Oncolytic virotherapy induced CSDE1 neo-antigenesis restricts VSV replication but can be targeted by immunotherapy

Abstract

In our clinical trials of oncolytic vesicular stomatitis virus expressing interferon beta (VSV-IFNβ), several patients achieved initial responses followed by aggressive relapse. We show here that VSV-IFNβ-escape tumors predictably express a point-mutated CSDE1^P5S form of the RNA-binding Cold Shock Domain-containing E1 protein, which promotes escape as an inhibitor of VSV replication by disrupting viral transcription. Given time, VSV-IFNβ evolves a compensatory mutation in the P/M Inter-Genic Region which rescues replication in CSDE1^P5S cells. These data show that CSDE1 is a major cellular co-factor for VSV replication. However, CSDE1^P5S also generates a neo-epitope recognized by non-tolerized T cells. We exploit this predictable neo-antigenesis to drive, and trap, tumors into an escape phenotype, which can be ambushed by vaccination against CSDE1^P5S, preventing tumor escape. Combining frontline therapy with escape-targeting immunotherapy will be applicable across multiple therapies which drive tumor mutation/evolution and simultaneously generate novel, targetable immunopeptidomes associated with acquired treatment resistance.

Introduction

Escape from frontline therapy is a major cause of treatment failure in cancer patients^1,2,3,4, wherein a subset of patients initially develop promising clinical responses, followed by aggressive, lethal, tumor growth. Hence, strategies that reduce treatment failure through tumor escape would be highly significant.

We have shown that treatment-escaped tumors differ significantly from primary tumors immunogenically, genetically, and phenotypically^{1,2,3,4,5,6,7,8,9,10,11,12}, due to mutational plasticity and selection of treatment-resistant clones^{1,2,3,4,9,10,11}. These treatment-escape phenotypes arise, at least in part, through the evolution of a pool of mutated tumor cells from which highly aggressive, treatment-resistant clones are rapidly selected^{1,2,3,4,9,10,11}. In this respect, APOBEC (apolipoprotein B mRNA editing enzyme, catalytic polypeptide-like) cytosine deaminases provide an endogenous source of DNA mutation-driving cancer evolution in response to a variety of different frontline treatments^{1,9,10,11,13,14,15,16,17,18,19,20}. APOBEC3 cytosine deaminases act as innate antiviral restriction factors and catalyze cytosine to uracil deamination of ssDNA (C–T transitions and C–G transversions)^1,16,17. Although the human genome encodes seven APOBEC3 enzymes (3A–H), the mouse encodes a single gene, mAPOBEC3^1,17,21, which has similar activities to those of human APOBEC3B (hAPOBEC3B)¹⁰. Consistent with APOBECs as drivers of tumor escape, the APOBEC3B signature is associated with therapeutic resistance in multiple cancers^{1,14,15,18,19,20}.

We have shown that APOBEC3 induction following frontline treatment with adoptive T-cell therapy, chemotherapy, or oncolytic virotherapy has profound consequences for the generation of escape variants from all three types of therapy^9,10,11. With respect to oncolytic virotherapy, we, and others, have developed vesicular stomatitis virus (VSV), a single-strand negative sense RNA virus (Rhabdovirus, Indiana serotype), as an oncolytic platform for clinical testing^{22,23,24,25,26,27,28,29,30,31}. VSV, which is highly sensitive to inhibition by interferon (IFN), shows selective replication in Type I IFN-defective tumor cells, while being rapidly shut down in (IFN-responsive) normal cells^22,23,24. For the clinical development of VSV³², we overexpressed the IFNβ gene in the virus²⁵ to enhance safety (increased antiviral IFN expressed in normal cells) and to increase the immunogenicity of infected/dying tumor cells, as IFNβ acts as a key signal 3 cytokine to facilitate priming of tumor-reactive T cells^25,32,33.

In our clinical trials of VSV-IFNβ as an oncolytic^{22,23,24,25,26,27,28,29,30,31}, several patients achieved initial responses followed by aggressive escape. To understand the mechanistic basis of these effects, we established in vitro and in vivo models in which suboptimal levels of VSV infection leads to the escape of virus-resistant cells^9,10. We showed that suboptimal infection with VSV induced Type I IFN-dependent hAPOBEC3B- or mAPOBEC3-induced mutation of the cell genome, degradation of the viral genome, and escape of virus/oncolysis-resistant (VSV-ESC) cells^9,10. VSV-ESC cells¹⁰ carried stable APOBEC3B mutational signatures in multiple genes¹¹, some of which might be critical for escape. Simultaneously, we reasoned that some of these mutations may also induce neo-antigenesis—the generation of neo-epitopes with increased major histocompatibility complex (MHC) binding and immunogenicity^34,35,36, rendering VSV-ESC cells susceptible to T-cell attack. Consistent with this hypothesis, we identified a C–T mutation in the CSDE1 gene (CSDE1^C-T) (proline to serine at αα5, CSDE1^P5S)¹¹ in VSV-ESC cells^10,11, which generated a heteroclitic^37,38,39 neo-epitope, which primed T-cell responses against both itself (CSDE1^P5S) and, to a lesser extent, wild-type CSDE1^[WT11. CSDE1, is multi-functional RNA-binding protein that regulates RNA translation^{40,41,42,43,44,45,46,47}….