Sequence Conservation Among γ-Herpesviruses

To date, more than 200 viral miRNAs have been identified. At the beginning of this work, the miRNA registry release v13.0 listed 146 viral miRNAs. These are mainly encoded by polyomaviruses and herpesviruses. Polyomaviruses encode for only one miRNA, whereas members of the herpesvirus family have been shown to encode for mutiple miRNAs.

Since miRNAs are abundantly expressed by γ-herpesviruses compared to other DNA viruses it was suggested that the ability to encode miRNAs is a conserved feature of herpesviruses. Furthermore, a conservation of miRNA sequences might translate into conserved functions, by regulating the same sets of genes. Nevertheless, only two viruses have been shown to encode for conserved miRNAs, the γ-herpesviruses Epstein-Barr virus (EBV) and the related rhesus lymphocryptovirus (rLCV). These had been previously reported to encode 23 and 16 miRNAs, respectively, seven of them being partially conserved (Cai et al., 2006; Grundhoff et al., 2006; Pfeffer et al., 2005b). It was suggested that there are more examples of conservation. Therefore, a detailed analysis of miRNAs encoded by the γ-herpesvirus family was performed.

The aim of this work was to get a better insight into γ-herpesvirus miRNA conservation and function.

Assuming that miRNAs are functionally important, their sequences should be more conserved than adjacent regions. The closer two viruses are related the more background conservation will occur and thus discrimination between false positives and true positives will increase with evolutionary distance.

However, pre-miRNAs that are predicted within coding regions might either be conserved due to protein conservation or due to their own importance. Additionally, viral miRNAs are primarily encoded within intergenic and non-coding regions. Thus, to facilitate the analysis and to minimize the number of false positives, only non-coding regions were analyzed.

First, pre-miRNA hairpins were predicted in all fully sequenced γ-herpesvirus genomes using the VMir program. VMir is a previously designed algorithm for the ab initio prediction of pre-miRNA hairpins and uses the RNAFold algorithm for prediction (Hofacker and Stadler, 2006). Four γ-herpesviruses belonging to the lymphocryptovirus genus and ten from the rhadinovirus genus were analyzed (see table 4-1).

Table 4-1 γ-Herpesviruses Analyzed in This Work

Genus Name Abbreviation Genbank

Callithricine herpesvirus-1 CaHV-3 NC_004367 Epstein-Barr virus type I EBV type I NC_007605 Epstein-Barr virus type II EBV type II NC_009334 lymphocryptovirus

Rhesus lymphocryptovirus rLCV NC_006146

Equidherpesvirus-2 EHV-2 NC_001650

Ovine herpesvirus-2 OvHV-2 NC_007646 Alcelaphine herpesvirus-1 AHV-1 NC_002531 Bovine herpesvirus-4 BoHV-4 NC_002665 Saimiriine herpesvirus-2 HVS NC_001350 Kaposi’s sarcoma associated herpesvirus type M KSHV type M NC_003409 Kaposi’s sarcoma associated herpesvirus type P KSHV type K NC_009333 Rhesus rhadinovirus RRV NC_003401 Japanese monkey herpesvirus JMHV NC_007016 rhadinovirus

Murine herpesvirus-68 MHV-68 NC_001826

Several potential pre-miRNA hairpins were predicted for all viruses and passed the filter criteria, which are described in detail in 3.6.1. A total number of 42,356 hairpins were predicted. Out of these, 810 passed the filter criteria. Due to the partially inverted nature of the RNA sequences able to form stem-loop structures, these 810 hairpins included stem-loops that are predicted for both strands of the genome. In general only one strand is transcribed in vivo and produces a bona fide miRNA. The only known exceptions so far are mouse cytomegalovirus (MCMV) and herpes simplex virus type 1 (HSV-1), in which miRNAs are transcribed from bidirectional loci (Buck et al., 2007; Dolken et al., 2007) Umbach 2008). Excluding hairpins from the reverse strand that is not suggested to produce miRNAs in vivo decreases the number to 607 hairpin loci.

The sensitivity of the analysis was demonstrated by considering previously identified miRNAs. In the initial prediction, all 68 known miRNAs were predicted and 64 of them were retained after filtering.

This elimination was performed to circumvent identification of miRNAs conserved, due to the fact that they are contained within overlapping coding regions, which generally show a higher conservation between viruses. For miRNAs encoded within ORFs, it is impossible to distinguish if miRNA conservation is due to conservation of the coding regions or due to the importance of the miRNA itself. On the basis of that, the number of predicted pre-miRNAs was further reduced to 61 miRNAs.

In conclusion 548 novel pre-miRNA hairpin loci were predicted.

To identify conserved pre-miRNAs, pairwise BLAST alignments were performed for all combinations. The resulting sequence identity is shown in figure 4-1 A. A conserved pre-miRNA had to fulfill different requirements. These were:

a) alignment to at least one putative orthologue pre-miRNA hairpin, b) conservation of six consecutive nt in one of the hairpin arms, and

c) a second BLAST alignment of the two orthologues resulted in an expect value ≤ 0.01.

The results of this analysis are depicted in figure 4-1 B. Figure 4-1 highlights the overall low sequence identity of γ-herpesviruses except for high identity between strain variants of a given virus (e.g. KSHV type P and M; EBV type I and II). Only two pairs show a significantly higher sequence identity, rhesus

Results

rhadinovirus (RRV) and Japanese monkey herpesvirus (JMHV), as well as Epstein-Barr virus (EBV) and rhesus lymphocryptovirus (rLCV). They had an overall sequence identity of about 88% and 60%, respectively. Interestingly, rLCV is the next related virus to EBV, infecting rhesus macaques. RRV and JMHV infect also rhesus macaques and are the related viruses to KSHV. So the question arises if these viruses express conserved miRNAs.

As depicted in figure 4-1, a correlation between the overall sequence conservation and the prediction of putative conserved pre-miRNAs was found. In accordance with the low sequence identity, in only two cases (apart from strain variants) the conservation of putative pre-miRNA hairpins was predicted.

Indeed, these were RRV and JMHV and EBV and rLCV.

Figure 4-1 Sequence Identity and Pre-miRNA Hairpin Conservation of γ-Herpesvirus Genomes

All fully sequenced γ-herpesviruses were subjected to pairwise BLAST alignments. The sequence identity is given as percentage and additionally indicated by shading (A). VMir prediction of pre-miRNA hairpins identified several putative pre-miRNAs in all tested genomes (white numbers on black background) and few conserved pre-miRNA hairpin loci (black numbers on grey background) (B) (Walz et al., 2010). (figure:

Copyright © American Society for Microbiology, [J Virol, 84, 716-728, 2010])