Kaposi's sarcoma-associated herpesvirus (KSHV) is the responsible agent for Kaposi's sarcoma (KS), primary effusion lymphoma and multicentric Castleman's disease. KSHV expresses multiple microRNAs that modulate human gene expression. Most microRNAs (miRNAs) repress target gene expression by destabilizing the mRNA transcript and decreasing translational efficiency. A goal of the project is to determine targets of viral miRNAs and understand why the virus has selected specific human target genes for inhibition. We hope to discover new functions of human genes as they relate to viral infection and cancer. Using a variety of expression profiling data, we constructed a dataset to integrate the expression data from multiple gain and loss of microRNA function experiments. We have tested over fifty predicted target genes and over thirty microRNA target genes were significantly inhibited by viral miRNAs using a variety of validation methods. A subset of these target genes has been further validated by looking at protein expression of endogenous target genes in response to viral microRNA expression, microRNA inhibition in infected cells and KSHV infection. To assess changes in protein expression, we utilize a near-infrared scanner to perform simultaneous two-color quantitative western blotting assays. In addition, we have mapped functional microRNA target sites in multiple human genes using site-directed mutagenesis. Furthermore, using KS biopsies from patient enrolled in clinical trials, we have determined multiple microRNA target genes that are inhibited in our cell culture systems are also inhibited at sites of KSHV infection in patients. We are currently investigating the functional roles of a couple of selected miRNA targets identified in these approaches. Using a combination of mRNA expression profiling and proteomic approaches with gain and loss of viral miRNA activity, we have previously reported the identification of numerous mRNA targets of KSHV-encoded miRNAs. Using these previous expression datasets, we integrated the data using a rank sum method in order to select genes with the highest expression changes in response to gain or loss of viral miRNA function. Importantly, in this analysis, we ignored miRNA seed-matching information and ranked potential miRNA targets solely by their expression changes in response to individual KSHV miRNAs and KSHV infection. We then cloned 49 different full-length 3'UTRs of these predicted miRNA target mRNAs downstream of the renilla luciferase gene into a reporter plasmid and performed reporter assays in cells in the presence of various KSHV miRNAs. Luciferase activity was normalized to control assays that used parental luciferase reporters (lacking inserted human 3'UTR sequences) and non-targeting negative control miRNA mimics. These data demonstrated that 28 of 49 3'UTRs were repressed by at least one viral miRNA, when using a p-value cutoff of 0.05. However, this is likely to be an underestimate of miRNA targets, since we did not test all combinations of all individual viral miRNAs and 3'UTRs. There were also several instances of a single 3'UTR being repressed by multiple KSHV miRNAs, indicating a high level of redundancy in KSHV miRNAs mediated targeting of cellular mRNAs. There were at least 50 instances of a miRNA repressing a 3'UTR, but the majority (28 of 50) of these target interactions were missed using a common bioinformatic tool for predicting miRNA targets. This suggested that some miRNA:target repression events may function through non-canonical sites, as has been demonstrated for human miRNAs . Using miR target prediction software with modified parameters, we were able to predict the existence of non-canonical miRNA binding sites in the 3'UTRs of many mRNAs. Indeed, integration with PAR-CLIP data identified Ago2-associated sequences within the 3'UTRs of mRNAs that do not contain perfect seed-matching sites to these miRNAs. These binding sites had 5mer/ 6mer miRNA binding regions, deviations from the 2-7 seed-matching region, and several G:U interactions spread across the length of the miRNA. To validate these non-canonical miRNA binding sites, we cloned these potential sites into luciferase reporters and repeated the 3'UTR luciferase reporter assays in the presence of KSHV miRNAs. We were able to identify functional miRNA binding sites in the 3'UTRs of multiple genes. Interestingly, luciferase activity of the non-canonical site identified in one 3'UTR, where the binding was mediated by nucleotides 3-9 of a KSHV miRNA, was repressed by nearly 75% compared to the control miRNA, suggesting strong interactions at this site. These non-canonical miRNA binding sites demonstrate the existence of functional sites that can be missed by scanning only for 7mer/8mer interactions in the 3'UTRs. Furthermore, a single 3'UTR can be targeted at multiple sites by multiple KSHV miRNAs, using both canonical and non-canonical interactions, demonstrating a high level of complexity in their regulation. These validated targets may be of interest in future studies of the functional roles of viral miRNAs. It is noteworthy to state approximately half of these microRNA:target interactions are not detected using common bioinformatic methods. This method of integrating data from expression screening, CLIP studies, and modified miRNA target prediction programs could be directly applied to studies of other viral and cellular miRNAs. Additionally, we are using network analysis to understand how newly identified microRNA target genes could be interacting with each other, based on published literature. This analysis has revealed multiple examples of how microRNA target genes are in the same signaling pathways. These examples highlight significant pathways targeted by KSHV miRNAs and focused our efforts on pathways with multiple miRNA targets. Some of our research has led to a discovery of using a combination of drugs that activates the viral lytic cycle, but suppresses productive lytic replication. This combination could potentially be used to target latent herpesvirus infections. After infection in primary endothelial cells with KSHV, growth arrest and DNA damage inducible gene 45 (GADD45) is one of the most repressed genes using genomic expression profiling. Endogenous GADD45 protein expression was also repressed when multiple KSHV miRNAs were introduced to uninfected cells. We hypothesized that KSHV miRNAs may target human GADD45 to protect cells from DNA damage consequences, which may be triggered upon viral infection. Expression of GADD45 protein is increased by the p53 activator, Nutlin-3, and a KSHV miRNA inhibits this induction. Using various flow cytometry assays, Nutlin-3 increased apoptosis and cell cycle arrest in primary endothelial cells. However, a specific KSHV miRNA protected primary endothelial cells from apoptosis and cell cycle arrest following Nutlin-3 treatment. Targeting GADD45 with siRNAs showed similar protective phenotypes as with the KSHV miRNA. The KSHV miRNA also repressed protein levels of cleaved standard apoptosis markers after Nutlin-3 treatment. In B lymphocytes latently infected with KSHV, delivery of specific inhibitors of the KSHV miRNA increased GADD45 expression and apoptosis, indicating that miRNA is important for reducing apoptosis in infected cells. Furthermore, ectopic expression of GADD45 in infected cells promoted apoptosis. Together, these results identify a new human target gene of KSHV miRNAs and demonstrate that KSHV miRNAs are important for protecting infected cells from DNA damage responses.
Showing the most recent 10 out of 13 publications
