Alexander Solovyov

A neoantigen fitness model predicts tumour response to checkpoint blockade immunotherapy

Checkpoint blockade immunotherapies enable the host immune system to recognize and destroy tumour cells. Their clinical activity has been correlated with activated T-cell recognition of neoantigens, which are tumour-specific, mutated peptides presented on the surface of cancer cells. Here we present a fitness model for tumours based on immune interactions of neoantigens that predicts response to immunotherapy. Two main factors determine neoantigen fitness: the likelihood of neoantigen presentation by the major histocompatibility complex (MHC) and subsequent recognition by T cells. We estimate these components using the relative MHC binding affinity of each neoantigen to its wild type and a nonlinear dependence on sequence similarity of neoantigens to known antigens. To describe the evolution of a heterogeneous tumour, we evaluate its fitness as a weighted effect of dominant neoantigens in the subclones of the tumour. Our model predicts survival in anti-CTLA-4-treated patients with melanoma and anti-PD-1-treated patients with lung cancer. Importantly, low-fitness neoantigens identified by our method may be leveraged for developing novel immunotherapies. By using an immune fitness model to study immunotherapy, we reveal broad similarities between the evolution of tumours and rapidly evolving pathogens.

Dengue virus NS2B protein targets cGAS for degradation and prevents mitochondrial DNA sensing during infection

During the last few decades, the global incidence of dengue virus (DENV) has increased dramatically, and it is now endemic in more than 100 countries. To establish a productive infection in humans, DENV uses different strategies to inhibit or avoid the host innate immune system. Several DENV proteins have been shown to strategically target crucial components of the type I interferon system. Here, we report that the DENV NS2B protease cofactor targets the DNA sensor cyclic GMP-AMP synthase (cGAS) for lysosomal degradation to avoid the detection of mitochondrial DNA during infection. Such degradation subsequently results in the inhibition of type I interferon production in the infected cell. Our data demonstrate a mechanism by which cGAS senses cellular damage upon DENV infection.

Host Dependent Evolutionary Patterns and the Origin of 2009 H1N1 Pandemic Influenza.

The origin of H1N1pdm constitutes an unresolved mystery, as its most recently observed ancestors were isolated in pigs nearly a decade before it emerged in humans. One theory suggests imperfect surveillance of swine viruses caused the virus to be missed in swine herds. Other hypotheses point to the possibility of laboratory error or an avian intermediary. We show substitution bias classification identifies the host where a virus has been evolving. Comparing the evolution of H1N1pdm ancestors with other influenza viruses, we show the evolutionary history in unsampled years is similar to the evolution of other swine viruses, presenting evidence it emerged from unsampled herds.

Diverse repetitive element RNA expression defines epigenetic and immunologic features of colon cancer

There is tremendous excitement for the potential of epigenetic therapies in cancer, but the ability to predict and monitor response to these drugs remains elusive. This is in part due to the inability to differentiate the direct cytotoxic and the immunomodulatory effects of these drugs. The DNA-hypomethylating agent 5-azacitidine (AZA) has shown these distinct effects in colon cancer and appears to be linked to the derepression of repeat RNAs. LINE and HERV are two of the largest classes of repeats in the genome, and despite many commonalities, we found that there is heterogeneity in behavior among repeat subtypes. Specifically, the LINE-1 and HERV-H subtypes detected by RNA sequencing and RNA in situ hybridization in colon cancers had distinct expression patterns, which suggested that these repeats are correlated to transcriptional programs marking different biological states. We found that low LINE-1 expression correlates with global DNA hypermethylation, wild-type TP53 status, and responsiveness to AZA. HERV-H repeats were not concordant with LINE-1 expression but were found to be linked with differences in FOXP3+ Treg tumor infiltrates. Together, distinct repeat RNA expression patterns define new molecular classifications of colon cancer and provide biomarkers that better distinguish cytotoxic from immunomodulatory effects by epigenetic drugs.

FBXW7 regulates a mitochondrial transcription program by modulating MITF

FBXW7 is well characterized as a tumor suppressor in many human cancers including melanoma; however, the mechanisms of tumor‐suppressive function have not been fully elucidated. We leveraged two distinct RNA sequencing datasets: human melanoma cell lines (n = 10) with control versus silenced FBXW7 and a cohort of human melanoma tumor samples (n = 51) to define the transcriptomic fingerprint regulated by FBXW7. Here, we report that loss of FBXW7 enhances a mitochondrial gene transcriptional program that is dependent on MITF in human melanoma and confers poor patient outcomes. MITF is a lineage‐specific master regulator of melanocytes and together with PGC‐1alpha is a marker for melanoma subtypes with dependence for mitochondrial oxidative metabolism. We found that inactivation of FBXW7 elevates MITF protein levels in melanoma cells. In vitro studies examining loss of FBXW7 and MITF alone or in combination showed that FBXW7 is an upstream regulator for the MITF/PGC‐1 signaling.

Global Cancer Transcriptome Quantifies Repeat Element Polarization between Immunotherapy Responsive and T Cell Suppressive Classes

SUMMARY It has been posited that anti-tumoral innate activation is driven by derepression of endogenous repeats. We compared RNA sequencing protocols to assess repeat transcriptomes in The Cancer Genome Atlas (TCGA). Although poly(A) selection efficiently detects coding genes, most non-coding genes, and limited subsets of repeats, it fails to capture overall repeat expression and co-expression. Alternatively, total RNA expression reveals distinct repeat co-expression subgroups and delivers greater dynamic changes, implying they may serve as better biomarkers of clinical outcomes. We show that endogenous retrovirus expression predicts immunotherapy response better than conventional immune signatures in one cohort yet is not predictive in another. Moreover, we find that global repeat derepression, including the HSATII satellite repeat, correlates with an immunosuppressive phenotype in colorectal and pancreatic tumors and validate in situ. In conclusion, we stress the importance of analyzing the full spectrum of repeat transcription to decode their role in tumor immunity.

Abstract A087: Quantifying the landscape of immunostimulatory tumoral RNA

Recent studies demonstrate an unexpected connection between aberrant transcription of noncoding RNA in tumors and innate immune system activation in the tumor microenvironment. Such RNA is often of unknown function and may consist of typically silenced interspersed elements, satellite repeats, and endogenous retroviruses. For instance, satellite RNA from the pericentromere (particularly HSATII) is abundantly transcribed in several solid tumors - such as pancreatic ductal adenocarcinoma - yet it is virtually silent in normal tissue. The genomic DNA repetitive regions this RNA is transcribed from frequently expand during tumorigenesis. Using novel quantitative methods, we have shown that a set of such repetitive elements, abundantly expressed in tumors, display sequence patterns typically associated with viruses. We therefore predicted they are immunogenic, particularly HSATII. In a novel, theory-experiment collaboration between the laboratories of Professors Benjamin Greenbaum and Nina Bhardwaj, the most significant set of these RNA have been validated as immunostimulatory (HSATII and murine GSAT), capable of activating antigen presenting cells - HSATII stimulated production of IL-6, IL-12 and TNFalpha (Tanne, et al., PNAS, 2015). At the same time a set of recent papers has shown that ERV transcription may be a predictor of immunotherapy response in melanoma. Hence, it is critical to profile key immunostimulatory endogenous RNA in the tumor microenviroment, understand which immune pathways different sets of such RNA activate, and assess the link between the specific pathways activated, prognosis, and immunotherapy. We further profile the landscape of activation and expression for endogenous elements. We have currently demonstrated a key set of noncoding RNAs preferentially expressed in cancer cells have sequence features that are immunostimulatory in humans, and have characterized the range and breadth of the expression of such elements in several solid tumors. We present several new results on the topic and the potential consequences of the aberrant expression of endogenous RNA that mimic pathogen features in cancer. Citation Format: Alexander Solovyov, Antoine Tanne, Luciana Muniz, Simona Cocco, Remi Monasson, Arnold Levine, David T. Ting, Nina Bhardwaj, Benjamin Greenbaum. Quantifying the landscape of immunostimulatory tumoral RNA [abstract]. In: Proceedings of the Second CRI-CIMT-EATI-AACR International Cancer Immunotherapy Conference: Translating Science into Survival; 2016 Sept 25-28; New York, NY. Philadelphia (PA): AACR; Cancer Immunol Res 2016;4(11 Suppl):Abstract nr A087.

Abstract 531: The landscape of immunostimulatory RNA transcription

Recent studies have demonstrated an unexpected connection between aberrant transcription of noncoding RNA (ncRNA) in tumors and innate immune system activation in the tumor microenvironment (TME). Such ncRNA are often of unknown function and may consist of typically silenced interspersed elements, satellite repeats, and endogenous retroviruses (ERVs). For instance, satellite RNA from the pericentromere (particularly HSATII) has been shown to be abundantly transcribed in several solid tumors, including the often lethal pancreatic ductal adenocarcinoma, yet is virtually silent in normal tissue. The genomic DNA repetitive regions this RNA is transcribed from frequently expand during tumerigenesis. Using a novel computational approach, we have shown that, unlike ncRNA expressed under normal conditions, a set of such repetitive elements, abundantly expressed in tumors, display patterns typically associated with viruses. We therefore predicted they are immunogenic, particularly HSATII. In a novel, theory-experiment collaboration between the laboratories of Professors Benjamin Greenbaum and Nina Bhardwaj, the most significant set of these ncRNA have been validated as immunogenic (HSATII and murine GSAT), capable of activating antigen presenting cells - HSATII stimulated production of IL-6, IL-12 and TNFalpha (Tanne, et al., PNAS, 2015). At the same time a set of recent papers has shown that ERV transcription may be a predictor of immunotherapy response in melanoma. Hence, it is critical to profile key “immunogenic” ncRNA (i-ncRNA) in the tumor microenviroment, understand which immune pathways different sets of i-ncRNA activate, and assess the link between the specific pathways activated, prognosis, and immunotherapy. Our preliminary work on melanoma transcriptomes has shown that activation of endogenous elements implies better prognosis. We further profile the landscape of activation for i-ncRNA candidates. We have currently demonstrated a key set of ncRNAs preferentially expressed in cancer cells have sequence features that are immunostimulatory in humans. Quantitatively, our analysis uses a new, unique set of computational methods to study the innate immune system in the TME, originally designed to find motifs in viruses targeted by innate immune receptors (Greenbaum, et al., PNAS, 2014). Our methods for motif characterization utilize a novel mathematical approach we created based on transfer matrix approaches in statistical physics, and are here utilized for the purpose of discovering immunogenic nucleic acids in the tumor microenvironment. They offer several advantages over previous approaches. In particular they are far more computationally efficient, allowing the discovery of anomalous, potentially immunostimulatory patterns of motif usage in far larger sequence datasets than was previously practical. Citation Format: Alexander Solovyov, Antoine Tanne, David Ting, Arnold Levine, Nina Bhardwaj, Benjamin Greenbaum. The landscape of immunostimulatory RNA transcription. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr 531.

Unraveling predicted immunomodulatory effects of novel cancer-associated non-coding RNAs

Cellular non-coding RNA (ncRNA) transcription and regulation has only recently been systematically investigated. Studies have demonstrated abundant transcription of a set of ncRNAs preferentially within tumors as opposed to normal tissue. In some cases these tumor-associated ncRNAs also associate with inflammatory cytokine production suggesting potential immunostimulatory qualities. Using a novel approach from statistical physics we quantify global transcriptome-wide motif usage for the first time in human and murine ncRNAs determining that most have motif usage consistent with the coding genome. However an outlier subset of tumor-associated ncRNAs typically of recent evolutionary origin has motif usage more associated with pathogen RNA. Here we show that a human repeat enhances immunostimulatory motifs CpG dinucleotides in AU-rich contexts which most of the human genome and human adapted viruses have evolved to avoid. We demonstrate that a key subset of these ncRNA function as immunostimulatory “self-agonists” and directly activate cells of the mononuclear phagocytic system to produce pro-inflammatory cytokines. These ncRNAs arise from endogenous repetitive elements that are normally silenced, yet are often very highly expressed in cancers. We conclude that the innate response in tumors partially originates from direct interaction of immunogenic ncRNAs expressed in cancer cells with innate pattern recognition receptors and thereby assign a new danger-associated function to a set of dark matter repetitive elements, potentially reconciling several observations concerning the role of ncRNA expression in cancers. We consider the possible roles they may play in the tumor microenvironment.