Detection of Proteome Diversity Resulted from Alternative Splicing is Limited by Trypsin Cleavage Specificity

Alternative splicing dramatically increases transcriptome complexity but its contribution to proteome diversity remains controversial. Exon-exon junction spanning peptides provide direct evidence for the translation of specific splice isoforms and are critical for delineating protein isoform complexity. Here we found that junction-spanning peptides are underrepresented in publicly available mass spectrometry-based shotgun proteomics data sets. Further analysis showed that evolutionarily conserved preferential nucleotide usage at exon boundaries increases the occurrence of lysine- and arginine-coding triplets at the end of exons. Because both lysine and arginine residues are cleavage sites of trypsin, the nearly exclusive use of trypsin as the protein digestion enzyme in shotgun proteomic analyses hinders the detection of junction-spanning peptides. To study the impact of enzyme selection on splice junction detectability, we performed in-silico digestion of the human proteome using six proteases. The six enzymes created a total of 161,125 detectable junctions, and only 1,029 were common across all enzyme digestions. Chymotrypsin digestion provided the largest number of detectable junctions. Our experimental results further showed that combination of a chymotrypsin-based human proteome analysis with a trypsin-based analysis increased detection of junction-spanning peptides by 37% over the trypsin-only analysis and identified over a thousand junctions that were undetectable in fully tryptic digests. Our study demonstrates that detection of proteome diversity resulted from alternative splicing is limited by trypsin cleavage specificity, and that complementary digestion schemes will be essential to comprehensively analyze the translation of alternative splicing isoforms. Alternative splicing dramatically increases transcriptome complexity but its contribution to proteome diversity remains controversial. Exon-exon junction spanning peptides provide direct evidence for the translation of specific splice isoforms and are critical for delineating protein isoform complexity. Here we found that junction-spanning peptides are underrepresented in publicly available mass spectrometry-based shotgun proteomics data sets. Further analysis showed that evolutionarily conserved preferential nucleotide usage at exon boundaries increases the occurrence of lysine- and arginine-coding triplets at the end of exons. Because both lysine and arginine residues are cleavage sites of trypsin, the nearly exclusive use of trypsin as the protein digestion enzyme in shotgun proteomic analyses hinders the detection of junction-spanning peptides. To study the impact of enzyme selection on splice junction detectability, we performed in-silico digestion of the human proteome using six proteases. The six enzymes created a total of 161,125 detectable junctions, and only 1,029 were common across all enzyme digestions. Chymotrypsin digestion provided the largest number of detectable junctions. Our experimental results further showed that combination of a chymotrypsin-based human proteome analysis with a trypsin-based analysis increased detection of junction-spanning peptides by 37% over the trypsin-only analysis and identified over a thousand junctions that were undetectable in fully tryptic digests. Our study demonstrates that detection of proteome diversity resulted from alternative splicing is limited by trypsin cleavage specificity, and that complementary digestion schemes will be essential to comprehensively analyze the translation of alternative splicing isoforms. RNA sequencing (RNA-Seq) studies have highlighted a key role of alternative splicing in increasing transcriptome complexity. It has been shown that 92–94% of human genes undergo alternative splicing, and about 86% have a minor isoform frequency of 15% or more (1.Wang E.T. Sandberg R. Luo S. Khrebtukova I. Zhang L. Mayr C. Kingsmore S.F. Schroth G.P. Burge C.B. Alternative isoform regulation in human tissue transcriptomes.Nature. 2008; 456: 470-476Crossref PubMed Scopus (3567) Google Scholar). However, the contribution of alternative splicing to proteomic complexity remains controversial (2.Tress M.L. Abascal F. Valencia A. Most Alternative Isoforms Are Not Functionally Important.Trends Biochem. Sci. 2017; 42: 408-410Abstract Full Text Full Text PDF PubMed Scopus (41) Google Scholar, 3.Tress M.L. Abascal F. Valencia A. Alternative Splicing May Not Be the Key to Proteome Complexity.Trends Biochem. Sci. 2017; 42: 98-110Abstract Full Text Full Text PDF PubMed Scopus (174) Google Scholar, 4.Blencowe B.J. The Relationship between Alternative Splicing and Proteomic Complexity.Trends Biochem. Sci. 2017; 42: 407-408Abstract Full Text Full Text PDF PubMed Scopus (82) Google Scholar). Although some translatome and proteome profiling studies suggest that alternative splicing contributes significantly to proteomic diversity (5.Weatheritt R.J. Sterne-Weiler T. Blencowe B.J. The ribosome-engaged landscape of alternative splicing.Nat. Struct. Mol. Biol. 2016; 23: 1117-1123Crossref PubMed Scopus (84) Google Scholar, 6.Liu Y. Gonzalez-Porta M. Santos S. Brazma A. Marioni J.C. Aebersold R. Venkitaraman A.R. Wickramasinghe V.O. Impact of Alternative Splicing on the Human Proteome.Cell Rep. 2017; 20: 1229-1241Abstract Full Text Full Text PDF PubMed Scopus (87) Google Scholar), a systematic analysis of over 100 published mass spectrometry (MS)-based shotgun proteomics data sets showed that most protein-coding genes have a single dominant isoform irrespective of tissue or cell type (7.Ezkurdia I. Rodriguez J.M. Carrillo-de Santa Pau E. Vazquez J. Valencia A. Tress M.L. Most highly expressed protein-coding genes have a single dominant isoform.J. Proteome Res. 2015; 14: 1880-1887Crossref PubMed Scopus (66) Google Scholar). In shotgun proteomics experiments, proteins are enzymatically digested into peptides, which are subsequently fractionated and analyzed by liquid chromatography (LC)-MS/MS. The most commonly used protease in proteomics is trypsin, which cleaves at the C terminus of lysine and arginine with high efficiency and specificity (8.Olsen J.V. Ong S.E. Mann M. Trypsin cleaves exclusively C-terminal to arginine and lysine residues.Mol. Cell. Proteomics. 2004; 3: 608-614Abstract Full Text Full Text PDF PubMed Scopus (869) Google Scholar) and produces peptides of optimal length and charge characteristics for tandem MS sequencing. A major challenge in using LC-MS/MS data for the confirmation of splice isoforms is their limited sequence coverage. Exon-exon junction spanning peptides provide direct evidence for the translation of splice isoforms and thus delineate protein isoform complexity and improve gene and isoform annotation (9.Tanner S. Shen Z. Ng J. Florea L. Guigo R. Briggs S.P. Bafna V. Improving gene annotation using peptide mass spectrometry.Genome Res. 2007; 17: 231-239Crossref PubMed Scopus (152) Google Scholar, 10.Brosch M. Saunders G.I. Frankish A. Collins M.O. Yu L. Wright J. Verstraten R. Adams D.J. Harrow J. Choudhary J.S. Hubbard T. Shotgun proteomics aids discovery of novel protein-coding genes, alternative splicing, and “resurrected” pseudogenes in the mouse genome.Genome Res. 2011; 21: 756-767Crossref PubMed Scopus (99) Google Scholar, 11.Castellana N. Bafna V. Proteogenomics to discover the full coding content of genomes: a computational perspective.J. Proteomics. 2010; 73: 2124-2135Crossref PubMed Scopus (132) Google Scholar). However, the ability to detect junction-spanning peptides in shotgun proteomics experiments is currently unknown. We studied the proteomic coverage of exon-exon junctions in three publicly available proteomics data sets and found that trypsin preferentially cleaves exon-exon junctions and thus hinders the detection of junction-spanning peptides. This phenomenon was explained by evolutionarily conserved preferential nucleotide usage at exon boundaries according to nucleotide sequence analysis of five eukaryotic genomes. Our in silico and experimental analyses showed that complementary digestion schemes are essential to study the translation of alternative mRNA splicing to proteome diversity. Three publicly available label-free shotgun proteomics data sets were used in this study, including the CPTAC_CRC data set (12.Zhang B. Wang J. Wang X. Zhu J. Liu Q. Shi Z. Chambers M.C. Zimmerman L.J. Shaddox K.F. Kim S. Davies S.R. Wang S. Wang P. Kinsinger C.R. Rivers R.C. Rodriguez H. Townsend R.R. Ellis M.J. Carr S.A. Tabb D.L. Coffey R.J. Slebos R.J. Liebler D.C. Nci C. Proteogenomic characterization of human colon and rectal cancer.Nature. 2014; 513: 382-387Crossref PubMed Scopus (937) Google Scholar), the NCI-60 cell lines data set (13.Moghaddas Gholami A. Hahne H. Wu Z. Auer F.J. Meng C. Wilhelm M. Kuster B. Global proteome analysis of the NCI-60 cell line panel.Cell Rep. 2013; 4: 609-620Abstract Full Text Full Text PDF PubMed Scopus (222) Google Scholar), and label-free data from ProteomicsDB (14.Wilhelm M. Schlegl J. Hahne H. Gholami A.M. Lieberenz M. Savitski M.M. Ziegler E. Butzmann L. Gessulat S. Marx H. Mathieson T. Lemeer S. Schnatbaum K. Reimer U. Wenschuh H. Mollenhauer M. Slotta-Huspenina J. Boese J.H. Bantscheff M. Gerstmair A. Faerber F. Kuster B. Mass-spectrometry-based draft of the human proteome.Nature. 2014; 509: 582-587Crossref PubMed Scopus (1312) Google Scholar). These data sets were selected based on their high quality and large sample size. The CPTAC_CRC data set includes proteomics data from 95 colorectal cancer tumor samples, whereas the NCI-60 cell lines data set includes proteomics data from 59 cancer cell lines. We previously generated proBAM files for these data sets using the proBAMsuite (15.Wang X. Slebos R.J. Chambers M.C. Tabb D.L. Liebler D.C. Zhang B. proBAMsuite, a Bioinformatics Framework for Genome-Based Representation and Analysis of Proteomics Data.Mol. Cell. Proteomics. 2016; 15: 1164-1175Abstract Full Text Full Text PDF PubMed Scopus (23) Google Scholar). The proBAM files map peptide spectrum matches (PSMs) 1The abbreviations used are: PSM(s), peptide spectrum match(es);CDS, coding DNA sequence;ProBAM, proteomics BAM;IGV, integrative genomics viewer;CPTAC, Clinical Proteomic Tumor Analysis Consortium. from proteomics studies to the genome, making it possible to analyze PSMs in the context of the genome. ProteomicsDB is a comprehensive human proteome database constructed using data from 16,857 LC-MS/MS experiments involving human tissues, cell lines, and body fluids (14.Wilhelm M. Schlegl J. Hahne H. Gholami A.M. Lieberenz M. Savitski M.M. Ziegler E. Butzmann L. Gessulat S. Marx H. Mathieson T. Lemeer S. Schnatbaum K. Reimer U. Wenschuh H. Mollenhauer M. Slotta-Huspenina J. Boese J.H. Bantscheff M. Gerstmair A. Faerber F. Kuster B. Mass-spectrometry-based draft of the human proteome.Nature. 2014; 509: 582-587Crossref PubMed Scopus (1312) Google Scholar). We downloaded a total of 550,904 unique proteotypic peptide sequences identified from label-free experiments in the database. Based on the genome annotation described below, we mapped these peptide sequences to the human genome using proBAMsuite and then generated proBAM files for downstream analyses. We used the function PrepareAnnotationRefseq in the R package customProDB (16.Wang X. Zhang B. customProDB: an R package to generate customized protein databases from RNA-Seq data for proteomics search.Bioinformatics. 2013; 29: 3235-3237Crossref PubMed Scopus (109) Google Scholar) to prepare genome annotation, including protein sequences and the loci of exons for all protein coding transcripts of Human (hg19), mouse (mm10) and C. elegans (ce10) from the UCSC table browser. The genome annotations of S. cerevisiae and S. pombe were collected from the Ensembl database (ENSEMBL FUNGI 26) using the function PrepareAnnotationEnsembl in the same package. Our entire data set contained 38248, 29934, 27721, 6692, and 5144 coding transcripts for human, mouse, C. elegans, S. pombe, and S. cerevisiae, respectively. To analyze the relationship between locations of the observed peptide termini and exon boundaries, we the between peptide termini and exon The locations of identified peptides a proBAM were with locations of the exons to peptides at to splice sites exon peptides at to splice sites exon junction peptides to the end of the and junction peptides to the of the exons. and we the and coding exons for The was used to locations that significantly more peptides. To at splice junctions of protein coding genes, we used the between exon locations and protein We the in exon in the data including both residues by the triplets of exons and junction residues by triplets exon-exon junctions. we the with its usage across all protein sequences in genome data To of splice junctions between we the number of unique splice junctions for genome. We coding transcripts only and exons in We nucleotide sequences at both and splice sites on exon and on of splice junctions in coding or were of protein coding transcripts from genome the nucleotide at the R package was used to the sequence for sequences and downstream from the splicing human protein sequences from the annotation from UCSC table were digested with and into peptides using from M.C. B. R. D.L. S. L. B. B. J. K. N. N. B. C. L. K. C. B. F. J. P. S. T. A. J. B. T. K. J. M. Tabb D.L. P. A for mass spectrometry and PubMed Scopus Google Scholar). peptides with cleavage between to were cell were in with was using a and an of total protein was used for digestion with trypsin and respectively. peptide was using a 100 peptide were to in and in A was performed with an and with a was The of as A and and as B. was using the from to in from to in from to in and at for an were collected the and were into by was then analyzed by liquid mass spectrometry LC-MS/MS analyses were on a in were on a 100 with chromatography was at at a of using a of in and in A full was for peptides in the of by were using of previously analyzed for were generated by of the peptide at a of to generate a of and as major files are data set generated using a specific S. Chambers M.C. Zimmerman L.J. B. Tabb D.L. protein with high peptide Proteome Res. PubMed Scopus Google Scholar) was used to the peptide that files by S. a database for 2014; PubMed Scopus Google Scholar) and D.L. Chambers M.C. highly tandem mass peptide by Proteome Res. 2007; PubMed Scopus Google Scholar) human protein database protein of to by to and peptides with peptides, for in selection peptide sequence and to and to to the termini of peptides with A peptide length of was and a number of was the same of a of an of peptides with peptides, for in peptide sequence and to and to to the termini of peptides with A peptide length of and a peptide length of were for were generated from with a of and to proBAM files as described A peptide length of was for all downstream analyses. results were using the proteomics data available at and for all junction peptides are available in We studied the locations of all peptides identified in a shotgun proteomics analysis of 95 colorectal cancer published by the Clinical Proteomic Tumor Analysis (12.Zhang B. Wang J. Wang X. Zhu J. Liu Q. Shi Z. Chambers M.C. Zimmerman L.J. Shaddox K.F. Kim S. Davies S.R. Wang S. Wang P. Kinsinger C.R. Rivers R.C. Rodriguez H. Townsend R.R. Ellis M.J. Carr S.A. Tabb D.L. Coffey R.J. Slebos R.J. Liebler D.C. Nci C. Proteogenomic characterization of human colon and rectal cancer.Nature. 2014; 513: 382-387Crossref PubMed Scopus (937) Google Scholar). proBAMsuite, a for and analysis of proteomics we mapped of peptide sequences identified in the CPTAC_CRC data set to unique locations (15.Wang X. Slebos R.J. Chambers M.C. Tabb D.L. Liebler D.C. Zhang B. proBAMsuite, a Bioinformatics Framework for Genome-Based Representation and Analysis of Proteomics Data.Mol. Cell. Proteomics. 2016; 15: 1164-1175Abstract Full Text Full Text PDF PubMed Scopus (23) Google Scholar). the mapped peptides, were exon-exon junction peptides that provide direct evidence for RNA splicing (15.Wang X. Slebos R.J. Chambers M.C. Tabb D.L. Liebler D.C. Zhang B. proBAMsuite, a Bioinformatics Framework for Genome-Based Representation and Analysis of Proteomics Data.Mol. Cell. Proteomics. 2016; 15: 1164-1175Abstract Full Text Full Text PDF PubMed Scopus (23) Google Scholar). of the peptide sequences mapped exons or at the exon This is the of based on a peptide length of in this data set and a length of for exons analysis of shotgun proteomics data set of the NCI-60 cell lines (13.Moghaddas Gholami A. Hahne H. Wu Z. Auer F.J. Meng C. Wilhelm M. Kuster B. Global proteome analysis of the NCI-60 cell line panel.Cell Rep. 2013; 4: 609-620Abstract Full Text Full Text PDF PubMed Scopus (222) Google Scholar) showed that of the peptide sequences exons or at the exon To further this in we used a of 550,904 proteotypic peptide sequences from the LC-MS/MS human proteome ProteomicsDB (14.Wilhelm M. Schlegl J. Hahne H. Gholami A.M. Lieberenz M. Savitski M.M. Ziegler E. Butzmann L. Gessulat S. Marx H. Mathieson T. Lemeer S. Schnatbaum K. Reimer U. Wenschuh H. Mollenhauer M. Slotta-Huspenina J. Boese J.H. Bantscheff M. Gerstmair A. Faerber F. Kuster B. Mass-spectrometry-based draft of the human proteome.Nature. 2014; 509: 582-587Crossref PubMed Scopus (1312) Google Scholar). We mapped of these peptide sequences to human genome based on the the mapped peptide were exon with in the cancer data of the peptide sequences mapped exons or at the exon To further the relationship between locations of the observed peptide termini and exon boundaries, we the between peptide termini and exon boundaries in the CPTAC_CRC data the for all observed peptides that or of a splice A more analysis was performed for all exon-exon junction peptides we observed an of peptide termini to exon boundaries, most for peptides that at a splice downstream of a splice at a splice or downstream of a splicing We to these as type peptides, These of peptides for of the mapped peptides. A was observed in the NCI-60 cell lines data set and the proteotypic peptide sequences from ProteomicsDB To the observed of type peptides, we the of the residues by all exons of human coding including residues by the triplets of exons and junction residues by triplets exon-exon junctions. with the for all human coding five were more the and junction including lysine arginine and These five most for of all and junction with the of of the five lysine and in of the and junction Because trypsin cleaves peptides on the of lysine and their and junction residues the of type peptides in shotgun proteomics studies using trypsin for protein digestion Because the same splice junction be it was by we further the number of unique exon-exon junctions in the genome and the number of unique junctions with lysine or arginine at or exon-exon junction to as shown in junctions of all unique junctions in the human coding genome. is by nucleotide usage and exon To further the of junctions, we the nucleotide sequences on of the boundaries for all coding genes in the human genome. We observed an at the exon boundaries, to the at the for we that of all triplets involving the nucleotide of an exon for lysine or which is more the of The peptide described be generated in by the of the splice junction a produces of peptides In the trypsin digestion is at the splicing thus type and type peptides. In trypsin cleaves at downstream of the splicing thus type and type peptides. We the nucleotide at exon boundaries in eukaryotic and found a shown in the to more We found junction these for the cerevisiae which is of the of in most cerevisiae The results that in to human trypsin digestion to of type peptides in shotgun proteomics analyses of of type peptides by trypsin digestion the of splice junctions by In the CPTAC_CRC data mapped peptide sequences into type and thus be used to splice junctions. mapped peptide sequences were type peptides, which have limited in splice junction only a nucleotide on the exon We then performed in silico digestion to the impact of enzyme selection on splice junction in silico and peptides were for and respectively. peptides were in proBAM which were used to coding genome, and junction for enzyme and showed the coding DNA sequence to of all of the human number of detectable junctions, and junctions from in-silico digestion of six proteases. length was to be to The was based on peptides mapped to genome. peptides were in the study of by the single nucleotide coverage on the junctions junctions of R and of of of and of of table in a However, trypsin digestion resulted in coverage for junctions, and the junction coverage to with the proteome coverage We that the detectable junctions were of by or and coverage these enzymes and respectively. and the coverage for junction and junction were to for the coding DNA sequence A was found for the of detectable junctions and junctions between the three Chymotrypsin and shown in the enzyme digestion a unique of detectable junction peptides, and their combination a more comprehensive the six enzymes created a total of 161,125 detectable junctions, and only were common across all enzyme Chymotrypsin digestion provided the largest number of detectable junctions and junctions We a complementary digestion by shotgun proteomics experiments on cell using trypsin and respectively. In to trypsin, cleaves at the of and which are and exon-exon junction residues trypsin digestion to a number of peptide The number of identified peptides with trypsin digestion was peptide whereas digestion identified peptide However, with in silico digestion generated a of type peptides trypsin digestion with the and for trypsin and respectively. We further the of junctions by proteomic data in the we limited to the mapped junction peptides. peptides were in the study of by the single nucleotide coverage on the exon peptides with only nucleotide coverage on the exon were the number of was We found and mapped junction peptide sequences in the trypsin and the to and junctions, The of junctions were and for trypsin and of the junctions in the trypsin resulted from peptides or by lysine or the junctions that were in the experiments, but the trypsin have been with tryptic peptides of the of lysine or arginine at splicing sites experimental data the in silico and the preferential cleavage by trypsin at splicing a previously that of protein isoform complexity. We use three from the data to further shown in an exon in gene protein was by a peptide identified from the but data from trypsin digestion to provide evidence for this of the of lysine at the junction that junction the proteomic confirmation of isoforms resulted from alternative has three isoform an with isoform in a protein with a C the exon of isoform only it is to mass evidence for isoform the protein is by trypsin at the junction a peptide from digestion the usage of an alternative in gene whereas the peptide from trypsin digestion to provide evidence on alternative usage We the preferential cleavage at exon-exon junction sites by trypsin digestion and evolutionarily conserved preferential nucleotide usage at exon Although the nucleotide of the exon boundaries have been for some alternative splicing 2004; PubMed Scopus Google Scholar), studies have the on of the proteins S. R. splice junctions map at protein PubMed Scopus Google Scholar, L. H. Splicing and the of proteins in Biol. 2007; PubMed Scopus Google Scholar, B. J. junctions are by protein Res. 2015; PubMed Scopus Google Scholar). Because trypsin is the most commonly used enzyme in proteomics experiments, has in the of proteomics experiments, the is to the translation of alternative splicing to the proteome The contribution of alternative splicing to proteome diversity is an of and is evidence as to the diversity be to proteins (5.Weatheritt R.J. Sterne-Weiler T. Blencowe B.J. The ribosome-engaged landscape of alternative splicing.Nat. Struct. Mol. Biol. 2016; 23: 1117-1123Crossref PubMed Scopus (84) Google Scholar, 6.Liu Y. Gonzalez-Porta M. Santos S. Brazma A. Marioni J.C. Aebersold R. Venkitaraman A.R. Wickramasinghe V.O. Impact of Alternative Splicing on the Human Proteome.Cell Rep. 2017; 20: 1229-1241Abstract Full Text Full Text PDF PubMed Scopus (87) Google Scholar, I. Rodriguez J.M. Carrillo-de Santa Pau E. Vazquez J. Valencia A. Tress M.L. Most highly expressed protein-coding genes have a single dominant isoform.J. Proteome Res. 2015; 14: 1880-1887Crossref PubMed Scopus (66) Google Scholar). The nearly exclusive use of trypsin as the protein digestion enzyme in shotgun proteomic analyses hinders the detection of junction-spanning peptides and it to delineate protein isoform complexity. This the Trypsin digestion to the limited in novel splicing peptides in studies (16.Wang X. Zhang B. customProDB: an R package to generate customized protein databases from RNA-Seq data for proteomics search.Bioinformatics. 2013; 29: 3235-3237Crossref PubMed Scopus (109) Google Scholar, K. The of mass spectrometry-based proteomic data for of novel alternative splice from RNA-Seq a 2010; PubMed Scopus Google Scholar, and mass analysis of novel peptides using Cell. Proteomics. 2013; Full Text Full Text PDF PubMed Scopus Google Scholar). that digestion to fully tryptic peptides T. C. E. Aebersold R. A. of protein digestion cleavage efficiency of tandem over trypsin Proteome Res. PubMed Scopus Google Scholar) will further the of exon-exon junctions in trypsin digested an analysis of mouse exon-exon junctions that be by shotgun proteomics in trypsin digested showed that the detection from to peptides were M. Saunders G.I. Frankish A. Collins M.O. Yu L. Wright J. Verstraten R. Adams D.J. Harrow J. Choudhary J.S. Hubbard T. Shotgun proteomics aids discovery of novel protein-coding genes, alternative splicing, and “resurrected” pseudogenes in the mouse genome.Genome Res. 2011; 21: 756-767Crossref PubMed Scopus (99) Google Scholar). The use of alternative enzymes with specificity to trypsin, as will only generate peptide sequences that identified based on trypsin but will provide peptide evidence for junctions that are to in tryptic digests. digestion schemes will more comprehensive study of RNA splicing proteome in The shotgun proteomics data of have been in the proteomics and are at including the for all junction peptides, and with files