In-depth analysis of Kaposi's sarcoma-associated herpesvirus microRNA expression provides insights into the mammalian microRNA-processing machinery

Abstract

We have used deep sequencing to analyze the pattern of viral microRNA (miRNA) expression observed in the B-cell line BC-3, which is latently infected with Kaposi's sarcoma-associated herpesvirus (KSHV). We recovered 14.6 x 10(6) total miRNA cDNA reads, of which a remarkable 92% were of KSHV origin. We detected 11 KSHV miRNAs as well as all 11 predicted miRNA or passenger strands from the miRNA duplex intermediate. One previously reported KSHV miRNA, miR-K9, was found to be mutationally inactivated. This analysis revealed that the 5' ends of 10 of the 11 KSHV miRNAs were essentially invariant, with significantly more variation being observed at the 3' end, a result which is consistent with the proposal that the 5'-proximal region of miRNAs is critical for target mRNA recognition. However, one KSHV miRNA, miR-K10-3p, was detected in two isoforms differing by 1 nucleotide (nt) at the 5' end that were present at comparable levels, and these two related KSHV miRNAs are therefore likely to target at least partially distinct mRNA populations. Finally, we also report the first detection of miRNA offset RNAs (moRs) in vertebrate somatic cells. moRs, which derive from primary miRNA (pri-miRNA) sequences that immediately flank the mature miRNA and miRNA strands, were identified flanking one or both sides of nine of the KSHV miRNAs. These data provide new insights into the pattern of miRNA processing in mammalian cells and indicate that this process is highly conserved during animal evolution.

FIG. 1.

Frequency of KSHV miRNA and miRNA* strands recovered from deep sequencing of short RNAs derived from uninduced BC-3 cells. Gray bars represent 5p miRNAs recovered, black bars represent 3p miRNAs recovered, and the white bar indicates edited miRNAs. The percent contribution of each set of miRNAs to the total number of viral miRNAs detected is indicated.

FIG. 2.

KSHV gene expression in uninduced and induced BC-3 and BC-1 cells. All values are expressed as fold expression over an uninfected control sample and are normalized to cellular β-actin mRNA levels. (A) Rta and Mta mRNA expression in uninduced (light gray) and TPA-induced (white) BC-3 cells and uninduced (black) and TPA-induced (dark gray) BC-1 cells. (B) Expression levels of KSHV miRNAs in uninduced (gray) and induced (black) BC-3 cells. (C) Same as panel B except using RNA derived from uninduced and TPA-induced BC-1 cells.

FIG. 3.

Comparison of the BC-1 and BC-3 miR-K9 genomic loci. (A) Alignment of the KSHV miR-K9 sequences found in BC-1 and BC-3 cells. Conserved nucleotides are indicated with asterisks, and the 5p and 3p strands of miR-K9 are boxed in gray. (B) Predicted secondary structures of the miR-K9 pri-miRNA stem-loops transcribed from the KSHV genomes present in BC-1 and BC-3 cells. In the BC-1 hairpin, the previously reported dominant miRNA is indicated in bold, Drosha cleavage sites are indicated by black arrowheads, and Dicer cleavage sites are indicated by gray arrowheads. In the defective BC-3 stem-loop structure, the mutated “miR-K9” is indicated in bold.

FIG. 4.

Sequence variation in KSHV miRNAs recovered from deep sequencing of BC-3 cells with 5′ ends indicated in gray and 3′ ends indicated in black. In each case, the end of the most abundant form of each miRNA is designated nucleotide position 0 and all other, less abundant isoforms are numbered accordingly. (A) Collective sequence variation of the mature miRNA strands recovered. (B) Same as panel A except with analysis of the less abundant miRNA* strands. (C) Collective sequence variation of all 5p miRNA strands recovered from deep sequencing. (D) Same as panel C except with analysis of the 3p miRNA strands. (E) Sequence variation detected at the 5′ and 3′ ends of mature miR-K10-3p.

FIG. 5.

Predicted structures of the KSHV miR-K12, miR-K4, and miR-K8 RC pri-miRNA stem-loops. The most common miRNA recovered is indicated in bold. Drosha cleavage sites are indicated by black triangles, and Dicer cleavage sites are indicated by dark gray triangles. The predominant termini observed for the moRs are indicated by light gray triangles. Boxed numbers indicate the frequency with which each processing fragment was recovered by deep sequencing.