cells Article Voluntary Wheel Running Did Not Alter Gene Expression in 5xfad Mice, but in Wild-Type Animals Exclusively after One-Day of Physical Activity Anna Wierczeiko 1,2,† , Lena Gammel 3,†, Konstantin Radyushkin 2, Vu Thu Thuy Nguyen 3 , Hristo Todorov 1, Susanne Gerber 1,* and Kristina Endres 3,* 1 Working Group Computational Systems Genetics (CSG), Institute of Human Genetics, University Medical Center, Johannes Gutenberg University, 55131 Mainz, Germany; Anna.Wierczeiko@lir-mainz.de (A.W.); hristo.todorov@uni-mainz.de (H.T.) 2 Working Group Mouse Behavioral Unit (MBU), Leibniz Institute for Resilience Research (LIR), 55122 Mainz, Germany; radyushkin@uni-mainz.de 3 Working Group Healthy Aging and Neurodegeneration, Department of Psychiatry and Psychotherapy, University Medical Center, Johannes Gutenberg University, 55131 Mainz, Germany; lenagammel@googlemail.com (L.G.); VuThuThuy.Nguyen@unimedizin-mainz.de (V.T.T.N.) * Correspondence: sugerber@uni-mainz.de (S.G.); kristina.endres@unimedizin-mainz.de (K.E.) † These authors contributed equally to this work. Abstract: Physical activity is considered a promising preventive intervention to reduce the risk of developing Alzheimer’s disease (AD). However, the positive effect of therapeutic administration  of physical activity has not been proven conclusively yet, likely due to confounding factors such  as varying activity regimens and life or disease stages. To examine the impact of different routines Citation: Wierczeiko, A.; Gammel, L.; of physical activity in the early disease stages, we subjected young 5xFAD and wild-type mice to Radyushkin, K.; Nguyen, V.T.T.; 1-day (acute) and 30-day (chronic) voluntary wheel running and compared them with age-matched Todorov, H.; Gerber, S.; Endres, K. sedentary controls. We observed a significant increase in brain lactate levels in acutely trained 5xFAD Voluntary Wheel Running Did Not Alter Gene Expression in 5xfad Mice, mice relative to all other experimental groups. Subsequent brain RNA-seq analysis did not reveal but in Wild-Type Animals Exclusively major differences in transcriptomic regulation between training durations in 5xFAD mice. In contrast, after One-Day of Physical Activity. acute training yielded substantial gene expression changes in wild-type animals relative to their Cells 2021, 10, 693. https://doi.org/ chronically trained and sedentary counterparts. The comparison of 5xFAD and wild-type mice 10.3390/cells10030693 showed the highest transcriptional differences in the chronic and sedentary groups, whereas acute training was associated with much fewer differentially expressed genes. In conclusion, our results Academic Editor: Pyotr A. Slominsky suggest that different training durations did not affect the global transcriptome of 3-month-old 5xFAD mice, whereas acute running seemed to induce a similar transcriptional stress state in wild-type Received: 10 February 2021 animals as already known for 5xFAD mice. Accepted: 17 March 2021 Published: 20 March 2021 Keywords: physical activity; Alzheimer’s disease; 5xFAD; chronic; acute; wheel running Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affil- 1. Introduction iations. As the population ages, the number of people with degenerative diseases is simulta- neously increasing. Dementia is the fifth most frequent cause of death worldwide, with Alzheimer’s disease (AD) as the most prevalent type of dementia in the elderly [1]. AD is characterized by a progressive decline in memory functions, behavioral impairments, Copyright: © 2021 by the authors. and the loss of social skills, which means that quality of life begins to deteriorate well Licensee MDPI, Basel, Switzerland. This article is an open access article before death [2,3]. Key neuropathological features of this devastating disorder include the distributed under the terms and accumulation of extracellular β-amyloid (Aβ) plaques, intracellular neurofibrillary tangles conditions of the Creative Commons (NFT) composed of hyperphosphorylated tau proteins, and region-specific synaptic as Attribution (CC BY) license (https:// well as cellular degeneration, together with an increase in inflammation and oxidative creativecommons.org/licenses/by/ stress [4–7]. In addition, metabolic dysfunctions such as abnormal glucose uptake and 4.0/). brain insulin resistance, likely increasing degeneration and cognitive impairment, have also Cells 2021, 10, 693. https://doi.org/10.3390/cells10030693 https://www.mdpi.com/journal/cells Cells 2021, 10, 693 2 of 18 been described in AD patients [8]. Besides the known pathological hallmarks, the leading cause of progressive brain atrophy is still unknown, explaining the lack of successful AD treatments. It became, however, generally accepted that the underlying mechanisms are polyfactorial and depend on the complex interplay of multiple (partly unknown) genetic and non-genetic variables [9–12]. Furthermore, it is widely recognized that AD already seems to develop decades before clinical symptoms occur [13–16]. Since prophylactic pharmacological therapies in advance of a possible disease outbreak might be ethically not reasonable or hampered by lack of a long-term compliance, lifestyle-modifying inter- ventions, and preventions such as diet and physical activity have become increasingly attractive in the neurodegenerative research field. Physical activity (e.g., aerobic exercise or resistance training) has already been associ- ated with several beneficial effects that reduce the risk of developing AD in both, humans and animal models [17–20]. The duration of physical activity being studied can be divided into acute, one single bout of physical activity, and chronic, consistent activity over an extended time period [21]. Several studies suggested that regular training can improve reaction time, increase the size of the hippocampus (the brain region mainly responsible for learning and memory), and delay age-related memory impairments as well as the overall cognitive decline in older individuals with and without mild-cognitive impairment (MCI) or early AD [18,22–24]. Even after just one exercise session, short- and long-term memory and immune cell function were presumably stimulated in normal aging and MCI cohorts [25–27]. In transgenic rodent AD models, chronically trained animals exhibited a reduction in Aβ deposition and tau phosphorylation, anti-inflammatory modifications, improved brain oxygenation, and general cognitive improvements as compared to seden- tary littermates [13,28–33]. Furthermore, another study showed that only three days of voluntary wheel running was associated with enhanced neurogenesis in rats [34]. However, there is still insufficient evidence that physical activity provides an effective method of fighting AD, as there are also several studies that have not observed a positive link between physical activity and brain function improvement [35–39]. Recently, Hansson et al. conducted a large longitudinal study over 20 years that indicated no reduction in the frequency of developing AD by comparing exercising participants (n = 197,685) with seden- tary individuals (n = 197,684) [37]. The same results were obtained in a subsequent study on chronically trained AD model mice (5xFAD) [37]. The evaluation of neuroinflammation and non-cognitive parameters in voluntarily running young 5xFAD mice, in which no pathological features had yet developed, also showed no changes compared to sedentary mice. Moreover, signs of disease acceleration could even be identified within the trained transgenic group [38]. The inconsistency of physical activity effects on cognitive function and AD could likely be explained by different starting points such as age or disease state, duration (e.g., acute or chronic), and activity types [18]. On the neurochemical level, specific neurotransmitters, neuromodulators, and lactate concentrations were shown to increase after short and long-term exercise [40]. Furthermore, some genes and proteins that are predominantly involved in neurogenesis were suggested to be differentially expressed after both, acute and regular physical activity, in humans and in animal models, with brain-derived neurotrophic factor (BDNF) being the most prominent [34,41,42]. The global transcriptional response to different types of physical activity in AD has rarely been investigated. To better understand the molecular biology behind physical activity in AD, we examined in this study the blood and brain lactate levels as well as the transcriptomes of 3-month-old transgenic 5xFAD and wild-type mice after one day (acute) or four weeks (chronic) voluntary wheel running. Comparing the gene expression of 5xFAD mice at an age at which the pathology is not severely manifested with age-matched controls could provide further insights into fundamental molecular mechanisms of short- and long-term exercise in both AD and healthy individuals. Cells 2021, 10, 693 3 of 18 2. Materials and Methods 2.1. Animals Male B6SJL-Tg(APPSwFlLon, PSEN1*M146L*L286V)6799Vas/Mmjax (5xFAD) mice (Jackson Lab, Bar Harbor, Maine, USA) were crossbred with female C57BL/6J mice and maintained as heterozygous transgenics on the C57BL/6J background. The animals were single-caged in type II cages with a 12-h day/night cycle. Food and water were available ad libitum. The non-transgenic littermates were used as the control. All procedures were performed in accordance with the European Communities Council Directive regarding care and use of animals for experimental procedures and were approved by local authorities (Landesuntersuchungsamt Rheinland-Pfalz; approval number G 14-1-087). Mice were randomly assigned to three groups: untrained (sedentary), access to a saucer wheel (15-cm diameter; med associates Inc.) for 30 days, or access to a saucer wheel for 18 h before sacrifice. The saucer wheel contained a WLAN connection to a computer for measuring running counts. Food consumption and body weight were assessed regularly. 2.2. Tissue Preparation Mice were sacrificed after isoflurane anesthesia, and truncal blood was collected. Blood glucose was measured immediately from two independent droplets with an AccuCheck mobile device (Roche). Heart, muscle, abdominal fat, and the left brain-hemisphere were dissected and shock-frozen in liquid nitrogen with subsequent storage at −80 ◦C. The right hemisphere was cut into small cubicles and incubated at Room Temperature for 20 min in RNAlater (Qiagen, Hilden, Germany) before being stored at −80 ◦C. Serum was obtained from truncal blood by two-times centrifugation after clotting and stored at −80 ◦C. 2.3. Lactate Measurement Brain samples were homogenized with a Tissue Lyzer, and stainless-steel beads (Qi- agen, Hilden, Germany) in 600 µL 0.5M KH2PO4-buffer supplemented with protease inhibitor (cOmplete, Roche) pH7 for 5min at 50 Hz. 20 µL of the homogenate and 2.5 µL of serum were subjected to the enzymatic assay. Lactate was measured following the protocol from Lin and colleagues [43]. In brief, the assay uses coupling of lactate-oxidase and peroxidase for the conversion of 2,2′-azino-bis (3-ethylbenzothiazoline-6-sulfonic acid) (ABTS, Sigma Aldrich, Steinheim, Germany). As a standard, lithium-lactate was used, and the absorbance of the chromogenic product was measured at 405 nm. The protein content of all samples was assessed with Roti-Nanoquant (Roth) and used for normalization. 2.4. RNA-Seq Library Preparation and Sequencing Next-generation sequencing library preparation was performed with Illumina’s TruSeq stranded mRNA LT Sample Prep Kit following Illumina’s standard protocol (Part # 15,031,047 Rev, E). The libraries were prepared with a starting amount of 302 ng and amplified in 12 PCR cycles, profiled in a DNA 1000 Chip on a 2100 Bioanalyzer (Agilent Technologies) and quantified using the Qubit dsDNA HS Assay Kit, in a Qubit 2,0 Fluo- rometer (Life technologies). All 24 samples were pooled in equimolar ratio and sequenced on 1 NextSeq 500 Highoutput FC, PE for 2 × 42 cycles plus 7 cycles for the index read. 2.5. RNA-Seq Analysis After merging five technical replicates and down-sampling one sequencing file, the reads were trimmed using BBDuk (version 38.06) [44]. For the differential expression analysis, the trimmed reads were mapped to the UCSC reference genome of Mus musculus obtained through Illumina iGenomes (version mm10, latest release in May 2012) [45], using the splice-aware mapper STAR (version 2.7.0d, default options of 2-pass mapping) [46]. The aligned reads were then counted per gene and sample (based on the mm10 UCSC annotation file from iGenomes) using FeatureCounts provided by SubRead (version 1.6.2, default options for single-end reads) [47]. Cells 2021, 10, 693 4 of 18 The differential expression analysis (DEA) was done in R (version 4.0.3) and RStudio (version 1.3.1073) using the R package DESeq2 (version 1.30.0) [48–50]. In order to reduce the impact of extreme expressions, three genes, Psen1, App, and Thy1, which are known to be highly differentially expressed due to the genotype of 5xFAD mice [51], were removed from the count tables. The normalized gene counts of these marker genes can be found in Supplementary Figure S1. After normalizing the filtered gene counts by the size factors provided by DESeq2, a principal component analysis was performed for all mice and each genotype separately [52]. Additionally, PCA was conducted based on the separate training groups and is shown in the Supplementary Figure S2. The p-values of the DEA were adjusted using the Benjamini–Hochberg method, and the cut-off for significant differential regulation was set at adjusted p-value < 0.05 [53]. Gene Ontology (GO) term enrichment and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses were performed for all subsets of differentially regulated genes using the R package clusterProfiler (version 3.18.0) [54]. Furthermore, the gene subsets were examined for brain cell-type enrichment using the Fisher’s Exact Test implemented in R [55]. The cell-type-associated gene list consisted of five brain cell types, including neurons, astrocytes, oligodendrocytes, microglia, and endothelial cells, and is based on a previously published brain transcription analysis [56,57]. The gene universe (i.e., the gene’s background set for the statistical analysis) was set to the genes considered for the differential expression analysis for all tests. After adjusting the p-values of the overrepresentation test using the Benjamini–Hochberg method, the terms with an adjusted p-value < 0.05 were considered as significantly enriched. All plots were created using the R packages ggplot2 (version 3.3.2) and venn (version 1.9) [58,59]. The raw fastq files and the count table were uploaded to Gene Expression Omnibus (GEO) under the accession number GSE164798 (see Data Availability Statement). 3. Results 3.1. Voluntary Running Wheel Usage in 5xFAD Mice Physical activity has often been discussed as beneficial in AD or animal models of the disorder. To analyze the potential impact of acute and chronic physical voluntary training, we investigated male mice with the 5xFAD genotype and their wild-type littermates, starting at the age of 2 months. At this age, no behavioral deficits have been reported in the mice; however, deposition of Aβ is observable already at 1.5 months of age [51]. Mice were single-caged and divided into three groups: an untrained group with no access to a saucer wheel, a chronically trained group with full access over the 30 days of the experiment, and the third group with access only within the last night before sacrifice (see Figure 1A). Running activity was assessed by a wireless reporting system: within the first night after adding the saucer wheel, animals of both genotypes quickly adjusted to the new enrichment during the first hours after being exposed to the saucer wheel. This is indicated by a significant count increase in both genotypes starting at 22 p.m. (comparison vs. starting point at 10 a.m., Figure 1B). No apparent differences occurred in the starting phase in both genotypes despite slight but non-significantly higher arousal in 5xFAD mice. Approximately 150 counts for wheel turning per hour were reached at 12 a.m. With prolonged access to the saucer wheel, counts as high as 2800 per hour were reached on average (Figure 1C) with a clear peak in the early dark phase (between 6 and 10 p.m). 5xFAD mice showed a statistically significant elevation of wheel usage at several points of time, resulting in an increased mean running distance (5xFAD: 8.6 ± 0.32 km per night, wild-type: 7.5 ± 0.41 km per night) but no change in the circadian rhythm of usage. Cells 2021, 10, 693 Cells 2021, 10, x FOR PEER REVIEW 5 of 18 5 of 19 Figure 1. TrFaiignuinrge g1r.ouTprasi nanindg thgero cuopmspaanridsotnh eofc foimrstp narigishotn anodf fichrsrtonniicg hsatuacnedr wchhreoenl iucssaaguec. e(Ar )w Shceheelmuastaicg eo.f the three groups used(A f)orS cthheem inatvicesotfigtahteiothnr. eAe lgl raonuipmsaulss ewdefroer tshinegilnev-ceastgiegdat iaonnd. Akellpatn iinm taylspew eIIr emsaincgrolelo-cna gceadgeasn dfokr e3p0t days. The untrained grinoutpyp headII nmo aaccrcoelsosn toc aag seasufcoerr 3w0hdeealy, st.heT chheruonnitcraalilnye tdraginroeudp grhoaudpn koepatc ctheses wtoheaels aouvecer rthwe hwehelo,lteh teime period, and the acuctehlryo ntriacianlleydt rmaiicne dhagdro aucpceksesp otntlhye fworh teheel olavsetr nthigehwt. hWolheeteilm tuerpneirnigo dc,oaunndtst hpera chuotuerly wtrearien medeamsiucreed for new access to theh saaduaccecre wsshoenel y(Bfo) ranthde tlhaes tcnhirgohnti.c Wushaegeel (tCur).n Dinagtac oauren tpsrepseernhteodu raws mereamn eoaf stuhree pdefrofrornmewanaccec oevssert othe 30 days of the chrontihcaelslya utrcaeirnwedh emelic(eB )(na n= d10th feorc hwroilndi-ctyupsea,g ne =(C 8) .foDra 5taxFaAreDp rmesiceen)t.e Edrraosrm beaarns aorfet hneotp veirsfuoramlizaendc eforv celrarity of the graph. Statisthtieca3l0 adnaalyyssoisf: tmheulcthiprolen uicnapllayirterdai nt-etedsmts i(c*e p( n< 0=.0150; f*o**r pw