The development of attenuated viruses that are adapted for preferential replication in solid tumors is an attractive approach to treatment of malignancies where more standard therapies are either ineffective or difficult to apply. HSV-1-oncolytic vectors used in early-phase clinical trials for the treatment of GBM have shown some success without serious side effects [33
]. Tumor specificity can be achieved by deleting viral genes that permit mutant virus replication in tumor cells while profoundly impairing virus replication in normal host cells [1
]. The vector prototype is G207, that is, deleted for γ
34.5 and produces a nonfunctional ICP6-LacZ fusion protein [35
]. ICP6 encodes the large subunit of the viral ribonucleotide reductase, a protein that permits virus growth in nondividing cells by maintaining the nucleotide pool, while γ
34.5 counteracts the virus-induced activation of the PKR pathway. The more advanced oncolytic vector examined in this study, MGH2, was derived from G207 by replacing lacZ
with eGFP at the ICP6 locus and inserting two antitumor genes, CYP2B1
(encoding cytochrome p450) and shiCE
(encoding secreted human intestinal carboxylesterase) [32
]. These enzymes activate the anticancer drugs cyclophosphamide and irinotecan, respectively, both of which are potent tumor toxic products. In one study, MGH2 showed oncolytic activity in vivo
only with the addition of cyclophosphamide and irinotecan [32
], indicating that the MGH2 vector backbone alone does not function as an effective OV. Some evidence suggests that vector replication in certain tumor cells may require γ
34.5 activity and that oHSV is susceptible to innate immune responses, potentially limiting the effectiveness of this and other oHSV vectors [36
]. We therefore sought to examine other mutant backbones that may overcome these limitations.
In this study we characterize an oHSV vector (JD0G) that is deleted for ICP0 and the joint elements of the viral genome. HSV-1 mutants defective for the production of ICP0 protein provide an attractive alternative to the MGH2 vector backbone. ICP0-deficient vectors are impaired for growth at low MOI in most cell lines, while replication preferentially occurs in certain tumor cell lines [22
]. Moreover, Hummel and colleagues have reported that an HSV-1 double mutant lacking VP16 and ICP0 protein expression replicates efficiently in a variety of tumor cells derived from prostate, lung, colon, and mammary carcinomas with evidence of oncolytic activity in animal models [24
]. Using the JD0G backbone, we sought to determine whether the unique interaction of an ICP0-deleted vector with tumor cells could be exploited to develop novel oncolytic viruses suitable for treatment of brain tumors. Our findings showed that JD0G replication was significantly enhanced in the U251 and SNB19 glioblastoma cell lines compared to HEL cells, while MGH2 replication was highly impaired, showing 2-3 logs lower yields on the glioblastoma lines than JD0G. These findings point to the possibility that genetic changes related to glioblastoma development relieve the need for ICP0 expression and permit JD0G replication, while the mutations in MGH2 are detrimental to vector growth and not complemented in these lines. Furthermore, the difference in replication between these two vectors was most pronounced at low multiplicity (data not shown), a condition that may be more relevant to circumstances in vivo
than high-MOI infections in cell culture. An oHSV vector that can replicate efficiently from a small number of initial infectious particles is most likely to be effective in tumors.
It is likely that the tumor-specific replication of the JD0G vector can be ascribed to the ICP0 deficiency and not to the removal of the joint components of the viral genome. The HSV genome consists of two unique elements (UL
) that are each flanked by repeated sequences (ab/b′a′; a′c′/ca) (). This arrangement creates two terminal repeat regions that together contain one copy of the genes for ICP34.5, ICP4, ICP0, and the latency-associated transcript (LAT), and an internal repeat element (joint) that contains a second copy of each gene. Thus, the joint deletion generated in the construction of JD0G removes a single copy of each gene, reducing but not eliminating gene expression. This modification alone has a minimal effect on virus replication (unpublished observations, [38
]). The benefit of this deletion is that it eliminates the possibility of genome isomerization events that typically occur between the repeated elements of the viral genome and generate four possible HSV isomers. Hence, the JD0G genome has enhanced stability over the wild-type KOS genome.
Following inoculation of oHSV in vivo,
can be produced by resident microglia, recruited macrophages, dendritic and NK cells at the growing tumor site [25
]. In response to IFNγ
, dendritic cells and tumor cells may produce IDO, whose enzymatic activity leads to the degradation of tryptophan and the production of toxic catabolites such as kynurenine and quinolinate [40
]. Additionally, tumor cells including glioblastoma cells have been reported to express IDO without IFNγ
]. Tryptophan depletion can impede the production of viral proteins and, together with kynurenine, can induce effector T-cell anergy [30
]. Therefore, IFNγ
can negatively affect both oncolytic virus replication and tumor-specific immunity at several levels. Although IFNγ
-mediated inhibition of oHSV replication may be overcome in part by the administration of immunosuppressive drugs such as cyclophosphamide, a drawback to this approach is that it will also inhibit the induction of tumor-specific immunity.
In view of these considerations, and to explore the tumor-related changes that may potentially influence the innate immune response to virus infection, KOS and JD0G viral replication in glioblastoma cells were tested for their sensitivity to IFNγ treatment. Our data demonstrate that both viruses were sensitive to IFNγ at low MOI, but the ability of IFNγ to control virus replication was largely lost at elevated MOI. However, we observed marked differences between the glioblastoma cell lines in both the IFNγ sensitivity of virus replication and the involvement of IDO activity in the inhibition of virus replication. SNB19 cells produced lower levels of IDO protein than U251 cells, and virus replication in SNB19 cells was less sensitive to IFNγ treatment. In the IFNγ-hypersensitive U251 cells, both KOS and JD0G were found to downregulate, but not eliminate, IDO mRNA and protein, and the IDO inhibitor 1-MT was able to improve the efficiency of virus replication. In contrast, both KOS and JD0G infection nearly eliminated IDO expression in IFNγ-treated SNB19 cells, and 1-MT did not improve virus replication. Together, these results indicate that IDO expression above a certain level can inhibit virus replication in glioblastoma cells, but also that the majority of IFNγ-induced inhibition of virus replication occurs via a different pathway. The limited role of IDO in controlling virus replication may be attributable to the virus-mediated downregulation of IDO observed on infection with both KOS and JD0G. This downregulation may not only allow for vector replication in the presence of IFNγ, but also minimize adverse effects on antitumor immunity. Thus, we propose that vector-induced IDO downregulation may be an important aspect of the therapeutic potential of oHSV vectors.