(KOMP)Wtsi strain (tm1b), or to C57BL/6N- Gt(ROSA)26Sortm1(FLP1) Dym/Wtsi transgenic mice expressing the Flp recombinase to delete the whole transgene cassette, thereby generating the Adgbtm1c(KOMP)Wtsi strain. The latter were further crossed with C57BL/6N-H prtTg(CMV- cre) Brd/Wtsi transgenic mice to delete exons 13 and 14, thereby generating the Adgbtm1d(KOMP)Wtsi (tm1d, knockout) strain. The Cre recombinase allele was bred out before any experiments were performed. RNA extraction and RT-qPCR Testes were frozen in liquid nitrogen and stored at –80°C. Tissues were homogenized using a Tissue- Lyser (Qiagen, Valencia, CA, USA). Subsequent RNA isolation and cDNA synthesis were performed as described previously (Keppner et al., 2019). In brief, RNA was extracted using an RNeasy Mini Kit (Qiagen) and reverse transcription was performed with 1.5 µg of total RNA and PrimeScript reverse transcriptase (Takara Bio Inc, Kusatsu, Japan). RT-q PCR was performed on a CFX96 C1000 real-t ime PCR cycler (Bio- Rad Laboratories, Hercules, CA) using SYBRgreen PCR master mix (Kapa Biosystems, London, UK). 21.5 ng of cDNA were loaded, and each sample was run as duplicate. mRNA levels were normalized to β-actin as previously described (De Backer et al., 2021). Primer sequences are displayed in Supplementary file 1. Cell culture and transfection HEK293 and A375 (ATCC CRL-1 619) cells were maintained in Dulbecco’s Minimum Essential Media (DMEM) (Gibco, Life Technologies, Carlsbad, CA, USA), containing L-g lutamine, supplemented with 10% heat- inactivated fetal bovine serum (FBS; PAN Biotech, Aidenbach, Germany) and 100 Units/ mL penicillin/100 μg/mL streptomycin (Gibco, Life Technologies, Carlsbad, CA, USA). Both cell lines were incubated in a humidified 5% CO2 atmosphere at 37°C and were routinely subcultured after trypsinization. For hypoxic experiments, cells were seeded out in six- well plates or 100 mm culture dishes. On the subsequent day, hypoxia experiments were carried out at 0.2% O2 and 5% CO2 in a gas- controlled glove box (InvivO2 400, Baker Ruskinn, Bridgend, UK) for 24 hr. Transfection of HEK293 cells was performed using calcium-p hosphate (Jordan et al., 1996), with 2 µg of plasmid DNA for regular immunoblotting experiments and 5 µg of plasmid DNA for immunoprecipitation experiments. Briefly, the DNA was diluted in sterile water and mixed with 250 mM CaCl2. 25 µM chloroquine was added to the cells and allowed to incubate for a minimum of 20 min. Prewarmed 37°C HBS buffer pH 7.05 (NaCl [280 mM], KCl [10 mM], Na2HPO4 [1.5 mM], D- glucose [12 mM], and HEPES [50 mM]) was added to the DNA solution (50% v/v), and the transfection mixture was added dropwise to the cells. The medium was replaced after 6 hr. Transfection of A375 cells was performed using JetOptimus (Polyplus- transfection SA, Illkirch-G rafffenstaden, France) according to the manufacturer’s instructions. Heme depletion To deprive HEK293T cells of intra- and extracellular heme, transfected HEK293T cells were cultured in DMEM medium supplemented with heme-d epleted FBS along with 3  mM succinylacetone to inhibit intracellular heme synthesis. Heme depletion in FBS was carried out by incubation with 10 mM ascorbic acid at 37°C for 7 hr, with subsequent dialysis in PBS for three times (Chen et al., 2012). This treatment was carried out for 24 hr prior to cell lysis and co-I P. SDS-PAGE and immunoblotting Tissues were homogenized, and proteins for immunoblotting, immunoprecipitation (IP), and proteomics were extracted as described (Keppner et al., 2019). Briefly, tissues were harvested and snap-f rozen in liquid nitrogen. The tissues were homogenized using a TissueLyser (Qiagen Valencia, CA, USA) in 1 mL protein extraction buffer (Tris-H Cl [50 mM], EDTA [1 mM], EGTA [1 mM], sucrose [0.27 mM], leupeptin [2 µg/mL], aprotinin [2 µg/mL], pepstatin [2 µg/mL], PMSF [1 mM], NaF [50 mM], Na- pyro- phosphate [5 mM], and Na3VO4 [1 mM]). The samples were centrifuged, the supernatant collected, and proteins quantified by Bradford assay. Cells were lysed in triton buffer (Tris- HCl [20 mM, pH 7.4], NaCl [150 mM], and triton X-1 00 [1%]), left on ice for 15 min, and centrifuged, and the proteins were quantified by Bradford assay. 25  µg of proteins were separated by SDS- PAGE on 10% gels, and proteins were electrotrans- ferred to nitrocellulose membranes (Amersham Hybond- ECL, GE Healthcare, Chicago, IL, USA). The membranes were incubated overnight at 4°C with primary antibody (Supplementary file 2) and for Keppner et al. eLife 2022;11:e72374. DOI: https://doi.org/10.7554/eLife.72374 17 of 26 Research article Cell Biology | Developmental Biology 1 hr with donkey anti- rabbit or anti- mouse IgG HRP- conjugated secondary antibody (1:5000, Amer- sham, Bukinghampshire, UK). All antibodies were diluted in TBS-t ween (1%) and dried milk (1%). The signal was revealed using ECL Prime (Amersham, Bukinghampshire, UK) on a C-D iGit Western blot scanner (LI- COR Biosciences) and quantified using ImageStudio program (LI-C OR Biosciences, Lincoln, NE, USA). The polyclonal anti-A dgb antibody was custom-m ade (Proteintech Group Inc, Rosemont, IL, USA). A fusion protein immunogen raised against the 409–745 amino acid region of mouse Adgb was used for the immunization of two rabbits over a period of 102 days. The antibodies in immune sera were affinity purified. Pre- bleeds, test bleeds, and purified antibodies were tested and validated by immunoblotting on wild- type and knockout testis extracts. Immunoprecipitation For immunoprecipitation (IP) of subsequent LC-M S/MS and immunoblotting analyses, 4 and 2 mg of proteins were used, respectively. The protein lysates were first pre-c leared for 24 hr at 4°C with protein G- sepharose beads (GE Healthcare, Chicago, IL, USA) coupled to rabbit IgG (Bethyl Labora- tories Inc, Montgomery, TX, USA). Samples were incubated overnight at 4°C with 2 µg primary anti- body (Supplementary file 2) or 2 µg rabbit IgG, followed by 4 hr with protein G- sepharose beads, then washed two times with wash buffer (Tris-H Cl [20 mM, pH 7.4], NaCl [300 mM], and LAP [1 mM]) and three times with equilibration buffer (Tris-H Cl [20 mM, pH 7.4], NaCl [150 mM], and LAP [1 mM]). Samples were eluted by boiling for 5 min at 95°C in 2× sample buffer and separated from the beads by centrifugation. LC-MS/MS analysis Washed IP beads were incubated with Laemmli sample buffer (Laemmli, 1970), and proteins were reduced with 1 mM DTT for 10 min at 75°C and alkylated using 5.5 mM iodoacetamide for 10 min at room temperature (RT). Protein samples were separated by SDS- PAGE on 4–12% gradient gels (ExpressPlus, Genscript, New Jersey, NJ, USA). Each gel lane was cut into six equal slices, the proteins were in- gel digested with trypsin (Promega, Madison, WI, USA), and the resulting peptide mixtures were processed on StageTips (Rappsilber et al., 2007; Shevchenko et al., 2006). LC-M S/MS measurements were performed on a Q Exactive Plus mass spectrometer (Thermo Fisher Scientific, Waltham, MA, USA) coupled to an EASY-n LC 1000 nanoflow HPLC (Thermo Fisher Scien- tific). HPLC-c olumn tips (fused silica) with 75 µm inner diameter were packed with Reprosil- Pur 120 C18- AQ, 1.9  µm (Dr. Maisch GmbH, Ammerbuch, Germany) to a length of 20  cm. A gradient of solvents A (0.1% formic acid in water) and B (0.1% formic acid in 80% acetonitrile in water) with increasing organic proportion was used for peptide separation (loading of sample with 0% B; separa- tion ramp: from 5 to 30% B within 85 min). The flow rate was 250 nL/min and for sample application 650 nL/min. The mass spectrometer was operated in the data-d ependent mode and switched auto- matically between MS (max. of 1 × 106 ions) and MS/MS. Each MS scan was followed by a maximum of 10 MS/MS scans using normalized collision energy of 25% and a target value of 1000. Parent ions with a charge state form z=1, and unassigned charge states were excluded from fragmentation. The mass range for MS was m/z=370–1750. The resolution for MS was set to 70,000 and for MS/MS to 17,500. MS parameters were as follows: spray voltage 2.3 kV; no sheath and auxiliary gas flow; ion- transfer tube temperature 250°C. The MS raw data files were uploaded into the MaxQuant software version 1.6.2.10 for peak detection, generation of peak lists of mass error corrected peptides, and for database searches (Tyanova et al., 2016). A full- length UniProt mouse (based on UniProt FASTA version April 2016) or human database (UniProt FASTA version March 2016) additionally containing common contaminants, such as keratins and enzymes used for in-g el digestion, was used as reference. Carbamidomethylcysteine was set as fixed modification and protein amino- terminal acetylation and oxidation of methionine were set as variable modifications. Three missed cleavages were allowed, enzyme specificity was trypsin/P, and the MS/MS tolerance was set to 20 ppm. The average mass precision of identified peptides was in general less than 1 ppm after recalibration. Peptide lists were further used by MaxQuant to identify and relatively quantify proteins using the following parameters: peptide and protein false discovery rates (FDRs), based on a forward- reverse database, were set to 0.01, minimum peptide length was set to 7, minimum number of peptides for identification and quan- titation of proteins was set to one which must be unique. The ‘match-b etween- run’ option (0.7 min) was used. Keppner et al. eLife 2022;11:e72374. DOI: https://doi.org/10.7554/eLife.72374 18 of 26 Research article Cell Biology | Developmental Biology Propidium iodide staining and flow cytometry Preparation of germ cell suspensions was achieved as described (Jeyaraj et al., 2003). Briefly, decap- sulated testes were incubated in 0.5 mg/mL collagenase type IV in PBS, washed with PBS, and incu- bated in 1 µg/mL DNase and 1 µg/mL trypsin. Soybean trypsin inhibitor was added, the suspension was filtered, washed in PBS, fixed with 70% ethanol, and stored at 4°C. DNA staining using propidium iodide was performed as described (Krishnamurthy et al., 2000). Propidium iodide- stained cells were analyzed in a FACScan flow cytometer (Becton-D ickinson Immunocytometry, San Jose, CA, USA). The fluorescent signals of propidium iodide-s tained cells were recorded, and a cytogram of DNA area vs. cell count was used to select cell populations on the basis of their DNA content. A total of 10,000 events was recorded for each histogram. Cell populations were selected based on their DNA content, and their relative numbers were calculated using Summit (Cytomation, CO, USA). Gating was performed as such that quantification represents the ratio of cell size to propidium iodide stainable nuclear DNA content. Histological analyses and immunofluorescence Testes were fixed in 4% paraformaldehyde (PFA) and embedded in paraffin. Preparation of sections and H&E staining was performed as described (Keppner et al., 2015). Pictures were taken using a Nikon Eclipse microscope (Nikon Corporation, Tokyo, Japan). For immunofluorescence, testes were fixed in 4% PFA for at least 1 week and incubated in 30% sucrose for another week. The testes were embedded in Optimal Cutting Temperature compound (O.C.T. Tissue- Tek, Sakura Finetek, Tokyo, Japan), and 5 µm thick sections were cut using a cryotome. For seminiferous tubule dissections and stainings, slides were prepared as previously described (Kotaja et al., 2004). For sperm stainings, cauda epididymal sperm was retrieved and diluted in PBS. A drop of the suspension was smeared on glass slides and fixed by drying for 15 min and by 4% PFA for 20 min. The slides were blocked in 10% normal goat serum and 0.5% triton X- 100 for 1 hr. Testis sections and sperm slides were incubated overnight at 4°C with primary antibodies (Supplementary file 2) in 5% normal goat serum and 0.25% triton X- 100. The slides were washed with PBS (3 × 10 min), incubated with secondary Alexa Fluor 488 or 594 coupled goat anti- mouse or anti- rabbit IgG (1:300, Invitrogen, Waltham, MA, USA) for 1 hr, washed again with PBS (3 × 10 min), and counterstained with Sudan Black (0.1%) for autofluorescence quenching. Slides were mounted with fluoromount mounting medium containing DAPI (Southern- Biotech, Birmingham, AL, USA) and visualized using a Nikon Eclipse fluorescent microscope (Nikon Corporation). In situ mRNA hybridization RNAscope in situ hybridization was performed using BaseScope Detection Reagent Kit v2 RED (Advanced Cell Diagnostics Inc, Newark, CA, USA, Cat. No. 323900) according to the manufacturer’s instructions. H2O2 treatment, antigen retrieval, and protease treatment were performed on 5 µm- thick sections prior to hybridization with probes for Adgb (BA- Mm- Adgb- 3zz-s t, Advanced Cell Diagnos- tics, Cat. No. 862141), DapB as negative control (BA-D apB-3 zz, Advances Cell Diagnostics, Cat. No. 701011), and Ppib as positive control (Ba-M m- Ppib-3 zz, Advanced Cell Diagnostics, Cat. No. 701071) at 40°C for 2 hr followed by eight amplification steps. The signal was revealed with Fast Red, and the sections were counterstained with Gill’s hematoxylin no. 1 and mounted with VectaMount (Vector Laboratories, Burlingame, CA, USA). The sections were visualized using a light microscope. Electron microscopy All electron microscopy experiments were performed at the Electron Microscopy Platform of the University of Lausanne, Switzerland. Mouse testes were fixed in 2.5% glutaraldehyde solution (EMS, Hatfield, PA, USA) in phosphate buffer (PB 0.1 M [pH 7.4]) for 1 hr at RT and post-fi xed in a fresh mixture of osmium tetroxide 1% (EMS) with 1.5% of potassium ferrocyanide (Sigma, St. Louis, MO, USA) in PB buffer for 1 hr at RT. The samples were then washed twice in distilled water and dehy- drated in acetone solution (Sigma) at graded concentrations (30%–40 min; 50%–40 min; 70%–40 min; 100%–2 × 1 hr). This was followed by infiltration in Epon resin (EMS, Hatfield, PA, USA) at graded concentrations (Epon 33% in acetone- 4 hr; Epon 66% in acetone- 4 hr; Epon 100%–2 × 8 hr) and finally polymerized for 48 hr at 60°C in an oven. Ultrathin sections of 50 nm were cut using a Leica Ultracut (Leica Mikrosysteme GmbH, Vienna, Austria), picked up on a copper slot grid of 2 × 1 mm (EMS, Keppner et al. eLife 2022;11:e72374. DOI: https://doi.org/10.7554/eLife.72374 19 of 26 Research article Cell Biology | Developmental Biology Hatfield, PA, USA), and coated with a polystyrene film (Sigma, St Louis, MO, USA). Sections were post- stained with uranyl acetate (Sigma, St. Louis, MO, USA) 4% in H2O for 10 min, rinsed several times with H2O followed by Reynolds lead citrate in H2O (Sigma, St Louis, MO, USA) for 10 min, and rinsed several times with H2O. Micrographs were taken with a TEM FEI CM100 (FEI, Eindhoven, The Netherlands) at an acceleration voltage of 80 kV with a TVIPS TemCamF416 digital camera (TVIPS GmbH, Gauting, Germany). RNAseq library preparation and transcriptome sequencing Total RNA from two independent samples of wild-t ype and Adgb-/- testis was extracted using the mirVana miRNA- Kit according to manufacturer’s instructions (Life Technologies, Carlsbad, USA). Prior to library construction, RNA quality was assessed using an Agilent 2100 Bioanalyzer and the Agilent RNA 6000 Nano Kit (Agilent Technologies, Santa Clara, CA, USA). RNA was quantified using Qubit RNA BR Assay Kit (Invitrogen, Waltham, MA, USA). Libraries were prepared starting from 1000 ng of total RNA using the RNA Sample Prep Kit v2 (Illumina Inc, San Diego, CA, USA) including a poly-A selection step following the manufacturer’s instructions and sequenced as 2 × 100 nt paired-e nd reads using an Illumina HiSeq 2500. Library preparation and sequencing were performed by the NGS Core Facility of the Department of Biology, Johannes- Gutenberg University (Mainz, Germany). RNAseq data are available from the European Nucleotide Archive under accession number PRJEB46499. Differential gene expression, GO term annotation, and pathway enrichment analyses Raw sequences were pre- processed to remove low quality reads and residual Illumina adapter sequences using BBduk from the BBtools suite (https://sourceforge.net/projects/bbmap/). The overall sequencing quality and the absence of adapter contamination were evaluated with FastQC. Mapping was performed with HISAT2, and quantification of gene expression was done using StringTie. Differ- entially expressed genes were determined using DESeq2. Genes were considered differentially expressed when presenting |fold change|>2 and FDR-c orrected p-v alue≤0.1. GO term enrichment analyses were performed using WebGestalt 2019 using the Overrepresentation Enrichment Analysis method, requiring a BH-c orrected p-v alue≤0.05 and a minimum enrichment of four genes for term/ pathway. Enrichment in canonical pathways were performed with Qiagen’s Ingenuity Pathway Analysis (IPA, Qiagen, Hilden, Germany), Core analysis tool using bias- corrected z-s core (when applicable), and BH- corrected p-v alues≤0.05. Cloning and construction of expression plasmids Generation of pLenti6-A DGB was described before (Bracke et  al., 2018), pLenti6- SEPT10- V5 was purchased from DNASU (clone ID HsCD00943271, DNASU Plasmid Repository, Arizona State University, AZ, USA). All additional recombinant genes were cloned into pFLAG-C MVTM- 6a expres- sion vector (Sigma) unless otherwise specified. All coding sequences were amplified by PCR using Phusion High- Fidelity DNA polymerase (Thermo Fisher Scientific). Recombinant ADGB, SEPT10, and SEPT7 with N-t erminal FLAG tag and C- terminal myc tag were constructed by amplifying and ligating their respective coding sequence with in- primer designed myc tag (EQKLISEEDL) into the expression vector, in- frame with the N-t erminal FLAG tag. A glycine-s erine (GSG) linker was added between the last codon of SEPT10 and the first codon of the myc tag. SEPT2, SEPT11, and SEPT12 expression vectors were cloned by amplifying and ligating their respective coding sequence into the pFLAG- CMV- 6a expression vector containing an N-t erminal FLAG tag. Truncated ADGB proteins consisting of the calpain- like domain, 350- residue uncharacterized domain and globin domain (N-t erminal mutant), or the 700- residue uncharacterized region (C-t erminal mutant) domain were designed with GFP tags at both N- and C- termini. The N-t erminal mutant ADGB was amplified between codons of residues 58 and 968 in ADGB, while the C-t erminal mutant ADGB was amplified between codons of residues 969 and 1667. Amplicons were designed with 5’- and 3’- overhangs compatible with two customized GFP amplicons designed to anneal at the 5’- and 3’-ends of the genes. GSG linkers (GGSGGGGSGG) were added to bridge the GFP tags and the truncated ADGB proteins. Similarly, N- terminally GFP- tagged isolated ADGB globin domain was cloned by amplifying and ligating amplicons of the ADGB globin coding sequence downstream to a GFP coding sequence with complementary overhangs, with the GSG linker added between the two proteins. With the same construction, GFP-t agged ADGB Keppner et al. eLife 2022;11:e72374. DOI: https://doi.org/10.7554/eLife.72374 20 of 26 Research article Cell Biology | Developmental Biology globin domains with a single mutation on the proximal histidine in helix F codon 8 (H824G) or the distal glutamine in helix E codon 7 (Q792G), or both (H824G/Q792G) were cloned by amplifying and ligating the globin domain coding sequence using primers designed to carry the mutated codon sequence. ADGB-ΔIQ, ADGB-Δcalpain, and ADGB-ΔCCD were constructed by amplifying designed ADGB amplicons with compatible overhangs and were ligated in-f rame to generate an ADGB coding sequence with deletions in the desired domains. For M2H assays, Gal4-C aM, Gal4-A DGB, VP16- globin domain, VP16- CaM, or VP16- ADGB were cloned into a pcDNA3.0 expression vector. Gal4 DNA- binding domain or VP16 transactivation domain sequences were amplified with the in-p rimer designed GSG linker at the 3’-end of the amplicons. The coding sequence of CALM3, ADGB globin domain, and ADGB full length were amplified with complementary 5’-end and ligated to the Gal4 and VP16 sequences to generate the fusion genes. Reporter gene assays For M2H assays, 2.15 × 105 HEK293 or 4 × 105 A375 cells were transiently transfected with 1 µg firefly luciferase reporter plasmid (5xGAL4-T ATA- luciferase, Addgene, 46756) (Sun et al., 1994) and 500 ng or 200 and 300 ng chimeric Gal4 and VP16 fusion protein vectors, respectively, in 12- or 6-w ell format using CaCl2 or JetOptimus. To control for differences in transfection efficiency and extract preparation, 25 ng or 50 ng pRL- SV40 Renilla luciferase reporter vector (Promega, Madison, WI, USA) was co-t ransfected, respectively for HEK293 and A375 cells. Cultures were evenly split onto 12-w ell plates 24 hr after transfection for A375 cells. For hypoxia control experiments, 4 × 105 A375 cells were transiently co- transfected with 500 ng firefly 5’/3’-hypoxia response element- dependent EPO promoter- driven luciferase reporter plasmid (Storti et al., 2014) and 50 ng pRL-S V40 Renilla lucif- erase reporter vector. Luciferase activities of duplicate wells were determined using the Dual Lucif- erase Reporter Assay System (Promega) as described before (Schörg et al., 2015). Reporter activities were expressed as relative firefly/Renilla luciferase activities. All reporter gene assays were performed at least three times independently. Testosterone quantification For intra- testis testosterone quantification, testis samples were homogenized twice using the Precellys 24 tissue homogenizer (Bertin Instruments; Rockville, MD, USA) (4°C, 3×, 30 s at 6500 rpm, cycle break 30 s) in chloroform- isopropanol (1 mL, 50/50%; v/v) containing ISTD. Combined super- natants were centrifuged (10  min, RT, 16,000  × g) and evaporated using a Genevac EZ- 2 evapo- rator (Stepbios, Muttenz, Switzerland) (3 hr, 35°C). Samples were reconstituted in methanol (50 µL, 10 min, RT, 1300 rpm) and sonicated (10 min, RT). Reconstituted samples were centrifuged (10 min, RT, 16,000 × g), and supernatants were transferred to LC-M S vials. Testosterone content was analyzed by ultra- performance liquid chromatography-M S/MS (UPLC- MS/MS) using an Agilent 1290 Infinity II UPLC coupled to an Agilent 6495 triple quadrupole mass spectrometer equipped with a jet- stream electrospray ionization interface (Agilent Technologies). Analyte separation was achieved using a reverse- phase column (1.7 µm, 2.1 mm × 150 mm; Acquity UPLC BEH C18; Waters). Data acquisi- tion and quantitative analysis were performed by MassHunter (Version B.10.0. Build 10.0.27, Agilent Technologies). TUNEL assay TUNEL assay was performed on paraffin- embedded testis sections using the DeadEnd Fluorometric TUNEL System according to manufacturer’s instructions (Promega, Madison, WI, USA). The positive control sections were pre-i ncubated with DNase I solution (Qiagen, Valencia, CA, USA), while negative control sections were incubated in incubation buffer without recombinant terminal deoxynucleotidyl transferase (rTdT enzyme). Slides were mounted with fluoromount mounting medium containing DAPI (SouthernBiotech, Birmingham, AL, USA) and visualized using a Nikon Eclipse fluorescent microscope (Nikon Corporation). 10 pictures were taken for each animal and TUNEL- positive cells were counted using FIJI software (Schindelin et al., 2012). ELISA quantification of LH and FSH Serum LH and FSH were quantified by ELISA according to manufacturer’s instructions using 50 µL of sample (Cloud- Clone Corp., Houston, TX, USA). Keppner et al. eLife 2022;11:e72374. DOI: https://doi.org/10.7554/eLife.72374 21 of 26 Research article Cell Biology | Developmental Biology Statistical analysis All values were presented as mean ± SEM. Differences in means between two groups were analyzed with unpaired two- tailed Student’s t- test (Figure 1A, C, D and E; Figure 6H right graph; Figure 1— figure supplement 2A,B Figure 1—figure supplement 3A-G ,I Figure 4—figure supplement 1H-N Figure  6—figure supplement 5A,B,C) and those among multiple groups with one-w ay ANOVA followed by Tukey post- hoc test (Figure 2D; Figure 6D, G and H left graph, Figure 6—figure supple- ment 2B). All statistics were performed with GraphPad Prism software 7.05. Values of p≤0.05 were considered statistically significant. Acknowledgements We thank Christine Roulin for technical assistance. We thank Damien De Bellis from the Electron Micros- copy Platform of the University of Lausanne for EM section preparation and image acquisition. This work was supported by the Swiss National Science Foundation to DH (grants 31,003 A_173000 and 310030_207460) and the German Research Foundation to DH (HO 5837/1–1) and TH (HA 2103/9–1). Additional information Funding Funder Grant reference number Author Schweizerischer 31003A_173000 David Hoogewijs Nationalfonds zur Förderung der Wissenschaftlichen Forschung Schweizerischer 310030_207460 David Hoogewijs Nationalfonds zur Förderung der Wissenschaftlichen Forschung Deutsche HO 5837/1-1 David Hoogewijs Forschungsgemeinschaft Deutsche HA 2103/9-1 Thomas Hankeln Forschungsgemeinschaft The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication. Author contributions Anna Keppner, Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Visu- alization, Writing – original draft; Miguel Correia, Formal analysis, Investigation, Visualization; Sara Santambrogio, Teng Wei Koay, Darko Maric, Denise V Winter, Frédéric Chalmel, Formal analysis, Inves- tigation, Methodology; Carina Osterhof, Formal analysis, Investigation, Methodology, Visualization; Angèle Clerc, Investigation; Michael Stumpe, Dieter Kressler, Data curation, Formal analysis, Meth- odology; Sylvia Dewilde, Formal analysis; Alex Odermatt, Formal analysis, Methodology, Resources; Thomas Hankeln, Data curation, Formal analysis, Funding acquisition, Methodology; Roland H Wenger, Formal analysis, Investigation, Resources, Writing - review and editing; David Hoogewijs, Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Supervision, Writing – original draft Author ORCIDs Carina Osterhof http://orcid.org/0000-0002-1699-7410 Michael Stumpe http://orcid.org/0000-0002-9443-9326 Dieter Kressler http://orcid.org/0000-0003-4855-3563 David Hoogewijs http://orcid.org/0000-0001-5547-6004 Keppner et al. eLife 2022;11:e72374. DOI: https://doi.org/10.7554/eLife.72374 22 of 26 Research article Cell Biology | Developmental Biology Ethics All experimental procedures and animal maintenance followed Swiss federal guidelines and the study was revised and approved by the "Service de la sécurité alimentaire et des affaires vétérinaires"(SAAV) of the canton of Fribourg, Switzerland (license number 2017_16_FR). Decision letter and Author response Decision letter https://doi.org/10.7554/eLife.72374.sa1 Author response https://doi.org/10.7554/eLife.72374.sa2 Additional files Supplementary files • Transparent reporting form • Supplementary file 1. List of primers used for RT- qPCR. • Supplementary file 2. List of antibodies used throughout the study. Data availability RNA- sequencing data have been submitted to ENA with accession number PRJEB46499 and is also available as supplemental dataset 1 (excel table). All data generated or analysed during this study are included in the manuscript and supporting files. Source data files are provided for Figures 1, 2, 4B, 4C, 4D, 4E, 6A, 6B, 6C, 6D, 6E, 6F, Fig. 1-fi g. suppl. 1, Fig. 4-fi g. suppl. 1A, B, C, D, E, F, G, Fig. 4-fi g. suppl. 2A, B, C, Fig. 4- fig. suppl. 3A, B, C, D, E, Fig. 4- fig. suppl. 5, Fig. 4- fig. suppl. 6B, Fig. 6- fig. suppl. 1A, Fig. 6- fig. suppl. 4, Fig. 6-fi g. suppl. 5D, E. References Andrews C, Xu Y, Kirberger M, Yang JJ. 2020. Structural Aspects and Prediction of Calmodulin- Binding Proteins. International Journal of Molecular Sciences 22:E308. DOI: https://doi.org/10.3390/ijms22010308, PMID: 33396740 Avidor- Reiss T, Ha A, Basiri ML. 2017. Transition Zone Migration: A Mechanism for Cytoplasmic Ciliogenesis and Postaxonemal Centriole Elongation. Cold Spring Harbor Perspectives in Biology 9:a028142. DOI: https://doi. org/10.1101/cshperspect.a028142, PMID: 28108487 Avidor- Reiss T, Carr A, Fishman EL. 2020. The sperm centrioles. Molecular and Cellular Endocrinology 518:110987. DOI: https://doi.org/10.1016/j.mce.2020.110987, PMID: 32810575 Bähler M, Rhoads A. 2002. Calmodulin signaling via the IQ motif. FEBS Letters 513:107–113. DOI: https://doi. org/10.1016/s0014-5793(01)03239-2, PMID: 11911888 Bracke A, Hoogewijs D, Dewilde S. 2018. Exploring three different expression systems for recombinant expression of globins: Escherichia coli, Pichia pastoris and Spodoptera frugiperda. Analytical Biochemistry 543:62–70. DOI: https://doi.org/10.1016/j.ab.2017.11.027, PMID: 29203135 Brown JM, Witman GB. 2014. Cilia and Diseases. Bioscience 64:1126–1137. DOI: https://doi.org/10.1093/biosci/ biu174, PMID: 25960570 Chen CY, Samuel TK, Krause M, Dailey HA, Hamza I. 2012. Heme utilization in the Caenorhabditis elegans hypodermal cells is facilitated by heme- responsive gene- 2. The Journal of Biological Chemistry 287:9601– 9612. DOI: https://doi.org/10.1074/jbc.M111.307694, PMID: 22303006 Darde TA, Sallou O, Becker E, Evrard B, Monjeaud C, Le Bras Y, Jégou B, Collin O, Rolland AD, Chalmel F. 2015. The ReproGenomics Viewer: an integrative cross- species toolbox for the reproductive science community. Nucleic Acids Research 43:W109–W116. DOI: https://doi.org/10.1093/nar/gkv345, PMID: 25883147 Darde TA, Lecluze E, Lardenois A, Stévant I, Alary N, Tüttelmann F, Collin O, Nef S, Jégou B, Rolland AD, Chalmel F, Kelso J. 2019. The ReproGenomics Viewer: a multi- omics and cross-s pecies resource compatible with single- cell studies for the reproductive science community. Bioinformatics (Oxford, England) 35:3133– 3139. DOI: https://doi.org/10.1093/bioinformatics/btz047, PMID: 30668675 Darszon A, Nishigaki T, Beltran C, Treviño CL. 2011. Calcium channels in the development, maturation, and function of spermatozoa. Physiological Reviews 91:1305–1355. DOI: https://doi.org/10.1152/physrev.00028. 2010, PMID: 22013213 De Backer J, Maric D, Bosman M, Dewilde S, Hoogewijs D. 2021. A reliable set of reference genes to normalize oxygen- dependent cytoglobin gene expression levels in melanoma. Scientific Reports 11:10879. DOI: https:// doi.org/10.1038/s41598-021-90284-6, PMID: 34035373 Dolat L, Hu Q, Spiliotis ET. 2014. Septin functions in organ system physiology and pathology. Biological Chemistry 395:123–141. DOI: https://doi.org/10.1515/hsz-2013-0233, PMID: 24114910 Egri SB, Ouch C, Chou HT, Yu Z, Song K, Xu C, Shen K. 2022. Cryo- EM structures of the human GATOR1- Rag- Ragulator complex reveal a spatial- constraint regulated GAP mechanism. Molecular Cell 82:1836-1849.. DOI: https://doi.org/10.1016/j.molcel.2022.03.002, PMID: 35338845 Keppner et al. eLife 2022;11:e72374. DOI: https://doi.org/10.7554/eLife.72374 23 of 26 Research article Cell Biology | Developmental Biology Ermolova N, Kramerova I, Spencer MJ. 2015. Autolytic activation of calpain 3 proteinase is facilitated by calmodulin protein. The Journal of Biological Chemistry 290:996–1004. DOI: https://doi.org/10.1074/jbc.M114. 588780, PMID: 25389288 Fainberg J, Kashanian JA. 2019. Recent advances in understanding and managing male infertility. F1000Research 8:F1000. DOI: https://doi.org/10.12688/f1000research.17076.1, PMID: 31143441 Ghossoub R, Hu Q, Failler M, Rouyez M-C , Spitzbarth B, Mostowy S, Wolfrum U, Saunier S, Cossart P, Jamesnelson W, Benmerah A. 2013. Septins 2, 7 and 9 and MAP4 colocalize along the axoneme in the primary cilium and control ciliary length. Journal of Cell Science 126:2583–2594. DOI: https://doi.org/10.1242/jcs. 111377, PMID: 23572511 Green CD, Ma Q, Manske GL, Shami AN, Zheng X, Marini S, Moritz L, Sultan C, Gurczynski SJ, Moore BB, Tallquist MD, Li JZ, Hammoud SS. 2018. A Comprehensive Roadmap of Murine Spermatogenesis Defined by Single- Cell RNA- Seq. Developmental Cell 46:651–667. DOI: https://doi.org/10.1016/j.devcel.2018.07.025, PMID: 30146481 Hermo L, Pelletier RM, Cyr DG, Smith CE, Pelletier RM. 2010a. Surfing the wave, cycle, life history, and genes/ proteins expressed by testicular germ cells. Part 1: background to spermatogenesis, spermatogonia, and spermatocytes. Microscopy Research and Technique 73:241–278. DOI: https://doi.org/10.1002/jemt.20783, PMID: 19941293 Hermo L, Pelletier RM, Cyr DG, Smith CE, Pelletier RM. 2010b. Surfing the wave, cycle, life history, and genes/ proteins expressed by testicular germ cells. Part 2: changes in spermatid organelles associated with development of spermatozoa. Microscopy Research and Technique 73:279–319. DOI: https://doi.org/10.1002/ jemt.20787, PMID: 19941292 Hermo L, Pelletier RM, Cyr DG, Smith CE, Pelletier RM. 2010c. Surfing the wave, cycle, life history, and genes/ proteins expressed by testicular germ cells. Part 3: developmental changes in spermatid flagellum and cytoplasmic droplet and interaction of sperm with the zona pellucida and egg plasma membrane. Microscopy Research and Technique 73:320–363. DOI: https://doi.org/10.1002/jemt.20784, PMID: 19941287 Hermo L, Pelletier RM, Cyr DG, Smith CE, Pelletier RM. 2010d. Surfing the wave, cycle, life history, and genes/ proteins expressed by testicular germ cells. Part 4: intercellular bridges, mitochondria, nuclear envelope, apoptosis, ubiquitination, membrane/voltage-g ated channels, methylation/acetylation, and transcription factors. Microscopy Research and Technique 73:364–408. DOI: https://doi.org/10.1002/jemt.20785, PMID: 19941288 Hermo L, Pelletier RM, Cyr DG, Smith CE, Pelletier RM. 2010e. Surfing the wave, cycle, life history, and genes/ proteins expressed by testicular germ cells. Part 5: intercellular junctions and contacts between germs cells and Sertoli cells and their regulatory interactions, testicular cholesterol, and genes/proteins associated with more than one germ cell generation. Microscopy Research and Technique 73:409–494. DOI: https://doi.org/10.1002/ jemt.20786, PMID: 19941291 Hoogewijs D, Ebner B, Germani F, Hoffmann FG, Fabrizius A, Moens L, Burmester T, Dewilde S, Storz JF, Vinogradov SN, Hankeln T. 2012. Androglobin: A Chimeric Globin in Metazoans That Is Preferentially Expressed in Mammalian Testes. Molecular Biology and Evolution 29:1105–1114. DOI: https://doi.org/10.1093/ molbev/msr246, PMID: 22115833 Hou S, Freitas T, Larsen RW, Piatibratov M, Sivozhelezov V, Yamamoto A, Meleshkevitch EA, Zimmer M, Ordal GW, Alam M. 2001. Globin-c oupled sensors: A class of heme-c ontaining sensors in Archaea and Bacteria. PNAS 98:9353–9358. DOI: https://doi.org/10.1073/pnas.161185598, PMID: 11481493 Hu Q, Milenkovic L, Jin H, Scott MP, Nachury MV, Spiliotis ET, Nelson WJ. 2010. A septin diffusion barrier at the base of the primary cilium maintains ciliary membrane protein distribution. Science (New York, N.Y.) 329:436– 439. DOI: https://doi.org/10.1126/science.1191054, PMID: 20558667 Ihara M, Kinoshita A, Yamada S, Tanaka H, Tanigaki A, Kitano A, Goto M, Okubo K, Nishiyama H, Ogawa O, Takahashi C, Itohara S, Nishimune Y, Noda M, Kinoshita M. 2005. Cortical organization by the septin cytoskeleton is essential for structural and mechanical integrity of mammalian spermatozoa. Developmental Cell 8:343–352. DOI: https://doi.org/10.1016/j.devcel.2004.12.005, PMID: 15737930 Jan SZ, Vormer TL, Jongejan A, Röling M, Silber SJ, de Rooij DG, Hamer G, Repping S, van Pelt AMM. 2017. Unraveling transcriptome dynamics in human spermatogenesis. Development (Cambridge, England) 144:3659– 3673. DOI: https://doi.org/10.1242/dev.152413, PMID: 28935708 Jégou B, Sankararaman S, Rolland AD, Reich D, Chalmel F. 2017. Meiotic Genes Are Enriched in Regions of Reduced Archaic Ancestry. Molecular Biology and Evolution 34:1974–1980. DOI: https://doi.org/10.1093/ molbev/msx141, PMID: 28444387 Jeyaraj DA, Grossman G, Petrusz P. 2003. Dynamics of testicular germ cell apoptosis in normal mice and transgenic mice overexpressing rat androgen- binding protein. Reproductive Biology and Endocrinology 1:48. DOI: https://doi.org/10.1186/1477-7827-1-48, PMID: 12831397 Jordan M, Schallhorn A, Wurm FM. 1996. Transfecting Mammalian Cells: Optimization of Critical Parameters Affecting Calcium- Phosphate Precipitate Formation. Nucleic Acids Research 24:596–601. DOI: https://doi.org/ 10.1093/nar/24.4.596, PMID: 8604299 Kaur P, Nagar S, Bhagwat M, Uddin M, Zhu Y, Vancurova I, Vancura A, Singh PK. 2021. Activated heme synthesis regulates glycolysis and oxidative metabolism in breast and ovarian cancer cells. PLOS ONE 16:e0260400. DOI: https://doi.org/10.1371/journal.pone.0260400 Keppner Anna, Andreasen D, Mérillat A-M , Bapst J, Ansermet C, Wang Q, Maillard M, Malsure S, Nobile A, Hummler E, Cheng X. 2015. Epithelial Sodium Channel-M ediated Sodium Transport Is Not Dependent on the Keppner et al. eLife 2022;11:e72374. DOI: https://doi.org/10.7554/eLife.72374 24 of 26 Research article Cell Biology | Developmental Biology Membrane- Bound Serine Protease CAP2/Tmprss4. PLOS ONE 10:e0135224. DOI: https://doi.org/10.1371/ journal.pone.0135224 Keppner A, Maric D, Sergi C, Ansermet C, De Bellis D, Kratschmar DV, Canonica J, Klusonova P, Fenton RA, Odermatt A, Crambert G, Hoogewijs D, Hummler E. 2019. Deletion of the serine protease CAP2/Tmprss4 leads to dysregulated renal water handling upon dietary potassium depletion. Scientific Reports 9:19540. DOI: https://doi.org/10.1038/s41598-019-55995-x, PMID: 31863073 Keppner A, Maric D, Correia M, Koay TW, Orlando IMC, Vinogradov SN, Hoogewijs D. 2020. Lessons from the post- genomic era: Globin diversity beyond oxygen binding and transport. Redox Biology 37:101687. DOI: https://doi.org/10.1016/j.redox.2020.101687 Khawar MB, Gao H, Li W. 2019. Mechanism of Acrosome Biogenesis in Mammals. Frontiers in Cell and Developmental Biology 7:195. DOI: https://doi.org/10.3389/fcell.2019.00195, PMID: 31620437 Kissel H, Georgescu MM, Larisch S, Manova K, Hunnicutt GR, Steller H. 2005. The Sept4 septin locus is required for sperm terminal differentiation in mice. Developmental Cell 8:353–364. DOI: https://doi.org/10.1016/j. devcel.2005.01.021, PMID: 15737931 Koay TW, Osterhof C, Orlando IMC, Keppner A, Andre D, Yousefian S, Suárez Alonso M, Correia M, Markworth R, Schödel J, Hankeln T, Hoogewijs D. 2021. Androglobin gene expression patterns and FOXJ1- dependent regulation indicate its functional association with ciliogenesis. The Journal of Biological Chemistry 296:100291. DOI: https://doi.org/10.1016/j.jbc.2021.100291, PMID: 33453283 Kotaja N, Kimmins S, Brancorsini S, Hentsch D, Vonesch JL, Davidson I, Parvinen M, Sassone- Corsi P. 2004. Preparation, isolation and characterization of stage-s pecific spermatogenic cells for cellular and molecular analysis. Nature Methods 1:249–254. DOI: https://doi.org/10.1038/nmeth1204-249, PMID: 16144087 Krishnamurthy H, Danilovich N, Morales CR, Sairam MR. 2000. Qualitative and quantitative decline in spermatogenesis of the follicle-s timulating hormone receptor knockout (FORKO) mouse. Biology of Reproduction 62:1146–1159. DOI: https://doi.org/10.1095/biolreprod62.5.1146, PMID: 10775161 Kuo YC, Lin YH, Chen HI, Wang YY, Chiou YW, Lin HH, Pan HA, Wu CM, Su SM, Hsu CC, Kuo PL. 2012. SEPT12 mutations cause male infertility with defective sperm annulus. Human Mutation 33:710–719. DOI: https://doi. org/10.1002/humu.22028, PMID: 22275165 Kwak S, Jung MW. 2019. Distinct roles of striatal direct and indirect pathways in value- based decision making. eLife 8:e46050. DOI: https://doi.org/10.7554/eLife.46050, PMID: 31310237 Laemmli UK. 1970. Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 227:680–685. DOI: https://doi.org/10.1038/227680a0, PMID: 5432063 Lin YH, Lin YM, Wang YY, Yu IS, Lin YW, Wang YH, Wu CM, Pan HA, Chao SC, Yen PH, Lin SW, Kuo PL. 2009. The expression level of septin12 is critical for spermiogenesis. The American Journal of Pathology 174:1857–1868. DOI: https://doi.org/10.2353/ajpath.2009.080955, PMID: 19359518 Lin YH, Huang CY, Ke CC, Wang YY, Lai TH, Liu HC, Ku WC, Chan CC, Lin YH. 2020. ACTN4 Mediates SEPT14 Mutation- Induced Sperm Head Defects. Biomedicines 8:E518. DOI: https://doi.org/10.3390/ biomedicines8110518, PMID: 33228246 Lukassen S, Bosch E, Ekici AB, Winterpacht A. 2018. Characterization of germ cell differentiation in the male mouse through single- cell RNA sequencing. Scientific Reports 8:6521. DOI: https://doi.org/10.1038/s41598- 018-24725-0, PMID: 29695820 Mecklenburg JM, Hermann BP. 2016. Mechanisms Regulating Spermatogonial Differentiation. Results and Problems in Cell Differentiation 58:253–287. DOI: https://doi.org/10.1007/978-3-319-31973-5_10, PMID: 27300182 Neto FTL, Bach PV, Najari BB, Li PS, Goldstein M. 2016. Spermatogenesis in humans and its affecting factors. Seminars in Cell & Developmental Biology 59:10–26. DOI: https://doi.org/10.1016/j.semcdb.2016.04.009, PMID: 27143445 Osako Y, Maemoto Y, Tanaka R, Suzuki H, Shibata H, Maki M. 2010. Autolytic activity of human calpain 7 is enhanced by ESCRT- III- related protein IST1 through MIT- MIM interaction. The FEBS Journal 277:4412–4426. DOI: https://doi.org/10.1111/j.1742-4658.2010.07822.x, PMID: 20849418 Platts AE, Dix DJ, Chemes HE, Thompson KE, Goodrich R, Rockett JC, Rawe VY, Quintana S, Diamond MP, Strader LF, Krawetz SA. 2007. Success and failure in human spermatogenesis as revealed by teratozoospermic RNAs. Human Molecular Genetics 16:763–773. DOI: https://doi.org/10.1093/hmg/ddm012, PMID: 17327269 Randriamboavonjy V, Isaak J, Elgheznawy A, Pistrosch F, Frömel T, Yin X, Badenhoop K, Heide H, Mayr M, Fleming I. 2012. Calpain inhibition stabilizes the platelet proteome and reactivity in diabetes. Blood 120:415– 423. DOI: https://doi.org/10.1182/blood-2011-12-399980, PMID: 22665935 Rappsilber J, Mann M, Ishihama Y. 2007. Protocol for micro- purification, enrichment, pre- fractionation and storage of peptides for proteomics using StageTips. Nature Protocols 2:1896–1906. DOI: https://doi.org/10. 1038/nprot.2007.261, PMID: 17703201 Ribet D, Boscaini S, Cauvin C, Siguier M, Mostowy S, Echard A, Cossart P. 2017. SUMOylation of human septins is critical for septin filament bundling and cytokinesis. The Journal of Cell Biology 216:4041–4052. DOI: https:// doi.org/10.1083/jcb.201703096, PMID: 29051266 Schindelin J, Arganda- Carreras I, Frise E, Kaynig V, Longair M, Pietzsch T, Preibisch S, Rueden C, Saalfeld S, Schmid B, Tinevez JY, White DJ, Hartenstein V, Eliceiri K, Tomancak P, Cardona A. 2012. Fiji: an open- source platform for biological- image analysis. Nature Methods 9:676–682. DOI: https://doi.org/10.1038/nmeth.2019, PMID: 22743772 Schörg A, Santambrogio S, Platt JL, Schödel J, Lindenmeyer MT, Cohen CD, Schrödter K, Mole DR, Wenger RH, Hoogewijs D. 2015. Destruction of a distal hypoxia response element abolishes trans- activation of the PAG1 Keppner et al. eLife 2022;11:e72374. DOI: https://doi.org/10.7554/eLife.72374 25 of 26 Research article Cell Biology | Developmental Biology gene mediated by HIF- independent chromatin looping. Nucleic Acids Research 43:5810–5823. DOI: https:// doi.org/10.1093/nar/gkv506, PMID: 26007655 Shen YR, Wang HY, Kuo YC, Shih SC, Hsu CH, Chen YR, Wu SR, Wang CY, Kuo PL. 2017. SEPT12 phosphorylation results in loss of the septin ring/sperm annulus, defective sperm motility and poor male fertility. PLOS Genetics 13:e1006631. DOI: https://doi.org/10.1371/journal.pgen.1006631, PMID: 28346465 Shen YR, Wang HY, Tsai YC, Kuo YC, Wu SR, Wang CY, Kuo PL. 2020. The SEPT12 complex is required for the establishment of a functional sperm head-t ail junction. Molecular Human Reproduction 26:402–412. DOI: https://doi.org/10.1093/molehr/gaaa031, PMID: 32392324 Shevchenko A, Tomas H, Havlis J, Olsen JV, Mann M. 2006. In- gel digestion for mass spectrometric characterization of proteins and proteomes. Nature Protocols 1:2856–2860. DOI: https://doi.org/10.1038/ nprot.2006.468, PMID: 17406544 Sigg MA, Menchen T, Lee C, Johnson J, Jungnickel MK, Choksi SP, Garcia G, Busengdal H, Dougherty GW, Pennekamp P, Werner C, Rentzsch F, Florman HM, Krogan N, Wallingford JB, Omran H, Reiter JF. 2017. Evolutionary Proteomics Uncovers Ancient Associations of Cilia with Signaling Pathways. Developmental Cell 43:744-762.. DOI: https://doi.org/10.1016/j.devcel.2017.11.014, PMID: 29257953 Skarnes WC, Rosen B, West AP, Koutsourakis M, Bushell W, Iyer V, Mujica AO, Thomas M, Harrow J, Cox T, Jackson D, Severin J, Biggs P, Fu J, Nefedov M, de Jong PJ, Stewart AF, Bradley A. 2011. A conditional knockout resource for the genome- wide study of mouse gene function. Nature 474:337–342. DOI: https://doi. org/10.1038/nature10163, PMID: 21677750 Spiliotis ET, Nakos K. 2021. Cellular functions of actin- and microtubule- associated septins. Current Biology 31:R651–R666. DOI: https://doi.org/10.1016/j.cub.2021.03.064, PMID: 34033796 Storti F, Santambrogio S, Crowther LM, Otto T, Abreu- Rodríguez I, Kaufmann M, Hu CJ, Dame C, Fandrey J, Wenger RH, Hoogewijs D. 2014. A novel distal upstream hypoxia response element regulating oxygen- dependent erythropoietin gene expression. Haematologica 99:e45-e48. DOI: https://doi.org/10.3324/ haematol.2013.102707, PMID: 24510339 Sun P, Enslen H, Myung PS, Maurer RA. 1994. Differential activation of CREB by Ca2+/calmodulin- dependent protein kinases type II and type IV involves phosphorylation of a site that negatively regulates activity. Genes & Development 8:2527–2539. DOI: https://doi.org/10.1101/gad.8.21.2527, PMID: 7958915 Thijs L, Vinck E, Bolli A, Trandafir F, Wan X, Hoogewijs D, Coletta M, Fago A, Weber RE, Van Doorslaer S, Ascenzi P, Alam M, Moens L, Dewilde S. 2007. Characterization of a globin-c oupled oxygen sensor with a gene- regulating function. The Journal of Biological Chemistry 282:37325–37340. DOI: https://doi.org/10.1074/ jbc.M705541200, PMID: 17925395 Toure A, Rode B, Hunnicutt GR, Escalier D, Gacon G. 2011. Septins at the annulus of mammalian sperm. Biological Chemistry 392:799–803. DOI: https://doi.org/10.1515/BC.2011.074, PMID: 21740329 Tüttelmann F, Ruckert C, Röpke A. 2018. Disorders of spermatogenesis: Perspectives for novel genetic diagnostics after 20 years of unchanged routine. Medizinische Genetik 30:12–20. DOI: https://doi.org/10.1007/ s11825-018-0181-7, PMID: 29527098 Tyanova S, Temu T, Cox J. 2016. The MaxQuant computational platform for mass spectrometry- based shotgun proteomics. Nature Protocols 11:2301–2319. DOI: https://doi.org/10.1038/nprot.2016.136, PMID: 27809316 Villalobo A, González- Muñoz M, Berchtold MW. 2019. Proteins with calmodulin- like domains: structures and functional roles. Cellular and Molecular Life Sciences 76:2299–2328. DOI: https://doi.org/10.1007/s00018-019- 03062-z Wang M, Liu X, Chang G, Chen Y, An G, Yan L, Gao S, Xu Y, Cui Y, Dong J, Chen Y, Fan X, Hu Y, Song K, Zhu X, Gao Y, Yao Z, Bian S, Hou Y, Lu J, et al. 2018. Single- Cell RNA Sequencing Analysis Reveals Sequential Cell Fate Transition during Human Spermatogenesis. Cell Stem Cell 23:599-614.. DOI: https://doi.org/10.1016/j.stem. 2018.08.007, PMID: 30174296 Wang YY, Lai TH, Chen MF, Lee HL, Kuo PL, Lin YH. 2019. SEPT14 Mutations and Teratozoospermia: Genetic Effects on Sperm Head Morphology and DNA Integrity. Journal of Clinical Medicine 8:E1297. DOI: https://doi. org/10.3390/jcm8091297, PMID: 31450874 Yuan G, Nanduri J, Bhasker CR, Semenza GL, Prabhakar NR. 2005. Ca2+/calmodulin kinase-d ependent activation of hypoxia inducible factor 1 transcriptional activity in cells subjected to intermittent hypoxia. The Journal of Biological Chemistry 280:4321–4328. DOI: https://doi.org/10.1074/jbc.M407706200, PMID: 15569687 Keppner et al. eLife 2022;11:e72374. DOI: https://doi.org/10.7554/eLife.72374 26 of 26   2.3 A role of Androglobin in cancer? – insights from an overexpression system and  transcriptome data mining  Carina  Osterhof,    : The association of Androglobin with cancer – insights from  an overexpression system. (unpublished)  Supplementary files: electronic Appendix  Own contributions to this publication:  ‐ Planning and initial tests for generation of an A549 Adgb overexpression system. Stable  cell lines were generated and analysed via RT‐PCR and immunofluorescence under my  supervision by MSc student   (2021).  ‐ Cultivation of A549 cells. RNA isolation, quantification, and quality control from 3 Adgb  OE samples as well as 3 empty control lines.   ‐ Quality processing and mapping of mRNA‐Seq data from OE system; quantification and  subsequent differential gene expression analysis; interpretation of gene lists via gene  ontology analysis and upstream regulator prediction.  ‐ Phenotypic analysis of Adgb OE and empty control lines via migration assays. Images  of migration assays were processed with help of  .   ‐ Reanalysis of interpretation of qRT‐PCR data (carried out by  ) from clinical  lung  and  testis  samples  and  comparison  to  reanalysed,  publicly  available  transcriptomic data on cancer progression.  ‐ Mining of public databases containing transcriptomic information on cell lines (CCLE,  HPA) and  clinical  tumour  samples  from various  tissues  (TCGA); Correlation analysis  with known regulatory factors.     Planning  of  experiments,  analysis  and  interpretation  of  data  as  well  as  drafting  of  the  manuscript were realized together with  . The project was managed by Prof.  Dr. T. Hankeln.       73          74      A role of Andoglobin in cancer? – insights from an overexpression system and transcriptome  data mining                           75      Abstract  Androglobin  is  a  phylogenetically  ancient, multimeric  protein  with  an  embedded  globin  domain. Although this globin domain is circularly permutated and interrupted by a calmodulin  IQ domain, it still displays heme binding properties and may therefore interact with gaseous  signaling molecules. For other members of the globin gene family, such as NGB, CYGB and MB,  either  tumor  suppressor or oncogenic  functions have been proposed – depending on  the  tissue  the malignancy  originated  from.  Also  for  ADGB,  a  potential  oncogenic  effect was  suggested  derived  from  data  generated  in  prostate  and  brain  cancer  cell models,  but  it  remained  unclear whether  ADGB  is  also  expressed  in  real  tumor  entities. We  therefore  investigated the expression pattern of ADGB in the context of in vivo cancer samples and its  potential influence on tumor malignancy in a lung cancer model.   We analyzed a wide  variety of  transcriptomic datasets  from  cancer  cell  lines,  tumors and  studies on tumor progression and found that ADGB is systematically downregulated in both  immortalized cell lines as well as cancer entities derived from endogenous ADGB expression  sites. In the rare cases ADGB was expressed in cell lines, only few exons were transcribed or  the ADGB gene locus was multiplied. To study the effect of ADGB on immortalized cells, we  established  a  stable  overexpression  system  in  A549  lung  cancer  cells.  Surprisingly,  transcriptomic analysis of ADGB+ lung cancer cells showed higher motility and restructuring  of  the  extracellular  matrix  –  both  hallmarks  of  elevated  malignancy  in  cancer  entities.   Phenotypically, ADGB+  cells had  longer  cellular appendages, possibly  reflecting  the higher  migratory capacity, but the effect was not pronounced enough to be observable in a migration  assay. Thus, ADGB has oncogenic potential  in vitro. These findings are  in  line with previous  studies describing an oncogenic effect of ADGB in a glioma and pancreatic cancer cell model.  However, our expression analysis on in vivo cancer entities shows that ADGB mRNA is simply  not expressed or even actively silenced in malignancies. We therefore conclude that despite  its potential oncogenic properties, ADGB is unsuitable as prognostic marker in cancer patients.       76      Introduction  Globins are an evolutionary old family of proteins, which form a hydrophobic pocket (the so‐ called globin‐fold) that allows the binding of a heme prosthetic group. Vertebrate globins, such  as Hemoglobin (Hb) or Myoglobin (Mb), are well known for their vital role in oxygen supply  and transport (Wittenberg & Wittenberg, 2003). In addition, both have the ability to either  produce or scavenge reactive oxygen species (ROS) depending on the concentration of oxygen  in their surroundings (Flögel et al., 2001; Hendgen‐Cotta et al., 2008). By this property, globins  have been reported to promote or inhibit tumor growth. Several members of the globin family  are  in  fact deregulated  in human cancer entities and may even have prognostic value. The  expression of Mb, for example, is upregulated in colon and breast cancer (Flonta et al., 2009;  Gorr et al., 2011; Kristiansen et al., 2010). Depending on the tissue of origin, the presence of  Mb is either associated with a favorable outcome (Breast cancer, Kristiansen et al., 2010; Head  and  neck  squamous  cell  carinoma, Meller  et  al.,  2016)  or with  a more  aggressive  tumor  phenotype (Glioma, Elsherbiny et al., 2021). Cytoglobin (Cygb), a vertebrate globin type mostly  expressed  in connective tissue cells of many organs,  is mostly downregulated  in tumors via  hypermethylation of its promotors, e.g. in esophageal cancer, non‐small cell lung cancer and  head and neck squamous carcinoma (as reviewed  in: Oleksiewicz et al., 2011). Knockout of  Cygb in two independent mouse models showed a higher susceptibility for neoplasia in liver,  lung, and colon (Thi Thanh Thuy et al., 2011; Yassin et al., 2018). In line with these findings,  decreased Cygb expression was correlated with more frequent tumor recurrence and poor  prognosis in glioma as well as in pancreatic ductal adenocarcinoma patients (Kono et al., 2021;  Xu et al., 2013). For Neuroglobin (Ngb), a globin variant mostly expressed in neurons of the  nervous system (Burmester et al., 2000), a dualistic function in cancer progression has been  proposed. While it may act as a tumor suppressor in hepatocellular carcinoma (J. Zhang et al.,  2013),  its expression  is upregulated  in a variety of cancers  (Emara et al., 2010). However,  antibody free hybridization analysis could not reproduce this upregulation on the mRNA level  (Gorr et al., 2011). In glioma and breast cancer models, endogenous Ngb has shown an anti‐ apoptotic function and promotes proliferation (Fiocchetti et al., 2018; B. Zhang et al., 2018).  Together, these results suggest that globins may present promising candidates not only for  cardiovascular related, but also for cancer research.   77      The most recently reported “novel” member of the vertebrate globin family  is Androglobin  (ADGB)  (Hoogewijs et al., 2012). ADGB  is  structurally very different  from  the other  family  members,  since  it´s  globin  domain  is  embedded  in  a  larger  protein  which  carries  high  resemblance with a calpain protease. In addition, the ADGB globin domain is also permutated  and  interrupted by a putative calmodulin binding site. In analogy to other family members,  the name was designed to hint at the predominant expression site of the gene, since ADGB  was initially reported to be mainly expressed in later stages of spermatogenesis in mammalian  testes (Hoogewijs et al., 2012). Subsequently, ADGB expression was found to be a hallmark of  multi‐ciliated  cells  from  lung,  brain  and  also  the  female  reproductive  tract,  where  its  expression is probably regulated by FoxJ1, the master transcription factor of ciliogenesis (Koay  et al., 2021). This strong association with ciliated cells suggests a direct function in ciliogenesis,  however, ADGB’s exact mode of action is unclear. ADGB has also been studied in the context  of cancer. An  in vitro cell culture  study  suggested  that ADGB could act as an oncogene  in  glioma formation (Huang et al., 2014). Another study proposed that epigenetic silencing of  ADGB via  the  lncRNA STXBP5‐AS1  reduces cell  stemness  in pancreatic cancer  (Chen et al.,  2020). Besides this, however, the expression pattern and functional role of ADGB  in cancer  entities remains elusive.   We therefore aimed at characterizing the expression  intensity of ADGB  in different human  cancer  entities  and  during  cancer  progression  to  analyze whether  ADGB  could  influence  tumorigenesis in vivo.       78      Material and Methods   RNA‐sequencing analysis  RNA‐seq libraries were prepared from 1000 ng of total RNA with Illumina´s TruSeq RNA Sample  Prep Kit v2, including poly‐A selection, and sequenced as 70 nt single‐end reads on an Illumina  HiSeq2500. Library preparation and sequencing was performed by the NGS Core Facility of the  Department of Biology, Johannes‐Gutenberg‐University  (Mainz, Germany). Raw sequencing  data are available at the European Nucleotide Archive under accession number XXXXX.   Processing of RNA‐seq data  Raw  transcriptome  data  were  downloaded  either  from  NCBI  or  ENA  web  servers  (https://www.ncbi.nlm.nih.gov/sra; https://www.ebi.ac.uk/ena). Trimming parameters were  assessed  for  each  dataset  via  inspection  with  FastQC  (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/).  Adapter  and  quality  trimming were performed with BBDuk (https://sourceforge.net/projects/bbmap/). Minimum  quality cutoff of all datasets was phred 20, minimal read length was 20bp. After processing,  the reads were mapped against the reference genome of Homo sapiens (Hsa 38) with HISAT2  under default parameters (Kim et al., 2019). Calculation of transcripts‐per‐million values (TPM)  was performed with StringTie (Pertea et al., 2015) in guided mode.  Differential expression and GO term analysis  Differential gene expression analysis was conducted  in R  (https://www.r‐project.org/) with  Bioconductor package DESeq2 (Love et al., 2014). Genes were considered significant with p‐ adjusted ≤ 0.1. No LFC cutoff was applied. GO term enrichment analysis on gene  lists with  subsequent  term  reduction and visualization was performed with Metascape  (Zhou et al.,  2019) and g:profiler (Raudvere et al., 2019).   Prediction of upstream regulators  Upstream regulators were predicted on lists of differentially expressed genes with the Core  Ingenuity  pathway  analysis  (IPA,  Qiagen).  Chemical  regulators  were  excluded  from  the  analysis. Upstream regulators with a Z‐Score > |2| and a BH‐corrected p‐value > 0.05 were  considered significant.     79      Analysis of TCGA datasets  TCGA datasets were filtered, analysed and visualized with the GEPIA 2 tool (Tang et al., 2019).  Parameters for differential testing of ADGb expression were |LFC|=0.5 and p‐value cutoff =  0.01.  Generation of a stable overexpression system  A549 cells were provided by Prof. Dr. Susanne Strand (University Medical Center, Mainz). Cells  were cultured  in DMEM supplemented with 10 % FBS and 1 % Pen/Strep  (PAN Biotech) at  room air with 5 % CO2 and 37°C. pcDNA3.1 Vector with either full length ADGB or an empty  backbone  as  control were  transfected  using  Lipofectamine  3000  (Invitrogen).  Cells were  selected with Geneticin  (Roth) over 2 weeks and clones were obtained  through  single cell  dilution. For each condition, several clones of independent integration events were used for  further analysis.   RNA extraction   RNA extraction of tumor tissues was performed from snap‐frozen samples with the RNeasy  Plus Universal Mini kit (Qiagen) according to the manufacturer’s instruction. ~50 mg of tissues  was grinded and homogenized with a MiniLys (Precellys) system using mixed ceramic beads  (Precellys Lysing CKMix, Precellys). RNA was eluted  in nuclease‐free water. RNA extraction  from cell culture was performed with RNeasy Mini kit (Qiagen) according to the instructions  of the manufacturer.   RNA quality was assessed via a Bioanalyzer chip (Agilent), and only samples with RIN > 7 (for  tissues) or > 9 (for cultured cells) were used for further analysis. RNA was quantified via Qubit  measurement using the Broad Range RNA Assay Kit (Thermo Fisher) and was stored at ‐80°C  until further use.   qRT‐PCR   To validate the bioinformatical findings, we performed quantitative reverse‐transcription PCR  (qRT‐PCR) on  tissues as described previously  (Koay et al. 2021). 1000 ng of  total RNA per  sample was used for reverse transcription with the SuperScript III enzyme (10000 units per  assay; Invitrogen) using an Oligo‐dT primer. In the absence of validated reference genes, the  amount  of mRNA  expression  was  normalized  on  the  adjusted  total  amount  of  RNA.  To  80      additionally control  for differences  in cDNA synthesis, 100 ng of Drosophila  total RNA was  added to the reaction as a spike‐in control. qRT‐PCR was carried out using GoTaq qPCR Master  Mix  (Promega)  on  the  ABI  Prism  7500  Fast  Detection  System  (SDS,  Applied  Biosystems,  Carlsbad, USA) and interpreted using 7500 Software Version 2.3. Quantification of ADGB cDNA  molecules  (fwd  primer:  5’‐CGGAAGGAAAACATTCAAACAGG‐3’;  rev  Primer:  5’‐ CGAAACTGATGAATTTCTTCCGC‐3’) was done in absolute numbers applying a calibration standard  curve with known amounts of target PCR product, previously cloned  into the pGEM T‐easy  vector system  (Promega). Copies of the Drosophila Globin 1  (Glob1) cDNA  (fwd primer: 5’‐  GGAGCTAAGTGGAAATGCTCG‐3’;  rev  primer:  5’‐GATGATTCCGTAGACATGGTC‐3’)  of  the  internal  control were measured  in  parallel  to  identify  samples with  substandard  reverse  transcription.   Immunofluorescence  A549 cells were seeded in 24 well plates on uncoated glass cover slips. After 24‐48h, cells were  fixed with 4% PFA, washed, blocked in blocking buffer (1xPBS‐TritonX with 5% Horse serum  and  1%  BSA)  and  incubated  with  corresponding  primary  antibody  (α‐acetylated  tubulin;  custom made α‐ADGB (Charles River laboratories)) 1:500 in blocking buffer overnight. After  several washing steps, cells were  incubated with secondary antibody (CF 488, CF 555 or CF  633 conjugated anti‐mouse/anti‐rabbit F(ab)’‐Fragment (Sigma Aldrich)) and counterstained  with DAPI. Cover slips were mounted in RotiMount Fluorcare antifading solution (Roth) and  sealed  with  Roti‐Mount  (Roth).  Image  acquisition  was  done  with  a  Leica  SP5.  Cellular  appendages were measured with ImageJ.   Migration assay  Cells  were  seeded  in  4‐well  culture  dishes  (ibidi)  with  silicone  insert,  which  allowed  comparison of multiple cell population at once with a defined gap size. The silicon insert was  removed after cells reached confluency (~24h). Every 4h, cells were observed under an inverse  microscope and pictures were taken. Gap area was segmented, measured and plotted with  the  ImageJ  MiToBo  ScratchAssayAnalyzer  Toolbox  (https://mitobo.informatik.uni‐ halle.de/index.php), with σ=2 and an entropy filter size of 25.        81      Results  ADGB is downregulated in tumorous tissues and cell lines  Advances in NGS techniques have led to a plethora of publicly available data on transcriptomic  expression changes associated with  tumorigenesis. We  therefore mined mRNA sequencing  data  from  several  big  consortia  that  analysed  large  numbers  of  transcriptomic  data  from  cancerous tissues as well as commonly used cell lines.    Since lung is an endogenous expression site of ADGB and also a very frequent cancer type, we  firstly reanalyzed two RNA‐Seq datasets on  lung cancer progression  in human. We found a  significantly  lower amount of ADGB expression  in fully developed squamous cell carcinoma  (Ooi et al., 2014) in comparison to normal basal cells (Figure 1A). There was also a tendency  towards lower levels of ADGB mRNA already in premalignant cell lesions, however, this was  not statistically significant. The second study (Morton et al., 2014) showed a similar result,  with  significantly  lower ADGB  levels  in  invasive  lung  carcinomas  (Figure 1B).    Stratified by  patients,  the  study  also  showed  that  ADGB mRNA  was  downregulated  in  every  patient  analysed (n=6, Figure 1C).   Next, we investigated the extensive TCGA datasets (https://portal.gdc.cancer.gov/; Figure 2A)  with  a  focus  on  tissues  with  endogenous  ADGB  expression  (Koay  et  al.,  2021).  The  downregulation of ADGB mRNA expression could again be verified in different tumor entities  from lung and testis. We validated these findings via qPCR in paired normal and tumour tissue  samples  from  human  lung  and  testis  of  an  independent  cohort  (Figure  1  D,E).  Absolute  expression  of  ADGB mRNA  was  lower  in  every  tumour  sample  compared  to  its  healthy  counterpart. The TCGA data also showed a (non‐significant) tendency towards higher levels of  ADGB mRNA in two types of uterine cancer (Figure 2A). However, there was no difference of  ADGB mRNA levels between glioblastoma multiforme, brain lower grade glioma, pancreatic  adenocarcinomas and their respective healthy tissues (Figure 2B).   In summary, ADGB mRNA expression decreases during cancer progression in testis and lung  tissue, but the starting amount of ADGB varies greatly in accordance with our earlier findings.  This  is  probably  due  to  ADGBs  strict  cell  type  specific  expression  pattern  and  different  proportions of ciliated cells in the analysed samples (Koay et al., 2021).     82        Figure 1: Expression of ADGB mRNA  in human tumor tissue.  (A) ADGB mRNA expression decreases  significantly during progression from normal basal cells to squamous cell carcinoma (RNA‐Seq data set  from Ooi  et  al. 2014)  (B)  +  (C) ADGB mRNA  expression  is  also  significantly  lower  in  invasive  lung  carcinoma (data set from Morton et al. 2014). Although there is high variability in the amount of ADGB  mRNA in normal tissue samples, it decreases in every patient analysed during tumour progression. (D)  + (E) qPCR analysis of human ADGB mRNA levels in paired normal and tumour tissue samples of testis  and lung cancer. ADGB mRNA expression is significantly lower in both types of tumour tissue biopsies.   * = p < 0.05, ** = p < 0.01    83        Figure 2: ADGB expression in a subset of tumours sequenced by the TCGA consortium. Red = tumorous  tissue; grey = normal  tissue. Number  in brackets  indicate  the number of  tumour or healthy  tissue  entities that were used for the analysis. (A) Expression of ADGB in tumours which resulted from tissues  with endogenous expression. ADGB is significantly downregulated in lung adenocarcinoma, lung small  cell carcinoma as well as testicular germ cell tumours. In contrast, ADGB mRNA expression is higher in  tumorous uterine tissues, but compared to normal tissue this difference is not significant. (B) ADGB  expression  in  glioblastoma  multiforme,  brain  lower  grade  glioma,  pancreatic  adenocarcinoma  compared to the corresponding healthy tissue. Only low levels of ADGB mRNA can be detected in all  these groups.       84      In light of other studies which suggested an oncogenic function of ADGB in vitro (Chen et al.,  2020; Huang et al., 2014), we also  screened a variety of publicly available databases with  transcriptomic information on cell lines. We found that of the 961 cell lines analyzed by the  Cancer Cell Line Encyclopedia  (CCLE)  (Barretina et al., 2012; Ghandi et al., 2019), only  four  showed  a  low  amount  of  ADGB mRNA  expression:  lung  cancer  NCIH2106  (TPM  12.09);  liposarcoma  KMLS1  (TPM  11.93),  Hodgkin  lymphoma  HDLM2  (TPM  4.33)  and  colon  adenocarcinoma C75  (TPM 2.68)  (Figure 3A).  Intriguingly, genomic  resequencing data and  subsequent  gene  copy  number  analysis  derived  from  the  DepMap  portal  suggested  that  elevated ADGB expression in NCIH2106 as well as KMLS1 is not due to modulation of ADGB  regulation, but rather a consequence of an increased number of gene copies in these cell lines  (Supplementary  Figure  1A).  Of  the  63  cell  lines  analyzed  by  the  Human  Protein  Atlas  consortium, again only HDLM‐2 expressed low amounts of ADGB (Figure 3B). We reanalysed  transcriptomic data of HDLM‐2 from 2 independent studies. ADGB was indeed expressed in  both (TPM 7.63, 6.48, 9.91 (PRJNA668062) and 1.85 (PRJEB30312)). Visual inspection of the  alignments revealed that the cDNA reads covered only the last 4 exons from the ADGB gene  locus (Supplementary Figure 1B). This indicates that the HDLM‐2 cell line may express only a  truncated version of  the gene which neither encompasses  the calpain, nor  the globin and  calmodulin‐binding domains of ADGB. We found no bioinformatical evidence for endogenous  ADGB mRNA  expression  in  glioblastoma  cell  lines U87  and U251  in  contrast  to what was  described previously (Huang et al., 2014) (Figure 3A, “Brain cancer”).   85        Figure 3: ADGB mRNA expression in cancer cell lines analysed by the CCLE (A) and Human Protein Atlas  consortia (B). Only a small subset of cancer cell lines shows endogenous expression of ADGB mRNA.       86      Correlation analysis implies complex regulatory mechanisms of ADGB expression  Recently,  a  possible  mechanism  for  the  observed  downregulation  of  ADGB  mRNA  was  proposed. A  study  suggested  that  the  expression  of  LncRNA  STXBP5‐AS1  and  subsequent  recruitment of transcription factor EZH2 lead to heavy methylation und thus silencing of the  ADGB promotor (Chen et al., 2020). As a consequence, tissues with high levels of STXBP5‐AS1  should  show  lower  levels of ADGB mRNA.  It  seems, however,  that  the downregulation of  ADGB in tumorous tissues might follow a different mechanism, since we did not find evidence  for the expected inverse correlation of STXBP5‐AS1 and ADGB in the TCGA data of tissues with  endogenous ADGB expression (Supplementary figure 2A). On the contrary, both genes show  a positive correlation in their expression pattern in this dataset. Since the STXBP5‐AS1 gene is  located  directly  downstream  of  the  ADGB  gene  on  chromosome  6,  an  open  chromatin  conformation  could  possibly  affect  both  genes  simultaneously.  We  also  analysed  the  correlation coefficient of ADGB and STXBP5‐AS1  in pancreatic  tissue only,  since  this  is  the  tissue for which the mechanism was initially proposed. Also in this sub‐dataset, we found no  evidence of a correlated expression pattern (Supplementary figure 2B).  Another study recently showed that overexpression of transcription factors FOXJ1 and RFX2  was able to  induce the expression of ADGB  in HEK293 cells  (Koay et al., 2021).  Indeed, we  found a positive correlation between ADGB and expression of FOXJ1 as well as RFX2  in the  TCGA  datasets.  Interestingly,  ADGB  mRNA  expression  can  take  place  with  virtually  no  expression of FOXJ1 at all (Supplementary figure 2C, red box). In contrast, there was no RFX2‐ negative  sample  that  proved  ADGB  positive  (Supplementary  figure  2D).  RFX2  should  mechanistically stabilize FOXJ1 binding during motile ciliogenesis  (Quigley & Kintner, 2017)  and  the cell culture experiments showed  that RFX2 alone  is not sufficient  to  induce ADGB  expression (Koay et al., 2021). The results of the present study therefore suggest the existence  of additional, FOXJ1‐independent regulatory mechanisms for ADGB mRNA expression in tissue  samples.           87      Overexpression of Androglobin induces morphological changes in A549 cells  As  shown  above,  none  of  the  analyzed  cell  lines  seemed  to  be  an  adequate model  of  endogenous ADGB expression to study its function: expression levels were either extremely  low, expression did not  cover  the  complete ADGB gene  locus,  cell  lines putatively  carried  duplications of the ADGB gene locus (complicating genetic KO experiments), or cell lines were  derived  from  tissues which  themselves  are  no  native  expression  site  of  ADGB  in  healthy  organisms.    To study which effect the expression of ADGB would have on a cancer cell line, we decided to  stably overexpress ADGB in the widely used A549 lung adenocarcinoma cell line. Overall, 10  clonal cell  lines were generated  through G418  selection, 5 of which expressed human  full  length ADGB and another 5 contained an empty vector backbone as control. We validated the  overexpression on mRNA  level via qPCR as well as on protein  level by  immunofluorescent  staining of fixed cell samples (Supplementary figure 3). ADGB antibody signal showed an even  cytoplasmic distribution. Although ADGB carries a putative nuclear localisation signal, there  was  no  evidence  of  ADGB  transport  to  the  nucleus  (Figure  4,  Supplementary  figure  3).  Intriguingly, ADGB positive A549 cells showed abnormally long cellular appendages (Figure 4),  also  including cytokinetic bridges, with a median  length of 10.52 µm vs. 4.49 µm  in empty  control cells. Rare outliers were even as long as 50 µm (Figure 4D).  88        Figure 4: Immunofluorescent staining of ADGB‐overexpressing A549 cells. ADGB = green, DAPI (DNA)  = blue; acetylated a‐Tubulin = red; ADGB  (isolated channel  in  (B))  is  localized  in the cytoplasm and  accumulates  in cytokinetic bridges  (C) which are also  rich  in Tubulin.  (D) Ectopic overexpression of  ADGB in A549 clones significantly increased the length of cellular appendages.            89      Transcriptomic changes in response to ADGB overexpression in A549 cells  To further characterize the molecular phenotype of ectopic ADGB mRNA overexpression, we  performed RNA‐seq analysis of our stably transfected A549 cell clones (n=3) in comparison to  empty control cell lines (n=3). TPM values for ADGB mRNA were 4.01, 1.90 and 0.57 and thus  roughly comparable to the amount of ADGB mRNA  in healthy  lung samples (see also qPCR  results in Supplementary figure 3).   Differential  gene  expression  analysis  identified  a  set  of  497  genes,  231  of  which  were  upregulated  in  response  to ADGB  overexpression  (Supplementary  file  1). Among  these,  a  distinct  set  of  genes was  either  fully  induced  (counts  in  vector  only  cells:  0  in  all  three  replicates) or downregulated  (final  counts: 0  in all  three  replicates) upon ADGB  induction  (Table 1). Noteworthy, many of  the upregulated genes are expressed  in  tissues which are  positive for endogenous ADGB mRNA expression. The expression of the lincRNA EWSAT1 for  example is strongly induced upon overexpression of ADGB. It has been described as a potential  oncogene in both colorectal cancer and osteosarcoma (Shen et al., 2021; R. Zhang et al., 2018),  but in healthy tissue it is predominantly expressed in the testis. This gene and the others are  potential candidates for further analyses of ADGB’s gene regulatory network.   We  then performed GO overrepresentation  analysis on  the  list of differentially expressed  genes (Figure 5; full  list: Supplementary file 2). ADGB overexpression  is strongly associated  with transcriptional changes of genes associated with, among others, the extracellular matrix  organisation (GO:0030198), cell adhesion (GO:0016477) and cell migration (GO:0016477).  A  similar  picture  was  also  seen  when  only  upregulated  genes  were  used  for  GO  analysis  (Supplementary file 3). Comparison of our gene list against databases dedicated to pathways  resulted  in hits for extracellular matrix organisation (REACTOME, Gillespie et al., 2022) and  epithelial‐to‐mesenchymal  transition  in  both  thyroid  cells  and  colorectal  cancer  (WikiPathways, Martens et al., 2021).  To  identify  potential  regulators  involved  in  the  transcriptional  changes  induced  by  ADGB  overexpression, we performed an upstream regulator analysis using the  Ingenuity Pathway  software  suite  (Qiagen).  The  resulting  activated  (blue) or deactivated  (red)  regulators  are  summarized in Supplementary table 1. We evaluated the literature on these regulators with  regard to their influence on the motility of cells. Due to the proposed involvement of ADGB in  cilia/flagella formation (Keppner et al., 2022; Koay et al., 2021) and its possible calcium, NO or  90      oxygen reactivity (Hoogewijs et al., 2012), we also investigated the upstream regulators with  respect  to  these  terms.  Indeed,  many  of  these  candidate  regulators  turned  out  to  be  associated with ciliogenesis or were even known to  influence the motility or  length of cilia.  Few, however, were connected with oxygen or NO metabolism, in literature. In line with the  GO term analysis of the target genes, many of these regulators are involved in modulating the  motility of cells or regulate the epithelial to mesenchymal transition.      Figure 5: GO terms enriched in differentially expressed genes upon overexpression of ADGB in A549  lung cancer cells. Differentially expressed genes: LFC > 0.5; adjusted p‐value < 0.05. Terms are strongly  associated with changes in the extracellular matrix and motility as well as adhesion of cells. Terms are  colour  coded  by  their  respective  database:  green  = GO: Biological  process;  blue  = GO: Molecular  function; red = GO: Cellular component; rose = KEGG; Purple = WikiPathways; orange = REACTOME  Pathways.  91      ADGB overexpressing cells do not exhibit higher motility   The bioinformatical findings predicted a molecular phenotype in motility and the composition  of the extracellular matrix. Thus, we  investigated the migratory capacity of our cell culture  system. Although transcriptomic data suggested a more mesenchymal phenotype in ADGB+  vs.  ADGB‐  cells,  cell  migration  assays  revealed  no  statistically  significant  difference  in  migratory capacity (Figure 6).     Figure 6: Migration assay of ADGB+ (red) and Empty control cells (black). Symbols indicate pairs of cell  lines that were grown in the same culture vessel. In pair A7/E9, a constant decline in gap size could be  measured, but no difference between the genotypes was observable.  In the 2 other pairs, gap size  increased unexpectedly after 12 hours, indicating technical difficulties with the experiment.           92      Discussion  Androglobin  is a chimeric protein harbouring a globin domain (Hoogewijs et al., 2012). The  other members of the globin protein  family are known  for differential regulation and both  oncogenic or protective functions in tumour entities. We therefore studied the expression of  ADGB in the context of cancer. Our analysis of a wide variety of transcriptome of cancer cell  lines and different malignancies revealed that ADGB is not expressed in the vast majority of  cell  lines  analysed.  In  addition, ADGB was  found downregulated  in  tumours derived  from  tissues with endogenous ADGB expression.  This could suggest that ADGB either has a tumour  suppressor function or simply no influence at all upon tumorigenesis. Counterintuitively, our  transcriptomic analysis of an ADGB overexpression system suggested a putative oncogenic  role  for  ADGB,  much  in  agreement  with  two  previous  studies,  which  showed  that  downregulation  of  ADGB  resulted  in  slower  proliferation  and  higher  rate  of  apoptosis  of  glioblastoma cells (Huang et al., 2014) and compromised stemness and metastatic capacity in  pancreatic cancer cell lines (Chen et al., 2020). Ectopic ADGB expression in A549 cells led to  transcriptional  changes which would promote  cell motility  and  changes of  the ECM, both  processes relevant for the epithelial to mesenchymal transition and a hallmark of malignant  cancer entities (e.g. Grelet et al., 2017; Thiery et al., 2009). However, this potential gain  in  motility  was  not  pronounced  enough  to  be  observable  in  a  classic  migration  assay.  Nevertheless,  ADGB  overexpressing  cells  showed  longer  cellular  appendages  including  cytokinetic bridges. These bridges are the final connection of two daughter cells which have  completed mitosis. Since the process of final abscission is carefully timed (Nähse et al., 2017),  longer  bridges  suggest  a  higher motility  of  the  connected  cells  or  a  change  in  adhesive  properties, which would be in line with the transcriptomic and regulator analysis of our cell  model. ADGB+ tumour cells may thus get a selective advantage in vitro – but ADGB positivity  simply does not happen in cancer in vivo.   In healthy tissues, ADGB is exclusively expressed in ciliated cells of lung, brain and the fallopian  tubes as well as during late stages of spermatogenesis (Koay et al., 2021). The overexpression  of FOXJ1 and RFX2, which are  important  ciliogenesis  transcription  factors, was  capable of  ADGB  induction  in vitro (Koay et al., 2021). We have found that tissues can be positive for  ADGB  mRNA  without  the  expression  of  FOXJ1,  possibly  hinting  at  additional  regulatory  mechanisms for ADGB in vivo.   93      Ciliogenesis is a process which is tightly connected to the cell cycle: the centriole acts both as  a initiation point for the spindle apparatus and as the basal body, which migrates to the cell  borders and serves as an axoneme nucleation point to  initiate cilium protrusion (Avasthi &  Marshall, 2012). The cilium can then be absorbed upon re‐entry  into mitosis. In contrast to  this, differentiation to a cell carrying motile cilia is not reversible any more. Overproliferative  cells such as some cancer cell lines and also most malignancies in vivo notably do not carry  cilia. It seems that loss of the differentiation status of cells and tissues that comes along with  cancer progression concomitantly leads to the loss of ADGB expression, indicating that ADGB  is specifically needed  in fully differentiated ciliated cells and also tightly regulated with this  process.   In conclusion, ADGB may have the capability to enhance malignancy in tumour entities, and  possibly does so in artificial overexpressing cell systems, but since its expression is simply not  induced  in vivo  in proliferative,  cancerous  tissues,  this has no prognostic value  for  cancer  patients.     Acknowledgements  The  results  reported  here  are  in  part  based  upon  data  generated  by  the  TCGA  Research  Network:  https://www.cancer.gov/tcga.  The  authors  thank    for  help  in  analysing qRT‐PCR data and   for critical reading of the manuscript. This work  was supported by the Deutsche Forschungsgemeinschaft (DFG) to T.H. (HA 2103/9‐1).  94      Table 1: Differentially expressed genes upon overexpression of ADGB  in A%$)  lung cancer cells. All genes were either fully  induced or downregulated  in all replicates. LFC =  LogFoldChange; p‐adjusted = BH corrected p‐value.  Gene Name  LFC  p‐adjusted  RNA‐type  main mRNA expression site (HPA + GTEX)  AC027088.2  10.99  8.71E‐16  linc‐RNA  testis, brain  ADGB  8.33  5.47E‐07  protein‐coding  fallopian tube, testis, lung, brain; ciliated cells  NMU  8.17  1.81E‐07  protein‐coding  esophagus, skin; squamous epithelial cells  LINC00648  7.69  4.39E‐06  linc‐RNA  broadly expressed  NELL2  7.69  8.23E‐05  protein‐coding  brains; neurons, T‐cells  LINC02081  7.69  4.59E‐06  linc‐RNA  testis, lung  CTNND2  7.58  2.79E‐05  protein‐coding  brain  MAGEA1  7.38  1.82E‐05  protein‐coding  testis; spermatogonia, spermatocytes  JPH1  6.90  2.23E‐03  protein‐coding  broadly expressed; highest in Cardiomyocytes  GIMAP2  6.74  5.51E‐03  protein‐coding  lymphoid tissue; B‐cells  GLI3  6.25  6.37E‐02  protein‐coding  broadly expressed; fibroblasts  SLC27A6  6.24  1.19E‐02  protein‐coding  broadly expressed, adrenal gland, brain, fallopian tube, heart muscle  RARRES3  4.48  5.32E‐02  protein‐coding  broadly expressed  WNT6  ‐4.00  5.63E‐04  protein‐coding  broadly expressed; Sertoli cells  OPRL1  ‐4.62  5.61E‐04  protein‐coding  testis; early &late spermatids  MUC5B  ‐4.98  6.64E‐07  protein‐coding  salivary gland; intestinal epithelial cells  ADGRL3  ‐5.62  1.91E‐04  protein‐coding  brain  AL645608.7  ‐5.99  1.45E‐04  linc‐RNA  broadly expressed, artery  S100P  ‐6.00  3.13E‐04  protein‐coding  bone marrow, stomach, urinary bladder  SELENBP1  ‐6.51  2.69E‐05  protein‐coding  intestine, liver  SLITRK4  ‐6.54  3.15E‐15  protein‐coding  adrenal gland, brain, skeletal muscle  LUM  ‐6.85  1.36E‐06  protein‐coding  gallbladder, placenta; fibroblasts  TNNC1  ‐7.20  1.47E‐07  protein‐coding  muscle; striated muscle cell  CCNYL2  ‐7.26  4.24E‐05  pseudogene  testis  MYOCD  ‐7.72  1.15E‐06  protein‐coding  broadly expressed; fibroblasts  HOXA9  ‐8.06  6.29E‐10  protein‐coding  broadly expressed, highest in kidney; proximal tubular cells  ASAH2  ‐9.65  3.51E‐15  protein‐coding  intestine; enterocytes  LINC00839  ‐9.85  7.83E‐15  linc‐RNA  broadly expressed, kidney, spleen  95      Supplementary table 1:  IPA upstream regulator prediction in ADGB+ cells: activated (blue) and deactivated (red)  transcription factors  Regulator  z‐score  p‐value  motility?  cilia?  calcium?  hypoxia/NO?  TGFB3  3,385  4,34E‐06  ↓ motility of cells (Han et  defect ↓ sperm motility al., 2012)  (Drozdzik et al., 2015)      2+  FGF2  3,171  1,12E‐04  ↑ sperm motility (Saucedo  controls ciliary length  ↑ intracellular Ca levels (Y. et al., 2015)  (Bosakova et al., 2019)  Liu & Schneider, 2014)    ↑ ciliary beat frequency  TNF 2,978  2,04E‐08  (dosage dependent) (Rhee    ↑ Ca 2+levels (González et al.,    et al., 1999; Weiterer et al.,  2016)    2014)  EGF  2,840  4,37E‐07  ↑ motility (Xie et al., 1998)  ↓ ciliogenesis(Kasahara et al., 2018)      regulates differentiation of  Tgf beta  2,674  3,09E‐05  motile cilia (Tözser et al.,  controls ciliary length  controls ECM via 2+   2015)  (Tözser et al., 2015)  Ca (Mukherjee et al., 2012)  ROS mediated ↑ of motility  destabilizes primary cilia (S.  regulated by calmodulin STAT1 2,599  1,29E‐04  dependent kinase (L. Wang  hypoxia ↓ STAT1 (Ivanov et   (J. Liu et al., 2019)  Wang et al., 2018)  et al., 2008)  al., 2007)  MSTN  2,412  2,50E‐03  ‐        SNAI2  2,395  1,71E‐03          IFNB1  2,385  7,78E‐04          HGF  2,244  2,48E‐06  ↑ epithelial cell motility (Spix et al., 2007)        Interferon alpha  2,219  4,64E‐07  cf. STAT1        TWIST2  2,205  7,93E‐06  ↑ EMT (T. Wang et al., 2014)        ↑ EMT (Gao et al., 2016;  PRMT1  2,186  2,78E‐04  Jackson‐Weaver et al.,  localizes to Flagellum  upregulated by Ca 2+ (M. Y.  hypoxia inducible in A549  2020)  (Mizuno & Sloboda, 2017)  Liu et al., 2020)  (Lim et al., 2013)  OSM  2,180  9,39E‐08          96      TFAP4  2,164  9,42E‐04          SNCA  2,138  2,66E‐04           RASSF1  ‐2,714  1,68E‐06  isoform a ↓ motility        isoform c ↑ motility  (Vlahov et al., 2015)  AHR  ‐2,599  2,23E‐04  ‐  essential for ciliogenesis  regulates Ca‐signalling  = oxygen sensor  (embryo) (Villa et al., 2016)  (Brinchmann et al., 2018)  (Rothhammer & Quintana,  2019)  BCL3  ‐2,415  1,45E‐03  ↑(Wakefield et al.,  ‐  ‐    2013)/↓(Wakefield et al.,  2008) motility  TRIM24  ‐2,170  1,40E‐05  ↓ motility (↑ adhesion)  ‐  ‐    (Appikonda et al., 2018)  NEUROG1  ‐2,121  3,89E‐05  ‐  ‐  ‐  hypoxia ↑ NSC proliferation  (Qi et al., 2017);  NEUROG1 ↑ differentiation  (Luo et al., 2010)  MYOC  ‐2,000  3,35E‐04  ↑ migration (Kwon &  localizes to cilia of  binds calcium (Donegan et    Tomarev, 2011)  photoreceptor cells (Kubota  al., 2012)  et al., 1997)    97          98        Supplementary figure 1: (A) ADGB expression in cell lines from the CCLE consortium : TPM values and  ADGB gene copy number appear to correlate. Cell lines NCIH2106 and KMLS1, which show the highest  endogenous  ADGB  expression,  also  harbour  multiple  copies  of  the  ADGB  gene.    (B)  Graphical  representation of ADGB read coverage in HDLM‐2 RNA‐Seq data. The 5’‐end of the gene is free from  reads (left) whereas the last 4 exons are expressed.  99        Supplementary figure 2: Correlation analysis of ADGB and potential transcription factors on mRNA  level.     100          101      Supplementary  figure  3:  Validation  of  ectopic  ADGB  mRNA  expression  via  qPCR  and  immunofluorescent staining. Green = α‐ADGB antibody, blue = DAPI. (A) and (B) show staining of ADGB  overexpressing lines 1 and 8, whereas (C) and (D) where stably transfected with an empty control (lines  5 and 9). (E) ADGB overexpressing cells stained only with a secondary antibody as control did not show  any fluorescence in the green channel. (F) Validation of ectopic ADGB mRNA expression in transfected  A549 cell clones (ADGB 1,5,7,8,11) via qPCR analysis. cDNA from  lung,  lung cancer, testis and testis  cancer were used as references to assess intensity of overexpression. Expression intensity is shown in  relation to the lung (Replicate 1, 100%). Weak PCR signals were obtained in vector transfected “Empty”  A549 cells and the no‐template control (NTC). Melting curve analysis showed that these signals were  due to short amplicons originating from primer dimerization. No melting point corresponding to those  artifactual amplicons could be detected in the other samples.      102      References  Appikonda, S., Thakkar, K. N., Shah, P. K., Dent, S. Y. R., Andersen, J. N., & Barton, M. C. (2018).  Cross‐talk  between  chromatin  acetylation  and  SUMOylation  of  tripartite  motif– containing protein 24  (TRIM24)  impacts cell adhesion.  Journal of Biological Chemistry,  293(19), 7476–7485.  Avasthi, P., & Marshall, W. F. (2012). Stages of Ciliogenesis and Regulation of Ciliary Length.  Differentiation; Research in Biological Diversity, 83(2), S30.  Barretina, J., Caponigro, G., Stransky, N., Venkatesan, K., Margolin, A. A., Kim, S., Wilson, C. J.,  Lehár, J., Kryukov, G. V., Sonkin, D., Reddy, A., Liu, M., Murray, L., Berger, M. F., Monahan,  J. E., Morais, P., Meltzer, J., Korejwa, A., Jané‐Valbuena, J., … Garraway, L. A. (2012). The  Cancer Cell Line Encyclopedia enables predictive modelling of anticancer drug sensitivity.  Nature, 483(7391), 603–607.  Bosakova, M. K., Nita, A., Gregor, T., Varecha, M., Gudernova, I., Fafilek, B., Barta, T., Basheer,  N., Abraham, S. P., Balek, L., Tomanova, M., Kucerova, J. F., Bosak, J., Potesil, D., Zieba, J.,  Song, J., Konik, P., Park, S., Duran, I., … Krejci, P. (2019). Fibroblast growth factor receptor  influences  primary  cilium  length  through  an  interaction  with  intestinal  cell  kinase.  Proceedings  of  the  National  Academy  of  Sciences  of  the  United  States  of  America,  116(10), 4316–4325.  Brinchmann, B. C.,  Le  Ferrec,  E., Bisson, W. H.,  Podechard, N., Huitfeldt, H.  S., Gallais,  I.,  Sergent,  O.,  Holme,  J.  A.,  Lagadic‐Gossmann,  D.,  &  Øvrevik,  J.  (2018).  Evidence  of  selective  activation  of  aryl  hydrocarbon  receptor  nongenomic  calcium  signaling  by  pyrene. Biochemical Pharmacology, 158, 1–12.  Burmester, T., Welch, B., Reinhardt, S., & Hankeln, T. (2000). A vertebrate globin expressed in  the brain. Nature, 407(6803), 520–523.  Chen, S., Huang, L., Li, G., Qiu, F., Wang, Y., Yang, C., Pan, J., Wu, Z., Chen, J., & Tian, Y. (2020).  LncRNA  STXBP5‐AS1  suppresses  stem  cell‐like  properties  of  pancreatic  cancer  by  epigenetically  inhibiting neighboring androglobin gene expression. Clinical Epigenetics,  12(1), 1–12.  Donegan, R. K., Hill, S. E., Turnage, K. C., Orwig, S. D., & Lieberman, R. L. (2012). The glaucoma‐ 103      associated olfactomedin domain of myocilin is a novel calcium binding protein. Journal  of Biological Chemistry, 287(52), 43370–43377.  Drozdzik, M., Kaczmarek, M., Malinowski, D., Bros, U., Kazienko, A., Kurzawa, R., & Kurzawski,  M.  (2015).  TGFβ3  (TGFB3) polymorphism  is  associated with male  infertility.  Scientific  Reports, 5, 71–74.  Elsherbiny, M.  E.,  Shaaban, M.,  El‐Tohamy,  R.,  Elkholi,  I.  E., Hammam, O. A., Magdy, M.,  Allalunis‐Turner, J., & Emara, M. (2021). Expression of Myoglobin in Normal and Cancer  Brain Tissues: Correlation With Hypoxia Markers. Frontiers in Oncology, 11, 1561.  Emara, M., Turner, A. R., & Allalunis‐Turner, J. (2010). Hypoxic regulation of cytoglobin and  neuroglobin expression  in human normal and tumor tissues. Cancer Cell  International,  10(1), 1–16.  Fiocchetti, M., Cipolletti, M., Ascenzi, P., & Marino, M. (2018). Dissecting the 17β‐estradiol  pathways necessary for neuroglobin anti‐apoptotic activity  in breast cancer. Journal of  Cellular Physiology, 233(7), 5087–5103.  Flögel, U., Merx, M. W., Gödecke, A., Decking, U. K. M., & Schrader, J. (2001). Myoglobin: A  scavenger of bioactive NO. Proceedings of the National Academy of Sciences of the United  States of America, 98(2), 735–740.  Flonta, S. E., Arena, S., Pisacane, A., Michieli, P., & Bardelli, A. (2009). Expression and functional  regulation of myoglobin in epithelial cancers. The American Journal of Pathology, 175(1),  201–206.  Gao, Y., Zhao, Y., Zhang, J., Lu, Y., Liu, X., Geng, P., Huang, B., Zhang, Y., & Lu, J. (2016). The  dual  function of PRMT1  in modulating epithelial‐mesenchymal  transition  and  cellular  senescence  in  breast  cancer  cells  through  regulation  of  ZEB1.  Scientific  Reports,  6(December 2015), 1–13.  Ghandi, M.,  Huang,  F. W.,  Jané‐Valbuena,  J.,  Kryukov,  G.  V.,  Lo,  C.  C., McDonald,  E.  R.,  Barretina, J., Gelfand, E. T., Bielski, C. M., Li, H., Hu, K., Andreev‐Drakhlin, A. Y., Kim, J.,  Hess, J. M., Haas, B. J., Aguet, F., Weir, B. A., Rothberg, M. V., Paolella, B. R., … Sellers, W.  R. (2019). Next‐generation characterization of the Cancer Cell Line Encyclopedia. Nature,  569(7757), 503–508.  104      Gillespie, M., Jassal, B., Stephan, R., Milacic, M., Rothfels, K., Senff‐Ribeiro, A., Griss, J., Sevilla,  C., Matthews,  L., Gong,  C., Deng,  C., Varusai,  T.,  Ragueneau,  E., Haider,  Y., May,  B.,  Shamovsky, V., Weiser, J., Brunson, T., Sanati, N., … D’Eustachio, P. (2022). The reactome  pathway knowledgebase 2022. Nucleic Acids Research, 50(D1), D687–D692.  González, C., Droguett, K., Rios, M., Cohen, N. A., & Villalón, M. (2016). TNF α Affects Ciliary  Beat Response  to  Increased Viscosity  in Human Pediatric Airway  Epithelium. BioMed  Research International, 2016, 1–9.  Gorr,  T.  A., Wichmann, D.,  Pilarsky,  C.,  Theurillat,  J.  P.,  Fabrizius,  A.,  Laufs,  T.,  Bauer,  T.,  Koslowski, M., Horn, S., Burmester, T., Hankeln, T., & Kristiansen, G. (2011). Old proteins  – new locations: myoglobin, haemoglobin, neuroglobin and cytoglobin in solid tumours  and cancer cells. Acta Physiologica, 202(3), 563–581.  Grelet, S., McShane, A., Geslain, R., & Howe, P. H.  (2017). Pleiotropic Roles of Non‐Coding  RNAs  in  TGF‐β‐Mediated  Epithelial‐Mesenchymal  Transition  and  Their  Functions  in  Tumor Progression. Cancers, 9(7).  Han, A., Bandyopadhyay, B., Jayaprakash, P., Lua, I., Sahu, D., Chen, M., Woodley, D. T., & Li,  W. (2012). The anti‐motility signaling mechanism of TGFβ3 that controls cell traffic during  skin wound healing. Biology Open, 1(12), 1169–1177.  Hendgen‐Cotta, U. B., Merx, M. W., Shiva, S., Schmitz, J., Becher, S., Klare, J. P., Steinhoff, H.  J., Goedecke,  A.,  Schrader,  J., Gladwin, M.  T.,  Kelm, M., &  Rassaf,  T.  (2008). Nitrite  reductase activity of myoglobin regulates respiration and cellular viability in myocardial  ischemia‐reperfusion  injury.  Proceedings  of  the National  Academy  of  Sciences  of  the  United States of America, 105(29), 10256–10261.  Hoogewijs, D., Ebner, B., Germani, F., Hoffmann, F. G., Fabrizius, A., Moens, L., Burmester, T.,  Dewilde, S., Storz, J. F., Vinogradov, S. N., & Hankeln, T. (2012). Androglobin: A Chimeric  Globin  in Metazoans That  Is Preferentially Expressed  in Mammalian Testes. Molecular  Biology and Evolution, 29(4), 1105–1114.  Huang, B., Lu, Y., Li, X., Zhu, Z., Li, K., Liu, J., Zheng, J., & Hu, Z. (2014). Androglobin knockdown  inhibits growth of glioma cell lines. Int J Clin Exp Pathol, 7(5), 2179–2184.  Ivanov, S. V., Salnikow, K., Ivanova, A. V., Bai, L., & Lerman, M. I. (2007). Hypoxic repression of  105      STAT1 and its downstream genes by a pVHL/HIF‐1 target DEC1/STRA13. Oncogene, 26(6),  802–812.  Jackson‐Weaver, O., Ungvijanpunya, N., Yuan, Y., Qian, J., Gou, Y., Wu, J., Shen, H., Chen, Y.,  Li, M., Richard, S., Chai, Y., Sucov, H. M., & Xu, J. (2020). PRMT1‐p53 Pathway Controls  Epicardial EMT and Invasion. Cell Reports, 31(10).  Kasahara, K., Aoki, H., Kiyono, T., Wang, S., Kagiwada, H., Yuge, M., Tanaka, T., Nishimura, Y.,  Mizoguchi,  A.,  Goshima,  N.,  &  Inagaki,  M.  (2018).  EGF  receptor  kinase  suppresses  ciliogenesis through activation of USP8 deubiquitinase. Nature Communications, 9(1).  Keppner, A., Correia, M., Santambrogio, S., Koay, T. W., Maric, D., Osterhof, C., Winter, D. V,  Clerc, A., Stumpe, M., Chalmel, F., Dewilde, S., Odermatt, A., Kressler, D., Hankeln, T.,  Wenger, R. H., & Hoogewijs, D.  (2022). Androglobin, a chimeric mammalian globin,  is  required for male fertility. ELife, 11.  Kim, D.,  Paggi,  J. M.,  Park, C., Bennett, C., &  Salzberg,  S.  L.  (2019). Graph‐based  genome  alignment  and  genotyping  with  HISAT2  and  HISAT‐genotype.  Nature  Biotechnology,  37(8), 907–915.  Koay, T. W., Osterhof, C., Orlando, I. M. C., Keppner, A., Andre, D., Yousefian, S., Alonso, M. S.,  Correia, M., Markworth, R., Schödel, J., Hankeln, T., & Hoogewijs, D. (2021). Androglobin  gene  expression  patterns  and  FOXJ1‐dependent  regulation  indicate  its  functional  association with ciliogenesis. Journal of Biological Chemistry, 296.  Kono, H., Hosomura, N., Amemiya, H., Kawaida, H., Furuya, S., Akaike, H., Kawaguchi, Y., Sudo,  M.,  &  Ichikawa,  D.  (2021).  Cytoglobin  as  a  Prognostic  Factor  for  Pancreatic  Ductal  Adenocarcinoma. Pancreas, 50(7), 994–999.  Kristiansen, G., Rose, M., Geisler, C., Fritzsche, F. R., Gerhardt,  J., Lüke, C., Ladhoff, A. M.,  Knüchel, R., Dietel, M., Moch, H., Varga, Z., Theurillat, J. P., Gorr, T. A., & Dahl, E. (2010).  Endogenous  myoglobin  in  human  breast  cancer  is  a  hallmark  of  luminal  cancer  phenotype. British Journal of Cancer, 102(12), 1736–1745.  Kubota, R., Noda, S., Wang, Y., Minoshima, S., Asakawa, S., Kudoh, J., Mashima, Y., Oguchi, Y.,  & Shimizu, N. (1997). A novel myosin‐like protein (myocilin) expressed in the connecting  cilium  of  the  photoreceptor: Molecular  cloning,  tissue  expression,  and  chromosomal  106      mapping. Genomics, 41(3), 360–369.  Kwon, H. S., & Tomarev, S. I. (2011). Myocilin, a glaucoma‐associated protein, promotes cell  migration through activation of  integrin‐focal adhesion kinase‐serine/threonine kinase  signaling pathway. Journal of Cellular Physiology, 226(12), 3392–3402.  Lim, S. K., Jeong, Y. W., Kim, D. Il, Park, M. J., Choi, J. H., Kim, S. U., Kang, S. S., Han, H. J., &  Park, S. H.  (2013). Activation of PRMT1 and PRMT5 mediates hypoxia‐ and  ischemia‐ induced apoptosis in human lung epithelial cells and the lung of miniature pigs: The role  of p38 and JNK mitogen‐activated protein kinases. Biochemical and Biophysical Research  Communications, 440(4), 707–713.  Liu,  J., Qu,  L., Meng,  L., &  Shou, C.  (2019).  Topoisomerase  inhibitors promote  cancer  cell  motility  via  ROS‐mediated  activation  of  JAK2‐STAT1‐CXCL1  pathway.  Journal  of  Experimental and Clinical Cancer Research, 38(1), 1–12.  Liu, M. Y., Hua, W. K., Chiou, Y. Y., Chen, C. J., Yao, C. L., Lai, Y. T., Lin, C. H., & Lin, W. J. (2020).  Calcium‐dependent methylation by PRMT1 promotes erythroid differentiation through  the p38α MAPK pathway. FEBS Letters, 594(2), 301–316.  Liu, Y., & Schneider, M. F. (2014). FGF2 activates TRPC and Ca2+ signaling leading to satellite  cell activation. Frontiers in Physiology, 5 FEB(February), 1–10.  Love, M. I., Anders, S., & Huber, W. (2014). Differential analysis of count data ‐ the DESeq2  package. In Genome Biology (Vol. 15, Issue 12).  Luo, Y., Shan, G., Guo, W., Smrt, R. D., Johnson, E. B., Li, X., Pfeiffer, R. L., Szulwach, K. E., Duan,  R., Barkho, B. Z., Li, W., Liu, C., Jin, P., & Zhao, X.  (2010). Fragile X mental retardation  protein regulates proliferation and differentiation of adult neural stem/progenitor cells.  PLoS Genetics, 6(4).  Martens, M., Ammar, A., Riutta, A., Waagmeester, A., Slenter, D. N., Hanspers, K., Miller, R.  A.,  Digles,  D.,  Lopes,  E.  N.,  Ehrhart,  F.,  Dupuis,  L.  J.,  Winckers,  L.  A.,  Coort,  S.  L.,  Willighagen, E. L., Evelo, C. T., Pico, A. R., & Kutmon, M. (2021). WikiPathways: connecting  communities. Nucleic Acids Research, 49(D1), D613–D621.  Meller, S., van Ellen, A., Gevensleben, H., Bicker, A., Hankeln, T., Gorr, T. A., Sailer, V., Dröge,  F., Schröck, F., Bootz, F., Schröck, A., Perner, S., Dietrich, D., & Kristiansen, G.  (2016).  107      Ectopic Myoglobin Expression Is Associated with a Favourable Outcome in Head and Neck  Squamous Cell Carcinoma Patients. Anticancer Research, 36(12), 6235–6242.  Mizuno,  K.,  &  Sloboda,  R.  D.  (2017).  Protein  arginine  methyltransferases  interact  with  intraflagellar  transport  particles  and  change  location  during  flagellar  growth  and  resorption. Molecular Biology of the Cell, 28(9), 1208–1222.  Morton, M. L., Bai, X., Merry, C. R., Linden, P. A., Khalil, A. M., Leidner, R. S., & Thompson, C.  L. (2014). Identification of mRNAs and lincRNAs associated with lung cancer progression  using next‐generation RNA sequencing  from  laser micro‐dissected archival FFPE tissue  specimens. Lung Cancer, 85(1), 31–39.  Mukherjee, S., Kolb, M. R. J., Duan, F., & Janssen, L. J. (2012). Transforming growth factor‐β  evokes  Ca2+ waves  and  enhances  gene  expression  in  human  pulmonary  fibroblasts.  American Journal of Respiratory Cell and Molecular Biology, 46(6), 757–764.  Nähse, V., Christ, L., Stenmark, H., & Campsteijn, C. (2017). The Abscission Checkpoint: Making  It to the Final Cut. Trends in Cell Biology, 27(1), 1–11.  Oleksiewicz, U.,  Liloglou,  T.,  Field,  J.  K., &  Xinarianos, G.  (2011).  Cytoglobin:  biochemical,  functional and clinical perspective of the newest member of the globin family. Cellular  and Molecular Life Sciences 2011 68:23, 68(23), 3869–3883.  Ooi, A. T., Gower, A. C., Zhang, K. X., Vick, J. L., Hong, L., Nagao, B., Wallace, W. D., Elashoff, D.  A., Walser, T. C., Dubinett, S. M., Pellegrini, M., Lenburg, M. E., Spira, A., & Gomperts, B.  N. (2014). Molecular profiling of premalignant lesions in lung squamous cell carcinomas  identifies mechanisms involved in stepwise carcinogenesis. Cancer Prevention Research,  7(5), 487–495.  Pertea, M., Pertea, G. M., Antonescu, C. M., Chang, T.‐C., Mendell, J. T., & Salzberg, S. L. (2015).  StringTie  enables  improved  reconstruction  of  a  transcriptome  from  RNA‐seq  reads.  Nature Biotechnology, 33(3).  Qi, C., Zhang, J., Chen, X., Wan, J., Wang, J., Zhang, P., & Liu, Y. (2017). Hypoxia stimulates  neural  stem  cell  proliferation  by  increasing HIF-1α  expression  and  activating Wnt/β‐ catenin signaling. Cellular and Molecular Biology, 63(7), 12.  Quigley, I. K., & Kintner, C. (2017). Rfx2 Stabilizes Foxj1 Binding at Chromatin Loops to Enable  108      Multiciliated Cell Gene Expression. PLoS Genetics, 13(1), 1–29.  Raudvere, U., Kolberg, L., Kuzmin, I., Arak, T., Adler, P., Peterson, H., & Vilo, J. (2019). g:Profiler:  a web  server  for  functional  enrichment  analysis  and  conversions  of  gene  lists  (2019  update). Nucleic Acids Research, 1.  Rhee, C.‐S., Hong, S. K., Min, Y.‐G., Lee, C. H., Lee, K. S., Ahn, S.‐H., Park, K.‐S., & Yi, W. J. (1999).  Effects of  IL‐1β, TNF‐α, and TGF‐β on Ciliary Beat Frequency of Human Nasal Ciliated  Epithelial Cells in Vitro. American Journal of Rhinology, 13(1), 27–30.  Rothhammer, V., & Quintana, F. J. (2019). The aryl hydrocarbon receptor: an environmental  sensor  integrating  immune  responses  in  health  and  disease.  Nature  Reviews  Immunology, 19(3), 184–197.  Saucedo, L., Buffa, G. N., Rosso, M., Guillardoy, T., Góngora, A., Munuce, M. J., Vazquez‐Levin,  M. H., & Marín‐Briggiler, C. (2015). Fibroblast growth factor receptors (FGFRs) in human  sperm: Expression, functionality and involvement in motility regulation. PLoS ONE, 10(5),  1–17.  Shen, D., Liu, Y., Liu, Y., Wang, T., Yuan, L., Huang, X., & Wang, Y. (2021). Long non‐coding RNA  EWSAT1 promoted metastasis and actin cytoskeleton changes via miR‐24‐3p sponging in  osteosarcoma. Journal of Cellular and Molecular Medicine, 25(2), 716–728.  Spix, J. K., Chay, E. Y., Block, E. R., & Klarlund, J. K. (2007). Hepatocyte growth factor induces  epithelial cell motility through transactivation of the epidermal growth factor receptor.  Experimental Cell Research, 313(15), 3319–3325.  Tang, Z., Kang, B., Li, C., Chen, T., & Zhang, Z. (2019). GEPIA2: an enhanced web server for  large‐scale expression profiling and interactive analysis. Nucleic Acids Research, 47(W1),  W556–W560.  Thi Thanh Thuy, L., Morita, T., Yoshida, K., Wakasa, K., Iizuka, M., Ogawa, T., Mori, M., Sekiya,  Y., Momen, S., Motoyama, H., Ikeda, K., Yoshizato, K., & Kawada, N. (2011). Promotion of  liver  and  lung  tumorigenesis  in DEN‐treated  cytoglobin‐deficient mice.  The American  Journal of Pathology, 179(2), 1050–1060.  Thiery,  J.  P.,  Acloque, H., Huang,  R.  Y.  J., & Nieto, M.  A.  (2009).  Epithelial‐Mesenchymal  Transitions in Development and Disease. Cell, 139(5), 871–890.  109      Tözser, J., Earwood, R., Kato, A., Brown, J., Tanaka, K., Didier, R., Megraw, T. L., Blum, M., &  Kato, Y. (2015). TGF‐β Signaling Regulates the Differentiation of Motile Cilia. Cell Reports,  11(7), 1000–1007.  Villa, M., Crotta,  S., Dingwell, K.  S., Hirst, E. M. A., Gialitakis, M., Ahlfors, H.,  Smith,  J. C.,  Stockinger, B., & Wack, A.  (2016). The aryl hydrocarbon  receptor controls cyclin O  to  promote epithelial multiciliogenesis. Nature Communications, 7, 1–11.  Vlahov, N., Scrace, S., Soto, M. S., Grawenda, A. M., Bradley, L., Pankova, D., Papaspyropoulos,  A., Yee, K. S., Buffa, F., Goding, C. R., Timpson, P., Sibson, N., & O’Neill, E. (2015). Alternate  RASSF1 transcripts control SRC activity, E‐cadherin contacts, and YAP‐mediated invasion.  Current Biology, 25(23), 3019–3034.  Wakefield, A., Piggott, L., Croston, D., Jiang, W., & Clarkson, R. (2008). Suppression of the NF‐ κB  cofactor  Bcl3  inhibits mammary  epithelial  cell  apoptosis  and,  in  breast  tumours,  correlates with poor prognosis. Breast Cancer Research, 10(S2), O4.  Wakefield, A., Soukupova, J., Montagne, A., Ranger, J., French, R., Muller, W. J., & Clarkson, R.  W. E. (2013). Bcl3 selectively promotes metastasis of ERBB2‐driven mammary tumors.  Cancer Research, 73(2), 745–755.  Wang, L., Tassiulas,  I., Park‐Min, K. H., Reid, A. C., Gil‐Henn, H., Schlessinger,  J., Baron, R.,  Zhang,  J.  J., &  Ivashkiv,  L. B.  (2008).  “Tuning” of  type  I  interferon‐induced  Jak‐STAT1  signaling by calcium‐dependent kinases in macrophages. Nature Immunology, 9(2), 186– 193.  Wang, S., Liu, A., Wu, G., Ding, H. F., Huang, S., Nahman, S., & Dong, Z. (2018). The CPLANE  protein  Intu protects kidneys  from  ischemia‐reperfusion  injury by  targeting STAT1  for  degradation. Nature Communications, 9(1), 1–11.  Wang, T., Li, Y., Wang, W., Tuerhanjiang, A., Wu, Z., Yang, R., Yuan, M., Ma, D., Wang, W., &  Wang, S. (2014). Twist2, the key Twist isoform related to prognosis, promotes invasion  of  cervical  cancer  by  inducing  epithelial‐mesenchymal  transition  and  blocking  senescence. Human Pathology, 45(9), 1839–1846.  Weiterer, S., Schulte, D., Müller, S., Kohlen, T., Uhle, F., Weigand, M. A., & Henrich, M. (2014).  Tumor necrosis factor alpha induces a serotonin dependent early increase in ciliary beat  110      frequency and epithelial transport velocity in murine tracheae. PLoS ONE, 9(3), 1–10.  Wittenberg,  J.  B., & Wittenberg,  B.  A.  (2003). Myoglobin  function  reassessed.  Journal  of  Experimental Biology, 206(12), 2011–2020.  Xie,  H.,  Pallero, M.  A., Gupta,  K.,  Chang,  P., Ware, M.  F., Witke, W.,  Kwiatkowski,  D.  J.,  Lauffenburger, D. A., Murphy‐Ullrich, J. E., & Wells, A. (1998). EGF receptor regulation of  cell motility: EGF induces disassembly of focal adhesions independently of the motility‐ associated PLCγ signaling pathway. Journal of Cell Science, 111(5), 615–624.  Xu, H. W., Huang, Y. J., Xie, Z. Y., Lin, L., Guo, Y. C., Zhuang, Z. R., Lin, X. P., Zhou, W., Li, M.,  Huang, H. H., Wei, X. L., Man, K., & Zhang, G. J. (2013). The expression of cytoglobin as a  prognostic factor in gliomas: a retrospective analysis of 88 patients. BMC Cancer, 13, 247.  Yassin, M., Kissow, H., Vainer, B., Joseph, P. D., Hay‐Schmidt, A., Olsen, J., & Pedersen, A. E.  (2018).  Cytoglobin  affects  tumorigenesis  and  the  expression  of  ulcerative  colitis‐ associated genes under chemically  induced colitis  in mice. Scientific Reports 2018 8:1,  8(1), 1–16.  Zhang, B.,  Liu, Y.,  Li, Y.,  Zhe, X.,  Zhang,  S., &  Zhang,  L.  (2018). Neuroglobin promotes  the  proliferation and  suppresses  the apoptosis of glioma  cells by activating  the PI3K/AKT  pathway. Molecular Medicine Reports, 17(2), 2757–2763.  Zhang, J., Lan, S. J., Liu, Q. R., Liu, J. M., & Chen, X. Q. (2013). Neuroglobin, a Novel Intracellular  Hexa‐Coordinated Globin, Functions as a Tumor Suppressor in Hepatocellular Carcinoma  via Raf/MAPK/Erk. Molecular Pharmacology, 83(5), 1109–1119.  Zhang, R., Li, J. B., Yan, X. F., Jin, K., Li, W. Y., Xu, J., Zhao, J., Bai, J. H., & Chen, Y. Z. (2018).  Increased  EWSAT1  expression  promotes  cell  proliferation,  invasion  and  epithelial‐ mesenchymal  transition  in  colorectal  cancer.  European  Review  for  Medical  and  Pharmacological Sciences, 22(20), 6801–6808.  Zhou, Y., Zhou, B., Pache, L., Chang, M., Khodabakhshi, A. H., Tanaseichuk, O., Benner, C., &  Chanda, S. K. (2019). Metascape provides a biologist‐oriented resource for the analysis  of systems‐level datasets. Nature Communications, 10(1).      111          112      2.4 Androglobin expression pattern in basal metazoans confirms its conserved  functional association with cilia    Carina Osterhof,  : Expression analysis of Androglobin at  the root of the metazoan tree (unpublished)  Supplementary files: electronic Appendix  Own contributions to this publication:  ‐ Mining of literature and public repositories to assess the data basis and choose suitable  organisms for comparative analysis  ‐ Quality filtering, mapping, and quantification of bulk RNA‐Seq data  ‐ Quality filtering, pre‐processing and cluster analysis of count matrices from single cell  sequencing data; annotation of cell identity via marker genes and literature analysis  ‐ Differential expression analysis of various sample types (bulk RNA‐Seq data, single cell  clusters) and correlation analysis of gene expression across cell populations  ‐ Gene ontology enrichment analysis of gene lists   ‐ Pseudotime  calculation  and  analysis  of  developmental  trajectories  in  embryonal  datasets  ‐ Visualisation of results (together with  )    Project planning, analysis, and  interpretation of data as well as wording of  the manuscript  were  realized  together with Prof. Dr. T. Hankeln. The project was managed by Prof. Dr. T.  Hankeln.       113          114      Androglobin  expression  pattern  in  basal  metazoans  confirms  its  conserved  functional  association with cilia  Carina Osterhof1,     and Thomas Hankeln1  1Institute of Organismic and Molecular Evolution, Molecular Genetics and Genome Analysis,  Johannes Gutenberg University, Mainz, Germany  *correspondence: hankeln@uni‐mainz.de      115      Abstract  Androglobin  is  the most distinctive member of  the globin protein  family:  its characteristic  globin domain  is embedded within a  large protein of 1500 amino acids, which additionally  contains a calpain protease domain and several other domains of currently unknown function.  In addition, the globin domain is permutated and interrupted by a calmodulin binding motif.  The Adgb gene/protein  is highly conserved across the metazoan phylogenetic tree. Initially,  mammalian Adgb was thought to be mainly expressed in testis and – at a lower level – in lung  tissue, but we could recently show that it is also found in the female reproductive tract. In all  these tissues, Adgb is expressed in cell types carrying either motile cilia or flagella. Analysis of  Adgb KO mice also revealed a pivotal function of Adgb during late stages of spermatogenesis.   Since  it  is difficult to observe motile cilia directly  in mammalian tissues, we studied Adgb’s  expression pattern in several other model organisms for developmental studies. Specifically,  we analysed publicly available bulk and single cell mRNA sequencing data from a variety of  taxa at the root of the metazoan tree. We confirmed that not only the Adgb gene structure,  but also  its mRNA expression pattern  is highly  conserved – as  is  the case  for many ciliary  proteins.  Androglobin was  expressed  in  different  types  of  ciliated  or  flagellated  cells,  for  example on the surface of embryos in sea urchins, in comb cells of the sea walnut, in ciliated  cells of the digestive tract of the starlet sea anemone and of course, in sperm cells of sponges.  The morphological  similarities  between  choanocytes,  a major  expression  site  of  Adgb  in  sponges,  and  choanoflagellates  could  indicate  that  also  these  unicellular  organisms  are  a  suitable model. All these organisms, either isolated or as a combination, can now be used to  further elucidate  the exact  function of Adgb  in  flagellar motility and biosynthesis, but we  propose  to especially  focus on  the  cnidarian Nematostella vectensis,  since  it  is genetically  manipulable and embryos as well as adults are translucent and thus easy to image.      116      Introduction  Globins are a family of small metalloproteins, which share a characteristic α‐helical protein  structure and an iron atom‐containing heme prosthetic group (Kendrew et al., 1958; Dickerson  & Geis, 1983; Bolognesi et al., 1997). In vertebrates, the best studied globins are the oxygen‐ supplying respiratory proteins hemoglobin (Hb), present in erythrocytes, and myoglobin (Mb),  which  is mainly  found  in  skeletal and heart muscle  tissue.  In  the past 20  years, however,  several  additional  representatives  of  the  gene  family were  discovered  in many metazoan  species:  neuroglobin  (Ngb)  in  neurons  of  the  central  and  peripheral  nervous  system  of  vertebrates (Burmester et al., 2000; Reuss et al., 2002) has possibly a neuroprotective function  (Van Acker et al., 2018); cytoglobin (Cygb), which  is present  in all vertebrate classes and  is  expressed in extracellular matrix‐producing cell types, neurons and melanocytes of mammals  (Kawada et al., 2001; Burmester et al., 2004; Fujita et al., 2014); Globin X (GbX) which exists in  fish,  amphibians  and  reptiles  and may  possibly  protect membranes  from  reactive  oxygen  species (Blank et al., 2011; Koch & Burmester, 2016); Globin E (GbE) as an respiratory protein  that ensures oxygen supply in the retina of birds and turtles (Rösner et al., 2005; Schwarze et  al., 2015) and possibly in lungfish oocytes (Lüdemann et al., 2019);  Globin Y (GbY) as a globin  type of unknown  function,  limited  to  fish, amphibians,  reptiles and monotreme mammals  (Fuchs et al., 2006; Schwarze & Burmester, 2013). The substantial amount of work on these  novel family members in the past years has led to the conclusion that globins are expressed  in many different tissues and cell types where they may fulfill additional, non‐classical cellular  functions  besides  oxygen  transport  and  storage,  such  as  the  protection  against  reactive  oxygen  or  nitrogen  species,  the  prevention  of  hypoxia‐triggered  apoptosis,  and/or  the  involvement in intracellular signalling processes (Burmester & Hankeln, 2014; Keppner et al.,  2020).  In 2012, androglobin (Adgb) was identified as the newest and ‐at the same time‐ most peculiar  member  of  the  vertebrate  globin  protein  family  (Hoogewijs  et  al.,  2012). Adgb  is  a  large  chimeric protein of about 1500 amino acids, which contains an embedded globin domain. This  globin  domain  is  permutated  with  respect  to  its  characteristic  alpha  helices,  and  it  is  interrupted by a calmodulin binding motif. Adgb was initially identified bioinformatically in the  echinoderm  species  Strongylocentrotus  purpuratus  and  the  cephalopod  Branchiostoma  floridae.  Subsequent phylogenetic analysis  revealed  that Adgb homologues exist  in nearly  117      every branch of the metazoan tree. Orthologous copies of the Adgb gene could be found in  basal metazoan taxa such as the cnidarian Nematostella vectensis (Cnidaria), the placozoan  Trichoplax  adherens  (Placozoa)  and  even  in  the  choanoflagellate  Monosiga  brevicollis  (Hoogewijs et al., 2012), which suggests an elementary and possibly conserved function for  Adgb  in animals. The gene name hints at the predominant expression  in mammalian testis  tissue, where  lower  levels of Adgb could be correlated with impaired fertility (Hoogewijs et  al., 2012). In addition, we could recently show that Adgb is expressed in multi‐ciliated cells of  the mammalian lung, brain and the female reproductive tract (Koay et al., 2021).  Upon ablation of flagellar genes, severe phenotypes can observed: A study in Adgb KO‐mice  confirmed  its  pivotal  role  in  reproduction,  showing  that  Adgb  ablation  leads  to  fertility  impairments  and  structural  changes  in  sperm morphology  (Keppner  et  al.,  2022).    These  fertility impairments greatly aggravate breeding of these animals and complicate functional  studies. In addition, motile ciliary defects in lungs or other tissues are difficult to study directly  in mammals,  since  beat  frequencies  need  to  be monitored  life  (Jing  et  al.,  2017)  or  in  sophisticated  cell  culture  systems  (Lee  et  al.,  2020).  Due  to  Adgb  gene  loss  events  in  Caenorhabditis  elegans  and  Drosophila  melanogaster,  these  two  non‐mammalian model  systems are not suitable for the functional analysis of Adgb as well. Both spermatogenesis and  ciliogenesis are extremely conserved processes (Mitchell, 2007).  We therefore examined the  expression pattern of Adgb  in  taxa at  the  root of  the metazoan  tree, which have  recently  emerged as new model systems so study highly conserved processes such as regeneration or  development (e.g. Nematostella vectensis, Röttinger, 2021). Here, we will test whether the  gene  structure and distinct expression of Adgb  is  conserved. We  intend  to  identify model  organisms which could facilitate unravel Adgb’s role in the development or maintenance of  metazoan ciliary and flagellar structures.       118      Material and Methods  Data acquisition  For this study, we only used publicly available transcriptomic data from previously published  studies. Raw data  for RNA‐Seq experiments was downloaded  from  the EBI2 or NCBI3 web  servers  in  fastq‐format.  Read  count  matrices  of  single‐cell  RNA‐Seq  data  were  also  downloaded  from  these  repositories  or  taken  from  the  supplementary  data  of  the  corresponding publication. The full list of datasets can be found in Supplementary file 1.  Processing of RNA‐Sequencing data  Quality scores of RNA‐Seq data were evaluated using fastqc4. Adapter trimming and quality  filtering  was  performed  using  bbduk  from  the  BBtools  suite5.  Parameters  were  chosen  depending on the sequence length and overall quality of the dataset and selected mode for  trimming was retained for all datasets of the same study. Reads were then mapped against  the corresponding reference genome (Supplementary table 1) with HISAT2 (Kim et al., 2019)  and expression levels quantified with StringTie as trancripts per million (TPM) (Pertea et al.,  2015).   Differential gene expression and ontology analysis  Differential gene expression analysis was performed in R using Bioconductor package DESeq2  (Love et  al., 2014). Gene ontology enrichment  analysis was  subsequently performed with  g:Profiler (Raudvere et al., 2019).   Processing of single cell RNA‐sequencing data  Single cell sequencing data were downloaded as count matrices  from the EBI or NCBI web  servers. Quality filtering and cluster analysis were performed  in R with Seurat 4 (Hao et al.,  2021).  Cell  identities were  assigned  via  expressed marker  genes,  in  accordance with  the  original publications. Differential expression analysis was performed using Seurat’s build‐in  “find.markers”  function  with  default  parameters.  Pearson  correlation  coefficients  were  calculated  from  expression  matrices.  Gene  ontology  enrichment  was  performed  using    2 https://www.ebi.ac.uk/ena/browser/  3 https://www.ncbi.nlm.nih.gov/sra  4 https://www.bioinformatics.babraham.ac.uk/projects/fastqc/  5 https://sourceforge.net/projects/bbmap/  119      g:Profiler. A subset of datasets or single clusters were subjected to pseudotime and trajectory  analysis using Monocle 3 (Qiu et al., 2017).         120      Results and Discussion  Data selection  To study the expression profile of Adgb in basal metazoans, we first identified organisms that  were already  studied extensively via  transcriptomics due  to  their  important  role as model  systems. We  searched  the  SRA  database  for  organisms,  for which  staged  developmental  profiles were available. Thus, we narrowed down the time point of highest Adgb expression  during development. We also focused on datasets from untreated wildtype animals to exclude  possible influences thereof . Whenever possible, we analysed bigger datasets which included  multiple  time  points  or  replicates  from  one  working  group  only  to  minimize  technical  variation.  We included mRNA ‐Seq runs of isolated tissues from adult stages, to study spatial  expression differences in these taxa. Finally, to study which cell type(s) Adgb mRNA expression  emanates from, we compared the results from bulk analysis to single cell mRNA‐Seq profiles.  Meeting  these  requirements,  we  analysed  RNA‐Seq  data  from  the  following  taxa:  Amphimedon  queenslandica  (Porifera),  Mnemiopsis  leidyi  (Ctenophora),  Nematostella  vectensis (Cnidaria), and Strongylocentrotus purpuratus (Echinoderm) (the full list of datasets  can  be  found  in  Supplementary  file  X).  A  phylogenetic  tree  describing  the  evolutionary  relationship between these taxa is given in Figure 1 (adapted and modified from: Sebé‐Pedrós  et al., 2017).    Figure 1: Phylogenetic  tree of basal metazoan  taxa  that  carry  the Adgb gene. Note  that  the exact position of Porifera and Ctenophora  in the metazoan clade  is still a matter of debate. Adapted and  modified from: Sebé‐Pedrós et al., 2017  121      In  the  Porifera  Amphimedon  queenslandica,  Adgb  is  associated  with  fully  differentiated  ciliated cells  Porifera are a basal metazoan taxon and consist of only a limited number of cell types. First,  we  reanalysed  transcriptomic  data  on  isolated  cell  populations which  are  thought  to  be  homologous to cell types present  in the  last common ancestor of multicellular animals and  unicellular holozoans (Sogabe et al., 2019): archeocytes, amoeboid and pluripotent precursor  cells;  choanocytes,  feeding  cells  that  propel water  via  the  characteristic  flagellated  collar  complex; and pinacocytes, epithelial cells which form the inner and outer lining of the sponge  (Funayama, 2013). With the only exception of one archeocyte sample, Adgb expression was  exclusively detected  in  choanocytes  (Figure 2A).  Intriguingly, only 5 out of 10  choanocyte  samples were positive  for Adgb mRNA, 4 of which were derived  from the same  individual.  Since choanocytes may dedifferentiate into the more stem‐cell like archeocytes (Nakanishi et  al.,  2014;  Sogabe  et  al.,  2016),  these  inter‐individual  differences  could  suggest  deviating  differentiation stages of these cells.  We  then analysed  single cell mRNA  sequencing data derived  from adult A. queenslandica.  Identities of clusters were derived from marker gene expression as described in the original  publication  (Sebé‐Pedrós et al., 2018). Adgb mRNA was expressed  in  sperm  cells and  in a  second cell cluster termed “collagen cells” (Figure 2B). To assess patterns of co‐expression, we  performed Pearson correlation analysis of Adgb and all other expressed genes from the global  transcriptomic data. GO‐enrichment analysis of the TOP100 correlated genes returned several  highly  significant  terms,  such  as  cilium  organization,  assembly,  and  motility  (full  list:  Supplementary  file  2),  reflecting  the  high  expression  in  sperm  cells. We  then  calculated  Pearson  correlation  coefficients  for  all  genes  of  the  second  Adgb  expressing  cluster.  Surprisingly, subsequent GO‐term analysis again showed a strong enrichment for cilia‐related  terms in these gene lists, suggesting a second ciliated cell type that does not cluster with the  first (Supplementary file 3, Figure 2D). Terms such as “dynein complex” and “motile cilium”  even suggested the presence of moving cilia in these cells. Due to the elongated shape of the  cell in the UMAP cluster and the more prominent Adgb expression towards the edge of the  cluster, we  assumed  an  underlying  developmental  process with  gradual  changes  in  gene  expression. Therefore, we computed trajectories and pseudotime values with Monocle 3 (Qiu  et  al.,  2017).  The  cells  from  the  collagen  cluster  could  indeed  be  arranged  along  a  122      developmental trajectory (Figure 2C), where Adgb expression was higher in cells with higher  pseudotime values. We also plotted for comparison the expression of a potential radial spoke  head gene homolog, which encodes a  constituent of motile  cilia, and we  found  the  same  distribution as for Adgb. This indicated that the collagen cluster represents cells differentiating  into a cell type carrying motile cilia and that Adgb mRNA expression was higher towards the  end of this differentiation process. In other demosponges closely related to A. queenslandica,  a rare cell type was described, which synthesizes high amounts of collagen and is also motile:  the lophocyte (reviewed in: Ehrlich et al., 2018; Garrone, 1985). The described collagen cell  cluster may thus represent lophocyte cells, however, due to scarce literature we were unable  to  identify additional potential markers for validation. In contrast to what was found  in the  bulk sequencing data, Adgb was only expressed in very few cells of the choanocyte cluster in  the single cell dataset, which could reflect the inter‐individual variability in gene expression  seen  before.  However,  the  cellular  assignments  of  the  bulk  RNA‐Seq  data  relied  on  morphological characterization whereas the single cell data were classified via marker gene  expression.  These  datasets  could  therefore  represent  differing  subsets  or  subtypes  of  choanocytes.   Finally, we analysed the expression of Adgb mRNA during development of A. queenslandica  (Figure 2E). Adgb mRNA levels rose during early embryogenesis, peaked in the spot stage and  then fell rapidly. Adgb was weakly expressed in precompetent larvae, but not in competent  and post‐competent stages. Expression levels rose again in juvenile and adult animals, as also  reflected  by  the  single  cell  sequencing  analysis.  In  the  second  study,  we  reanalysed  transcriptomic  profiles  from  single  embryos.  The  bulk mRNA  samples were  clustered  and  visualised  in  dimensionally  reduced  embeddings  using  Monocle3  and  subjected  to  pseudotime  and  developmental  trajectory  analysis  (Figure  3F).  Competent  and  post‐ settlement  larval  stages were  assigned  using  differentially  expressed  genes  from  another  transcriptomic  study  (Conaco  et  al.,  2012),  to  determine  the  correct  direction  of  the  developmental trajectory. Also in this single embryo dataset, Adgb expression peaked during  early development (red circle), with residual expression in (pre‐)competent larvae. GO term  enrichment  analysis  performed  on  the  top  correlated  genes  showed  again  that  Adgb  expression  coincides  with  genes  enriched  for  ciliary  GO  terms,  as  exemplified  by  the  expression pattern of radial spoke head homolog 3 (Supplementary file 4). Morphological data  confirmed this association: the  first ciliated cells appear after cleavage  in A. queenslandica  123      embryos. During the spot stage, pigment cells accumulate at the posterior pole (the pigment  spot) and  the  ciliated  cells migrate and  form an outer  ciliated  layer  (Degnan et al., 2005;  Borisenko  et  al.,  2019).  Studies  in  two  demosponges  (Reneira  sp.  and  A.  queenslandica)  showed that after settlement, when Adgb expression drops to zero in our analysis, these cells  resorb their cilia, wander  inwards and transdifferentiate  into choanocytes  (Leys & Degnan,  2002; Sogabe et al., 2016).   In  conclusion,  Adgb  mRNA  was  expressed  in  a  variety  of  ciliated  cells  from  different  developmental stages in sponges.                 ↓Figure  2:  Expression  of  Adgb  mRNA  in  different  tissues  and  developmental  stages  of  A.  queenslandica.  (A) Bulk mRNA sequencing data on  isolated cell populations show high amounts of  Adgb  in  Choanocytes,  but  not  in  every  individual.  (B)  single  cell  sequencing  analysis  of  an  adult  specimen detected Adgb mRNA in sperm cells and collagen cells. (C) Pseudotime analysis on collagen  cell cluster shows an underlying developmental profile, with higher expression of Adgb and the radial  spoke head  protein  3  homolog  as  a  ciliary marker  in  cells with higher pseudotime.    (D) GO  term  enrichment  analysis  on  the  TOP100  genes  from  the  collagen  cell  cluster  with  highest  Pearson  correlation  coefficient  to  Adgb  showed  high  occurrence  of  terms  related  to motile  ciliogenesis,  indicating that this cluster rather represented another subtype of ciliated cells. Terms with a size ≤ 5  were  excluded  from  visualisation  (E)  Adgb mRNA  expression  derived  from  bulk  RNA  sequencing  datasets  during  development  of  A.  queenslandica.  Adgb  expression  increased  during  early  development until it peaks in the “spot” stage and decreased again shortly after. (F) Adgb expression  in  transcriptomes  from  single  embryos  during  development  of  A.  queenslandica.  Emb  –  embryogenesis; cl – competent larvae; pl – post‐competent larvae.  124            125      Adgb is associated with cilia in a variety of tissues and developmental stages of Mnemiopsis  leidyi  First, we analysed Adgb mRNA expression during early development of Mnemiopsis  leidyi.  Adgb mRNA expression was lower in early stages and rose – though not in all replicates – from  6 hours post fertilization (hpf) onwards.  In 1 replicate from the 9hpf timepoint, expression  rose as high as 92 TPM. After 9 hours of development, the first ciliary bands appear on the M.  leidyi embryos and thus coincide with the surge in Adgb mRNA expression.   We  then  reanalysed  RNA‐sequencing  data  on  dissected  comb  rows  and  tentacle  bulbs  (Babonis  et  al.,  2018)  as  wells  as  statocysts  (PRJNA787267,  unpublished)  from  adult  specimens. Adgb mRNA expression  in the ciliated comb rows was ten times higher than  in  tentacle bulbs and three times higher than  in statocysts (Figure 3B). Subsequent single cell  expression analysis of adult M.  leidyi showed a distinct cluster with a high amount of Adgb  mRNA expressing cells (Figure 3D). In accordance with the original publication, we identified  this cluster as comb cells. Lower amounts of Adgb could also be identified in second cluster  (hereafter  called: Adgb  cluster  2),  corresponding  to  clusters  from  the  original  publication  which were not assigned a clear cell type  identity. Gene ontology analysis on marker genes  from  this  cluster  showed  again  a  strong  association  with  cilia  (Figure  3G,  full  list:  Supplementary file 5). A potential homolog of FoxJ1, the master transcription factor of motile  ciliogenesis that regulates Adgb  in mammals (Koay et al., 2021), was also expressed  in this  additional clusters (Figure 3F). Differential expression analysis of comb cells and Adgb cluster  2 with subsequent GO term analysis revealed that Adgb cluster 2 showed higher transcription  of  genes  related  to  ribosome  biogenesis  and  translation, whereas  there was  no  sign  for  elevated transcriptional activity in comb cells. This could mean that Adgb cluster 2 represents  a  precursor  of  comb  cells,  although  it  showed  no  proximity  to  the  comb  cell  cluster  in  dimensionality  reduced  plots  and  no  affinity  in  trajectory  analysis  (data  not  shown).  Alternatively, these cells may represent a different type of ciliated cells. Adgb expression  is  strongly linked to flagella biosynthesis in sperm cells of mammals (Keppner et al., 2022) and  could also be found in sperm cells of A. queenslandica. In the initial description of this M. leidyi  dataset,  the authors did not clearly  identify germ cell clusters. Therefore  it  is  tempting  to  speculate that Adgb cluster 2 might represent cells differentiating into sperm cells.   126            127      Figure 3: Adgb mRNA expression analysis in the Ctenophore Mnemiopsis leidyi. (A) schematic depiction  of relevant tissues in adult M. leidyi. (B) Adgb mRNA expression in adult tissues. Note that statocysts  were analysed in a different study and direct comparison of expression values should be performed  carfully. (C) Adgb mRNA expression during early stages of development. (D) Adgb mRNA expression in  single cell sequencing data from adult M.  leidyi. Adgb was expressed  in comb cells and a subset of  unknown cells (cf. E), and both clusters coexpress the ciliary transcription factor FoxJ1 (F). (G) GO term  enrichment  analyis  based  on marker  genes  from  the  unknown  Adgb  expressing  clusters  showed,  among others, strong enrichment for ciliary terms.       128      Expression pattern of Adgb in tissues of Nematostella vectensis and other cnidarians also hints  at ciliated cells  During  development  of  the  cnidarian  N.  vectensis,  Adgb  was  expressed  at  low  levels  throughout all stages.  In adult N. vectensis, Adgb mRNA expression was 10 times higher  in  testis tissue than in ovaries whereas there was only little expression in muscle tissue (Figure  4D). There was also moderate expression in mesenterial filaments, an organ associated with  digestion.  Expression was  again  higher  in male  individuals.  Since  the  gonads  are  in  close  proximity to these filaments, this could be due to residual testis tissue. However,  in female  individuals, mesenterial expression of Adgb mRNA was highest among the organs analysed.  Histological analysis of mesenterial filaments in another Anthozoan species, Lophelia pertusa,  described a ciliated cell type next to cnidocytes, mucocytes and secretory cells (Strömberg &  Östman, 2017). Either of those could possibly be the source of Adgb mRNA expression in N.  vectensis mesenteries.  In  any  case, Adgb mRNA  expression was  also  detected  in  isolated  nematosomes. These are small and mobile cellular aggregates containing cnidocytes and a  flagellated cell type, but no gland cells (Robson, 1957; Williams, 1979). Transcriptomic analysis  on cell  fractions enriched  for or devoid of cnidocytes  from Cnidarian  tentacles showed no  difference  in  Adgb  expression,  indicating  that  Adgb  is  not  expressed  in  these  cells  (Supplementary  Table  2,  Sunagar  et  al.,  2018).  The  higher  expression  in mesenteries  and  nematosomes is therefore most probably resulting from the presence of cilia in one or more  different cell types.   Unfortunately, we  could  not  validate  either  hypothesis  on  the  single‐cell  level. No  Adgb  expression  was  detected  in  adult  or  juvenile  samples  of  Nematostella  vectensis  (Supplementary Figure 1). These datasets were generated by the MARS Seq technique (Jaitin  et al., 2014), which possibly detects a lower amount of expressed genes in total and results in  a high amount of drop‐outs (Ziegenhain et al., 2017). Hence, if endogenous Adgb expression  was low, it would be difficult to detect. However, other single cell sequencing studies of adult  tissues such as A. queenslandica or M.  leidyi discussed above that were also based on this  protocol successfully detected Adgb mRNA expression. In a single cell sequencing study on the  adult polyp of Hydra vulgaris, Adgb mRNA expression was more abundant (Siebert et al., 2019;  Supplementary Figure 1C). The highest expression could be found in the male germ line cluster  (mgl1+2, upper  left  corner).  In  the  single‐cell  study  from N.  vectensis, no germ  cells were  129      identified, only the somatic portion of the gonad. In summary, although single cell evidence is  scarce,  Adgb  appears  as  predominantly  expressed  in male  germ  cells  of  cnidarians  and  hypothetically in ciliated cell types of the mesenteries.          Figure 4: Adgb expression analysis in bulk sequencing data of the cnidarian Nematostella vectensis. (A)  schematic overview of  tissues  in adult N. vectensis.  (B) Adgb mRNA  is constantly expressed during  development at low levels. (C) In larvae, Adgb mRNA expression is higher in aboral tissues than body  tissues in both early and later stages. (D) Adgb mRNA expression in adult N. vectensis tissues. Dotted  line separates the different studies that were analysed. Adgb expression is highest in male gonads and  mesenteries.  In  female  specimens, Adgb expression  is highest  in mesenteries. Lower  levels can be  found in nematosomes, whereas there is nearly no expression detectable in tentacles, muscle tissue  and female gonads.     130      Expression of Adgb in the echinoderm Strongylocentrotus purpuratus  The first study we revaluated focused on the early stages of S. purpuratus embryogenesis. The  amount of Adgb mRNA expression varied greatly during development of S. purpuratus (Figure  5D). Adgb was already maternally deposited  in unfertilized eggs. The mRNA then vanished  quickly within  the  first  few hours. With  the onset of embryonal  transcription, Adgb mRNA  levels rose again during the blastula stage until they peaked during gastrulation (~40 hpf). This  surge  in expression coincides with  the  formation of a ciliary  tuft at  the blastula  stage and  subsequent ciliation on the whole surface of the early gastrula embryo (Eldon et al., 1990;  Cameron et al., 1993). Adgb mRNA levels then remained elevated until pluteus stage. A second  study on the transcriptome the larval stages of S. purpuratus showed that Adgb mRNA levels  surged in the motile larval stages (Figure 5E). In addition, Adgb levels increased from the four  arm stage to vestibular invagination stage, where ciliary bands develop into so called epaulets,  which assist the  larvae  in feeding and swimming (Smith et al., 2008). Adgb expression then  dropped sharply at the end of metamorphosis when the sea urchin becomes sessile.  In the adult sea urchin, Adgb levels were higher in testis than in ovary tissue, possibly reflecting  an association with flagellar structures (Figure 5C). Highest expression  levels were found  in  the gut and low levels of expression in the axial gland or radial nerve (Figure 5B). Intriguingly,  histological  analysis  of  the  gut  from  sea  urchin  Strongylocentrotus  intermedius  showed  stratified epithelia, but no signs of cellular protrusions such as microvilli or cilia  (Hu et al.,  2021),  in  contrast  to,  for  example,  the  composition  of  mesenteries  in  N.  vectensis.  Unfortunately, no single cell sequencing data from adult S. purpuratus specimen was available  so that no definitive cell type allocation could be made here.     ↓ Figure 5: Adgb mRNA expression in different stages of Strongylocentrotus purpuratus development.  (A) schematic depiction and duration of stages  in development. Adgb mRNA expression analysis  in  different adult tissues (B) and gonads (C) showed high expression levels in testis tissue and the gut.  During development  (D), Adgb mRNA was deposited  in  the unfertilized egg. Expression  then drops  sharply. In later stages, Adgb mRNA levels increased again and were highest during gastrulation and in  the late pluteus. In later developmental stages, Adgb levels increased during further development of  the larvae and then dropped again upon onset of metamorphosis (E).     131            132      Conclusion  In mammals, Adgb is prominently expressed in late stages of spermatogenesis and in ciliated  epithelial cells of the female reproductive tract, lung, and brain (Koay et al., 2021). Our work  here shows that this expression pattern is highly conserved and that association of Adgb with  ciliated  cell  types  can  be  traced  back  to  the most  basal  representatives  of multicellular  organisms. Some expression modalities such as the strong association with fully differentiated  ciliated cells such as collagen cells from A. queenslandica show striking similarities with our  findings in mammals, indicating that also the function of Adgb is highly conserved across the  metazoan tree. Arguing for an even deeper phylogenetic origin, Adgb seems also to be present  in  the  ciliated protozoan Tetrahymena  thermophila  (Joachimiak et al., 2021), a unicellular  eukaryote related to Paramecia (Cavalier‐Smith, 1993). Whereas effects of Adgb ablation in  mammals are severe and lead to infertility and abnormal sperm morphology (Keppner et al.,  2022),  loss  of  Adgb  in  T.  thermophila  only  caused minor modification  of  ciliary motility  (Joachimiak et al., 2021). It is therefore advisable to expand the set of model organisms for  future functional studies of Adgb. Unfortunately, classical model organisms such as Drosophila  melanogaster or Caenorhabditis elegans are not suitable for this, since both taxa  lost their  copy of Adgb (Hoogewijs et al., 2012) as well as motile cilia on somatic cell types (Choksi et al.,  2014).  The  work  presented  here  can  help  identifying  suitable  taxa  for  future  functional  analysis. In the simple metazoan A. queenslandica, archeocytes differentiate into choanocytes  on the one hand and sperm cells on the other. Both differentiated cell types (and possibly the  somewhat elusive lophocytes as well) expressed Adgb and would therefore be an interesting  model to study Adgb’s regulatory mechanisms and to study whether Adgb expression impacts  cellular fate or is resulting thereof. However, there are no protocols established for genetic  manipulation  of  A.  queenslandica  and  also  for  no  other  closely  related  taxon  of  the  demosponge genus. N. vectensis, on the other hand, has been used  frequently as a model  system for regeneration (Amiel & Röttinger, 2021) and can also be genetically manipulated  (Ikmi et al., 2014; Karabulut et al., 2019). Since the expression data  in N. vectensis was not  fully conclusive, it would be interesting to perform in situ expression analysis on this organism  to  determine  the  cell  type  expressing  Adgb  in mesenteries  and  to  confirm  the  potential  expression  in  germ  cells.  S.  purpuraturus,  an  echinoderm  and  thus  sister  clade  to  the  vertebrates, could also be used to study Adgb’s function. It is an established model organisms  in ciliary research, but in comparison to A. queenslandica and N. vectensis it is more complex  133      S. purpuratus also lacks some intriguing properties such as N. vectensis’ translucency during  adulthood which greatly facilitates imaging and thus the analysis of internal cilia (DuBuc et al.,  2014). In summary, we propose to focus on N. vectensis as a model for future studies of Adgb’s  function.   It could also be desirable to study Adgb’s function not in a multicellular, but unicellular model  system. Flagellated choanocytes of  sponges, possibly a major expression  site of Adgb, are  structurally  very  similar  to  choanoflagellates.  Both  cells  are  ciliated  and  carry  a  collar  of  microvilli around the flagellum. Choanoflagellates are also the closest contemporary relatives  of multicellular animals and carry a putative Adgb gene (Hoogewijs et al., 2012).  It has been  proposed  that  multicellularity  itself  evolved  from  a  unicellular,  flagellated,  and  collared  ancestor  and  that  choanocytes  and  choanoflagellates  are homologous  to each other  (e.g.  Cavalier‐Smith, 2017; Nielsen, 2008). However, recent transcriptomic (Sogabe et al., 2019) and  morphological analysis (Mah et al., 2014) suggest that both cell types evolved convergently,  and that multicellularity rather is derived from transdifferentiating cells resembling modern,  pluripotent stem cells (Sogabe et al., 2019). Nonetheless, the structural resemblance and the  possibility of genetic manipulation (Hoffmeyer & Burkhardt, 2016) make choanoflagellates an  intriguing candidate to complement functional studies on Adgb in e.g. N. vectensis.   Here, we  could generate a  convincing hypothesis  that Adgb’s expression pattern  is highly  conserved over a variety of organisms relying only on publicly available bulk and single cell  sequencing data, without performing any additional wet lab experiments. The additional point  of view on these data even led to potential new classifications of unknown or poorly described  cell clusters. These types of reanalyses can generate new knowledge and help annotating cell  cluster identities from sequencing efforts, which do not have the resources to rely on expert  annotation  in  large  consortia,  such  as  the  Fly  Cell Atlas  (Li  et  al.,  2022). Now  these  new  hypotheses may be validated with appropriate wet lab experiments to confirm these cell type  identities to test whether Adgb can be used as a marker for ciliated and flagellated cells of  basal metazoans in vivo.   Acknowledgments  This work was supported by the German Research Foundation to T.H. (HA 2103/9‐1).      134      Supplementary table 2: List of organisms, their corresponding reference genomes and annotation versions used  in bulk RNA‐Seq analysis in this study.  Organism  Genome version  annotation  Amphimedon queenslandica  Aqu1  Aqu1.52 (gene set v2.1)  (GCA_000090795.1)  Mnemiopsis leidyi  MneLei_Aug2011  MneLei_Aug2011.52  (GCA_000226015.1)  Nematostella vectensis  ASM20922v1  ASM20922v1.52  (GCA_000209225.1)  Strongylocentrotus purpuratus  Spur_5.0  Spur_5.0.49  (GCA_000002235.4)        Supplementary  table  2:  Adgb  mRNA  expression  in  Cnidocytes,  remapped  and  quantified  from  study  PRJNA391807  by  Sunagar  et  al.,  2018.  There  is  no  detectable  difference  between  the  Cnidocyte‐positive,  ‐ superpositive and ‐negative samples.   RUN SAMPLE ID GROUP ADGB EXPRESSION (TPM) SRR5864937 Neg1 negative 4.466142 SRR6706292 Neg2 negative 7.61013 SRR6706323 Neg3 negative 7.219751 SRR6710013 Neg5 negative 5.05338 SRR6710014 Neg6 negative 7.842983 SRR6710017 Neg7 negative 5.67276 SRR6713982 Pos1 positive 2.161914 SRR6713983 Pos2 positive 4.428854 SRR6713984 Pos3 positive 2.994717 SRR6713985 Pos5 superpositive 5.155812 SRR6713986 Pos6 superpositive 2.35182 SRR6713987 Pos7 superpositive 2.752329     135          Supplementary figure 4: Single cell mRNA expression of Adgb in Cnidarians. Adgb mRNA was not enriched in any  of the cell clusters in (A) juvenile and (B) adult Nematostella vectensis, but was enriched in male germ cells of  the interstitial lineage from Hydra vulgaris (Siebert et al., 2019; Figure from Single cell atlas of Broad institute,  https://singlecell.broadinstitute.org/single_cell).    136      References  Amiel, A. R., & Röttinger,  E.  (2021).  Experimental  Tools  to  Study Regeneration  in  the  Sea  Anemone Nematostella vectensis. Methods in Molecular Biology, 2219, 69–80.  Babonis, L. S., DeBiasse, M. B., Francis, W. R., Christianson, L. M., Moss, A. G., Haddock, S. H.  D., Martindale, M. Q., &  Ryan,  J.  F.  (2018).  Integrating  Embryonic Development  and  Evolutionary  History  to  Characterize  Tentacle‐Specific  Cell  Types  in  a  Ctenophore.  Molecular Biology and Evolution, 35(12), 2940–2956.  Blank, M., Wollberg, J., Gerlach, F., Reimann, K., Roesner, A., Hankeln, T., Fago, A., Weber, R.  E., & Burmester, T. (2011). A membrane‐bound vertebrate globin. PLoS ONE, 6(9).  Bolognesi,  M.,  Bordo,  D.,  Rizzi,  M.,  Tarricone,  C.,  &  Ascenzi,  P.  (1997).  Nonvertebrate  hemoglobins:  Structural  bases  for  reactivity.  In  Progress  in  Biophysics  and Molecular  Biology (Vol. 68, Issue 1, pp. 29–68). Elsevier Ltd.  Borisenko, I., Podgornaya, O. I., & Ereskovsky, A. V. (2019). From traveler to homebody: Which  signaling mechanisms sponge larvae use to become adult sponges? Advances in Protein  Chemistry and Structural Biology, 116, 421–449.  Burmester, T., Haberkamp, M., Mitz, S., Roesner, A., Schmidt, M., Ebner, B., Gerlach, F., Fuchs,  C., & Hankeln, T.  (2004). Neuroglobin and Cytoglobin: Genes, Proteins and Evolution.  IUBMB Life, 56(11), 703–707.  Burmester,  T.,  &  Hankeln,  T.  (2014).  Function  and  evolution  of  vertebrate  globins.  Acta  Physiologica, 211(3), 501–514.  Burmester, T., Welch, B., Reinhardt, S., & Hankeln, T. (2000). A vertebrate globin expressed in  the brain. Nature, 407(6803), 520–523.  Cameron, R. A., Britten, R. J., & Davidson, E. H. (1993). The embryonic ciliated band of the sea  urchin,  Strongylocentrotus  purpuratus  derives  from  both  oral  and  aboral  ectoderm.  Developmental Biology, 160(2), 369–376.  Cavalier‐Smith, T. (1993). Kingdom protozoa and its 18 phyla. Microbiological Reviews, 57(4),  953.  Cavalier‐Smith,  T.  (2017).  Origin  of  animal  multicellularity:  precursors,  causes,  137      consequences—the choanoflagellate/sponge transition, neurogenesis and the Cambrian  explosion.  Philosophical  Transactions  of  the  Royal  Society  B:  Biological  Sciences,  372(1713).  Choksi,  S. P.,  Lauter, G.,  Swoboda, P., & Roy,  S.  (2014).  Switching on  cilia:  transcriptional  networks regulating ciliogenesis. Development, 141(7), 1427–1441.  Conaco, C., Neveu, P., Zhou, H., Arcila, M. L., Degnan, S. M., Degnan, B. M., & Kosik, K. S. (2012).  Transcriptome  profiling  of  the  demosponge  Amphimedon  queenslandica  reveals  genome‐wide events that accompany major life cycle transitions. BMC Genomics, 13(1),  1–19.  Degnan, B. M., Leys, S. P., & Larroux, C. (2005). Sponge Development and Antiquity of Animal  Pattern Formation. Integrative and Comparative Biology, 45(2), 335–341.  Dickerson, R. E., & Geis, I. (1983). Hemoglobin : structure, function, evolution, and pathology.  Benjamin/Cummings Pub. Co.  DuBuc, T. Q., Dattoli, A. A., Babonis, L. S., Salinas‐Saavedra, M., Röttinger, E., Martindale, M.  Q., & Postma, M. (2014). In vivo imaging of Nematostella vectensis embryogenesis and  late development using fluorescent probes. BMC Cell Biology, 15(1), 1–14.  Ehrlich,  H., Wysokowski, M.,  Zółtowska‐Aksamitowska,  S.,  Petrenko,  I.,  &  Jesionowski,  T.  (2018). Collagens of Poriferan Origin. Marine Drugs, 16(3).  Eldon, E. D., Montpetit, I. C., Nguyen, T., Decker, G., Valdizan, M. C., Klein, W. H., & Brandhorst,  B. P. (1990). Localization of the sea urchin Spec3 protein to cilia and Golgi complexes of  embryonic ectoderm cells. Genes & Development, 4(1), 111–122.  Fujita, Y., Koinuma, S., De Velasco, M. A., Bolz,  J., Togashi, Y., Terashima, M., Hayashi, H.,  Matsuo, T., & Nishio, K. (2014). Melanoma transition is frequently accompanied by a loss  of cytoglobin expression in melanocytes: A novel expression site of cytoglobin. PLoS ONE,  9(4).  Funayama, N. (2013). The stem cell system in demosponges: Suggested involvement of two  types  of  cells:  Archeocytes  (active  stem  cells)  and  choanocytes  (food‐entrapping  flagellated cells). Development Genes and Evolution, 223(1–2), 23–38.  138      Garrone, R. (1985). The Collagen of the Porifera. In Bairati, A. & R. Garrone (Eds.), Biology of  Invertebrate and Lower Vertebrate Collagens (pp. 157–175). Springer US.  Hao, Y., Hao, S., Andersen‐Nissen, E., Mauck, W. M., Zheng, S., Butler, A., Lee, M. J., Wilk, A.  J., Darby, C., Zager, M., Hoffman, P., Stoeckius, M., Papalexi, E., Mimitou, E. P., Jain, J.,  Srivastava, A., Stuart, T., Fleming, L. M., Yeung, B., … Satija, R. (2021). Integrated analysis  of multimodal single‐cell data. Cell, 184(13), 3573‐3587.e29.  Hoffmeyer, T. T., & Burkhardt, P. (2016). Choanoflagellate models — Monosiga brevicollis and  Salpingoeca rosetta. Current Opinion in Genetics & Development, 39, 42–47.  Hoogewijs, D., Ebner, B., Germani, F., Hoffmann, F. G., Fabrizius, A., Moens, L., Burmester, T.,  Dewilde, S., Storz, J. F., Vinogradov, S. N., & Hankeln, T. (2012). Androglobin: A Chimeric  Globin  in Metazoans That  Is Preferentially Expressed  in Mammalian Testes. Molecular  Biology and Evolution, 29(4), 1105–1114.  Hu, F., Chi, X., Yang, M., Ding, P., Yin, D., Ding, J., Huang, X., Luo, J., Chang, Y., & Zhao, C. (2021).  Effects of eliminating interactions in multi‐layer culture on survival, food utilization and  growth  of  small  sea  urchins  Strongylocentrotus  intermedius  at  high  temperatures.  Scientific Reports 2021 11:1, 11(1), 1–11.  Ikmi, A., McKinney, S. A., Delventhal, K. M., & Gibson, M. C. (2014). TALEN and CRISPR/Cas9‐ mediated  genome  editing  in  the  early‐branching  metazoan  Nematostella  vectensis.  Nature Communications, 5.  Jaitin, D. A., Kenigsberg,  E., Keren‐Shaul, H.,  Elefant, N., Paul,  F.,  Zaretsky,  I., Mildner, A.,  Cohen, N., Jung, S., Tanay, A., & Amit, I. (2014). Massively parallel single‐cell RNA‐seq for  marker‐free decomposition of tissues into cell types. Science (New York, N.Y.), 343(6172),  776–779.  Jing, J. C., Chen, J. J., Chou, L., Wong, B. J. F., & Chen, Z. (2017). Visualization and Detection of  Ciliary Beating Pattern and Frequency in the Upper Airway using Phase Resolved Doppler  Optical Coherence Tomography. Scientific Reports 2017 7:1, 7(1), 1–11.  Joachimiak, E., Osinka, A., Farahat, H., Świderska, B., Sitkiewicz, E., Poprzeczko, M., Fabczak,  H., & Wloga, D.  (2021). Composition and  function of  the C1b/C1f  region  in  the ciliary  central apparatus. Scientific Reports 2021 11:1, 11(1), 1–17.  139      Karabulut, A., He, S., Chen, C. Y., McKinney, S. A., & Gibson, M. C. (2019). Electroporation of  short hairpin RNAs for rapid and efficient gene knockdown in the starlet sea anemone,  Nematostella vectensis. Developmental Biology, 448(1), 7–15.  Kawada, N., Kristensen, D. B., Asahina, K., Nakatani, K., Minamiyama, Y., Seki, S., & Yoshizato,  K.  (2001). Characterization of a Stellate Cell Activation‐associated Protein  (STAP) with  Peroxidase Activity Found  in Rat Hepatic Stellate Cells. Journal of Biological Chemistry,  276(27), 25318–25323.  Kendrew, J. C., Bodo, G., Dintzis, H. M., Parrish, R. G., Wyckoff, H., & Phillips, D. C. (1958). A  three‐dimensional model of the myoglobin molecule obtained by x‐ray analysis. Nature,  181(4610), 662–666.  Keppner, A., Correia, M., Santambrogio, S., Koay, T. W., Maric, D., Osterhof, C., Winter, D. V,  Clerc, A., Stumpe, M., Chalmel, F., Dewilde, S., Odermatt, A., Kressler, D., Hankeln, T.,  Wenger, R. H., & Hoogewijs, D.  (2022). Androglobin, a chimeric mammalian globin,  is  required for male fertility. ELife, 11.  Keppner,  A., Maric,  D.,  Correia, M.,  Koay,  T. W., Orlando,  I. M.  C.,  Vinogradov,  S.  N., &  Hoogewijs,  D.  (2020).  Lessons  from  the  post‐genomic  era:  Globin  diversity  beyond  oxygen binding and transport. Redox Biology, 37, 101687.  Kim, D.,  Paggi,  J. M.,  Park, C., Bennett, C., &  Salzberg,  S.  L.  (2019). Graph‐based  genome  alignment  and  genotyping  with  HISAT2  and  HISAT‐genotype.  Nature  Biotechnology,  37(8), 907–915.  Koay, T. W., Osterhof, C., Orlando, I. M. C., Keppner, A., Andre, D., Yousefian, S., Alonso, M. S.,  Correia, M., Markworth, R., Schödel, J., Hankeln, T., & Hoogewijs, D. (2021). Androglobin  gene  expression  patterns  and  FOXJ1‐dependent  regulation  indicate  its  functional  association with ciliogenesis. Journal of Biological Chemistry, 296.  Koch, J., & Burmester, T. (2016). Membrane‐bound globin X protects the cell from reactive  oxygen  species. Biochemical  and  Biophysical  Research  Communications,  469(2),  275– 280.  Lee,  D.  D.  H.,  Petris,  A.,  Hynds,  R.  E.,  &  O’Callaghan,  C.  (2020).  Ciliated  epithelial  cell  differentiation  at  air‐liquid  interface  using  commercially  available  culture  media.  140      Methods in Molecular Biology (Clifton, N.J.), 2109, 275.  Leys,  S.  P., & Degnan, B. M.  (2002).  Embryogenesis  and metamorphosis  in  a  haplosclerid  demosponge: Gastrulation and transdifferentiation of larval ciliated cells to choanocytes.  Invertebrate Biology, 121(3), 171–189.  Li, H., Janssens, J., de Waegeneer, M., Kolluru, S. S., Davie, K., Gardeux, V., Saelens, W., David,  F.  P.  A.,  Brbić, M.,  Spanier,  K.,  Leskovec,  J., McLaughlin,  C. N.,  Xie, Q.,  Jones,  R.  C.,  Brueckner, K., Shim, J., Tattikota, S. G., Schnorrer, F., Rust, K., … Aerts, S. (2022). Fly Cell  Atlas: A single‐nucleus transcriptomic atlas of the adult fruit fly. Science, 375(6584).  Love, M.  I.,  Huber, W.,  &  Anders,  S.  (2014). Moderated  estimation  of  fold  change  and  dispersion for RNA‐seq data with DESeq2. Genome Biology, 15(12), 550.  Lüdemann,  J., Verissimo, K. M., Dreger, K., Fago, A., Schneider,  I., & Burmester, T.  (2019).  Globin E is a myoglobin‐related, respiratory protein highly expressed in lungfish oocytes.  Scientific Reports 2019 9:1, 9(1), 1–11.  Mah, J. L., Christensen‐Dalsgaard, K. K., & Leys, S. P. (2014). Choanoflagellate and choanocyte  collar‐flagellar  systems  and  the  assumption  of  homology.  Evolution  &  Development,  16(1), 25–37.  Mitchell, D. R.  (2007). The evolution of eukaryotic  cilia and  flagella as motile and  sensory  organelles. Advances in Experimental Medicine and Biology, 607, 130.  Nakanishi, N., Sogabe, S., & Degnan, B. M. (2014). Evolutionary origin of gastrulation: Insights  from sponge development. BMC Biology, 12(1), 1–9.  Nielsen,  C.  (2008).  Six major  steps  in  animal  evolution:  Are  we  derived  sponge  larvae?  Evolution and Development, 10(2), 241–257.  Pertea, M., Pertea, G. M., Antonescu, C. M., Chang, T. C., Mendell, J. T., & Salzberg, S. L. (2015).  StringTie  enables  improved  reconstruction  of  a  transcriptome  from  RNA‐seq  reads.  Nature Biotechnology, 33(3), 290–295.  Qiu, X., Mao, Q., Tang, Y., Wang, L., Chawla, R., Pliner, H. A., & Trapnell, C. (2017). Reversed  graph embedding resolves complex single‐cell trajectories. Nature Methods 2017 14:10,  14(10), 979–982.  141      Raudvere, U., Kolberg, L., Kuzmin, I., Arak, T., Adler, P., Peterson, H., & Vilo, J. (2019). g:Profiler:  a web  server  for  functional  enrichment  analysis  and  conversions  of  gene  lists  (2019  update). Nucleic Acids Research, 1.  Reuss, S., Saaler‐Reinhardt, S., Weich, B., Wystub, S., Reuss, M. H., Burmester, T., & Hankeln,  T.  (2002).  Expression  analysis  of  neuroglobin mRNA  in  rodent  tissues. Neuroscience,  115(3), 645–656.  Robson, E. A. (1957). A Sea‐Anemone from Brackish Water. Nature 1957 179:4563, 179(4563),  787–788.  Rösner,  A.,  Fuchs,  C.,  Hankeln,  T.,  &  Burmester,  T.  (2005).  A  Globin  Gene  of  Ancient  Evolutionary Origin  in Lower Vertebrates: Evidence for Two Distinct Globin Families  in  Animals. Molecular Biology and Evolution, 22(1), 12–20.  Röttinger,  E.  (2021).  Nematostella  vectensis,  an  Emerging  Model  for  Deciphering  the  Molecular and Cellular Mechanisms Underlying Whole‐Body Regeneration. Cells, 10(10).  Schwarze, K., Singh, A., & Burmester, T. (2015). The Full Globin Repertoire of Turtles Provides  Insights into Vertebrate Globin Evolution and Functions. Genome Biology and Evolution,  7(7), 1896–1913.  Sebé‐Pedrós, A., Chomsky, E., Pang, K., Lara‐Astiaso, D., Gaiti, F., Mukamel, Z., Amit, I., Hejnol,  A., Degnan, B. M., & Tanay, A. (2018). Early metazoan cell type diversity and the evolution  of multicellular gene regulation. Nature Ecology and Evolution, 2(7), 1176–1188.  Sebé‐Pedrós, A., Degnan, B. M., & Ruiz‐Trillo,  I. (2017). The origin of Metazoa: a unicellular  perspective. Nature Reviews Genetics 2017 18:8, 18(8), 498–512.  Siebert, S., Farrell, J. A., Cazet, J. F., Abeykoon, Y., Primack, A. S., Schnitzler, C. E., & Juliano, C.  E. (2019). Stem cell differentiation trajectories in Hydra resolved at single‐cell resolution.  Science, 365(6451).  Smith, M. M., Smith, L. C., Cameron, R. A., & Urry, L. A. (2008). The  larval stages of the sea  urchin, Strongylocentrotus purpuratus. Journal of Morphology, 269(6), 713–733.  Sogabe, S., Hatleberg, W.  L., Kocot, K. M., Say, T. E., Stoupin, D., Roper, K. E., Fernandez‐ Valverde, S. L., Degnan, S. M., & Degnan, B. M.  (2019). Pluripotency and the origin of  142      animal multicellularity. Nature 2019 570:7762, 570(7762), 519–522.  Sogabe, S., Nakanishi, N., & Degnan, B. M.  (2016). The ontogeny of choanocyte chambers  during metamorphosis in the demosponge Amphimedon queenslandica. EvoDevo, 7(1),  1–13.  Sunagar, K., Columbus‐Shenkar, Y. Y.,  Fridrich, A., Gutkovich, N., Aharoni, R., & Moran, Y.  (2018). Cell type‐specific expression profiling unravels the development and evolution of  stinging cells in sea anemone. BMC Biology, 16(1), 108.  Van Acker, Z. P., Luyckx, E., & Dewilde, S. (2018). Neuroglobin Expression in the Brain: a Story  of Tissue Homeostasis Preservation. Molecular Neurobiology 2018 56:3, 56(3), 2101– 2122.  Williams, R. B.  (1979). Studies on  the nematosomes of Nematostella vectensis Stephenson  (Coelenterata: Actiniaria). Journal of Natural History, 13(1), 69–80.  Ziegenhain, C., Vieth, B., Parekh, S., Reinius, B., Guillaumet‐Adkins, A., Smets, M., Leonhardt,  H., Heyn, H., Hellmann, I., & Enard, W. (2017). Comparative Analysis of Single‐Cell RNA  Sequencing Methods. Molecular Cell, 65(4), 631‐643.e4.                              143          144      3 Discussion and Outlook  Androglobin, a phylogenetically ancestral member of the globin  family,  is highly conserved  across the metazoan tree and predominantly expressed in mammalian testis tissue, with lower  levels of expression  in both  lung and brain tissue (Hoogewijs et al., 2012).  In this work, we  provided a detailed description of a novel expression site of Adgb, the female reproductive  tract. Adgb is expressed in fallopian tubes and endometrial tissue in humans, mice and cattle.   Via re‐analysis of publicly available bulk and single cell RNA‐Seq data we observed that Adgb  expression  in  lung,  brain  and  the  female  reproductive  tract  always  originates  from  fully  differentiated cells carrying motile cilia. As a result, Adgb presumably exerts the same function  in all  these organs and  this  function  is hypothetically also associated with  the presence or  development of motile cilia. In addition, we have identified potential regulatory elements for  the Adgb  gene  locus.  FoxJ1,  the master  transcription  factor  for motile  ciliogenesis,  is  co‐ expressed with Adgb. Our cooperation partners could subsequently proof that FoxJ1 can bind  to the Adgb promotor and induce its expression. Overexpression of Adgb also increases the  proportion of ciliated cells in vivo. Adgb could therefore exhibit a function in ciliogenesis itself  (Chapter 2.1; Koay et al., 2021).   In  testis  tissue,  Adgb  is  detected  in  round  and  elongating  spermatids  and  thus  closely  associated with the emergence of the flagellum, a special form of a cilium pivotal for motility  of sperm cells and reproductive success. Upon Adgb ablation, male mice suffer from infertility.  We have analysed the transcriptomic changes in Adgb KO testis tissue compared to wildtype  littermates and found a small, but distinct set of genes that suggest defects in late stages of  spermatogenesis but revealed no pathway in particular. Closer phenotypic analysis conducted  by  our  cooperation  partners  of  spermatids  via microscopy  showed  impaired maturation,  abnormal  sperm  shape  and  a  variety  of  ultrastructural  defects  in  organisation  of  both  microtubules and mitochondria. Mature sperm displayed a variety of flagellar deformations.  Mass spectrometry analysis of immunoprecipitated Adgb also conducted in the Hoogewijs lab  suggests an interaction with Sept10. The specific function of this protein is unknown, however,  it forms complexes with other members of the septin family which were already described as  vital for the proper formation of the connecting midpiece of sperm (Shen et al., 2020). The  interaction of Sept10 and Adgb was subsequently confirmed both in vivo and in vitro. Upon  ablation  of  Adgb,  Sept10  also  mislocalizes  in  sperm  cells.  Mechanistically,  Adgb  could  145      contribute to the proteolysis of Sept10, which is required for its activation. All in all, these data  suggest a pivotal role for Adgb in the maturation of structurally sound sperm cells (Chapter  2.2; Keppner et al., 2022).    In the initial publication from 2012, the Adgb gene was identified in nearly every branch of the  metazoan tree. Given its pivotal function in reproduction, we interrogated whether not only  the Adgb  gene,  but  also  its  expression  pattern  as well  as  the  association with  ciliated  or  flagellated  cells  is  conserved  beyond  vertebrates.  We  therefore  analysed  Adgb  mRNA  expression in bulk and single cell data from taxa at the root of the metazoan tree. Adgb was  detected in a variety of cell types carrying motile cilia or flagella, such as sperm cells (sponge),  comb cells (sea walnut), ciliated cells on the surface of embryos (sea urchin) as well as ciliated  cells from the digestive tract (sea anemone). Overall, these results show a strong conservation  of Adgb’s expression pattern and presumably also  its cilia‐association function  in metazoan  taxa (Chapter 2.4).   Many globin family members such as Mb or Cygb have previously been studied in the context  of cancer. Depending on the globin  family member as well as the cancer type  in question,  globins may have both oncogenic and protective function  in malignancies and can serve as  prognostic markers (see also Chapter 1.3). To analyse Adgb’s role in cancer, we have stably  overexpressed Adgb in a lung cancer cell model. Subsequent transcriptomic analysis showed,  among others, hallmarks of enhanced motility, which suggest an oncogenic effect of Adgb –  in line with the findings of other studies in brain and prostate cancer models. However, these  effects were not pronounced enough to be detectable in phenotypic motility assays. We then  analysed  a  wide  variety  of  transcriptomic  data  from  in‐vivo  cancer  samples.  Despite  its  presumable  oncogenic  role, we  found  that  Adgb  is  consistently  downregulated  in  cancer  entities  derived  from  tissues with  endogenous  Adgb  expression.  Adgb mRNA  levels  also  decreased with degree of malignancy and are not detectable  in metastases. We validated  these  findings  via  qPCR  in  an  independent  cohort  of  lung  and  testis  cancer  samples. We  concluded that while Adgb may have oncogenic properties in cell culture models, it is simply  not expressed  in vivo, such that Adgb undergoes sophisticated regulatory mechanisms that  silence  the gene upon dedifferentiation of  tissues. Adgb  is  thus not  suitable as prognostic  marker in cancer (Chapter 2.3).       146      3.1 Androglobin as part of the ciliary proteome  Androglobin is mainly expressed in cells carrying motile cilia or flagellar structures (Chapter  2.1;  Koay  et  al.,  2021)  and  this  expression  pattern  is  highly  conserved  throughout  the  metazoan tree (Chapter 2.4). A proteomic study of cilia  in a sea urchin (Strongylocentrotus  purpuratus), a  sea anemone  (Nematostella vectensis), and a choanoflagellate  (Salpingoeca  rosetta) suggested that Adgb is not only expressed in these ciliated cell types, but localizes to  the  cilium  itself  (Table  3,  Sigg  et  al.,  2017): mass  spectrometry  analysis  on  isolated  cilia  identified peptides derived from Adgb in all three organisms in nearly all replicates analysed  (exception: replicate 2 of S. purpuratus). The authors also subdivided cilia from S. purpurates  into an axonemal fraction and one containing the membrane and matrix of the cilium before  analysis via mass spectrometry. Adgb counts tend to be higher  in the axonemal fraction of  isolated cilia. In S. rosetta, Adgb was even exclusively found in ciliary samples, whereas it was  not detectable in the cell bodies or the microvilli.    Table 3: Adgb peptide counts  in proteomic studies of  isolated cilia  (from: Sigg et al 2017). Protein  composition of motile cilia  from S. purpuratus, N. vectensis, and S. rosetta were analysed via mass  spectrometry. Peptides derived from Adgb protein could be identified in cilia from all three organisms.  No peptides were identified in the cell bodies and microvilli from S. rosetta.   Organism (Adgb acc.) →  S. purpuratus  N. vectensis  S. rosetta  Analysed cell fraction ↓  (W4XW92)  (A7RWT3)  (F2TY84)  Whole cilia  4.66 (n=3)  6 (n=2)  2 (n=2)  Axoneme  6 (n=3)  ‐  ‐  Membrane + matrix  3.66 (n=3)  ‐  ‐  Cell bodies   ‐  ‐  0 (n=1)  microvilli  ‐  ‐  0 (n=1)    Protein import from the cytosol into the cilium is a tightly controlled process (Kee et al., 2012).  It is orchestrated by the ciliary pore complexes (CPC), which carry a high resemblance to the  nuclear pore complexes (NPC) that regulate protein traffic into the nucleus. Both structures  allow for passive diffusion of smaller proteins; however, large proteins (>~70 kDa) require a  signalling peptide  for active  transport  (Kee & Verhey, 2013). Since  the deciliation protocol  used in the proteomic study by Sigg et al. (2017) excluded the basal body region (Stephens,  147      1995), a colocalisation to ciliary structures from the cytosolic side is unlikely for Adgb. Due to  its large size, the full length Adgb protein likely needs to be transported actively into the cilium.   Upon first discovery, a nuclear localisation signal (NLS) has been predicted bioinformatically  at the C‐terminus of Adgb (Hoogewijs et al., 2012). However, we could find no evidence of  Adgb in the nucleus in a variety of immunofluorescent stainings. In support of this, stainings  of  B.  taurus  testis  sections  remained  devoid  of  nuclear  Adgb  signal  (Chapter  2.1).  The  asynchronous nature of  spermatogenesis provides all meiotic  stages  in one  tissue  section,  which shows that  in B. taurus there  is also no cell cycle‐dependent nuclear  localisation. So,  either the NLS is inactive, or it has a different function. Studies performed on the NLS of the  interflagellar transport component KIF17 have indeed shown that it can serve both as nuclear  and ciliary localisation signal (CLS). The full‐length protein localizes to the cilium, whereas a  fragmented version carrying the C‐terminus and the signalling peptide only is transported into  the nucleus (Dishinger et al., 2010). The NPC and the CPC share common constituents. It  is  thus tempting to speculate that the NLS of Adgb could in fact be a CLS. Preliminary results of  transfection  assays with NLS  reporter  constructs  in  starved  A549  cells  could  not  verify  a  translocation  to  the  cilium. However,  A549  cells  carry  only  non‐motile  cilia, whereas  the  proteome studies were conducted in organisms with motile cilia. In addition, we analysed the  NLS in isolation of the rest of the protein. The studies performed on KIF17 have already shown  that the protein context can be relevant for the mode of transportation (as discussed in: Kee  & Verhey,  2013).  Studies  on  the  ciliary  protein  PC1  have  for  example  demonstrated  that  disruption of  its coiled coil motif  leads to decreased ciliary  localisation efficiency (Su et al.,  2015). Adgb does also carry a coiled‐coil motif upstream of the putative NLS/CLS (Hoogewijs  et al., 2012), which could be relevant for it passing through the CPC.   In  future  studies,  localisation  experiments  could  be  performed  in  a  different model,  for  example in primary human tracheal epithelial cells differentiated under air liquid interphase  conditions.  These  cells  develop  functioning,  beating  cilia  (e.g.:  Coles  et  al.,  2020).  First,  immunofluorescent stainings should be performed to confirm Adgb’s  localisation to motile  cilia  in this system,  if possible even by electron microscopy. Via subsequent transfection of  different reporter constructs, which carry the NLS/CLS, the C‐terminal region upstream and  signalling peptide, or the full‐length version of Adgb, it can be determined which protein parts  are relevant for the correct localisation of Adgb in ciliated cells.  148      3.2 Loss of Androglobin is an indicator of ciliary or flagellar specialization in  metazoans  The  Androglobin  gene  is  highly  conserved  across  the  metazoan  tree.  Our  work  on  the  expression patterns in basal multicellular animals (Chapter 2.4) shows that not only the gene  itself,  but  also  the  mode  of  expression  associated  with  flagellar  or  ciliary  structures  is  conserved.  Since  the  proper  assembly  of  these  structures  is  pivotal  for  embryonic  development  as  well  as  sexual  reproduction,  and  Adgb  has  already  been  shown  to  be  important in murine male gamete generation, it is reasonable to assume that Adgb also serves  a notable function in these organisms.   Nonetheless, the Adgb gene was lost (or heavily mutated) independently several times during  evolution, for example in the model species Caenorhabditis elegans (Hoogewijs et al., 2012).  In C. elegans, loss of Adgb coincides with the absence of motile cilia. Only a subset of sensory  neurons  is ciliated  in  this organism and several  important substructures, such as  the basal  body, show specific abnormalities (Nechipurenko & Sengupta, 2017). Sperm in C. elegans are  motile, but they do not propel themselves via flagellar beating. The cells form pseudopods  and migrate in an amoeboid‐like fashion. A similar situation is described for Daphnia pulex, a  crustacean which carries no Adgb ortholog as well. Sperm of Daphnia are either tailless or  form short pseudopodia (Ebert, 2005). A study on the evolution of the dynein motor protein  gene family shows that C. elegans lost all but 2 copies of the axonemal dyneins, in line with its  lack of motile ciliary structures. A reduction in number of axonemal dynein genes can also be  found in Branchiopoda such as Daphnia: the two‐headed inner‐arm dyneins DHC5 and DHC6  are  absent here, which  suggests  structural  changes or even  impairment of  ciliary motility  (Kollmar,  2016).  The  same  study  identifies  additional  organisms whose  axonemal  dynein  content  is  greatly  reduced  and  which  could  therefore  be  ideal  candidates  to  test  the  hypothesis of ciliary reduction with concomitant loss of Adgb: the cnidarian Kodoa iwatai, the  nematode  Strongyloides  ratti,  and  the  arachnid  Tetranychus  urticae  all  show  the  same  reduction  in axonemal gene diversity as C. elegans. These genomes should be searched for  Adgb orthologs in future studies.   In D. melanogaster, another taxon without an Adgb gene, ciliation is again different than in  mammals: Drosophila is a model system for basal body and primary cilia research, but all cilia  are  immotile.  Only  sperm  cells  carry  flagella, which  confer motility  during  insemination.  149      Intriguingly,  Drosophila  harbours  two  species‐specific  globin  genes  that  are  exclusively  expressed  in  the  testis:  glob2  and  glob3  (Gleixner  et  al.,  2012;  Prothmann  et  al.,  2020).  Preliminary  analysis  conducted  on  a  RNAi  mediated  glob3‐knockdown  strain  suggest  a  function  of  glob3  in  spermatogenesis.  Male  glob3  knockdown  flies  show  structural  irregularities at the post‐meiotic stage and display reduced fertility (Prothmann, 2019). The  ablation of Adgb in mice leads also to defects in spermatogenesis. As described in Chapter 2.2  (Keppner et al., 2022), spermatids from Adgb KO‐mice develop structural deficiencies from  the post‐meiotic stage onwards. Maturation  is  impaired and sperm cells exhibit acrosomal  defects  and  malformed  flagella.  The  effect  is  more  pronounced  than  the  phenotypical  manifestation of a Drosophila glob3, however, residual expression levels and compensation  through  the paralogous  glob2  gene  could  attenuate  severe defects  in  spermatogenesis.  If  Adgb is substituted for by glob2 and 3 in Drosophila, functional analysis of these two genes  could generate additional hypotheses of Adgb’s exact mode of action.   Curiously, Adgb is not the only globin whose expression could be linked to ciliated cells: GbY,  a globin confined  to  fish, amphibians,  reptiles excluding birds, and platypus, was  found  in  multi‐ciliated cells of the epidermis in adult Xenopus laevis. RNA‐Seq data suggests that Adgb  is not co‐expressed here. In contrast to this X. laevis data, single cell RNA sequencing analysis  shows Adgb expression in ciliated cells from a closely related sister species, Xenopus tropicalis  – while GbY is expressed to a much lesser extent in this species (Michel Seiwert, unpublished  data). This might be a second instant were another globin potentially substitutes for Adgb and  thus another model system which could be considered for further functional studies.       150      3.3 Adgb beyond metazoans  As described in Chapter 2.4, Adgb is expressed during the motile blastula or gastrula stages of  a variety of marine invertebrate embryos as well as in a diverse set of flagellated or ciliated  cell  types  in  adult  organisms.  In  addition,  Adgb  has  been  described  in  Choanoflagellates  (Hoogewijs et al., 2012) as well as in the alveolate Tetrahymena thermophila – a rather distant  relative of the metazoan clade (Burki et al., 2020).  In Tetrahymena, Adgb ablation  leads to  mild perturbation of ciliary motility, indicating also some functional conservation (Joachimiak  et  al., 2021). Despite  the  large  topological distance  in  the metazoan  tree Adgb’s  function  seems  to be conserved, which  indicates  that Adgb could also be present  in other protists.  Indeed,  a  search  of  the  NCBI  protein  database  for  “Androglobin”  in  taxa  beyond  the  metazoans  resulted  in  several  hits:  in  the  ciliate  Stylonychia  lemnae,  the  dinoflagellate  Symbiodinium  microadriaticum  (both  Alveolata),  and  the  haptophyte  phytoplankton  Chrysochromulina tobinii. All these automatically annotated sequences are sufficiently  long  (~1000‐1500 AA) and carry a putative 5’ protease domain, as  indicated by NCBIs conserved  domains feature. A globin domain however was not predicted. Preliminary BLASTp analysis  suggests that additional sequences exist in taxa closely related to these examples. In the ciliate  clade,  putative  Adgb  sequences  were  identified  in  Halteria  grandinella,  Tetrahymena  thermophilia, Blepharisma stoltei, Ichtyophthitirus spec and several taxa from the Paramecium  clade.  The  sequence  from  Symbiodinium  microadriaticum  resulted  in  hits  in  additional  Symbodinium  taxa,  as  well  as  in  2  other  dinoflagellates  and  the  Chromideran  Vitrella  brassicaformis.  In  the haptophyte  lineage, 1 additional  sequence was  identified  (results  in  Appendix).   For orientation, figure 3‐1 shows one of the most recent eucaryotic phylogenetic trees (Burki  et  al.,  2020).  Putative  Adgb  sequences  could  be  identified  in  taxa  from  at  least  three  “supergroups”:  the TSAR, Haptista, and Amorphea  clades. Under  the assumption  that  the  topology is correct, this distribution implies that the Adgb gene could additionally be present  in other phyla,  since  these  three  groups do not  share  a  common  ancestor.  In  analogy  to  metazoans,  taxa with  the Adgb gene  could be  ciliated or  flagellated and  indeed, with  the  ciliates and the dinoflagellates, two taxa famously known for their appendages were already  identified as Adgb‐positive above (e.g. Kahl, 1930; Sarjeant, 1974). But flagellar structures are  far more widespread and diverse in protists. They occur on nearly every branch, can vary in  151      size and number and may have associated structures such as loricas or thecas  (as reviewed  in: Jeuck & Arndt, 2013).   Initially, Adgb was thought to be absent  in fungi (Hoogewijs et al., 2012), since no putative  Adgb  sequences could be  identified  in  the genomes of Saccharomyces cerevisiae, Candida  albicans and Aspergillus niger. However, all these taxa belong to the phylum of ascomycetes,  which  propagate  themselves  through  non‐motile  ascospores  propelled  via  the wind  (e.g.  Neiman, 2005). It is possible that the absence of Adgb is again due to a secondary loss here.  There are fungal taxa which reproduce via motile zoospores, e.g. Chytridiomycota, Oomycota  or Myxomycetes (Webster & Weber, 2007). These groups should be considered first  in the  search for fungal Adgb sequences in future studies.    In  contrast  to  this, Adgb will probably not be  identified  in  the Amoebozoa  sister  clade  to  Ophistokonta from the Amorphea group (Figure 3‐1). As the name indicates, Amoebozoa are  not flagellated and crawl via protrusion of pseudopodia (amoeboid movement). They share  this  characteristic with a variety of other  taxa. For example,  the  former  clade of Heliozoa  consist of ameboid protists with axopodia, stiff pseudopodia stabilized by microtubules. The  clade was dissolved after molecular data revealed that the different subgroups spread over  the phylogenetic tree (Nikolaev et al., 2004), however, the morphological similarities hint at  organisms  that  could  have  all  lost  the  Adgb  gene. Many  amoeboid  taxa  belong  to  the  superphylum of Rhizaria  (Nikolaev et al., 2004), a  sister clade  to  the  flagellated alveolates  (Burki et al., 2020). This position renders both branches compelling for comparative analyses.   Despite  the  shared  name,  prokaryotic  and  eukaryotic  flagellar  have  nothing  in  common.  Eukaryotic  flagellar are  intracellular protrusions  that  carry  the  canonical microtubular 9+2  conformation and are composed of a variety of proteins (e.g. Smith & Yang, 2004). Motility is  conferred though sliding of microtubular filaments via dynein motor proteins, which results in  bending of the flagellum (Satir, 1968). Also bacterial flagella have been studied extensively –  and have  even  served  as  the  key witness  in  a  court  case on  “evolution  versus  intelligent  design”  (Pallen & Matzke, 2006) – but  they are structurally more simple, extracellular and  consist only of a single group of proteins, the flagellins (Namba & Vonderviszt, 1997). Their  motion comes from a special motor protein complex at the basal membrane of the flagellum,  which  translates  an  ion  flux  into  rotation  (as  reviewed  in: Nakamura & Minamino, 2019).  Archaeal  flagellar  are  structurally  similar  to  those  of  bacteria  (Metlina,  2004),  but  are  152      assembled from archaea specific genes and thus a product of convergent evolution (Thomas  et al., 2001). It has been suggested that these kingdoms developed different forms of flagellar  due  to  energy  constraints:  the  energetically  expensive  eukaryotic  structures  can  only  be  sustained  by  sufficiently  big  cells, whereas  archaea  and  bacteria  had  to  develop  cheaper  systems  for motility  (Schavemaker & Lynch, 2022). Ciliates,  for example, could only evolve  their high number of cellular appendages due to their  larger size  in comparison with other  unicellular organisms (Schavemaker & Lynch, 2022). Because of these substantial differences  in  composition,  structure,  and  energy  expenditure, we hypothesize  that Adgb will not be  found in genomes from either Bacteria or Archaea.   It will be very interesting to elucidate whether Adgb indeed occupies a far more basal position  in the phylogenetic tree than previously anticipated and to test the hypothesis that it could  have already been present, together with the flagellum,  in the  last common ancestor of all  contemporary eucaryotes (Mitchell, 2007).       Figure 3‐1: Occurrence of putative Adgb sequences across the eucaryotic tree. Adgb was identified in  Choanoflagellates  and  Metazoans  (Ophistokonta)  and  Ciliates  (Alveolata).  Database  search  and  subsequent  BLAST  analysis  identified  putative  sequences  in  Dinoflagellates  (Alveolata)  and  the  Haptophyta clade.        153      3.4 Androglobin: the maverick of the protein family?  In many regards, Adgb can be considered as a distant cousin to the globin protein family. Its  multi‐domain  structure  is unique  among metazoan  globins,  and  its  globin domain  is both  rearranged and interrupted by a calmodulin binding IQ motif. In addition, the globin domain  lacks one of the two crucial histidines needed for proper binding of the prosthetic heme group  (although the substituting GlnE7 has been described in other proper globins before). It also  diverged  very  early  from  the  other  globin  family members, which  arose  from  each  other  through duplication events (Storz 2018, see also Chapter 1.2). It is therefore unlikely that the  full length protein exerts solely a classical globin function, such as NO homeostasis or oxygen  storage – even if its absence sometimes coincides with the atypical occurrence of other globin  family members.   Androglobin’s putative role  in the maintenance of motile cilia could hypothetically emerge  mainly  from  its  calpain  domain. Calpains  are  calcium‐dependent  cysteine  endopeptidases  from the C2 family and activated by autodigestion (Hosfield et al., 1999; Barrett & Rawlings,  2001). Adgb shows highest sequence similarity with human calpain 7, but lacks the cysteine  residues which are important for the proteolytic function (Hoogewijs et al., 2012; Rawlings,  2015). The same lack of critical residues has been described for human calpain 6: Intriguingly,  calpain 6 is also known to influence ciliogenesis. It is involved in regulation and stabilization of  microtubule  dynamics  in  embryonic  tissue  (Tonami  et  al.,  2007)  and  calpain  6  deficiency  reduces the proportion of ciliated cells in vitro (Kim et al., 2019). Overexpression of Adgb can  increase the proportion of ciliated cells in a cell culture model (Chapter 2.1), so it would be  possible that this effect stems from its calpain protease domain. There is also evidence that  Adgb is proteolytically cleaved: in several culture system and with different antibody systems,  Western Blot analysis revealed additional bands upon overexpression of Adgb (Chapter 2.2,  Bracke et al., 2018).   As  also  described  in  Chapter  2.2,  a  mechanism  has  been  proposed  in  which  Adgb  proteolytically cleaves (directly or  indirectly) Sept10, and that proper colocalization of both  proteins  is  critical  for  the  formation of  flagellar  structures.  This  cleavage  is depending on  calmodulin  (CaM)  binding  to  Adgb’s  IQ  domain,  but  not  influenced  by  the  O2  level.  Overexpression of mutated globin constructs in combination with heme‐depleting conditions  showed  that  the  interaction of Adgb with CaM  is  independent of  the  incorporation of  the  154      heme  (Keppner  et  al.,  2022)  –  adding  further  evidence  that  Adgb  should  not  only  be  considered as a member of the globin protein family.   As a cautionary remark, all findings except for the direct interaction of Adgb and Sept10 were  analysed in vitro through overexpression analysis in a cell culture model without endogenous  expression of Adgb. As described in Chapter 2.3, these systems can produce artificial results.  Expression  data  from  the  human  protein  atlas  (Uhlen  et  al.,  2015;  Karlsson  et  al.,  2021)  suggests  that  expression  of Adgb  and  Sept10  is  not  strictly  correlated  in  human  samples  (Figure 3‐2). Sept10 is expressed ubiquitously in tissues (Figure 3‐2 A), but there is no apparent  co‐expression with ADGB at the cell type level (Figure 3‐2 B). However, it would be possible  that the mode of normalization used by the HPA could exaggerate cell type specificity and  cover low expression levels. This would be in line with the expression data reported in Chapter  2.2 (reanalysed from M. Wang et al., 2018), which shows ubiquitous, but  low and constant  expression levels for Sept10 during human spermatogenesis. Interestingly, Sept10 also shows  a different expression pattern in murine single cell data compared to human samples, where  there is a surge in Sept10 expression in round and elongating spermatids, in line with Adgb’s  increased expression  (Chapter 2.2,  reanalysed  from: Green et al., 2018;  Jung et al., 2019;  Lukassen et al., 2018). Therefore,  it would be  interesting to  further analyse the expression  pattern of Sept10 in murine tissues other than the testis and, if feasible, also in human tissues  and sperm cells to see whether this interaction can be translated to the human system.   Finally, the binding of the heme group by the globin domain may not be necessary for the  direct interaction of Adgb and CaM, however, it could still be possible that it could act as a  signalling/activating component. The domain‐specific perturbations of Adgb were introduced  into these systems via overexpression, and not in the context of the whole Adgb protein. These  experiments could be repeated  in a more native setting, for example by creating a domain  specific knockouts  in a system with endogenous Adgb expression. Such a system could be  either a more  sophisticated mammalian  cell model,  such as  the air  liquid  interface model  introduced  above,  or  a  completely  different,  taxonomically  basal  organism which would  ideally facilitate genomic manipulation.   155        156      Figure 3‐2: Expression data on Sept10 and Adgb from the Human protein atlas. (A) expression values  derived from bulk sequencing data and antibody stainings. Adgb is ascribed to the “ciliated cells/cilium  organization” mRNA  expression  cluster;  antibody  stainings  show  protein  expression  in  the  upper  respiratory tract and male and female reproductive organs (in line with our own results, Chapter 2.1).  Sept10  is expressed broadly on protein  level;  it  is assigned to the “squamous epithelial cell cluster”  deducted from bulk mRNA sequencing data. (B) Expression data on Adgb and Sept10 based on single  cell mRNA sequencing data. Adgb expression was found in ciliated cells and during spermatogenesis,  whereas Sept10 is expressed in trophoblasts, muscle cells, adipocytes, fibroblasts, and immune cells.                                                     157            158      4 Summary  Androglobin  is  a  phylogenetically  ancient member  of  the  globin  protein  family.  Its multi‐ domain  structure with  rearranged  globin  domain  is  unique  among metazoan  globins.  As  already indicated by its name, Adgb is mainly expressed in testis tissue and, to a lower extent,  in the lung and the brain. Here, we describe the female reproductive tract as a new expression  site for Adgb. Adgb is expressed in the fallopian tubes and endometria of different mammals.  We  analysed  public  single‐cell  mRNA‐Seq  data  and  found  that  Adgb  expression  always  originates from cells with motile cilia. We therefore hypothesize that Adgb exerts the same  cilia‐associated function in all of these organs. In addition, we found that Adgb is strongly co‐ expressed  with  FoxJ1,  the master  transcription  factor  of  ciliogenesis,    which  is  thus  an  intriguing candidate for the regulation of Adgb. In testis tissue, Adgb is detected in later stages  of  spermatogenesis  and  thus  closely  associated  with  the  production  of  the  flagellum,  a  structure very similar to motile cilia. The knockout of Adgb in mice leads to male sterility. In  this context, we analysed RNA‐Seq data from Adgb KO testis compared to wildtype tissue and  found a small, but distinct set of deregulated genes. GO enrichment analysis suggested defects  in late spermatogenesis but unfortunately did not identify specific deregulated pathways. In  the initial publication, the Adgb gene was found on nearly every branch of the metazoan tree.  Given its strong cilia association in mammals, we investigated whether its expression pattern  is conserved beyond vertebrates. We therefore analysed RNA‐Seq data from a variety of basal  metazoan taxa and found that Adgb is indeed expressed in many cell types carrying flagella or  motile cilia, such as sperm cells from sponges or epidermal ciliated cells in the sea walnut and  embryos from sea urchins. These results indicate a strong conservation of Adgb’s expression  pattern, and we hypothesize that also its cilia‐associated function is conserved in metazoans.  Many other members of the globin family have been studied in the context of human cancer.  Depending on the globin type analysed and also the cancer tissue type, globins can have either  oncogenic or suppressive properties  in malignancies. We therefore analysed Adgb’s role  in  cancer. Transcriptomic analysis of an Adgb overexpressing A549  lung cancer model  initially  suggested  a  possible  oncogenic  effect of Adgb,  in  line with  previous  studies  in  brain  and  prostate cancer models. However, a thorough analysis of transcriptomes from cell lines and  in vivo cancer entities showed that Adgb is simply not expressed in those tumours, which hints  at  sophisticated  regulatory  mechanisms  that  silence  the  Adgb  gene  in  cancer.  Adgb  is  therefore not suitable as a prognostic marker in malignancies.  159          160      5 Zusammenfassung  Androglobin  ist  ein  phylogenetisch  altes  Mitglied  der  Globinproteinfamilie.  Seine  Multi‐ Domänen‐Struktur mit einer rearrangierten Globin‐Domäne ist einzigartig unter den Globinen  der Metazoen. Wie sein Name bereits andeutet, wird Adgb vorwiegend im Hodengewebe und  in  geringerem Maße  in  der  Lunge  und  im Gehirn  exprimiert.  In  dieser  Arbeit wurde  der  weibliche Fortpflanzungstrakt als ein neuer Expressionsort für Adgb identifiziert. Adgb wird in  den  Eileitern  und  Endometrien  verschiedener  Säugetiere  exprimiert.  Durch  die  Analyse  öffentlich zugänglicher mRNA‐Seq‐Daten und single‐cell RNA‐Seq wurde festgestellt, dass die  Adgb‐Expression immer in differenzierten Zellen mit motilen Zilien erfolgt. Adgb könnte daher  in  all  diesen  Organen  die  gleiche,  zilienassoziierte  Funktion  ausüben.  Zudem  wurde  beobachtet, dass Adgb stark mit FoxJ1, dem Haupttranskriptionsfaktor der Zilienbildung, co‐ exprimiert,  der  somit  einen  interessanten  Kandidaten  für  die  Untersuchung  der  Regulationsmechanismen von Adgb darstellt.  Im Hodengewebe wird Adgb vor allem  in den  späteren Stadien der Spermatogenese nachgewiesen und steht damit  in engem zeitlichem  Zusammenhang mit der Bildung des Flagellums, einer Struktur, die den beweglichen Zilien  sehr  ähnlich  ist.  Der  Knockout  von  Adgb  in Mäusen  führt  zu männlicher  Sterilität.  Eine  differentielle  Expressionsanalyse  von RNA‐Seq‐Daten  von Adgb‐KO‐Hoden  im Vergleich  zu  Wildtypen ergab nur eine kleine, aber sehr distinkte Gruppe von deregulierten Genen. Die GO‐ Anreicherungsanalyse  dieser  Liste  deutete  auf  Defekte  in  den  späten  Stadien  der  Spermatogenese  hin,  aber  leider wurde  kein  spezieller  differenziell  regulierter  Signalweg  identifiziert.  In  der  Erstveröffentlichung  wurde  das  Adgb‐Gen  auf  fast  jedem  Ast  des  Stammbaums der Metazoen  gefunden. Angesichts  seiner  starken Assoziation mit  Zilien  in  Säugetieren wollten wir untersuchen, ob auch sein Expressionsmuster über die Wirbeltiere  hinaus  konserviert  ist.  Bulk‐  und  single‐cell  RNA‐Seq‐Daten  einer  Auswahl  basaler  Metazoenarten wurden analysiert und zeigten, dass Adgb tatsächlich in vielen verschiedenen  Zelltypen exprimiert wird, die Geißeln oder bewegliche Zilien tragen (z. B. in Spermazellen von  Schwämmen  oder  epidermalen  Zilienzellen  in  der  Meeresnuss  und  in  Embryonen  von  Seeigeln). Diese Ergebnisse zeigen, dass das Expressionsmuster von Adgb stark konserviert ist.  Dies kann daher auch für seine möglicherweise mit Zilien assoziierte Funktion  in Metazoen  gelten.  Viele  andere  Mitglieder  der  Globinfamilie  wurden  im  Zusammenhang  mit  menschlichem  Krebs  untersucht.  Je  nach  untersuchtem Globin  und  auch  je  nach  Art  des  Krebsgewebes  können  Globine  sowohl  onkogene  als  auch  schützende  Eigenschaften  bei  161      bösartigen  Erkrankungen  haben.  Daher  wurde  die  mögliche  Rolle  von  Adgb  in  Krebs  untersucht.  Die  Transkriptomanalyse  eines  A549‐Lungenkrebsmodells,  welches  Adgb  überexprimiert,  deutete  auf  eine  onkogene Wirkung  von  Adgb  hin, was  im  Einklang mit  früheren Studien an Gehirn‐ und Prostatakrebsmodellen steht. Eine umfassende Analyse der  Transkriptome von Zelllinien und Krebsentitäten zeigte jedoch, dass Adgb in diesen Entitäten  einfach nicht exprimiert wird, was auf komplexe Mechanismen hinweist, die das Adgb‐Gen in  vivo    in  Tumoren  abschalten.  Adgb  ist  daher  nicht  als  prognostischer Marker  für  Krebs  geeignet.                                         162      6 References  Ascenzi, P., di Masi, A., Leboffe, L., Fiocchetti, M., Nuzzo, M. T., Brunori, M., & Marino, M.  (2016). Neuroglobin:  From  structure  to  function  in  health  and  disease.  In Molecular  Aspects of Medicine (Vol. 52, pp. 1–48). Elsevier Ltd.  Barrett, A. J., & Rawlings, N. D. (2001). Evolutionary  lines of cysteine peptidases. Biological  Chemistry, 382(5), 727–733.  Blank, M., & Burmester, T. (2012). Widespread Occurrence of N‐Terminal Acylation in Animal  Globins and Possible Origin of Respiratory Globins from a Membrane‐Bound Ancestor.  Molecular Biology and Evolution, 29(11), 3553–3561.  Blank, M., Kiger, L., Thielebein, A., Gerlach, F., Hankeln, T., Marden, M. C., & Burmester, T.  (2011). Oxygen supply from the bird’s eye perspective: globin E is a respiratory protein in  the chicken retina. The Journal of Biological Chemistry, 286(30), 26507–26515.  Blank, M., Wollberg, J., Gerlach, F., Reimann, K., Roesner, A., Hankeln, T., Fago, A., Weber, R.  E., & Burmester, T. (2011). A membrane‐bound vertebrate globin. PLoS ONE, 6(9).  Blum, A., Wang, P., & Zenklusen, J. C. (2018). SnapShot: TCGA‐Analyzed Tumors. Cell, 173(2),  530.  Bracke, A., Hoogewijs, D., & Dewilde, S. (2018). Exploring three different expression systems  for recombinant expression of globins: Escherichia coli, Pichia pastoris and Spodoptera  frugiperda. Analytical Biochemistry, 543(September 2017), 62–70.  Brunori, M., Giuffrè, A., Nienhaus, K., Nienhaus, G. U., Scandurra, F. M., & Vallone, B. (2005).  Neuroglobin, nitric oxide, and oxygen: functional pathways and conformational changes.  Proceedings  of  the  National  Academy  of  Sciences  of  the  United  States  of  America,  102(24), 8483–8488.  Burki, F., Roger, A. J., Brown, M. W., & Simpson, A. G. B. (2020). The New Tree of Eukaryotes.  Trends in Ecology & Evolution, 35(1), 43–55.  Burmester, T., Ebner, B., Weich, B., & Hankeln, T.  (2002). Cytoglobin: A Novel Globin Type  Ubiquitously  Expressed  inVertebrate  Tissues. Molecular  Biology  and  Evolution,  19(4),  416–421.  163      Burmester,  T., & Hankeln,  T.  (2009). What  is  the  function of neuroglobin?  The  Journal  of  Experimental Biology, 212(Pt 10), 1423–1428.  Burmester,  T.,  &  Hankeln,  T.  (2014).  Function  and  evolution  of  vertebrate  globins.  Acta  Physiologica, 211(3), 501–514.  Burmester, T., Welch, B., Reinhardt, S., & Hankeln, T. (2000). A vertebrate globin expressed in  the brain. Nature, 407(6803), 520–523.  Cen, D., Huang, H., Yang, L., Guo, K., & Zhang,  J.  (2019). Long noncoding RNA STXBP5‐AS1  inhibits  cell  proliferation,  migration,  and  invasion  through  inhibiting  the  PI3K/AKT  signaling pathway in gastric cancer cells. OncoTargets and Therapy, 12, 1929–1936.  Chen, S., Huang, L., Li, G., Qiu, F., Wang, Y., Yang, C., Pan, J., Wu, Z., Chen, J., & Tian, Y. (2020).  LncRNA  STXBP5‐AS1  suppresses  stem  cell‐like  properties  of  pancreatic  cancer  by  epigenetically  inhibiting neighboring androglobin gene expression. Clinical Epigenetics,  12(1), 1–12.  Coles,  J. L., Thompson,  J., Horton, K. L., Hirst, R. A., Griffin, P., Williams, G. M., Goggin, P.,  Doherty, R., Lackie, P. M., Harris, A., Walker, W. T., O’callaghan, C., Hogg, C., Lucas, J. S.,  Blume, C., & Jackson, C. L. (2020). A revised protocol for culture of airway epithelial cells  as a diagnostic tool for primary ciliary dyskinesia. Journal of Clinical Medicine, 9(11), 1– 16.  Corti, P., Xue, J., Tejero, J., Wajih, N., Sun, M., Stolz, D. B., Tsang, M., Kim‐Shapiro, D. B., &  Gladwin, M. T.  (2016). Globin X  is a six‐coordinate globin that reduces nitrite to nitric  oxide  in  fish  red blood  cells. Proceedings of  the National Academy of Sciences of  the  United States of America, 113(30), 8538–8543.  Dishinger, J. F., Kee, H. L., Jenkins, P. M., Fan, S., Hurd, T. W., Hammond, J. W., Truong, Y. N.  T., Margolis, B., Martens, J. R., & Verhey, K. J. (2010). Ciliary entry of the kinesin‐2 motor  KIF17 is regulated by importin‐beta2 and RanGTP. Nature Cell Biology, 12(7), 703–710.  Doğan,  A.,  Demirci,  S.,  Kıratlı,  B.,  &  Şahin,  F.  (2017).  Cytoglobin:  a  potential marker  for  adipogenic differentiation in preadipocytes in vitro. Cytotechnology, 69(1), 157.  Dröge, J., Pande, A., Englander, E. W., & Makałowski, W. (2012). Comparative Genomics of  Neuroglobin Reveals Its Early Origins. PLOS ONE, 7(10), e47972.  164      Ebert, D. (2005).  Introduction to Daphnia Biology ‐ Ecology, Epidemiology, and Evolution of  Parasitism in Daphnia [Internet]. In Ecology, Epidemiology, and Evolution of Parasitism in  Daphnia.  Elkholi,  I. E., Elsherbiny, M. E., & Emara, M. (2022). Myoglobin: From physiological roles to  potential implications in cancer. Biochimica et Biophysica Acta (BBA) ‐ Reviews on Cancer,  1877(3), 188706.  Elsherbiny, M.  E.,  Shaaban, M.,  El‐Tohamy,  R.,  Elkholi,  I.  E., Hammam, O. A., Magdy, M.,  Allalunis‐Turner, J., & Emara, M. (2021). Expression of Myoglobin in Normal and Cancer  Brain Tissues: Correlation With Hypoxia Markers. Frontiers in Oncology, 11, 1561.  Emara, M., Salloum, N., & Allalunis‐Turner, J. (2009). Expression and hypoxic up‐regulation of  neuroglobin in human glioblastoma cells. Molecular Oncology, 3(1), 45–53.  Emara, M., Turner, A. R., & Allalunis‐Turner, J. (2010). Hypoxic regulation of cytoglobin and  neuroglobin expression  in human normal and tumor tissues. Cancer Cell  International,  10(1), 1–16.  Fabrizius, A., Andre, D., Laufs, T., Bicker, A., Reuss, S., Porto, E., Burmester, T., & Hankeln, T.  (2016).  Critical  re‐evaluation  of  neuroglobin  expression  reveals  conserved  patterns  among mammals. Neuroscience, 337, 339–354.  Fago, A., Giangiacomo, L., D’Avino, R., Carratore, V., Romano, M., Boffi, A., & Chiancone, E.  (2001). Hagfish Hemoglobins: Structure, function and oxygen‐linked association. Journal  of Biological Chemistry, 276(29), 27415–27423.  Fago, A., & Weber, R. E.  (1995). The hemoglobin  system of  the hagfish Myxine glutinosa:  aggregation state and functional properties. Biochimica et Biophysica Acta (BBA) ‐ Protein  Structure and Molecular Enzymology, 1249(1), 109–115.  Fiocchetti, M., Cipolletti, M., Ascenzi, P., & Marino, M. (2018). Dissecting the 17β‐estradiol  pathways necessary for neuroglobin anti‐apoptotic activity  in breast cancer. Journal of  Cellular Physiology, 233(7), 5087–5103.  Flonta, S. E., Arena, S., Pisacane, A., Michieli, P., & Bardelli, A. (2009). Expression and functional  regulation of myoglobin in epithelial cancers. The American Journal of Pathology, 175(1),  201–206.  165      Fordel,  E.,  Thijs,  L., Martinet, W.,  Lenjou, M.,  Laufs,  T.,  Van  Bockstaele, D., Moens,  L., &  Dewilde, S. (2006). Neuroglobin and cytoglobin overexpression protects human SH‐SY5Y  neuroblastoma  cells against oxidative  stress‐induced  cell death. Neuroscience  Letters,  410(2), 146–151.  Fuchs,  C.,  Burmester,  T., &  Hankeln,  T.  (2006).  The  amphibian  globin  gene  repertoire  as  revealed by  the Xenopus genome. Cytogenetic and Genome Research, 112(3–4), 296– 306.  Fujita, Y., Koinuma, S., De Velasco, M. A., Bolz,  J., Togashi, Y., Terashima, M., Hayashi, H.,  Matsuo, T., & Nishio, K. (2014). Melanoma transition is frequently accompanied by a loss  of cytoglobin expression in melanocytes: A novel expression site of cytoglobin. PLoS ONE,  9(4).  Gallagher, M. D., & Macqueen, D. J. (2017). Evolution and Expression of Tissue Globins in Ray‐ Finned Fishes. Genome Biology and Evolution, 9(1), 32–47.  Gleixner, E., Herlyn, H., Zimmerling, S., Burmester, T., & Hankeln, T.  (2012). Testes‐specific  hemoglobins  in Drosophila evolved by a combination of sub‐ and neofunctionalization  after gene duplication. BMC Evolutionary Biology, 12(1), 34.  Gorr,  T.  A., Wichmann, D.,  Pilarsky,  C.,  Theurillat,  J.  P.,  Fabrizius,  A.,  Laufs,  T.,  Bauer,  T.,  Koslowski, M., Horn, S., Burmester, T., Hankeln, T., & Kristiansen, G. (2011). Old proteins  – new locations: myoglobin, haemoglobin, neuroglobin and cytoglobin in solid tumours  and cancer cells. Acta Physiologica, 202(3), 563–581.  Green, C. D., Ma, Q., Manske, G. L., Shami, A. N., Zheng, X., Marini, S., Moritz, L., Sultan, C.,  Gurczynski,  S.  J., Moore, B. B., Tallquist, M. D.,  Li,  J.  Z., & Hammoud,  S.  S.  (2018). A  Comprehensive Roadmap of Murine Spermatogenesis Defined by Single‐Cell RNA‐Seq.  Developmental Cell, 46(5), 651‐667.e10.  Hardison, R. C.  (1996). A brief history of hemoglobins: plant, animal, protist, and bacteria.  Proceedings of the National Academy of Sciences, 93(12), 5675–5679.  Hardison, R. C. (1998). Hemoglobins from bacteria to man: evolution of different patterns of  gene expression. The Journal of Experimental Biology, 201(Pt 8), 1099–1117.  Helbo, S., Dewilde, S., Williams, D. R., Berghmans, H., Berenbrink, M., Cossins, A. R., & Fago,  166      A.  (2012).  Functional  differentiation  of myoglobin  isoforms  in  hypoxia‐tolerant  carp  indicates  tissue‐specific  protective  roles. American  Journal  of  Physiology.  Regulatory,  Integrative and Comparative Physiology, 302(6), 693–701.  Herman, J. L., Challis, C. J., Novák, Á., Hein, J., & Schmidler, S. C. (2014). Simultaneous Bayesian  Estimation of Alignment and Phylogeny under a  Joint Model of Protein Sequence and  Structure. Molecular Biology and Evolution, 31(9), 2251–2266.  Hoffmann, F. G., Opazo, J. C., Hoogewijs, D., Hankeln, T., Ebner, B., Vinogradov, S. N., Bailly,  X., & Storz, J. F. (2012). Evolution of the globin gene family in deuterostomes: Lineage‐ specific patterns of diversification and attrition. Molecular Biology and Evolution, 29(7),  1735–1745.  Hoffmann, F. G., Opazo, J. C., & Storz, J. F. (2010). Gene cooption and convergent evolution of  oxygen  transport  hemoglobins  in  jawed  and  jawless  vertebrates.  Proceedings  of  the  National Academy of Sciences of the United States of America, 107(32), 14274–14279.  Hoffmann, F. G., Storz,  J. F., Kuraku, S., Vandewege, M. W., & Opazo,  J. C.  (2021). Whole‐ Genome  Duplications  and  the  Diversification  of  the  Globin‐X  Genes  of  Vertebrates.  Genome Biology and Evolution, 13(10).  Hoogewijs, D., Ebner, B., Germani, F., Hoffmann, F. G., Fabrizius, A., Moens, L., Burmester, T.,  Dewilde, S., Storz, J. F., Vinogradov, S. N., & Hankeln, T. (2012). Androglobin: A Chimeric  Globin  in Metazoans That  Is Preferentially Expressed  in Mammalian Testes. Molecular  Biology and Evolution, 29(4), 1105–1114.  Hosfield, C. M., Elce, J. S., Davies, P. L., & Jia, Z. (1999). Crystal structure of calpain reveals the  structural basis  for Ca(2+)‐dependent protease  activity  and  a novel mode of enzyme  activation. The EMBO Journal, 18(24), 6880.  Huang, B., Lu, Y., Li, X., Zhu, Z., Li, K., Liu, J., Zheng, J., & Hu, Z. (2014). Androglobin knockdown  inhibits growth of glioma cell lines. Int J Clin Exp Pathol, 7(5), 2179–2184.  Huang, J., Xie, N., Huang, H., Yao, J., & Hu, W. (2018). Long noncoding RNA STXBP5‐AS1 inhibits  cell proliferation, migration, and  invasion via preventing  the PI3K/AKT against STXBP5  expression  in  non‐small‐cell  lung  carcinoma.  Journal  of  Cellular  Biochemistry,  120(5),  7489–7498.  167      Hutter, C., & Zenklusen, J. C. (2018). The Cancer Genome Atlas: Creating Lasting Value beyond  Its Data. Cell, 173(2), 283–285.  Jeuck,  A.,  &  Arndt,  H.  (2013).  A  Short  Guide  to  Common  Heterotrophic  Flagellates  of  Freshwater Habitats Based on the Morphology of Living Organisms. Protist, 164(6), 842– 860.  Joachimiak, E., Osinka, A., Farahat, H., Świderska, B., Sitkiewicz, E., Poprzeczko, M., Fabczak,  H., & Wloga, D.  (2021). Composition and  function of  the C1b/C1f  region  in  the ciliary  central apparatus. Scientific Reports 2021 11:1, 11(1), 1–17.  Jung, M., Wells, D., Rusch,  J., Ahmad, S., Marchini,  J., Myers, S. R., & Conrad, D. F.  (2019).  Unified single‐cell analysis of testis gene regulation and pathology in five mouse strains.  ELife, 8.  Kahl, A. (1930). Allgemeiner Teil und Prostomata. In Urtiere oder Protozoa I, Wimpertiere oder  Ciliata (Infusoria). Fischer.  Karlsson, M.,  Zhang, C., Méar,  L.,  Zhong, W., Digre, A., Katona, B.,  Sjöstedt,  E., Butler,  L.,  Odeberg,  J., Dusart, P., Edfors, F., Oksvold, P., von Feilitzen, K., Zwahlen, M., Arif, M.,  Altay, O.,  Li, X., Ozcan, M., Mardonoglu, A., …  Lindskog, C.  (2021). A  single–cell  type  transcriptomics map of human tissues. Science Advances, 7(31).  Kawada, N., Kristensen, D. B., Asahina, K., Nakatani, K., Minamiyama, Y., Seki, S., & Yoshizato,  K.  (2001). Characterization of a Stellate Cell Activation‐associated Protein  (STAP) with  Peroxidase Activity Found  in Rat Hepatic Stellate Cells. Journal of Biological Chemistry,  276(27), 25318–25323.  Kee, H. L., Dishinger, J. F., Lynne Blasius, T., Liu, C. J., Margolis, B., & Verhey, K. J. (2012). A size‐ exclusion permeability barrier and nucleoporins characterize a ciliary pore complex that  regulates transport into cilia. Nature Cell Biology 2012 14:4, 14(4), 431–437.  Kee, H. L., & Verhey, K. J. (2013). Molecular connections between nuclear and ciliary import  processes. Cilia, 2(1), 1–10.  Keppner, A., Correia, M., Santambrogio, S., Koay, T. W., Maric, D., Osterhof, C., Winter, D. V,  Clerc, A., Stumpe, M., Chalmel, F., Dewilde, S., Odermatt, A., Kressler, D., Hankeln, T.,  Wenger, R. H., & Hoogewijs, D.  (2022). Androglobin, a chimeric mammalian globin,  is  168      required for male fertility. ELife, 11.  Keppner,  A., Maric,  D.,  Correia, M.,  Koay,  T. W., Orlando,  I. M.  C.,  Vinogradov,  S.  N., &  Hoogewijs,  D.  (2020).  Lessons  from  the  post‐genomic  era:  Globin  diversity  beyond  oxygen binding and transport. Redox Biology, 37, 101687.  Kim, B. H., Kim, D. Y., Oh, S., Ko, J. Y., Rah, G., Yoo, K. H., & Park, J. H. (2019). Deficiency of  calpain‐6 inhibits primary ciliogenesis. BMB Reports, 52(10), 619.  Koay, T. W., Osterhof, C., Orlando, I. M. C., Keppner, A., Andre, D., Yousefian, S., Alonso, M. S.,  Correia, M., Markworth, R., Schödel, J., Hankeln, T., & Hoogewijs, D. (2021). Androglobin  gene  expression  patterns  and  FOXJ1‐dependent  regulation  indicate  its  functional  association with ciliogenesis. Journal of Biological Chemistry, 296.  Koch, J., & Burmester, T. (2016). Membrane‐bound globin X protects the cell from reactive  oxygen  species. Biochemical  and  Biophysical  Research  Communications,  469(2),  275– 280.  Kollmar, M.  (2016).  Fine‐Tuning Motile  Cilia  and  Flagella:  Evolution  of  the Dynein Motor  Proteins  from Plants  to Humans at High Resolution. Molecular Biology and Evolution,  33(12), 3249–3267.  Kono, H., Hosomura, N., Amemiya, H., Kawaida, H., Furuya, S., Akaike, H., Kawaguchi, Y., Sudo,  M.,  &  Ichikawa,  D.  (2021).  Cytoglobin  as  a  Prognostic  Factor  for  Pancreatic  Ductal  Adenocarcinoma. Pancreas, 50(7), 994–999.  Kristiansen, G., Hu, J., Wichmann, D., Stiehl, D. P., Rose, M., Gerhardt, J., Bohnert, A., Ten Haaf,  A., Moch, H., Raleigh, J., Varia, M. A., Subarsky, P., Scandurra, F. M., Gnaiger, E., Gleixner,  E., Bicker, A., Gassmann, M., Hankeln,  T., Dahl,  E., & Gorr,  T. A.  (2011).  Endogenous  myoglobin in breast cancer is hypoxia‐inducible by alternative transcription and functions  to  impair mitochondrial activity: a role  in tumor suppression? The Journal of Biological  Chemistry, 286(50), 43417–43428.  Kristiansen, G., Rose, M., Geisler, C., Fritzsche, F. R., Gerhardt,  J., Lüke, C., Ladhoff, A. M.,  Knüchel, R., Dietel, M., Moch, H., Varga, Z., Theurillat, J. P., Gorr, T. A., & Dahl, E. (2010).  Endogenous  myoglobin  in  human  breast  cancer  is  a  hallmark  of  luminal  cancer  phenotype. British Journal of Cancer, 102(12), 1736–1745.  169      Kugelstadt, D., Haberkamp, M., Hankeln, T., & Burmester, T. (2004). Neuroglobin, cytoglobin,  and  a  novel,  eye‐specific  globin  from  chicken.  Biochemical  and  Biophysical  Research  Communications, 325(3), 719–725.  Li, R. C., Morris, M. W., Lee, S. K., Pouranfar, F., Wang, Y., & Gozal, D. (2007). Neuroglobin  protects PC12 cells against oxidative stress. Brain Research, 1190(1), 159–166.  Liu, X., El‐Mahdy, M. A., Boslett, J., Varadharaj, S., Hemann, C., Abdelghany, T. M., Ismail, R.  S., Little, S. C., Zhou, D., Thuy, L. T. T., Kawada, N., & Zweier,  J. L.  (2017). Cytoglobin  regulates  blood  pressure  and  vascular  tone  through  nitric  oxide metabolism  in  the  vascular wall. Nature Communications 2017 8:1, 8(1), 1–14.  Lüdemann,  J., Verissimo, K. M., Dreger, K., Fago, A., Schneider,  I., & Burmester, T.  (2019).  Globin E is a myoglobin‐related, respiratory protein highly expressed in lungfish oocytes.  Scientific Reports 2019 9:1, 9(1), 1–11.  Lukassen, S., Bosch, E., Ekici, A. B., & Winterpacht, A.  (2018). Characterization of germ cell  differentiation in the male mouse through single‐cell RNA sequencing. Scientific Reports,  8(1), 8–14.  Mannino, M. H., Patel, R. S., Eccardt, A. M., Perez Magnelli, R. A., Robinson, C. L. C., Janowiak,  B.  E.,  Warren,  D.  E.,  &  Fisher,  J.  S.  (2019).  Myoglobin  as  a  versatile  peroxidase:  Implications  for  a more  important  role  for  vertebrate  striated muscle  in  antioxidant  defense. Comparative Biochemistry and Physiology Part B: Biochemistry and Molecular  Biology, 234, 9–17.  Mathai, C., Jourd’heuil, F. L., Lopez‐Soler, R. I., & Jourd’heuil, D. (2020). Emerging perspectives  on cytoglobin, beyond NO dioxygenase and peroxidase. Redox Biology, 32, 101468.  Meller, S., van Ellen, A., Gevensleben, H., Bicker, A., Hankeln, T., Gorr, T. A., Sailer, V., Dröge,  F., Schröck, F., Bootz, F., Schröck, A., Perner, S., Dietrich, D., & Kristiansen, G.  (2016).  Ectopic Myoglobin Expression Is Associated with a Favourable Outcome in Head and Neck  Squamous Cell Carcinoma Patients. Anticancer Research, 36(12), 6235–6242.  Metlina,  A.  L.  (2004).  Bacterial  and  archaeal  flagella  as  prokaryotic  motility  organelles.  Biochemistry (Moscow) 2004 69:11, 69(11), 1203–1212.  Mitchell, D. R.  (2007). The evolution of eukaryotic  cilia and  flagella as motile and  sensory  170      organelles. Advances in Experimental Medicine and Biology, 607, 130.  Nakamura, S., & Minamino, T. (2019). Flagella‐Driven Motility of Bacteria. Biomolecules, 9(7).  Nakatani, K., Okuyama, H., Shimahara, Y., Saeki, S., Kim, D. H., Nakajima, Y., Seki, S., Kawada,  N.,  &  Yoshizato,  K.  (2004).  Cytoglobin/STAP,  its  unique  localization  in  splanchnic  fibroblast‐like cells and function in organ fibrogenesis. Laboratory Investigation; a Journal  of Technical Methods and Pathology, 84(1), 91–101.  Namba, K., & Vonderviszt, F. (1997). Molecular architecture of bacterial flagellum. Quarterly  Reviews of Biophysics, 30(1), 1–65.  Nechipurenko, I. V., & Sengupta, P. (2017). The rise and fall of basal bodies in the nematode  Caenorhabditis elegans. Cilia, 6(1), 1–6.  Neiman,  A.  M.  (2005).  Ascospore  Formation  in  the  Yeast  Saccharomyces  cerevisiae.  Microbiology and Molecular Biology Reviews, 69(4), 565.  Nikolaev, S. I., Berney, C., Fahrni, J. F., Bolivar, I., Polet, S., Mylnikov, A. P., Aleshin, V. V., Petrov,  N. B., & Pawlowski, J. (2004). From the Cover: The twilight of Heliozoa and rise of Rhizaria,  an emerging supergroup of amoeboid eukaryotes. Proceedings of the National Academy  of Sciences of the United States of America, 101(21), 8066.  Nishi, H.,  Inagi, R., Kawada, N., Yoshizato, K., Mimura,  I., Fujita, T., & Nangaku, M.  (2011).  Cytoglobin,  a  novel member  of  the  globin  family,  protects  kidney  fibroblasts  against  oxidative stress under  ischemic conditions. The American Journal of Pathology, 178(1),  128–139.  Oleksiewicz, U.,  Liloglou,  T.,  Field,  J.  K., &  Xinarianos, G.  (2011).  Cytoglobin:  biochemical,  functional and clinical perspective of the newest member of the globin family. Cellular  and Molecular Life Sciences 2011 68:23, 68(23), 3869–3883.  Opazo, J. C., Lee, A. P., Hoffmann, F. G., Toloza‐Villalobos, J., Burmester, T., Venkatesh, B., &  Storz,  J. F.  (2015). Ancient duplications and expression divergence  in  the globin gene  superfamily of vertebrates: Insights from the elephant shark genome and transcriptome.  Molecular Biology and Evolution, 32(7), 1684–1694.  Pallen, M.  J., & Matzke, N.  J.  (2006). From The Origin of Species  to  the origin of bacterial  171      flagella. Nature Reviews Microbiology 2006 4:10, 4(10), 784–790.  Petersen,  M.  G.,  Dewilde,  S.,  &  Fago,  A.  (2008).  Reactions  of  ferrous  neuroglobin  and  cytoglobin with nitrite under anaerobic  conditions.  Journal of  Inorganic Biochemistry,  102(9), 1777–1782.  Prothmann, A. (2019). Analyse der Evolution und Funktion von Globinen in Arthropoden [JGU  Mainz].  Prothmann,  A., Hoffmann,  F. G., Opazo,  J.  C., Herbener,  P.,  Storz,  J.  F.,  Burmester,  T., &  Hankeln,  T.  (2020).  The Globin Gene  Family  in Arthropods:  Evolution  and  Functional  Diversity. Frontiers in Genetics, 11, 858.  Qiu, Y., Maillett, D. H., Knapp,  J., Olson,  J.  S., & Riggs, A.  F.  (2000).  Lamprey hemoglobin.  Structural basis of the bohr effect. The Journal of Biological Chemistry, 275(18), 13517– 13528.  Rawlings, N. D.  (2015).  Bacterial  calpains  and  the  evolution  of  the  calpain  (C2)  family  of  peptidases. Biology Direct, 10(1).  Reeder, B. J., Svistunenko, D. A., & Wilson, M. T. (2011). Lipid binding to cytoglobin leads to a  change in haem co‐ordination: a role for cytoglobin in lipid signalling of oxidative stress.  The Biochemical Journal, 434(3), 483–492.  Reuss, S., Saaler‐Reinhardt, S., Weich, B., Wystub, S., Reuss, M. H., Burmester, T., & Hankeln,  T.  (2002).  Expression  analysis  of  neuroglobin mRNA  in  rodent  tissues. Neuroscience,  115(3), 645–656.  Ribeiro, E. A., & Ramos, C. H. I. (2005). Circular permutation and deletion studies of myoglobin  indicate  that  the correct position of  its N‐terminus  is  required  for native stability and  solubility but not for native‐like heme binding and folding. Biochemistry, 44(12), 4699– 4709.  Rösner,  A.,  Fuchs,  C.,  Hankeln,  T.,  &  Burmester,  T.  (2005).  A  Globin  Gene  of  Ancient  Evolutionary Origin  in Lower Vertebrates: Evidence for Two Distinct Globin Families  in  Animals. Molecular Biology and Evolution, 22(1), 12–20.  Sarjeant,  W.  A.  S.  (1974).  Living  Dinoflagellates:  Morphology.  In  Fossil  and  Living  172      Dinoflagellates (pp. 13–32). Academic Press.  Satir,  P.  (1968).  STUDIES  ON  CILIA :  III.  Further  Studies  on  the  Cilium  Tip  and  a  “Sliding  Filament” Model of Ciliary Motility. The Journal of Cell Biology, 39(1), 77.  Schavemaker, P. E., & Lynch, M. (2022). Flagellar energy costs across the tree of life. ELife, 11.  Schwarze,  K., &  Burmester,  T.  (2013).  Conservation  of  globin  genes  in  the  “living  fossil”  Latimeria chalumnae and reconstruction of the evolution of the vertebrate globin family.  Biochimica et Biophysica Acta ‐ Proteins and Proteomics, 1834(9), 1801–1812.  Schwarze, K., Campbell, K. L., Hankeln, T., Storz, J. F., Hoffmann, F. G., & Burmester, T. (2014).  The  globin  gene  repertoire  of  lampreys:  Convergent  evolution  of  hemoglobin  and  myoglobin  in  jawed and  jawless vertebrates. Molecular Biology and Evolution, 31(10),  2708–2721.  Schwarze, K., Singh, A., & Burmester, T. (2015). The Full Globin Repertoire of Turtles Provides  Insights into Vertebrate Globin Evolution and Functions. Genome Biology and Evolution,  7(7), 1896–1913.  Shen, Y. R., Wang, H. Y., Tsai, Y. C., Kuo, Y. C., Wu, S. R., Wang, C. Y., & Kuo, P. L. (2020). The  SEPT12  complex  is  required  for  the  establishment  of  a  functional  sperm  head‐tail  junction. Molecular Human Reproduction, 26(6), 402–412.  Sigg, M. A., Menchen,  T.,  Lee,  C.,  Johnson,  J.,  Jungnickel, M.  K.,  Choksi,  S.  P., Garcia, G.,  Busengdal, H., Dougherty, G. W., Pennekamp, P., Werner, C., Rentzsch, F., Florman, H.  M.,  Krogan,  N.,  Wallingford,  J.  B.,  Omran,  H.,  &  Reiter,  J.  F.  (2017).  Evolutionary  Proteomics  Uncovers  Ancient  Associations  of  Cilia  with  Signaling  Pathways.  Developmental Cell, 43(6), 744‐762.e11.  Smagghe, B. J., Trent, J. T., & Hargrove, M. S. (2008). NO dioxygenase activity in hemoglobins  is ubiquitous in vitro, but limited by reduction in vivo. PloS One, 3(4).  Smith, E. F., & Yang, P. (2004). The Radial Spokes and Central Apparatus: Mechano‐Chemical  Transducers That Regulate Flagellar Motility. Cell Motility and the Cytoskeleton, 57(1), 8.  Sosa, V., Moliné, T., Somoza, R., Paciucci, R., Kondoh, H., & LLeonart, M. E. (2013). Oxidative  stress and cancer: an overview. Ageing Research Reviews, 12(1), 376–390.  173      Stephens, R. E.  (1995). Ciliogenesis  in sea urchin embryos‐‐a subroutine  in  the program of  development. BioEssays : News and Reviews  in Molecular, Cellular and Developmental  Biology, 17(4), 331–340.  Storz, J. F. (2018). Hemoglobin. Oxford University Press.  Storz, J. F., Opazo, J. C., & Hoffmann, F. G. (2013). Gene duplication, genome duplication, and  the  functional  diversification  of  vertebrate  globins.  Molecular  Phylogenetics  and  Evolution, 66(2), 469–478.  Su,  X., Wu, M.,  Yao, G.,  El‐Jouni, W.,  Luo, C.,  Tabari, A., &  Zhou,  J.  (2015). Regulation  of  polycystin‐1  ciliary  trafficking by motifs at  its C‐terminus and polycystin‐2 but not by  cleavage at the GPS site. Journal of Cell Science, 128(22), 4063–4073.  Thi Thanh Hai, N., Thuy, L. T. T., Shiota, A., Kadono, C., Daikoku, A., Hoang, D. V., Dat, N. Q.,  Sato‐Matsubara, M.,  Yoshizato,  K., &  Kawada, N.  (2018).  Selective  overexpression  of  cytoglobin  in  stellate  cells  attenuates  thioacetamide‐induced  liver  fibrosis  in  mice.  Scientific Reports 2018 8:1, 8(1), 1–15.  Thi Thanh Thuy, L., Morita, T., Yoshida, K., Wakasa, K., Iizuka, M., Ogawa, T., Mori, M., Sekiya,  Y., Momen, S., Motoyama, H., Ikeda, K., Yoshizato, K., & Kawada, N. (2011). Promotion of  liver  and  lung  tumorigenesis  in DEN‐treated  cytoglobin‐deficient mice.  The American  Journal of Pathology, 179(2), 1050–1060.  Thomas, N. A., Bardy, S. L., & Jarrell, K. F. (2001). The archaeal flagellum: a different kind of  prokaryotic motility structure. FEMS Microbiology Reviews, 25(2), 147–174.  Thuy, L. T. T., Hai, H., & Kawada, N. (2020). Role of cytoglobin, a novel radical scavenger, in  stellate cell activation and hepatic fibrosis. Clinical and Molecular Hepatology, 26(3), 280.  Thuy, L. T. T., Van Thuy, T. T., Matsumoto, Y., Hai, H., Ikura, Y., Yoshizato, K., & Kawada, N.  (2016).  Absence  of  cytoglobin  promotes multiple  organ  abnormalities  in  aged mice.  Scientific Reports, 6.  Tiso, M., Tejero, J., Basu, S., Azarov, I., Wang, X., Simplaceanu, V., Frizzell, S., Jayaraman, T.,  Geary, L., Shapiro, C., Ho, C., Shiva, S., Kim‐Shapiro, D. B., & Gladwin, M. T. (2011). Human  Neuroglobin Functions as a Redox‐regulated Nitrite Reductase. The Journal of Biological  Chemistry, 286(20), 18277.  174      Tonami, K., Kurihara, Y., Aburatani, H., Uchijima, Y., Asano, T., & Kurihara, H. (2007). Calpain 6  is  involved  in microtubule  stabilization  and  cytoskeletal  organization. Molecular  and  Cellular Biology, 27(7), 2548–2561.  Trachootham, D., Alexandre, J., & Huang, P. (2009). Targeting cancer cells by ROS‐mediated  mechanisms: a radical therapeutic approach? Nature Reviews Drug Discovery 2009 8:7,  8(7), 579–591.  Trent,  J.  T.,  &  Hargrove, M.  S.  (2002).  A  ubiquitously  expressed  human  hexacoordinate  hemoglobin. Journal of Biological Chemistry, 277(22), 19538–19545.  Uhlen, M., Fagerberg, L., Hallstrom, B. M., Lindskog, C., Oksvold, P., Mardinoglu, A., Sivertsson,  A., Kampf, C., Sjostedt, E., Asplund, A., Olsson,  I., Edlund, K., Lundberg, E., Navani, S.,  Szigyarto, C. A.‐K., Odeberg,  J., Djureinovic, D., Takanen,  J. O., Hober, S., … Ponten, F.  (2015).  Tissue‐based  map  of  the  human  proteome.  Science,  347(6220),  1260419– 1260419.  Van Acker, Z. P., Luyckx, E., & Dewilde, S. (2018). Neuroglobin Expression in the Brain: a Story  of Tissue Homeostasis Preservation. Molecular Neurobiology 2018 56:3, 56(3), 2101– 2122.  Vinogradov,  S.  N.,  Hoogewijs,  D.,  Bailly,  X.,  Mizuguchi,  K.,  Dewilde,  S.,  Moens,  L.,  &  Vanfleteren, J. R. (2007). A model of globin evolution. Gene, 398(1–2), 132–142.  Vinogradov,  S. N.,  Tinajero‐Trejo, M.,  Poole,  R.  K., & Hoogewijs, D.  (2013).  Bacterial  and  archaeal globins  ‐ A  revised perspective. Biochimica et Biophysica Acta  ‐ Proteins and  Proteomics, 1834(9), 1789–1800.  Wang, M., Liu, X., Chang, G., Chen, Y., An, G., Yan, L., Gao, S., Xu, Y., Cui, Y., Dong, J., Chen, Y.,  Fan, X., Hu, Y., Song, K., Zhu, X., Gao, Y., Yao, Z., Bian, S., Hou, Y., … Qiao, J. (2018). Single‐ Cell  RNA  Sequencing  Analysis  Reveals  Sequential  Cell  Fate  Transition  during  Human  Spermatogenesis. Cell Stem Cell, 23(4), 599‐614.e4.  Webster, J., & Weber, R. (2007).  Introduction to Fungi.  In  Introduction to Fungi. Cambridge  University Press.  Wright, T. J., & Davis, R. W. (2015). Myoglobin oxygen affinity in aquatic and terrestrial birds  and mammals. Journal of Experimental Biology, 218(14), 2180–2189.  175      Xu, H. W., Huang, Y. J., Xie, Z. Y., Lin, L., Guo, Y. C., Zhuang, Z. R., Lin, X. P., Zhou, W., Li, M.,  Huang, H. H., Wei, X. L., Man, K., & Zhang, G. J. (2013). The expression of cytoglobin as a  prognostic factor in gliomas: a retrospective analysis of 88 patients. BMC Cancer, 13, 247.  Yassin, M., Kissow, H., Vainer, B., Joseph, P. D., Hay‐Schmidt, A., Olsen, J., & Pedersen, A. E.  (2018).  Cytoglobin  affects  tumorigenesis  and  the  expression  of  ulcerative  colitis‐ associated genes under chemically  induced colitis  in mice. Scientific Reports 2018 8:1,  8(1), 1–16.  Zhang, B.,  Liu, Y.,  Li, Y.,  Zhe, X.,  Zhang,  S., &  Zhang,  L.  (2018). Neuroglobin promotes  the  proliferation and  suppresses  the apoptosis of glioma  cells by activating  the PI3K/AKT  pathway. Molecular Medicine Reports, 17(2), 2757–2763.  Zhang, J., Lan, S. J., Liu, Q. R., Liu, J. M., & Chen, X. Q. (2013). Neuroglobin, a Novel Intracellular  Hexa‐Coordinated Globin, Functions as a Tumor Suppressor in Hepatocellular Carcinoma  via Raf/MAPK/Erk. Molecular Pharmacology, 83(5), 1109–1119.        176      7 Appendix    7.1 Supplementary data          177      7.2 Index of electronic supplement      Electronic version of this thesis  Appendix of Chapter 2.1  ‐ Digital version of the publication  ‐ Supplemental figures and tables  ‐ Supplementals files 1‐2  Appendix of Chapter 2.2  ‐ Digital version of the publication  ‐ Digital version of the figures  ‐ Supplementary files 1‐2  Appendix of Chapter 2.3  ‐ Supplementary files 1‐3  Appendix of Chapter 2.4  ‐ Supplementary files 1‐5  Appendix of Discussion  ‐ Blast results for basal metazoan Adgb sequences      178    7.3 Acknowledgements 179   7.4 Eidesstattliche Erklärung    Hiermit  versichere  ich,  dass  ich  diese  Dissertation  selbständig  angefertigt  und  alle  verwendeten Hilfsmittel in der Arbeit angegeben habe. Ich habe die Dissertation, oder Teile  davon, an keiner anderen Fakultät, bzw. einem anderen Fachbereich eingereicht.  Ich habe  keinen anderen Promotionsversuch unternommen.        Mainz, September 2022      ______________________________________                  Carina Osterhof      180      7.5 Curriculum vitae      181          182