Interfering With Tumor Pathways That Augment Viral Oncolysis

Abstract

I have always encouraged my graduate students to take advanced virology courses, not just to gain an understanding of the biology of these highly evolved parasites but because the study of virus–host interactions has formed, and continues to form, the basis of much of our understanding of the molecular mechanisms that control mammalian cell growth and development. The apparent genetic simplicity of viruses belies the sophistication of the strategies they use to co-opt their much more complex cellular hosts by exploiting the signaling pathways that regulate mammalian cell growth. A recent article in Cancer Cell by Mahoney and colleagues1 demonstrates that we still have much to learn from studying the interactions between viruses and the cells they infect. The study describes a functional genomics approach that identified unexpected cancer-specific pathways that can be manipulated to augment oncolytic virus killing of tumor cells. Studying viruses has led to many of the major breakthroughs in our understanding of the molecular biology of the cell. For example, Richard Roberts and Phil Sharp discovered the principles of gene splicing by studying the biology of adenoviruses,2 and much of our current understanding of the regulation of mammalian protein translation and the discovery of internal ribosome entry site elements by Nahum Sonenberg's group comes from studying picornaviruses.3,4 Principles of DNA replication and DNA repair were deciphered using viruses as probes, and the cellular Abl, Src, and Ras oncogenes were originally identified as the transforming components of oncogenic retroviruses.5 Knowing that viruses exploit many of the key pathways that control cell growth and that dysregulation of these same pathways contributes to malignancies has led to efforts to engineer viruses to target cancer cells.6,7 These so-called oncolytic viruses (OVs)—based on vaccinia, herpes, measles, and reovirus platforms—have shown promise in early clinical studies.8,9,10,11,12 The aim of oncolytic virotherapy is to engineer tumor-specific viral parasites that can infect and commandeer the metabolic machinery of the cancer cell. Once in control of the cell, the OV would replicate and ultimately lead to the manufacture and assembly of progeny that could continue to kill the tumor in successive waves. A key aspect of this class of therapeutics is that they not infect or replicate within normal tissues. Although some OVs in preclinical development are designed to be able to discriminate between normal and cancer cells via the recognition of receptors specifically expressed on the malignant cell surface,13,14 all the OVs currently being tested in the clinic recognize receptors found on the surface of both normal and cancerous cells. Indeed, the selectivity of most OV therapeutics currently in the clinic is instead based on the specific intracellular signaling pathways that are dysregulated in the target cancer cell. For example, the vaccinia virus–based therapeutic JX-594 has an engineered deletion of its virally encoded thymidine kinase gene and thus is dependent, in part, on the overexpression of the cellular thymidine kinase gene alone or in combination with other proteins in this metabolic pathway that are characteristic of many malignancies.15 Both JX-594 and the reovirus-based therapeutic Reolysin have a predilection for growing in tumor cells that have an activated epidermal growth factor receptor (EGFR)–Ras pathway.7,8 The herpesvirus-based therapeutic OncoVEX lacks the viral ICP34.5 gene, which normally plays a critical role in counteracting the antiviral programs initiated by interferon.16 Indeed, the apparent defective interferon response that is characteristic of many different kinds of tumors17 has a role in the selective replication of a great number of oncolytic viruses. Whereas in an ideal world the selectivity of an OV would depend on absolutes found in tumor cells and absent in normal tissues, the reality is that—as with most other therapeutics—the differential activities of the relevant signaling pathways in normal and tumor cells are not always clear-cut. Activation of the EGFR–Ras pathway can occur at many levels and to different extents; the degree of overexpression of enzymes involved in DNA metabolism is variable; the interferon response of tumor cells varies from nearly normal to completely absent. The genetic heterogeneity of tumors can thus lead to variable response to any therapeutic, and this is certainly true of OVs. The dream of creating a “replicating machine” that can rapidly eat through a tumor may be restricted to the few cancers that have, for instance, an absolute loss of interferon response or superactivation of the EGFR pathway. On the other hand, the genomic chaos characteristic of many malignancies can create a situation that lends itself to the development of synthetic lethality. Thus, mutations leading to even partial sensitivity to an OV might be complemented by drugs that target a second pathway such that it uniquely sensitizes the tumor, but not normal cells, to the killing properties of a virus infection. In normal cells, there are many levels of redundancy to protect against invasion by microbes, genotoxic damage, or stress. In cancers, mutations that occur in critical growth control or apoptotic genes can reduce these layers of protection; it should therefore be possible to identify compounds that can enhance the ability of OVs to kill tumor cells without sensitizing normal cells to OV infection. Indeed, this has already been shown to be possible using high-throughput screens of small-molecule libraries on virus-infected tumor cells.18,19 Molecules that further cripple the already weakened antiviral response of tumor cells or compounds that enhance the expression of enzymes involved in DNA metabolism have been shown to sensitize refractory tumor cells to OV infection.19,20 Although drug screens are useful, they are limited in that the exact target of action of the newly identified drug may remain elusive—and with it a wealth of information on “druggable” pathways. In the new study, Mahoney et al.1 take a different approach to identifying synthetic lethal mutations that could be revealed during OV infection. Their strategy involved treating partially OV-sensitive tumor cell lines with RNA interference (RNAi) directed against expressed cellular genes. Using an arrayed library of approximately 18,000 genes, they then used the oncolytic Maraba virus21 to probe for genes that sensitized tumor cells to viral oncolysis. Remarkably, they uncovered RNAi-targetable genes that could specifically sensitize tumor cells over 10,000-fold to Maraba infection. The RNAi screen identified a number of gene products involved in the unfolded protein response (UPR), including dedicated transcription factors (ATF6α, ATF6B), the endoribonuclease IRE1α and its downstream product XBP-1. They also identified proteins associated with the endoplasmic reticulum (ER)-associated degradation (ERAD) pathway that removes misfolded polypeptides from the ER and targets them for proteolytic degradation. The striking enrichment of proteins involved in the UPR and ERAD pathways suggested that Mahoney and colleagues had identified a key pathway that could complement cell killing by Maraba and perhaps other rhabdoviruses. Importantly, combination of Maraba infection with knockdown of UPR/ERAD genes did not sensitize normal skin or lung fibroblasts or normal human astrocytes. To close the loop, the workers then chemically synthesized known inhibitors of IREα and demonstrated that these could block UPR and synergize with Maraba in tumor cell killing. The authors' experiments also revealed that tumor cells had “rewired” their UPR/ERAD pathways—in the sense that they have come to a new equilibrium with respect to ER stress—leading to a tumor-specific activation of an apoptotic pathway triggered by OV infection that is caspase 2–dependent. This led to a prediction that an “ER preload” by RNAi inactivation of IRE1α could lead to enhanced tumor cell killing by other chemical compounds that work through caspase 2 activation. Indeed, doxorubicin treatment following IRE1α knockdown specifically increased tumor cell killing. Once again, exploration of virus–host interactions has led to a new understanding of the myriad pathways that control the life and death of mammalian cells. Many questions remain. Is tumor cell killing by all OVs enhanced by ER preload or is Maraba virus uniquely sensitive? Can other synthetic lethal mutations be identified by screening with different OVs? How frequently do tumor cells rewire their UPR/ERAD pathways? Although it remains unknown whether rhabdoviruses such as Maraba will become viable cancer therapeutics, the synthetic lethal screening approach described by Mahoney et al.1 illustrates the value of studying how oncolytic viruses replicate within and kill cancer cells. Although many of us believe that OVs will eventually become viable anticancer therapeutics, the results from this group suggest that, at a minimum, studying the biology of OV–host interactions will reveal previously unappreciated cancer-specific pathways that could potentially identify combination drug approaches that might be less toxic, and yet more effective, in cancer patients.