Although most of the studies on the expression of KSHV miRNAs have been conducted in a limited number of PEL cell lines, some teams have employed different models of study ranging from other cultured cell lines (e.g. immortalised endothelial cell lines, numerous diverse PEL cell lines) to clinical samples (e.g. peripheral-blood mononuclear cells, saliva, or biopsy samples from patients with classic KS, AIDS-KS, or MCD), as well as animal models.
Viral miRNAs have been shown to be expressed both during the latent and lytic stages, and can be detected in infected tissues or in the lesions associated with virus pathology (Hansen et al., 2010; Marshall et al., 2007; O’Hara et al., 2009; Umbach and Cullen, 2010).
Some variability was reported in pre-miRNA expression levels across different PEL cell lines and KS biopsies, but the greater differences were observed when comparing different models of infection, with patient-derived samples and cell lines showing a greater expression of the viral pre-miRNAs expression compared to in vitro infection models (O’Hara et al., 2009).
Sequence analysis of KSHV pri-miRNA regions, pre-miRNAs and miRNAs from 5 PEL cell lines and 17 diverse types of clinical samples revealed 2 distinct clusters of the intronic pri- miRNA sequences (a major and a variant cluster), but a high degree of conservation of the
disease states and geographical locations, thus suggesting that KSHV miRNA genes are under tight selection in vivo and suggest that they have an important role for the virus cycle, and associated pathogenesis. Nonetheless, some variability that could affect processing was observed in some of the pre-miRNAs sequences (Marshall et al., 2007). This correlates with the expression variability observed in the study from O’Hara et al., and with some observation reported in previous studies. Indeed, it was previously shown by Gottwein et al.
that the BCBL-1 PEL cell line presented a very low expression of miR-K12-5 compared to another PEL cell line (BC-1). This is caused by a single nt polymorphism in the pre-miR- K12-5 expressed in BCBL-1 cells, which therefore importantly alters its processing (Gottwein et al., 2006). These results were also confirmed by the Sullivan laboratory in a deep sequencing analysis, in which they observed that pre-miR-K12-5 was the fourth most abundant KSHV pre-miRNA that was cloned, but that its processed mature form was one of the lowest expressed KSHV miRNAs (Lin et al., 2010). Pre-miR-K12-9 also shows a high degree of variability, in particular in the PEL cell lines and was even hyper-mutated in one of them (BCP-1), suggesting that this miRNA is not essential for maintenance of the virus in cell culture (Marshall et al., 2007). A lack of miR-K12-9 expression has later on been observed in another PEL cell line (BC-3) through in depth analysis of KSHV miRNAs expression, and sequence analysis confirmed extensive mutations of the pre-miRNA (Umbach and Cullen, 2010).
To date, there is no evidence of a specific promoter for the cluster of KSHV miRNAs, which would allow the expression of KSHV miRNAs independently from an intron or from the Kaposin transcript. The transcription of the pri-miRNA seems to be driven by the same promoters that also drive the expression of the transcript encoding Kaposin, vFLIP, vCyc. The expression of the pri-miRNA is currently reported to be always associated to a transcript that also contains at least the Kaposin CDS (Cai and Cullen, 2006; Pearce et al., 2005).
Nonetheless, studies that have compared viral miRNA expression during latent and lytic phases have shown a significant increase in pre-miR-K12-10 and -12 expression during lytic infection (Lin et al., 2010; Samols et al., 2005); this is due to an RTA-inducible unspliced Kaposin transcript expressed as a DE lytic gene (Ganem and Ziegelbauer, 2008). The biological relevance of such a higher expression of these two miRNAs is yet unknown.
Indeed, although KSHV miRNAs are also expressed during lytic infection, it is currently unclear if they could play a role in the replicative cycle, when compared to the robust shutoff of host gene expression induced by viral lytic proteins such as the KSHV shutoff and exonuclease protein (SOX - ORF73) (Glaunsinger and Ganem, 2004) (discussed in section
6.2 of the introduction), and it has even been suggested that a long ncRNA expressed antisense to the latency-associated region during the lytic phase could act as a miRNA sponge to inhibit miRNA accumulation or function during lytic infection (Chandriani et al., 2010).
KSHV miRNAs are abundantly expressed, as they can represent from 30% to about 70% of all miRNAs in a cell (Gottwein et al., 2011; Lin et al., 2010). As suggested by Pfeffer and Voinnet, such high expression levels of viral miRNAs could saturate the processing machinery, leading to a global downregulation in the expression of cellular miRNAs, and thus to a perturbation of cellular homeostasis, which could be linked to oncogenesis (discussed in section 6.1 of the introduction). However, there seems to be enough room to accommodate a certain number of extra miRNA precursors, and that with some exceptions (e.g. adenovirus VA RNA), cellular miRNA levels do not seem to be globally affected by the expression of virus-encoded miRNAs.
Finally, even if most of the KSHV miRNAs are expressed from the same transcripts, their relative abundances are dramatically different (Gottwein et al., 2011; Umbach and Cullen, 2010). Deep sequencing analysis of the BC-3 PEL cell line reported miR-K12-4-3p and miR- K12-3 as the two most abundant KSHV miRNAs, with respectively 78% and 12% of the KSHV miRNA sequence reads obtained, whereas the nine other miRNAs (miR-12-9 being not expressed in BC-3) ranged from 2.4% to 0.19% of the total number of sequence reads (Umbach and Cullen, 2010). However, this study did not look specifically at the miRNAs ioncorporated into RISC, but the overall miRNA population present in the cells, and therefore these results cannot be definitely stated as representative of KSHV miRNAs activity. Indeed Gottwein et al., in their PAR-CLiP study, observed a different distribution of the viral miRNAs loaded into Ago2. In BC-3 cells, miR-K12-4-3p was still the most represented, with this time 33% of all reads pertaining to KSHV miRNAs, and was followed by miR-K12-6-3p with 20% of the viral miRNA total (3.5% in Lin et al. study). On the other hand, miR-K12-3 was surprisingly the lowest in the number of sequence reads (0.17%) when compared to the other non-cited above non-star species, which accounted for up to 6.2% of all KSHV miRNA sequence reads (Gottwein et al., 2011). However, it is noteworthy that the number of reads for a given sequence (e.g. miRNA) obtained by deep sequencing is not always representative of the real absolute quantities of the sequences analysed. Indeed, it was recently demonstrated that some cloning bias could exist in the small RNA libraries generated, thus altering quantitation (Sorefan et al., 2012). Nevertheless, these results still remain good indicators of the discrepancies in the levels of expression of KSHV miRNAs, as when considering the star
These different expression levels might be caused by distinct efficiencies of pre-miRNA processing, due for example to their more or less optimal hairpin structure, accessibility for the microprocessor complex within the cluster, secondary structures, or by additional cellular or viral factors involved in the processing or even in the stability of the mature miRNAs. It would be thus tempting to speculate that the different levels of expression of KSHV miRNAs could reflect their evolutionary apparition within the cluster as discussed in section 2.2.2, and therefore that they would be indicators of their functional relevance. However, other factors have to be taken in to account for such considerations, such as the fact that viral miRNAs are not under the same selection pressure compared to their cellular counterparts (see section 6.2.3), in addition to the fact that the different types of functional viral miRNAs (i.e.
analogues of cellular miRNAs, or viral-specific miRNAs) are subjected to different evolutionary constraints. Therefore, to be able to establish a link between such discrepancies in the levels of expression of viral miRNAs and their biological functionality, a better comprehension of their targeting functions is required.
2. Function of KSHV miRNAs and their potential involvement in viral oncogenesis