Reovirus Variants with Mutations in Genome ... - Journal of Virology

Reovirus Variants with Mutations in Genome Segments S1 and L2 Exhibit Enhanced Virion Infectivity and Superior Oncolysis Maya Shmulevitz,a,c Shashi A. Gujar,a Dae-Gyun Ahn,a Adil Mohamed,c and Patrick W. K. Leea,b Department of Microbiology & Immunology,a and Department of Pathology,b Dalhousie University, Halifax, Nova Scotia, Canada, and Department of Medical Microbiology and Immunology, University of Alberta, Edmonton, Alberta, Canadac

Reovirus preferentially replicates in transformed cells and is being explored as a cancer therapy. Immunological and physical barriers to virotherapy inspired a quest for reovirus variants with enhanced oncolytic potency. Using a classical genetics approach, we isolated two reovirus variants (T3v1 and T3v2) with superior replication relative to wild-type reovirus serotype 3 Dearing (T3wt) on various human and mouse tumorigenic cell lines. Unique mutations in reovirus ␭2 vertex protein and ␴1 cell attachment protein were associated with the large plaque-forming phenotype of T3v1 and T3v2, respectively. Both T3v1 and T3v2 exhibited higher infectivity (i.e., a higher PFU-to-particle ratio) than T3wt. A detailed analysis of virus replication revealed that virus cell binding and uncoating were equivalent for variant and wild-type reoviruses. However, T3v1 and T3v2 were significantly more efficient than T3wt in initiating productive infection. Thus, when cells were infected with equivalent input virus particles, T3v1 and T3v2 produced significantly higher levels of early viral RNAs relative to T3wt. Subsequent steps of virus replication (viral RNA and protein synthesis, virus assembly, and cell death) were equivalent for all three viruses. In a syngeneic mouse model of melanoma, both T3v1 and T3v2 prolonged mouse survival compared to wild-type reovirus. Our studies reveal that oncolytic potency of reovirus can be improved through distinct mutations that increase the infectivity of reovirus particles.

H

uman reovirus is a nonpathogenic member of the family Reoviridae (35) with an inherent ability to target transformed cells (7, 10, 14, 42) and is presently being explored as a potential cancer virotherapy. An explosion of studies have demonstrated the potential of reovirus oncotherapy using cell culture systems, in vivo mouse models, ex vivo human tumor specimens, and volunteers (19). Phase III clinical trials are under way using Reolysin (respiratory enteric orphan virus, a reovirus formulation) in combination with paclitaxel and carboplatin in patients with head and neck cancers. Large strides have also been made toward understanding the molecular mechanisms of reovirus oncolysis. At least four steps of reovirus replication are enhanced by constitutive Ras signaling (25): (i) disassembly (uncoating) of incoming virions (1, 25), (ii) infectivity of progeny reovirus particles (25), (iii) release of progeny virions through enhanced reovirus-induced apoptosis (25, 38), and (iv) impaired expression of beta interferon (IFN-␤) (34, 37) (thereby promoting reovirus dissemination from cell to cell). The use of reovirus in vivo, however, faces several potential obstacles, not the least of which is the inevitable invocation of an immune response against the virus. While it is desirable to design strategies to specifically dampen these antivirus responses, a complementary approach would be the development of fast-acting reovirus variants capable of launching a preemptive strike against cancer cells before triggering a major antivirus response. Studies of reovirus oncolysis have almost exclusively used natural unmodified reovirus, which inherently possesses a preference for transformed cells. To our knowledge, only three modified reoviruses have been explored as potential alternatives to wild-type reovirus for optimized oncolysis. Kim et al. isolated an attenuated reovirus with a truncated cell attachment protein (␴1) that retains its oncolytic potential while reducing host toxicity (21). A second study, by van den Wollenberg et al., revealed that ␴1 can be genetically modified to target an artificial receptor (45). Furthermore, Rudd and Lemay isolated an interferon-hypersensitive reovirus

July 2012 Volume 86 Number 13

mutant that became increasingly dependent on Ras activation (34). These studies show that reovirus modifications can enhance the safety and specificity of oncolytic reovirus. In the present study, we sought to produce reovirus variants with increased potency (i.e., enhanced ability to replicate and spread in cancer cells) relative to the wild-type reovirus serotype 3 routinely used for oncolysis studies. We used a classical genetics approach to isolate and characterize two reovirus variants, called T3v1 and T3v2, with enhanced infection of already-susceptible cells. Although having mutations in different viral proteins, these variants share an advantage in virus replication that promotes infection of many cancer cell lines. Both reovirus variants have improved infectivity, exhibit enhanced activation of early transcription, and establish a productive infection more efficiently than wild-type reovirus. Additionally, both variants performed remarkably well in vivo, showing improved survival over wildtype reovirus in a syngeneic murine melanoma model in the absence of host toxicity. Altogether, our studies demonstrate the feasibility of generating reovirus variants with enhanced oncolytic potential without compromising specificity and safety. MATERIALS AND METHODS Cell lines and virus. NIH 3T3, PANC-1, H1299, and L929 cells were purchased from the American Type Culture Collection and maintained according to instructions. ID8 and B16 cells were maintained in RPMI and 10% fetal bovine serum. Reovirus serotypes and variants were grown

Received 4 February 2012 Accepted 13 April 2012 Published ahead of print 24 April 2012 Address correspondence to Maya Shmulevitz, [email protected], or Patrick Lee, [email protected]. Copyright © 2012, American Society for Microbiology. All Rights Reserved. doi:10.1128/JVI.00304-12

Journal of Virology

p. 7403–7413

jvi.asm.org

7403

Shmulevitz et al.

in L929 cells and purified by ultracentrifugation on cesium chloride gradients as previously described (27). Virus-infected cells were freezethawed and twice extracted with Vertrel XF (Dymar Chemicals) as previously described (27) and then layered onto 1.25- to 1.45-g/ml CsCl gradients. Virus was banded at 23,000 rpm for 5 h and dialyzed extensively against virus dilution buffer (150 mM NaCl, 15 mM MgCl2, 10 mM Tris, pH 7.4). Generation of reovirus variants and sequencing. L929 monolayers were coinfected with reovirus serotypes 1, 2, and 3 at a multiplicity of infection (MOI) of 3 each. At 80% cytolysis (determined visually), freezethawed lysates were amplified on fresh L929 monolayers at a 1/1,000 dilution. After four rounds of amplification, lysates were plaque titrated on large-scale L929 monolayers and plaques were visualized at day 6 with neutral red solution (Sigma). Plaques produced by coinfected lysates that were notably larger than T3wt plaques (by visual comparison) were plaque purified three times for clonal purity. All 10 genome segments of T3wt, T3v1, and T3v2 were reverse transcribed with Superscript (Invitrogen) and PCR amplified with high-fidelity Vent polymerase (New England BioLabs), using 3=-end and 5=-end primers specific to each segment. PCR products from two preparations were then sequenced in both directions (MCLAB Sequencing Services), using forward and reverse primers that spanned the entire sequence in 300- to 400-nucleotide intervals for each direction. Sequences were assembled using Vector NTI software (Invitrogen). Plaque size assay and reovirus titers. To compare plaque size, wildtype and variant reoviruses were plaque titrated on the indicated cell lines as previously described (37). Plaques were visualized by immunohistochemical staining as previously described (37) or with 1% (wt/vol) crystal violet solution, and plaque size was quantified with the ImageQuant TL (Amersham Biosciences) colony-counting feature. Reovirus titers from freeze-thawed infected cell lysates were determined by plaque titration on L929 cells and counted at 3 (T3v1 and T3v2) or 5 (T3wt) days postinfection to avoid bias of plaque size on titers. T3v1 and T3v2 plaques were counted 2 days prior to plaques produced by T3wt in order to ensure that virus titer measurements were not obscured by differences in plaque size. Radiolabeling, Western blot analysis, and dsRNA analysis. For radiolabeling experiments, [35S]methionine (PerkinElmer) was diluted in methionine-free minimal essential medium (MEM) at 100 ␮Ci/ml. To purify radiolabeled and unlabeled whole virions from infected cells, cell lysates were clarified by low-speed centrifugation and subjected to high-speed ultracentrifugation (100,000 ⫻ g) for 1 h over a 30% sucrose cushion (wt/vol). Cell lysates, CsCl-purified virions, or sucrose cushionpurified virions were subjected to SDS-PAGE and visualized by phosphorimaging, Imperial Coomassie dye staining (Pierce), or Western blot analysis with rabbit polyclonal anti-reovirus antibodies as previously described (37). To visualize double-stranded RNA (dsRNA), CsCl-purified virions were diluted 1:1 in 2⫻ protein sample buffer (Bio-Rad), heated at 65°C for 5 min, and subjected to SDS-PAGE at 6 mA (per gel) for 22 h prior to visualization with ethidium bromide staining. The results of enhanced chemiluminescence (ECL) fluorescence imaging and phosphorimaging on the Typhoon 9400 (Amersham) were quantified with ImageQuant TL software (Amersham). Coomassie blue and ethidium bromide staining was visualized on the Molecular Imager GelDoc XR⫹ system (Bio-Rad), and bands were quantified using the Quantity One 1-D Analysis Software (Bio-Rad). Flow cytometric analysis of productively infected cells and cell death. For flow cytometric quantification of infected cells, cells were trypsinized at 12 to 15 h postinfection, washed once with phosphatebuffered saline (PBS), and fixed with 4% fresh paraformaldehyde for 30 min on ice. Cells were then incubated in PBS containing 0.1% Triton X-100 and 3% bovine serum albumin (PBST/3% BSA) for 1 h. Cells were incubated with polyclonal anti-reovirus antibodies (1:5,000 in PBST/3% BSA) and secondary anti-rabbit Cy2- or Cy5-conjugated antibodies (at 1:500 in PBST/3% BSA) for 1 h each, with three PBS washes following each antibody incubation. For quantification of cell death, live cells were

7404

jvi.asm.org

stained with 7-aminoactinomycin D (7-AAD) and annexin V (BD Biosciences) exactly as per manufacturer’s instructions. Cell fluorescence was quantified with a FACScan flow cytometer (Becton Dickinson) and analyzed using WinDMI Version 2.8 (Scripps Research Institute). RNA extraction and qRT-PCR. For real-time quantitative reverse transcription-PCR (qRT-PCR), RNA was purified with TRIzol reagent (Invitrogen) followed by RNeasy mini columns (Qiagen) according to the manufacturer’s protocol. cDNA was synthesized using Moloney murine leukemia virus (MMLV) reverse transcriptase (Invitrogen) and random primers or sense-specific glyceraldehyde-3-phosphate dehydrogenase (GAPDH), reovirus S4, or M2 gene-specific primers. qPCR was performed using Go-Taq (Promega) according to the manufacturer’s instructions on the Stratagene MX3000P thermocycler and analyzed with MxPro (Stratagene). Standard curves were generated for each primer set as described previously (37). Statistical analysis was performed using Prism software (GraphPad). In vivo oncolysis experiments. A total of 20 C57B/L6 mice were injected subcutaneously with B16 melanoma cells (1 ⫻ 105 cells/mouse). When tumors were palpable (approximately 25 mm2), mice were segregated into four equivalent groups (based on tumor volume; five mice per group) and injected intratumorally at two-day intervals with high-purity T3wt, T3v1, or T3v2 at equivalent infectious doses (5 ⫻ 108 PFU/tumor) or with UV-inactivated T3wt. Tumor volumes were measured with a digital caliper and calculated as length (longest tumor dimension) ⫻ width (shortest tumor dimension). The animals were sacrificed when tumor dimensions reached 150 mm2. The statistical difference between survival curves was calculated with a log rank test using Prism software (Graphpad).

RESULTS

Reovirus variants T3v1 and T3v2 are more infectious than wildtype T3 on L929 cells. Three serotypes of human reovirus (T1L, serotype 1 Lang; T2J, serotype 2 Jones; and T3D, serotype 3 Dearing) exhibit distinct virulence and tropism in animal models and unique characteristics in cell culture, such as rate of inclusion body formation and apoptosis (26, 43, 44). Reovirus serotype 3 has superior cytolytic activity (44) and is used routinely for studies of reovirus oncolysis. Our goal was therefore to enhance the potency of the already-oncolytic T3D reovirus serotype (here referred to as T3wt). A classical genetics screen was employed to isolate reovirus variants that produced larger plaques than T3wt on highly permissive murine L929 cells. Plaque size was used as a convenient measure of virus fitness, reflecting the efficiency of both virus replication and dissemination. We hypothesized that T3wt potency could be improved by one of two strategies. First, mutations to T3wt proteins could enhance specific stages of reovirus replication and oncolysis. The mutation rate of porcine rotavirus, another member of the Reoviridae family, has been estimated at 5 ⫻ 10⫺5 mutations per nucleotide site (4). Assuming a similar mutation rate and the reovirus genomic size of 23.5 kb, natural mutations are expected to arise approximately once per replication cycle. Alternatively, reassortment between T3wt, T1L, and T2J genome segments could introduce new characteristics to T3wt that are beneficial for replication on cancer cells. Reovirus genomes are composed of 10 individual segments, permitting efficient gene reassortment between reovirus serotypes. To capture both possibilities (i.e., mutations and reassortants), new reovirus variants that outperform T3wt were selected from a lysate of L929 cells coinfected by all three reovirus serotypes. Since T3wt produces larger plaques than T1L and T2J (data not shown), we used T3wt as a means of comparison to isolate reovi-

Journal of Virology

Reovirus Variants with Enhanced Oncolysis

FIG 1 Reovirus variants T3v1 and T3v2 exhibit enhanced infection relative to wild-type reovirus. (A) Reovirus plaques produced by T3v1, T3v2, and wild-type parental serotype 3 Dearing (T3wt) on L929 mouse fibroblast cells from three independent experiments (exp. 1, 2, and 3) visualized by crystal violet staining (exp. 1 and 2) or immunohistochemistry with reovirus-specific antibodies (exp. 3). (Right) Plaque size (area represented in pixels) was quantified from results of four independent experiments using the ImageQuant colony-counting feature at 6 days postinfection (dpi). For each group, the 25 to 75% quartiles are indicated by the box, the median is represented by a horizontal line within the box, and minimum and maximum values are shown with short horizontal lines (n ⫽ 4; ***, P ⬍ 0.001 by the Student t test; 21 to 95 plaques per virus in each experiment). (B) One-step virus growth curve for T3wt, T3v1, and T3v2 on L929 cells infected at an MOI of 1 (mean ⫾ standard error of the mean [SEM]; n ⫽ 3). Cells were infected with 100 ␮l of virus, and total titers were collected from cells plus medium in a 1-ml mixture. T3v1 and T3v2 titers are significantly higher than the T3wt titer at 18 and 24 hpi (P ⬍ 0.001, Student t test). (C) Virus growth curves for L929 cells infected at an MOI of 0.01 (mean ⫾ SEM; n ⫽ 2). T3v1 and T3v2 titers are significantly higher than the T3wt titer at 24 to 120 hpi (Student t test). (D) Mean plaque sizes for T3v1, T3v2, and T3wt at 5, 6, and 7 dpi (pixel area, ⫾ SD, 19 to 44 plaques per virus).

rus variants with improved potency. Two reovirus variants (T3v1 and T3v2) were isolated on the basis of producing significantly larger plaques than T3wt on L929 cells (Fig. 1A). Reovirus plaques were routinely visualized by crystal violet staining, where zones of cell death reflect virus-induced lysis (Fig. 1A, experiments 1 and 2). Immunohistochemical staining of plaques with reovirus-specific antibodies indicated that larger plaques reflect an increased number of virus-infected cells (Fig. 1A, experiment 3). Digital images of these plaques were then obtained, and their sizes (in pixels) were compared. While the average plaque size for T3wt was 1,580 ⫾ 112 pixels, T3v1 and T3v2 produced plaques with 3,241 ⫾ 342 and 5,488 ⫾ 576 pixels, respectively (Fig. 1A, right panel). Furthermore, T3v2 plaques were larger than plaques produced by T3v1 (Fig. 1A, right panel), suggesting that T3v1 and T3v2 are likely unique variants. T3v1 and T3v2 plaques were also larger than plaques produced by reovirus type 3 Dearing stocks provided by two independent laboratories (data not shown), making it unlikely that our T3wt is a small-plaque variant of the wild-type virus. Increased plaque size can result from enhanced virus replication, cytolysis, or cell-to-cell virus spread. One-step growth curves have previously shown that a single round of reovirus replication in L929 cells is complete by approximately 18 to 24 h (17, 22). When the variants were infected at a multiplicity of infection (MOI) of 1, titers of both T3v1 and T3v2 were consistently higher


than that of T3wt by 9 h postinfection (hpi) and persisted over the 24-hour course of infection (Fig. 1B). T3v1 and T3v2 titers at 24 hpi were significantly higher than that of T3wt (4.8- ⫾ 1.7-fold and 6.3- ⫾ 1.8-fold higher for T3v1 and T3v2 relative to T3wt, respectively). When cells were infected at a low multiplicity of infection (MOI ⫽ 0.01) to permit multiple rounds of replication and dissemination (Fig. 1C), T3v1 and T3v2 titers increased relative to that of T3wt at each round of replication. T3v1 and T3v2 titers were 4.8- ⫾ 0.2-fold and 5.2- ⫾ 0.2-fold higher than that of T3wt at 24 hpi (P ⬍ 0.001 and P ⬍ 0.01, respectively), and 8.6- ⫾ 0.5-fold and 8.8- ⫾ 0.5-fold higher at 48 hpi (P ⬍ 0.05). The rapid saturation of infection by 48 hpi precluded analysis of subsequent rounds of infection. In a plaque assay where reovirus spread to neighboring cells can continue over many rounds of infection, T3v1 and T3v2 plaques became increasingly larger than T3wt plaques at 24-hour intervals (Fig. 1D). Together, these results suggest that both reovirus variants are superior to wild-type reovirus at each cycle of virus replication and spread. Mutations in ␭2 and ␴1 correlate with enhanced replication of T3v1 and T3v2, respectively. A genetic analysis was employed to associate specific gene modifications to the enhanced replication of T3v1 and T3v2. The double-stranded RNA genome of reovirus comprises four small (S), three medium (M), and three large (L) segments. Since T3v1 and T3v2 were isolated from a lysate of L929 cells coinfected by all three reovirus serotypes (T1L,

jvi.asm.org 7405

Shmulevitz et al.

FIG 2 Genetic characterization of reovirus variants T3v1 and T3v2. (A) The three large (L), three medium (M), and four small (S) dsRNA segments of T3wt, T3v1, T3v2, and reovirus serotypes 1 Lang (T1L) and 2 Jones (T2J) were resolved by SDS-PAGE and visualized by ethidium bromide staining. Extended electrophoresis was necessary to separate L and M genome segments (right panel) (B) Cell lysates produced by infection of L cells with T3wt alone (MOI ⫽ 20) or T3v1 alone (MOI ⫽ 10) or coinfection of T3wt and T3v1 (MOI ⫽ 20 and MOI ⫽ 10, respectively) were plaque titrated on fresh L cell monolayers. A box-and-whisker plot shows the sizes of plaques in relative pixel area, calculated using ImageQuant TL software. (C) Ten large and nine small plaques were isolated from T3wt ⫻ T3v1 coinfection lysates and subjected to sequencing at nucleotide mutation sites in L1, L2, and L3 segments. Mutant (C1216T, G3316A, or A425G in L1, L2, or L3, respectively) and wild-type (wt) sequences are indicated. (D) Location and nomenclature of reovirus structural proteins (43). Structural proteins that correspond to mutations in T3v1 or T3v2 are underlined, and corresponding genome segments are provided in brackets.

T2J, and T3D), we examined whether T3v1 or T3v2 represented reassortant viruses. The genome segments of each reovirus serotype produced distinct RNA profiles by SDS-PAGE (Fig. 2A). With the exception of S2 (not discernible between T3D and T1L)

7406

jvi.asm.org

and L1 segments (not discernible between T3D and T2J), the remaining genome segments in T3v1 and T3v2 were distinctly of reovirus serotype 3 origin. To further explore genetic changes in reovirus variants, the genomes of T3wt, T3v1, and T3v2 were fully sequenced. All 10 genome segments of T3wt, T3v1, and T3v2 were reverse transcribed, PCR amplified (with high-fidelity polymerase), and sequenced in both directions with a minimum coverage of three overlapping sequences at every nucleotide. Sequencing results confirmed that all 10 genome segments from T3v1 and T3v2 were of serotype 3 Dearing origin. Relative to T3wt, reovirus variant 1 had single nucleotide mutations in L1, L2, and L3 genome segments (Table 1, C1216T, G3316A, and A425G, respectively). Reovirus variant 2 only had one mutation (G65T) in the S1 genome segment. With the exception of these four point mutations, all remaining genome sequences of T3v1 and T3v2 were identical to that of T3wt. All four point mutations corresponded to missense amino acid substitutions (Table 1). Based on published sequences of natural reovirus serotype 3 isolates, these mutations in S1, L1, L2, and L3 reovirus segments are unique to T3v2 and T3v1. Since reovirus T3v1 was associated with single mutations in L1, L2, and L3, we next determined which L segment mutations are required for superior infection by T3v1. We generated reassortant viruses between T3wt and T3v1 by coinfection of L929 cells (Fig. 2B). Plaques produced by T3wt/T3v1 reassortants were a mixture of small and large sizes. A total of 10 large and 9 small plaques were isolated and examined specifically for mutations in L1, L2, and L3 by sequencing. The G3316A mutation in L2 was associated with all 10 large-plaque-forming reassortant viruses (Fig. 2C). Importantly, four of the large plaque-forming reassortants had a G3316A mutation in L2 alone, with wild-type sequences for L1 and L3. These results strongly suggested that the G3316A L2 mutation is essential and sufficient for enhanced replication relative to T3wt. Reassortant isolate 12 caused small plaques despite having both G3316A and A425G mutations in L2 and L3 segments. We hypothesized that the small-plaque phenotype of isolate 12 may be a consequence of “deleterious” mutations elsewhere in the genome. If this hypothesis is correct (and if the deleterious mutations are not also on the L2 segment), then coinfection by isolate 12 and T3wt should segregate the L2 and deleterious mutations and produce a proportion of large plaques. Indeed, reassortants produced by coinfection of isolate 12 and T3wt exhibited both large- and small-plaque phenotypes (data not shown). The A425G mutation in L3 was associated equally with small and large reassortant plaques and was insufficient to promote the large-plaque phenotype alone (Fig. 2C, isolates 11, 13, 16, and 17). TABLE 1 Mutations in T3v1 and T3v2 reovirus variants Mutation positiona Reovirus variant

Reovirus gene

Reovirus protein

Nucleotide

Amino acid

Accession no.

T3v2 T3v1

S1 L1 L2 L3

␴1 ␭3 ␭2 ␭1

G65 ¡T C1216 ¡T G3316 ¡A A425 ¡G

S18¡I P400¡S M1101¡I N138¡D

JQ599138 JQ599139 JQ599140 JQ599141

a Nucleotide mutations in S1, L1, L2, and L3 of T3v1 and T3v2 relative to T3wt and corresponding mutations in the deduced amino acid sequences of gene products ␴1, ␭3, ␭2, and ␭3, respectively.

Journal of Virology


The C1216T mutation in L1 associated with only 2 out of 10 largeplaque reassortant viruses but was not found in small-plaque reassortants. These studies suggest that L1 and L3 mutations are not essential for the T3v1 large-plaque phenotype, although we cannot exclude that these mutations affect (positively or negatively) infection by T3v1. The L2 gene, mutated in T3v1, encodes the reovirus ␭2 protein which forms the pentameric turrets at reovirion vertices (depicted in Fig. 2D), which anchor the cell attachment protein ␴1 in complete virions (9). Following cell entry, reovirus partially disassembles into transcriptionally active core particles devoid of ␴1 and outer capsid proteins ␮1C and ␴3 (18). The ␭2 turrets then function as gateways between the inside of reovirus cores and the cytoplasm of cells, permitting the entry of cellular metabolites (e.g., nucleotides, SAM, etc.) and the exit of newly transcribed viral RNAs (44). The ␭2 proteins also possess guanylyltransferase and methyltransferase activities (6, 23, 33) involved in 5= capping of positive-sense RNAs transcribed by the reovirus RNA-dependent RNA polymerase (L1-encoded ␭3 [41]). Reovirus variant 2 (T3v2) only had one missense amino acid substitution in the S1 segment, with all remaining sequences being identical to that of T3wt. The S1 gene mutation (G65T) therefore directly accounts for the increased plaque size of T3v2. The S1 genome segment encodes the well-characterized ␴1 cell attachment protein (depicted in Fig. 2D), which is shed from reovirions (i.e., disengaged from ␭2) following cell attachment and entry. ␴1 is also involved in reovirus-induced apoptosis (8, 44). The S1 segment also encodes the nonstructural protein ␴1 small (␴1s) (36) involved in reovirus-induced cell cycle arrest (32), apoptosis (16), and pathogenesis (5). The G65T mutation in T3v2, however, lies upstream of the coding sequence for ␴1s and could affect ␴1s only by indirect mechanisms (e.g., altered ␴1s expression levels). T3v1 and T3v2 have increased particle infectivity relative to T3wt. In order to determine the impact of S1 and L2 mutations on reovirus infection, we undertook a more extensive analysis of T3v1, T3v2, and T3wt reovirus particles. To standardize the relative quantity of reovirus particles between T3v1, T3v2, and T3wt, 2-fold serial dilutions of cesium chloride gradient-purified reovirus particles were separated by SDS-PAGE, and dominant reovirus protein bands (␭1 and 2, ␮1, ␴2 and ␴3) were quantified by densitometry (Fig. 3A). Purified T3v1, T3v2, and T3wt preparations were also subjected to plaque titration. Intriguingly, titers of T3v1 and T3v2 were significantly higher than that of T3wt for equivalent reovirus proteins (Fig. 3A). Quantification of titers versus proteins from three purified virion preparations demonstrated that T3v1 and T3v2 reovirus variants have 3.9 ⫾ 0.5-fold and 3.1 ⫾ 0.6-fold higher infectivity-per-particle ratios, respectively, relative to T3wt (Fig. 3C) (mean ⫾ standard deviation [SD]; P ⫽ 0.0002 and P ⫽ 0.0012 for T3v1 and T3v2 relative to 1, one-sample t test). Both reovirus variants also exhibited increased titers relative to T3wt for equivalent virus particles when titration was performed on the human non-small cell lung carcinoma cell line NCI-H1299. It is important to note that for reasons poorly understood, a majority of wild-type reovirus virus particles (⬎99%) are noninfectious (13). The increased PFU-to-particle ratios for T3v1 and T3v2 may therefore represent an improved capacity of reovirions to undergo productive replication. Genomic dsRNAs were also quantified from serial dilutions of purified T3wt, T3v1, and T3v2 virions to determine if infectivity differences are a consequence of variation in genomic RNA con-


FIG 3 T3v1 and T3v2 virus particles have enhanced infectivity per particle relative to T3wt. Serial 2-fold dilutions of cesium chloride gradient-purified T3v1, T3v2, and T3wt were separated by SDS-10% polyacrylamide gel electrophoresis and stained with Imperial Coomassie dye (A) or ethidium bromide (B). Dominant bands were quantified using Quantity One 1-D Analysis Software (Bio-Rad). Standard curves were generated based on T3wt values for each reovirus protein (A) or double-stranded RNA genome segment (B) (except for M1 and M2, which were quantified together), and the relative quantity of corresponding proteins and dsRNAs from T3v1 and T3v2 were deduced from polynomial trendline equations. (A) Relative protein values represent averages for all four proteins. PFU per lane were calculated from titers obtained for corresponding virus preparation. The relative PFU-to-particle ratio was calculated by dividing relative PFU by relative protein values. (B) Relative RNA values represent averages for all dsRNA segments. The relative RNA-to-protein ratio was calculated as the ratio of relative RNA to relative protein obtained in the experiment shown in panel A. (C) The relative PFU-to-particle ratio was based on the values obtained in the experiment shown in panel A, and titers were obtained on L929 cells (mean ⫾ SD; n ⫽ 5) or NCI-H1299 cells (mean ⫾ SD; n ⫽ 2).

tent. Reovirus dsRNA quantities were proportional to reovirus protein amounts for T3wt, T3v1, and T3v2 (Fig. 3B). Electrophoretic analysis therefore suggested that a greater proportion of T3v1 and T3v2 reovirus particles are infectious despite having equivalent dsRNA genomic content. T3v1 and T3v2 have enhanced early onset of infection relative to T3wt. A thorough analysis of hallmark steps of reovirus replication was undertaken to explore the basis for increased in-

jvi.asm.org 7407

Shmulevitz et al.

fectivity of T3v1 and T3v2 virus particles. The extent of reovirus binding was evaluated by exposing cells to equivalent particle doses of T3wt, T3v1, and T3v2 for 1 h at 4°C , followed by quantification of cell-associated reovirus ␮1 proteins by Western blot analysis. T3wt, T3v1, and T3v2 showed similar associations with L cells, suggesting that binding is equivalent for all three viruses (Fig. 4A). A 4-fold reduction in particle dose resulted in 4-fold-lower cell-associated virus, demonstrating that cell receptors were not saturated in our experiments. After receptor binding and endocytosis, reovirus particles partially disassemble/uncoat into intermediate subvirion particles (ISVPs). ISVPs are formed through the degradation of outer-coat ␴3 protein and cleavage of ␮1c to ␦ (30, 31), mediated by lysosomal cathepsins proteases (2, 11). To compare the rates of endocytosis and reovirus uncoating, we monitored the hallmark cleavage of ␮1c to ␦. Cells were exposed to equivalent particle doses of T3wt, T3v1, and T3v2 for 1 h at 4°C to promote virus binding, followed by incubation at 37°C for 0 to 4 h to monitor the entry and uncoating process. The rate of ␮1c proteolysis to ␦ was similar for wild-type and variant reoviruses, suggesting that ISVP production is equally efficient (Fig. 4B). Reovirus ISVPs must further disassemble into transcriptionally active core particles. Production of reovirus cores is associated with release of ␴1 from vertices (18, 40), significant conformational changes to ␭2 turret proteins (9, 28, 33), and activation of reovirus transcription machinery (polymerase, ␭3, and capping enzymes ␭2 [6], ␮2 [20], and ␭1 [3]). Quantitative reverse transcription-PCR was undertaken to monitor the onset of virus transcription (Fig. 4C). Similar levels of T3v1, T3v2, and T3wt reovirus RNAs were associated with cells at 1 and 3 h postinfection (hpi), reflecting equivalent input levels of reovirus genomic RNAs. However, significantly increased levels of reovirus transcripts were associated with cells infected with T3v1 and T3v2 relative to T3wt by 6 hpi. These studies suggest that T3v1 and T3v2 both have an advantage over T3wt early during infection, likely at or preceding the onset of viral transcription. Differences in the levels of reovirus transcripts following exposure to equivalent viral particle inputs of T3v1, T3v2, or T3wt and comparable cell binding and uncoating efficiencies could reflect two unique scenarios. The first scenario is that T3v1 and T3v2 manifest a more rapid or efficient onset of viral replication, resulting in increased levels of viral RNA transcripts at a given time point. In this scenario, an equivalent number of cells would be productively infected by any of the three viruses, but those infected by T3v1 or T3v2, compared to those infected by T3wt, would produce more viral transcripts, proteins, and progeny per infected cell. The alternative scenario is that a greater proportion of T3v1 and T3v2 virus particles can initiate a productive infection. In this scenario, more cells should become productively infected by T3v1 and T3v2 relative to T3wt but the viral RNA and protein levels per infected cells would be similar for all three viruses. To distinguish between these two possibilities, cells exposed to 2-fold serial dilutions of T3v1, T3v2, and T3wt were stained with polyclonal anti-reovirus antibodies and reovirus protein expression was measured by flow cytometry. Despite equivalent binding of all three viruses (Fig. 4A, B, and C), a greater number of cells stained positive for reovirus protein expression following infection by reovirus variants relative to T3wt (Fig. 4D). The

7408

jvi.asm.org

FIG 4 Enhanced onset of productive infection is common to both T3v1 and T3v2. (A) L929 cells were exposed to equal particle doses of T3wt, T3v1, and T3v2. Cell-associated reovirus ␮1c protein was detected after 1 h of incubation at 4°C, extensive washing, and Western blot analysis. Densitometric analysis of ␮1c quantities relative to ␤-actin is provided below each lane. To ensure quantitative results, cells were exposed to 4-fold dilutions of reovirus particles. (B) Cleavage of reovirus ␮1c to ␦, a hallmark of reovirus disassembly, was monitored for T3wt, T3v1, and T3v2 by Western blot analysis following exposure to virus for 1 h at 4°C and then at 37°C for the indicated hours postinfection (h.p.i.). Ratios of ␮1c to ␦ provided below corresponding lanes were calculated by densitometry. (C) Relative levels of reovirus S4 RNAs following infection with equivalent particle doses at indicated hours postinfection (n ⫽ 3; mean ⫾ standard error of the mean [SEM]; *, P ⬍ 0.05; **, P ⬍ 0.01; ***, P ⬍ 0.001 by the Student t test). (D) Cells exposed to serial 4-fold dilutions of purified reovirus were incubated for 12 h prior to staining with anti-reovirus serum and flow cytometric analysis.

mean fluorescence intensity of cells positively stained with reovirus-specific antibodies was similar for all three viruses, suggesting that reovirus protein expression per infected cell is likely equivalent between reovirus strains (as is explored fur-

Journal of Virology


ther below). Together, these studies suggest that T3v1 and T3v2 virus particles have an increased capacity to establish a productive infection. Enhanced potency of T3v1 and T3v2 is independent of subsequent stages of reovirus replication. To determine if T3v1 and T3v2 had additional advantages in replication over T3wt at stages subsequent to the onset of productive infection, L929 cells were infected with equivalent PFU (infectious doses) of wild-type and variant reovirus. Using equivalent infectious doses of T3v1, T3v2, and T3wt resulted in equivalent numbers of reovirus protein-expressing (i.e., productively infected) cells (Fig. 5A). A closer examination was then performed to determine if specific steps of replication were more efficient for reovirus variants within productively infected cells. When differences in particle infectivity were neutralized by using equivalent infectious doses, reovirus RNA levels (measured by qRT-PCR) were found to be equivalent for all three viruses at early (6 to 9 hpi) and late (12 to 24 hpi) stages of infection (Fig. 5B). Virus protein accumulation (by Western blotting [Fig. 5C]) and de novo synthesis (by [35S]methionine labeling [Fig. 5D]) were also equivalent. These studies suggest that while T3v1 and T3v2 have a higher proportion of infectious particles relative to T3wt (Fig. 4), the rates of virus gene transcription and protein expression per infectious particle are equivalent for all three viruses. We next asked whether progeny virus assembly was more efficient for reovirus variants. Infected cells were pulse-labeled with [35S]methionine and chased for 1 h to permit assembly. Virions were then separated from detergent lysates by high-speed ultracentrifugation through a sucrose cushion to ensure separation between pellet (virus) and free proteins. Serial dilution of resuspended pellets showed equivalent quantities of newly produced virions from cells infected with T3wt, T3v1, and T3v2 (Fig. 5E). The quantity of outer-capsid ␴3 protein in pellets reflects the extent of complete virion assembly. Furthermore, quantities of negative-sense reovirus RNAs were equivalent for all three viruses, suggesting similar dsRNA synthesis within newly assembled progeny core particles (Fig. 5F). Importantly (and as expected), despite having similar particle numbers according to Western blot analysis (Fig. 5G), reovirus particles purified from T3v1 and T3v2-infected cells had higher titers than those purified from T3wt-infected cells. The size of virus plaques reflects the efficiency of virus replication but also the capacity of progeny viruses to spread from cell to cell. Therefore, in addition to evaluating the steps in a single round of infection, we also assessed whether T3v1 and T3v2 had an advantage in dissemination. The efficiency of reovirus dissemination partially depends on the extent of virus release that follows lysis of infected cells. To compare the extent of cytolysis induced by wildtype and variant reovirus, infected cells were stained with annexin V and 7-aminoactinomycin D to monitor apoptosis and cell death, respectively. Flow cytometric analysis indicated that cells infected by T3v1, T3v2, and T3wt underwent equivalent cytolysis when standardized for equal numbers of infected cells (Fig. 5H). Altogether, our in-depth analysis of T2v1 and T3v2 replication suggests that mutations in ␴1 and lambda proteins support enhanced virus replication primarily by increasing the proportion of progeny virus particles that are infectious. T3v1 and T3v2 are more infectious than T3wt on permissive cancer cells. Next, we wanted to establish whether enhanced in-


fectivity of T3v1 and T3v2 over T3wt was maintained on various well-established murine and human cancer cell lines. Plaque size was used as an effective measure of reovirus replication over several rounds of infection. T3v1 and T3v2 caused visibly larger plaques than T3wt on the highly permissive human non-small cell lung carcinoma cell line NCI-H1299 (Fig. 6). T3v1 and T2v2 also produced larger plaques on several human and mouse cell lines with moderate susceptibility to reovirus, including HCT116 human colorectal carcinoma cells, PANC-1 pancreatic ductal carcinoma cells, and ID8 murine ovarian cancer cells. In contrast, replication of both wild-type and reovirus variants was restricted on nontransformed NIH 3T3 mouse fibroblast cells (Fig. 6). These experiments suggest that while T3 variants exhibit superior potency of infection on already-permissive cells, the host specificity of T3 variants resembles that of T3wt. T3v1 and T3v2 show superior oncolytic activities in vivo compared to T3wt. T3v1 and T3v2 showed a clear advantage over wild-type reovirus for replication in cancer cells in vitro. To assess these viruses in vivo, we made use of the B16 melanoma syngeneic immunocompetent mouse model. This model was chosen because our previous studies demonstrated that wild-type reovirus could drastically reduce (but not eliminate) the tumor arising from B16 melanoma cells (12). Furthermore, we reasoned that differences in oncolytic virus potency would become more evident in an immunocompetent model, where antiviral immunity constrains reovirus replication and dissemination (15, 39). We first determined whether T3v1 and T3v2 exhibited enhanced infectivity and oncolysis of B16 melanoma cells in vitro. As seen for other permissive cancer cell lines (Fig. 6), T3v1 and T3v2 produced higher titers than T3wt in B16 cells (Fig. 7A) (titers at 48 hpi for T3wt, T3v1, and T3v2 were 1.25 ⫾ 0.12 ⫻ 108, 4.90 ⫾ 0.38 ⫻ 108, and 7.70 ⫾ 0.53 ⫻ 108, respectively). Plaques produced by T3v1 and T3v2 were also visibly larger than T3wt plaques on B16 cell monolayers (Fig. 7B). C57B/L 6 mice were injected subcutaneously with B16 cells. When tumor volumes reached 25 mm2, animals were treated intratumorally with three injections of T3wt, T3v1, or T3v2 at equivalent infectious doses (5 ⫻ 108 PFU/ tumor) or with UV-inactivated T3wt (Fig. 7C). Equivalent infectious doses were used to provide all three viruses an equal opportunity to establish an infection initially. All three reovirus preparations slowed tumor growth and prolonged survival relative to UV-inactivated T3wt-treated mice (Fig. 7D) (P ⫽ 0.008, 0.002, and 0.002, respectively, for T3wt, T3v1, and T3v2). Importantly, T3v2 and especially T3v1 showed remarkable improvement over T3wt in promoting animal survival (P ⫽ 0.008 for T3v1; P ⫽ 0.04 for T3v2). The median survival times of tumorbearing animals treated with UV-inactivated T3wt, T3wt, T3v1, and T3v2 were 22, 30, 41, and 33 days, respectively. Importantly, mice injected with T3wt or reovirus variants appeared healthy and showed no visible signs of distress or host toxicity. DISCUSSION

Reovirus is currently in phase III clinical trials as a potential cancer therapeutic. Our study reveals that wild-type reovirus serotype 3, presently considered the best candidate for reovirus oncotherapy, can achieve higher oncolytic potency. For reasons not well understood, a majority of reovirus particles (⬎99%) are noninfectious. The generation of reovirus variants with a higher infectious/noninfectious particle ratio would therefore be highly desirable, as they should theoretically be more efficient in target cell killing. Used as oncolytic agents in

jvi.asm.org 7409

Shmulevitz et al.

FIG 5 Enhanced potency of T2v1 and T3v2 is independent of subsequent stages of reovirus replication. L cells were infected with T3wt, T3v1, and T3v1 at equivalent infectious doses (MOI ⫽ 1). (A) Percent of cells stained positive with anti-reovirus antibodies by flow cytometry at 12 hpi. (B) Levels of reovirus M1 RNA by qRT-PCR (positive and negative sense; n ⫽ 3; mean ⫾ standard error of the mean [SEM]; no significant differences between viruses at a common time point by the Student t test). (C) Total levels of ␤-actin (loading control) and reovirus ␮1 and ␴3 proteins by Western blot analysis at indicated hours postinfection (hpi). (D) Infected cells labeled at 12 hpi with [35S]methionine for 30 min followed by immunoprecipitation with anti-reovirus antibodies monitor de novo synthesis of viral proteins. (E) [35S]Methionine-labeled reovirus particles purified by high-speed ultracentrifugation. Two-fold dilutions of resuspended virus pellets were visualized by SDS-PAGE and autoradiography. (F) Ratios of negative-sense (reflecting dsRNA) to positive-sense (reflecting mRNA) reovirus M1 RNAs by qRT-PCR (n ⫽ 2; mean ⫾ SEM; no significant differences between viruses by the Student t test). (G) Representative experiment shows 2-fold dilutions of sucrose cushion-purified reovirus from T3wt, T3v1, and T3v2-infected cells at 12 hpi, analyzed by Western blot analysis using anti-reovirus antiserum. Densitometric analysis shows relative levels of ␮1c and ␴3 proteins. Band intensities at higher concentrations were nonlinear and therefore are not presented. Titers (PFU/ml) are provided for each undiluted preparation. (H) Cell death at 24 hpi by flow cytometric analysis of live cells stained with 7-AAD (y axis) and annexin V (x axis). The percentages of cell death are shown in the top and bottom right-hand quadrants as fractions of infected cells for three independent experiments (mean ⫾ SEM; no significant differences by the Student t test).

vivo, these variants should cause a more rapid reduction of tumor burden prior to host immune virus clearance. Our analysis of two new reovirus variants, T3v1 and T3v2, demonstrates that a greater proportion of infectious reovirus par-

7410

jvi.asm.org

ticles (relative to T3wt) can be achieved. On average, T3v1 and T3v2 had 3.9 ⫾ 0.5-fold and 3.1 ⫾ 0.6-fold higher infectivity (PFU) per particle. The increased particle infectivity of T3v1 and T3v2 progeny would result in higher reovirus titers per infected

Journal of Virology


FIG 6 Reovirus variants exhibit superior replication on cancer cell lines. T3wt, T3v1, and T3v2 plaques on human (NCI-H1299 lung, PANC-1 pancreatic, HCT-116 colorectal) and mouse (ID8 ovarian) cancer cell lines and on nontransformed NIH-3T3 cells, visualized by reovirus-specific immunohistochemical staining at 6 (6d) or 8 (8d) days postinfection. Note that representative photographs were chosen to permit visualization of plaque size and do not reflect relative titers. Graphical representation of plaques size (area in pixels) on respective cell lines as quantified using the ImageQuant colony-counting feature (mean ⫾ standard error of the mean [SEM]; minimum 21 plaques).

cell at every round of infection. The exponential rise in reovirus infection at every replication cycle would thereby produce large infection differences over time. The superior potency of the two reovirus variants was indeed evident both in cell culture systems (producing larger plaques and higher titers in permissive cells) and in an in vivo immunocompetent murine melanoma model (promoting prolonged survival). It is intriguing that while T3v2 produced larger plaques than T3v1 in vitro, T3v1 was more potent than T3v2 in the in vivo B16 melanoma model. It is conceivable that additional parameters such as pH, extracellular proteases, or extent of interferon signal-

ing exist between our in vitro and in vivo systems that impact the stability and/or infectivity of T3v1 and T3v2 reovirus particles. It would be interesting to compare the oncolytic efficiency of T3v1 versus that of T3v2 in other animal cancer models to determine the relative value of T3v1 versus T3v2 in cancer therapy. Reovirus variants not only exhibited enhanced potency but also retained their specificity for cancer cells in vitro and showed no host toxicity in vivo. These variants showed superior replication on highly susceptible (L929 and NCI-H1299) and moderately susceptible (HCT116, PANC-1, and ID8) cancer cell lines. In contrast, nontransformed cells (NIH-3T3) remained resistant to all

FIG 7 Oncolytic potency and safety of reovirus variants in vivo. (A) Titers of virus inoculum and progeny produced in B16 mouse melanoma cells in culture at 48 hpi (n ⫽ 2 in quadruplicate; mean ⫾ SEM; ***, P ⬍ 0.001 relative to T3wt, Student t test). Differences between inocula were not significant (Student t test). (B) T3wt, T3v1, and T3v2 plaques on B16 melanoma cells visible by immunohistochemical staining with anti-reovirus antibodies. Wild-type C57BL/6 mice were subcutaneously (s.c.) grafted with native B16 cells, injected with UV-T3wt, T3wt, T3v1, or T3v2 intratumorally (i.t.) at the indicated time points (C), and monitored for tumor growth and survival (n ⫽ 5 mice per group) (D).


jvi.asm.org 7411

Shmulevitz et al.

three viruses. We anticipate that previously published mechanisms that suppress reovirus replication in nontransformed cells (1, 25, 34, 37, 38) also restrict replication of T3v1 and T3v2. Our studies also revealed the molecular basis for enhanced oncolytic potency of T3v1 and T3v2 over T3wt. Genetic analysis combined with a detailed examination of reovirus replication steps demonstrated that both T3v1 and T3v2 achieved a strategy to enhance an early step(s) of virus infection, leading to enhanced levels of early virus transcription. Our data suggest that the two variant viruses are more efficient than the wild-type virus in the generation of “transcription-competent” subviral particles that are nonetheless indistinguishable in all subsequent steps of viral replication. Thus, on a per-particle basis, T3v1 and T3v2 are more infectious than T3wt, whereas on a per-infectious-unit basis, all three viruses behave similarly. It is remarkable that enhancement of this early step of reovirus infection is achieved by independent mutations in viral ␭2 and ␴1 proteins (for T3v1 and T3v2, respectively), both of which are located at the 12 vertices of the viral icosahedron. The N-terminal segment of ␴1, where the T3v2 mutation resides, is responsible for anchoring the protein to the inside wall of the ␭2 turret (24, 29). The mutation in T3v1 lies within the C-terminal immunoglobulin-like “flaps” on the outer surface of reovirions. These ␭2 flaps are hypothesized to anchor ␴1 (9) and undergo substantial rearrangement from closed to open conformations during the transition from ISVPs to transcriptionally active cores (9, 28, 33). Accordingly, it is conceivable that specific mutations in either protein could facilitate ␴1 disengagement or promote capsid structural changes necessary for generation of transcriptionally active cores during infection. Furthermore, it is possible that higher particle infectivity results in additional effects such as lower induction of antiviral signaling, which would further promote virus cell-to-cell spread, plaque size, and in vivo tumor clearance. Studies are under way to understand the precise effects of mutations in ␭2 and ␴1 proteins on early steps of reovirus replication and on cell-to-cell spread. The present study demonstrates the feasibility of generating oncolytic reovirus with enhanced potency by classical genetic means. It is likely that additional reovirus variants harboring mutations in other viral genes will be identified in the future. Genetic crossing between such variants could conceivably lead to the generation of “supervariants” with superior oncolytic properties. While the capability of these viruses to launch a preemptive strike against cancer is a major desirable outcome, it is imperative that they remain relatively harmless to normal cells and the host. Based on the observations made in the present study, such a goal should be deemed achievable. ACKNOWLEDGMENTS We thank the outstanding personnel at the FACS, imaging, and animal care facilities at Dalhousie University. This work was supported by an operating grant from the Canadian Institutes for Health Research (CIHR) and the Canadian Cancer Society Research Institute through the Terry Fox Foundation (CCSRI and TFRI) to P.W.K.L., Cancer Research Training Program postdoctoral fellowships, through the Beatrice Hunter Cancer Research Institute (to M.S., S.A.G., and D.-G.A.), CIHR postdoctoral fellowships (to M.S. and S.A.G.), and the National Cancer Institute of Canada (NCIC) postdoctoral fellowship (M.S.).

REFERENCES 1. Alain T, et al. 2007. Proteolytic disassembly is a critical determinant for reovirus oncolysis. Mol. Ther. 15:1512–1521.

7412

jvi.asm.org

2. Baer GS, Ebert DH, Chung CJ, Erickson AH, Dermody TS. 1999. Mutant cells selected during persistent reovirus infection do not express mature cathepsin L and do not support reovirus disassembly. J. Virol. 73:9532–9543. 3. Bisaillon M, Lemay G. 1997. Characterization of the reovirus lambda1 protein RNA 5=-triphosphatase activity. J. Biol. Chem. 272:29954 –29957. 4. Blackhall J, Fuentes A, Magnusson G. 1996. Genetic stability of a porcine rotavirus RNA segment during repeated plaque isolation. Virology 225: 181–190. 5. Boehme KW, Frierson JM, Konopka JL, Kobayashi T, Dermody TS. 2011. The reovirus ␴1s protein is a determinant of hematogenous but not neural virus dissemination in mice. J. Virol. 85:11781–11790. 6. Cleveland DR, Zarbl H, Millward S. 1986. Reovirus guanylyltransferase is L2 gene product lambda 2. J. Virol. 60:307–311. 7. Coffey MC, Strong JE, Forsyth PA, Lee PW. 1998. Reovirus therapy of tumors with activated Ras pathway. Science 282:1332–1334. 8. Connolly JL, Barton ES, Dermody TS. 2001. Reovirus binding to cell surface sialic acid potentiates virus-induced apoptosis. J. Virol. 75:4029 – 4039. 9. Dryden KA, et al. 1993. Early steps in reovirus infection are associated with dramatic changes in supramolecular structure and protein conformation: analysis of virions and subviral particles by cryoelectron microscopy and image reconstruction. J. Cell Biol. 122:1023–1041. 10. Duncan MR, Stanish SM, Cox DC. 1978. Differential sensitivity of normal and transformed human cells to reovirus infection. J. Virol. 28:444 – 449. 11. Ebert DH, Deussing J, Peters C, Dermody TS. 2002. Cathepsin L and cathepsin B mediate reovirus disassembly in murine fibroblast cells. J. Biol. Chem. 277:24609 –24617. 12. Gujar SA, Marcato P, Pan D, Lee PW. 2010. Reovirus virotherapy overrides tumor antigen presentation evasion and promotes protective antitumor immunity. Mol. Cancer Ther. 9:2924 –2933. 13. Hand R, Tamm I. 1973. Reovirus: effect of noninfective viral components on cellular deoxyribonucleic acid synthesis. J. Virol. 11:223–231. 14. Hashiro G, Loh PC, Yau JT. 1977. The preferential cytotoxicity of reovirus for certain transformed cell lines. Arch. Virol. 54:307–315. 15. Hirasawa K, et al. 2003. Systemic reovirus therapy of metastatic cancer in immune-competent mice. Cancer Res. 63:348 –353. 16. Hoyt CC, et al. 2005. Nonstructural protein sigma1s is a determinant of reovirus virulence and influences the kinetics and severity of apoptosis induction in the heart and central nervous system. J. Virol. 79:2743–2753. 17. Jane-Valbuena J, et al. 1999. Reovirus virion-like particles obtained by recoating infectious subvirion particles with baculovirus-expressed sigma3 protein: an approach for analyzing sigma3 functions during virus entry. J. Virol. 73:2963–2973. 18. Joklik WK. 1972. Studies on the effect of chymotrypsin on reovirions. Virology 49:700 –715. 19. Kelly K, et al. 2009. Reovirus-based therapy for cancer. Expert. Opin. Biol. Ther. 9:817– 830. 20. Kim J, Parker JS, Murray KE, Nibert ML. 2004. Nucleoside and RNA triphosphatase activities of orthoreovirus transcriptase cofactor mu2. J. Biol. Chem. 279:4394 – 4403. 21. Kim M, et al. 2011. Attenuated reovirus displays oncolysis with reduced host toxicity. Br. J. Cancer 104:290 –299. 22. Loh PC, Oie HK, Camyre KP. 1971. Role of lysine in the replication of reovirus. II. Characterization of empties and effect on macromolecular synthesis. Arch. Gesamte Virusforsch. 35:114 –125. 23. Luongo CL, Reinisch KM, Harrison SC, Nibert ML. 2000. Identification of the guanylyltransferase region and active site in reovirus mRNA capping protein lambda2. J. Biol. Chem. 275:2804 –2810. 24. Mah DC, Leone G, Jankowski JM, Lee PW. 1990. The N-terminal quarter of reovirus cell attachment protein sigma 1 possesses intrinsic virion-anchoring function. Virology 179:95–103. 25. Marcato P, Shmulevitz M, Pan D, Stoltz D, Lee PW. 2007. Ras transformation mediates reovirus oncolysis by enhancing virus uncoating, particle infectivity, and apoptosis-dependent release. Mol. Ther. 15:1522–1530. 26. Mbisa JL, Becker MM, Zou S, Dermody TS, Brown EG. 2000. Reovirus mu2 protein determines strain-specific differences in the rate of viral inclusion formation in L929 cells. Virology 272:16 –26. 27. Mendez II, Hermann LL, Hazelton PR, Coombs KM. 2000. A comparative analysis of freon substitutes in the purification of reovirus and calicivirus. J. Virol. Methods 90:59 – 67. 28. Mendez II, Weiner SG, She YM, Yeager M, Coombs KM. 2008. Con-

Journal of Virology


29. 30.

31. 32. 33. 34. 35. 36. 37.

formational changes accompany activation of reovirus RNA-dependent RNA transcription. J. Struct. Biol. 162:277–289. Nibert ML, Dermody TS, Fields BN. 1990. Structure of the reovirus cell-attachment protein: a model for the domain organization of sigma 1. J. Virol. 64:2976 –2989. Nibert ML, Odegard AL, Agosto MA, Chandran K, Schiff LA. 2005. Putative autocleavage of reovirus mu1 protein in concert with outercapsid disassembly and activation for membrane permeabilization. J. Mol. Biol. 345:461– 474. Odegard AL, et al. 2004. Putative autocleavage of outer capsid protein micro1, allowing release of myristoylated peptide micro1N during particle uncoating, is critical for cell entry by reovirus. J. Virol. 78:8732– 8745. Poggioli GJ, Keefer C, Connolly JL, Dermody TS, Tyler KL. 2000. Reovirus-induced G2/M cell cycle arrest requires ␴1s and occurs in the absence of apoptosis. J. Virol. 74:9562–9570. Reinisch KM, Nibert ML, Harrison SC. 2000. Structure of the reovirus core at 3.6 Å resolution. Nature 404:960 –967. Rudd P, Lemay G. 2005. Correlation between interferon sensitivity of reovirus isolates and ability to discriminate between normal and Rastransformed cells. J. Gen. Virol. 86:1489 –1497. Sabin AB. 1959. Reoviruses. A new group of respiratory and enteric viruses formerly classified as ECHO type 10 is described. Science 130:1387– 1389. Sarkar G, et al. 1985. Identification of a new polypeptide coded by reovirus gene S1. J. Virol. 54:720 –725. Shmulevitz M, Pan LZ, Garant K, Pan D, Lee PW. 2010. Oncogenic Ras


38.

39. 40.

41. 42.

43.

44.

45.

promotes reovirus spread by suppressing IFN-beta production through negative regulation of RIG-I signaling. Cancer Res. 70:4912– 4921. Smakman N, van den Wollenberg DJ, Borel RI, Hoeben RC, Kranenburg O. 2005. Sensitization to apoptosis underlies KrasD12-dependent oncolysis of murine C26 colorectal carcinoma cells by reovirus T3D. J. Virol. 79:14981–14985. Smakman N, et al. 2006. Immunosuppression promotes reovirus therapy of colorectal liver metastases. Cancer Gene Ther. 13:815– 818. Smith RE, Zweerink HJ, Joklik WK. 1969. Polypeptide components of virions, top component and cores of reovirus type 3. Virology 39:791– 810. Starnes MC, Joklik WK. 1993. Reovirus protein lambda 3 is a poly(C)dependent poly(G) polymerase. Virology 193:356 –366. Strong JE, Coffey MC, Tang D, Sabinin P, Lee PW. 1998. The molecular basis of viral oncolysis: usurpation of the Ras signaling pathway by reovirus. EMBO J. 17:3351–3362. Tyler KL, Fields BN. 2001. Mammalian reoviruses, p 1729. In Knipe DM, et al (ed), Fields virology, 4th ed. Lippincott Williams & Wilkins, Philadelphia, PA. Tyler KL, et al. 1995. Differences in the capacity of reovirus strains to induce apoptosis are determined by the viral attachment protein sigma 1. J. Virol. 69:6972– 6979. van den Wollenberg DJ, et al. 2008. A strategy for genetic modification of the spike-encoding segment of human reovirus T3D for reovirus targeting. Gene Ther. 15:1567–1578.

jvi.asm.org 7413