Cancer Gene Therapy (2012) 19, 38–48 r 2012 Nature America, Inc. All rights reserved 0929-1903/12 www.nature.com/cgt ORIGINAL ARTICLE Molecular network pathways and functional analysis of tumor signatures associated with development of resistance to viral gene therapy T-J Song1,2,*, D Haddad1,3,*, P Adusumilli1, T Kim1, B Stiles1, M Hezel1, ND Socci1,4,MGo¨nen5 and Y Fong1 1Department of Surgery, Memorial Sloan-Kettering Cancer Center, New York, NY, USA; 2Department of Surgery, New York College of Medicine, Korea University, Seoul, South Korea; 3Institute for Biochemistry and Virchow Center for Experimental Biomedicine, University of Wuerzburg, Wuerzburg, Germany; 4Department of Bioinformatics, Memorial Sloan-Kettering Cancer Center, New York, NY, USA and 5Department of Epidemiology and Biostatistics, Memorial Sloan-Kettering Cancer Center, New York, NY, USA Replication-competent attenuated herpes simplex viruses have proven effective in killing many cancer cell lines. However, determinants of resistance to oncolytic therapy are mostly unknown. We developed viral therapy-resistant cells and examined changes in gene-expression pattern compared with therapy-sensitive parental cells. Colon cancer cell line HT29 and hepatoma cell line PLC5 were exposed to increasing concentrations of virus G207. Therapy-resistant cells were isolated and grown in vitro. Tumorigenicity was confirmed by ability of cell lines to form tumors in mice. Human Genome U133A complementary DNA microarray chips were used to determine gene-expression patterns, which were analyzed in the context of molecular network interactions, pathways and gene ontology. In parental cell lines, 90–100% of cells were killed by day 7 at 1.0 multiplicity of infection. In resistant cell lines, cytotoxicity assay confirmed 200- to 400-fold resistance. Microarray analysis confirmed changes in gene expressions associated with resistance: cell surface proteins affecting viral attachment and entry, cellular proteins affecting nucleotide pools and proteins altering apoptotic pathways. These changes would decrease viral infection and replication. Our study identifies gene-expression signatures associated with resistance to oncolytic viral therapy. These data provide potential targets to overcome resistance, and suggest that molecular assays may be useful in selecting patients for trial with this novel treatment. Cancer Gene Therapy (2012) 19, 38–48; doi:10.1038/cgt.2011.64; published online 21 October 2011 Keywords: herpes simplex virus; oncolytic viral therapy; G207; molecular networks; signaling pathways Introduction expression analysis using complementary DNA (cDNA) GeneChip microarray Human Genome U133A Genetically engineered, attenuated herpes simplex virus (Affymetrix, Santa Clara, CA) to determine the possible (HSV) is a promising new modality to treat cancer. Such predictive gene changes in viral-resistant cells compared HSV viruses have been shown to be effective against with viral-sensitive cells. These changes may enable many cancer types, including colorectal, hepatoma, lung, identification of cancers resistant to viral therapy, thus prostate, bladder and breast.1–11 Among these viruses, avoiding undesirable side effects associated with higher G207 and NV1020 are currently in clinical trials. The doses of viral treatment. Moreover, if gene-expression efficacy of HSV therapy depends on both infectivity and changes associated with resistance to therapy are identi- replication. Although initial higher cell kill is achieved fied, new strategies can potentially be developed in order with a higher concentration of virus, this in turn has an to overcome them. increased potential for toxicity in vivo and may therefore not be desirable. In this study, we conducted gene- Materials and methods Correspondence: Dr Y Fong, Department of Surgery, Memorial Sloan-Kettering Cancer Center, 1275 York Avenue, New York, NY Cell lines and cell culture 10065, USA. One colon cancer (HT29) and one hepatoma cell line E-mail: [email protected] (PLC5) were used for constructing resistance to oncolytic *These authors contributed equally to this work. HSV. Colon cancer cell line HT29 (HTB-38) was grown in Received 22 June 2011; accepted 16 August 2011; published online American Type Culture Collection (ATCC) medium 21 October 2011 (Manassas, VA; McCoy’s 5a medium with 1.5 mmol l–1 Gene expression in therapy-resistant cancer T-J Song et al 39 L-glutamine and 10% fetal calf serum), and maintained were noted to be resistant up to an MOI of 400 via under standard cell culture conditions. cytotoxic assay. These resistant cells were then frozen and The human hepatoma cell line PLC5 (PLC/PRF/5) was used for subsequent experiments, as described below. This grown in ATCC medium (Eagle’s minimum essential resulted in a pressure selection of viral-resistant tumor –1 medium with 2 mmol l L-glutamine and Earle’s balanced cells with a differing cellular makeup than the parent cells. salt solution adjusted to contain 1.5 g l–1 sodium bicarbo- Cytotoxicity profiles were assessed by cellular LDH nate, 0.1 mmol l–1 nonessential amino acids and release assay. 1.0 mmol l–1 sodium pyruvate with 10% fetal calf serum) under standard cell culture conditions. Tumorigenicity assay To confirm the tumor origin of the resistant cell lines, 7 Virus normal parent cells (1 Â 10 ) were injected subcutaneously in the left flank of athymic nude mice, whereas resistant G207 is an engineered HSV type 1 (HSV-1) replication- 7 competent, second-generation oncolytic HSV. Con- cells (1 Â 10 ) were injected in the right flank, to assess for structed from R3616, it has a mutation of the ICP6 gene, tumor growth over a period of 14 days. deletion of both copies of the g134.5 genes and an insertion of the Escherichia coli lac Z marker gene into Preparation of RNA for microarray the U 39 gene.11,12 The g 34.5 gene deletion decreases Total mRNA preparation was performed in a 10-cm L 1 7 neurovirulence of the virus,11–13 and disruption of the (diameter) petri dish culture vessel (monolayer, 1 Â 10 cells UL39 gene eliminates ribonucleotide reductase activity, in confluence). Lysis was performed directly in the vessel thereby increasing specificity of the virus to proliferating following RNeasy mini kit protocol (Qiagen, Valencia, cells such as tumor cells.12,14–16 The G207 virus was CA). The mRNA samples were measured by spectro- propagated as previously described, and titer levels were photometer for proof of purity and hybridized to Human determined by standard viral plaque assay.17 Genome U133A cDNA microarray chips (Affymetrix) by the genomic core laboratory at Memorial Sloan-Kettering Cancer Center. Two samples each of parent and G207- Lactate dehydrogenase cytotoxicity assay resistant tumor cells were prepared and used. CytoTox 96 nonradioactive cytotoxic assay kit (Promega, Madison, WI) was used for lactate dehydrogenase (LDH) cytotoxicity assays. This is a colorimetric assay, where the Data analysis The chip images were scanned and processed to CEL intensity of color formation post addition of reagent is files using the standard GeneChip Operating Software proportional to the number of lysed cells (which release (GCOS) analysis suite (Affymetrix). The CEL files were LDH), thus allowing quantification of the proportion of then normalized and processed to signal intensities using cells killed by infection with the virus. the gcRMA algorithm from the Bioconductor library Assays were performed for both nonresistant parent for the R statistical programming system (Institute for cells and resistant cells. 1 Â 104 cells per well were plated Statistics and Mathematics, Wein, Austria; http://cran. in 24-well plates in 1 ml medium per well. After r-project.org/). All subsequent analyses were done on the incubation for 6 h, cells were infected with variable log (base 2) transformed data. To find differentially combinations of multiplicity of infection (MOI), which expressed genes, a moderated t-test was used as imple- describes the proportion of viral particles per cell. Viral mented in the Bioconductor LIMMA package (Fred toxicity was measured at certain time points over 7 days. Hutchinson Cancer Research Center, Seattle, WA, USA; Infected cells were washed with phosphate-buffered saline http://www.bioconductor.org). To control for multiple and lysed by 1.5% Triton X-100 (Dow, Midland, MI) to testing, the false discovery rate (FDR) method was used, release intracellular LDH. A total of 50 ml of the lysates with a cutoff of 0.05. Genes were then grouped into was then placed in 96-well reading plates, and after specific ontology functional groups using the Ingenuity addition of 50 ml of assay reagent-buffer solution, the Pathways Analysis software (Ingenuity Systems, Red- visible wavelength absorbance (490 nm) data were col- wood City, CA), also known as IPA. Network and lected using a 96-well plate reader. All samples were pathway analysis was then performed to find direct and analyzed in triplicate and results expressed as the indirect interactions on common same-directional up- or surviving fraction of cells, as determined by the measured down-regulated genes 42-fold. Functional analysis was absorbance of each sample relative to control untreated performed, also utilizing IPA. cell lysates. Isolation of virus-resistant cells The two hepatoma and two colon cancer cell lines were Results cultured and serially infected with increasing G207 doses Generation of viral-resistant cells at variable MOIs ranging from 0.001 to 200 plaque- Serial passaging with increasing MOIs of G207 in the one forming units. When B80% cell kill was achieved with hepatoma and one colon cancer cell line resulted in the each dose, the surviving cells were re-cultured in fresh development of virally resistant tumor cells with differing medium, allowed to proliferate and subsequently infected cellular makeup compared with the parent cell lines. with a higher MOI of virus. This was repeated until cells These were used for our study. Cancer Gene Therapy Gene expression in therapy-resistant cancer T-J Song et al 40 Tumorigenicity the total common genes) and 38 downregulated (51.35%).