Multiple instance learning to predict immune checkpoint blockade efficacy using neoantigen candidates.

Previous studies showed that the neoantigen candidate load is an imperfect predictor of immune checkpoint blockade (ICB) efficacy. Further studies provided evidence that the response to ICB is also affected by the qualitative properties of a few or even single candidates, limiting the predictive power based on candidate quantity alone. Here, we predict ICB efficacy based on neoantigen candidates and their neoantigen features in the context of the mutation type, using Multiple-Instance Learning via Embedded Instance Selection (MILES). Multiple instance learning is a type of supervised machine learning that classifies labeled bags that are formed by a set of unlabeled instances. MILES performed better compared with neoantigen candidate load alone for low-abundant fusion genes in renal cell carcinoma. Our findings suggest that MILES is an appropriate method to predict the efficacy of ICB therapy based on neoantigen candidates without requiring direct T cell response information.

CoVigator-A Knowledge Base for Navigating SARS-CoV-2 Genomic Variants.

BACKGROUND: The outbreak of the severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) resulted in the global COVID-19 pandemic. The urgency for an effective SARS-CoV-2 vaccine has led to the development of the first series of vaccines at unprecedented speed. The discovery of SARS-CoV-2 spike-glycoprotein mutants, however, and consequentially the potential to escape vaccine-induced protection and increased infectivity, demonstrates the persisting importance of monitoring SARS-CoV-2 mutations to enable early detection and tracking of genomic variants of concern. RESULTS: We developed the CoVigator tool with three components: (1) a knowledge base that collects new SARS-CoV-2 genomic data, processes it and stores its results; (2) a comprehensive variant calling pipeline; (3) an interactive dashboard highlighting the most relevant findings. The knowledge base routinely downloads and processes virus genome assemblies or raw sequencing data from the COVID-19 Data Portal (C19DP) and the European Nucleotide Archive (ENA), respectively. The results of variant calling are visualized through the dashboard in the form of tables and customizable graphs, making it a versatile tool for tracking SARS-CoV-2 variants. We put a special emphasis on the identification of intrahost mutations and make available to the community what is, to the best of our knowledge, the largest dataset on SARS-CoV-2 intrahost mutations. In the spirit of open data, all CoVigator results are available for download. The CoVigator dashboard is accessible via covigator.tron-mainz.de. CONCLUSIONS: With increasing demand worldwide in genome surveillance for tracking the spread of SARS-CoV-2, CoVigator will be a valuable resource of an up-to-date list of mutations, which can be incorporated into global efforts.

MC38 colorectal tumor cell lines from two different sources display substantial differences in transcriptome, mutanome and neoantigen expression.

Introduction: The cell line MC38 is a commonly used murine model for colorectal carcinoma. It has a high mutational burden, is sensitive to immune checkpoint immunotherapy and endogenous CD8+ T cell responses against neoantigens have been reported. Methods: Here, we re-sequenced exomes and transcriptomes of MC38 cells from two different sources, namely Kerafast (originating from NCI/NIH, MC38-K) and the Leiden University Medical Center cell line collection (MC38-L), comparing the cell lines on the genomic and transcriptomic level and analyzing their recognition by CD8+ T cells with known neo-epitope specificity. Results: The data reveals a distinct structural composition of MC38-K and MC38-L cell line genomes and different ploidies. Further, the MC38-L cell line harbored about 1.3-fold more single nucleotide variations and small insertions and deletions than the MC38-K cell line. In addition, the observed mutational signatures differed; only 35.3% of the non-synonymous variants and 5.4% of the fusion gene events were shared. Transcript expression values of both cell lines correlated strongly (p = 0.919), but we found different pathways enriched in the genes that were differentially upregulated in the MC38-L or MC38-K cells, respectively. Our data show that previously described neoantigens in the MC38 model such as Rpl18mut and Adpgkmut were absent in the MC38-K cell line resulting that such neoantigen-specific CD8+ T cells recognizing and killing MC38-L cells did not recognize or kill MC38-K cells. Conclusion: This strongly indicates that at least two sub-cell lines of MC38 exist in the field and underlines the importance of meticulous tracking of investigated cell lines to obtain reproducible results, and for correct interpretation of the immunological data without artifacts. We present our analyses as a reference for researchers to select the appropriate sub-cell line for their own studies.

Accurate detection of tumor-specific gene fusions reveals strongly immunogenic personal neo-antigens.

Cancer-associated gene fusions are a potential source for highly immunogenic neoantigens, but the lack of computational tools for accurate, sensitive identification of personal gene fusions has limited their targeting in personalized cancer immunotherapy. Here we present EasyFuse, a machine learning computational pipeline for detecting cancer-specific gene fusions in transcriptome data obtained from human cancer samples. EasyFuse predicts personal gene fusions with high precision and sensitivity, outperforming previously described tools. By testing immunogenicity with autologous blood lymphocytes from patients with cancer, we detected pre-established CD4+ and CD8+ T cell responses for 10 of 21 (48%) and for 1 of 30 (3%) identified gene fusions, respectively. The high frequency of T cell responses detected in patients with cancer supports the relevance of individual gene fusions as neoantigens that might be targeted in personalized immunotherapies, especially for tumors with low mutation burden.

Identification of neoantigens for individualized therapeutic cancer vaccines.

Somatic mutations in cancer cells can generate tumour-specific neoepitopes, which are recognized by autologous T cells in the host. As neoepitopes are not subject to central immune tolerance and are not expressed in healthy tissues, they are attractive targets for therapeutic cancer vaccines. Because the vast majority of cancer mutations are unique to the individual patient, harnessing the full potential of this rich source of targets requires individualized treatment approaches. Many computational algorithms and machine-learning tools have been developed to identify mutations in sequence data, to prioritize those that are more likely to be recognized by T cells and to design tailored vaccines for every patient. In this Review, we fill the gaps between the understanding of basic mechanisms of T cell recognition of neoantigens and the computational approaches for discovery of somatic mutations and neoantigen prediction for cancer immunotherapy. We present a new classification of neoantigens, distinguishing between guarding, restrained and ignored neoantigens, based on how they confer proficient antitumour immunity in a given clinical context. Such context-based differentiation will contribute to a framework that connects neoantigen biology to the clinical setting and medical peculiarities of cancer, and will enable future neoantigen-based therapies to provide greater clinical benefit.

T-cell Receptor Therapy Targeting Mutant Capicua Transcriptional Repressor in Experimental Gliomas.

PURPOSE: Gliomas are intrinsic brain tumors with a high degree of constitutive and acquired resistance to standard therapeutic modalities such as radiotherapy and alkylating chemotherapy. Glioma subtypes are recognized by characteristic mutations. Some of these characteristic mutations have shown to generate immunogenic neoepitopes suitable for targeted immunotherapy. EXPERIMENTAL DESIGN: Using peptide-based ELISpot assays, we screened for potential recurrent glioma neoepitopes in MHC-humanized mice. Following vaccination, droplet-based single-cell T-cell receptor (TCR) sequencing from established T-cell lines was applied for neoepitope-specific TCR discovery. Efficacy of intraventricular TCR-transgenic T-cell therapy was assessed in a newly developed glioma model in MHC-humanized mice induced by CRISPR-based delivery of tumor suppressor-targeting guide RNAs. RESULTS: We identify recurrent capicua transcriptional repressor (CIC) inactivating hotspot mutations at position 215 CICR215W/Q as immunogenic MHC class II (MHCII)-restricted neoepitopes. Vaccination of MHC-humanized mice resulted in the generation of robust MHCII-restricted mutation-specific T-cell responses against CICR215W/Q. Adoptive intraventricular transfer of CICR215W-specific TCR-transgenic T cells exert antitumor responses against CICR215W-expressing syngeneic gliomas. CONCLUSIONS: The integration of immunocompetent MHC-humanized orthotopic glioma models in the discovery of shared immunogenic glioma neoepitopes facilitates the identification and preclinical testing of human leukocyte antigen (HLA)-restricted neoepitope-specific TCRs for locoregional TCR-transgenic T-cell adoptive therapy.

Large-scale analysis of SARS-CoV-2 spike-glycoprotein mutants demonstrates the need for continuous screening of virus isolates.

Due to the widespread of the COVID-19 pandemic, the SARS-CoV-2 genome is evolving in diverse human populations. Several studies already reported different strains and an increase in the mutation rate. Particularly, mutations in SARS-CoV-2 spike-glycoprotein are of great interest as it mediates infection in human and recently approved mRNA vaccines are designed to induce immune responses against it. We analyzed 1,036,030 SARS-CoV-2 genome assemblies and 30,806 NGS datasets from GISAID and European Nucleotide Archive (ENA) focusing on non-synonymous mutations in the spike protein. Only around 2.5% of the samples contained the wild-type spike protein with no variation from the reference. Among the spike protein mutants, we confirmed a low mutation rate exhibiting less than 10 non-synonymous mutations in 99.6% of the analyzed sequences, but the mean and median number of spike protein mutations per sample increased over time. 5,472 distinct variants were found in total. The majority of the observed variants were recurrent, but only 21 and 14 recurrent variants were found in at least 1% of the mutant genome assemblies and NGS samples, respectively. Further, we found high-confidence subclonal variants in about 2.6% of the NGS data sets with mutant spike protein, which might indicate co-infection with various SARS-CoV-2 strains and/or intra-host evolution. Lastly, some variants might have an effect on antibody binding or T-cell recognition. These findings demonstrate the continuous importance of monitoring SARS-CoV-2 sequences for an early detection of variants that require adaptations in preventive and therapeutic strategies.

Comprehensive Genomic and Transcriptomic Analysis of Three Synchronous Primary Tumours and a Recurrence from a Head and Neck Cancer Patient.

Synchronous primary malignancies occur in a small proportion of head and neck squamous cell carcinoma (HNSCC) patients. Here, we analysed three synchronous primaries and a recurrence from one patient by comparing the genomic and transcriptomic profiles among the tumour samples and determining the recurrence origin. We found remarkable levels of heterogeneity among the primary tumours, and through the patterns of shared mutations, we traced the origin of the recurrence. Interestingly, the patient carried germline variants that might have predisposed him to carcinogenesis, together with a history of alcohol and tobacco consumption. The mutational signature analysis confirmed the impact of alcohol exposure, with Signature 16 present in all tumour samples. Characterisation of immune cell infiltration highlighted an immunosuppressive environment in all samples, which exceeded the potential activity of T cells. Studies such as the one described here have important clinical value and contribute to personalised treatment decisions for patients with synchronous primaries and matched recurrences.

NeoFox: annotating neoantigen candidates with neoantigen features.

SUMMARY: The detection and prediction of true neoantigens is of great importance for the field of cancer immunotherapy. Wesearched the literature for proposed neoantigen features and integrated them into a toolbox called NEOantigen Feature toolbOX (NeoFox). NeoFox is an easy-to-use Python package that enables the annotation of neoantigen candidates with 16 neoantigen features. AVAILABILITY AND IMPLEMENTATION: NeoFox is freely available as an open source Python package released under the GNU General Public License (GPL) v3 license at https://github.com/TRON-Bioinformatics/neofox. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

A Set of Cell Lines Derived from a Genetic Murine Glioblastoma Model Recapitulates Molecular and Morphological Characteristics of Human Tumors.

Glioblastomas (GBM) are the most aggressive tumors affecting the central nervous system in adults, causing death within, on average, 15 months after diagnosis. Immunocompetent in-vivo models that closely mirror human GBM are urgently needed for deciphering glioma biology and for the development of effective treatment options. The murine GBM cell lines currently available for engraftment in immunocompetent mice are not only exiguous but also inadequate in representing prominent characteristics of human GBM such as infiltrative behavior, necrotic areas, and pronounced tumor heterogeneity. Therefore, we generated a set of glioblastoma cell lines by repeated in vivo passaging of cells isolated from a neural stem cell-specific Pten/p53 double-knockout genetic mouse brain tumor model. Transcriptome and genome analyses of the cell lines revealed molecular heterogeneity comparable to that observed in human glioblastoma. Upon orthotopic transplantation into syngeneic hosts, they formed high-grade gliomas that faithfully recapitulated the histopathological features, invasiveness and immune cell infiltration characteristic of human glioblastoma. These features make our cell lines unique and useful tools to study multiple aspects of glioblastoma pathomechanism and to test novel treatments in an intact immune microenvironment.

Integrative analysis of structural variations using short-reads and linked-reads yields highly specific and sensitive predictions.

Genetic diseases are driven by aberrations of the human genome. Identification of such aberrations including structural variations (SVs) is key to our understanding. Conventional short-reads whole genome sequencing (cWGS) can identify SVs to base-pair resolution, but utilizes only short-range information and suffers from high false discovery rate (FDR). Linked-reads sequencing (10XWGS) utilizes long-range information by linkage of short-reads originating from the same large DNA molecule. This can mitigate alignment-based artefacts especially in repetitive regions and should enable better prediction of SVs. However, an unbiased evaluation of this technology is not available. In this study, we performed a comprehensive analysis of different types and sizes of SVs predicted by both the technologies and validated with an independent PCR based approach. The SVs commonly identified by both the technologies were highly specific, while validation rate dropped for uncommon events. A particularly high FDR was observed for SVs only found by 10XWGS. To improve FDR and sensitivity, statistical models for both the technologies were trained. Using our approach, we characterized SVs from the MCF7 cell line and a primary breast cancer tumor with high precision. This approach improves SV prediction and can therefore help in understanding the underlying genetics in various diseases.

Multi-Omics Characterization of the 4T1 Murine Mammary Gland Tumor Model.

Background: Tumor models are critical for our understanding of cancer and the development of cancer therapeutics. The 4T1 murine mammary cancer cell line is one of the most widely used breast cancer models. Here, we present an integrated map of the genome, transcriptome, and immunome of 4T1. Results: We found Trp53 (Tp53) and Pik3g to be mutated. Other frequently mutated genes in breast cancer, including Brca1 and Brca2, are not mutated. For cancer related genes, Nav3, Cenpf, Muc5Ac, Mpp7, Gas1, MageD2, Dusp1, Ros, Polr2a, Rragd, Ros1, and Hoxa9 are mutated. Markers for cell proliferation like Top2a, Birc5, and Mki67 are highly expressed, so are markers for metastasis like Msln, Ect2, and Plk1, which are known to be overexpressed in triple-negative breast cancer (TNBC). TNBC markers are, compared to a mammary gland control sample, lower (Esr1), comparably low (Erbb2), or not expressed at all (Pgr). We also found testis cancer antigen Pbk as well as colon/gastrointestinal cancer antigens Gpa33 and Epcam to be highly expressed. Major histocompatibility complex (MHC) class I is expressed, while MHC class II is not. We identified 505 single nucleotide variations (SNVs) and 20 insertions and deletions (indels). Neoantigens derived from 22 SNVs and one deletion elicited CD8+ or CD4+ T cell responses in IFNγ-ELISpot assays. Twelve high-confidence fusion genes were observed. We did not observe significant downregulation of mismatch repair (MMR) genes or SNVs/indels impairing their function, providing evidence for 6-thioguanine resistance. Effects of the integration of the murine mammary tumor virus were observed at the genome and transcriptome level. Conclusions: 4T1 cells share substantial molecular features with human TNBC. As 4T1 is a common model for metastatic tumors, our data supports the rational design of mode-of-action studies for pre-clinical evaluation of targeted immunotherapies.

Bioinformatics for Cancer Immunotherapy.

Our immune system plays a key role in health and disease as it is capable of responding to foreign antigens as well as acquired antigens from cancer cells. Latter are caused by somatic mutations, the so-called neoepitopes, and might be recognized by T cells if they are presented by HLA molecules on the surface of cancer cells. Personalized mutanome vaccines are a class of customized immunotherapies, which is dependent on the detection of individual cancer-specific tumor mutations and neoepitope (i.e., prediction, followed by a rational vaccine design, before on-demand production. The development of next generation sequencing (NGS) technologies and bioinformatic tools allows a large-scale analysis of each parameter involved in this process. Here, we provide an overview of the bioinformatic aspects involved in the design of personalized, neoantigen-based vaccines, including the detection of mutations and the subsequent prediction of potential epitopes, as well as methods for associated biomarker research, such as high-throughput sequencing of T-cell receptors (TCRs), followed by data analysis and the bioinformatics quantification of immune cell infiltration in cancer samples.

ArtiFuse-computational validation of fusion gene detection tools without relying on simulated reads.

MOTIVATION: Gene fusions are an important class of transcriptional variants that can influence cancer development and can be predicted from RNA sequencing (RNA-seq) data by multiple existing tools. However, the real-world performance of these tools is unclear due to the lack of known positive and negative events, especially with regard to fusion genes in individual samples. Often simulated reads are used, but these cannot account for all technical biases in RNA-seq data generated from real samples. RESULTS: Here, we present ArtiFuse, a novel approach that simulates fusion genes by sequence modification to the genomic reference, and therefore, can be applied to any RNA-seq dataset without the need for any simulated reads. We demonstrate our approach on eight RNA-seq datasets for three fusion gene prediction tools: average recall values peak for all three tools between 0.4 and 0.56 for high-quality and high-coverage datasets. As ArtiFuse affords total control over involved genes and breakpoint position, we also assessed performance with regard to gene-related properties, showing a drop-in recall value for low-expressed genes in high-coverage samples and genes with co-expressed paralogues. Overall tool performance assessed from ArtiFusions is lower compared to previously reported estimates on simulated reads. Due to the use of real RNA-seq datasets, we believe that ArtiFuse provides a more realistic benchmark that can be used to develop more accurate fusion gene prediction tools for application in clinical settings. AVAILABILITY AND IMPLEMENTATION: ArtiFuse is implemented in Python. The source code and documentation are available at https://github.com/TRON-Bioinformatics/ArtiFusion. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Publisher Correction: Actively personalized vaccination trial for newly diagnosed glioblastoma.

The additional author support information was erroneously omitted from the Supplementary Information. This has been corrected online.

Improving mRNA-Based Therapeutic Gene Delivery by Expression-Augmenting 3' UTRs Identified by Cellular Library Screening.

Synthetic mRNA has emerged as a powerful tool for the transfer of genetic information, and it is being explored for a variety of therapeutic applications. Many of these applications require prolonged intracellular persistence of mRNA to improve bioavailability of the encoded protein. mRNA molecules are intrinsically unstable and their intracellular kinetics depend on the UTRs embracing the coding sequence, in particular the 3' UTR elements. We describe here a novel and generally applicable cell-based selection process for the identification of 3' UTRs that augment the expression of proteins encoded by synthetic mRNA. Moreover, we show, for two applications of mRNA therapeutics, namely, (1) the delivery of vaccine antigens in order to mount T cell immune responses and (2) the introduction of reprogramming factors into differentiated cells in order to induce pluripotency, that mRNAs tagged with the 3' UTR elements discovered in this study outperform those with commonly used 3' UTRs. This approach further leverages the utility of mRNA as a gene therapy drug format.

Actively personalized vaccination trial for newly diagnosed glioblastoma.

Patients with glioblastoma currently do not sufficiently benefit from recent breakthroughs in cancer treatment that use checkpoint inhibitors1,2. For treatments using checkpoint inhibitors to be successful, a high mutational load and responses to neoepitopes are thought to be essential3. There is limited intratumoural infiltration of immune cells4 in glioblastoma and these tumours contain only 30-50 non-synonymous mutations5. Exploitation of the full repertoire of tumour antigens-that is, both unmutated antigens and neoepitopes-may offer more effective immunotherapies, especially for tumours with a low mutational load. Here, in the phase I trial GAPVAC-101 of the Glioma Actively Personalized Vaccine Consortium (GAPVAC), we integrated highly individualized vaccinations with both types of tumour antigens into standard care to optimally exploit the limited target space for patients with newly diagnosed glioblastoma. Fifteen patients with glioblastomas positive for human leukocyte antigen (HLA)-A*02:01 or HLA-A*24:02 were treated with a vaccine (APVAC1) derived from a premanufactured library of unmutated antigens followed by treatment with APVAC2, which preferentially targeted neoepitopes. Personalization was based on mutations and analyses of the transcriptomes and immunopeptidomes of the individual tumours. The GAPVAC approach was feasible and vaccines that had poly-ICLC (polyriboinosinic-polyribocytidylic acid-poly-L-lysine carboxymethylcellulose) and granulocyte-macrophage colony-stimulating factor as adjuvants displayed favourable safety and strong immunogenicity. Unmutated APVAC1 antigens elicited sustained responses of central memory CD8+ T cells. APVAC2 induced predominantly CD4+ T cell responses of T helper 1 type against predicted neoepitopes.

HLA and proteasome expression body map.

BACKGROUND: The presentation of HLA peptide complexes to T cells is a highly regulated and tissue specific process involving multiple transcriptionally controlled cellular components. The extensive polymorphism of HLA genes and the complex composition of the proteasome make it difficult to map their expression profiles across tissues. METHODS: Here we applied a tailored gene quantification pipeline to 4323 publicly available RNA-Seq datasets representing 55 normal tissues and cell types to examine expression profiles of (classical and non-classical) HLA class I, class II and proteasomal genes. RESULTS: We generated the first comprehensive expression atlas of antigen presenting-related genes across 56 normal tissues and cell types, including immune cells, pancreatic islets, platelets and hematopoietic stem cells. We found a surprisingly heterogeneous HLA expression pattern with up to 100-fold difference in intra-tissue median HLA abundances. Cells of the immune system and lymphatic organs expressed the highest levels of classical HLA class I (HLA-A,-B,-C), class II (HLA-DQA1,-DQB1,-DPA1,-DPB1,-DRA,-DRB1) and non-classical HLA class I (HLA-E,-F) molecules, whereas retina, brain, muscle, megakaryocytes and erythroblasts showed the lowest abundance. In contrast, we identified a distinct and highly tissue-restricted expression pattern of the non-classical class I gene HLA-G in placenta, pancreatic islets, pituitary gland and testis. While the constitutive proteasome showed relatively constant expression across all tissues, we found the immunoproteasome to be enriched in lymphatic organs and almost absent in immune privileged tissues. CONCLUSIONS: Here, we not only provide a reference catalog of tissue and cell type specific HLA expression, but also highlight extremely variable expression of the basic components of antigen processing and presentation in different cell types. Our findings indicate that low expression of classical HLA class I molecules together with lack of immunoproteasome components as well as upregulation of HLA-G may be of key relevance to maintain tolerance in immune privileged tissues.

Targeting immune checkpoints potentiates immunoediting and changes the dynamics of tumor evolution.

The cancer immunoediting hypothesis postulates a dual role of the immune system: protecting the host by eliminating tumor cells, and shaping the tumor by editing its genome. Here, we elucidate the impact of evolutionary and immune-related forces on editing the tumor in a mouse model for hypermutated and microsatellite-instable colorectal cancer. Analyses of wild-type and immunodeficient RAG1 knockout mice transplanted with MC38 cells reveal that upregulation of checkpoint molecules and infiltration by Tregs are the major tumor escape mechanisms. Our results show that the effects of immunoediting are weak and that neutral accumulation of mutations dominates. Targeting the PD-1/PD-L1 pathway using immune checkpoint blocker effectively potentiates immunoediting. The immunoediting effects are less pronounced in the CT26 cell line, a non-hypermutated/microsatellite-instable model. Our study demonstrates that neutral evolution is another force that contributes to sculpting the tumor and that checkpoint blockade effectively enforces T-cell-dependent immunoselective pressure.