Graph-pMHC: graph neural network approach to MHC class II peptide presentation and antibody immunogenicity.

Antigen presentation on MHC class II (pMHCII presentation) plays an essential role in the adaptive immune response to extracellular pathogens and cancerous cells. But it can also reduce the efficacy of large-molecule drugs by triggering an anti-drug response. Significant progress has been made in pMHCII presentation modeling due to the collection of large-scale pMHC mass spectrometry datasets (ligandomes) and advances in machine learning. Here, we develop graph-pMHC, a graph neural network approach to predict pMHCII presentation. We derive adjacency matrices for pMHCII using Alphafold2-multimer and address the peptide-MHC binding groove alignment problem with a simple graph enumeration strategy. We demonstrate that graph-pMHC dramatically outperforms methods with suboptimal inductive biases, such as the multilayer-perceptron-based NetMHCIIpan-4.0 (+20.17% absolute average precision). Finally, we create an antibody drug immunogenicity dataset from clinical trial data and develop a method for measuring anti-antibody immunogenicity risk using pMHCII presentation models. Our model increases receiver operating characteristic curve (ROC)-area under the ROC curve (AUC) by 2.57% compared to just filtering peptides by hits in OASis alone for predicting antibody drug immunogenicity.

Tumor-intrinsic expression of the autophagy gene Atg16l1 suppresses anti-tumor immunity in colorectal cancer.

Microsatellite-stable colorectal cancer (MSS-CRC) is highly refractory to immunotherapy. Understanding tumor-intrinsic determinants of immunotherapy resistance is critical to improve MSS-CRC patient outcomes. Here, we demonstrate that high tumor expression of the core autophagy gene ATG16L1 is associated with poor clinical response to anti-PD-L1 therapy in KRAS-mutant tumors from IMblaze370 (NCT02788279), a large phase III clinical trial of atezolizumab (anti-PD-L1) in advanced metastatic MSS-CRC. Deletion of Atg16l1 in engineered murine colon cancer organoids inhibits tumor growth in primary (colon) and metastatic (liver and lung) niches in syngeneic female hosts, primarily due to increased sensitivity to IFN-γ-mediated immune pressure. ATG16L1 deficiency enhances programmed cell death of colon cancer organoids induced by IFN-γ and TNF, thus increasing their sensitivity to host immunity. In parallel, ATG16L1 deficiency reduces tumor stem-like populations in vivo independently of adaptive immune pressure. This work reveals autophagy as a clinically relevant mechanism of immune evasion and tumor fitness in MSS-CRC and provides a rationale for autophagy inhibition to boost immunotherapy responses in the clinic.

Pattern recognition receptor agonists in pathogen vaccines mediate antitumor T-cell cross-priming.

BACKGROUND: Cancer immunotherapies are generally effective in patients whose tumors contain a priori primed T-cells reactive to tumor antigens (TA). One approach to prime TA-reactive T-cells is to administer immunostimulatory molecules, cells, or pathogens directly to the tumor site, that is, in situ vaccination (ISV). We recently described an ISV using Flt3L to expand and recruit dendritic cells (DC), radiotherapy to load DC with TA, and pattern recognition receptor agonists (PRRa) to activate TA-loaded DC. While ISV trials using synthetic PRRa have yielded systemic tumor regressions, the optimal method to activate DCs is unknown. METHODS: To discover optimal DC activators and increase access to clinical grade reagents, we assessed whether viral or bacterial components found in common pathogen vaccines are an effective source of natural PRRa (naPRRa). Using deep profiling (155-metric) of naPRRa immunomodulatory effects and gene editing of specific PRR, we defined specific signatures and molecular mechanisms by which naPRRa potentiate T-cell priming. RESULTS: We observed that vaccine naPRRa can be even more potent in activating Flt3L-expanded murine and human DCs than synthetic PRRa, promoting cross-priming of TA-reactive T-cells. We developed a mechanistically diverse naPRRa combination (BCG, PedvaxHIB, Rabies) and noted more potent T-cell cross-priming than with any single naPRRa. The naPRRa triplet-as part of Flt3L-primed ISV-induced greater intratumoral CD8 T-cell infiltration, T-cells reactive to a newly defined tumorous neoantigen, durable tumor regressions. CONCLUSIONS: This work provides rationale for the translation of pathogen vaccines as FDA-approved clinical-grade DC activators which could be exploited as immune-stimulants for early phase trials.

The peptide woods are lovely, dark and deep: Hunting for novel cancer antigens.

Harnessing the patient's immune system to control a tumor is a proven avenue for cancer therapy. T cell therapies as well as therapeutic vaccines, which target specific antigens of interest, are being explored as treatments in conjunction with immune checkpoint blockade. For these therapies, selecting the best suited antigens is crucial. Most of the focus has thus far been on neoantigens that arise from tumor-specific somatic mutations. Although there is clear evidence that T-cell responses against mutated neoantigens are protective, the large majority of these mutations are not immunogenic. In addition, most somatic mutations are unique to each individual patient and their targeting requires the development of individualized approaches. Therefore, novel antigen types are needed to broaden the scope of such treatments. We review high throughput approaches for discovering novel tumor antigens and some of the key challenges associated with their detection, and discuss considerations when selecting tumor antigens to target in the clinic.

Expanding cross-presenting dendritic cells enhances oncolytic virotherapy and is critical for long-term anti-tumor immunity.

Immunotherapies directly enhancing anti-tumor CD8+ T cell responses have yielded measurable but limited success, highlighting the need for alternatives. Anti-tumor T cell responses critically depend on antigen presenting dendritic cells (DC), and enhancing mobilization, antigen loading and activation of these cells represent an attractive possibility to potentiate T cell based therapies. Here we show that expansion of DCs by Flt3L administration impacts in situ vaccination with oncolytic Newcastle Disease Virus (NDV). Mechanistically, NDV activates DCs and sensitizes them to dying tumor cells through upregulation of dead-cell receptors and synergizes with Flt3L to promote anti-tumor CD8+ T cell cross-priming. In vivo, Flt3L-NDV in situ vaccination induces parallel amplification of virus- and tumor-specific T cells, including CD8+ T cells reactive to newly-described neoepitopes, promoting long-term tumor control. Cross-presenting conventional Type 1 DCs are indispensable for the anti-tumor, but not anti-viral, T cell response, and type I IFN-dependent CD4+ Th1 effector cells contribute to optimal anti-tumor immunity. These data demonstrate that mobilizing DCs to increase tumor antigen cross-presentation improves oncolytic virotherapy and that neoepitope-specific T cells can be induced without individualized, ex vivo manufactured vaccines.

Molecular determinants of response to PD-L1 blockade across tumor types.

Immune checkpoint inhibitors targeting the PD-1/PD-L1 axis lead to durable clinical responses in subsets of cancer patients across multiple indications, including non-small cell lung cancer (NSCLC), urothelial carcinoma (UC) and renal cell carcinoma (RCC). Herein, we complement PD-L1 immunohistochemistry (IHC) and tumor mutation burden (TMB) with RNA-seq in 366 patients to identify unifying and indication-specific molecular profiles that can predict response to checkpoint blockade across these tumor types. Multiple machine learning approaches failed to identify a baseline transcriptional signature highly predictive of response across these indications. Signatures described previously for immune checkpoint inhibitors also failed to validate. At the pathway level, significant heterogeneity is observed between indications, in particular within the PD-L1+ tumors. mUC and NSCLC are molecularly aligned, with cell cycle and DNA damage repair genes associated with response in PD-L1- tumors. At the gene level, the CDK4/6 inhibitor CDKN2A is identified as a significant transcriptional correlate of response, highlighting the association of non-immune pathways to the outcome of checkpoint blockade. This cross-indication analysis reveals molecular heterogeneity between mUC, NSCLC and RCC tumors, suggesting that indication-specific molecular approaches should be prioritized to formulate treatment strategies.

An Integrated Genomic, Proteomic, and Immunopeptidomic Approach to Discover Treatment-Induced Neoantigens.

All nucleated mammalian cells express major histocompatibility complex (MHC) proteins that present peptides on cell surfaces for immune surveillance. These MHC-presented peptides (pMHC) are necessary for directing T-cell responses against cells harboring non-self antigens derived from pathogens or from somatic mutations. Alterations in tumor-specific antigen repertoires - particularly novel MHC presentation of mutation-bearing peptides (neoantigens) - can be potent targets of anti-tumor immune responses. Here we employed an integrated genomic and proteomic antigen discovery strategy aimed at measuring how interferon gamma (IFN-γ) alters antigen presentation, using a human lymphoma cell line, GRANTA-519. IFN-γ treatment resulted in 126 differentially expressed proteins (2% of all quantified proteins), which included components of antigen presentation machinery and interferon signaling pathways, and MHC molecules themselves. In addition, several proteasome subunits were found to be modulated, consistent with previous reports of immunoproteasome induction by IFN-γ exposure. This finding suggests that a modest proteomic response to IFN-γ could create larger alteration to cells' antigen/epitope repertoires. Accordingly, MHC immunoprecipitation followed by mass spectrometric analysis of eluted peptide repertoires revealed exclusive signatures of IFN-γ induction, with 951 unique peptides reproducibly presented by MHC-I and 582 presented by MHC-II. Furthermore, an additional set of pMHCs including several candidate neoantigens, distinguished control and the IFN-γ samples by their altered relative abundances. Accordingly, we developed a classification system to distinguish peptides which are differentially presented due to altered expression from novel peptides resulting from changes in antigen processing. Taken together, these data demonstrate that IFN-γ can re-shape antigen repertoires by identity and by abundance. Extending this approach to models with greater clinical relevance could help develop strategies by which immunopeptide repertoires are intentionally reshaped to improve endogenous or vaccine-induced anti-tumor immune responses and potentially anti-viral immune responses.

Antigen presentation in cancer: insights into tumour immunogenicity and immune evasion.

Immune checkpoint blockade, which blocks inhibitory signals of T cell activation, has shown tremendous success in treating cancer, although success still remains limited to a fraction of patients. To date, clinically effective CD8+ T cell responses appear to target predominantly antigens derived from tumour-specific mutations that accumulate in cancer, also called neoantigens. Tumour antigens are displayed on the surface of cells by class I human leukocyte antigens (HLA-I). To elicit an effective antitumour response, antigen presentation has to be successful at two distinct events: first, cancer antigens have to be taken up by dendritic cells (DCs) and cross-presented for CD8+ T cell priming. Second, the antigens have to be directly presented by the tumour for recognition by primed CD8+ T cells and killing. Tumours exploit multiple escape mechanisms to evade immune recognition at both of these steps. Here, we review the tumour-derived factors modulating DC function, and we summarize evidence of immune evasion by means of quantitative modulation or qualitative alteration of the antigen repertoire presented on tumours. These mechanisms include modulation of antigen expression, HLA-I surface levels, alterations in the antigen processing and presentation machinery in tumour cells. Lastly, as complete abrogation of antigen presentation can lead to natural killer (NK) cell-mediated tumour killing, we also discuss how tumours can harbour antigen presentation defects and still evade NK cell recognition.

In silico tools for accurate HLA and KIR inference from clinical sequencing data empower immunogenetics on individual-patient and population scales.

Immunogenetic variation in humans is important in research, clinical diagnosis and increasingly a target for therapeutic intervention. Two highly polymorphic loci play critical roles, namely the human leukocyte antigen (HLA) system, which is the human version of the major histocompatibility complex (MHC), and the Killer-cell immunoglobulin-like receptors (KIR) that are relevant for responses of natural killer (NK) and some subsets of T cells. Their accurate classification has typically required the use of dedicated biological specimens and a combination of in vitro and in silico efforts. Increased availability of next generation sequencing data has led to the development of ancillary computational solutions. Here, we report an evaluation of recently published algorithms to computationally infer complex immunogenetic variation in the form of HLA alleles and KIR haplotypes from whole-genome or whole-exome sequencing data. For both HLA allele and KIR gene typing, we identified tools that yielded >97% overall accuracy for four-digit HLA types, and >99% overall accuracy for KIR gene presence, suggesting the readiness of in silico solutions for use in clinical and high-throughput research settings.

Cancer neoantigens and immunogenicity: mutation position matters.

Cancer mutations can elicit protective immunity. Computational methods are critical for selecting these neoantigens for immunotherapy. While significant progress has been made in the field in predicting peptide presentation, our understanding of which mutated peptide is recognized as foreign by T cells remains limited. We used mouse vaccination studies to examine the features of immunogenic neoantigens and demonstrated that the mutation position is an important criterion for predicting neoantigens.

Mutation position is an important determinant for predicting cancer neoantigens.

Tumor-specific mutations can generate neoantigens that drive CD8 T cell responses against cancer. Next-generation sequencing and computational methods have been successfully applied to identify mutations and predict neoantigens. However, only a small fraction of predicted neoantigens are immunogenic. Currently, predicted peptide binding affinity for MHC-I is often the major criterion for prioritizing neoantigens, although little progress has been made toward understanding the precise functional relationship between affinity and immunogenicity. We therefore systematically assessed the immunogenicity of peptides containing single amino acid mutations in mouse tumor models and divided them into two classes of immunogenic mutations. The first comprises mutations at a nonanchor residue, for which we find that the predicted absolute binding affinity is predictive of immunogenicity. The second involves mutations at an anchor residue; here, predicted relative affinity (compared with the WT counterpart) is a better predictor. Incorporating these features into an immunogenicity model significantly improves neoantigen ranking. Importantly, these properties of neoantigens are also predictive in human datasets, suggesting that they can be used to prioritize neoantigens for individualized neoantigen-specific immunotherapies.

Transposable element expression in tumors is associated with immune infiltration and increased antigenicity.

Profound global loss of DNA methylation is a hallmark of many cancers. One potential consequence of this is the reactivation of transposable elements (TEs) which could stimulate the immune system via cell-intrinsic antiviral responses. Here, we develop REdiscoverTE, a computational method for quantifying genome-wide TE expression in RNA sequencing data. Using The Cancer Genome Atlas database, we observe increased expression of over 400 TE subfamilies, of which 262 appear to result from a proximal loss of DNA methylation. The most recurrent TEs are among the evolutionarily youngest in the genome, predominantly expressed from intergenic loci, and associated with antiviral or DNA damage responses. Treatment of glioblastoma cells with a demethylation agent results in both increased TE expression and de novo presentation of TE-derived peptides on MHC class I molecules. Therapeutic reactivation of tumor-specific TEs may synergize with immunotherapy by inducing inflammation and the display of potentially immunogenic neoantigens.

Late-differentiated effector neoantigen-specific CD8+ T cells are enriched in peripheral blood of non-small cell lung carcinoma patients responding to atezolizumab treatment.

BACKGROUND: There is strong evidence that immunotherapy-mediated tumor rejection can be driven by tumor-specific CD8+ T cells reinvigorated to recognize neoantigens derived from tumor somatic mutations. Thus, the frequencies or characteristics of tumor-reactive, mutation-specific CD8+ T cells could be used as biomarkers of an anti-tumor response. However, such neoantigen-specific T cells are difficult to reliably identify due to their low frequency in peripheral blood and wide range of potential epitope specificities. METHODS: Peripheral blood mononuclear cells (PBMC) from 14 non-small cell lung cancer (NSCLC) patients were collected pre- and post-treatment with the anti-PD-L1 antibody atezolizumab. Using whole exome sequencing and RNA sequencing we identified tumor neoantigens that are predicted to bind to major histocompatibility complex class I (MHC-I) and utilized mass cytometry, together with cellular 'barcoding', to profile immune cells from patients with objective response to therapy (n = 8) and those with progressive disease (n = 6). In parallel, a highly-multiplexed combinatorial tetramer staining was used to screen antigen-specific CD8+ T cells in peripheral blood for 782 candidate tumor neoantigens and 71 known viral-derived control peptide epitopes across all patient samples. RESULTS: No significant treatment- or response associated phenotypic difference were measured in bulk CD8+ T cells. Multiplexed peptide-MHC multimer staining detected 20 different neoantigen-specific T cell populations, as well as T cells specific for viral control antigens. Not only were neoantigen-specific T cells more frequently detected in responding patients, their phenotypes were also almost entirely distinct. Neoantigen-specific T cells from responder patients typically showed a differentiated effector phenotype, most like Cytomegalovirus (CMV) and some types of Epstein-Barr virus (EBV)-specific CD8+ T cells. In contrast, more memory-like phenotypic profiles were observed for neoantigen-specific CD8+ T cells from patients with progressive disease. CONCLUSION: This study demonstrates that neoantigen-specific T cells can be detected in peripheral blood in non-small cell lung cancer (NSCLC) patients during anti-PD-L1 therapy. Patients with an objective response had an enrichment of neoantigen-reactive T cells and these cells showed a phenotype that differed from patients without a response. These findings suggest the ex vivo identification, characterization, and longitudinal follow-up of rare tumor-specific differentiated effector neoantigen-specific T cells may be useful in predicting response to checkpoint blockade. TRIAL REGISTRATION: POPLAR trial NCT01903993 .

Regulation of Tumor-Associated Myeloid Cell Activity by CBP/EP300 Bromodomain Modulation of H3K27 Acetylation.

Myeloid-derived suppressor cells (MDSCs) are found in most cancer malignancies and support tumorigenesis by suppressing immunity and promoting tumor growth. Here we identify the bromodomain (BRD) of CBP/EP300 as a critical regulator of H3K27 acetylation (H3K27ac) in MDSCs across promoters and enhancers of pro-tumorigenic target genes. In preclinical tumor models, in vivo administration of a CBP/EP300-BRD inhibitor (CBP/EP300-BRDi) alters intratumoral MDSCs and attenuates established tumor growth in immunocompetent tumor-bearing mice, as well as in MDSC-dependent xenograft models. Inhibition of CBP/EP300-BRD redirects tumor-associated MDSCs from a suppressive to an inflammatory phenotype through downregulation of STAT pathway-related genes and inhibition of Arg1 and iNOS. Similarly, CBP/EP300-BRDi decreases differentiation and suppressive function of human MDSCs in vitro. Our findings uncover a role of CBP/EP300-BRD in intratumoral MDSCs that may be targeted therapeutically to boost anti-tumor immunity.

Publisher Correction: Clinical activity and molecular correlates of response to atezolizumab alone or in combination with bevacizumab versus sunitinib in renal cell carcinoma.

In the version of this article originally published, there was an error in Fig. 2n. The top line of the HR comparison chart originally was Atezo + bev vs sun. It should have been Atezo + bev vs atezo. The error has been corrected in the HTML and PDF versions of this article.

Clinical activity and molecular correlates of response to atezolizumab alone or in combination with bevacizumab versus sunitinib in renal cell carcinoma.

We describe results from IMmotion150, a randomized phase 2 study of atezolizumab (anti-PD-L1) alone or combined with bevacizumab (anti-VEGF) versus sunitinib in 305 patients with treatment-naive metastatic renal cell carcinoma. Co-primary endpoints were progression-free survival (PFS) in intent-to-treat and PD-L1+ populations. Intent-to-treat PFS hazard ratios for atezolizumab + bevacizumab or atezolizumab monotherapy versus sunitinib were 1.0 (95% confidence interval (CI), 0.69-1.45) and 1.19 (95% CI, 0.82-1.71), respectively; PD-L1+ PFS hazard ratios were 0.64 (95% CI, 0.38-1.08) and 1.03 (95% CI, 0.63-1.67), respectively. Exploratory biomarker analyses indicated that tumor mutation and neoantigen burden were not associated with PFS. Angiogenesis, T-effector/IFN-γ response, and myeloid inflammatory gene expression signatures were strongly and differentially associated with PFS within and across the treatments. These molecular profiles suggest that prediction of outcomes with anti-VEGF and immunotherapy may be possible and offer mechanistic insights into how blocking VEGF may overcome resistance to immune checkpoint blockade.

TGFß attenuates tumour response to PD-L1 blockade by contributing to exclusion of T cells.

Therapeutic antibodies that block the programmed death-1 (PD-1)-programmed death-ligand 1 (PD-L1) pathway can induce robust and durable responses in patients with various cancers, including metastatic urothelial cancer. However, these responses only occur in a subset of patients. Elucidating the determinants of response and resistance is key to improving outcomes and developing new treatment strategies. Here we examined tumours from a large cohort of patients with metastatic urothelial cancer who were treated with an anti-PD-L1 agent (atezolizumab) and identified major determinants of clinical outcome. Response to treatment was associated with CD8+ T-effector cell phenotype and, to an even greater extent, high neoantigen or tumour mutation burden. Lack of response was associated with a signature of transforming growth factor ß (TGFß) signalling in fibroblasts. This occurred particularly in patients with tumours, which showed exclusion of CD8+ T cells from the tumour parenchyma that were instead found in the fibroblast- and collagen-rich peritumoural stroma; a common phenotype among patients with metastatic urothelial cancer. Using a mouse model that recapitulates this immune-excluded phenotype, we found that therapeutic co-administration of TGFß-blocking and anti-PD-L1 antibodies reduced TGFß signalling in stromal cells, facilitated T-cell penetration into the centre of tumours, and provoked vigorous anti-tumour immunity and tumour regression. Integration of these three independent biological features provides the best basis for understanding patient outcome in this setting and suggests that TGFß shapes the tumour microenvironment to restrain anti-tumour immunity by restricting T-cell infiltration.

SUV420H2 is an epigenetic regulator of epithelial/mesenchymal states in pancreatic cancer.

Epithelial-to-mesenchymal transition is implicated in metastasis, where carcinoma cells lose sessile epithelial traits and acquire mesenchymal migratory potential. The mesenchymal state is also associated with cancer stem cells and resistance to chemotherapy. It might therefore be therapeutically beneficial to promote epithelial identity in cancer. Because large-scale cell identity shifts are often orchestrated on an epigenetic level, we screened for candidate epigenetic factors and identified the histone methyltransferase SUV420H2 (KMT5C) as favoring the mesenchymal identity in pancreatic cancer cell lines. Through its repressive mark H4K20me3, SUV420H2 silences several key drivers of the epithelial state. Its knockdown elicited mesenchymal-to-epithelial transition on a molecular and functional level, and cells displayed decreased stemness and increased drug sensitivity. An analysis of human pancreatic cancer biopsies was concordant with these findings, because high levels of SUV420H2 correlated with a loss of epithelial characteristics in progressively invasive cancer. Together, these data indicate that SUV420H2 is an upstream epigenetic regulator of epithelial/mesenchymal state control.

Repression of Stress-Induced LINE-1 Expression Protects Cancer Cell Subpopulations from Lethal Drug Exposure.

Maintenance of phenotypic heterogeneity within cell populations is an evolutionarily conserved mechanism that underlies population survival upon stressful exposures. We show that the genomes of a cancer cell subpopulation that survives treatment with otherwise lethal drugs, the drug-tolerant persisters (DTPs), exhibit a repressed chromatin state characterized by increased methylation of histone H3 lysines 9 and 27 (H3K9 and H3K27). We also show that survival of DTPs is, in part, maintained by regulators of H3K9me3-mediated heterochromatin formation and that the observed increase in H3K9me3 in DTPs is most prominent over long interspersed repeat element 1 (LINE-1). Disruption of the repressive chromatin over LINE-1 elements in DTPs results in DTP ablation, which is partially rescued by reducing LINE-1 expression or function.

Characterizing neoantigens for personalized cancer immunotherapy.

Somatic mutations can generate neoantigens that are presented on MHC molecules and drive effective T cells responses against cancer. Mutation load in cancer patients predicts response to immune checkpoint blockade therapy. Additionally, vaccination targeting neoantigens controls established tumor growth in preclinical models. These recent findings led to a renewed interest in the field of cancer vaccines and the development of antigen-targeted cancer immunotherapies. However, targeting neoantigens is challenging, as most mutations are unique to each cancer patient. In addition, only a small fraction of the mutations are immunogenic and therefore their accurate prediction is critical. In this review, we discuss the properties of neoantigens that influence their immunogenicity, along with questions that remain to be addressed in order to improve prediction algorithms.