MicroRNA-mediated transformation by the Kaposi's sarcoma-associated herpesvirus Kaposin locus

Abstract

The human oncogenic Kaposi's sarcoma-associated herpesvirus (KSHV) expresses a set of ∼20 viral microRNAs (miRNAs). miR-K10a stands out among these miRNAs because its entire stem-loop precursor overlaps the coding sequence for the Kaposin (Kap) A/C proteins. The ectopic expression of KapA has been reported to lead to transformation of rodent fibroblasts. However, these experiments inadvertently also introduced miR-K10a, which raises the question whether the transforming activity of the locus could in fact be due to miR-K10a expression. To answer this question, we have uncoupled miR-K10a and KapA expression. Our experiments revealed that miR-K10a alone transformed cells with an efficiency similar to that when it was coexpressed with KapA. Maintenance of the transformed phenotype was conditional upon continued miR-K10a but not KapA protein expression, consistent with its dependence on miRNA-mediated changes in gene expression. Importantly, miR-K10a taps into an evolutionarily conserved network of miR-142-3p targets, several of which are expressed in 3T3 cells and are also known inhibitors of cellular transformation. In summary, our studies of miR-K10a serve as an example of an unsuspected function of an mRNA whose precursor is embedded within a coding transcript. In addition, our identification of conserved miR-K10a targets that limit transformation will point the way to a better understanding of the role of this miRNA in KSHV-associated tumors.

Importance: Kaposi's sarcoma-associated herpesvirus (KSHV) is a human tumor virus. The viral Kaposin locus has known oncogenic potential, which has previously been attributed to the encoded KapA protein. Here we show that the virally encoded miR-K10a miRNA, whose precursor overlaps the KapA-coding region, may account for the oncogenic properties of this locus. Our data suggest that miR-K10a mimics the cellular miRNA miR-142-3p and thereby represses several known inhibitors of oncogenic transformation. Our work demonstrates that functional properties attributed to a coding region may in fact be carried out by an embedded noncoding element and sheds light on the functions of viral miR-K10a.

FIG 1

The miR-K10a miRNAs overcome contact inhibition in the absence of KapA in NIH/3T3 cells. (A) Schematic of the parental lentiviral vector pTRIPZ and wild-type (WT) or frameshift mutant (FS) pTRIPZ-based KapA/miR-K10a/+1 expression vectors. For KapA/miR-K10a/+1 expression, the tRFP-sh-miR cassette was replaced with the WT or FS KapA ORF. LTR, long terminal repeat; tRFP, turboRFP reporter; TRE, tetracycline-inducible promoter; sh-miR, miR-30-based siRNA expression cassette; rtTA3, reverse tetracycline transactivator 3; UBC, human ubiquitin C promoter; IRES, internal ribosomal entry site; PURO^R puromycin resistance gene; WRE, woodchuck hepatitis virus posttranscriptional regulatory element; SIN-LTR, self-inactivating LTR. (B) Western blot analysis demonstrating the Dox-dependent (+, ON) expression of 3× Flag-tagged KapA in WT- but not FS-transduced cells. (C) Primer extension analysis demonstrating the Dox-dependent (ON) expression of miR-K10a and miR-K10a+1 in both WT- and FS-transduced cells. Expression levels were slightly below those observed in the PEL cell line BC-3. 5S RNA served as an input RNA control. (D) Focus formation assay shows equivalent numbers of Dox-dependent foci in WT- and FS-transduced cells. Foci were scored 6 weeks into the assay. n = 3 independent experiments, with three technical replicates each. Error bars, standard deviations (SD); n.s., not significant.

FIG 2

Continued transgene expression is required to maintain transformed status. (A) Schematic outlining the experimental setup used for secondary focus formation assays. (B) TaqMan qRT-PCR confirms loss of miR-K10a/+1 expression after withdrawal (−) of Dox for 10 days. miR-K10a/+1 expression levels in cells under continued Dox treatment (+) were in the physiological range observed in the PEL cell lines BC-1 and BC-3. (C) Western blot analysis demonstrates loss of KapA expression in WT cells after withdrawal (−) of Dox for 10 days. FS cells did not express KapA, as expected. (D) Loss of transgene expression after withdrawal (−) of Dox for 10 days results in lower cell densities. Cells were counted 10 days into the assay. Statistically significant differences are indicated (*, P < 0.05; **, P < 0.01 [unpaired t test]; error bars, SD; n = 3). (E) Loss of transgene expression results in reduced proliferation under confluent conditions. WT or FS cells were cultured in the presence (+) or absence (−) of Dox for 10 days, plated close to confluence, and, 14 days later, assayed for incorporation of BrdU. (F) Loss of transgene expression after withdrawal (OFF) of Dox for 10 days results in loss of the transformed phenotype. Data were obtained from 3 independent clones, which were representative of >5 clones each.

FIG 3

Inhibition of mature miR-K10a/+1 results in loss of the transformed phenotype in FS-transformed NIH 3T3 cells. (A) Dual-luciferase reporter assays confirm inhibition of miR-K10a/+1 activity by the miR-K10a/+1 sponge. Activities from the mir-K10a/+1-firefly luciferase reporter were normalized to those from an internal RLuc control. Normalized values obtained from cells expressing miR-K10a/+1 were normalized to those from control cells lacking miR-K10a/+1 expression, which were set at 1. (B) Sponge-mediated inhibition of mature miR-K10a/+1 results in loss of the transformed phenotype in NIH 3T3 cells transformed by FS vectors. The results shown are representative of 3 independent experiments. (C) Sponge-mediated inhibition of mature miR-K10a/+1 results in loss of the transformed phenotype in NIH 3T3 cells transformed by WT vectors. The results shown are representative of 3 independent experiments. (D) Western blot analysis showing unexpected reduction of 3× Flag-tagged KapA expression in WT-cells transduced with the miR-K10a/+1-specific sponge.

FIG 4

Inhibition of KapA expression does not affect the transformed phenotype in WT-transformed NIH 3T3 cells. (A) Western blotting confirms reduction of KapA expression (Flag) in cells expressing a KapA-directed sh-miR (shKap). (B) TaqMan qRT-PCR shows that miR-K10a/+1 expression was not affected by shKap. (C) shKap-induced inhibition of KapA does not affect the transformed phenotype. The results shown are representative of 3 independent experiments.

FIG 5

Conservation of PAR-CLIP-identified miRNA binding sites between human and mouse. Sites were scored as conserved if at least a minimal match to the miRNA was present in a two-way alignment between the human and mouse genomes (i.e., base pairing of nt 2 to 8 or of nt 2 to 7 with an A across from nt 1 of the miRNA). White, viral miRNAs that mimic cellular miRNAs; black, cellular miRNAs; gray, viral miRNAs with seed sequences unrelated to cellular miRNAs.

FIG 6

miR-K10a/+1 inhibit p27 and p120 expression in NIH 3T3 and human endothelial cells. (A and B) Sequence alignment showing the previously validated, PAR-CLIP-identified binding sites of miR-K10a/+1 in the CDKN1B and CTNND1 3′UTRs. (C and D) Quantitative Western blot analysis of p27 and p120 expression in clonal FS-derived cell lines cultured for 10 days in the presence or absence of Dox, plated, and measured 14 days into the assay. Differences were statistically significant (P < 0.05 [unpaired t test]; error bars, SD; n = 4 independent experiments). (E and F) Telomerase-immortalized human microvascular endothelial cells (iHMVEC) were transfected with control mimic or equimolar concentrations of miR-K10a/+1 mimics. p27 and p120 expression was analyzed 2 days after transfection. Differences were statistically significant (P < 0.05 [unpaired t test]; error bars, SD; n = 5 independent experiments). (G) PAR-CLIP-identified binding sites for miR-K10a and/or miR-K10a/+1 in the 3′UTRs of TNFRSF12A (encoding TWEAKR) and BCLAF mRNAs. Two-nucleotide seed match mutations introduced in each binding site are indicated above the wt sequence (arrows). (H) miRNA binding sites were tested for regulation in dual luciferase reporter assays using human WT and binding site mutant 3′UTR reporters and either control miRNA or viral miRNA mimics. Data were normalized to an internal RLuc control, control mimic transfected cells, and 3′UTR mutants (error bars, SD; n = 3 independent experiments).