|Home | About | Journals | Submit | Contact Us | Français|
Oncolytic herpes simplex viruses (oHSVs) are promising anticancer therapeutics. We sought to characterize the functional genomic response of human cancer cells to oHSV infection using G207, an oHSV previously evaluated in a phase I trial. Five human malignant peripheral nerve sheath tumor cell lines, with differing sensitivity to oHSV, were infected with G207 for 6 h. Functional genomic analysis of virus-infected cells demonstrated large clusters of downregulated cellular mRNAs and smaller clusters of those upregulated, including 21 genes commonly upregulated in all five lines. Of these, 7 are known to be HSV-1 induced and 14 represent novel virus-regulated genes. Gene ontology analysis revealed that a majority of G207-upregulated genes are involved in Janus kinase/signal transducer and activator of transcription signaling, transcriptional regulation, nucleic acid metabolism, protein synthesis and apoptosis. Ingenuity networks highlighted nodes for AP-1 subunits and interferon signaling via STAT1, suppressor of cytokine signaling-1 (SOCS1), SOCS3 and RANTES. As biological confirmation, we found that virus-mediated upregulation of SOCS1 correlated with sensitivity to G207 and that depletion of SOCS1 impaired virus replication by >10-fold. Further characterization of roles provided by oHSV-induced cellular genes during virus replication may be utilized to predict oncolytic efficacy and to provide rational strategies for designing next-generation oncolytic viruses.
Oncolytic viruses are rapidly advancing as anticancer therapeutics.1,2 These viruses reduce tumor burden via mechanisms including direct cell lysis (virus replication), disruption of tumor vasculature and induction of antitumor immunity.3 Early clinical trials for oncolytic viruses have shown strong safety profiles and a modest antitumor effect.4 Specifically, oncolytic herpes simplex virus (oHSV) mutants have demonstrated robust antitumor activity against preclinical human tumor models and strong attenuation in nonhuman primates.5–10
G207, a multi-attenuated oHSV deleted at the UL39 and γ134.5 loci and with a truncation at the UL3 locus, was evaluated in a phase I trial that showed clinical safety and radiographic evidence of decreased tumor burden.11,12 As early clinical trials have shown potential utility of oncolytic viruses, it is important to further characterize the interaction between oncolytic virus and the infected cancer cell. Understanding the environment of the infected cancer cell on a molecular level may allow for future engineering of oncolytic viruses with enhanced antitumor activity while retaining a robust safety profile.
Functional genomic studies to identify cellular genes regulated by HSV-1 have utilized viruses, including wild-type strains (F, KOS, 17, McKrae), and mutants at the UL41 (R2621) and γ134.5 (17termA, R3616) loci.13–22 With the exception of one array study that utilized the human glioma cell line U87, all previous studies examined effects of HSV-1 infection in either rodent cells/tissues or growth-arrested human cells. These studies gave insight into the function of viral proteins and began to reveal the transcriptomic effect of HSV-1 infection. Although HSV-1 infection downregulated a majority of cellular mRNAs, a more striking finding was that a subset of cellular genes, especially those regulating transcriptional regulation and the antiviral response, were consistently upregulated. Overall, these reports suggest that the cellular response to infection is deliberate and may be crucial for virus propagation.23 Although these studies provided insight into the cellular response to HSV infection, functional genomic studies assessing oncolytic viruses should be performed by infection of human cancer cells. As antitumor efficacy of oHSV is determined by its capacity to replicate, it is important to identify cellular processes required for efficient viral propagation. We hypothesized that genomic profiling of oHSV infection in human cancer cells would allow for identification of cellular genes or processes that govern cellular sensitivity to viral infection.
In the studies described here, we characterized the transcriptomic effect of G207 infection in a panel of human malignant peripheral nerve sheath tumor (MPNST) cells. We utilized Ingenuity software to analyze lists of G207-upregulated genes. These studies identified genes involved in transcriptional regulation, nucleic acid binding/metabolism and Janus kinase (JAK)/signal transducer and activator of transcription (STAT) signaling. Ingenuity network analysis indicated interactions for Fos, STAT1 and EGR1. Confirmatory reverse transcription-PCR studies demonstrated that sensitivity of MPNST cell lines to G207 correlated with suppressor of cytokine signaling-1 (SOCS1) upregulation. Finally, shRNA-mediated depletion of SOCS1 impaired the efficiency of G207 virus production. These data suggest that SOCS1 upregulation is required for maximal G207 replication. Studies to further define the role of SOCS1 and the other upregulated genes in oncolytic virus replication are warranted.
G207 has been well described and is derived from HSV-1 F strain with deletions for UL39 and both γ134.5 loci.11,24 Culture conditions for Vero and human MPNST (S462, 90-8, ST8814, STS26T and T265p21) cells have been described.25 For shRNA studies, lentiviral particles containing control or shRNA-expressing plasmids were purchased from Sigma (St Louis, MO) and manufacturer protocols followed. Cells were grown in 2 µgml−1 puromycin (Sigma).
Cells were infected with G207 at a multiplicity of infection = 5. After 6 h, cells were overlayered with 1.5 ml of TriZol (Invitrogen, Carlsbad, CA). The homogenate was centrifuged for 20 min (4 °C) at 12 000 × g, supernatant transferred to a fresh tube and precipitated by addition of 1 ml chloroform. The sample was incubated for 3 min at RT and centrifuged for 15 min (4 °C) at 10 000 × g. The aqueous phase was precipitated by the addition of 2.5 ml isopropanol. The precipitate was incubated for 10 min at RT and centrifuged for 20 min (4 °C) at 10 000 × g. The sample was washed 1 × with 70% ethanol, air-dried and resuspended in diethyl pyrocarbonate water. RNA was purified by RNeasy Clean up column (Qiagen, Valencia, CA), quantified and checked for purity by comparison of absorbance at 260 and 280nm in Nanodrop Spectrophotometer (Nanodrop, Rockland, DE). Total RNA was analyzed for integrity and concentration by microanalysis in an Agilent 2100 bioanalyzer (Agilent, PaloAlto, CA). A total of 10 µg RNA was combined with 100 pmol of T7-dT-primer and incubated at 70 °C for 10 min. A total of 4 µl of 5 × first-strand synthesis buffer, 2 µl of 0.1 M dithiothreotol and 1 µl of 10 mm dNTP mix was added to the reaction and incubated for 2 min at 42 °C. A total of 1 µl (200U µl−1) of reverse transcriptase was added and incubated for 1 h at 42 °C. Second-strand synthesis was accomplished by addition of 91 µl of diethyl pyrocarbonate water, 30 µl of 5 × second-strand synthesis buffer, 3 µl of 10 mm dNTP mix (200 µm each), 4 µl (40 U) of DNA polymerase, 1 µl (10 U) of DNA ligase (E. coli 10 U µl−1). The reaction was incubated at 16 °C for 2 h. An additional 2 µl (20 U) of T4 DNA polymerase was added to the mix and incubated for an additional 5 min at 16 °C. The reaction was stopped by the addition of 10 µl of 0.5 M EDTA and was cleaned by absorption over a Phase Lock Gel light spin column (Eppendorf, Westbury, NY). Biotinylation and amplification of cDNA was accomplished by Enzo BioArray High Yield RNA Transcript Labeling Kit (Enzo Life Sciences, Farmingdale, NY). Entire quantity of synthesized cDNA was added to 4 µl of HY reaction buffer, 4 µl of 10 × biotin labeled ribonucleotides, 4 µl of dithiothreotol, 4 µl of 10 × RNase inhibitor mix, 2.0 µl of 20× T7 RNA polymerase and diethyl pyrocarbonate water to bring the volume to 40 µl. The reaction was incubated for 4 h at 37 °C followed by purification by absorption over an RNeasy column (Qiagen, Valencia, CA). Before hybridization, 15 µg of cRNA was fragmented by incubation for 35 min at 94 °C. Hybridization to U133A plus 2.0 human genome chips was performed according to Affymetrix protocols. A total of 15 µg of fragmented cRNA was added to 5 µl of control oligonucleotide B2 (3 nm), 15 µl of 20× eukaryotic hybridization controls (bioB, bioC, bioD and cre), 3 µl of herring sperm DNA (10 mg ml−1), 3 µl of acetylated bovine serum albumin (50 mg ml−1), 150 µl of 2 × hybridization buffer and H2O to a final volume of 300 µl. The hybridization cocktail was heated to 95 °C for 5 min. Affymetrix arrays were prepared by wetting for 10 min at 45 °C with 1 × hybridization buffer. The hybridization cocktail was transferred to a 45 °C heat block for 5 min and was clarified by centrifugation. The buffer solution was removed from the array and the clarified hybridization cocktail was added to the array cartridge. The array was hybridized for 16 h in a 45 °C rotisserie oven (30 r.p.m.). Hybridized arrays were washed, labeled with phycoerythrin-conjugated streptavidin (Molecular Probes, Eugene, OR) and scanned on the Affymetrix scanner. The image file was captured on an Affymetrix GeneChip Scanner 3000. Genechip Operating Software 1v4 (Affymetrix, Santa Clara, CA) was then used to generate .cel files. Standard Affymetrix internal control genes were used to check the quality of the assay by the signals of the 3′ probe set to the 5′ probe set of the internal control genes, GAPDH and β-actin, with acceptable 3′–5′ ratios between 1 and 3. Eukaryotic spike controls confirmed proper hybridization of target RNA to the array.
Raw expression data in .cel files were imported into Genespring (Silicon Genetics, Santa Clara, CA) and RNA processed using quantile normalization. Data were normalized in Genespring via data transformation, per chip normalization to the 50th percentile and per gene normalization to the average of the corresponding time-matched mock-infected samples. For statistical analysis, we performed a one-way ANOVA with nonparametric analysis with a false discovery rate of 0.025 with multiple testing correction by Benjamin and Hochberg FDR. The number of probe sets passing these criteria was 7817. This list was filtered on expression level for <0.7 and >1.3 giving 7351 probe sets. K-means clustering was performed for a maximum of 5 clusters.
Malignant peripheral nerve sheath tumor cells were infected with G207 at a multiplicity of infection = 5 and total RNA collected at 6 h using Trizol. A total of 2 µg total RNA was combined with 1 µg of oligo-dT (Invitrogen) and incubated at 70 °C for 10 min, and then chilled on ice. A total of 4 µl of 5 × first-strand synthesis buffer (Invitrogen), 2 µl of 0.1 M dithiothreotol, 0.5 µl of 0.1M dNTP, 0.5 µl of RNAguard (Amersham Biosciences, Piscataway, NJ) and 1 µl of M-MLV RT (Promega, San Luis Obispo, CA) were added and incubated at 37 °C for 1 h. The reaction was stopped by 10 min incubation at 70 °C. A total of 1 µl of synthesized cDNA was combined with primers (5 pmol) and 10 µl of 2 × DyNamo HS SYBR Green qPCR master mix (New England BioLabs, Ipswich, MA). A test run of gradient temperature setting was done to optimize annealing temperature. RT-qPCR was performed in DNA Engine Opticon 2 Continuous Fluorescence Detection System (Bio-Rad, Hercules, CA). Primers are listed in Supplementary Table 2.
Malignant peripheral nerve sheath tumor cells were infected with G207 at a multiplicity of infection = 5. At 6 hpi, protein was harvested using M-PER (Pierce, Rockford, IL) with 50mM NaF, 1mm NaVO3 and 1 × protease inhibitor (Roche, Indianapolis, IN), sonicated and ultracentrifuged (18 000 × g). Western blots were performed as described, and primary incubation was performed with α-SOCS1 (Abcam, Cambridge, MA) or α-actin (1:1000) (in house).25
We previously documented the sensitivity of human MPNST cell lines S462, ST8814, 90-8, STS26T and t265p21 (listed in order of sensitivity from high to low) to productive infection by G207.25 To identify cell response gene signatures to G207 infection, we undertook a transcriptional profiling approach with these five human MPNST cell lines that were either mock-treated or G207-infected for 6 h. This time point was chosen as it has shown maximal virus-mediated alteration of cellular gene expression.13,18 Gene expression changes induced by infection were identified by referencing the gene expression level for each probe to that in the corresponding uninfected culture for each cell line.
A gene list containing probe sets differentially regulated between infected and mock samples and further filtered for virus sensitivity contained 7351 probe sets. Graphical representation by hierarchical clustering of these probe sets across MPNST cell lines revealed four large clusters of downregulated genes, containing 1375, 1306, 928 and 656 genes, and a single cluster of 1025 upregulated genes (Figures 1a and b; Table 1). As predicted for a virion host shutoff (UL41)-expressing virus, the major effect was a large-scale loss of cellular mRNAs consistent with virus-mediated degradation. As HSV-1 has evolved mechanisms to degrade cellular mRNA by both specific and nonspecific routes, we focused on further analysis of G207-upregulated probe sets. Of the 1677 upregulated probe sets (Supplementary Table 1), 343 were upregulated by > 2.5-fold. Of these 343 probe sets, 172, 63, 38, 31 and 38 (representing 21 genes) were shared by one, two, three, four or all five MPNST cell lines, respectively (Figure 2). To illustrate the validity of our experimental approach, of the 21 genes commonly upregulated in all five lines, 7 have been previously reported to be HSV-1 induced (Table 2, asterisks). These 21 genes add to the emerging transcriptome for HSV-1 infection in human cancer cells. Included in this list are 14 potentially novel HSV-1- regulated genes involved in mRNA splicing, protein translation, transcriptional regulation and cell survival. Of note are the highly conserved genes, tncRNA and MALAT1, that encode large noncoding nuclear mRNAs that likely function in mRNA metabolism.26 Interestingly, TncRNA suppresses expression of class II major histocompatibility complex, as well as the allogenic B-cell response.27
The gene ZFP36 or tristetraprolin (TTP) is known to accumulate during HSV-1 infection.28–30 TTP regulates mRNA stability of many cellular genes including GM-CSF, IL-3, TNFα, IL-2, and FOS, acting to limit the inflammatory response.31,32 TTP activation and interaction with viral virion host shutoff protein enables preferential translation of viral transcripts among a background of host mRNA degradation.18,30 Cellular genes controlling translational initiation and elongation including eIF4A, EIF1, eIF4G2 and EIF3S9 were upregulated and may act to form a ribosomal complex uniquely configured to promote efficient translation of viral transcripts. Specifically, eukaryotic translation initiation factor 4A (eIF4A1) was upregulated in all five MPNST lines and regulates translation as a DEAD-box RNA helicase. Its role may be crucial for the increased protein production during virus replication or perhaps to accommodate HSV-derived transcripts that contain a high degree of secondary structure.33 Interactions between viral virion host shutoff and cellular translation factors eIF4A and eIF4H may mediate selective mRNA degradation during HSV-1 infection.34
G207-induced genes contained DNA-interacting genes including GADD45a, GADD45b, GADD34 and c-FOS, each previously identified to be upregulated during HSV-1 infection.18,35,36 Interestingly, cellular transcription factors including Oct-1 and Egr-1 were upregulated by G207. Oct-1 interacts with VP16 and is a key cellular factor required for efficient expression of immediate early HSV-1 genes.37,38 The stress response protein Egr-1 has been shown to repress expression of HSV-1 immediate early proteins ICP4 and ICP22 and is thought to play a role during virus latency and reactivation.39,40
The gene list of G207-upregulated genes were clustered on the basis of known process, function or cell component using the GeneOntology database. Classification by biological process revealed importance of the JAK/STAT pathway and regulation of cellular metabolism at levels including nucleic acid metabolism, transcriptional regulation, DNA packaging, apoptosis and protein synthesis (Table 3). Classification by molecular function revealed additional genes involved in Notch signaling, poly(A) binding and ribosomal components. Classification based upon cellular compartment showed a significant number of nuclear proteins (P = 0.007) and genes involved in the viral envelope and capsid (P = 0.039) (data not shown). Cellular processes are highly manipulated during HSV infection and highlight the high degree to which HSV infection is host cell dependent.
Ingenuity pathway analysis-generated lists for genes upregulated by G207 and involved in gene expression, immune/antiviral response or cell death were further analyzed via Ingenuity-defined functional interaction networks. Genes involved in transcriptional regulation included transcription factors such as AP-1 subunits, STAT1, NFATC1 and EGR1, while genes involved in cell death included BCL2L11 (BIM), BCL6 and GADD45β (Supplementary Figure 1a–c). Ingenuity pathway analysis network analysis for genes upregulated in the immune/antiviral response contained nodes for Rantes (CCL5), IL-1R, EGR1, STAT1, and SOCS1 (Figure 3). Genes involved in JAK/STAT signaling were significantly upregulated by G207. We noticed that upregulation of the JAK/STAT signaling inhibitor SOCS1 occurred only in cells sensitive to G207. As SOCS1 acts downstream to limit JAK/STAT phosphorylation,41 we speculated that upregulation of SOCS1 may play a role to facilitate oHSV replication.
We utilized quantitative real-time RT-PCR to verify consistent upregulation of genes including GADD45β and FOS. Examination of transcript levels for STAT1 and SOCS1 were interesting in that the oHSV-sensitive cell line S462 showed little if any STAT1 upregulation and robust SOCS1 upregulation (Figure 4). The converse was true for the oHSV-resistant cell line t265p21, with the other lines falling into an intermediate range. Overall, using biologic replicates, we achieved ~53% validation of 12 genes.
Our previous studies revealed that S462 supports robust G207 replication but t265p21 is nonpermissive.25 Although the absolute levels of SOCS1 did not correlate with virus replication, the stimulation of SOCS1 expression by virus infection in the permissive S462 cells and its absence in the nonpermissive t265p21 cells suggested a possible role for SOCS1 upregulation in supporting virus replication. To further determine if SOCS1 plays a role in G207 replication in human MPNST cells, we generated cells stably expressing a control or SOCS1-targeted shRNA. Knockdown of SOCS1 protein was verified by western blot at baseline and upon infection with G207 (Figure 5a). No effect on cellular proliferation was noted between SOCS1-depleted, control shRNA and parental cells. Parental S462 cells are highly permissive for G207 infection showing a 4-log increase in virus production in 48 h. In contrast, SOCS1-depleted S462 cells showed significantly impaired virus replication, reduced by >10-fold (Figure 5b). This result suggests that upregulation of SOCS1 is important for maximally productive G207 infection.
We identified genes induced by G207 infection in human MPNST cells. Of these genes, 7 are known to play roles in HSV-1 infection. Comparison of genes discordantly regulated in cells that differed in virus sensitivity identified potentially important cellular genes crucial for virus replication. Ingenuity analysis revealed nodes representing functional interactions for G207-upregulated genes including Egr-1, c-Fos, STAT1, SOCS1, SOCS3 and Bcl6. Knockdown of one of these genes, SOCS1, led to a significant impairment in G207 virus replication in MPNST cells. Because SOCS1-upregulation was not observed in the cell line that does not support virus replication, it is possible that this gene represents a surrogate predictor of oHSV sensitivity. More cell lines that do not upregulate SOCS1 would be needed to confirm this hypothesis. Our finding that SOCS1 plays a significant role in virus replication suggests that inhibition of JAK/STAT signaling may enhance productive replication of G207.
The SOCS proteins are potent negative feedback regulators of the JAK/STAT and c-Kit signaling pathways. SOCS proteins serve to limit the immune response and are required for normal homeostasis. Inhibition of STAT1 signaling by SOCS1 regulates innate immune recognition via double-stranded RNA-activated pathways in human keratinocytes.42 SOCS1, via its SH2-binding domain, has been reported to possess tumor suppressor activity and inhibit JAK/STAT, c-Kit and IL-3 receptor signaling.43,44 Intriguingly, a role for SOCS1 in antitumor dendritic cell vaccines has also been identified. SOCS1 depletion allowed for dendritic cells to break self-tolerance and induce antitumor immunity via elevated IL-12 signaling and persistent antigen presentation.45 A recent report showed that human neuroendocrine tumor cells upregulate SOCS1 to effectively desensitize themselves to growth suppression by type I interferon; SOCS1 downregulation induced apoptosis.46 As we did not observe increased cell death in cells exposed to SOCS1 shRNA, induction of apoptosis was not likely the mechanism of reduced virus replication in our studies. Another SOCS family member has also been reported to regulate HSV-1 infection. Yokota et al.47,48 showed that induction of SOCS3 was required for efficient replication of wild-type HSV-1 (strain RV3) in human cell lines. Similar to our study, these authors found that the upregulation of SOCS3 was important to determine virus sensitivity. SOCS family members play important roles in immune recognition of many human diseases and are regulated by multiple cellular pathways. It is possible that manipulation of SOCS proteins may allow for enhancement of oncolytic virus activity against human cancers.
Herpes simplex virus-1 has evolved mechanisms to bypass host immunity by degradation of cellular mRNAs encoding major histocompatibility complex proteins, by ICP47-mediated suppression of antigen presentation and others.49,50 Viral proteins interact with cellular proteins to effectively facilitate replication of virus progeny.51 HSV-1-mediated degradation of cellular mRNAs via viral virion host shutoff protein and inhibition of mRNA splicing and maturation by ICP27 reprogram cellular processes. HSV proteins (ICP6, ICP34.5 and US11) interact with protein translation machinery to promote preferential translation of HSV-derived transcripts.51 Translation of HSV-derived mRNAs that suppress apoptosis may be required for efficient virus replication. In our study, we found numerous G207-upregulated mRNAs that controlled cellular gene expression, translation and antiviral immunity.
As the field of oncolytic viruses moves forward, it will be important to fine-tune these biologic therapeutics. Rational vector design may be achieved by strengthening our understanding of virus-induced alterations of intracellular gene expression and extracellular cytokine signaling. As SOCS family proteins have been implicated in replication of human viruses, these proteins hold promise to achieve increased efficacy of therapeutic viral vectors. For example, our findings predict that forced, exogenous SOCS1 expression as a transgene in an oncolytic HSV vector may confer virus replication and oncolysis in cells such as t265p21 that do not normally support virus infection. Functional genomic analysis of oHSV infection of human xenografted tumors in vivo will additionally help identify extracellular factors that promote or inhibit intratumoral viral infection and spread.
We thank MediGene Inc. for providing G207. This work was supported by the Cincinnati Children’s Hospital Medical Center Division of Hematology and Oncology, by a University of Cincinnati Functional Genomics Seed Grant to YYM, and NIH grant R01-CA114004 to TPC.
The authors have no conflicting financial interests.
Supplementary Information accompanies the paper on Cancer Gene Therapy website (http://www.nature.com/cgt)