Immune cell infiltrates have proven highly relevant for colorectal carcinoma prognosis, making colorectal cancer a promising candidate for immunotherapy. Because tumors interact with the immune system via HLA-presented peptide ligands, exact knowledge of the peptidome constitution is fundamental for understanding this relationship. Here, we comprehensively describe the naturally presented HLA ligandome of colorectal carcinoma and corresponding nonmalignant colon (NMC) tissue. Mass spectrometry identified 35,367 and 28,132 HLA class I ligands on colorectal carcinoma and NMC, attributable to 7,684 and 6,312 distinct source proteins, respectively. Cancer-exclusive peptides were assessed on source protein level using the Kyoto Encyclopedia of Genes and Genomes (KEGG) and protein analysis through evolutionary relationships (PANTHER), revealing pathognomonic colorectal carcinoma–associated pathways, including Wnt, TGFβ, PI3K, p53, and RTK-RAS. Relative quantitation of peptide presentation on paired colorectal carcinoma and NMC tissue further identified source proteins from cancer- and infection-associated pathways to be overrepresented merely within the colorectal carcinoma ligandome. From the pool of tumor-exclusive peptides, a selected HLA-ligand subset was assessed for immunogenicity, with the majority exhibiting an existing T-cell repertoire. Overall, these data show that the HLA ligandome reflects cancer-associated pathways implicated in colorectal carcinoma oncogenesis, suggesting that alterations in tumor cell metabolism could result in cancer-specific, albeit not mutation-derived, tumor antigens. Hence, a defined pool of unique tumor peptides, attributable to complex cellular alterations that are exclusive to malignant cells, might comprise promising candidates for immunotherapeutic applications.

Significance: Cancer-associated pathways are reflected in the antigenic landscape of colorectal cancer, suggesting that tumor-specific antigens do not necessarily have to be mutation-derived but may also originate from other alterations in cancer cells. Cancer Res; 78(16); 4627–41. ©2018 AACR.

HLA-presented peptides on the cell surface constitute a key feature of adaptive immunity by presenting a showcase of the cellular interior to T cells. This ligandome can be of decisive importance for cell fate, constituting a reflection of cellular processes distorted by allele-specific peptide binding (1). While the colorectal carcinoma genome and transcriptome have widely been studied (2–4), analyses of the HLA ligandome are lacking and the identification of tumor-associated antigens has proven challenging. This is inter alia due to the limited correlation of genomic and transcriptomic alterations with the presentation of HLA-restricted ligands (5, 6). Direct elution and mass spectrometric analysis of naturally presented HLA-restricted peptides in contrast enables direct mapping of HLA ligandomes with relatively low experimental bias (7). Modern ligandomics studies provide considerable depth of ligandome analysis, and enable pinpointing cellular alterations presented on HLA to interact with T cells (8, 9). Profound knowledge of cancer-specific alterations and their influence on HLA presentation of antigenic peptides is particularly relevant for targeted immune interventions. The choice of suitable tumor-associated and/or tumor-specific antigens is crucial for preventing on-target off-tumor effects, induced by recognition of naturally presented HLA ligands on nonmalignant tissues (10, 11).

Despite increasing efforts in early detection and prevention, colorectal carcinoma persists among the most common cancers worldwide (12). Whereas curative therapies are available for early stage cancers, mortality in advanced/recurrent disease remains high (13).

Growing evidence suggests that unleashing the immune system by immune checkpoint inhibition is beneficial for the rare subset of patients with microsatellite instable (MSI) colorectal carcinomas (14), attributed to high mutational loads and dense infiltrates of tumor-infiltrating lymphocytes (15). Vice versa, mutated neoepitopes (for the terminology used in this article, please refer to Table 1) appear to have minor relevance for non-MSI colorectal carcinomas, demanding the identification of other suitable target structures. Beyond somatic mutations, tumor-specific alterations may include changes on every level of cellular metabolism, generating distinct tumor-associated or even tumor-specific—albeit not mutation-derived—variant peptides, also representing neoantigens. Such HLA-presented peptides can induce clinically relevant immune responses (8, 16) or represent suitable structures for targeted therapies.

Because inflammation is established to be associated with colorectal carcinoma occurrence and corresponding immune infiltrates have proven clinically relevant in tumors with low mutational burden (17), the expression of immunogenic peptides can be expected. Therefore, specific knowledge of the antigenic landscape in colorectal carcinoma is highly relevant (16).

In this study, we provide comprehensive data on the HLA-presented antigenic repertoire of a solid cancer, giving insights into molecular pathways represented within the ligandomic landscape of colorectal cancer.

Ethics approval and informed consent

This study was conducted in accordance with the Declaration of Helsinki and approved by the local Institutional Review Board of the University Hospital Tübingen. All participants provided written informed consent.

Patient materials

Primary colorectal carcinoma (HLA-class I: n = 30; HLA class II: n = 19) and corresponding nonmalignant colon (NMC; HLA class I, n = 35; HLA class II, n = 20) tissue samples were obtained from surgical specimens (Supplementary Table S1). Colorectal carcinoma diagnosis was confirmed as adenocarcinomas in all cases.

Peripheral blood was obtained from additional patients with colorectal carcinoma (n = 50) and blood donors (n = 15 each for all eight most frequent Caucasian HLA allotypes; ref. 18). For in vitro priming buffy coats from anonymous blood donors (Department of Transfusion Medicine, Tübingen) were employed. Peripheral blood mononuclear cells (PBMC) were isolated from blood or buffy coats by density-gradient centrifugation (LSM 1077, PAA Laboratories).

Patients with colorectal carcinoma (tissue cohort)

The study cohort (Supplementary Table S2) comprised 15 women and 22 men (median age, 69 years; 27–85 years at surgery). Tumors were localized in the caecum (n = 1), colon (n = 21), sigma (n = 13), or rectum (n = 2). Curatively intended resection was performed in all patients.

Patients were followed for a median of 34.8 months (0.3–109.2 months), median overall survival was not reached.

Patients with colorectal carcinoma (PBMC cohort)

PBMC samples from 50 further patients with colorectal carcinoma (Supplementary Table S2) were obtained, comprising 17 women and 33 men (median age, 62 years; 24–82 years).

Median follow-up was 36.0 months (2–167 months). Median overall survival after initial diagnosis was not reached. During follow-up, 16 and 34 patients were without detectable or with active disease, respectively. Forty-six patients remained alive.

HLA typing

High-resolution HLA typing (using clinically validated LUMINEX and sequence-based typing) was performed for HLA-A, HLA-B, HLA-C, HLA-DRB1, and HLA-DQB1 (tissue cohort, Supplementary Table S1). Molecular HLA typing (HLA-A, HLA-B, HLA-C) was performed for control patients and blood donors.

Isolation of HLA ligands from primary colorectal carcinoma and NMC tissues

HLA class I and II molecules were isolated by immunoaffinity purification using the pan–HLA-class I monoclonal antibody W6/32, the HLA-DR monoclonal antibody L243 and the pan–HLA-class II monoclonal antibody Tü39 (all produced in-house; ref. 19). Separate purification of HLA class I and II ligands was performed eluting peptides with 0.2% trifluoroacetic acid.

Analysis of HLA ligands by LC-MS/MS

HLA-ligand extracts separated for HLA class I and II were analyzed in up to five technical replicates each, as previously described (19). In brief, peptide samples were separated by nanoflow uHPLC (UltiMate 3,000 RSLCnano System, Thermo Fisher Scientific) using a 50 μm × 25 cm column (PepMap RSLC, Thermo Fisher Scientific) and a gradient ranging from 2.4% to 32.0% acetonitrile over the course of 90 minutes. Eluting peptides were analyzed in an online coupled LTQ Orbitrap XL mass spectrometer (Thermo Fisher Scientific) using a top 5 collision-induced dissociation fragmentation method.

Database search and spectral annotation

The Mascot search engine (Mascot version 2.2.04, Matrix Science) was used to search the human proteome in the Swiss-Prot database (20,279 reviewed protein sequences, September 2013) without enzymatic restriction (19). Oxidized methionine was allowed as a dynamic modification. The false discovery rate was estimated using the Percolator algorithm (20) and set to 5%. Peptide lengths were limited from 8 to 12 amino acids for HLA class I and 12 to 25 amino acids for HLA class II. Protein inference was disabled, allowing for multiple protein annotations of peptides. HLA class I annotation was performed using an in-house version of SYFPEITHI (21), and NetMHC (version 4.0; ref. 22). Annotation of HLA class I ligands was based on the HLA typing performed for each patient, ensuring high confidence annotations due to the restricted search space. HLA class I peptide immunogenicity was predicted using the IEDB prediction tool (tools.iedb.org/immunogenicity; ref. 23).

For further reference, an in-house database of HLA-ligandome data from 132 nonmalignant tissues was used (complete benign dataset), comprising NMC (n = 35), PBMCs (n = 33), kidney (n = 31), liver (n = 12), bone marrow (n = 10), or other tissues (n = 11).

Data availability

MS raw data for colorectal carcinoma/NMC samples have been deposited to the ProteomeXchange Consortium (http://proteomecentral.proteomexchange.org) via the PRIDE partner repository with the dataset identifier PXD009602.

Annotation of source proteins to metabolic pathways

Considering exclusively detected source proteins from colorectal carcinoma and NMC tissue only, Kyoto Encyclopedia of Genes and Genomes (KEGG; ref. 24) pathway analysis and Protein Analyses Through Evolutionary Relationships (PANTHER, version 10; ref. 25) were executed using PAVER software (Decodon) as described previously (26). Here, the tile color codes for exclusivity among colorectal carcinoma or NMC source proteins. The log2 detection quotient on colorectal carcinoma versus NMC tissue was coded as follows: orange depicts log2 ratios above zero (exclusively detected in colorectal carcinoma tissue), blue: log2 ratios below zero (exclusively detected in NMC). In these graphs, the size of single tiles is without relevance, whereas the polygon size attributable to distinct items correlates with the number of included source proteins.

Word clouds were created using an internet-based tool (www.worditout.com). Here, font size correlates with the relative abundance of colorectal carcinoma/NMC samples presenting HLA ligands attributable to the respective source proteins.

Relative quantitation of HLA-presented peptides

For label-free quantitation (LFQ) of relative HLA-ligand abundances, the total injected peptide amounts of all paired colorectal carcinoma/NMC samples were normalized and LC-MS/MS analysis was performed in five technical replicates for each sample. For normalization, the relative sample amounts in the respective ligand extracts were estimated based on the summed intensities of all peptide identifications detected in dose-finding mass spectrometry runs. Injected sample amounts were diluted for subsequent LFQ runs accordingly. Relative quantification of HLA ligands was performed using the area of precursor extracted ion chromatograms (XIC) using Proteome Discoverer version 1.4 (Thermo Fisher Scientific).

For Volcano plots, the ratios of the mean area of individual peptides in five replicate LFQ runs per condition were calculated (fold change; x-axis) and two-tailed Student t tests comparing peptide abundance in colorectal carcinoma and NMC tissue performed, applying Benjamini–Hochberg correction for multiple testing (y-axis), using an R script (version 3.2) as described previously (27). Only peptides detectable at least in two LFQ runs on either colorectal carcinoma or NMC were included. For imputation of missing peptide identifications, the median area of the five lowest intensity identifications in the same run was utilized as an estimate for the limit of detection (LOD). Peptides were considered significantly overrepresented on either colorectal carcinoma or NMC, if the fold change was above 2 log2 (4-fold overrepresented) with P values <0.01.

For all significantly overrepresented peptides on colorectal carcinoma or NMC, pathway annotation was performed using an internet-based Search Tool for the Retrieval of INteracting Genes/proteins (STRING, www.string-db.org, version 10, String Consortium, Swiss Institute for Bioinformatics, Zürich, Switzerland; ref. 28).

Immunohistochemistry

Thirty representative colorectal carcinoma samples were stained for HLA class I (clone HCA2, BIOZOL, diluted 1:200) and HLA class II (Clone M0775, DAKO, diluted 1:200). Staining was performed on formalin-fixed, paraffin-embedded tissue sections on a Lab Vision Autostainer (Thermo Scientific Scientific), following the manufacturer's protocols.

Peptide and HLA-peptide monomer synthesis

Peptides were synthesized by automated peptide synthesizers (EPS 221, Abimed; ABI 433A, Applied Biosystems) using the 9-fluorenylmethyl-oxycarbonyl/tert-butyl (Fmoc/tBu) strategy (29). Biotinylated recombinant HLA molecules and fluorescent HLA-peptide tetramers were produced as described previously (30).

Priming of peptide-specific T cells using aAPC

For the generation of artificial antigen-presenting cells (aAPC), streptavidin-coated polystyrene particles (Bangs Laboratories) were resuspended in PBE [PBS/BSA/ETDA (Gibco, Thermo Fisher Scientific)], containing biotinylated HLA-peptide monomer and anti-human biotinylated CD28 antibody (produced in-house) and incubated at room temperature for 30 minutes under continuous shaking. After washing, aAPCs were stored at 4°C prior to use (31). T cells from healthy blood donors were enriched by CD8+ positivity using magnetic cell sorting (Miltenyi Biotec). Stimulations were performed in 96-well plates with 1 × 106 T cells plus 2 × 105 aAPC complemented with human IL12 (PromoKine). IL2 (R&D) was added on days 5, 12, and 19. aAPC stimulation was repeated on days 10 and 17, for a total of 3 cycles. Cells were evaluated on day 29 in multimer stainings.

Presensitization of antigen-specific T cells

T cells were presensitized with peptides (1 μg/mL) and IL2 for 12 days (32). Cells were restimulated afterward and evaluated in IFNγ ELISpot.

IFNγ ELISpot assay

Presensitized or primed cells were stimulated (24 hours) with peptides (1 μg/mL) in antibody precoated nitrocellulose plates. Positive and negative controls (Supplementary Table S3), reagents and instruments were used as reported previously (32). Duplicate wells were considered positive if at least 10 spots per 250,000 PBMCs were detectable in either well, and the mean number of spots exceeded the spots in the negative control more than 3-fold. Unstimulated PBMCs and phytohemagglutinin (PHA, Sigma-Aldrich) were used as additional controls.

Multimer staining

Multimers were produced by addition of streptavidin-R-phycoerythrin conjugate (SAPE, Thermo Fisher Scientific) successively to a monomer, with 30 minutes of incubation while rotating overhead between each SAPE addition. Primed CD8+ T cells were stained with Life/Dead Fixable Aqua (Thermo Fisher Scientific), CD8-PE-Cy7 (Beckmann Coulter) and multimers and measured using a FACSCanto (BD Biosciences). Unstimulated and unstained PBMCs and ionomycin/phorbol 12-myristate-13-acetate (PMA) were used as controls.

Staining was only considered specific when at least a population of 0.5% of multimer-specific T cells was detectable.

Software and statistical analysis

R and Python scripts were used for the HLA-peptide plateau regression analysis (8).

Further statistical tests including two-sided and paired Student t tests, as well as Kaplan–Meier regression analysis, were performed using IBM SPSS version 22 (2011, IBM Corporation). P values <0.05 were considered statistically significant.

Identification of HLA-presented peptides on colorectal carcinoma and NMC tissue

Immunohistochemical staining for HLA class I and II molecules was confirmed on malignant cells as well as in the tumor microenvironment (Fig. 1A). The composition of naturally presented HLA class I and class II peptidomes of colorectal carcinoma (n = 30 and n = 19 for HLA classes I and II, respectively) and nonmalignant colon (NMC) tissue (n = 35 and n = 20 for HLA classes I and II, respectively) was assessed by mass spectrometry (MS/MS) after HLA immunoprecipitation (Fig. 1B).

The mean number of identified HLA class I–restricted peptides per colorectal carcinoma or NMC tissue sample was 1,171 (322–2,407) and 796 (59-1,802), respectively. The HLA allotypes included in this analysis covered a calculated 95.6% of a Caucasian population (Supplementary Fig. S1; ref. 18). Based on the HLA-allele distribution of our patient cohort, we estimated maximum attainable quantities of HLA class I and II ligand source proteins for colorectal carcinoma and NMC (Fig. 1C; Supplementary Fig. S2A and S2B, left and middle). For a further reference, an additional saturation analysis assessing a diverse set of benign tissues (n = 132) was performed (Supplementary Fig. S2A and S2B, right). The total number of distinct source proteins represented by the identified HLA ligands in colorectal carcinoma surpassed those in NMC (Fig. 1D; Supplementary Fig. S2).

Source proteins were either shared between colorectal carcinoma and NMC [n = 5,318 (61.3%) and n = 1,032 (49.6%) for HLA class I and II, respectively) or exclusive to colorectal carcinoma [n = 2,366 (27.3%) and 570 (27.4%) for HLA class I and II, respectively) or else exclusive to NMC [n = 994 (11.4%) and 477 (23.0%) for HLA class I and II, respectively; Fig. 1D). After subtracting all proteins contained in the complete benign dataset (Supplementary Fig. S3A and S3B), 758 (6.2% of total colorectal carcinoma source proteins for HLA class I, Supplementary Fig. S3C) and 310 (7.5% of total colorectal carcinoma source proteins for HLA class II, Supplementary Fig. S3D) source proteins, remained exclusively represented by HLA ligands on colorectal carcinoma tissue.

Target selection and immunogenicity evaluation

First, we assessed predicted immunogenicity of all identified peptides using IEDB class I immunogenicity prediction (Fig. 2A and B; y-axis). No differences between colorectal carcinoma– and NMC-exclusive, as well as shared peptides were obvious. Because immunogenicity can be associated with HLA-binding affinity (33), HLA-binding scores were assessed using NetMHC 4.0. Here again, no apparent differences were observed (Fig. 2B and C; x-axis), indicating that target selection based on predicted immunogenicity remains insufficient and requires extensive characterization of a range of colorectal carcinoma–exclusive peptides.

Nevertheless, we aimed to characterize the T-cell repertoire of a manageable number of rationally selected tumor-associated HLA class I ligands. Therefore, 5 peptides each for the seven most frequent HLA class I alleles in a Caucasian population (HLA-A*01, -A*02, -A*03, -A*24, -B*07, -B*08; -B*44) plus 3 peptides for HLA-C*07 were selected (Fig. 2D).

Peptide selection was based on the following criteria: Candidate HLA ligands were filtered for exclusive identification on colorectal carcinoma tissue (as compared with the complete benign dataset). Peptides derived from colorectal carcinoma–exclusive source proteins (n = 24, 63%) as well as colorectal carcinoma–exclusive peptides derived from nontumor exclusive proteins (n = 14, 37%; ★ in Fig. 2D) were selected and ranked by their prevalence within their respective HLA allotype (y-axis). Binding to their cognate HLA alleles was confirmed by peptide-monomer refolding (except for HLA-B*44 and -C*07 for technical reasons).

Immunogenicity of selected peptides was evaluated using two different approaches: (i) assessment of de novo induction of T-cell responses by aAPC in vitro priming using T cells from healthy volunteers (Fig. 2D, bottom and Fig. 2E) and (ii) analysis of pre-existing T-cell responses by IFNγ ELISpot in an independent group of patients with colorectal carcinoma (Fig. 2F) and healthy volunteers (Supplementary Table S4). Among the five peptides, previously described as T-cell epitopes, an existing T-cell repertoire was reconfirmed in two cases. Nineteen additional peptides were confirmed as putative T-cell epitopes in this study (Fig. 2E–G; Supplementary Table S4). The remaining peptides comprised 10 previously eluted and four newly identified naturally presented HLA ligands confirmed by MS. A full characterization as T-cell epitopes, including lysis assays, was unfeasible due to lacking patient material and beyond the scope of this work.

Our selection approach yielded a majority of peptides (63%: 8% in patients with colorectal carcinoma, 8% in blood donors; 47% in aAPC priming), against which, an existing T-cell repertoire could be verified.

Tumor association was further assessed by searching the scientific literature (Fig. 2H; Supplementary Table S4) as well as considering GTEx datasets (www.gtexportal.org), including highly proliferative tissue types. Sixteen of the ligand source proteins had been described as associated with colorectal carcinoma and further 16/38 proteins were reported as associated with other cancers. Of note, no previous study identified any of these ligands on native colorectal carcinoma tissue (Supplementary Table S4).

Tumor-associated alterations of the HLA ligandome

We then aimed to determine whether the HLA-ligand source proteins of colorectal carcinoma and NMC might be attributable to molecular interactions. We submitted all detected source proteins of either colorectal carcinoma (n = 30) or NMC (n = 35) to KEGG pathway analyses (26) revealing no global changes.

Next, we aimed to attribute colorectal carcinoma– and NMC-exclusive source proteins to signaling pathways. In PANTHER analyses, we observed an array of pathways emerging on colorectal carcinoma tissue including established tumor-associated pathways comprising Wnt, Integrin, and Cadherin signaling as well as p53 and inflammation-mediated cytokine signaling. We observed a prominent presence of growth factor signaling represented by FGF, EGF, and PDGF. Representation of distinct colorectal carcinoma–/NMC-exclusive source proteins was further depicted by word clouds (Fig. 3A for HLA class I, Fig. 3B for HLA class II, high-resolution image in Supplementary Fig. S4).

By comparative KEGG pathway analyses of the exclusive source proteins, colorectal carcinoma–exclusive source proteins were particularly attributable to epigenetic regulations (i.e., histone modifications) for both HLA class I– and class II–restricted peptides. Further, significantly overrepresented pathways in colorectal carcinoma discernable by HLA class I–restricted peptides comprised ribosome biosynthesis, protein processing in the endoplasmic reticulum, and RIG-I-like receptor signaling. With regard to HLA class II–restricted peptides, significantly overrepresented pathways in the colorectal carcinoma HLA ligandome included focal adhesion and cytokine cytokine-receptor interaction (Fig. 4A for HLA class I, Fig. 4B for HLA class II, high-resolution image in Supplementary Fig. S5).

Overrepresentation and tumor exclusivity of detected HLA ligands

To further elucidate tumor-associated alterations of the HLA ligandome, we evaluated HLA-ligand presentation on colorectal carcinoma versus matched autologous NMC tissue of individual patients by label-free quantitation (exemplified results are depicted in Fig. 5A; Supplementary Fig. S6A for HLA classes I and II, respectively).

The median abundance of 25.3% of identified HLA ligands (10.1%–44.8% per sample) appeared to be significantly changed. We detected overrepresentation on colorectal carcinoma [median: 12.6% of peptides (5.3–25.6%)] and overrepresentation on NMC [median: 12.1% of peptides (3.4%–19.9%)], indicating substantial differences in HLA-peptidome composition on colorectal carcinoma versus its nonmalignant tissue counterpart (Supplementary Tables S5 and S6). For HLA class II peptides, in median 17.1% (2.8%–23.9%) and 13.0% (0.5%–33.6%) of peptides were found to be overrepresented on colorectal carcinoma or NMC, respectively (Supplementary Tables S7 and S8). These significantly overrepresented peptides were again mapped to their source proteins (Fig. 5B; Supplementary Fig. S6B). Based on these analyses, we enriched ligands that were significantly overrepresented on colorectal carcinoma, and whose source proteins remained tumor exclusive as well as being shared by as many colorectal carcinoma samples as possible (≥3/13; increasing this limit abolished the respective overlap). The requirements of tumor exclusivity and significant overrepresentation on colorectal carcinoma are not absolutely interdependent, as the semi-random sampling of automated data dependent acquisition (DDA) MS may sporadically lead to the exclusive identification of a peptide in one specific condition. To control for some of the stochasticity of DDA MS [i.e., the possibility that some peptides are only identified exclusively on tumor tissue due to the semi-random sampling of the instrument (34)], we employed a two-tier filtering approach, which requires a protein to be exclusively represented on tumor tissue but also robust and reproducible detection of the corresponding HLA ligands. Therefore, the significance calculation included in the LFQ strategy was implemented to mitigate this limitation. Consequently, all source proteins covered by significantly overrepresented peptides on colorectal carcinoma (Fig. 5B) were compared with the complete benign dataset and evaluated for their HLA-restricted presentation on multiple colorectal carcinoma samples (Fig. 5C). This combined qualitative, semi-quantitative filtering strategy aims to prioritize tumor-associated antigens fulfilling all complimentary requirements and led to the identification of three tumor-exclusive source proteins represented by 10 HLA ligands (Fig. 5C). These peptides represent candidates for shared nonmutated (i.e., wild-type sequence) tumor-specific antigens.

Identical analyses of peptide representation were performed for HLA class II. Here, we identified only a single tumor-exclusive source protein (IL6R) represented by one derived peptide with significant overrepresentation on ≥2/7 colorectal carcinomas (Supplementary Fig. S6C).

Ligandome composition reflects cancer-associated pathways

To comprehensively assess potential interactions of HLA class I represented source proteins, significantly overrepresented source proteins (colorectal carcinoma and NMC) were further evaluated using a database of functional protein–protein interaction networks (www.string-db.org, version 10; ref. 28).

In the LFQ dataset (n = 13 matched sample pairs), source proteins of overrepresented peptides on colorectal carcinoma shared more protein–protein interactions (median 2.19-fold) as compared with source proteins of overrepresented peptides on NMC (Supplementary Table S9). Comprehensive functional protein association network analysis revealed associations of cancer- and infection-associated pathways in colorectal carcinoma overrepresented source proteins only. Pathways involved in cell adhesion and protein metabolism were likewise overrepresented in colorectal carcinoma, while pathways involved in degenerative processes and normal cell metabolism were found overrepresented within the ligandome of NMC (Fig. 6A; Supplementary Table S10). Including merely colorectal carcinoma overrepresented source proteins attributable to cancer-associated pathways (Fig. 6B), we observed close protein–protein interactions of all proteins except for PIK3AP1. Furthermore, five of the source proteins were directly linked to colorectal carcinoma (Fig. 6B; dark gray). When assessing infection related pathways only, we observed seven source proteins also related to pathways in cancer (Fig. 6C; dark gray nodes) and eleven source proteins linked to EBV infection (Fig. 6D; dark gray nodes).

Selection of potential candidate antigens for human application

Finally, these principally separate selection strategies (tumor exclusivity, Fig. 1; immunogenicity, Fig. 2; PANTHER association, Fig. 3; KEGG-pathway association, Fig. 4; overrepresentation, Fig. 5; STRING analysis, Fig. 6) were combined to identify potential target structures for use in immunotherapeutic strategies with vaccination, MHC antibody and chimeric antigen receptor (CAR) T-cell approaches in mind (Supplementary Table S11). The main selection criteria comprised most importantly tumor exclusivity (ideally on source protein level), as well as cancer association of source proteins, overrepresentation on tumor tissue and confirmation of a T-cell repertoire. Frequent detection on various tumors without evidence for HLA-restricted presentation on nonmalignant tissue was also considered. After manual assessment, 12 naturally presented HLA ligands were chosen as the most promising candidate antigens (Supplementary Table S12). This selection represents a subjective choice by the authors, which is based on a rational selection strategy based on the obtained data. All selected antigens supposedly represent wild-type sequence nonmutated neoantigens.

With its breakthrough success, immunotherapy, especially immune checkpoint inhibition, has been introduced into the standard treatment of different tumors, including melanoma (35) and non–small cell lung cancer (36). In colorectal carcinoma, it has proven clinical efficacy in patients with MSI tumors only (14). Presumably high loads of mutated neoepitopes do not constitute the only mechanism of immune recognition in colorectal carcinoma, because it has been observed that defined immune infiltrates in colorectal carcinoma correlate with survival and are more important than the mutational status (37). These data suggest that immunotherapy with relevant colorectal carcinoma antigens is worthwhile, requiring a profound knowledge of the HLA antigenome (38). Most previous studies have focused on the identification of mutation-derived neoepitopes and showed that—especially in native human tumor tissue—such antigens are hard to find and if detectable limited to individual tumors, indicating that applicability of these neoantigens has to be strictly personalized (9, 39).

We here present a first of its kind integrative in-depth analysis of the HLA-presented ligandome, comparing tissue from colorectal carcinoma and NMC including analyses of paired samples. The observed ligand numbers are in line with current observations by other investigators (9).

So far comprehensive knowledge on the HLA ligandome in malignant disease is fragmentary, as there is a strong bias toward describing ligands eluted from malignant tissue only (9) or toward comparison with unrelated tissues, when the matching nonmalignant autologous counterpart is unavailable (8). In contrast, our analysis allows the relative quantitation of HLA-ligand presentation on matched malignant and nonmalignant tissue obtained from the same patient.

Direct identification of peptides using MS provides high accuracy in identifying naturally presented HLA ligands, which is in sharp contrast to predictions inferring the HLA ligandome from genomic data (5). Current research furthermore suggests that only a small percentage of the expressed genome is actually presented on HLA under physiologic conditions (40).

In our study, tumor-associated and/or -exclusive peptides were assessed using two principally different approaches:

  • (i) Direct comparison of HLA-ligand source proteins between colorectal carcinoma and NMC, as well as an in-house database of 132 benign tissues from different organs, revealed about 750 source proteins (6.2% of the assessed proteome) that remained exclusive to colorectal carcinoma. This dataset was assembled inter alia to reflect potentially infiltrating blood cells into cancers to prevent cross-contamination with ligands derived from nonresident cells. This aspect is of special importance for HLA class II presented peptides, because HLA class II is mostly lacking in colorectal carcinomas (as also confirmed for our tissue cohort). With this approach, peptides derived from proteins were detected that were missing in our NMC dataset, indicating a differential regulation of the proteome and/or antigen processing machinery in colorectal carcinoma. Comparing the HLA-ligandome size estimated based on our data by regression analysis to the proteome size of the colon (http://www.proteinatlas.org/humanproteome/colon), we theoretically expect to find HLA ligands mapping to source proteins that would cover about 60% of the colon proteome in NMC. This estimated ligandome size coincides with the 59% share of identified HLA ligands that mapped to genes expressed in lymphocytes previously published (40). The number of MS-detected distinct source proteins represented on HLA of NMC reached ∼80% of the extrapolated saturation level. Further, the numbers of distinct source proteins that constitute the HLA ligandome in colorectal carcinoma assessed by MS exceed the amount of distinct source proteins covered by ligands in NMC. Respective datasets were further functionally annotated, comparing the colorectal carcinoma and NMC presented HLA peptidomes by KEGG and PANTHER analyses (26). Evaluating the colorectal carcinoma–associated HLA-class I ligandome, major pathways implicated in colorectal carcinoma oncogenesis, including Wnt, p53, RAS, and PI3K (4) were found represented. Further TGFβ signaling emerged prominently, which, although usually characterized by mutational inactivation, in colorectal carcinoma paradoxically constitutes a characteristic feature (41).

  • (ii) Assessment of overrepresented HLA ligands showed significant differences in their relative abundance on colorectal carcinoma versus NMC. Comparing matched colorectal carcinoma/NMC pairs, we analyzed the differential presentation of nonmutated peptides by relative quantification. With this approach, we were able to identify altered HLA-ligand representation within the ligandome. Both significantly overrepresented source proteins on colorectal carcinoma or NMC were evaluated using tree maps visualization (28). Solely in colorectal carcinoma, source proteins of HLA class I presented peptides related to cancer-associated pathways as well as infection were shown to be significantly overrepresented, again documenting a prominent representation of signaling pathways involved in carcinogenesis (4). This indicates that tumor-specific and/or -associated HLA ligands can be influenced by metabolic alterations in tumor cells (42).

Taken together, these different analytical approaches can be combined for a comprehensive portrayal of the HLA ligandome and may even be used to select peptide candidates for clinical application including vaccination, CAR T cells (43) and TCR mimic antibodies (44). The latter two approaches have gained relevant interest for clinical development and require a profound knowledge of the HLA ligandome to prevent on-target off-tumor effects. Here, characterization of the immunogenicity of HLA-presented peptides themselves is of subordinate importance whereas robust HLA-ligand presentation is crucial.

Clinical translation is therefore not restricted to vaccination approaches only, requiring a naturally occurring T-cell repertoire (which repeatedly yielded disappointing results in the past). Considering vaccine approaches, a full characterization of the immunologic relevance and lytic capacity of T cells against the identified HLA ligands might be worthwhile and may overcome limitations of previous attempts but requires extensive and time-consuming efforts (45). However, it should be noted that full immunologic characterization of potential vaccine antigens has never been a prerequisite before clinical application (8, 46). Even such a characterization in vitro could not prevent failure in clinical trials due to tumor immune escape mechanisms including antigen loss, development of T-cell anergy, and immune-suppressive tumor microenvironment, etc. (47, 48).

With regard to active immunization and definition of suitable sets of vaccine candidates, the computational preselection based on predicted immunogenicity appears ineffective so far because differences between colorectal carcinoma– and NMC-exclusive or shared peptides are negligible. Against this background, a selected set of HLA ligands covering a subset of the most frequent HLA alleles among Caucasians was chosen and assessed for immunogenicity in our work. With this rational approach, 63% of selected HLA ligands were attested as putative T-cell epitopes and thus promising targets. Due to limited patient material, no further evaluation for their lytic capacity was possible.

Interestingly, we identified also T-cell reactivity against some peptides in anonymous blood donors, preventing correlation with clinical parameters (e.g., occurrence of colorectal adenomas, chronic inflammatory bowel disease). These observations are in line with previous reports on circulating melanoma-specific T cells without any obvious physiological relevance (49).

Although our observations in principle have the ability to identify immunologically promising target structures, establishing a clinically applicable warehouse or recommending T-cell targets for clinical use was beyond the scope of this work. The selection of suitable target structures remains a major challenge, especially for vaccination approaches. Such approaches might include therapeutic vaccination with strong adjuvantation, T-cell receptor gene therapy and adoptive transfer of T cells with specificity for the identified nonmutated cancer-specific antigens.

To our knowledge, none of the previous approaches for evaluation of the HLA ligandome (antigenic repertoire) as well as the immunopeptidome (immunogenic repertoire) considered the complex multifaceted alterations of cancer cells. These insufficiently elucidated processes may alter physiologic protein metabolism profoundly (42, 50), leading to cancer-exclusive presentation of unexpected target structures with therapeutic implications.

In summary, we provide a comprehensive dataset of the HLA ligandome in a solid cancer and its autologous nonmalignant counterpart. The approach, combining mass spectrometry with functional assessment, also led to the identification of tumor exclusive as well as significantly overrepresented peptides, which—interestingly—were attributable to signaling pathways known to be pathognomonic for colorectal carcinoma (4). This indicates that not only genetic mutations may give rise to tumor-specific antigens but suggests that complex alterations of protein metabolism can influence the ligandomic landscape extensively, leading to tumor-specific HLA-restricted peptides constituting nonmutated neoantigens.

M.W. Löffler, D.J. Kowalewski, H. Schuster, S. Stevanović, and S.P. Haen are the inventors of patents owned by Immatics Biotechnologies GmbH. D.J. Kowalewski, L. Backert, and H. Schuster are currently employees of Immatics Biotechnologies GmbH. H.-G. Rammensee has ownership interest (including patents) in Immatics, CureVac, and Synimmune. No potential conflicts of interest were disclosed by the other authors.

Conception and design: M.W. Löffler, H. Schuster, I. Königsrainer, O. Kohlbacher, A. Königsrainer, H.-G. Rammensee, S. Stevanović, S.P. Haen

Development of methodology: D.J. Kowalewski, H. Schuster, S.P. Haen

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M.W. Löffler, D.J. Kowalewski, P. Adam, F. Dengler, D. Backes, S. Beckert, S. Wagner, L. Kanz, A. Königsrainer, S.P. Haen

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M.W. Löffler, D.J. Kowalewski, L. Backert, J. Bernhardt, P. Adam, H. Schuster, F. Dengler, A. Königsrainer, H.-G. Rammensee, S.P. Haen

Writing, review, and/or revision of the manuscript: M.W. Löffler, D.J. Kowalewski, L. Backert, P. Adam, F. Dengler, H.-G. Kopp, S. Beckert, I. Königsrainer, O. Kohlbacher, L. Kanz, A. Königsrainer, H.-G. Rammensee, S. Stevanović, S.P. Haen

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M.W. Löffler, D.J. Kowalewski, P. Adam, H.-G. Kopp, S. Beckert, S. Wagner, S.P. Haen

Study supervision: M.W. Löffler, S. Beckert, O. Kohlbacher, A. Königsrainer, S.P. Haen

The authors wish to thank Claudia Falkenburger, Patricia Hrstic, Nicole Bauer, Katharina Fiedler, and Beate Pömmerl for production of peptides and monoclonal antibodies, as well as excellent technical support. We gratefully acknowledge the support and helpful discussions with Cécile Gouttefangeas on T-cell immunology and Damian Szklarczyk, Swiss Institute of Bioinformatics (Zürich, Switzerland), on STRING analyses. We also thank Reinhild Klein for HLA typing, as well as the nurses and staff of the outpatient clinic of the Department for Oncology and Hematology such as the staff of the study center of the Department of General, Visceral and Transplant Surgery for provision of samples and support in patient care as well as all the participating patients. Further, we would like to express our gratitude to Falko Fend, Institute of Pathology and Neuropathology, University of Tübingen, for the provision of sample material, as well as the Department of Transfusion Medicine, University of Tübingen, for the provision of sample materials from healthy donors.

This work was supported by grants of the Deutsche Krebshilfe (Project No. 110465) to S.P. Haen and the European Research Council (ERC AdG 339842 MUTAEDITING) to H.G. Rammensee, and received funding by the Deutsche Forschungsgemeinschaft (DFG, SFB 685) to I. Königsrainer, A. Königsrainer, O. Kohlbacher, H.G. Rammensee, and S. Stevanović.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Rammensee
HG
,
Falk
K
,
Rotzschke
O
. 
Peptides naturally presented by MHC class I molecules
.
Annu Rev Immunol
1993
;
11
:
213
44
.
2.
Robles
AI
,
Traverso
G
,
Zhang
M
,
Roberts
NJ
,
Khan
MA
,
Joseph
C
, et al
Whole-exome sequencing analyses of inflammatory bowel disease-associated colorectal cancers
.
Gastroenterology
2016
;
150
:
931
43
.
3.
Haan
JC
,
Labots
M
,
Rausch
C
,
Koopman
M
,
Tol
J
,
Mekenkamp
LJ
, et al
Genomic landscape of metastatic colorectal cancer
.
Nat Commun
2014
;
5
:
5457
.
4.
The Cancer Genome Atlas Network
. 
Comprehensive molecular characterization of human colon and rectal cancer
.
Nature
2012
;
487
:
330
7
.
5.
Weinzierl
AO
,
Lemmel
C
,
Schoor
O
,
Muller
M
,
Kruger
T
,
Wernet
D
, et al
Distorted relation between mRNA copy number and corresponding major histocompatibility complex ligand density on the cell surface
.
Mol Cell Proteomics
2007
;
6
:
102
13
.
6.
Bassani-Sternberg
M
,
Pletscher-Frankild
S
,
Jensen
LJ
,
Mann
M
. 
Mass spectrometry of human leukocyte antigen class I peptidomes reveals strong effects of protein abundance and turnover on antigen presentation
.
Mol Cell Proteomics
2015
;
14
:
658
73
.
7.
Abelin
JG
,
Keskin
DB
,
Sarkizova
S
,
Hartigan
CR
,
Zhang
W
,
Sidney
J
, et al
Mass spectrometry profiling of HLA-associated peptidomes in mono-allelic cells enables more accurate epitope prediction
.
Immunity
2017
;
46
:
315
26
.
8.
Kowalewski
DJ
,
Schuster
H
,
Backert
L
,
Berlin
C
,
Kahn
S
,
Kanz
L
, et al
HLA ligandome analysis identifies the underlying specificities of spontaneous antileukemia immune responses in chronic lymphocytic leukemia (CLL)
.
Proc Natl Acad Sci U S A
2015
;
112
:
E166
75
.
9.
Bassani-Sternberg
M
,
Braunlein
E
,
Klar
R
,
Engleitner
T
,
Sinitcyn
P
,
Audehm
S
, et al
Direct identification of clinically relevant neoepitopes presented on native human melanoma tissue by mass spectrometry
.
Nat Commun
2016
;
7
:
13404
.
10.
Morgan
RA
,
Chinnasamy
N
,
Abate-Daga
D
,
Gros
A
,
Robbins
PF
,
Zheng
Z
, et al
Cancer regression and neurological toxicity following anti-MAGE-A3 TCR gene therapy
.
J Immunother
2013
;
36
:
133
51
.
11.
Linette
GP
,
Stadtmauer
EA
,
Maus
MV
,
Rapoport
AP
,
Levine
BL
,
Emery
L
, et al
Cardiovascular toxicity and titin cross-reactivity of affinity-enhanced T cells in myeloma and melanoma
.
Blood
2013
;
122
:
863
71
.
12.
Miller
KD
,
Siegel
RL
,
Lin
CC
,
Mariotto
AB
,
Kramer
JL
,
Rowland
JH
, et al
Cancer treatment and survivorship statistics, 2016
.
CA Cancer J Clin
2016
;
66
:
271
89
.
13.
Haggar
FA
,
Boushey
RP
. 
Colorectal cancer epidemiology: incidence, mortality, survival, and risk factors
.
Clin Colon Rectal Surg
2009
;
22
:
191
7
.
14.
Le
DT
,
Uram
JN
,
Wang
H
,
Bartlett
BR
,
Kemberling
H
,
Eyring
AD
, et al
PD-1 blockade in tumors with mismatch-repair deficiency
.
N Engl J Med
2015
;
372
:
2509
20
.
15.
Maby
P
,
Tougeron
D
,
Hamieh
M
,
Mlecnik
B
,
Kora
H
,
Bindea
G
, et al
Correlation between density of CD8+ T-cell infiltrate in microsatellite unstable colorectal cancers and frameshift mutations: a rationale for personalized immunotherapy
.
Cancer Res
2015
;
75
:
3446
55
.
16.
Heemskerk
B
,
Kvistborg
P
,
Schumacher
TN
. 
The cancer antigenome
.
EMBO J
2013
;
32
:
194
203
.
17.
Mlecnik
B
,
Tosolini
M
,
Kirilovsky
A
,
Berger
A
,
Bindea
G
,
Meatchi
T
, et al
Histopathologic-based prognostic factors of colorectal cancers are associated with the state of the local immune reaction
.
J Clin Oncol
2011
;
29
:
610
8
.
18.
Bui
HH
,
Sidney
J
,
Dinh
K
,
Southwood
S
,
Newman
MJ
,
Sette
A
. 
Predicting population coverage of T-cell epitope-based diagnostics and vaccines
.
BMC Bioinformatics
2006
;
7
:
153
.
19.
Kowalewski
DJ
,
Stevanovic
S
. 
Biochemical large-scale identification of MHC class I ligands
.
Methods Mol Biol
2013
;
960
:
145
57
.
20.
Kall
L
,
Canterbury
JD
,
Weston
J
,
Noble
WS
,
MacCoss
MJ
. 
Semi-supervised learning for peptide identification from shotgun proteomics datasets
.
Nat Methods
2007
;
4
:
923
5
.
21.
Rammensee
H
,
Bachmann
J
,
Emmerich
NP
,
Bachor
OA
,
Stevanovic
S
. 
SYFPEITHI: database for MHC ligands and peptide motifs
.
Immunogenetics
1999
;
50
:
213
9
.
22.
Andreatta
M
,
Nielsen
M
. 
Gapped sequence alignment using artificial neural networks: application to the MHC class I system
.
Bioinformatics
2016
;
32
:
511
7
.
23.
Calis
JJ
,
Maybeno
M
,
Greenbaum
JA
,
Weiskopf
D
,
De Silva
AD
,
Sette
A
, et al
Properties of MHC class I presented peptides that enhance immunogenicity
.
PLoS Comput Biol
2013
;
9
:
e1003266
.
24.
Kanehisa
M
,
Goto
S
,
Sato
Y
,
Furumichi
M
,
Tanabe
M
. 
KEGG for integration and interpretation of large-scale molecular data sets
.
Nucleic Acids Res
2012
;
40
:
D109
14
.
25.
Mi
H
,
Poudel
S
,
Muruganujan
A
,
Casagrande
JT
,
Thomas
PD
. 
PANTHER version 10: expanded protein families and functions, and analysis tools
.
Nucleic Acids Res
2016
;
44
:
D336
42
.
26.
Liebermeister
W
,
Noor
E
,
Flamholz
A
,
Davidi
D
,
Bernhardt
J
,
Milo
R
. 
Visual account of protein investment in cellular functions
.
Proc Natl Acad Sci USA
2014
;
111
:
8488
93
.
27.
Kowalewski
DJ
,
Walz
S
,
Backert
L
,
Schuster
H
,
Kohlbacher
O
,
Weisel
K
, et al
Carfilzomib alters the HLA-presented peptidome of myeloma cells and impairs presentation of peptides with aromatic C-termini
.
Blood Cancer J
2016
;
6
:
e411
.
28.
Szklarczyk
D
,
Franceschini
A
,
Wyder
S
,
Forslund
K
,
Heller
D
,
Huerta-Cepas
J
, et al
STRING v10: protein-protein interaction networks, integrated over the tree of life
.
Nucleic Acids Res
2015
;
43
:
D447
52
.
29.
Behrendt
R
,
White
P
,
Offer
J
. 
Advances in Fmoc solid-phase peptide synthesis
.
J Pept Sci
2016
;
22
:
4
27
.
30.
Garboczi
DN
,
Hung
DT
,
Wiley
DC
. 
HLA-A2-peptide complexes: refolding and crystallization of molecules expressed in Escherichia coli and complexed with single antigenic peptides
.
Proc Natl Acad Sci U S A
1992
;
89
:
3429
33
.
31.
Britten
CM
,
Gouttefangeas
C
,
Welters
MJ
,
Pawelec
G
,
Koch
S
,
Ottensmeier
C
, et al
The CIMT-monitoring panel: a two-step approach to harmonize the enumeration of antigen-specific CD8+ T lymphocytes by structural and functional assays
.
Cancer Immunol Immunother
2008
;
57
:
289
302
.
32.
Mirza
N
,
Zierhut
M
,
Korn
A
,
Bornemann
A
,
Vogel
W
,
Schmid-Horch
B
, et al
Graft versus self (GvS) against T-cell autoantigens is a mechanism of graft-host interaction
.
Proc Natl Acad Sci U S A
2016
;
113
:
13827
32
.
33.
Wang
S
,
Li
J
,
Chen
X
,
Wang
L
,
Liu
W
,
Wu
Y
. 
Analyzing the effect of peptide-HLA-binding ability on the immunogenicity of potential CD8+ and CD4+ T cell epitopes in a large dataset
.
Immunol Res
2016
;
64
:
908
18
.
34.
Aebersold
R
,
Mann
M
. 
Mass spectrometry-based proteomics
.
Nature
2003
;
422
:
198
207
.
35.
Hodi
FS
,
O'Day
SJ
,
McDermott
DF
,
Weber
RW
,
Sosman
JA
,
Haanen
JB
, et al
Improved survival with ipilimumab in patients with metastatic melanoma
.
N Engl J Med
2010
;
363
:
711
23
.
36.
Reck
M
,
Rodriguez-Abreu
D
,
Robinson
AG
,
Hui
R
,
Csoszi
T
,
Fulop
A
, et al
Pembrolizumab versus chemotherapy for PD-L1-positive non-small-cell lung cancer
.
N Engl J Med
2016
;
375
:
1823
33
.
37.
Mlecnik
B
,
Bindea
G
,
Angell
HK
,
Maby
P
,
Angelova
M
,
Tougeron
D
, et al
Integrative analyses of colorectal cancer show immunoscore is a stronger predictor of patient survival than microsatellite instability
.
Immunity
2016
;
44
:
698
711
.
38.
Caron
E
,
Kowalewski
DJ
,
Chiek
KC
,
Sturm
T
,
Schuster
H
,
Aebersold
R
. 
Analysis of major histocompatibility complex (MHC) immunopeptidomes using mass spectrometry
.
Mol Cell Proteomics
2015
;
14
:
3105
17
.
39.
Fritsche
J
,
Rakitsch
B
,
Hoffgaard
F
,
Romer
M
,
Schuster
H
,
Kowalewski
DJ
, et al
Translating immunopeptidomics to immunotherapy—decision making for patient and personalized target selection
.
Proteomics
2018
.
e1700284
.
doi: 10.1002/pmic.201700284
.
40.
Pearson
H
,
Daouda
T
,
Granados
DP
,
Durette
C
,
Bonneil
E
,
Courcelles
M
, et al
MHC class I-associated peptides derive from selective regions of the human genome
.
J Clin Invest
2016
;
126
:
4690
701
.
41.
Calon
A
,
Espinet
E
,
Palomo-Ponce
S
,
Tauriello
DV
,
Iglesias
M
,
Cespedes
MV
, et al
Dependency of colorectal cancer on a TGF-beta-driven program in stromal cells for metastasis initiation
.
Cancer Cell
2012
;
22
:
571
84
.
42.
Diederichs
S
,
Bartsch
L
,
Berkmann
JC
,
Frose
K
,
Heitmann
J
,
Hoppe
C
, et al
The dark matter of the cancer genome: aberrations in regulatory elements, untranslated regions, splice sites, non-coding RNA and synonymous mutations
.
EMBO Mol Med
2016
;
8
:
442
57
.
43.
Maus
MV
,
Plotkin
J
,
Jakka
G
,
Stewart-Jones
G
,
Riviere
I
,
Marghoub
T
, et al
An MHC-restricted antibody-based chimeric antigen receptor requires TCR-like affinity to maintain antigen specificity
.
Mol Ther Oncolytics
2017
;
3
:
1
9
.
doi: 10.1038/mto.2016.23
.
44.
Dubrovsky
L
,
Dao
T
,
Gejman
RS
,
Brea
EJ
,
Chang
AY
,
Oh
CY
, et al
T cell receptor mimic antibodies for cancer therapy
.
Oncoimmunology
2015
;
5
:
e1049803
.
doi: 10.1080/2162402X.2015.1049803
.
45.
Laske
K
,
Shebzukhov
YV
,
Grosse-Hovest
L
,
Kuprash
DV
,
Khlgatian
SV
,
Koroleva
EP
, et al
Alternative variants of human HYDIN are novel cancer-associated antigens recognized by adaptive immunity
.
Cancer Immunol Res
2013
;
1
:
190
200
.
46.
Rini
BI
,
Stenzl
A
,
Zdrojowy
R
,
Kogan
M
,
Shkolnik
M
,
Oudard
S
, et al
IMA901, a multipeptide cancer vaccine, plus sunitinib versus sunitinib alone, as first-line therapy for advanced or metastatic renal cell carcinoma (IMPRINT): a multicentre, open-label, randomised, controlled, phase 3 trial
.
Lancet Oncol
2016
;
17
:
1599
611
.
47.
Staveley-O'Carroll
K
,
Sotomayor
E
,
Montgomery
J
,
Borrello
I
,
Hwang
L
,
Fein
S
, et al
Induction of antigen-specific T cell anergy: an early event in the course of tumor progression
.
Proc Natl Acad Sci U S A
1998
;
95
:
1178
83
.
48.
Beatty
GL
,
Gladney
WL
. 
Immune escape mechanisms as a guide for cancer immunotherapy
.
Clin Cancer Res
2015
;
21
:
687
92
.
49.
Pittet
MJ
,
Valmori
D
,
Dunbar
PR
,
Speiser
DE
,
Lienard
D
,
Lejeune
F
, et al
High frequencies of naive Melan-A/MART-1-specific CD8(+) T cells in a large proportion of human histocompatibility leukocyte antigen (HLA)-A2 individuals
.
J Exp Med
1999
;
190
:
705
15
.
50.
Tsherniak
A
,
Vazquez
F
,
Montgomery
PG
,
Weir
BA
,
Kryukov
G
,
Cowley
GS
, et al
Defining a cancer dependency map
.
Cell
2017
;
170
:
564
76
.