Over the past decade, the presence of cytomegalovirus (CMV) has been detected in an increasing number of malignancies. The first discovery occurred in glioblastomas (GBMs). Multiple groups have independently corroborated that CMV DNA, RNA, and protein are present in a large majority of GBMs. CMV has also been detected in
medulloblastomas, colorectal cancer, and prostate cancer. However, the role of CMV in tumors remains controversial since other laboratories were unable to initially replicate the findings. Although CMV is capable of activating several oncogenic pathways, it has never been shown to be a transforming virus. Recent studies have suggested mechanisms by which CMV can modulate the tumor environment. Additionally, CMV has been shown to dysregulate several important cancer pathway including p53, STAT3 and Rb pathways. Viral contribution to tumorigenesis has been established in a growing number of cancers; however, a role of CMV in cancer remains highly controversial.
I hypothesized that CMV infection can modify the rate of glioma pathogenesis. To study the controversial role of CMV in gliomas, I infected Mut3 (GFAP-cre; Nf1loxP/+; Trp53-/+) mice with mouse CMV (MCMV). Mut3 mice develop normally but eventually succumb to spontaneous malignant astrocytomas, including GBMs, as an adult. After viral inoculation, mice exhibited infection throughout the body, including the brain. Mut3 mice infected perinatally succumbed to glioma significantly earlier than mock-infected Mut3 mice. MCMV-infected mice developed GBMs (WHO grade IV) at a higher rate than less severe tumors, whereas, mock-infected mice developed less severe gliomas (WHO grade III). To validate these results, I used an additional syngeneic model of intracranial mouse GBM. Wild type mice were perinatally MCMV- or mock-infected and injected intracranially with mouse GBM cells as adults. MCMV-infected mice with intracranial tumors died significantly sooner than mock-infected mice.
I interrogated changes in signaling in this population via microarray analysis of cultured neural stem cells (NSCs) in MCMV- versus mock-infected Mut3 mice. Integrated Pathway Analysis identified platelet-derived growth factor (PDGF-B) as a central
signaling hub in MCMV-infected neurospheres. In vitro infection of mouse NSCs or tumorspheres and human GBM tumorspheres with CMV validated that PDGF-B, a molecule shown to cause gliomas de novo, was highly upregulated after CMV infection. Additionally, in vitro infection of human GBM neurospheres with HCMV or exogenous PDGF-B activated signal transducer and activator of transcription 3(P-STAT3), a key regulator of gliomas. Analysis of pre-tumorigenic MCMV-infected Mut3 mice showed upregulation of PDGF-B and revealed a significant increase of P-STAT3 in the
subventricular zone, a hypothesized origin of GBMs. Additionally, MCMV-infected mice with intracranial syngeneic GBMs expressed higher P-STAT3 in tumors. Furthermore, I showed that CMV increased proliferation in orthotopic mouse tumors and in human GBM neurospheres via STAT3 activation. Taken together, my data suggest that CMV accelerates GBM progression by increasing proliferation through activation of oncogenic PDGF-B/STAT3 signaling.
During the course of the glioma study, many MCMV-infected mice developed large nonbrain tumors. I hypothesized that CMV could cooperate with germline Trp53 mutation present in the Mut3 mice. To test this premise, mice with a heterozygous Trp53 mutation
(Trp53+/-) were infected with MCMV 2 days (P2) or 4 weeks after birth. P2 MCMV-infected Trp53+/- mice developed tumors at a high frequency (43%) by 9 months of age. In contrast, less than 3% of mock-infected or mice infected at 4 weeks developed tumors by 9 months. A majority of non-brain tumors from P2 MCMV-infected mice were pleomorphic rhabdomyosarcomas (RMS) that harbored MCMV DNA, RNA, and protein. Of clinical relevance, I found that human RMS (embryonal, alveolar, and pleomorphic) also harbored human CMV-IE1 and -pp65 protein and CMV RNA. These data suggest that MCMV contributes to development of mouse pleomorphic RMS in the context of TP53 mutations. Given the high frequency of TP53 mutations in human RMS, my data suggest that HCMV infection also contributes to human RMS.