Cytomegalovirus infection of glioblastoma cells leads to NF-κB dependent upregulation of the c-MET oncogenic tyrosine kinase

Abstract

Cytomegalovirus (CMV) is widespread in humans and has been implicated in glioblastoma (GBM) and other tumors. However, the role of CMV in GBM remains poorly understood and the mechanisms involved are not well-defined. The goal of this study was to identify candidate pathways relevant to GBM that may be modulated by CMV. Analysis of RNAseq data after CMV infection of patient-derived GBM cells showed significant upregulation of GBM-associated transcripts including the MET oncogene, which is known to play a role in a subset of GBM patients. These findings were validated in vitro in both mouse and human GBM cells. Using immunostaining and RT-PCR in vivo we confirmed c-MET upregulation in a mouse model of CMV-driven GBM progression and in human GBM. siRNA knockdown showed that MET upregulation was dependent on CMV-induced upregulation of NF-κB signaling. Finally, proneural GBM xenografts overexpressing c-MET grew much faster in vivo than controls, suggesting a mechanism by which CMV infection of tumor cells could induce a more aggressive mesenchymal phenotype. These studies implicate the CMV-induced upregulation of c-MET as a potential mechanism involved in the effects of CMV on GBM growth.

1. INTRODUCTION

The relationship between cytomegalovirus (CMV) and cancers including glioblastoma (GBM) has been under investigation for many years but remains poorly understood [1]. GBM is the most common malignant primary tumor of the central nervous system (CNS) [2]. The current standard of care for patients with GBM is surgical resection, followed by alkylating chemotherapy treatment with temozolomide (TMZ) and irradiation. Despite treatment, the median survival of GBM patients remains a mere 15 months [3,4] GBM is characterized by heterogeneous genetic alterations involving the tumor suppressor genes p53 and RB, and activation of tyrosine kinase signaling via alterations in EGFR, as well as other tyrosine kinase receptors including c-MET [5,6]. Transcriptomic classification can further distinguish GBM into classical, mesenchymal, and proneural subtypes [7]. These transcriptional subtypes were shown to be of clinical relevance, being predictors of resistance to therapy and overall survival [7–9]. Despite recent advances in our understanding of molecular alterations in GBM, effective therapeutic control remains highly challenging, due to multiple factors including tumor heterogeneity, diffuse infiltrative growth, treatment resistance, and tumor recurrence [4,10].

c-MET is considered a proto-oncogene that is highly relevant to GBM biology. It is a tyrosine-protein kinase receptor encoded by the MET gene that is considered a mesenchymal signature gene in GBM and multiple other malignancies [11]. HGF is the major ligand of c-MET, capable of activating signalling pathways that promote proliferation, invasion, and chemoresistance [11–13]. High levels of c-MET have been correlated with shorter overall survival, and low c-MET expression correlates with increased sensitivity of GBM to the alkylating chemotherapeutic drug temozolomide (TMZ) [14, 15]. In addition, expression of c-MET and its co-receptor, CD44, have been shown to mediate resistance to radiotherapy, a clinical mainstay of GBM treatment [16].

Cytomegalovirus (CMV) is a member of the β-herpesvirus family [17] that has been associated with GBM [17–21]. This association is not without controversy [1,22,23], but there are some compelling mechanistic findings that support its role. Multiple reports have implicated various CMV genes in tumor growth in vitro [24–26] and clinical trials targeting CMV in GBM using anti-viral drugs [27–29] or immunotherapeutic approaches have shown promising results [30–34], supporting a role for CMV in GBM progression. Other studies have identified therapy induced CMV reactivation as a cause of encephalopathy and early death in a subset of GBM patients, which can be reversed by treatment with valganciclovir [35–37].

To interrogate the potential role of CMV in GBM in a physiologic context, we have established a murine model in which syngeneic GBM is implanted in mice with murine CMV (MCMV) infection. This shows a phenotype of MCMV-accelerated in vivo tumor growth in GBM that is associated with increased angiogenesis, and which can be reversed with anti-viral treatment [38]. This phenotype is robust and occurs reproducibly in all GBM lines tested so far [Lawler Lab in preparation, 38, 39]. Other murine studies have also shown acceleration of breast cancer metastases in the presence of MCMV [40].

To study the influence of CMV on tumor cells, we performed RNAseq of CMV-infected human GBM cells. In the current report, we demonstrate upregulation of several known GBM-linked transcripts, including the oncogenic tyrosine kinase receptor c-MET [11–13]. Upregulation of these genes is associated with CMV-associated activation of nuclear factor kappa B (NF-κB). We confirm c-MET upregulation in our MCMV-enhanced GBM murine model and also in human GBM specimens, suggesting another potential mechanism by which CMV potentiates tumor growth in GBM.

2. MATERIALS AND METHODS

2.1. Cell culture and virus propagation

Human CMV (Towne strain expressing GFP under control of the UL127 early promoter (Bill Britt, University of Alabama) and murine CMV (Δm157 Smith Strain) were propagated in MRC-5 and NIH/3T3 cells respectively, purified by centrifugation and viral titers were measured as previously described [32]. GL261Luc2 murine GBM cells were purchased from Perkin-Elmer (Boston, MA). Primary human glioma stem cells (GSCs) (G81, G88, G44, G157) were obtained by dissociation of gross tumor samples and cultivated in neurobasal media under an IRB approved protocol as previously described [8]. Mycoplasma testing was routinely done by PCR, and STR profiling was used to confirm cell identity.

2.2. Human Specimens

Human tissue specimens were obtained as approved by the Dana-Farber Cancer Institute (DFCI) IRB.

2.3. CMV infection and in vitro assays

For CMV infection in vitro, we seeded up to 10⁶ cells in 6-well plates and treated with CMV or mock (PBS) the next day. We infected GL261Luc2 cells as described [32] with murine CMV Δm157 Smith strain [multiplicity of infection (MOI), 1.0] or patient-derived human GBM neurospheres with Towne strain human CMV (MOI, 1.0) for 2 hours. Cells were then rinsed with Dulbecco’s phosphate-buffered saline (DPBS) and covered with fresh culture medium. For inhibition of c-MET, we used the small molecule ATP-competitive c-MET receptor PHA665752 [41].

2.4. Lentiviral infection and transfection

Lentiviral packaging plasmids pLP1 (3 μg/10cm² dish), pLP2 (2 μg/10cm² dish) and pLP/VSVG (2 μg/10cm² dish) were mixed with c-MET-pCMV vector (Addgene, Plasmid #37560) (6 μg/10cm² dish) or pCMV control vector (Addgene, Plasmid #17448). HEK293 cells were infected with prepared constructs with Lipofectamine 2000 (Invitrogen) following the manufacturer’s instructions. Medium from transfected cells was collected after 24 and 48 hours and replaced with fresh DMEM/10% FBS. Collected medium was centrifuged for 5 min at 350 g. Supernatant was collected, filtered, and centrifuged at 20,000 g for 2 hours, at 4°C. The pellet was dried and dissolved in 50 μL Neurobasal medium. For lentiviral transduction a pellet of 100,000 viable proneural G44 GSCs was mixed with 50 μL of prepared virus. Cells were incubated for 15 min on ice, 10 min at room temperature, 15 min at 37°C and seeded in 25cm² flasks. Selection with 1μg/ml of puromycin started after 48h following transfection.

2.5. Transwell cell migration assay

Light transmission blocking transwell inserts (FluoroBlock, Corning, NY) with a pore size of 8 μm were used for transwell migration studies. 1 × 10⁶ cells were trypsinized and added to the transwell compartment and incubated at 37°C in a cell culture incubator. Images were taken after 24 h and 48 h. For each assay, 15 images from three separate wells were analysed. Each assay was repeated three times.

2.6. Quantitative PCR

Quantitative PCR was done as previously described [32]. Total RNA was extracted using Trizol (Life Technologies, Carlsbad, CA) and treated with RNase-free DNase (Qiagen, Germany). mRNA expression analysis was carried out using Power SYBR Green (Applied Biosystems, CA). RNA concentration was quantified using a Nanodrop RNA 6000 (ThermoFisher) and analysed using the STEP one PLUS Applied Biosystems PCR machine. Primer sequences were as follows; CD133: FOR 5’-ACTCCCATAAAGCTGGACCC-3’, REV 5’-TCAATTTTGGATTCATATGCCTT-3’; Sox 2: FOR 5’-ACCGGCGGCAACCAGAAGAACAG-3’; REV 5’- GCGCCGCGGCCGGTATTTAT-3’; Notch 1: FOR 5’- AGTGTGAAGCGGCCAATG-3’, REV 5’- ATAGTCTGCCACGCCTCTG-3’; Nestin: FOR 5’- -3’, REV 5’- -3’; Nestin: FOR 5’-CAGCGTTGGAACAGAGGTTGG-3’, REV 5’-TGGCACAGGTGTCTCAAGGGTAG-3’; Olig2: FOR 5’- CTCCTCAAATCGCATCCAGA -3’, REV 5’- AGAAAAAGGTCATCGGGCTC -3’; BCL2A1: FOR 5’- ATGGATAAGGCAAAACGGAG-3’, REV 5’- TGGAGTGTCCTTTCTGGTCA-3’; c-MET: FOR 5’- CATGCCGACAAGTGCAGTA-3’, REV 5’- TCTTGCCATCATTGTCCAAC-3’; Lyn: FOR 5’-CTGAACTCAAGTCACCGTGG -3’, REV 5’-TCCATCGTCACTCAAGCTGT-3’; WT1: FOR 5’- TTAAAGGGAGTTGCTGCTGG-3’, REV 5’- GACACCGTGCGTGTGTATTC-3’; CD44: FOR 5’- CCCAGATGGAGAAAGCTCTG-3’, REV 5’-ACTTGGCTTTCTGTCCTCCA-3’;

2.7. Immunoblotting

Cells were lysed using RIPA buffer containing 1% protease inhibitor (Merck Millipore, MA) and 5% phosphatase inhibitor cocktail (Roche). Total protein concentration was measured using the Bradford protein assay. Primary antibodies used were against HCMV and HCMV pp65 (Virusys Corporation, MD #CA150–1 (1:1000), and #CA003–100 (1:1000) respectively), anti-β-actin (Cell Signaling, MA #4967, 1:1000), anti-c-MET (Cell Signaling, #8198, 1:1000), anti-phospho-c-MET (Tyr1234/1235, Cell Signaling #3077), anti-p65 (Cell Signaling, #8242, 1:1000) and anti-phospho-p65 (Cell Signaling, #3033, 1:1000).

2.8. Immunohistochemistry

Mice were euthanized using CO₂ inhalation and subsequently perfused with neutral-buffered formalin 4% (Sigma-Aldrich, MO) for fixation. Cryoprotection was performed using 30% sucrose. All mouse brain slides were obtained from frozen sections. Permeabilization was done using 1% Triton-X-100 (Sigma-Aldrich) in PBS pH 7.4 for 10 min. Slides were then incubated with the primary antibody (1:100 in normal serum) overnight at 4°C. For detection of the primary antibody, species-matched fluorophore coupled antibodies were incubated for 1 h at room temperature. Slides were then covered with anti-fade mounting medium (Vectashield, Vectorlabs, CA) and coverslipped. All fluorescent and bright-field microscopy-based assays were observed using a Nikon Eclipse Ti microscope (Nikon, Tokyo, Japan). High resolution confocal fluorescent microscopy was performed using a Zeiss LSM 710 confocal microscope system (Carl Zeiss Inc., Germany) and visualized using the ZEN Zeiss Imaging software.

2.9. Oligonucleotide transfection

The siRNA oligonucleotides specific to p65 RELA (CAT#: SR321602) were purchased from OriGene. Negative control siRNA (S103650318) with no homology p65 RELA was used as a negative control. Dissociated single cells from neurospheres were treated with individual siRNA against p65 RELA according to the manufacturer’s instructions. The cells were then incubated for 48 h in supplemented neurobasal media, and RNA/proteins were isolated from the treated cells.

2.10. RNA sequencing

Sequencing of proneural GSCs (G44) prior and after infection with HCMV was performed as previously described [32]. All RNAseq data have been deposited in the ArrayExpress database (E-MTAB-7613). Significantly altered genes with log2 fold-change > 2 were analyzed using ShinyGO v0.61 (ShinyGO v0.61: Gene Ontology Enrichment Analysis + more (sdstate.edu)), with a p-value cutoff of 0.05.

2.11. Mouse GBM models

Mouse GBM models were established according as previously described [32]. The CMV mouse model was established by intraperitoneal inoculation of mouse pups with murine CMV Δm157 Smith strain, followed by GL261 murine intracranial tumor implantation at 16 weeks of age [32]. For the growth of human xenografts, 50,000 G44 cells [42] (mock, or c-MET overexpressing) were implanted into the striatum of nude mice (Charles River) as described [8].

2.12. Statistical Analysis

Imaging assays were quantified using ImageJ (http://rsb.info.nih.gov.ezpprod1.hul.harvard.edu/ij/), including the Analyze Particles function of binary images with automatic threshold. Data are expressed as mean ± SD. The unpaired two-tailed Student’s t-test was used for comparison between two groups. Each group was tested for Gaussian distribution, if one-way ANOVA was passed, followed by Bonferroni’s test. If this failed, the Kruskal-Wallis test followed by Dunn’s correction was conducted to test for significance among multiple groups. Pearson’s correlation with nonlinear regression analysis was performed to compute Pearson’s r, and p-values. Statistical analyses were performed using Microsoft Office Excel 2011 or Graph Pad Prism 6 software. P < 0.05 was considered statistically significant.

3. RESULTS

3.1. Alterations of GBM tumor-promoting gene expression after CMV infection of GSCs

To understand how CMV infection influences GBM cells, we performed RNA-seq on human patient-derived glioblastoma stem-like GSCs (G44) 72 hours post-infection with Towne Strain human CMV. At baseline, G44 cells show proneural transcriptional programs [42], and are permissive to CMV infection. RNAseq analysis after CMV infection showed that there were 3,460 significant differentially expressed genes in G44 cells (Figure 1A, B). Numerous CMV transcripts were also detected, confirming CMV infection (Figure 1C, Supplementary Table 1). Pathway analysis showed upregulation of several pathways, including “Cancer Pathways” and “NF-kB signaling” (Supplementary Table 2). By comparing the list of genes from RNAseq analysis with significantly differentially expressed genes in the Cancer Genome Atlas (TCGA) GBM dataset [5], we found that 242 differentially expressed genes in our CMV dataset were also altered in GBM specimens compared with normal brain (Figure 1D). Of particular interest was the GBM oncogene MET [11–13], which was one of the most upregulated genes after CMV infection (8.13 log₂, p < 0.005) (Figure 1A, E). Because of its known importance in GBM, we therefore investigated the functional consequences of changes to c-MET expression in more detail.

Figure 1: Transcriptome changes in GBM stem cells after infection with HCMV Towne strain.

A total of 3460 genes were differentially expressed more than two-fold (log2) after HCMV infection of G44 proneural human GSCs (p adjusted < 0.005) (A + B) (n = 3). (C) Proportion of CMV transcripts of the total sequenced transcripts post-CMV infection. (D) 159 genes that are upregulated 2-fold (log2) or higher were found to overlap with gene expression changes found in glioblastoma compared to normal brain (TCGA dataset), with 83 downregulated genes also overlapping. (E) 7 out of the 10 most upregulated genes post-CMV infection of GBM cells are likewise commonly overexpressed in GBM compared with normal brain (All genes shown, glioblastoma associated genes are shown in bold)

3.2. Upregulation of c-MET by CMV infection of GBM cells

Hepatocyte growth factor (HGF) signaling via the c-MET receptor is known to contribute to tumor progression through increased proliferation and angiogenesis [11–13]. c-MET, along with its co-receptor, CD44, are typically associated with the mesenchymal transcriptional subtype of GBM [8,37]. Upregulation of c-MET in the proneural tumor type G44 suggested that CMV might tip cells to favor a mesenchymal program. To study this further, we evaluated c-MET and CD44 expression in a panel of patient-derived GSCs, classified as proneural or mesenchymal, according to their transcriptional signatures [7–9]. Both proneural and mesenchymal GSCs were permissive to infection with GFP-expressing Towne strain CMV in vitro (Figure 2A). Over a period of 14-days, CMV infection did not block tumor sphere growth or lead to increased cell death, observed via propidium iodide staining (Figure 2A, B). CMV infection was confirmed by RT-PCR for CMV gB, which showed increased levels over time (Figure 2C). We were not able to detect CMV virus in the conditioned media from these cells (Supplementary Figure 1). CMV infection of G44 GSCs led to a general increase in expression of a panel of mesenchymal markers and a general reduction of proneural markers, suggesting a partial transition to a more mesenchymal-like subtype (Figure 2D). c-MET was clearly the most upregulated of these signature genes, with the greatest increases observed in proneural GSCs based on analysis of GSCs from four different GBM patients (Figure 2E).

Figure 2: CMV infects both proneural and mesenchymal glioblastoma stem cells (GSCs) and significantly upregulates transcription of c-MET.

(A) Human GSCs with proneural (PN – upper panels) or mesenchymal (MES – lower panels) expression subtypes are permissive for CMV infection (assessed by the intensity of GFP fluorescence). Representative images are shown. Bar = 100 μm (B) CMV infection does not increase cell death in GSC neurospheres. The number of the dead cells was assessed by propidium iodide staining. Bar graphs represent the difference in the percent of dead cells between CMV infected and non-infected neurospheres at the times shown (experiments performed in at least 3 biological replicates). (C) CMV Glycoprotein B gene expression increases in a time-dependent manner over the course of infection. Bar graphs represent an absolute quantification of gB gene copy number over time. (Statistical analysis was performed using unpaired t test, **P < 0.05; ***P < 0.005, n= 3) (D) qRT-PCR data showing fold change of genes associated with the proneural and mesenchymal subtypes 72 h after CMV infection (MOI 1), mean ± SD (n = 3). (E) Relative expression of c-MET in HCMV infected proneuronal and mesenchymal GSCs, 72h after infection with CMV (MOI 1), mean ± SD (n = 3).

At the protein level, upregulation of c-MET after CMV infection of human GSCs was confirmed by Western blotting (Figure 3A). At baseline, there were higher c-MET levels in mesenchymal GSCs, than in proneural GSCs. While each cell type showed an increase in c-MET protein levels after CMV infection, GBM cells with a proneural phenotype showed significantly higher c-MET upregulation in comparison to the mesenchymal subtype (Figure 3A). c-MET levels increased in parallel with CMV transcripts over time (Figure 3B) and was partially blocked by incubation with the small molecule anti-viral drug cidofovir (Figure 3C).

Figure 3. CMV causes upregulation of c-MET protein in infected human GSCs.

(A-B) CMV infection (72h, MOI 1) drives the upregulation of c-MET in proneural G44 GSCs, as assessed by Western blotting. Corresponding densitometry measurements are represented in the bar graphs. (C) c-MET upregulation due to CMV infection is reversible in response to cidofovir (CDV), a commonly used antiviral therapeutic, at 48 hours post infection.(D) CMV promotes c-MET phosphorylation and upregulates the c-MET co-receptor, CD44, in uninfected GSCs. Both p-MET and CD44 expression increase as a result of treatment with HGF, a ligand for the CD44 receptor. This upregulation is further augmented by treating CMV-infected GSCs with HGF. Bar graphs show densitometric analysis from at least 5 independent experiments. The expression level of the targeted proteins was normalized to the loading control, GAPDH. Statistical analysis was performed using the Student’s t test. *P < 0.05; **P < 0.005, ***P < 0.0005. Graphical results are a product of 5 independent experiments.

However there was an absence of activating c-MET receptor phosphorylation (Tyr1234/1235), indicating a lack of activation.(Figure 3D). To investigate this further we stimulated CMV infected GSCs with the c-MET ligand HGF, which showed an increase in c-MET phosphorylation indicating the functionality of the upregulated receptor (Figure 3D). In further support of this observation, we found that the c-MET co-receptor CD44 [13] was also upregulated (Figure 3D). Thus, CMV infection of GBM cells results in upregulation of the c-MET/CD44 receptor/coreceptor, which is responsive to HGF in vitro.

3.3. Upregulation of c-MET is associated with CMV in vivo

To determine whether a relationship between c-MET and CMV exists in vivo we performed immunofluorescence analysis of human GBM specimens separated into two groups according patients CMV status determined by IgG seropositivity. RT-PCR analysis of the GBM tissue was performed which showed significant elevation of c-MET in the CMV seropositive patients, whereas CD44 levels remained unchanged (Figure 4A). In addition to detecting CMV by immunofluorescence in GBM specimens of CMV-seropositive patients, we also observed a pronounced upregulation of c-MET in seropositive patients, where it was closely colocalized with positive CMV staining as shown in Figure 4B and Supplementary Figure 2. This data suggests a correlation with c-MET and CMV in vivo in human GBM.

Figure 4: CMV infection is associated with tumoral c-MET expression and an accelerated tumor growth in GBM in vivo.

(A,B) CMV and c-MET are detected in human GBM specimens via qPCR analysis and immunofluorescence (n = 5). Data from both detection methods correlates with patients’ HCMV serostatus. (C) Increased c-MET expression in GL261Luc2 mouse glioblastoma cells after MCMV infection (72h, MOI 1) in vitro (n = 3). (D,E) Immunofluorescence comparing c-MET expression in GL261Luc2 tumors extracted from uninfected versus MCMV-infected mice. The relative difference in c-MET expression was quantified from 3 independent slides (5 fields per section) and is represented in the corresponding bar graph. (F) Staining for Ki-67 (green) and cell nuclei (blue) as visualized by immunofluorescence of sections from GL261Luc2 tumors from 3 independent slides (5 fields per section). The corresponding bar graph compares the relative quantitation of ki67 expression in uninfected versus CMV-infected tumors. Error bars indicate SD. Statistical analysis was performed using the Student’s t test. ***P < 0.005. Bar = 50 μm

To further confirm this, we turned to our murine system, in which mice are infected with MCMV perinatally, followed by tumor implantation 16 weeks post-infection. We previously demonstrated accelerated tumor growth in this model compared to uninfected controls [38] Here we showed that murine GL261luc2 GBM cells which are permissive to MCMV infection show upregulation of c-MET protein by Western blotting similar to human GBM cells post infection in vitro (Figure 4C). RT-PCR showed elevation of c-MET expression in vivo associated with the CMV latent mouse model, and anti-MCMV and anti-c-MET antibodies revealed co-localization in tumors in MCMV infected mice (Figure 4D,E). Finally, tumors grown in the context of MCMV infection displayed elevated ki67 positivity, indicating greater tumor proliferation (Figure 4F) as we previously described [38]. Overall, these experiments support the existence of a MET/CMV relationship in vivo.

3.4. Upregulation of c-MET by CMV requires NF-κB signalling.

Because CMV infection activates the cellular transcription factor NF-κB [43–45], and the c-MET promoter has a potential NF-κB binding site in its promoter region (−745 to −697) [46], we hypothesized that CMV induced activation of NF-κB induces c-MET expression in GBM. Comparison of upregulated genes with known NF-κB targets supported this hypothesis, showing significant overlap between our two datasets (Figure 5A). A p65/RelA siRNA was able to knockdown p65/RelA expression effectively (Figure 5B). Infection of proneural G44 human GSCs increased phosphorylation of the p65/RelA NF-κB subunit, but overall expression of p65 remained unchanged after infection (Figure 5C). Knockdown of p65/RelA using siRNA inhibited p65 expression and phosphorylation and abolished CMV-associated c-MET upregulation (Figure 5C). Thus, c-MET upregulation may be a consequence of activation of NF-κB by CMV infection in proneural GSCs.

Figure 5: c-MET overexpression in human GSCs is NF-kB dependent.

(A) CMV-induced overexpression of NF-kB target genes in human GSCs. (B, C) Graphical representation confirming the siRNA knockdown of p65 RELA expression (NF-kB pathway) based on qRT-PCR data. Western blotting showing that NF-kB pathway knockdown counters the CMV-dependent upregulation of c-MET (mean ± SD). The experiment was performed in 3 independent biological replicates.

3.5. Overexpression of c-MET in the absence of CMV leads to increased growth and migration in vitro and in vivo.

To better understand the importance of c-MET upregulation in GBM, GSCs stably overexpressing c-MET were generated. Overexpression of c-MET in proneural G44 human GSCs is shown by Western blot analysis (Figure 6A). Growth of GSCs in vitro was analysed using the growth in low attachment (GILA) neurosphere assay [47]. Here, proneural wild-type, empty vector, and c-MET transfected GSCs showed no statistically significant growth differences (Figure 6B). This is consistent with the lack of c-MET signalling activity in the absence of its ligand HGF, as described above (Figure 3). However, after addition of HGF, GSCs overexpressing c-MET showed a significant acceleration of growth (p < 0.0001) (Figure 6B).

Figure 6: c-MET overexpression leads to tumor formation and accelerated tumor growth in human GSCs and murine xenograft models.

(A) Western blot for lentiviral induced overexpression of c-MET in GSCs with a constituently low expression of c-MET (proneural). (B) GILA proliferation assay of c-MET overexpressing GSCs. (C) Transwell migration of lenti-c-MET GSCs (mean ± SD). (D) H&E and immunofluorescence staining of tumor sections. Lenti-c-MET shown in green, cell nuclei in blue. (E) Kaplan-Meier survival curve of c-MET overexpressing proneuronal tumor-bearing athymic mice relative to mice bearing wild-type mesenchymal and proneuronal GSC tumors. ***P < 0.005, Student’s t test. Error bars indicate SD. All in vitro experiments were performed in at least 3 independent biological replicates.

Transwell assays showed that migration of GSCs expressing c-MET was increased in the presence of HGF, compared to those transfected with the empty vector. This increase in migration was abrogated by the small molecule ATP-competitive c-MET receptor kinase inhibitor PHA665752 (Figure 6C). To validate our results in vivo, we implanted human proneural G44 human GSCs overexpressing c-MET in athymic nude mice. Here, proneural GSCs do not lead to tumor formation, while mesenchymal GSCs lead to solid tumor formation and progression resulting in a median survival of 42 days. Proneural GSCs transfected with the empty vector likewise showed no tumor growth, while transfection with c-MET led to tumor formation and a substantially shortened survival (median survival of 9 days) (Figure 6D). Brain sections stained with H&E at end-point confirmed the presence of large tumors in proneural GSCs expressing lenti-c-MET, and no tumors in those with the empty vector (Figure 6D). Immunofluorescence detection of GFP expressed by the transfected plasmid showed single tumor cells along the needle track in those injected with proneural GSCs empty-vector. In sharp contrast, there were large tumors with poorly defined borders and numerous invading cells in those animals receiving proneural GSCs expressing c-MET (Figure 6E). Together these data illustrate the importance of c-MET expression in GBM growth.

4. DISCUSSION

CMV has been linked with GBM progression, but the underlying mechanisms remain largely unknown. Because of this, the linkage between CMV and GBM has become clouded with some dispute, with numerous studies supporting the association [1,17–21, 24–26], while others suggest no association [22, 23]. To address this, we have developed a murine model incorporating both CMV and GBM. In this model, C57/BL6 mice are infected at birth with the Δm157 strain of MCMV, which allows efficient infection by evading NK cell responses. In our model, tumors are implanted 16 weeks post-CMV infection [38], which is a time point consistent with latency and when we see no viral activity. This allows evaluation of the impact of CMV on GBM progression in an isolated experimental system, confirming significant impact of previous CMV infection on GBM progression [38]. Accordingly, we have now turned our attention to identifying the mechanisms of GBM potentiation by CMV.

In our previous work, we detected CMV transcription in tumors [38], leading us to hypothesize that CMV might potentiate GBM progression by infecting naïve cells. This hypothesis is supported by the observation that antiviral treatment impairs tumor growth in our model [38], strongly suggesting that infection of new cells is important. One potential infection target is tumor cells, and in the current report we have confirmed that both murine and human GBM can be infected by their species relevant CMV, and that cells so infected live up to 2 weeks and are not lysed (in contrast to HSV). The next logical step was to evaluate what impact such infection has on tumor cells, leading to our RNA-seq analyses to understand the impact of CMV on the GBM cell transcriptome.

RNA-seq analysis showed that CMV infected GBM cells upregulate oncogenic signalling pathways, suggesting potential for driving tumor progression. Of specific interest, one of the most highly upregulated genes is MET, encoding the c-MET receptor tyrosine kinase. Our follow-up RT-PCR and protein studies confirmed that this upregulation leads to functional c-MET receptors that are rapidly activated by HGF. Consistent with other studies, this upregulation of c-MET appears consequent to NF-κB activation, which is known to occur during CMV infection [43–45] and can occur at different stages of the infection cycle. NF-κB also is a major driver of mesenchymal transition in GBM, likely due to upregulation of multiple inflammatory cytokines, as well as via c-MET as we suggest here. Indeed, one of our major findings is that CMV appears to promote a more aggressive mesenchymal phenotype in GBM We also speculate that upregulation of c-MET may offer an advantage to the virus by upregulating cell metabolism or promoting immunoevasion. These areas would be an interesting area of future investigation.

The importance of c-MET in GBM is illustrated by the current report, which shows greatly accelerated tumor growth and early mortality in murine xenograft models when proneural GBM tumor cells overexpress c-MET. Interestingly, although c-MET was upregulated by CMV in GBM cells, exogenous HGF was required for its activation, suggesting paracrine HGF production in the tumor microenvironment is required for receptor activation.

To corroborate the relevance of our observations in vivo, we show correlation of c-MET and CMV in human GBM tissue. This was further confirmed by our murine model, which also shows co-localization of MCMV and c-MET, and significant c-MET expression in MCMV+ mice when compared to naïve controls. Admittedly, previous results [38] suggest that intratumoral virus levels are much lower than those achievable in vitro. This may in part help to explain the difficulty others have had in detecting CMV in human tumors [22, 23]. It is nevertheless intriguing to see obvious c-MET upregulation in tumors in MCMV+ mice in our model, especially because this model has been associated with upregulated angiogenesis [38], a process already linked to c-MET expression [36].

Together, these results suggest a model by which CMV may contribute to GBM tumor progression. Patients with GBM undergo chemo- and radiotherapy, which could cause viral reactivation in seropositive patients [35–37]. GBM tumor cells are susceptible to CMV infection, and after infection could activate NF-κB with upregulation of c-MET in these cells, propelling tumor progression. Concomitantly, CMV drives PDGF-D production in the tumor microenvironment, causing recruitment of pericytes that serve as a scaffold for angiogenesis and tumor progression [38]. Indeed, irradiation has been shown to promote the mesenchymal phenotype in GBM [16] and based on this data, CMV may contribute further to this effect.

In conclusion, we have shown that CMV can infect GBM tumor cells with subsequent extensive alterations of the cellular transcriptome. Further, we found that infection causes NF-κB activation that leads to upregulation of the proto-oncogene c-MET. Given the well described relationship between c-MET and tumor progression, these provocative results suggest another possible mechanism by which CMV enhances GBM progression.

HIGHLIGHTS.

• Cytomegalovirus upregulates a number of tumor-associated genes upon infection of GBM cells in vitro including the GBM oncogene MET.

• MET upregulation is linked to CMV in murine GBM models, and in human GBM specimens

• MET upregulation is dependent on CMV-induced Nf-κB activation

• MET upregulation leads to aggressive growth of proneural GBM in vivo in a murine xenograft model.