Date Published: August 6, 2018
Publisher: Public Library of Science
Author(s): Nathan Ungerleider, Monica Concha, Zhen Lin, Claire Roberts, Xia Wang, Subing Cao, Melody Baddoo, Walter N. Moss, Yi Yu, Michael Seddon, Terri Lehman, Scott Tibbetts, Rolf Renne, Yan Dong, Erik K. Flemington, Bryan R. Cullen.
Our appreciation for the extent of Epstein Barr virus (EBV) transcriptome complexity continues to grow through findings of EBV encoded microRNAs, new long non-coding RNAs as well as the more recent discovery of over a hundred new polyadenylated lytic transcripts. Here we report an additional layer to the EBV transcriptome through the identification of a repertoire of latent and lytic viral circular RNAs. Utilizing RNase R-sequencing with cell models representing latency types I, II, and III, we identified EBV encoded circular RNAs expressed from the latency Cp promoter involving backsplicing from the W1 and W2 exons to the C1 exon, from the EBNA BamHI U fragment exon, and from the latency long non-coding RPMS1 locus. In addition, we identified circular RNAs expressed during reactivation including backsplicing from exon 8 to exon 2 of the LMP2 gene and a highly expressed circular RNA derived from intra-exonic backsplicing within the BHLF1 gene. While expression of most of these circular RNAs was found to depend on the EBV transcriptional program utilized and the transcription levels of the associated loci, expression of LMP2 exon 8 to exon 2 circular RNA was found to be cell model specific. Altogether we identified over 30 unique EBV circRNAs candidates and we validated and determined the structural features, expression profiles and nuclear/cytoplasmic distributions of several predominant and notable viral circRNAs. Further, we show that two of the EBV circular RNAs derived from the RPMS1 locus are detected in EBV positive clinical stomach cancer specimens. This study increases the known EBV latency and lytic transcriptome repertoires to include viral circular RNAs and it provides an essential foundation and resource for investigations into the functions and roles of this new class of EBV transcripts in EBV biology and diseases.
Epstein Barr virus (EBV) is a human oncogenic gamma herpesvirus that is carried by greater than 90% of the world’s population. While infection with EBV is generally asymptomatic, the virus persists for the lifetime of the host through an intricate and dynamic interplay with the host immune system, achieved in part through the virus’ utilization of multiple distinct gene expression programs. Initial infection through salivary exchange results in infection of the oral epithelium where the full repertoire of viral “lytic” genes is expressed to facilitate local amplification of virus titers. The virus is then transmitted to circulating naïve B-cells where a “latency type III” viral gene expression program is utilized (Latent Membrane Proteins (LMPs) -1 and -2, Epstein Barr Nuclear Antigens (EBNAs) -1, -2, -3A, -3B, -3C, and -LP and the non-coding transcripts, EBER1, EBER2, v-sisRNAs, v-snoRNA, RPMS1, and a set of viral miRNAs encoded within the introns of RPMS1) [1–4]. The expression of this full repertoire of latency genes facilitates potent B-cell activation and proliferation, a unique mechanism to expand the infected B-cell population in the host (i.e. independently of new virus production and de novo infection). Once an adaptive immune response is mounted to the antigenic latency proteins, the virus adapts through transitioning to (and/or selection for) more restricted latency gene expression programs, type II latency (EBER1, EBER2, RPMS1, viral miRNAs, EBNA1 and LMP1 and LMP2), type I latency (EBER1, EBER2, RPMS1, viral miRNAs and EBNA1) or type 0 latency (only non-coding RNAs, EBER1, EBER2, RPMS1, and viral miRNAs) [1, 2] where the virus persists at low levels with little detriment to the host. In the context of systemic immune suppression (e.g. HIV infection or clinically induced), variable expression of growth promoting type III latency genes can be better tolerated, which often leads to the development of B-cell lymphomas. In immune-competent individuals the viral utilization of non-coding RNAs in addition to low level type I or type II protein coding latency gene expression, perhaps tolerated through local tumor-immune suppression, provides one or more “hits” towards oncogenic progression with minimal impact from the immune system.
Despite more than 50 years of study since the discovery of EBV, our appreciation for the extent and diversity of the EBV transcriptome has grown substantially over the past 15 years with findings of hundreds of new viral RNAs including EBV encoded microRNAs [42, 43], a viral (v)-snoRNA , stable intronic sequence (sis)RNAs  and scores of previously unknown polyadenylated and non-polyadenylated lytic transcripts [25, 27, 28, 34, 44]. Here we report that EBV expresses a repertoire of yet another class of RNAs, circRNAs, in both latency and reactivation. Many of the EBV encoded circRNAs are expressed at levels that are comparable to or higher than the majority of cellular encoded circRNAs (Fig 2B), supporting the contention of potential functional relevance. Some EBV circRNAs are expressed broadly across latency types (e.g. circRPMS1_E4_E3a, circRPMS1_E4_E2, and circEBNA_U) suggesting roles in fundamental processes during latent infection. Further, circRPMS1_E4_E3a and circRPMS1_E4_E2 were found to be expressed in two of two EBV positive stomach cancer biopsies tested, supporting in vivo relevance and possible roles in supporting the tumor phenotype. Though expressed during latency, the expression of circEBNA_W1_C1 and circEBNA_W2_C1 is restricted to type III latency and could be involved in type III latency specific processes such as facilitating Cp initiated transcript diversity, Cp promoter regulation, or other as yet unappreciated type III latency functions. circBHLF1 is detected in most latency cell models but displays extraordinarily high expression under reactivation conditions (S3 Fig). The linear form of BHLF1 is expressed at low levels during latency through the activity of an alternative latency promoter [45, 46]. Whether the observed expression of circBHLF1 in latency cell lines is due to transcription from the latency promoter and/or whether its expression derives from a small percentage of spontaneously reactivating cells is unclear at this time. Nevertheless, its high expression during reactivation, its proximity to OriLyt, and the known association of BHLF1 RNAs with OriLyt supports speculation that it could play a role in reactivation such as facilitating lytic DNA replication. Expression of circLMP2_E8_E2 displayed tissue specificity, being detected in both reactivation models tested but not in the type III latency cell lines, Jijoye and JY (with low levels detected in IB4). This data supports the involvement of cis and trans mechanisms that facilitate the regulation of LMP2 back-splicing and they support a potential unique role for circLMP2_E8_E2 in reactivation. Together, the findings reported here reveal a spectrum of EBV circRNAs with diverse expression profiles and likely unique roles in latent and lytic infection.