Comprehensive analysis of Rhesus lymphocryptovirus microRNA expression

Abstract

Rhesus lymphocryptovirus (rLCV) and Epstein-Barr virus (EBV) are closely related gammaherpesviruses that infect and cause disease in rhesus monkeys and humans, respectively. Thus, rLCV is an important model system for EBV pathogenesis. Both rLCV and EBV express microRNAs (miRNAs), several conserved in sequence and genomic location. We have applied deep sequencing technology to obtain an inventory of rLCV miRNA expression in latently rLCV-infected monkey B cells. Our data confirm the presence of all previously identified mature rLCV miRNAs and have resulted in the discovery of 21 new mature miRNAs arising from previously identified precursor miRNAs (pre-miRNAs), as well as two novel pre-miRNAs (rL1-34 and rL1-35) that together generate four new mature miRNAs. Thus, the total number of rLCV-encoded pre-miRNAs is 35 and the total number of rLCV mature miRNAs is 68, the most of any virus examined. The exact 5' and 3' ends of all mature rLCV miRNAs were pinpointed, many showing marked sequence and length heterogeneity that could modulate function. We further demonstrate that rLCV mature miRNAs associate with Argonaute proteins in rLCV-infected B cells.

FIG. 1.

Summary of deep sequencing results. (A) Proportions of host and viral high-quality, filtered sequencing reads detected in latently infected 309-98 cells. (B) Summary of total read numbers, including all 75-nt reads obtained (total reads) and reads with perfect 3′ linker sequences (linkers mapped; no errors). Some of these high-quality reads mapped to the rhesus monkey genome (Macaca mulatta aligned), and others mapped to the rLCV genome (rLCV aligned). The rLCV reads included miRNA-associated sequences, fragments of the EBER noncoding RNAs, and other random sequences (rLCV misc. fragments). (C) All high-quality rLCV miRNA sequencing reads (above) are ordered relative to the rLCV genome (below). The genome is drawn to scale according to reference , with key latent (white) and lytic (black) genes. Sequence reads mapping to intervening loci not associated with known, predicted, or novel miRNAs are labeled “betw,” and the BART and BHRF1 miRNA clusters are indicated above the map.

FIG. 2.

Northern blotting validation of novel rLCV miRNAs derived from known pre-miRNAs. Expression of low-abundance novel miRNAs was validated by probing total RNA purified from uninfected BJAB B cells (lane 1) and rLCV-infected 211-98 and 309-98 cells (lanes 2 and 3, respectively) with DNA oligonucleotides antisense to the sequence with the highest number of reads for each miRNA listed. Rhesus monkey (M. mulatta) miRNA miR-16 is shown as a loading control.

FIG. 3.

Novel rLCV pre-miRNAs and the corresponding mature miRNAs. (A) Predicted structures for novel rLCV pre-miRNAs are minimal free-energy folds from the mfold algorithm (34, 66). Mature miRNAs representing the most frequently sequenced isoform are highlighted. (B) Northern blot analysis of novel mature miRNAs as in Fig. 2. (C) Sequence conservation of the major sequenced isoforms of rL1-2-5p and rL1-2-3p miRNAs with those from the similar EBV miRNAs, BHRF1-2-5p and BHRF1-2-3p, respectively. Nonidentical nucleotides are in bold.

FIG. 4.

Specific coimmunoprecipitation of rLCV miRNAs with Argonaute proteins. (A) Lysed 309-98 cells were subjected to coimmunoprecipitation with anti-Argonaute antibody 2A8 or negative controls (normal mouse serum [MS] and mouse monoclonal anti-HA IgG [IgG]). RNAs isolated from equal fractions of the input lysate (IN, lane 1), the cleared supernatant (S, lanes 2, 4, and 6), and immunoprecipitates (IP, lanes 3, 5, and 7) were probed with DNA oligonucleotides antisense to the miRNA indicated (see Table S1 in the supplemental material or miRBase). Precursor miRNAs do not efficiently coimmunoprecipitate (pre-rL1-2 is an example). U6 was provided a loading control. Length estimates are based on a 10-bp DNA ladder (right). (B) Coimmunoprecipitation, as in panel A, of the highly abundant novel mature rLCV miRNAs.