Received: 30 December 2020 Revised: 30 November 2021 Accepted: 2 February 2022
DOI: 10.1002/elps.202000393
RESEARCH ARTICLE
Protein speciation is likely to increase the chance of
proteins to be determined in 2-DE/MS
Marcel Kwiatkowski1 Madlen Hotze1 Julia Schumacher2 Abdul R. Asif3
Jose Miguel Ramos Pittol1 Bertram Brenig4 Sanja Ramljak5 Hans Zischler6
Holger Herlyn6
1Department of Biochemistry and Center
for Molecular Biosciences Innsbruck, Abstract
University of Innsbruck, Innsbruck, Multiple spotting due to protein speciation might increase a protein’s chance of
Austria being captured in a random selection of 2-DE spots. We tested this expectation in
2Abbott GmbH, Core Diagnostics,
Wiesbaden, Germany new (PXD015649) and previously published 2-DE/MS data of porcine and human
3Department of Clinical tissues. For comparison, we included bottom-up proteomics studies (BU-LC/MS)
Chemistry/UMG-Laboratories, University of corresponding biological materials. Analyses of altogether ten datasets pro-
Medical Center, Göttingen, Germany posed that amino acid modification fosters multispotting in 2-DE. Thus, the
4Department of Molecular Biology of
Livestock, Institute of Veterinary number of 2-DE spots containing a particular protein more tightly associated
Medicine, University of Göttingen, with a peptide diversity measure accounting for amino acid modification than
Göttingen, Germany with an alternative one disregarding it. Furthermore, every 11th amino acid was
5Digital Diagnostics AG, Mainz, Germany a post-translational modification candidate site in 2-DE/MS proteins, whereas in
6Institute of Organismic and Molecular BU-LC/MS proteins this was merely the case in every 21st amino acid. Alterna-
Evolution, Anthropology, University of
Mainz, Mainz, Germany tive splicing might contribute to multispotting, since genes encoding 2-DE/MS
proteins were found to have on average about 0.3 more transcript variants
Correspondence
Holger Herlyn, Institute of Organismic than their counterparts from BU-LC/MS studies. Correspondingly, resolution
and Molecular Evolution, Anthropology, completeness as estimated from the representation of transcript variant-rich
University of Mainz, genes was higher in 2-DE/MS than BU-LC/MS datasets. These findings suggest
Anselm-Franz-von-Bentzel-Weg 7, 55128
Mainz, Germany. that the ability to resolve proteomes down to protein species can lead to enrich-
Email: herlyn@uni-mainz.de ment of multispotting proteins in 2-DE/MS. Low sensitivity of stains and MS
instruments appears to enhance this effect.
Color online: See article online to view
Figs. 1–3 in color.
KEYWORDS
Funding information
University of Mainz alternative splicing, post-translational modification, protein abundance, protein species, two-
dimensional gel electrophoresis
Abbreviations: AS, Alternative splicing; 2-DE/MS, 2-DE followed by mass spectrometry; BU-LC/MS, bottom-up proteomics based on liquid
chromatography coupled to MS; IEF, isoelectric focussing; ∆PD, extent to which post-translational modifications increase peptide diversity; FA,
formic acid; FDR, False discovery rate; GO, gene ontology; NTV, number of transcript variants per gene; PD, peptide diversity with modifications
disregarded; PMI, post-translational modification index; PTM, post-translational modification; RT, room temperature; TPM, transcripts per million.
This is an open access article under the terms of the Creative Commons Attribution-NonCommercial License, which permits use, distribution and reproduction in any
medium, provided the original work is properly cited and is not used for commercial purposes.
© 2022 The Authors. Electrophoresis published by Wiley-VCH GmbH.
Electrophoresis 2022;43:1203–1214. www.electrophoresis-journal.com 1203
 15222683, 2022, 11, Downloaded from https://analyticalsciencejournals.onlinelibrary.wiley.com/doi/10.1002/elps.202000393 by Cochrane Germany, Wiley Online Library on [06/02/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1204 KWIATKOWSKI et al.
1 INTRODUCTION Laurasiatheria (Mammalia), that is, domestic pig (Sus
scrofa) and anatomical modern human (Homo sapiens)
Two-dimensional gel electrophoresis (2-DE) has proven to [37, 38]. The protein sets covered the soma and germline
be a powerful tool for resolving proteomes [1, 2]. However, of single species (e.g., human testis and heart) and cor-
it also appears that, depending on the details of a protocol, responding biological material of different species (e.g.,
protein sets from 2-DE/MS can have specific properties. human and porcine testis). In this study, we generated
For example, underrepresentation of membrane proteins a new set of testicular proteins from boars by 2-DE/MS
can occur [3], whereas highly abundant proteins might be (Supporting information Table S1). The nine other protein
overrepresented [4]. Samples from 2-DE/MS may also be sets were retrieved from tabular surveys of previous
enriched or depleted in proteins showing higher molec- publications as detailed in Table 1 [39–47] and Supporting
ular mass [5, 6]. Such shifts in the composition are not information Table S2.
attributable to 2-DE/MS per se [3, 6–8], which offers the
advantage of high resolution down to the content of single
spots instead [6–8]. The question addressed here is if the 2.2 Porcine testis proteome: 2-DE,
occurrence of single proteins in multiple spots might raise tryptic digestion, and LC-MS/MS analysis
their chance of being collected in 2-DE and determined in
downstream MS. Tissue pieces were cut from the ultra-frozen testicles of
Leaving aside differential conformation and complex three boars. Thawing at 4◦Cwas followed by solubilization
formation of proteins [9], protein speciation due to post- of the samples at room temperature (RT) in lysis buffer
translational modifications (PTMs) will be a major factor containing 7 M urea, 2 M thiourea, 4% CHAPS, 1.5% DTT
causing multiple spotting of proteins in 2-DE [8, 10–14]. (w/v), 1% ampholytes (pH 3−10; Bio-Rad), and 10 µL pro-
In fact, modifications, such as phosphorylation [15, 16], tease inhibitor cocktail (Sigma-Aldrich). After pelleting of
deamidation [17], oxidization [18], and glycosylation [19, cell debris by centrifugation (1min atmaximum speed), we
20], have the potential to drastically raise the number purified the supernatant using the ReadyPrep 2-DCleanup
of protein species which partially will differ in MW Kit (Bio-Rad) according to the manufacturer’s protocol.
and pI. Alternative splicing (AS) of mRNAs could be Proteins were rehydrated in the same buffer as above but
an additional source of multiple spotting [10–13]. The without protease inhibitor cocktail. Upon determination
mechanism is likely to be important for cell and tissue of protein concentration (Bradford assay; Bio-Rad), 250–
differentiation [21–25] and strongly increases transcript 400 µg of total protein was placed on immobilized pH
diversity [26–28]. In humans, for example, AS produces gradient strips (7 cm IPG ReadyStrip, pH 5–8; Bio-Rad), for
more than 140 000 transcript variants [29–31] from about passive rehydration (2 h at RT). Subsequently, strips were
20 300 [14] or 22 500 (Ensembl Genes 103) protein-coding covered with mineral oil for active rehydration for 14 h at
genes. Notwithstanding the nonsense-mediated decay of 50 V. One-dimensional separation by isoelectric focussing
many transcripts, some of the splice variants will be trans- (IEF) was conducted at 20◦C in a Protean IEF cell (Bio-
lated into alternative protein species [32–36] with specific Rad) according to a previously published protocol [48]: 2 h
migration patterns in 2-DE. Whether by AS or alternative at 200 V, 2 h at 500 V, and 5 or 6 h at 4000 V (increments
amino acid modification, the resulting multispotting with rapid ramp). We equilibrated IPG strips for 25 min in
should increase the probability of a protein being detected a buffer containing 6 M urea, 0.375 M Tris (pH 8.8), 30%
in shotgun 2-DE/MS in at least one protein species. In the glycerol, 2% SDS, and 2% DTT. Following this, the strips
current study, we test this prediction in new 2-DE/MS data were exposed for 25 min to the same buffer supplemented
on the porcine testicular proteome. A meta-analysis of with 2.5% iodoacetamide instead of DTT, and 20 µL of 1%
five datasets from 2-DE/MS and another five comparative bromophenol blue. Equilibrated strips were placed on 10%
datasets from bottom-up proteomics (BU-LC/MS) follows. polyacrylamide gels, followed by electrophoresis at 120 or
Indeed, effects related to the 2D resolution of proteomes 140 V. We ran three gels (technical replicates) for each of
should not be reflected in the latter type of studies. the three boars (biological replicates). The MW standard
used was Kaleidoscope (Bio-Rad). Following staining with
colloidal Coomassie-Brilliant Blue G250 (Roti-Blue, Roth),
2 MATERIALS ANDMETHODS gels were scanned with a GS-800 calibrated densitometer
(Bio-Rad).
2.1 Protein sets We excised arbitrarily selected 2-DE spots using biopsy
punches. Gel slices were vacuum dried, soaked with
We collected five protein sets from 2-DE/MS and digestion buffer supplemented with trypsin (0.01 µg/µL),
BU-LC/MS studies, each, representing two species of and incubated overnight at 37◦C in the same buffer
 15222683, 2022, 11, Downloaded from https://analyticalsciencejournals.onlinelibrary.wiley.com/doi/10.1002/elps.202000393 by Cochrane Germany, Wiley Online Library on [06/02/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
KWIATKOWSKI et al. 1205
TABLE 1 References and methods applied for gathering the protein sets analyzed
Species Tissues Methods Instruments References
Pig Testis 2-DE/MS LC-MS/MS This study
BU-LC/MS LC-MS/MS [39]
Human Testis 2-DE/MS MALDI-TOF [40]
BU-LC/MS LC-MS/MS [41]
Human Prefrontal cortex 2-DE/MS MALDI-TOF [42]
BU-LC/MS LC-MS/MS [43]
Human Heart 2-DE/MS MALDI-TOF [44]
BU-LC/MS LC-MS/MS [45]
Human Kidney 2-DE/MS MALDI-TOF [46]
BU-LC/MS LC-MS/MS [47]
Note. BU-LC/MS, bottom-up proteomics; 2-DE/MS, two-dimensional gel electrophoresis followed by proteomic analysis of single spots.
devoid of trypsin (5 mM CaCl2, 25 mM NH4HCO3). to 4 Da and MS/MS fragmentation amplitude to 1.25 V.
Peptides were extracted the next day by sonication in Precursor ions were actively excluded for 1 min. Peak lists
solvents with increasing ACN contenteration. Following for MS/MS database search (*.mgf) were generated using
vacuum-centrifugation and subsequent reconstitution in Data Analysis Software for 6300 Series Ion Trap LC/MS
0.1% formic acid (FA), peptides were separated by RPLC version 3.4 (Agilent Technologies) by applying default soft-
(C18; buffer A: 0.1% FA dissolved in HPLC-H2O; buffer B: ware settings, except for the following variations: (i) for
0.1% FA, dissolved in ACN) using a flow rate of 0.4 µL/min mass spectrum calculation, additional background sub-
and a gradient of 2–30% in 30 min. Eluting peptides were traction was performed using peak start and end spectra;
introduced via an ESI interface into a Q-TOF mass spec- (ii) after extraction of MS/MS data, spectra were deconvo-
trometer (one boar, Q-TOF Ultima, Micromass/Waters, luted using default parameters.
Manchester, UK) and an ion-trap mass spectrometer (two For Q-TOF analysis, the MS instrument was coupled to
other boars, XCT ion-trap, Agilent Technologies, Wald- a CapLC system (Waters, Eschborn, Germany). A total of
bronn, Germany), and analyzed as described elsewhere 1 µL per sample was loaded onto a μ-precolumn cartridge
[49, 50]. (C18 pepMap, 300 µm × 5 mm; 5 µm particle size, LC
The XCT ion-trap was equipped with an Agilent HPLC- Packings, Germering, Germany) using a flow rate of
Chip Cube interface integrated into an 1100 series HPLC 5 µL/min and 5% buffer B (buffer A: 0.1% FA prepared in
system (Agilent Technologies). The HPLC-chip (Large 5% ACN; buffer B: 0.1% FA prepared in 95% ACN). The
capacity chip, Agilent Technologies) was fitted with an peptides were separated on an analytical column (C18
enrichment column (internal volume 160 nL, 5 µm Zor- pepMap100 nano Series, 75 µm × 15 cm; 3 µm particle size,
bax 300 SB-C18 material), a separation column (150 mm LC Packings) using a flow rate of 0.25 µL/min and a gradi-
× 0.075 mm, 5 µm Zorbax 300 SB-C18 material), and ent from 10 to 95% buffer B in 50 min. The Q-TOF Ultima
a nanospray emitter. In detail, samples were loaded MS system was equipped with a nanoflow ESI Z-spray
(8 µL/sample) on the enrichment column using a flow rate source. MS analysis was performed in positive ion mode.
of 4 µL/min with the mix of the two following mobile The nanospray needle was held at 2000 V and the source
phases at a ratio 98:2 (mobile phase A: 0.2% FA in H2O; temperature at 40◦C. MS/MS analysis was performed in
mobile phase B: 100% ACN). Tryptic peptides trapped on DDA mode. MS1 scans were recorded over a m/z range of
the enrichment column were disseminated on the sepa- 400–1600. Multiply charged precursor ions were selected
ration column using a flow rate of 0.4 µL/min and a lin- in the quadrupole using a selection window of 2 Da, frag-
ear gradient of 2–40% eluent B within 40 min. For IT-MS mented in the hexapole collision cell using a collision ramp
analysis, the following parameters were used: MS scan- from 20 to 35 eV and their fragment ions were recorded
ning range from 300 to 2000m/z, electrospray voltage was in the TOF mass analyzer over an m/z range of 100–1600.
set to −1750 V, nitrogen was used as drying gas for des- Fragmented precursor ions were excluded from DDA
olvation at a flow rate of 4 L/min, and 325◦C. Precur- analysis for 1 min. Data acquisition was performed using
sor ion mass spectra were acquired in positive ion mode MassLynx (v 4.0, Waters). MS/MS peak lists (*.pkl) were
with automated data-dependent MS/MS of the three most generated using ProteinLynx Global Server bioinformatics
intense ions of each precursor MS scan. Doubly charged tool (PLGS; v 2.2; Waters) and the following parameters:
ions were isolated preferentially. Isolation width was set raw spectra were smoothed (savitzky golay, smooth
 15222683, 2022, 11, Downloaded from https://analyticalsciencejournals.onlinelibrary.wiley.com/doi/10.1002/elps.202000393 by Cochrane Germany, Wiley Online Library on [06/02/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1206 KWIATKOWSKI et al.
window channel: 3, number of smooths: 2), centered newly generated data on the porcine testicular proteome
(min peak width at half height: 4, centroid to: 80%), noise which measure closer associates with the number of 2-
reduction 10%, medium deisotoping with 3% threshold. DE spots containing a particular protein, peptide diver-
The generated pkl-files were converted to mgf-files. sity with modifications disregarded (PD), or the extent to
We used ProteomeDiscoverer 2.4 (Thermo Fisher Scien- which modifications increased peptide diversity (∆PD).
tific, San Jose, USA) for peptide and protein identification. Both variables and their basic data are reported in Sup-
Peak lists (*.mgf) were searched with Sequest HT against porting information Table S1. For the inference of PD, we
the S. scrofa reference proteome database (UniProt Pro- divided the number of peptides with modifications dis-
teome ID: UP000008227, 49 793 proteins). The searches regarded (thus, when considering EGWsDSTYGVTK and
were performed with precursor mass tolerance of 30 ppm EGWSDSTYGVTK as a single peptide) by the amino acid
(Q-TOF) or 2 Da (IT) and fragment mass tolerance of span of the particular protein species determined. By doing
0.1 Da (Q-TOF) or 0.5 Da (IT). For identification, the fol- so, we accounted for the expectation of larger peptide
lowing variable modifications were considered: oxidation diversity in longer amino acid sequences. For deriving
of methionine, deamidation of glutamine and asparagine, ∆PD, we subtracted the number of peptides with modifi-
phosphorylation of threonine, serine and tyrosine, acety- cations disregarded from the peptide number inclusively
lation of lysine, protein N-terminal acetylation, protein N- modifications (thus, when regarding EGWsDSTYGVTK
terminal methionine loss, protein N-terminal methionine and EGWSDSTYGVTK as two peptides), prior to division
loss plus acetylation. A carbamidomethylation of cysteine by amino acid span. Partial rank correlations were then
residues was considered as a static modification. The Per- carried out between 2-DE spot number on the one hand
colator function implemented in Sequest was used to cal- and either PD or∆PD on the other hand, using SPSS 23 V5
culate posterior error probabilities and q-values for the R. The influence of the uncorrelated third parameter was
identified peptide-spectrum matches. Threshold for cor- always controlled for.
rect peptide identification was set to a q-value of 0.05. The Using ShinyGO, we continued with investigating
peptides determined are listed in Supporting information whether frequencies in the number of transcript vari-
Table S1. The results were verified exemplarily for some of ants per gene (NTV) differed between the genes encoding
the samples in the Mascot search engine. We matched the porcine testicular proteins from the current 2-DE/MS anal-
proteins derived in the present study and two previous 2- ysis (query set) and the rest of porcine genes (Chi-square
DE/MS analyses (MALDI-TOF and MALDI-TOF/TOF) of test). We also evaluated whether the NTV frequency dis-
the porcine testicular proteome [51, 52], to recognize repro- tribution in our query set differed from the genome-wide
duced proteins. expectation. For comparison, ShinyGO was additionally
run on porcine testicular proteins from BU-LC/MS anal-
ysis. An analogous procedure was applied to four pairs of
2.3 Bioinformatics and statistics 2-DE/MS and BU-LC/MS studies on human testis, kidney,
liver, and prefrontal cortex (Table 1).
The applications outlined below were run with Ensembl Using Biomart (Ensembl Genes 103), we retrieved NTV
andUniProt IDs aswell as gene symbols. Thesewere either values for single genes in the 2-DE/MS and BU-LC/MS
copied directly from publications referenced in Table 1 or studies which are reported in Supporting information
had been retrieved beforehand using Ensembl’s Biomart, Table S3 along with the IDs of the encoded transcript
UniProt’s ID mapping tool, STRING version 11.0 (https: variants. Based on this, we quantified the difference
//string-db.org/), or ShinyGO version 0.61 which relies on between mean NTV (N̄TV) in 2-DE/MS and BU-LC/MS
annotations in EnsemblGenes 96 [53]. For easier handling, datasets, separately for each species. We additionally used
we confined analyses of larger gene and protein lists to the NTV values for derivation of the proportion of tran-
2000 randomly selected IDs per study. Thus, the gene lists script variant-rich genes, separately for the species and
(Supporting information Table S2) that entered the cur- analysis pipelines. For definition of transcript variant-rich
rent bioinformatics pipeline were smaller than in the ref- genes, we arbitrarily set thresholds between the lower 80%
erences fromwhich they were extracted. In addition, some and upper 20% of the NTV ranges in the aforementioned
genes and proteins were notmatched in single approaches. frequency distributions from ShinyGO analyses. Thus,
We considered this unproblematic as it should not have porcine genes having >4 and human genes having >11
affected the results. protein-coding transcript variants were regarded as tran-
We started with partial correlation analyses of the script variant-rich. Their numbers were divided by the
newly generated porcine data. In the case of two proteins, total numbers of protein-coding genes in 2-DE/MS and
which had been identified in two protein species, each, BU-LC/MS datasets, to obtain proportions of transcript
we included the longer ones. First, we examined in the variant-rich genes. Corresponding values were compared
 15222683, 2022, 11, Downloaded from https://analyticalsciencejournals.onlinelibrary.wiley.com/doi/10.1002/elps.202000393 by Cochrane Germany, Wiley Online Library on [06/02/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
KWIATKOWSKI et al. 1207
between 2-DE/MS and BU-LC/MS datasets, but also
between each of these and shares of transcript variant-rich
genes in the entire porcine and human protein-coding
genes, respectively. By doing so, we estimated the level of
resolution attained in MS analyses.
Downstream steps focused on human proteins due to
better data availability. Thus, we collected naturally occur-
ring PTMs from ProteomeScout (http://proteomescout.
wustl.edu/compendia; visited on 03/04/2021) [54]. Their
number divided by amino acid number was taken as a
proxy for the propensity to form protein species due to
PTM (Supporting information Table S4). We refer to this
proxy as to the PTM index (PMI). In addition, transcript
abundances (transcripts per million, TPM) were extracted
for human genes (Supporting information Table S5) from
the gtexGeneV8 database at the European mirror site of
the UCSC Table Browser (http://genome-euro.ucsc.edu/ F IGURE 1 Two-dimensional SDS-PAGE gel run withtesticular lysate of boar 1. The gel image corresponds to technical
cgi-bin/hgTables). Levels of PMI and TPM were then con- replicate A in this boar and is representative of three technical
trasted between human 2-DE/MS and BU-LC/MS datasets replicates per each of the three boars analyzed (compare Supporting
by the MWU test in SPSS. information Figures S2, S3). Circles highlight 2-DE spots examined
For the human data, we further tested for correlations by MS. Their numbering matches with respective entries in
between the three parameters collected: PMI, TPM, and Supporting information Table S1. Raw scans of all technical
NTV. For this purpose, we pooled the proteins from the 2- replicates are provided as Supporting information
DE/MS and BU-LC/MS studies of the human tissues and Figures S3-S11
kept only those for which all three parameters were avail-
able. Subsequent analyses were again performed as partial
rank correlations in SPSS, thus, controlling for the uncor- we foundmatches in previous 2-DE/MS studies on porcine
related variable once again. testis (Supporting information Table S1).
All tests conducted were two-sided. Alpha error rates (p A closer look at the data revealed that amino acid modi-
values) were converted into false discovery rates (FDRs) ficationsmight have raised a protein’s chance of being cap-
applying the method of Benjamini and Hochberg [55], tured. In fact, 46% of the proteins were determined from
accounting for altogether 17 tests (ten Chi-squared tests, two or more 2-DE spots (Figure 2; Supporting informa-
five partial rank correlations, and two MWU tests). tion Table S1). Subsequent partial rank correlations under-
Finally, we performed gene ontology (GO) analyses to scored that variation due to amino acid modifications
assess the physiological involvements of the proteins in the could have fostered multispotting in 2-DE of the porcine
two porcine and four human dataset pairs. For this pur- testis proteome (Table 2). Thus, the number of 2-DE spots
pose, we called PANTHER GO-Slim Molecular Function from which a protein was recovered more tightly asso-
annotations (http://pantherdb.org/, visited on 05/12/2021) ciated with ∆PD, that is, the extent to which modifica-
for the IDs reported in current Supporting information tions increased peptide diversity, than with PD, that is,
Table S2. the diversity of peptides with modifications disregarded.
This was reflected in a more than threefold larger coef-
ficient in partial rank correlation of 2-DE spot number
3 RESULTS with ∆PD (rho = 0.429; FDR < 0.001) relative to a corre-
sponding correlationwith PD (rho= 0.128; FDR< 0.05). In
Present 2-DE/MS analysis of porcine testis led to the detec- line with this, peptides with alternative modification pat-
tion of 248 protein species from 82 2-DE spots (Figure 1; terns were repeatedly detected in different 2-DE spots. For
Supporting information Figures S1-S11). Average coverage example, TFTDcFNcLPIAAIVDEK and TFTDcFncLPI-
was 24%, with a span of 3 to 93% (Supporting informa- AAIVDEK were recovered from 2-DE spots nos. 23 and
tion Table S1). Theoretical pI and MW ranged from 4.77 28, respectively (A0A287A6R4 in Supporting information
to 11.85 and 14.2 to 116.1 kDa, respectively. Moreover, two Table S1).
proteins (cathepsin D and pre-mRNA-processing factor 19) According to ShinyGO analysis (Ensembl Genes 96), AS
were determined in two protein species, each, which were could have played a contributory role in the multispotting
derived from alternative transcripts. For 22 of the proteins, of porcine testicular proteins. Thus, distributions of NTV
 15222683, 2022, 11, Downloaded from https://analyticalsciencejournals.onlinelibrary.wiley.com/doi/10.1002/elps.202000393 by Cochrane Germany, Wiley Online Library on [06/02/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1208 KWIATKOWSKI et al.
F IGURE 2 Frequency distribution of the number of 2-DE spots, from which a protein was determined, in present 2-DE/MS analysis of
the porcine proteome. Out of the 246 proteins determined, 46% occurred in more than a single 2-DE spot (light bars)
TABLE 2 Results of partial rank correlations tations, according to which genes underlying 2-DE/MS-
Controlled determined porcine proteins (N̄TV = 3.253) on average
Parameter pair variable NProteins rho FDRa had 0.265 more transcript variants than the ones encoding
∆PD vs. N2-DE Spots PD 246 0.429 <0.001 proteins from BU-LC/MS analysis of porcine testis (N̄TV
PD vs. N2-DE Spots ∆PD 246 0.128 <0.05 = 2.988) (Supporting information Table S3). Consistently,
TPM vs. N PMI 5365 0.133 <0.001 the proportion of transcript variant-rich genes was higherTV
TPM vs. PMI N 5365 0.363 <0.001 in the 2-DE/MS than in the BU-LC/MS study, althoughTV
NTV vs. PMI TPM 5365 –0.055 0.001
this proportion was increased in both experimental
<
datasets when compared to the genome-wide expectation
Note.∆PD, degree bywhich amino acidmodifications increased peptide diver-
sity; N2-DE Spots, number of 2-DE spots from which a protein was recovered;
(Table 4).
NTV, number of transcript variants (per coding gene); PD, peptide diversity; We observed similar patterns in 2-DE/MS and BU-
PMI, post-translational modification index; rho, correlation coefficient; TPM, LC/MS representing human tissues, particularly testis, but
transcripts per million. also heart, kidney, and prefrontal cortex. In all correspond-
aFalse discovery rates as derived from partial rank correlations. ing query sets, NTV frequencies differed from the rest of
genes in the human genome (FDR < 0.01, Chi-square test,
frequencies significantly differed between the genes cod- each; Table 3). Furthermore, frequency distributions per-
ing for the proteins determined by us and the rest of the sistentlyweremore strongly shifted toward higherNTV val-
genes in the pig (FDR ues in human query sets from 2-DE/MS than BU-LC/MS< 0.001, Chi-square test; Table 3).
Compared to genome-wide frequencies, the distribution studies (Supporting information Figure S12). Correspond-
was shifted toward higher N values in genes encoding ingly, a human gene coding for a 2-DE/MS-determinedTV
porcine testicular proteins from present 2-DE/MS analysis protein (N̄TV = 5.806) had on average 0.330 more protein-
(Figure 3A). In their counterparts coding for pig proteins coding transcript variants than its counterpart encod-
from a BU-LC/MS study, the frequency distribution was ing a BU-LC/MS protein (N̄TV = 5.476; Supporting infor-
also shifted toward higher N values, relative to the mation Table S3). Also, the representation of transcriptTV
genome-wide expectation and the rest of porcine genes variant-rich genes more strongly exceeded the genome-
(FDR < 0.001, Chi-square test; Table 3). Nevertheless, the wide expectation in human genes coding for 2-DE/MS-
shift was stronger in the 2-DE/MS than BU-LC/MS sample determined proteins than in the ones encoding proteins
(Figure 3). This was verified in Ensembl Genes 103 anno- from BU-LC/MS studies (Table 4).
 15222683, 2022, 11, Downloaded from https://analyticalsciencejournals.onlinelibrary.wiley.com/doi/10.1002/elps.202000393 by Cochrane Germany, Wiley Online Library on [06/02/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
KWIATKOWSKI et al. 1209
TABLE 3 Sample sizes and comparison of per-gene number of transcript variants in query sets with the rest of genes in a given species
Species Biological material Methods N a bproteins FDR
Pig Testis 2-DE/MS 230 <0.001
BU-LC/MS 1396 <0.001
Human Prefrontal cortex 2-DE/MS 146 <0.001
BU-LC/MS 2000 <0.001
Human Heart 2-DE/MS 97 <0.001
BU-LC/MS 2000 <0.01
Human Kidney 2-DE/MS 180 <0.001
BU-LC/MS 1953 <0.001
Human Testis 2-DE/MS 433 <0.001
BU-LC/MS 2000 <0.001
Note. BU-LC/MS, bottom-up proteomics based on liquid chromatography coupled to MS; 2-DE-MS, two-dimensional gel electrophoresis followed by MS.
aNumber of protein-coding genes as recognized by ShinyGO v. 0.61.
bFalse discovery rates as derived from Chi-squared tests comparing NTV distributions between query sets and the rest of genes in a given biological species.
TABLE 4 Proportions of transcript variant-rich genes
Proportion of
transcript
Species Data variant-rich genes
Domestic pig Whole genome 0.170
2-DE/MS 0.201
BU-LC/MS 0.185
Human Whole genome 0.052
2-DE/MS 0.093
BU-LC/MS 0.087
Note. For abbreviations see legend to Table 5.
TABLE 5 Comparison of variation and abundance parameters
between human proteins from 2-DE/MS and BU-LC/MS and
respective transcripts
Species Parameter Method NProteins FDRa
Human TPM 2-DE/MS 823 <0.001
BU-LC/MS 7824
F IGURE 3 Frequency distribution of the per-gene number of Human PMI 2-DE/MS 808 <0.001
transcript variants (NTV) in porcine testicular proteins. (A) The
query set contained 230 genes matched by ShinyGO version 0.61, out BU-LC/MS 7358
of 246 ones coding for 2-DE/MS-determined proteins. Distributions Note. BU-LC/MS, bottom-up proteomics based on liquid chromatography cou-
of NTV differed between query set (red bars) and the genome-wide pled to MS, 2-DE/MS, two-dimensional gel electrophoresis followed by MS;
expectation (grey bars). (B) The shift toward higher N values was PMI, post-translational modification index; TPM, transcripts per million.TV aFalse discovery rates as derived fromMWU tests comparing parameter levels
less pronounced in a query set including 1396 genes that coded for between 2-DE/MS and BU-LC/MS datasets.
porcine testicular proteins from a BU-LC/MS reference study [39].
The results shown here are representative of analyses carried out on
altogether five study pairs (Supporting information Figure S12) is, the number of PTMs per amino acid (FDR < 0.001,
MWU test; Table 5; Supporting information Table S4).
With a mean PMI (PMI) of 0.092, every 11th amino acid
Subsequent analyses, which focused on the aforemen- of a 2-DE/MS protein was a PTM candidate. In contrast,
tioned human protein samples due to better data availabil- every 21st amino acid in BU-LC/MS proteins might carry
ity, underscored that amino acid modifications might pro- a PTM (PMI = 0.049). In addition, transcripts had an
mote multispotting in 2-DE. In particular, pooled human overall higher abundance (TPM) when encoding 2-DE/MS
proteins from the four 2-DE/MS studies overall exceeded rather than BU-LC/MS proteins (FDR < 0.001, MWU test;
their BU-LC/MS counterparts in the level of PMI, that Table 5). On average, transcript abundance was 67% higher
 15222683, 2022, 11, Downloaded from https://analyticalsciencejournals.onlinelibrary.wiley.com/doi/10.1002/elps.202000393 by Cochrane Germany, Wiley Online Library on [06/02/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1210 KWIATKOWSKI et al.
F IGURE 4 Representation of molecular functions in porcine (A) and human testicular proteins (B) as determined by 2-DE/MS (left)
and BU-LC/MS. Fewer gene ontology (GO) terms were obtained for 2-DE/MS than BU-LC/MS samples. The emergence of additional GOs in
BU-LC/MS datasets (molecular adaptor activity and molecular transducer activity) was accompanied by smaller shares of GOs relating to
binding and catalytic activity, when compared to 2-DE/MS datasets. The pattern described was reproduced in altogether five study pairs (see
Supporting information Figure S13)
for genes of human 2-DE/MS proteins relative to genes recurring GOs in porcine and human study sets. These
underlying human BU-LC/MS proteins (Supporting infor- were GOs relating to binding, catalytic activity, molecular
mation Table S5). function regulation, structural molecule activity, and
The findings outlined above were paralleled by trends transporter activity (Figure 4, Supporting information
in partial rank correlation analyses of pooled values from Figure S13). Involvements in binding and catalytic activ-
2-DE/MS and BU-LC/MS studies (FDR < 0.001, each; ity accounted for at least about 75% of the GO terms.
Table 2). In particular, TPM significantly correlated with Beyond these similarities, there were clear differences
NTV and PMI in 5365 genes/proteins, for which all three between 2-DE/MS and BU-LC/MS datasets. Specifically,
parameters were available. Thus, the higher the tran- BU-LC/MS sets persistently had more GOs than the
script abundance was, the more transcript variants a gene corresponding 2-DE/MS sets. These additional GOs in
tended to have (rho = 0.133). In addition, higher tran- BU-LC/MS samples were molecular transducer activity
script amounts associated with elevated PMI values (rho= and molecular adaptor activity. We take the results of the
0.363). Last, NTV and PMI were correlated, although with GO analyses as post-hoc affirmation that the 2-DE/MS and
a negligible coefficient (rho = –0.055). BU-LC/MS studies were each sufficiently homogeneous to
Complementary functional characterization (PAN- conduct the comparisons between both groups as outlined
THER: GO-Slim Molecular Function) revealed five above.
 15222683, 2022, 11, Downloaded from https://analyticalsciencejournals.onlinelibrary.wiley.com/doi/10.1002/elps.202000393 by Cochrane Germany, Wiley Online Library on [06/02/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
KWIATKOWSKI et al. 1211
4 DISCUSSION age. The same applies to our proxy for assessing the level
of protein speciation due to AS, that is, NTV.
The usefulness of 2-DE for separating proteins [1, 2] and In further support of a prominent role of amino acid
protein species [10–13, 56, 57] is well established. In the modifications inmultiple spotting, we noticed that human
current investigation (Figure 1; Supporting information proteins from 2-DE/MS are more likely to occur in mul-
Figures S1–S11), 2-DE/MS enabled the detection of 248 pro- tiple PTM variants than their counterparts from BU-
tein species representing 246 proteins (Supporting infor- LC/MS analyses (PMI in Table 2; Supporting information
mation Table S1), of which 22 were previously discovered Table S4). Thus, every 11th amino acid site was found to
in porcine testis using 2-DE/MS [51, 52]. More importantly, be a PTM candidate in human 2-DE/MS proteins on aver-
the present findings add to previous hints that, depending age, while it was every 21st site in BU-LC/MS-determined
on the detailed protocol applied, 2-DE/MSmight affect the ones. Compared to this, the contribution of AS to mul-
composition of protein samples [3–5]. Thus, present results tispotting is probably small. Indeed, many proteins have
suggest that shotgun 2-DE/MS could preferentially deter- only a single dominant isoform despite the occurrence of
mine multispotting proteins. Consistent with such a possi- additional transcript variants [31, 33]. This is matched in
bility, a considerable percentage of proteins from porcine the present study by only a slight increase in the NTV (per
testis were identified from more than a single 2-DE spot coding gene) in the 2-DE/MS samples compared to the BU-
(Supporting information Table S1). The exact percentage LC/MS samples (Figure 3; Table 3; Supporting informa-
of such multiple determinations (Figure 2; 46%) was in the tion Figure S12; Supporting information Table S3). Nev-
range of previous 2-DE/MS-based investigations, including ertheless, the increase was traceable (plus ∼0.3) and our
those on glioblastoma (32%) [10],HeLa cells (50%) [58], and transcript-based proxy giving resolution in electrophore-
Mycobacterium tuberculosis (55%) [8]. In this respect, the sis was also higher in the 2-DE/MS than BU-LC/MS stud-
present study could be representative of a general trend. ies (Table 4). Thus, elevated protein speciation parameters
As mentioned before, protein speciation through amino probably reflect an overrepresentation of protein species-
acid modification and AS can cause multispotting in 2- rich proteins in 2-DE/MS and not their underrepresenta-
DE [56]. Consistently, we found the number of spots from tion in BU-LC/MS.
which a porcine testicular protein was recovered to be In addition to the occurrence in variants, present
more closely associated with a peptide diversity measure findings (Table 5) suggest that higher protein abundance,
which accounted for alternative amino acid modification as approximated by TPM, can raise the determination
patterns than with a measure which did not (Table 2). It probability of a protein in 2-DE/MS [4]. However, such an
should be noted though that, with a maximum of 93%, the effect is not an inherent property of 2-DE/MS but reflects
detected peptides only incompletely covered the protein modest sensitivity of a particular pipeline rather. Indeed,
species determined in the current 2-DE/MS. Potentially, the detection limits of the staining protocols applied in the
this reflects their overall larger size compared to protein 2-DE/MS studies included here were in the low nanogram
species for which 100% sequence coverage was achieved range ([40, 42, 44, 46], present study). Compared to this,
elsewhere. In fact, complete coverage was attained in pre- the sensitivity of MS instruments—which is the limiting
vious 2-DE/MS studies for protein species of up to about factor in BU-LC/MS—is an order of magnitude higher. For
14 kDa [8, 59], whereas such a MW marked the lower example, ESI-QTOF spectrometers may detect peptides
boundary in the present examination of the porcine tes- down to 10–100 fmol and modern Orbitrap MS instru-
ticular proteome. Yet, almost all protein species deter- ments can trace peptides in concentrations of 1–10 amol
mined by us had larger MWs of up to about 115 kDa [11, 61]. It is meanwhile even possible to determine more
(Supporting information Table S1). In addition, the reso- than 400 proteins from a single 2-DE spot with modern
lution of 2-DE was higher in the reference studies [8, 59] MS instruments [10–13]. This is far from what has been
than in the present one (Figure 1; Supporting informa- reached in the 2-DE/MS analyses considered here ([40, 42,
tion Figures S1–11). Furthermore, a combination of cleav- 44, 46], Supporting information Table S1). Probably for this
age reagents is considered to be favorable to attain 100% reason, fewer GO terms were represented in protein sets
sequence coverage [60] while the porcine testis proteome from 2-DE/MS than from BU-LC/MS pipelines included
was merely trypsin-digested in the present investigation. in the current meta-analysis (Figure 4; Supporting infor-
Either way, we do not expect that complete sequence cov- mation Figure S13). In fact, the additional GOs in the
erage would change the results of the present correlation BU-LC/MS studies referred to molecular adapter activity
analyses. This is because the current proxies for address- and molecular transducer activity and, thus, to processes
ing the issue do not depend on complete sequence cover- which usually are exerted by low-abundance proteins.
 15222683, 2022, 11, Downloaded from https://analyticalsciencejournals.onlinelibrary.wiley.com/doi/10.1002/elps.202000393 by Cochrane Germany, Wiley Online Library on [06/02/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1212 KWIATKOWSKI et al.
5 CONCLUDING REMARKS CONFL ICT OF INTEREST
The authors have declared no conflict of interest.
Present findings suggest that random or arbitrary selec-
tion of 2-DE spots can result in an overrepresentation DATA AVAILAB IL ITY STATEMENT
of multispotting proteins in downstream MS analysis. New MS data on the porcine testicular proteome have
The deeper cause for this is probably protein speciation been deposited to the ProteomeXchange Consortium via
due to alternative amino acid modification patterns and, the PRIDE partner repository with record ID PXD015649.
to a lesser extent, AS. Although demonstrated here for Identifiers and symbols of the corresponding proteins and
the pig and human, we expect the effect to be valid genes are listed in Supporting information Tables S1 and
beyond the taxonomic group they represent, Laurasiathe- S2. Supporting information Table S2 additionally contains
ria [37, 38]. This is because PTM and AS have proba- ID lists for the other nine protein sets analyzed, all of
bly evolved, at least in their basic characteristics, prior which were taken from Tables in previous publications
to the divergence of Eukaryota [26, 62–67]. The occur- [39–47]. Transcript IDs are reported in Supporting infor-
rence of protein speciation in other taxa, such as Heli- mation Table S3.
cobacter pylori, could support this view [3]. However, the
explanatory approach taken here should not depend on ORCID
whether multispotting is due to naturally occurring pro- HolgerHerlyn https://orcid.org/0000-0003-3157-7243
tein speciation which has evolved once in a common
stem line or severalfold independently in the tree of life. REFERENCES
Rather, artificial amino acid modifications, which can 1. Klose J. Protein mapping by combined isoelectric focusing and
occur during sample preparation, may also contribute to electrophoresis of mouse tissues. A novel approach to test-
multispotting. We have accounted for both possibilities in ing for induced point mutations in mammals. Humangenetik.
present analysis of mass spectra by including natural (e.g., 1975;26:231–43.
STY phosphorylation) and artificial modifications (e.g., 2. O’Farrell PH. High resolution two-dimensional electrophoresisof proteins. J Biol Chem. 1975;250:4007–21.
M oxidation). 3. Jungblut PR, Schiele F, Zimny-Arndt U, Ackermann R, Schmid
Furthermore, we do not claim that preferential deter- M, Lange S, et al. Helicobacter pylori proteomics by 2-
mination of multispotting proteins is reflected in every DE/MS, 1-DE-LC/MS and functional data mining. Proteomics.
2-DE/MS study. In particular, the combination of high- 2010;10:182–93.
resolution 2-DE with high-sensitivity LC-MS/MS raises 4. Gygi SP, Corthals GL, Zhang Y, Rochon Y, Aebersold R. Eval-
the prospect that proteomes will be detected entirely, or uation of two-dimensional gel electrophoresis-based proteome
at least almost completely [10–13, 58]. This should coun- analysis technology. Proc Natl Acad Sci USA. 2000;97:9390–5.
5. Klont F, Bras L, Wolters JC, Ongay S, Bischoff R, Halmos
teract potential biases in the composition of 2-DE/MS- GB, et al. Assessment of sample preparation bias in mass
determined protein samples. Still, awareness that shot- spectrometry-based proteomics. Anal Chem. 2018;90:5405–13.
gun 2-DE/MS using less sensitive staining protocols and 6. Schmidt F, Donahoe S, Hagens K, Mattow J, Schaible UE,
MS instruments can lead to an overrepresentation of pro- Kaufmann SH, et al. Complementary analysis of the Mycobac-
tein species-rich proteins might prove useful in the future. terium tuberculosis proteome by two-dimensional electrophore-
Future investigations might also address whether high- sis and isotope-coded affinity tag technology. Mol Cell Pro-
abundance proteins in general tend to be rich in protein teomics. 2004;3:24–42.
species, much as the present results suggest (Table 2). 7. Jungblut P, Thiede B, Zimny-Arndt U, Müller EC, SchelerC, Wittmann-Liebold B, et al. Resolution power of two-
Finally, the proportion of transcript variant-rich genes dimensional electrophoresis and identification of proteins from
could turn out to be a useful measure for assessing the gels. Electrophoresis. 1996;17:839–47.
degree of resolution reached. It should be advantageous 8. Lange S, Rosenkrands I, Stein R, Andersen P, Kaufmann SH,
here that 100% sequence coverage in MS analyses is not Jungblut PR. Analysis of protein species differentiation among
a prerequisite for using this measure and that representa- mycobacterial low-Mr-secreted proteins by narrow pH range
tive data on transcript variants are publicly available for Immobiline gel 2-DE-MALDI-MS. J Proteomics. 2014;97:235–44.
numerous biological species. 9. Deng X, Hahne T, Schröder S, Redweik S, Nebija D, Schmidt H,
et al. The challenge to quantify proteins with charge trains due
to isoforms or conformers. Electrophoresis. 2012;33:263–9.
ACKNOWLEDGEMENTS 10. Zhan X, Yang H, Peng F, Li J, Mu Y, Long Y, et al. How many
Funding was provided by University of Mainz (Stufe-1 to proteins can be identified in a 2DE gel spot within an analysis
of a complex human cancer tissue proteome? Electrophoresis.
HH). We thank two anonymous reviewers and the editor 2018;39:965–80.
handling the review process for constructive comments 11. Zhan X, Li N, Zhan X, Qian S. Revival of 2DE-LC/MS in pro-
and useful suggestions. We are further grateful to Mario teomics and its potential for large-scale study of human proteo-
Dejung (IMB, Mainz) for technical assistance. forms. Med One. 2018;3:e180008.
 15222683, 2022, 11, Downloaded from https://analyticalsciencejournals.onlinelibrary.wiley.com/doi/10.1002/elps.202000393 by Cochrane Germany, Wiley Online Library on [06/02/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
KWIATKOWSKI et al. 1213
12. Zhan X, Li B, Zhan X, Schlüter H, Jungblut PR. Innovating the 32. Irimia M, Roy SW. Origin of spliceosomal introns and alterna-
concept and practice of two-dimensional gel electrophoresis in tive splicing. Cold Spring Harb Perspect Biol. 2014;6, a016071.
the analysis of proteomes at the proteoform level. Coorssen J R 33. Ezkurdia I, Rodriguez JM, Carrillo-de Santa Pau E, Vázquez
Proteomes. 2019;7:36. J, Valencia A, Tress ML. Most highly expressed protein-
13. Naryzhny S, Klopov N, Ronzhina N, Zorina E, Zgoda V, Kleyst coding genes have a single dominant isoform. J Proteome Res.
O, et al. A database for inventory of proteoform profiles: "2DE- 2015;14:1880–7.
pattern". Electrophoresis. 2020;41:1118–24. 34. KerenH, Lev-Maor G, Ast G. Alternative splicing and evolution:
14. Aebersold R, Agar JN, Amster IJ, Baker MS, Bertozzi CR, Boja diversification, exon definition and function. Nat Rev Genet.
ES, et al. How many human proteoforms are there? Nat Chem 2010;11:345–55.
Biol. 2018;14:206–14. 35. Mudge JM, Frankish A, Fernandez-Banet J, Alioto T, Derrien
15. Lim YP. Mining the tumor phosphoproteome for cancer mark- T, Howald C, et al. The origins, evolution, and functional
ers. Clin Cancer Res. 2005;11:3163–9. potential of alternative splicing in vertebrates. Mol Biol Evol.
16. Raggiaschi R, Gotta S, Terstappen GC. Phosphoproteome analy- 2011;28:2949–59.
sis. Biosci Rep. 2005;25:33–44. 36. TressML, Abascal F, Valencia A. Alternative splicingmay not be
17. Sarioglu H, Lottspeich F,Walk T, Jung G, Eckerskorn C. Deami- the key to proteome complexity. Trends BiochemSci. 2017;42:98–
dation as a widespread phenomenon in two-dimensional poly- 110.
acrylamide gel electrophoresis of human blood plasma proteins. 37. Murphy WJ, Eizirik E, Johnson WE, Zhang YP, Ryder OA,
Electrophoresis. 2000;21:2209–18. O’Brien SJ. Molecular phylogenetics and the origins of placental
18. Kleinert P, Kuster T, Arnold D, Jaeken J, HeizmannCW, Troxler mammals. Nature. 2001;409:614–8.
H. Effect of glycosylation on the protein pattern in 2-D-gel elec- 38. Stuart GW, Moffett K, Leader JJ. A Comprehensive vertebrate
trophoresis. Proteomics. 2007;7:15–22. phylogeny using vector representations of protein sequences
19. Barrabés S, Sarrats A, Fort E, De Llorens R, Rudd PM, from whole genomes. Mol Biol Evol. 2002;19:554–62.
Peracaula R. Effect of sialic acid content on glycoprotein pI 39. Ma R, Zhang Y, Liu H, Ning P. Proteome profile of swine tes-
analyzed by two-dimensional electrophoresis. Electrophoresis. ticular cells infected with porcine transmissible gastroenteritis
2010;31:2903–12. coronavirus. PLoS One. 2014;9:e110647.
20. ZhouH, Liu Y, Chui J, Guo K, ShunQ, LuW, et al. Investigation 40. Guo X, Zhao C, Wang F, Zhu Y, Cui Y, Zhou Z, et al. Investiga-
on glycosylation patterns of proteins from human liver cancer tion of human testis protein heterogeneity using 2-dimensional
cell lines based on the multiplexed proteomics technology. Arch electrophoresis. J Androl. 2010;31:419–29.
Biochem Biophys. 2007;459:70–8. 41. Zhang Y, Li Q,Wu F, Zhou R, Qi Y, SuN, et al. Tissue-based pro-
21. Venables JP. Aberrant and alternative splicing in cancer. Cancer teogenomics reveals that human testis endows plentiful missing
Res. 2004;64:7647–54. proteins. J Proteome Res. 2015;14:3583–94.
22. Tazi J, Bakkour N, Stamm S. Alternative splicing and disease. 42. McManus CA, Polden J, Cotter DR, Dunn M. Two-dimensional
Biochim Biophys Acta. 2009;1792:14–26. reference map for the basic proteome of the human dorsolat-
23. Ward AJ, Cooper TA. The pathobiology of splicing. J Pathol. eral prefrontal cortex (dlPFC) of the prefrontal lobe region of the
2010;220:152–163. brain. J Proteomics. 2010;10:2551–5.
24. Scotti MM, Swanson MS. RNA mis-splicing in disease. Nat Rev 43. Cabello-Arreola A, Ho AM, Ozerdem A, Cuellar-Barboza AB,
Genet. 2016;17:19–32. Kucuker MU, Heppelmann CJ, et al. Differential dorsolateral
25. Bush SJ, Chen L, Tovar-Corona JM, Urrutia AO. Alternative prefrontal cortex proteomic profiles of suicide victims with
splicing and the evolution of phenotypic novelty. Philos Trans mood disorders. Genes. 2020;11:256.
R Soc London B Biol Sci. 2017;372:20150474. 44. Westbrook JA, Wheeler JX, Wait R, Welson SY, Dunn MJ. The
26. Chen SY, Li C, Jia X, Lai SJ. Sequence and evolutionary features human heart proteome: Two-dimensional maps using narrow-
for the alternatively spliced exons of eukaryotic genes. Int J Mol range immobilised pH gradients. Electrophoresis. 2006;27:1547–
Sci. 2019;20:3834. 55.
27. Schreiber K, Csaba G, HaslbeckM, Zimmer R. Alternative splic- 45. Doll S, Dreßen M, Geyer PE, Itzhak DN, Braun C, Doppler SA,
ing in next generation sequencing data of Saccharomyces cere- et al. Region and cell-type resolved quantitative proteomic map
visiae. PLoS One. 2015;10, e0140487. of the human heart. Nat Commun. 2017;8:1469.
28. Grau-Bové X, Ruiz-Trillo I, Irimia M. Origin of exon skipping- 46. Yoshida Y, Miyazaki K, Kamiie J, Sato M, Okuizumi S,
rich transcriptomes in animals driven by evolution of gene archi- Kenmochi A, et al. Two-dimensional electrophoretic pro-
tecture. Genome Biol. 2018;19:135. filing of normal human kidney glomerulus proteome and
29. Pan Q, Shai O, Lee LJ, Frey BJ, Blencowe BJ. Deep surveying of construction of an extensible markup language (XML)-based
alternative splicing complexity in the human transcriptome by database. Proteomics. 2005;5:1083–96.
high-throughput sequencing. Nat Genet. 2008;40:1413–5. 47. Sigdel TK, Piehowski PD, Roy S, Liberto J, Hansen JR, Swensen
30. Wang ET, Sandberg R, Luo S, Khrebtukova I, Zhang L, Mayr C, AC, et al. Near-single-cell proteomics profiling of the proximal
et al. Alternative isoform regulation in human tissue transcrip- tubular and glomerulus of the normal human kidney. Front
tomes. Nature. 2008;456:470–6. Med. 2020;7:499.
31. Gonzàlez-Porta M, Frankish A, Rung J, Harrow J, Brazma 48. Ramljak S, Asif AR, Armstrong VW, Wrede A, Groschup MH,
A. Transcriptome analysis of human tissues and cell lines Buschmann A, et al. Physiological role of the cellular prion pro-
reveals one dominant transcript per gene. Genome Biol. 2013;14, tein (PrP c): protein profiling study in two cell culture systems.
R70. J Proteome Res. 2008;7:2681–95.
 15222683, 2022, 11, Downloaded from https://analyticalsciencejournals.onlinelibrary.wiley.com/doi/10.1002/elps.202000393 by Cochrane Germany, Wiley Online Library on [06/02/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1214 KWIATKOWSKI et al.
49. Trusch M, Tillack K, Kwiatkowski M, Bertsch A, Ahrends R, 59. Okkels LM, Müller EC, Schmid M, Rosenkrands I, Kaufmann
Kohlbacher O, et al. Displacement chromatography as first SH, Andersen P, et al. CFP10 discriminates between nonacety-
separating step in online two-dimensional liquid chromatog- lated and acetylated ESAT-6 of Mycobacterium tuberculosis by
raphy coupled to mass spectrometry analysis of a complex differential interaction. Proteomics. 2004;4:2954–60.
protein sample–the proteome of neutrophils. J Chromatogr A. 60. Meyer B, Papasotiriou DG, KarasM. 100% protein sequence cov-
2012;1232:288–94. erage: amodern form of surrealism in proteomics. AminoAcids.
50. Kwiatkowski M, Krösser D, Wurlitzer M, Steffen P, Barcaru A, 2011;41:291–310.
Krisp C, et al. Application of displacement chromatography to 61. Miller I, Crawford J, Gianazza E. Protein stains for proteomic
online two-dimensional liquid chromatography coupled to tan- applications: which, when, why? Proteomics. 2006;6:5385–5408.
dem mass spectrometry improves peptide separation efficiency 62. Irimia M, Rukov JL, Penny D, Roy SW. Functional and evolu-
and detectability for the analysis of complex proteomes. Anal tionary analysis of alternatively spliced genes is consistent with
Chem. 2018;90:9951–58. an early eukaryotic origin of alternative splicing. BMCEvol Biol.
51. Huang S-Y, Lin J-H, Chen Y-H, Chuang C-K, Lin E-C, Huang 2007;4:188.
M-C, et al. A reference map and identification of porcine testis 63. Nilsen TW, Graveley BR. Expansion of the eukaryotic proteome
proteins using 2-DE and MS. Proteomics. 2005;5:4205–12. by alternative splicing. Nature. 2010;463:457–63.
52. Huang SY, Lin JH, Teng SH, Sun HS, Chen YH, Chen HH, et al. 64. Roy SW, Irimia M. Splicing in the eukaryotic ancestor: form,
Differential expression of porcine testis proteins during postna- function and dysfunction. Trends Ecol Evol. 2009;24:447–55.
tal development. Anim Reprod Sci. 2011;123:221–33. 65. Rogozin IB, Carmel L, Csuros M, Koonin EV. Origin and evolu-
53. Ge SX, JungD, YaoR. ShinyGO: a graphical gene-set enrichment tion of spliceosomal introns. Biol Direct. 2012;7:11.
tool for animals and plants. Bioinformatics 2020;36:2628–29. 66. Beltrao P, Bork P, Krogan NJ, van Noort V. Evolution and func-
54. Matlock MK, Holehouse AS, Naegle KM. ProteomeScout: a tional cross-talk of protein post-translational modifications. Mol
repository and analysis resource for post-translational modifica- Syst Biol. 2013;9:714.
tions and proteins. Nucleic Acids Res. 2015;43:D521–30. 67. Yeo GW, Van Nostrand E, Holste D, Poggio T, Burge CB. Identi-
55. Benjamini Y, Hochberg Y. Controlling the false discovery rate: fication and analysis of alternative splicing events conserved in
a practical and powerful approach to multiple testing. J R Stat human and mouse. Proc Natl Acad Sci USA. 2005;102:2850–5.
Soc. 1995;57:289–300.
56. Aslebagh R, Channaveerappa D, Arcaro KF, Darie CC. Compar-
ative two-dimensional polyacrylamide gel electrophoresis (2D- SUPPORT ING INFORMATION
PAGE) of humanmilk to identify dysregulated proteins in breast Additional supporting information may be found in the
cancer. Electrophoresis. 2018;39:1723–34. online version of the article at the publisher’s website.
57. Hentschker C, Dewald C, Otto A, Büttner K, Hecker M,
Becher D. Global quantification of phosphoproteins combining
metabolic labeling and gel-based proteomics in B. pumilus. Elec-
trophoresis. 2018;39:334–43. How to cite this article: Kwiatkowski M, Hotze
58. Thiede B, Koehler CJ, Strozynski M, Treumann A, Stein R, M, Schumacher J, Asif AR, Pittol JMR, Brenig B,
Zimny-Arndt U, et al. High resolution quantitative proteomics et al. Protein speciation is likely to increase the
of HeLa cells protein species using stable isotope labeling with
amino acids in cell culture (SILAC), two-dimensional gel elec- chance of proteins to be determined in 2-DE/MS.
trophoresis (2DE), and nano-liquid chromatograpohy coupled Electrophoresis. 2022;43:1203–14.
to an LTQ-OrbitrapMass spectrometer. Mol Cell Proteomics. https://doi.org/10.1002/elps.202000393
2013;12:529–38.
 15222683, 2022, 11, Downloaded from https://analyticalsciencejournals.onlinelibrary.wiley.com/doi/10.1002/elps.202000393 by Cochrane Germany, Wiley Online Library on [06/02/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License