We report here that IFN-β expression from a B18R-deleted vaccinia results in a systemically effective, highly selective oncolytic virus. To our knowledge, this is the first time that such a rationally designed combination of attenuating viral gene deletion and transgene expression has been incorporated into a systemically deliverable vector, and the first successful report of systemic IFN-β gene delivery to a tumor.
Oncolytic viruses hold promise for the treatment of cancer, but improvements are needed [2
]. In particular, systemic efficacy against metastatic tumors will be required in order to have a major impact on cancer patient survival. Novel approaches are therefore needed to improve intravenous safety and efficacy [28
]. We hypothesized that IFN-β expression from an oncolytic vaccinia virus could achieve both objectives. First, intravenous delivery of vaccinia viruses appears to be feasible in immunocompetent animal tumor models [29
]. Second, IFN-β antiviral effects in normal tissues should enhance safety. Finally, IFN-β expression in tumors should increase efficacy over that mediated by oncolysis alone.
Because type-I IFNs also possess antiviral properties, vaccinia has evolved to express both intracellular (K3L, E3L, and H1L) and extracellular (B18R) gene products that interfere with type-I IFN activity [20
]. In particular, the viral B18R
gene product is secreted from infected cells, binding and neutralizing extracellular IFN-α and -β [22
]. Therefore, to prevent neutralization of IFN-β after secretion from infected cancer cells, we deleted the B18R
gene from the vaccinia virus backbone used to express the IFN-β
It appears that the B18R-deleted vaccinia virus (WR-delB18R) itself is capable of tumor-specific replication. Whereas all primary cells tested were capable of inducing an antiviral state when pretreated with type-I IFN, most (but not all) cancer cell lines were incapable of responding to this cytokine. However, when IFN-α was added 5 h postinfection, to better mimic the likely order of exposure in vivo, wild-type vaccinia (WR), but not WR-delB18R, was capable of preventing a subsequent block in viral replication in susceptible cells. As a result, WR-delB18R was attenuated in normal cells, but not in most tumor cells, when type-I IFN was added postinfection; this sequence would be expected to occur in vivo. Further examination of these effects revealed that in primary cells, WR-delB18R, like wild-type WR, was effective at preventing release of IFN-β from infected cells, but that uninfected neighboring cells could be induced to produce IFN-β (presumably through TLR binding), and so induce an antiviral IFN-β response. This antiviral response could be blocked by expression of B18R or by addition of anti–IFN-β neutralizing antibody (unpublished data), and was irrelevant in many cancer cells that were deficient in their ability to produce (C33A) or respond to (A2780 and C33A) IFN-β.
The oncolytic potential of WR-delB18R was confirmed in immunocompetent mice, with the virus rapidly removed from all tissues other than the tumor, and capable of producing 100% complete responses after local delivery. Antitumor effects were also seen following intravenous delivery, demonstrating the systemic potential of this virus. However, fewer complete responses were witnessed, indicating that an increase in tumor cell–killing potential for this virus (such as by transgene expression) may be needed for optimal systemic efficacy. Intravenous delivery also exposes more nonmalignant tissues and organs to the potential of viral infection, and so tumor-selectivity may become more critical for this delivery route. Although no toxicity was observed with WR-delB18R at therapeutic doses, deletion of intracellular IFN-resistance genes may lead to similar effects as B18R deletion, and the combination of both deletions may act together to further attenuate this virus in normal tissues, if necessary.
This new oncolytic vector was also shown to be capable of targeting and infecting human colorectal tumor explants ex vivo, and of destroying tumor cells by multiple mechanisms of action, one of which, to the best of our knowledge, has not been previously described. First, cancer-selective replication results in direct oncolysis. Second, as rechallenge of mice with tumors following complete responses to treatment resulted in tumor rejection, it appeared that the virus was capable of inducing an antitumor immune state within the animal. Induction of tumor-specific CTLs by oncolytic virus treatment was reported previously with HSV [31
], but has not been shown for vaccinia virus. Although the exact mechanisms have not been proven, they are likely to include recruitment of antigen-presenting cells, induction of immunostimulatory cytokines, and release of tumor-associated and viral antigens following cell lysis, leading to in situ vaccination against the tumor. Finally, we report the infection of tumor-associated endothelial cells by the oncolytic virus, resulting in reduced tumor vascularity. Tumor-associated endothelial cell lysis can lead to tissue factor release and intratumoral vascular thrombosis. Endothelial cells are attractive targets for oncolytic viruses, given their accessibility to infection by intravascular virus [32
]. Tumor-associated endothelial cells may be specifically susceptible to this vaccinia mutant for several reasons [33
]. First, these cells tend to be hyperproliferative, and therefore may be generally more susceptible to vaccinia infection. Second, epithelial growth factor (EGF) receptors are frequently expressed on these cells. Vaccinia replication is enhanced by EGF receptor binding and activation by vaccinia growth factor (VGF). However, further research is needed to elucidate the mechanisms involved and to take full advantage of this novel antitumoral approach. It will also be interesting to determine whether tumor–endothelial cell targeting occurs with other vaccinia virus mutants and/or other oncolytic viruses.
Recombinant IFN-β has been administered systemically for the treatment of several cancer types. The protein has been delivered by intramuscular, intravenous, or intratumoral routes, with common toxicities including myelosuppression, transaminitis, and neurotoxicity (include seizures), indicating that localized, tumor-specific delivery of the cytokine would be desirable. Antitumoral efficacy was reported, however, both in patients with brain tumors (including glioblastoma multiforme) [34
] and in a patient with colorectal carcinoma [36
]. IFN-β therefore represents a promising cytokine for use in cancer therapy. However, because the effects of the recombinant protein are locally mediated and are short-lived in vivo, and its systemic administration leads to toxicity, expression of IFN-β from a gene therapy or oncolytic virus within the tumor represents a promising means to apply this cytokine [37
]. However, previous approaches have suffered from a lack of targeted gene delivery [38
The expression of IFN-β from WR-delB18R
therefore represents a promising strategy. Transgene “arming” of oncolytic viruses has frequently been utilized to enhance antitumoral efficacy and for noninvasive imaging purposes [10
]. However, in addition to increasing the antitumor effects of the virus, IFN-β expression serves to further reduce viral replication in normal tissues. To date, vaccinia virus–expressed transgenes have not been utilized to inhibit viral replication and enhance clearance from normal tissues; because IFN-β has potent antiviral properties, we predicted viral inhibition would occur in this case. A similar strategy with IFN-α was recently described for an oncolytic adenovirus vector [39
]. However, this vector did not demonstrate systemic delivery or efficacy potential, and efficacy was limited even with multiple (more than ten) intratumoral injections. Vector replication and selectivity were not studied in normal nonimmortalized cells, and no primary human tissue was tested as reported in this study. Cancer selectivity was not studied in vivo, either, because tumor-free animals were studied for toxic effects to the liver only. In addition, because vaccinia genes expressed from early/late promoters will be expressed at low levels even during nonproductive infection of resistant cells, a small amount of IFN-β will be expressed in any normal tissues exposed to the virus, allowing the early production of an antiviral state. Of note, these levels are nontoxic, and high levels of gene expression are linked tightly to viral replication within tumor tissue. In this way, transgene expression will be linked to permissive infection, and thereby restricted to tumor cells.
gene was inserted into the TK
gene, as this deletion has also been demonstrated to be tumor targeting [25
]. The resulting virus, JX-795
, has deletions in both B18R
genes, and expressed IFN-β, as well as luciferase, for preclinical imaging purposes. It was shown to be highly specific for cancer cells in vitro, without the requirement for addition of exogenous cytokine. JX-795
was also found to produce high levels of IFN-β in vivo, which remained localized within the tumor. Viral gene expression was also highly tumor-restricted in vivo, and so this report represents the first description of a system for the systemic delivery of type-I IFN to tumors. Furthermore, this virus was capable of effectively destroying established tumors in mouse models. We therefore demonstrated that JX-795
is highly tumor selective and capable of potent antitumor effects in vivo. The highly tumor-restricted luciferase
gene expression seen with JX-795
indicates that this vector could also be used as a gene-delivery vehicle for any further therapeutic transgenes whose expression might lead to toxicity if expressed in any nonmalignant tissues.
In addition to B18R
, vaccinia virus expresses several other type-I IFN-resistance proteins. These include several intracellular proteins that prevent production of IFN from infected cells through inhibition of PKR (e.g., E3L and K3L), or blockade of nuclear factor-κB (NF-κB) activation and interferon-regulatory factor (IRF) signaling (A52R, A46R, and N1L). In addition, additional proteins are expressed that may prevent infected cells from responding to IFN by blocking STAT1 signaling (H1L) [20
]. Future research may demonstrate that the normal tissue clearance of JX-795
is further enhanced by deletion of one or more of these genes or regions within these genes. However, although clearance from normal tissues is advantageous, overly rapid clearance from tumor tissue may reduce efficacy. A balance will need to be achieved for future viruses derived from JX-795
The rational design of oncolytic viruses combining tumor-targeting viral deletions with specific transgenes capable of complimenting, or even synergizing with, the phenotype of the attenuated virus represents a promising strategy for the design of virotherapeutics. In addition, the potential to destroy the tumor by a multitude of different mechanisms, as seen with oncolytic vaccinia strains, and the targeting of not only the malignant cells, but also other cells (e.g., endothelial and immune cells) within the tumor environment, may be the most effective approach to applying biological therapies.