Serotype-Specific Killing of Large Cell Carcinoma Cells by Reovirus

Reovirus is under development as a therapeutic for numerous types of cancer. In contrast to other oncolytic viruses, the safety and efficacy of reovirus have not been improved through genetic manipulation. Here, we tested the oncolytic capacity of recombinant strains (rs) of prototype reovirus laboratory strains T1L and T3D (rsT1L and rsT3D, respectively) in a panel of non-small cell lung cancer (NSCLC) cell lines. We found that rsT1L was markedly more cytolytic than rsT3D in the large cell carcinoma cell lines tested, whereas killing of adenocarcinoma cell lines was comparable between rsT1L and rsT3D. Importantly, non-recombinant T1L and T3D phenocopied the kinetics and magnitude of cell death induced by recombinant strains. We identified gene segments L2, L3, and M1 as viral determinants of strain-specific differences cell killing of the large cell carcinoma cell lines. Together, these results indicate that recombinant reoviruses recapitulate the cell killing properties of non-recombinant, tissue culture-passaged strains. These studies provide a baseline for the use of reverse genetics with the specific objective of engineering more effective reovirus oncolytics. This work raises the possibility that type 1 reoviruses may have the capacity to serve as more effective oncolytics than type 3 reoviruses in some tumor types.


Introduction
Oncolytic viruses are an emerging class of cancer therapeutics that selectively replicate in and kill transformed cells while sparing non-transformed cells [1,2]. Mammalian orthoreovirus (reovirus) is one of a number of viruses with oncolytic potential that are in different stages of development [2][3][4][5][6]. Reoviruses are non-enveloped viruses with a segmented double-stranded RNA genome that productively infect and lyse several different tumor cell types in vitro and in experimental animal models [3,4,7]. In some cases, reovirus shows efficacy as a virotherapeutic agent for aggressive and refractory human tumors [8]. Phases I and II clinical trials in the United States, Canada, and Europe demonstrate that pelareorep (Reolysin), a first-generation reovirus therapeutic is tolerated and safe, even in patients with advanced cancer who have undergone extensive chemotherapy [9,10]. A Phase III clinical trial to test reovirus efficacy against head and neck cancers was recently completed.
Phase I and II trials indicate that the majority of oncolytic viruses under development, including reovirus, are safe for use in humans [11]. However, the therapeutic efficacy of many of these agents is limited [2]. The therapeutic potency of oncolytic viruses can be enhanced by genetic modifications that increase viral replication in and killing of cancer cells, re-target the virus to infect new tumor types, or introduce immunomodulatory factors that enhance immune-mediated killing of tumors [2]. For example, talimogene laherparepvec (T-VEC), which in 2015 became the first oncolytic virus approved by the US Food and Drug Administration (FDA) for use in humans, is a genetically-modified carcinoma cell lines. This finding indicates that differences in viral replication do not underlie the strain-specific disparity in killing of the large cell carcinoma cell lines. Finally, we identified reovirus gene segments, L2, L3, and M1 as viral determinants of differential cell killing of large cell carcinoma cell lines. Together, our results indicate that that the oncolytic potency of reoviruses generated by reverse genetics is comparable to non-recombinant strains. Our results further suggest that a T1L-based reovirus may be more effective therapeutics than T3D-derived viruses in certain cell or tumor types.

Recombinant Reovirus Strains rsT1L and rsT3D Differ in the Capacity to Kill Large Cell Carcinoma Cell Lines
To assess the oncolytic capacity of recombinant reoviruses, we infected a panel of NSCLC cell lines (Table 1) with recombinant (rsT1L or rsT3D) or non-recombinant (T1L or T3D) reoviruses. The panel included two large cell carcinoma lines (H661 and H1299) and three adenocarcinoma lines (H1437, H1563, and H1975) ( Table 1). Each cell line was mock infected or infected with rsT1L, rsT3D, T1L, or T3D at multiplicities of infection (MOIs) of 10, 100, or 1000 plaque forming units (PFU)/cell and ATP content was measured at 24, 48, and 72 h post-infection as an indicator of cell viability (Figure 1). We observed a dramatic difference in cell viability between rsT1L and rsT3D in the two large cell carcinoma cell lines ( Figure 1A,B). In H661 cells, each dose of rsT1L induced more cell death than rsT3D at 48 and 72 h. The peak loss of cell viability following rsT3D infection was approximately 40% of control, whereas rsT1L reduced cell viability by greater than 95%. In H1299 cells, viability of rsT3D-infected cells never dropped below 60% of control, whereas rsT1L caused almost complete cell death at MOIs of 100 and 1000 PFU/cell by 72 h. In contrast to the large cell carcinoma cell lines, rsT1L and rsT3D induced comparable dose-and time-dependent cell killing in the adenocarcinoma cell lines ( Figure 1C-E). Importantly, cell death induced in each NSCLC line by non-recombinant T1L and T3D phenocopied the cell killing induced by rsT1L and rsT3D, including enhanced killing of the two large cell carcinoma cell lines (H661 and H1299) by T1L compared to T3D ( Figure 1A,B). These results indicate that the cell killing capacity of reovirus generated by plasmid-based reverse genetics does not differ from that of native, non-recombinant strains. These data also revealed that the two large cell carcinoma lines were more resistant to killing by rsT3D than rsT1L. These findings suggest that T1 reoviruses may be more effective oncolytics than T3 reoviruses in certain tumor types, including large cell carcinomas.

Reovirus Infectivity in NSCLC Cell Lines
We performed a fluorescent focus assay to assess reovirus infectivity in the NSCLC cell panel ( Figure 2). Each cell line was infected with rsT1L or rsT3D at an MOI of 100 PFU/cell, which was the median viral dose tested in the cell killing experiments. At 18 h post-infection, the cells were fixed Error bars indicate standard deviation (SD). * p < 0.05; ** p < 0.01; *** p < 0.001 as determined for rsT1L vs. rsT3D by Student's t-test. @, p < 0.05; @@, p < 0.01; @@@, p < 0.001 as determined for T1L vs. T3D by Student's t-test. The data shown is compiled from three independent experiments.

Reovirus Infectivity in NSCLC Cell Lines
We performed a fluorescent focus assay to assess reovirus infectivity in the NSCLC cell panel ( Figure 2). Each cell line was infected with rsT1L or rsT3D at an MOI of 100 PFU/cell, which was the median viral dose tested in the cell killing experiments. At 18 h post-infection, the cells were fixed and stained using reovirus-specific polyclonal antiserum and DAPI. A higher percentage of H661, H1473, H1563, and H1975 cells were infected by rsT3D compared to rsT1L. H1299 cells were the only cell line that rsT1L infected to a greater degree than rsT3D (~80% vs.~60%). The largest difference was observed in H1437 cells, where rsT3D infected an approximately 2-fold greater number of cells than rsT1L. In H661, H1299, and H1563 lines, the majority of cells were infected by both reovirus strains (>60%). In contrast, a low percentage (<40%) of H1437 and H1975 cells were infected by rsT1L or rsT3D. These findings indicate that each lung cancer cell lines tested is susceptible to infection by recombinant reoviruses. It is of note that although a higher percentage of H661 cells were infected with rsT3D than rsT1L, rsT1L induced markedly more cell death in H661 cells than rsT3D. Thus, it is unlikely that the strain-specific difference in killing of H661 cells is due to differential infectivity between the two viruses.
Viruses 2017, 9, 140 5 of 19 and stained using reovirus-specific polyclonal antiserum and DAPI. A higher percentage of H661, H1473, H1563, and H1975 cells were infected by rsT3D compared to rsT1L. H1299 cells were the only cell line that rsT1L infected to a greater degree than rsT3D (~80% vs. ~60%). The largest difference was observed in H1437 cells, where rsT3D infected an approximately 2-fold greater number of cells than rsT1L. In H661, H1299, and H1563 lines, the majority of cells were infected by both reovirus strains (>60%). In contrast, a low percentage (<40%) of H1437 and H1975 cells were infected by rsT1L or rsT3D. These findings indicate that each lung cancer cell lines tested is susceptible to infection by recombinant reoviruses. It is of note that although a higher percentage of H661 cells were infected with rsT3D than rsT1L, rsT1L induced markedly more cell death in H661 cells than rsT3D. Thus, it is unlikely that the strain-specific difference in killing of H661 cells is due to differential infectivity between the two viruses. Although rsT1L and rsT3D infectivity did not differ in H661 cells, rsT1L infected H1299 cells more efficiently than rsT3D. A greater level of infection by rsT1L could result in more killing of H1299 cells than rsT3D. To test whether the disparity in infectivity caused strain-specific differences in killing of H1299 cells, we performed cell viability assays in which the infectious dose of rsT1L and rsT3D was normalized based on fluorescent focus unit (FFU) titers that were determined on H1299 cells. We infected H1299 cells with 10,000, 1000, or 100 FFU/cell and measured ATP at 72 h post-infection as an indicator of cell viability ( Figure 3). For reovirus, the FFU:PFU ratio is approximately 10:1 [36]. Thus, the FFU doses tested approximate the MOI range used in experiments initiated using PFUs. We found that when infection was normalized by infectivity, rsT1L retained the capacity to kill H1299 cells to a greater degree than rsT3D ( Figure 3A). H1563 cells, which were equally susceptible to rsT1L and rsT3D infection, were killed to comparable levels by rsT1L and T3D when infection was initiated based on normalize FFU titers that were determined on H1563 cells ( Figure 3B). This result indicates that differences in virus infectivity do not lead to strain-specific differences in killing of H1299 cells. Error bars indicate SD. * p < 0.05 difference between rsT1L and rsT3D as determined by Student's t-test. The data presented is compiled from two independent experiments. Although rsT1L and rsT3D infectivity did not differ in H661 cells, rsT1L infected H1299 cells more efficiently than rsT3D. A greater level of infection by rsT1L could result in more killing of H1299 cells than rsT3D. To test whether the disparity in infectivity caused strain-specific differences in killing of H1299 cells, we performed cell viability assays in which the infectious dose of rsT1L and rsT3D was normalized based on fluorescent focus unit (FFU) titers that were determined on H1299 cells. We infected H1299 cells with 10,000, 1000, or 100 FFU/cell and measured ATP at 72 h post-infection as an indicator of cell viability ( Figure 3). For reovirus, the FFU:PFU ratio is approximately 10:1 [36]. Thus, the FFU doses tested approximate the MOI range used in experiments initiated using PFUs. We found that when infection was normalized by infectivity, rsT1L retained the capacity to kill H1299 cells to a greater degree than rsT3D ( Figure 3A). H1563 cells, which were equally susceptible to rsT1L and rsT3D infection, were killed to comparable levels by rsT1L and T3D when infection was initiated based on normalize FFU titers that were determined on H1563 cells ( Figure 3B). This result indicates that differences in virus infectivity do not lead to strain-specific differences in killing of H1299 cells.

rsT1L and rsT3D Gene Expression and Replication in NSCLC Cell Lines
To assess viral protein production in the NSCLC cell panel, each cell line was mock infected or infected with rsT1L or rsT3D at an MOI of 10 PFU/cell ( Figure 4). Whole cell lysates prepared at 6, 12, 18, or 24 h post-infection were separated by SDS-PAGE and immunoblotted with reovirus-specific polyclonal antiserum. Reovirus proteins were detected in each cell line by 18 h. In H1299 and H1975 cells rsT1L produced modestly higher viral protein levels than rsT3D. In H1299 cells, rsT1L produced 3.8-and 3.6fold more μ1 than rsT3D at 18 h and 24 h, respectively. Approximately 22-fold more σ3 was detected in rsT1L-infected cells relative to rsT3D at 18 h, but the difference decreased to 3.7-fold at 24 h. In H1975 cells, rsT1L produced 3-and 3.8-fold more μ1 at 18 h and 24 h, respectively. The differences in σ3 were 3.7-fold at 18 h and 2-fold at 24 h. For H1299, the difference could result from an increased number of cells becoming infected by rsT1L compared to rsT3D. However, no difference in reovirus gene expression was observed between rsT1L and rsT3D in H661, H1437, or H1563 cells. Together, these results demonstrate that recombinant reoviruses efficiently express viral proteins in NSCLC cell lines.

rsT1L and rsT3D Gene Expression and Replication in NSCLC Cell Lines
To assess viral protein production in the NSCLC cell panel, each cell line was mock infected or infected with rsT1L or rsT3D at an MOI of 10 PFU/cell ( Figure 4). Whole cell lysates prepared at 6, 12, 18, or 24 h post-infection were separated by SDS-PAGE and immunoblotted with reovirus-specific polyclonal antiserum. Reovirus proteins were detected in each cell line by 18 h. In H1299 and H1975 cells rsT1L produced modestly higher viral protein levels than rsT3D. In H1299 cells, rsT1L produced 3.8-and 3.6-fold more µ1 than rsT3D at 18 h and 24 h, respectively. Approximately 22-fold more σ3 was detected in rsT1L-infected cells relative to rsT3D at 18 h, but the difference decreased to 3.7-fold at 24 h. In H1975 cells, rsT1L produced 3-and 3.8-fold more µ1 at 18 h and 24 h, respectively. The differences in σ3 were 3.7-fold at 18 h and 2-fold at 24 h. For H1299, the difference could result from an increased number of cells becoming infected by rsT1L compared to rsT3D. However, no difference in reovirus gene expression was observed between rsT1L and rsT3D in H661, H1437, or H1563 cells. Together, these results demonstrate that recombinant reoviruses efficiently express viral proteins in NSCLC cell lines.

rsT1L and rsT3D Gene Expression and Replication in NSCLC Cell Lines
To assess viral protein production in the NSCLC cell panel, each cell line was mock infected or infected with rsT1L or rsT3D at an MOI of 10 PFU/cell ( Figure 4). Whole cell lysates prepared at 6, 12, 18, or 24 h post-infection were separated by SDS-PAGE and immunoblotted with reovirus-specific polyclonal antiserum. Reovirus proteins were detected in each cell line by 18 h. In H1299 and H1975 cells rsT1L produced modestly higher viral protein levels than rsT3D. In H1299 cells, rsT1L produced 3.8-and 3.6fold more μ1 than rsT3D at 18 h and 24 h, respectively. Approximately 22-fold more σ3 was detected in rsT1L-infected cells relative to rsT3D at 18 h, but the difference decreased to 3.7-fold at 24 h. In H1975 cells, rsT1L produced 3-and 3.8-fold more μ1 at 18 h and 24 h, respectively. The differences in σ3 were 3.7-fold at 18 h and 2-fold at 24 h. For H1299, the difference could result from an increased number of cells becoming infected by rsT1L compared to rsT3D. However, no difference in reovirus gene expression was observed between rsT1L and rsT3D in H661, H1437, or H1563 cells. Together, these results demonstrate that recombinant reoviruses efficiently express viral proteins in NSCLC cell lines.  To assess viral replication of the NSCLC cells, we infected each cell line with rsT1L or rsT3D at an MOI of 1 PFU/cell and quantified viral titers at 0, 24, and 48 h post-infection ( Figure 5). We used a lower multiplicity infection for assessing viral replication than cell killing (1 PFU/cell versus 10-1000 PFU/cell) to provide a larger dynamic range in which to measure viral yields. We found that rsT1L and rsT3D replicated in each lung cancer cell line, with progeny yields ranging from 100-to 10,000-fold increase over input. We found that rsT1L and rsT3D produced comparable yields in H661 and H1299 cells. This result indicates that differences viral in replication do not account for differential killing of H661 and H1299 cells by rsT1L and rsT3D. H1437 cells produced the lowest viral yields of the cell lines tested. Yields of rsT1L and rsT3D were equivalent in H1437 cells at 24 h. However, rsT3D produced higher progeny yields than rsT1L at 48 h. Although rsT1L and rsT3D replicated in H1563 and H1975, rsT1L produced higher progeny yields than rsT3D in these cell lines. These results indicate that recombinant reoviruses replicate efficiently in all of the NSCLC cell lines tested. To assess viral replication of the NSCLC cells, we infected each cell line with rsT1L or rsT3D at an MOI of 1 PFU/cell and quantified viral titers at 0, 24, and 48 h post-infection ( Figure 5). We used a lower multiplicity infection for assessing viral replication than cell killing (1 PFU/cell versus 10-1000 PFU/cell) to provide a larger dynamic range in which to measure viral yields. We found that rsT1L and rsT3D replicated in each lung cancer cell line, with progeny yields ranging from 100-to 10,000fold increase over input. We found that rsT1L and rsT3D produced comparable yields in H661 and H1299 cells. This result indicates that differences viral in replication do not account for differential killing of H661 and H1299 cells by rsT1L and rsT3D. H1437 cells produced the lowest viral yields of the cell lines tested. Yields of rsT1L and rsT3D were equivalent in H1437 cells at 24 h. However, rsT3D produced higher progeny yields than rsT1L at 48 h. Although rsT1L and rsT3D replicated in H1563 and H1975, rsT1L produced higher progeny yields than rsT3D in these cell lines. These results indicate that recombinant reoviruses replicate efficiently in all of the NSCLC cell lines tested.

Recombinant Reoviruses Kill NSCLC Cell Lines by a Caspase-Independent Mechanism
In many cell types, including L929 and HeLa cells, T3 reoviruses induce higher levels of apoptosis than T1 reoviruses [25]. To determine whether differences in apoptosis underlie differential cell killing by rsT1L and rsT3D, we assessed caspase-3/7 activity ( Figure 6) and poly-ADP ribose polymerase (PARP) cleavage ( Figure 7) in each NSCLC cell line following rsT1L and rsT3D infection. Tumor necrosis factor-α/cycloheximide (TNF-α/CHX) induced caspase-3/7 and PARP cleavage in each NSCLC cell line, indicating that apoptotic responses are functional in the cell panel. Infection with rsT1L or rsT3D did not elicit caspase-3/7 activity ( Figure 6A,B) or PARP cleavage ( Figure 7A,B) in the two large cell carcinoma lines (H661 and H1299). We detected modest caspase-3/7 activity in H1437 cells following infection with the highest multiplicity of rsT3D ( Figure 6C). However, minimal PARP cleavage was observed in this cell line following reovirus infection ( Figure 7C). In H1563 cells, rsT1L and rsT3D induced caspase-3/7 activity in a dose-dependent manner at 24 and 48 h (Figure 6d). However, rsT1L and rsT3D did not induce PARP cleavage in H1563 cells ( Figure 7D). In H1975 cells, rsT1L and rsT3D elicited caspase-3/7 activity at 48 h ( Figure 6E), but minimal PARP cleavage at all of the time points tested ( Figure 7E).

Recombinant Reoviruses Kill NSCLC Cell Lines by a Caspase-Independent Mechanism
In many cell types, including L929 and HeLa cells, T3 reoviruses induce higher levels of apoptosis than T1 reoviruses [25]. To determine whether differences in apoptosis underlie differential cell killing by rsT1L and rsT3D, we assessed caspase-3/7 activity ( Figure 6) and poly-ADP ribose polymerase (PARP) cleavage (Figure 7) in each NSCLC cell line following rsT1L and rsT3D infection. Tumor necrosis factor-α/cycloheximide (TNF-α/CHX) induced caspase-3/7 and PARP cleavage in each NSCLC cell line, indicating that apoptotic responses are functional in the cell panel. Infection with rsT1L or rsT3D did not elicit caspase-3/7 activity ( Figure 6A,B) or PARP cleavage ( Figure 7A,B) in the two large cell carcinoma lines (H661 and H1299). We detected modest caspase-3/7 activity in H1437 cells following infection with the highest multiplicity of rsT3D ( Figure 6C). However, minimal PARP cleavage was observed in this cell line following reovirus infection ( Figure 7C). In H1563 cells, rsT1L and rsT3D induced caspase-3/7 activity in a dose-dependent manner at 24 and 48 h ( Figure 6D). However, rsT1L and rsT3D did not induce PARP cleavage in H1563 cells ( Figure 7D). In H1975 cells, rsT1L and rsT3D elicited caspase-3/7 activity at 48 h ( Figure 6E), but minimal PARP cleavage at all of the time points tested ( Figure 7E). To directly test whether caspases are required for reovirus killing of NSCLC cells, we assessed the effect of the pan-caspase inhibitor Z-VAD-FMK (Z-VAD) on reovirus cell killing (Figure 8). Each lung cancer cell line was infected with rsT1L or rsT3D at an MOI of 100 PFU/cell in the absence or presence of 25 μM Z-VAD and ATP content was measured at 48 h post-infection as an indicator of cell viability. TNF/CHX-induced caspase-3/7 activity was completely inhibited in each cell line by both Z-VAD concentrations, indicating that the dose of Z-VAD tested was functional. As before, rsT1L killed H661 and H1299 cells to a greater extent than rsT3D ( Figure 8A,B). Similarly, rsT1L and rsT3D induced comparable levels of cell death in H1437, H1563, and H1975 cells ( Figure 8C-E). We To directly test whether caspases are required for reovirus killing of NSCLC cells, we assessed the effect of the pan-caspase inhibitor Z-VAD-FMK (Z-VAD) on reovirus cell killing (Figure 8). Each lung cancer cell line was infected with rsT1L or rsT3D at an MOI of 100 PFU/cell in the absence or presence of 25 µM Z-VAD and ATP content was measured at 48 h post-infection as an indicator of cell viability. TNF/CHX-induced caspase-3/7 activity was completely inhibited in each cell line by both Z-VAD concentrations, indicating that the dose of Z-VAD tested was functional. As before, rsT1L killed H661 and H1299 cells to a greater extent than rsT3D ( Figure 8A,B). Similarly, rsT1L and rsT3D induced comparable levels of cell death in H1437, H1563, and H1975 cells ( Figure 8C-E). We found that Z-VAD had little effect on killing by rsT1L or rsT3D in any of the NSCLC cell lines. Together, these results indicate that caspase activity is dispensable for reovirus-induced cell death in the NSCLC cell lines tested. These findings also suggest that strain-specific differences in apoptosis induction are not likely the underlying reason for the disparity in cell killing between rsT1L and rsT3D in H661 and H1299 cells. found that Z-VAD had little effect on killing by rsT1L or rsT3D in any of the NSCLC cell lines. Together, these results indicate that caspase activity is dispensable for reovirus-induced cell death in the NSCLC cell lines tested. These findings also suggest that strain-specific differences in apoptosis induction are not likely the underlying reason for the disparity in cell killing between rsT1L and rsT3D in H661 and H1299 cells.   found that Z-VAD had little effect on killing by rsT1L or rsT3D in any of the NSCLC cell lines. Together, these results indicate that caspase activity is dispensable for reovirus-induced cell death in the NSCLC cell lines tested. These findings also suggest that strain-specific differences in apoptosis induction are not likely the underlying reason for the disparity in cell killing between rsT1L and rsT3D in H661 and H1299 cells.

Reovirus Gene Segments L2, L3, and M1 Correlate with Strain-Specific Differences in Cell Killing
To identify reovirus gene segments responsible for differential killing of the two large cell carcinoma lines by rsT1L and rsT3D, we used reverse genetics to engineer reciprocal panels of T1L × T3D single-gene reassortant viruses. We generated a complete set of 20 single-gene reassortant viruses in which all ten gene segments were individually replaced in rsT1L or rsT3D genetic backgrounds. We infected H661 and H1299 cells with rsT1L, rsT3D, or the reassortant viruses at an MOI of 100 PFU/cell and measured ATP content at 72 h post-infection as an indicator of cell viability (Figure 9). In both cell lines, the parental viruses retained the serotype-specific differences in cell killing described above (Figure 1). No single gene from T3D reduced the cell killing capacity of rsT1L to levels observed for rsT3D in either cell line. In H661 cells, replacement of T1L L2, L3, M1, M3, S3, or S4 gene with the T3D allele decreased killing by viruses with an otherwise T1L genetic background ( Figure 9A, upper panel). In H1299 cells, all of the individual T3D gene replacements except for M2 decreased the cell killing capacity of the virus ( Figure 9B, upper panel). In both H661 and H1299 cells, T1L genes L2, L3, and M1 enhanced cell killing of an otherwise T3D virus, although not to the level of rsT1L ( Figure 9A,B, lower panels). These findings indicate that the L2, L3, and M1 gene segments have the highest correlation with differences between rsT1L-and rsT3D-mediated cell killing in H661 and H1299 cells.

Reovirus Gene Segments L2, L3, and M1 Correlate with Strain-Specific Differences in Cell Killing
To identify reovirus gene segments responsible for differential killing of the two large cell carcinoma lines by rsT1L and rsT3D, we used reverse genetics to engineer reciprocal panels of T1L × T3D single-gene reassortant viruses. We generated a complete set of 20 single-gene reassortant viruses in which all ten gene segments were individually replaced in rsT1L or rsT3D genetic backgrounds. We infected H661 and H1299 cells with rsT1L, rsT3D, or the reassortant viruses at an MOI of 100 PFU/cell and measured ATP content at 72 h post-infection as an indicator of cell viability (Figure 9). In both cell lines, the parental viruses retained the serotype-specific differences in cell killing described above (Figure 1). No single gene from T3D reduced the cell killing capacity of rsT1L to levels observed for rsT3D in either cell line. In H661 cells, replacement of T1L L2, L3, M1, M3, S3, or S4 gene with the T3D allele decreased killing by viruses with an otherwise T1L genetic background ( Figure 9A, upper panel). In H1299 cells, all of the individual T3D gene replacements except for M2 decreased the cell killing capacity of the virus ( Figure 9B, upper panel). In both H661 and H1299 cells, T1L genes L2, L3, and M1 enhanced cell killing of an otherwise T3D virus, although not to the level of rsT1L ( Figure 9A,B, lower panels). These findings indicate that the L2, L3, and M1 gene segments have the highest correlation with differences between rsT1L-and rsT3D-mediated cell killing in H661 and H1299 cells. We next assessed replication of L2, L3, and M1 single-gene reassortment viruses in H661 and H1299 cells (Figure 10). We focused on L2, L3, and M1 because those genes exhibited the highest correlation with differences in cell killing in the reassortant panel. In H661 cells, rsT1L/T3D-L3 and rsT1L/T3D-M1 produced lower yields at 24 h relative to rsT1L. However, only rsT1L/T3D-M1 was reduced significantly at 48 h. In H1299 cells, only replication of rsT1L/T3D-L3 was substantially lower We next assessed replication of L2, L3, and M1 single-gene reassortment viruses in H661 and H1299 cells (Figure 10). We focused on L2, L3, and M1 because those genes exhibited the highest correlation with differences in cell killing in the reassortant panel. In H661 cells, rsT1L/T3D-L3 and rsT1L/T3D-M1 produced lower yields at 24 h relative to rsT1L. However, only rsT1L/T3D-M1 was reduced significantly at 48 h. In H1299 cells, only replication of rsT1L/T3D-L3 was substantially lower than rsT1L at 24 h. However, yields for all three T3D single gene replacements (rsT1L/T3D-L2, rsT1L/T3D-L3, and rsT1L/T3D-M1) were at least 10-fold reduced compared to wild type rsT1L at 48 h. In the T3D genetic background, only the virus with the T1L L2 gene replicated to higher levels than rsT3D at 48 h in both H661 and H1299 cells. These findings indicate that reovirus replication in the large cell carcinoma cell lines is influenced by the genetic complement of the virus.
Viruses 2017, 9, 140 11 of 19 than rsT1L at 24 h. However, yields for all three T3D single gene replacements (rsT1L/T3D-L2, rsT1L/T3D-L3, and rsT1L/T3D-M1) were at least 10-fold reduced compared to wild type rsT1L at 48 h. In the T3D genetic background, only the virus with the T1L L2 gene replicated to higher levels than rsT3D at 48 h in both H661 and H1299 cells. These findings indicate that reovirus replication in the large cell carcinoma cell lines is influenced by the genetic complement of the virus.

Discussion
Our results indicate that recombinant reoviruses effectively kill NSCLC cell lines (Table 2). Importantly, cell killing by non-recombinant reoviruses was indistinguishable from cell death induced by recombinant strains, indicating that there are not significant differences in cell death induction between recombinant and native reoviruses. These results suggest that use of recombinant reoviruses to treat cancer can be as effective as the native strains. We also found that rsT1L killed the large cell carcinoma cell lines tested to a greater extent than rsT3D, suggesting that rsT1L or T1-based vectors may be more effective oncolytics than T3-based viruses in certain tumor types. A similar observation was made in a mouse mammary tumor line, where T1L induced more cell death than T3D, indicating that differential susceptibility of cancer cells to reovirus strains is not limited to the cells tested in this study [37].
It remains to be determined why the large cell carcinoma lines (H661 and H1299) differed in susceptibility to reovirus killing (Figure 1). One possibility is that H661 and H1229 cells harbor mutations in cancer-related genes or pathways that differ from those in the adenocarcinoma lines (H1437, H1563, and H1975). H1299 cells contain a mutation in the NRAS gene. H661 cells harbor mutations in the genes for CDKN2A, SMARCA4, and TP53. All three adenocarcinoma cell lines contain mutations in the CDKN2A gene, H1437 and H1975 cells contain mutations in the TP53 gene, and H1975 cells have mutations in the EGFR and PIK3CA genes. It is possible that mutations in other cancer-related genes associate with strain-specific differences in cell killing. However, based on the current information available, no common genetic marker correlates with differences in cell death induction between rsT1L and rsT3D in H661 and H1299 cells. A second possibility is whether the cell lines derive from primary or metastatic tumors (Table 1). Large cell carcinoma cell lines H661 and H1299 were isolated from metastatic growths, whereas adenocarcinoma cell lines H1563 and H1975 derive from a primary tumor mass. However, H1437 cells were generated from a lung metastasis. Moreover, data from mouse models and clinical trials indicate that T3D (Reolysin) can target and kill

Discussion
Our results indicate that recombinant reoviruses effectively kill NSCLC cell lines (Table 2). Importantly, cell killing by non-recombinant reoviruses was indistinguishable from cell death induced by recombinant strains, indicating that there are not significant differences in cell death induction between recombinant and native reoviruses. These results suggest that use of recombinant reoviruses to treat cancer can be as effective as the native strains. We also found that rsT1L killed the large cell carcinoma cell lines tested to a greater extent than rsT3D, suggesting that rsT1L or T1-based vectors may be more effective oncolytics than T3-based viruses in certain tumor types. A similar observation was made in a mouse mammary tumor line, where T1L induced more cell death than T3D, indicating that differential susceptibility of cancer cells to reovirus strains is not limited to the cells tested in this study [37].
It remains to be determined why the large cell carcinoma lines (H661 and H1299) differed in susceptibility to reovirus killing (Figure 1). One possibility is that H661 and H1229 cells harbor mutations in cancer-related genes or pathways that differ from those in the adenocarcinoma lines (H1437, H1563, and H1975). H1299 cells contain a mutation in the NRAS gene. H661 cells harbor mutations in the genes for CDKN2A, SMARCA4, and TP53. All three adenocarcinoma cell lines contain mutations in the CDKN2A gene, H1437 and H1975 cells contain mutations in the TP53 gene, and H1975 cells have mutations in the EGFR and PIK3CA genes. It is possible that mutations in other cancer-related genes associate with strain-specific differences in cell killing. However, based on the current information available, no common genetic marker correlates with differences in cell death induction between rsT1L and rsT3D in H661 and H1299 cells. A second possibility is whether the cell lines derive from primary or metastatic tumors (Table 1). Large cell carcinoma cell lines H661 and H1299 were isolated from metastatic growths, whereas adenocarcinoma cell lines H1563 and H1975 derive from a primary tumor mass. However, H1437 cells were generated from a lung metastasis. Moreover, data from mouse models and clinical trials indicate that T3D (Reolysin) can target and kill metastases in mouse models and humans [22,38,39]. Thus, whether cell lines derive from primary versus metastatic tumors is unlikely to underlie strain-specific differences in cell killing. Third, differential susceptibility to reovirus-induced cell death could reflect properties of the cell type of origin. Large cell carcinomas are undifferentiated tumors that originate from lung transformed epithelial cells [40], whereas adenocarcinomas derive from epithelial glands or ducts [41]. Although we found differential susceptibility to rsT3D between large cell carcinoma and adenocarcinoma cell lines, a previous study found that different adenocarcinoma and large cell carcinoma cell lines were equivalently susceptible to killing by Reolysin [42]. However, the same study identified bronchiolar carcinoma and lung squamous cell carcinoma cell lines that were Reolysin resistant [42]. Variable resistance to reovirus is observed in other cancer cell lines, including head and neck squamous cell carcinoma cell lines and pancreatic cancer cell lines [43,44]. It is common for human cancers to vary in numerous properties, such as growth rate, drug sensitivity, and invasiveness [45]. The same type of tumor can differ greatly between patients, and individual tumors are comprised of strikingly heterogeneous cell populations [45]. Tumor heterogeneity is a primary source of resistance to conventional chemotherapeutics [45]. Presumably, the mixed cell populations that comprise tumors will contain at least some cells that are resistant to reovirus. In clinical trials, some tumors respond well to reovirus therapy, whereas others were refractory to treatment [23].
Our results indicate that cancer cell lines resistant to one strain of reovirus may be susceptible to a different strain. In addition, serial passage in cancer cells can be used to select reovirus variants with enhanced cancer cell killing capacity [24]. Thus, many opportunities exist to use reverse genetics to expand the oncolytic range or potency of reovirus.   The difference in cell killing between rsT1L and rsT3D in H661 and H1299 cells also could result from distinct cell death mechanisms induced by T1 and T3 reoviruses. In cultured cells and in vivo, T3 reoviruses induce apoptosis, whereas T1 reoviruses induce minimal apoptosis and kills cells by an undefined mechanism [46][47][48][49]. Reolysin induces apoptosis in many different primary cancer cells and cancer cell lines, tumor xenografts in mice, and patients [44,[50][51][52]. It is possible that H661 and H1299 cells have impaired apoptotic responses, which is a common feature of transformed cells. The inability to undergo apoptosis could prevent killing by rsT3D, but not rsT1L. However, we found that TNF-α/CHX induced apoptotic markers (caspase-3/7 activation ( Figure 6) and PARP cleavage (Figure 7)) in H661 and H1299 cells, indicating that apoptotic responses remain intact in both cell lines. Moreover, minimal caspase-3/7 activity ( Figure 6) and PARP cleavage (Figure 7) were observed following infection with either rsT1L or rsT3D in any of the cell lines tested. Furthermore, treatment with the broad-spectrum caspase inhibitor Z-VAD-FMK did not prevent killing by either reovirus strain in any of the cell lines tested (Figure 8). Taken together, these findings indicate that rsT1L and rsT3D can induce caspase-independent cell death in each NSCLC cell line used in this study. Caspase activity also is dispensable for killing of head and neck cancer cell lines [44]. In cancer patients, the mechanism of reovirus oncolysis is not well understood. In addition to apoptosis, reovirus also can induce autophagy and programmed necrosis [53,54]. It is possible that reovirus elicits multiple cell death pathways and that the cellular environment dictates the mechanism of cell killing.
We found that the L2, L3, and M1 genes associate with differences in killing between rsT1L and rsT3D in H661 and H1299 cells. The L2, L3, and M1 genes encode viral proteins λ2, λ1, and µ2, respectively [25]. Protein λ2 is an outer capsid component that forms pentamers at the 5-fold vertices of the reovirus virion [55,56]. A channel in the center of the λ2 pentamer serves as the insertion site for the σ1 protein [57]. Once σ1 dissociates during viral entry into target cells, the channel functions as the exit site for newly synthesized viral RNAs [58]. The λ2 protein also has guanylyltransferase [59] and methyltransferase [57] activity that is required for addition of 5 cap to nascent reovirus transcripts. The λ1 protein is an inner capsid constituent that is present in low amounts in the virion (~120 copies) [60]. Protein λ1 has RNA helicase activity that may function in unwinding of genomic dsRNAs to initiate transcription [61][62][63]. The λ1 protein also has RTPase activity [63,64], which in combination with its helicase activity suggests that λ1 contributes to capping of reovirus mRNAs. The µ2 protein also is a low abundance inner capsid component (~24 copies) [60]. Protein µ2 is hypothesized to function as a polymerase cofactor that enhances reovirus replication in a number of cell lines by enhancing RNA synthesis [57]. However, the mechanism by which µ2 promotes reovirus RNA production is undefined. The µ2 protein also is implicated in serotype-specific differences in type 1-interferon induction and sensitivity [65][66][67]. It is not known whether λ1, λ2, and µ2 function independently or in concert to enhance rsT1L-mediated killing of H661 and H1299 cells. Although rsT1L and rsT3D replication in H661 and H1299 cells did not differ ( Figure 5), replication of several of the single-gene reassortant viruses differed from the parental strains. We observed a general trend that insertion of T3D genes into a T1L genetic background decreased viral replication, whereas insertion of T1L genes into the T3D genetic background increased viral progeny production ( Figure 10). Although differences in viral replication do not underlie differential cell killing by the parental viruses, variation in cell killing by single gene reassortant viruses may simply reflect differences in viral replication.
It is important to note that although rsT3D and Reolysin originally derive from laboratory strain T3D, each has a unique passage history and genetic differences between the two strains that could impact oncolytic efficacy. It remains to be determined whether H661 and H1299 cells are more resistant to Reolysin than rsT1L. Next generation sequencing of Reolysin stocks revealed 32 nucleotide changes between Reolysin and T3D reference sequences in Genbank, 18 of which resulted in protein coding changes [68]. A subsequent comparison of the Reolysin sequence and sequences of plasmid-encoded rsT3D gene segments identified 20 amino acid differences between the viruses [69]. These findings indicate that Reolysin and rsT3D are not highly divergent. Construction of the Reolysin polymorphisms in the rsT3D background will be important for mechanistic studies of reovirus oncolysis and an important step in moving recombinant reovirus to the clinic.
In this study, we assessed the oncolytic potential of recombinant reoviruses generated by plasmid-based reverse genetics. To our knowledge, this is the first characterization of cancer cell killing by recombinant reoviruses generated using plasmid-based reverse genetics [20]. Our findings suggest that in certain types of tumors, T1 reovirus-based oncolytics may be more effective than T3-based vectors. Importantly, the cell killing profiles of recombinant reoviruses mirrors that of the native strains, suggesting that recombinant viruses will be as effective as traditional strains with respect to their oncolytic capacity. This work will inform future efforts to improve the efficacy and safety of reovirus oncolytics.

Cell Viability Assay
Monolayers of cells in 96-well plates (1 × 10 4 cells/well) were mock infected or adsorbed in triplicate with reovirus at room temperature (RT) for 1 h. The MOIs were calculated on PFU titers determined on L929 cells or FFU titers determined on H1299 or 1563 cells, as indicated. Monolayers were washed twice with phosphate-buffered saline (PBS) and incubated at 37 • C in completed media for various intervals. CellTiter-Glo (Promega, Madison, WI, USA) was used to assess cell viability [73][74][75]. Briefly, cells were cooled to RT, 100 µL of CellTiter-Glo was added to each well, and the plate was placed on a multi-purpose rotator for 2 min. Plates were incubated at RT for 10 min and luminescence was measured using a FluoStar Omega plate reader (BMG LabTech, Ortenberg, Germany). Where indicated, cells were treated with 25 µM Z-VAD-FMK (Sigma-Aldrich) or DMSO vehicle control for 1 h prior to infection. Following adsorption, the cells were overlaid with media containing DMSO or Z-VAD-FMK.

Virus Replication
Monolayers of cells in 24-well plates (1 × 10 5 cells/well) were adsorbed with reovirus at a range of multiplicities at RT for 1 h. The monolayers were washed twice with PBS and incubated in completed media at 37 • C. At the indicated time points, cells were frozen and thawed twice, and viral titers in lysates were determined using plaque assay on L929 cells [72]. Viral yields were calculated using the following formula: log 10yieldtx = log 10(PFU/mL)tx − log 10(PFU/mL)t0 (1) where tx is the time post-infection.

Active Caspase-3/7 Assay
Monolayers of cells (1 × 10 4 cells/well) in 96-well plates were mock infected or adsorbed in triplicate with reovirus at the indicated multiplicities at RT for 1 h. The monolayers were washed twice with PBS and incubated at 37 • C in completed media for various intervals. The cells were cooled to RT, 100 µL of Caspase-Glo 3/7 (Promega) was added to each well and the plates were incubated in the dark at RT for 1 h. Luminescence was measured using a FluoStar Omega plate reader (BMG LabTech).