NMD inhibition by 5-azacytidine augments presentation of immunogenic frameshift-derived neoepitopes

•Immunopeptidomic neoepitope discovery workflow employing public DNA sequencing data•Identification of immunogenic frameshift-derived InDel neoepitopes in an MSI model•NMD inhibition increases HLA class I-mediated presentation neoepitopes•CKAP2 frameshift mutation is highly recurrent different types cancer Frameshifted protein sequences elicit tumor-specific T cell-mediated immune responses microsatellite-unstable (MSI) cancers if presented by I molecules. However, their expression and are limited nonsense-mediated RNA decay (NMD). We employed unbiased immunopeptidomics to analyze HCT-116 cells identified >10,000 I-presented peptides including five neoepitopes. Notably, pharmacological NMD with 5-azacytidine stabilizes frameshift-bearing transcripts the The underlying one colorectal cell lines primary patient samples, immunization corresponding induces strong CD8+ HLA-A?02:01 transgenic mouse model. Our data show directly that augments thus highlighting clinical potential anti-cancer immunotherapy strategies. Microsatellite-unstable (MSI CRCs) account for approximately 15% all CRCs, hence about 275,000 cases per year (Bray et al., 2018Bray F. Ferlay J. Soerjomataram I. Siegel R.L. Torre L.A. Jemal A. Global statistics 2018: GLOBOCAN estimates incidence mortality worldwide 36 185 countries.CA Cancer Clin. 2018; 68: 394-424https://doi.org/10.3322/caac.21492Crossref PubMed Scopus (45300) Google Scholar). These characterized accumulation somatic mutations, mainly small insertion/deletion (InDel) mutations repetitive stretches termed microsatellites (Cancer Genome Atlas, 2012Cancer Atlas N. Comprehensive molecular characterization human colon rectal cancer.Nature. 2012; 487: 330-337https://doi.org/10.1038/nature11252Crossref (5622) Importantly, two-thirds cause a shift reading frame, encoding (Kloor von Knebel Doeberitz, 2016Kloor M. Doeberitz biology cancer.Trends Cancer. 2016; 2: 121-133https://doi.org/10.1016/j.trecan.2016.02.004Abstract Full Text PDF (106) Human leukocyte antigen (HLA) (HLAp) derived from such frameshifted (FS) (termed neoepitopes) allow patrolling identify target tumor presenting (Schumacher Schreiber, 2015Schumacher T.N. Schreiber R.D. Neoantigens immunotherapy.Science. 2015; 348: 69-74https://doi.org/10.1126/science.aaa4971Crossref (2759) In contrast single amino acid (aa) change SNP neoepitopes), FS can potentially harbor several binding capacity allotypes. Moreover, have been suggested possess higher immunogenicity caused fundamental difference endogenous self-antigens originating wild-type (WT) proteins (Turajlic 2017Turajlic S. Litchfield K. Xu H. Rosenthal R. McGranahan Reading J.L. Wong Y.N.S. Rowan Kanu Al Bakir al.Insertion-and-deletion-derived tumour-specific neoantigens phenotype: pan-cancer analysis.Lancet Oncol. 2017; 18: 1009-1021https://doi.org/10.1016/s1470-2045(17)30516-8Abstract (0) Several studies support presence neoepitope-specific cytotoxic both healthy individuals patients CRC (Leoni 2020Leoni G. D'Alise A.M. Cotugno Langone Garzia De Lucia Fichera Vitale Bignone V. Tucci F.G. al.A genetic vaccine shared treat tumors microsatellite instability.Cancer Res. 2020; 80: 3972-3982https://doi.org/10.1158/0008-5472.CAN-20-1072Crossref (20) Scholar; Roudko 2020Roudko Bozkus C.C. Orfanelli T. McClain C.B. Carr C. O’Donnell Chakraborty L. Samstein Huang K.-l. Blank S.V. al.Shared poly-epitope unstable tumors.Cell. 183: 1634-1649.e17https://doi.org/10.1016/j.cell.2020.11.004Abstract (42) Schwitalle 2008Schwitalle Y. Kloor Eiermann Linnebacher Kienle P. Knaebel H.P. Tariverdian Benner Immune response against frameshift-induced neopeptides HNPCC carriers.Gastroenterology. 2008; 134: 988-997https://doi.org/10.1053/j.gastro.2008.01.015Abstract (261) Recently, we others showed induction neoepitope-directed after vaccination shared, silico-predicted mice Scholar) trial mismatch repair-deficient 2020Kloor Reuschenbach Pauligk Karbach Rafiyan M.-R. Al-Batran S.-E. Jäger E. A peptide neoantigen-based cancers: phase I/IIa trial.Clin. 26: 4503-4510https://doi.org/10.1158/1078-0432.Ccr-19-3517Crossref it was shown cells, re-activated checkpoint inhibition, further emphasizing crucial role immunotherapeutic strategies (Gubin 2014Gubin M.M. Zhang X. Schuster Caron Ward J.P. Noguchi Ivanova Hundal Arthur C.D. Krebber W.J. al.Checkpoint blockade targets mutant antigens.Nature. 2014; 515: 577-581https://doi.org/10.1038/nature13988Crossref (1362) these rely on sufficient quantities, most translation-dependent quality control pathway recognizes degrades mRNAs premature termination codons introduced nonsense or (Lavysh Neu-Yilik, 2020Lavysh D. Neu-Yilik UPF1-Mediated decay-danse macabre cloud.Biomolecules. 10: 999https://doi.org/10.3390/biom10070999Crossref (10) central factors UPF1, UPF2, SMG1, SMG6, SMG7 expressed substantially more strongly compared microsatellite-stable CRCs restricting production consequently recognition (Bokhari 2018Bokhari Jonchere Lagrange Bertrand Svrcek Marisa Buhard O. Greene Demidova Jia al.Targeting mRNA instability.Oncogenesis. 7: 70https://doi.org/10.1038/s41389-018-0079-xCrossref (32) El-Bchiri 2005El-Bchiri Penard-Lacronique Thomas Hamelin Duval Differential mediated mutated repair deficient cancers.Hum. Mol. Genet. 2005; 14: 2435-2442https://doi.org/10.1093/hmg/ddi245Crossref (46) Scholar, 2008El-Bchiri Guilloux Dartigues Loire Mercier Sobhani de la Grange Auboeuf Praz al.Nonsense-mediated impacts MSI-driven carcinogenesis anti-tumor immunity cancers.PLoS One. 3: e2583https://doi.org/10.1371/journal.pone.0002583Crossref (36) laboratory approved drug (5AZA) as potent inhibitor limits without interfering synthesis at therapeutic concentrations, distinguishing 5AZA other known inhibitors (Bhuvanagiri 2014Bhuvanagiri Lewis Putzker Becker Leicht Krijgsveld Batra Turnwald B. Jovanovic Hauer al.5-azacytidine inhibits MYC-dependent fashion.EMBO Med. 6: 1593-1609https://doi.org/10.15252/emmm.201404461Crossref (37) Therefore, hypothesized could increase host's system. Leveraging full generating effective novel possibly personalized requires reliable identification peptides. Neoepitopes commonly evaluating silico candidates using screening technologies. analyses high rates false-positive results (Andreatta Nielsen, 2016Andreatta Nielsen Gapped sequence alignment artificial neural networks: application MHC system.Bioinformatics. 32: 511-517https://doi.org/10.1093/bioinformatics/btv639Crossref (539) Bassani-Sternberg 2015Bassani-Sternberg Pletscher-Frankild Jensen L.J. Mann Mass spectrometry peptidomes reveals effects abundance turnover presentation.Mol. Cell Proteomics. 658-673https://doi.org/10.1074/mcp.M114.042812Abstract (257) Cui 2020Cui J.J. Wang L.Y. Tan Z.R. Zhou H.H. Zhan Yin J.Y. spectrometry-based therapy.Mass Spectrom. Rev. 39: 523-552https://doi.org/10.1002/mas.21620Crossref (18) Freudenmann 2018Freudenmann L.K. Marcu Stevanovic Mapping tumour ligandome mass spectrometry.Immunology. 154: 331-345https://doi.org/10.1111/imm.12936Crossref (63) Recent breakthroughs sensitivity reproducibility (MS) help overcome disadvantages enable exploration global immunopeptidome New methodological MS approaches dual-fragmentation electron-transfer/higher-energy collision dissociation (EThcD) expand detectable confidence identifications (Mommen 2014Mommen G.P. Frese C.K. Meiring H.D. van Gaans-van den Brink Jong A.P. Els C.A. Heck A.J. Expanding repertoire (EThcD).Proc. Natl. Acad. Sci. 111: 4507-4512https://doi.org/10.1073/pnas.1321458111Crossref (137) Furthermore, novo allows not included standard proteomics databases 2017Schumacher F.R. Delamarre Jhunjhunwala Modrusan Z. Phung Q.T. Elias J.E. Lill J.R. Building proteomic tool boxes monitor II peptides.Proteomics. 17: 1600061https://doi.org/10.1002/pmic.201600061Crossref (22) Here, report first unbiased, direct previously unknown MS, provide experimental evidence surface frequency highlights findings developing new cancers. line chosen model system this study. express six alleles, common allele, allowing broad spectrum (Scholtalbers 2015Scholtalbers Boegel Bukur Byl Goerges Sorn Loewer Sahin U. Castle J.C. TCLP: online catalogue integrating type, predicted neo-epitopes, virus gene expression.Genome 118https://doi.org/10.1186/s13073-015-0240-5Crossref competence determined transiently transfected dual-luciferase reporter (Boelz 2006Boelz Gehring N.H. Hentze M.W. Kulozik A.E. chemiluminescence-based decay.Biochem. Biophys. Commun. 2006; 349: 186-191https://doi.org/10.1016/j.bbrc.2006.08.017Crossref exhibit efficiency demonstrated significantly lower Renilla luciferase signal NS39 (0.055 ± 0.038 normalized WT signal, p ? 0.0001; Figure S1). NMD-restricting effect tested assessment transcript levels quantitative real-time PCR (qPCR). Treatment 5 ?M 24 h induced significant stabilization (p 0.0001) ATF3 (3.6-fold), SC35C (2.9-fold), SC35D (2.7-fold) (Figure MS-based offers only method (neo-)epitopes actually via molecules cells. extended recently published, high-throughput HLA-presented (Chong 2018Chong Marino Pak Racle Daniel R.T. Müller Gfeller Coukos High-throughput sensitive platform profound interferon?-mediated remodeling ligandome.Mol. 533-548https://doi.org/10.1074/mcp.TIR117.000383Abstract (86) investigate level 1). Briefly, immunoprecipitation (IP) pan-HLA antibody S2) subsequent separation bound molecules, HLAp were analyzed liquid chromatography-tandem mss (LC-MS/MS) applying fragmentation methods. Higher-energy collisional (HCD) mode acquiring high-resolution fast speed therefore provides in-depth coverage immunopeptidome. EThcD, combination HCD electron transfer dissociation, generates complex spectra leading ensure high-confidence identification. combined EThcD precursor selection targeting low-abundance precursors (lowEThcD) compensate slower acquisition fragmentation. Finally, obtained raw subjected sequencing-assisted database search, which improves accuracy enables above-described used isolated treated either solvent DMSO h. total 10,030 unique stringent false rate (FDR) 1% 2A). Of these, 3,098 (31% dataset) datasets recorded (EThcD/lowEThcD), whereas 4,907 (49%) fragmentation; 2,025 (20%) datasets. MS1 signals quantified wide range intensity spanning orders magnitude (log2 intensity: 12.75–32.55, mean log2 20.19). Peptides three methods slightly average intensity/abundance when alone 21.58 versus 19.77). note, subset 851 (mean 18.52) lowEThcD method, preferentially less abundant summary, illustrate benefit number providing comprehensive view controls performed dataset. First, calculated sequence-specific hydrophobicity index (HI), orthogonal parameter validate correct correlates experimentally observed retention times (Krokhin, 2006Krokhin O.V. Sequence-specific calculator. Algorithm prediction ion-pair RP-HPLC: 300- 100-Å pore size C18 sorbents.Anal. Chem. 78: 7785-7795https://doi.org/10.1021/ac060777wCrossref (185) HI tight correlation (Pearson's coefficient HCD: 0.96, EThcD: lowEThcD: 0.95; 2B). Next, HLA-associated properties I-typical length distribution nonamers 2C). Investigation entire reduces priori-introduced bias selectively surveying shortlisted predictions. posteriori represents suitable validation Using NetMHCpan, found 90% (8,988 peptides; 7,286 binder [SB], 1,702 weak [WB]) bind least 2D). Considering similarity HLAp, allele-specific binding-mediating anchor residues, clustered into groups four distinct motifs correspond consensus HLA-A?01:01, HLA-A?02:01, HLA-B?18:01, HLA-B?45:01 2E). Although HLA-C?05:01 HLA-C?07:01 be defined probably due low alleles motif redundancy HLA-A B (Bassani-Sternberg Neisig 1998Neisig Melief C.J.M. Neefjes Reduced HLA-C restricted stable TAP interaction.J. Immunol. 1998; 160: 171-179PubMed source 4,767 proteins, 2,524 (53%) represented one, 1,054 (22%) two, 516 (11%) surface. remaining 673 (14%) HLAp. general, associated cellular localization. previous reports, top 10% enriched clusters nuclear, cytoskeletal, ribosomal Walz 2015Walz Stickel J.S. Kowalewski D.J. Weisel Backert Kahn Nelde Stroh Handel al.The antigenic landscape multiple myeloma: (re)defines T-cell-based immunotherapy.Blood. 126: 1203-1213https://doi.org/10.1182/blood-2015-04-640532Crossref (71) Taken together, our dataset representative After having created high-quality, next sought query existence To constructed custom, line-specific based publicly available data. COSMIC CCLE contain 883 entries. lengths FS, 13 393 aa (median: 31 aa), 26,044 nine aa, preferred 11% (2,743 895 SB, 1,848 WB) S3). contains 1,260 reported analysis workflows, PEAKS reports scores, matched UniProt entry “de only” spectra. high-scoring subsequently searched custom databases. FDR 1%, InDel- SNP-neoepitope 3A Tables 1 2). Three candidates, BLASTp parameters short input rule out (and leucine/isoleucine permutations them) match sequences. excluded matching 14 part during generation 3A).Table 1Overview neoepitopes.GenePeptideLocation mutationRepeat & typeNumber aaHLA (Kd nM)CKAP2SLMEQIPHLchr13: 52474899A8, m114HLA-A?02:01 (2),HLA-C?05:01 (1,250),HLA-C?07:01 (1,665)NFAT5KRSSTILRLchr16: 69691085T5, m114HLA-C?07:01 (203)PSMC6REKHSWHEPchr14: 52713943A4, m128HLA-B?45:01 (1,218)STK38ISERDLLQYchr6: 36497806T7, m170HLA-A?01:01 (33)TUBGCP3GVWEKPRRVchr13: 112486018G5, m1108no binderHLA NetMHCpan 4.0. Underlined NFAT5-derived originate NFAT5 sequence. m1, minus base pair deletion. See also 3 Table S4. Open table tab 2Overview (GRCh37)aa changeHLA nM)CHMP7QTDQMVFNTY†chr8: 23116254p.A324THLA-A?01:01 (16),HLA-B?18:01 (4,799)NR1D1YSDNSNDSF†chr17: 38253572p.G39DHLA-A?01:01 (166),HLA-C?05:01 (23)PCMT1AAAPVVPQVchr6: 150117635p.A226VHLA-C?05:01 (2,282),HLA-C?07:01 (4,776)RBBP7EERVIDEEY†chrX: 16887311p.N61DHLA-B?18:01 (107)RGP1RLDPGEPKSYchr9: 35750729p.S110PHLA-A?01:01 (2,083),HLA-C?05:01 (6,887)HLA Mutated acids SNPs underlined. marked dagger. S4 S5. confirm identity/aa neoepitopes, those synthetic counterparts. samples applied pools confirmed identities (Figures 3B S4B). Intensities fragment ions very correlations correlation: 0.917–0.999) methods, much 0.073–0.673). substantial differences between identifications. Binding epitopes but validated genomic Sanger 3C S4C). validity 2 overview focused recurrence repeats genes CKAP2, NFAT5, PSMC6, STK38, TUBGCP3 analyzing (Table addition HCT-116, CKAP2 (KM12 [minus deletion (m1)], VaCo6 [m1], HROC24 [plus insertion (p1)], LS411