Amino acid shortages lead to amino acid misincorporations in human cancer cells To examine specific alterations in protein synthesis following tryptophan depletion induced by IFNγ treatment, we used MD55A3 melanoma cells stably expressing V5-ATF4(1-63)-tGFP in-frame and +1 out-of-frame reporters (Fig. 1a). In these vectors, a tryptophan-less turboGFP gene (tGFP) is placed either in-frame or out-of-frame (+1) downstream of an unstable V5-tagged-ATF4(1-63) fragment that includes a single tryptophan at position 93 (W93) (Bartok et al., 2021. Nature 590: 332-337). The +1 vector leads to the expression of a shorter protein due to a premature stop codon (marked *). We expected that IFNγ-induced tryptophan depletion would block in-frame mRNA translation downstream of the single tryptophan codon while inducing the out-of-frame tGFP, as previously reported (Bartok et al., 2021. Nature 590: 332-337). We therefore treated the cells with IFNγ for 48 hours, and subsequently subjected them to either mock or proteasome inhibition (MG132) for additional 4 hours to collect the proteins expressed during this period (Fig. 1b). Surprisingly, immunoblotting with anti-V5 revealed the synthesis of both the out-of-frame tGFP and the in-frame proteins from the +1 construct under IFNγ treatment condition (Fig. 1b, compare lanes 8 vs 7 and 6 vs 5). Similar results were obtained with tryptophan depletion (data not shown). Moreover, we observed in-frame production following either IFNγ treatment or tryptophan depletion also in the case of the Frame construct (Fig. 1b (lanes 1-4) and data not shown). The production of tGFP was verified by anti-tGFP staining. The persistent in-frame production in depleted tryptophan conditions, while the reporter contains a tryptophan codon, suggests that ribosomes somehow managed to bypass the tryptophan codon and stay in-frame despite the absence of tryptophan (data not shown)).

Deprivation of amino acids was shown in bacteria and yeast to lead to regulated codon reassignment following amino acid deprivation (Mordret et al., 2019. Mol Cell 75: 427-441; Drummond and Wilke, 2009. Nat Rev Genet 10: 715- 724). Whether this occurs in human cells at the proteome level is yet unknown. Translational bypass, a process by which ribosomes skip three or more nucleotides without decoding (Baranov et al., 2015. Nat Rev Genet 16: 517-529), and theoretically also codon skipping could be other possibilities leading to in-frame protein product (Fig. 1c). To test these options experimentally, we performed mass spectrometry (MS) analysis of V5-Tag immunoprecipitated (IP) proteins (V5- IP/MS) from lysates obtained from either mock or IFNγ-treated +1-vector- expressing MD55A3 cells. We then searched for out-of-frame tGFP peptides as a control for the treatment, as well as for all possible codon reassignment, translational bypass and codon skipping events that may arise at codon W93. We confirmed the occurrence of out-of-frame tGFP peptides following IFNγ treatment, as we previously reported (Bartok et al., 2021. Nature 590: 332-337). While no codon skipping or translational bypass at the tryptophan codon were detected, a specific and exclusive tryptophan to phenylalanine (W93F) substitution, resulting from a regulated codon reassignment, appeared only in the IFNγ treatment condition (data not shown). Globally, while IFNγ-treatment decreased the detection of peptides originating from the in-frame sequence, including the peptide spanning the single W93, both out-of-frame tGFP peptides and the W93F substitution peptide, were markedly induced (Fig. Id). Besides, no other amino acid substitution was enriched upon IFNγ treatment at this tryptophan codon or its surrounding, confirming the specificity and exclusivity of the W93F substitution (Fig. 1e).

The extent of W93F substitution indicates that this is a major translational event following tryptophan shortage. To strengthen the above observations and to probe for the cause of this substitution directly, we depleted the same reporter cells of tryptophan, and performed V5-IP/MS analyses. Corroborating the results from the IFNγ treatment, also tryptophan depletion induced frameshifting and a specific and exclusive W93F substitution (data not shown). In comparison, depletion of either tyrosine (Y) or phenylalanine (F) did not lead to any W93F substitutions, but rather to Y99F (upon Y-depletion), and upon F-depletion to F76Y and F76L (leucine) mis-incorporations (Fig. 1f). Taken together, these results indicate that in the context of our reporter vector, amino acid deprivations induce substitutions, codon reassignments that are highly specific with respect to the mis-incorporated amino acid.

Next, it was addressed whether tryptophan to phenylalanine misincorporations (W>F) facilitate the generation of full-length protein in the absence of tryptophan. For this purpose, we took advantage of the tryptophan-less tGFP vector and substituted a conserved phenylalanine at position 26 with tryptophan (tGFPF26W). When expressed in MD55A3 cells (MD55A3-tGFPF26W), this F26W mutant abolished the green fluorescence signal of tGFP as determined by flow cytometry analysis (data not shown). We, therefore, hypothesized that if tryptophan shortage can lead to stable W>F substitutions, green fluorescence should appear following the generation of mature wild-type proteins. Indeed, in response to either IFNγ-treatment or tryptophan-depletion, the fluorescent signal increased in comparison to the control (Fig. 1g). Additionally, IFNγ-induced tGFP signal was negated by IDO1 inhibition (IDOi), excluding IFNγ- induced effects other than the induction of a tryptophan shortage (Fig. 1g). In contrast to tGFPF26W, neither IFNγ-treatment nor direct tryptophan depletion of cells containing a control tGFPF26A construct affected the green fluorescent signal (data not shown). We further expanded this analysis to other cell lines and observed a tryptophan- depletion-induced increase in tGFPF26W signal in the cancer cell lines HCT-116 (colorectal origin) and MDA-MB-231 (breast), indicating that the W>F substitution is neither a phenomenon restricted to MD55A3 cells nor to melanoma (data not shown). The observation that not all cell lines showed increased tGFPF26W signal by tryptophan depletion (i.e. the breast cancer MCF7, the retinal epithelial RPE-1, and the breast tissue MCF10A cell lines) may indicate for a regulatory process ((data not shown).

Compromised activity of tryptophanyl-tRNA synthetase (WARSI) is likely to account for phenylalanine misincorporation at tryptophan codons in the absence of tryptophan

The observed substitutions W93F, F76Y and Y99F (Fig. If), which are aromatic amino acids substituted for another aromatic amino acid, cannot be explained by codon/anticodon interactions. In contrast, it suggests an error in aminoacylation in the absence of the cognate depleted amino acid. To examine this issue, we tested the capacity of recombinant human tryptophanyl-tRNA synthetase 1 (WARSI), the enzyme that catalyzes the amino acylation of tRNATrp, to activate different amino acids (Yu et al., 2020. Biochim Biophys Acta Mol Cell Res 1868: 118889). WARSI was able to activate tryptophan and phenylalanine, while it did not possess such activity towards control amino acids such as serine, methionine, and glycine (Fig. Ih). This result indicates that phenylalanine is a potential substrate of WARSI in the absence of tryptophan, and suggests that compromised specificity of amino acid tRNA synthetases can be the source of this codon reassignment.

Proteome-wide tryptophan to phenylalanine substitutions following tryptophan shortage

Next, we enquired whether codon reassignments appear in endogenous proteins when cells are treated with IFNγ. We exposed MD55A3 melanoma cells to IFNγ, performed 2D LC/MS/MS in duplicate, and searched for W>F substitutions in the entire proteome. After stringent filtering for peptides with such substitutions, we identified a significantly higher number of W>F instances induced by IFNγ as compared to the control treatment (Table 2). The enrichment in the IFNγ-treated condition was unique to W>F substitutions, and was not observed for other tested substitutions (i.e. proline (P), alanine (A), cysteine (C), histidine (H), leucine (L), glutamine (Q), and tyrosine (Y), data not shown). Additionally, the observed W>F enrichment was significant (pval < 2.2e-16) in the background of reduced global expression of peptides in the whole proteome (due to enhanced proteolysis and reduced mRNA translation by IFNγ) (Fig. 2a). Further analysis of the proteins carrying the W>F substitutions indicated that they are neither highly expressed nor differentially expressed before or after IFNγ-treatment (data not shown), suggesting that this is a global phenomenon not directly associated to the abundance of proteins.

Similar to IFNγ, we observed enrichment of W>F substitutions when MD55A3 cells were depleted of tryptophan (Table 3). The enrichment of tryptophan- depletion-induced mis-incorporation was specific to W>F, and not observed for tryptophan to alanine (W>A), cysteine (W>C) and histidine (W>H) substitutions (data not shown). Furthermore, the W>F enrichment was specifically observed upon W-depletion and not upon Y-depletion (data not shown). The enrichment for tryptophan- depletion-induced W>F peptides (44 vs 4) is highly significant, considering the globally reduced protein expression (pval < 2.2e-16, Fig. 2b). Altogether, these data demonstrate that amino acid deprivations induce site- specific amino acid substitutions. We therefore named these inducible codon re assignments at the protein level as “substitutants” - to distinguish them from genetically encoded mutants.

Inspired by the proteome-wide detection of substitutants in IFNγ-induced tryptophan depletion melanoma cell lines, we next interrogated their appearance in other cell types. To this end, we analysed the glioblastoma cell lines (RA and HROG02) treated with IFNγ and similarly uncovered enrichment of W>F substitutants (Fig. 2c, d) (Forlani et al., 2021. Mol Cell Proteomics 20: 100032). Reassuringly, the enrichment was specific to W>F and was not detected for W>Y, W>A and W>H (Fig. 2c, d; Table 6). Altogether, our findings indicate that W>F substitutants appear in a wide variety of cancer cell lines following IFNγ treatments. This mis-incorporation allows mRNA translation to proceed in frame, occasionally compromising the quality and functionality of the synthesized proteins.

Interestingly, 11 W>F substitutants were detected both in tryptophan- depletion and IFNγ-treatment conditions of MD55A3 cells, confirming specificity (data not shown). This list included peptides from categories of proteins with different functionalities, such as RNA binding, ribosome constituents, and oxoreductase activity. Of particular interest, the W>F substitution observed in the Peptidylprolyl Isomerase A (PPIA) protein has already been reported to reduce its enzymatic activity. This W>F substitution leads to the loss of affinity of the protein for cyclosporin, causing loss of function (Davis et al., 2010 PLoS Biol 8: e 1000439; Liu et al., 1991. Biochemistry 30: 2306-2310). Also, the W>F substitution in YBX1 protein was shown to cause decreased binding to C5-methlycytosine (m5C) containing mRNAs leading to the decreased discrimination between m5C- containing and unmethlyated RNA (Chen et al., 2019. Nat Cell Biol 21: 978-990). A more global structural analysis of W>F substitutions in the detected proteins suggests a wide range of effects enforced by tryptophan shortage on protein activity and function (data not shown). This indicates that the depletion of tryptophan leads to the expression of proteins with substitutants that alter, reduce, or inactivate, their function. Additionally, some of the detected W>F peptides contained 2, and even 3 substitutants (e.g., Prolyl 3-Hydroxylase 3 (P3H3) (Tables 2, 3), indicating effective and iterative bypass of tryptophan “starved” codons. Detection of W>F substitutants in cancer proteom.es The results thus far suggest that W>F substitutants may appear in tumours characterized by high level of infiltrating T cells, IFNγ signaling, and IDO1 expression. To examine this comprehensively, we interrogated the proteomes of Clinical Proteomic Tumour Analysis Consortium (CPTAC) dataset (Edwards et al., 2015. J Proteome Res 14: 2707-2713). Initially, we examined squamous cell lung cancer (LSCC; Satpathy et al., 2021. Cell 184: 4348-4371), a large-scale collection of 205 samples (104 tumours and 101 adjacent normal tissues). Proteomics analysis with highly stringent filtering criteria uncovered a large number of W>F substitutants, which was significantly higher than any other substitutants (Fig. 3a). The detected W>F substitutions were expressed less consistently across samples as compared to any other substitutants (data not shown), suggesting variability in their expression across samples.

Since the LSCC proteomics dataset contained tumours as well as adjacent normal tissues, we next examined the relative enrichment of W>F substitutants and stratified by IDO1 expression. This analysis identified a significantly higher number of W>F substitutants in tumours as compared to adjacent normal tissues (Fig. 3b), suggesting that such regulated codon reassignment is a tumour-enriched phenomenon. Moreover, the association of W>F substitutants to IDO1 expression level was specific to tumours and not seen in normal tissues (Fig. 3b). In contrast, W>Y substitutants showed neither enrichment in number, nor association to tumours or with IDO1 expression (Fig. 3b). These analyses of LSCC proteomes indicate that W>F substitutions are significantly enriched in tumours and appear associated with IFNγ signalling, while other tryptophan substitutions appear minimal, possibly due to either technical or biological noise (e.g. misinterpretation of posttranslational modifications in mass spectrometry data, genetic mutations, transcriptional errors).

Intrigued by this result, we searched in an unbiased manner for gene expression signatures that expand W>F substitutants landscape in tumours relative to adjacent normal tissues. We calculated the number of substitutants in tumour samples with high (log intensity >0) and low (log intensity <0) expression of every gene. This analysis led to clustering of genes into three categories (Fig. 3c). The major cluster represents the vast majority of genes that do not contribute to the expansion of W>F substitutants landscape. The two other clusters represent genes whose expression is relatively high either in tumours with high W>F substitutants (W>F High), or with low expression of W>F substitutants (W>F Low) (Fig. 3c). Examination of ontologies enriched in W>F High cluster linked this type of substitutants with immune response, vesicular transport and peptide presentation (Fig. 3c). In contrast, ontologies related to chromatin, TP53 activity and SUMOlyation were enriched in the W>F Low cluster (Fig. 3c). Furthermore, adjacent normal tissues showed no clustering nor enrichment with gene-expression signature, indicating that W>F substitutants and their associated pathways are a tumour-specific phenomenon (Fig. 3c). Finally, a similar analysis of control W>Y substitutions detected no gene expression clusters, providing evidence for the specific appearance of W>F substitutants in human tumours (Fig. 3d). Gene Set Enrichment Analysis (GSEA) (Subramanian et al., 2005. Proc Natl Acad Sci U S A 102: 15545-155500), further substantiated the link between W>F substitutants and T-cell activation index, specifically in tumours and not in adjacent normal tissues (Fig.3e).

Next, we examined kinase-signaling pathways associated with high production of W>F substitutants in LSCC tumours by interrogating phospho- proteomic datasets (data not shown). This analysis linked the phosphorylation of p38-MAPK, BCR, EGFR and VEGF signalling pathways to the expression of W>F substitutants (Fig. 31). As expected, this association was specific to tumours and not observed in adjacent normal tissues (Fig. 31). Altogether, global tryptophan substitution analysis of an LSCC tumour panel indicated W>F as substitutants whose expression is linked to tumour infiltrating T cells and potentially regulated by oncogenic signalling pathways such as p38-MAPK and EGFR.

We then extended our proteomics analysis to 211 Pancreatic Ductal Adenocarcinoma (PDA) proteomes (Cao et al., 2021. Cell 184: 5031-5052) (137 Tumours, 74 adjacent normal tissues). This showed a more specific expression of W>F as compared to other substitutions (data not shown), and identified a link between W>F substitutions and immune response pathways in tumours but not in adjacent normal tissues (Fig. 3g, data not shown) that was confirmed by GSEA (Fig. 3h). Thus, an independent proteomics analysis of pancreatic cancer supported the expression of W>F substitutants in LSCC tumours with high infiltrating T cells.

Observing W>F substitutants in two independent tumour-types led us to complete the analysis of all available tumour-types present in CPTAC database, including Liver (HCC, 331 samples (Gao et al., 2019. Cell 179: 1240), Head and Neck (HNSCC, 158 Samples) (Huang et al., 2021. Cancer Cell 39: 361-379), Uterine carcinoma (UCEC (Dou et al., 2020. Cell 180: 729-748; 149 Samples), Breast cancer (BC, 108 Samples) (Mertins et al., 2016. Nature 534, 55-62), Glioblastoma (GBM, 111 samples; Wang et al., 2021. Cancer Cell 39: 509), Ovarian cancer (OV, 107 Samples; McDermott et al., 2020. Cell Rep Med 1: doi: 10.1016/j.xcrm.2020.100004; Hu et al., 2020. Cell Rep 33: 108276), Renal cancer (CCRCC, 194 Samples; Clark, D. J. et al. Integrated Proteogenomic Characterization of Clear Cell Renal Cell Carcinoma. Cell 179, 964-983 e931, doi: 10.1016/j. cell.2019.10.007 (2019). Clark et al., 2020. Cell 179: 964-983 e931), and Lung adenocarcinoma (LUAD, 217 Samples; Gillette et al., 2020. Cell 182: 200-225). Figure 3i shows that W>F is highly enriched (in comparison to average W-substitutants) in multiple tumour-types, and is significantly higher than W>Y in every case except LUAD. Thresholding for the fold-differences between enrichment of W>F and W>Y substitutants led to the classification of all tumour types but CCRCC and LUAD as W>F substitutants- enriched tumour types. Similar to LSCC and PDA, the expression of W>F substitutants in the W>F-enriched category was more sample specific than any other substitutants (data not shown) These results indicate that W>F substitutants is a widespread phenomenon in human cancer, but also that there are factors that impede their expression in certain tumour-types (Fig. 3i). These factors may be linked to tumour vascularization, efficient tryptophan supply by the microenvironment, and nutrient replenishment by autophagy.

Finally, to support the link between tumour immunological microenvironment and W>F expression we analysed a dataset of breast cancer tumour xenografts expanded in immuno- deficient mice (PDX, 27 samples). While breast cancer tumours in native microenvironment demonstrated enrichment of W>F substitutants, PDX samples failed to do so (Fig. 3i-j). This result indicates that the host tumour microenvironment and a compatible immune response is imperative for synthesis of W>F substitutants. This is corroborated by the association of T-cell activation pathway with W>F substitutants in all tumour- types except for GBM and OV (Fig. 3k, and data not shown).

Altogether, global analysis of several panels of cancer proteomes and phospho-proteomes exposed the specific appearance of W>F substitutants and uncovered molecular pathways associated with their efficient expression. Intra- tumour T cell activity, high level of IDO 1, and certain oncogenic signalling pathways are hence proposed to stimulate W>F substitutants.

Substitutants can be presented on the cell surface to provoke immune-specific recognition

Next, we tested whether the induction of W>F mis-incorporations following IFNγ-treatment can impact the landscape of antigens presented at the cell surface. For this purpose, we used the model peptide SIINFEKL from chicken ovalbumin that binds to mouse H2-Kb, and in this context can be recognized by a monoclonal anti-H2-Kb-bound SIINFEKL antibody in a flow cytometry assay (Dersh et al., 2019. Methods Mol Biol 1988: 109-122). The SIINFEKL peptide contains a phenylalanine residue at its center, and therefore can be exploited to examine the impact of W>F substitutants on the presentation of peptides at the cell surface. The H2-Kb binding affinity prediction using NetMHC4.0 server (available at cbs.dtu.dk/services/NetMHC/; Jurtz et al., 2017. J Immunol 199: 3360-3368) indicated that while SIINFEKL is a strong binder to H2-Kb with an affinity below 20nM, a peptide version with the phenylalanine substituted by a tryptophan (SIINwEKL) is a weak binder with more than 400 nM affinity (Fig. 4a). Thus, an F>W substitution of the SIINFEKL peptide is likely to reduce both cellular presentation and recognition by the anti-SIINFEKL antibody (Fig. 4b). We therefore engineered MD55A3 melanoma cells to express H2-Kb, and SIINFEKL/SIINwEKL/SIINaEKL/SIINFwKL downstream of V5-ATF4(1- 63;W93Y)-tGFP, (Fig. 4c). Immunoblot analysis with anti-tGFP antibodies reveals a comparable intracellular level of both protein products (Fig. 4d).

Flow cytometry analysis indicated that while SIINFEKL resulted in the expected antibody recognition at the cell surface of untreated cells, the SIINwEKL peptide did not (Fig. 4e). We therefore expected that if IFNγ-treatment results in W>F substitutions, this should lead to sufficient presentation of SIINFEKL peptides originating from a SIINwEKL-encoding DNA sequence, followed by efficient recognition by the anti H-2Kb-bound SIINFEKL antibody. Indeed, a significant SIINFEKL signal was detected in the IFNγ-treated SIINwEKL cells (Fig. 4e). This signal was negated by IDOi treatment, confirming the role of tryptophan depletion in this presentation process. In contrast, alternative genetically encoded SIINFEKL mutations could not be reverted at the peptide level and showed no signal (Fig. 4e). To examine whether the presentation of IFNγ- induced substitutants is sufficient to provoke a T cell reaction, we co-cultured MD55A3-H2-Kb-SIINFEKL and SIINwEKL cells with T cells derived from OT-1 mice (harbouring the anti-SIINFEKL transgenic T cell receptor; Dersh et al., 2019. Methods Mol Biol 1988: 109-122), and examined their reactivity by measuring positivity for intracellular IFNγ and TNF by flow cytometry. To prevent T-cell inactivation by IFNγ-induced kynurenine (Triplett et al., 2018. Nat Biotechnol 36: 758-764), we added Kynureninase to IFNγ-treated samples (Triplett et al., 2018. Nat Biotechnol 36: 758-764). While SIINFEKL-presenting cells provoked a response in ~20% of the T cells, SIINwEKL-producing cells did not induce any response. Importantly, treatment of MD55A3-H2-Kb-SIINwEKL cells with IFNγ led to T cell activation of ~5% of the T cells (Fig. 41). This effect was due to tryptophan depletion mediated by IFNγ-induced IDO1, as it was negated by the addition of IDOi to the treatment (Fig. 41). Similar but more potent results were obtained with HT29 colorectal cancer cell line. IFNγ-treated HT29-H2-Kb- SIINwEKL cells showed a high cell-surface recognition signal by anti-SIINFEKL staining and OT-1 T-cell reactivation, which was negated by IDOi -treatment and enhanced by a combined treatment of IFNγ with tryptophan depletion (Fig. 4g, h). Importantly, while SIINwEKL presentation provoked a weak T-cell recognition and killing (~7% and ~35%, respectively, Fig. 4h,i), the extent of SIINwEKL>SIINFEKL substitution of cells pre-treated with IFNγ and tryptophan depletion (~30%, Fig. 4h) was sufficient to stimulate efficient T-cell killing (~80%, Fig. 4i). Thus, W>F substitutants produced by tryptophan depletion can induce specific alterations in antigen presentation at the cell surface that can lead to potent T cell recognition, activation and killing of cancer cells.

IFNy-induced W>F substitutants identified by inimunopeptidoniics

Inspired by the SIINwEKL experiment, we next sought to examine the presentation of endogenous peptides with W>F substitutions in a high-throughput manner. First, we searched for the appearance of W>F substitutants in a published immunopeptidomics dataset of four microsatellite stable (MSS) colorectal cancer (CRC) organoids treated or not with IFNγ for 48 hrs (Newey et al., 2019. J Immunother Cancer 7: 309). It was reported that IFNγ treatment did not expand the neoantigen landscape of these CRC organoids (Newey et al. 2019. J Immunother Cancer 7: 309). However, we observed IFNγ-treatment-specific W>F substitutions in each of the four organoids (data not shown). Specifically, after filtration, we observed 41 W>F substitution events across organoids upon IFNγ treatment, as opposed to only 4 in control conditions (Table 4). In contrast, a search for W>Y, W>N, and W>A, substitutions revealed no detectable peptides, vouching for the specificity of IFNγ-mediated W>F substitutant presentation (data not shown). The variability in the identity of the identified W>F substitutant- peptides across organoids can be explained by the diversity of HLA-I molecules expressed in each of them. Of interest, organoids number 4 and 8 have several HLA-I molecules in common (HLA-A*03:01, C*04:01 and C*05:01) and share one W>F substitutant peptide originating from the RPS18 gene (KIPDfFLNR where f marks the W>F substitutant of RPS18). This substitutant reproducibly appeared in all ten biological replicates of these two organoids. Even though from different tissues (colorectal vs melanoma) and different experimental setups (proteomics vs peptidomics), the same W>F substitutant in the RPS18 protein was observed in the proteomics analysis of melanoma MD55A3 cells after IFNγ treatment (Fig. 2a, QYKIPDfFLNR, where f marks the W>F substitutant of RPS18). Thus, our analyses point to the specificity by which W>F substitutants are generated upon IFNγ treatment, and to the role of HLA-I alleles in their presentation at the cell surface.

Second, we employed an immunopeptidomics protocol with a pan-anti-HLA-I antibody, and searched for the appearance of W>F substitutants in the glioblastoma cell line (RA) in which we previously detected their enrichment following 48 hr of IFNγ-treatment (Fig. 2c, d). Indeed, specific enrichment of W>F, but not W>A, >V, >Y, >H, >P, >G, >Q, >S or >T, substitutions appeared in the immunopeptidome of RA cells following IFNγ treatment (Tables 5 and 6; data not shown). The identification of five substitutant peptides was validated with targeted mass spectrometry analysis where synthetic standard peptides labelled with stable heavy isotopes were spiked into newly generated samples of HLA bound peptides eluted from RA cells treated with IFNγ (data not shown). We then enquired whether these substitutant peptides can be immunogenic. Based on HLA- restriction (RA expresses HLA-A*24:02) we selected 6 substitutant peptides from the 21 identified in our RA immunopeptidomics analysis and asked whether they can prime naive CD8+ T cells isolated from healthy HLA-A*24:02pos donors (Ali et al., 2019. Nat Protoc 14: 1926-1943). Monocyte -derived dendritic cells isolated from peripheral blood mononuclear cells (PBMCs) from healthy individuals were pulsed with substitutant-peptides and co-cultured with autologous naive CD8+ T cells. After co-culture, combinatorial peptide-MHC multimer staining analysed by flow cytometry showed T cells reactivity towards the two substitutant KLHL4 and Bl1 derived-pep tides that have been confirmed by the targeted MS analysis (Fig. 4j; Table 5), demonstrating immunogenicity.

Example 2

Materials and methods

As in Example 1.

Results

Broad anti-cancer activity mediated by a TCR T cell against a single neoepitope

To demonstrate the concept of targeting specific neoepitopes, we focused on one substitutant epitope (sub#l) identified by immunopeptidomics in a glioblastoma cell line (RA) expressing an HLA*A24:02 haplotype (Figure 5A). As a first step, we screened PBMCs and identified reactive T cells (Figure 5B). We subsequently sequenced the TCRs and performed activity assays. These tests uncovered one TCR with high affinity to sub#l (TCRsub#1) and high selectivity compared with the corresponding wild-type epitope (Figure 5C). Furthermore, in co-culture experiments, T cells armed with TCRsub#1 required both the host target gene expression and IDO1 activity to elicit effective cancer cell recognition and killing following IFNγ treatment (Figure 5D and E). This function of TCRsub#l was broad, provided that the target cancer cells expressed HLA*A24:02 and that the peptide presentation pathway was intact (Figure 6).

Table 2. Endogenous IFN-induced W>F substitutant peptides detected in massspectrometry. Two independent experiments. Ctr: control treatment. IFN: interferon gamma treatment.

Table 3. Endogenous amino acid depletion-induced W>F substitutant peptides detected in massspectrometry. Two independent experiments. Ctr: control treatment. MinusW: depletion of tryptophan. MinusY: depletion of tyrosine.

Table 4. Endogenous amino acid depletion-induced W>F substitutant peptides detected in immunopeptidomics of Colorectal Cancer organoids.

Table 5. Endogenous tryptophan depletion-induced W>F substitutant peptides detected in immunopeptidomics of Glioblastoma (RA) cells. RA_IFN(l-4) denote glioblastoma cell line RA treated with interferon gamma.

*: Reactive T cells in PBMCs of healthy individuals were identified against these peptides.

Table 6.

A. Endogenous tryptophan depletion-induced W>F substitutant peptides detected in immunopeptidomics of Glioblastoma (RA).

B. Endogenous tryptophan depletion-induced W>F substitutant peptides detected in proteomics of Glioblastoma (HR0G2).

C. Endogenous tryptophan depletion-induced W>F substitutant peptides detected in proteomics of cancer cells.

* HepG2, hepatocellular carcinoma; RA, glioblastoma; PC-3, prostate cancer; H1299, PC9, non-small cell lung carcinoma; HT29, Colo320 colon cancer; MDA-MB-231, breast cancer; MiaPACA2, pancreatic cancer.

Claims

1. A T cell epitope comprising 8-22 amino acid residues, more preferred 8-13 amino acid residues, of a cellular protein, said epitope comprising a phenylalanine residue that is not encoded by the cell's genome, whereby said cellular protein preferably is a protein selected from any one of Tables 2-6 having the indicated W>F amino acid alteration.

2. The T cell epitope according to claim 1, which is selected from the proteins listed in Tables 2-3.

3. A polyepitope, comprising 2-50, preferably 5-25 individual T cell epitopes according to claim 1 or 2, which individual epitopes may be alternated by spacer sequences of, preferably, 1-10 amino acid residues.

4. A nucleic acid molecule, encoding the T cell epitope of claim 1 or 2, or the polyepitope of claim 3, said nucleic acid molecule preferably being a RNA molecule that expresses said epitope or polyepitope upon delivery into a suitable cell.

5. A T-cell receptor (TCR) that specifically recognizes the T cell epitope according to claim 1 or 2, or the polyepitope according to claim 3, preferably wherein the TCR is expressed by a T-cell.

6. A method of inducing an immune response in an individual, said method comprising providing said individual with the T cell epitope of claim 1 or 2, the polyepitope of claim 3, the nucleic acid molecule of claim 4, or a combination thereof.

7. A method of treating an individual suffering from a tumor, comprising providing said individual with the T cell epitope of claim 1 or 2, the polyepitope of claim 3, the nucleic acid molecule of claim 4, the TCR according to claim 5, or a combination thereof.

8. The method according to claim 6 or 7, wherein said individual comprises a cell, especially a tumor cell, that expresses the protein comprising a phenylalanine that is not encoded by the cell's genome, wherein a tryptophan residue is encoded by the cell's genome at the position of said phenylalanine in the protein, and wherein said protein preferably is selected from any one of Tables 2-6 having the indicated W>F amino acid alteration.

9. The method according to any one of claims 6-8, wherein said individual comprises a glioblastoma cell, prostate cancer cell, pancreatic cancer cell, non-small cell lung carcinoma cell, melanoma cell, breast cancer cell, or a colorectal cancer cell.

10. The method of any one of claims 6-9, wherein said individual is further provided with interferon gamma, an immune checkpoint inhibitor, or both, whereby said interferon gamma and/or immune checkpoint inhibitor may be administered prior to, simultaneously with, or following administration of the T cell epitope of claim 1 or 2, the polyepitope of claim 3, the nucleic acid molecule of claim 4, the TCR according to claim 5, or combination thereof.

11. A pharmaceutical composition, comprising the T cell epitope of claim 1 or 2, the polyepitope of claim 3, the nucleic acid molecule of claim 4, the TCR according to claim 5, or a combination thereof.

12. The pharmaceutical composition according to claim 11, further comprising means for reducing the amount of tryptophan in a cell, an immune checkpoint inhibitor, or both.

13. The pharmaceutical composition according to claim 11 or 12, further comprising an accessory molecule such as an adjuvant, an immune stimulating molecule such as a chemokine and/or a cytokine, or a combination thereof.

14. A method of producing and identifying a human cellular protein comprising a phenylalanine residue that is not encoded by the cell's genome, said method comprising reducing the amount of tryptophan in a cell, thereby producing a protein comprising said phenylalanine residue that is not encoded by the cell's genome, said method further comprising identifying said protein comprising said phenylalanine residue that is not encoded by the cell's genome, whereby a tryptophan residue is encoded by the cell's genome at the position of said phenylalanine residue, wherein said protein is selected from any one of Table 2-6 having the indicated amino acid alteration.

15. A method of producing and identifying a human cellular protein comprising a phenylalanine residue is not encoded by the cell's genome, said method comprising reducing the amount of tryptophan in a cell, thereby producing a protein comprising said phenylalanine residue that is not encoded by the cell's genome, said method further comprising identifying said protein comprising said phenylalanine residue that is not encoded by the cell's genome, whereby a tryptophan residue is encoded by the cell's genome at the position of said phenylalanine residue.

16. The method of claim 14 or 15, wherein the amount of tryptophan is reduced in said cell by providing a growth medium that is depleted of tryptophan, by incubating the cells in the presence of interferon gamma, by activating indoleamine 2, 3- dioxygenase 1 (IDO1), indoleamine 2, 3-dioxygenase 2 (IDO2), and/or tryptophan 2, 3-dioxygenase (TDO) in the cells, or by a combination thereof.

17. The method of any one of claims 14-16, wherein the cell presents a peptide of 8-22 amino acid residues of the protein by MHC on the surface of said cell, preferably a peptide of 8-13 amino acid residues that is presented by MHC class I, said peptide comprising a phenylalanine residue at a position at which a tryptophan residue is encoded by the cell's genome.

18. The method of any one of claims 14-17, wherein the cell is a tumor cell such as a glioblastoma cell, prostate cancer cell, pancreatic cancer cell, non-small cell lung carcinoma cell, melanoma cell, breast cancer cell, or colorectal cancer cell. 72

19. A method of identifying a cellular protein comprising a phenylalanine residue that is not encoded by the cell's genome, said method comprising: providing a cell; reducing the amount of tryptophan in said cell; and identifying a protein comprising a phenylalanine residue that is not encoded by the cell's genome, preferably by identifying a peptide of 8-22 amino acid residues that is presented by MHC on the surface of said cell, whereby said phenylalanine residue is not encoded by the cell's genome, but a tryptophan residue is encoded by the cell's genome at the position of said phenylalanine in the protein, wherein said protein preferably is selected from any one of Tables 2-6.