Changes in microRNA expression profiles in HIV-1-transfected human cells

1MicroRNAs (miRNAs) are small RNAs of 18–25 nucleotides (nt) in length that are involved in the regulation of a variety of biological processes including developmental timing, signal transduction, apoptosis, cell proliferation and tumorigenesis. Recent studies indicate that cellular miRNAs can variably inhibit or promote viral replication. Viruses, on the other hand, seem to have developed strategies which include virus-encoded RNAi suppressors and/or virus-encoded miRNAs. Mechanistically, a current view is that miRNAs function to silence gene expression through imperfect base-pairing with cognate transcripts. Since RNA silencing mediated by miRNA does not require perfect sequence complementarity, one miRNA can target multiply different mRNAs. It is conceivable that viruses may seek to alter cellular miRNA expression in ways that benefit viral replication. Extant findings support such a notion since several viruses have been found to encode RNAi suppressors which could function to influence the cell's overall miRNA milieu.

2For HIV-1, it has been proposed, based on in vitro assays, that Tat can partially repress the processing activity of Dicer. Because Dicer is involved in the maturation of cellular miRNAs, we wondered how miRNA profiles in human cells that express HIV-1 proteins might differ from counterpart cells that do not express viral genes. To ask if HIV-1 alters the expression of host miRNAs, we employed a high throughput microarray approach to quantify changes in miRNA expression. We used a platform based on the RNA-primed Array-based Klenow Enzyme (RAKE) assay. RAKE originally described by Nelson and colleagues is a microarray assay which uses on-slide enzymatic reactions and primer extension. We printed specific DNA oligonucleotide probes which contain three distinct elements onto a microarray glass slide (Fig 1A). The three different elements include a 5' linker containing a constant nucleotide sequence with amine-modified 5'end for effective slide conjugation; a 3' anti-miRNA element of variable sequence which is complementary to specific miRNA; and a poly-thymidine region which allows for primer extension and labeling of hybridized miRNAs (Fig 1B). It is important to note that RAKE does not employ a sample amplification step; and the enzymes (Klenow and exonuclease I) used in this assay work in an unbiased, substrate sequence-independent way. Thus, RAKE-signals faithfully reflect the true amount of miRNAs in the samples being tested. This contrasts with some conventional microarray methods which use RNA ligase to add linkers on both ends of transcripts for subsequent sample amplification. The enzyme kinetics of RNA ligase varies depending on substrate sequences; thus, amplified samples may inaccurately represent that in the original starting population. Moreover, complete sequence complementarity of the 3'end of miRNA with the DNA oligonucleotide probe used in RAKE is absolutely required for the primer extension step. Since many mature miRNAs differ from their precursor forms and their paralogs in the 3'end sequence, this property offers a specificity advantage to RAKE over several other microarray methodologies.

3To validate and optimize our RAKE analysis, we first printed, based on the published miRNA literature, a small number of DNA probes on glass slides. Our initial sampling set was designed to distinguish between miRNAs reported to be expression-specific for Jurkat versus HeLa cells (Fig 2Aa). We wanted to verify that if we hybridized our slides with miRNAs isolated from HeLa cells, then only HeLa-specific signals would appear in our RAKE assay. Similarly, we wanted to validate the converse for Jurkat miRNAs. When we performed the assays, we indeed replicated the expected cell-specific miRNA expression patterns, with a single exception for hsa-miR-142-3p. Hsa-miR-142-3p was reported by others to be expressed in Jurkat cells, but was not detected by us in those cells (Fig 2Ab). It is unclear why hsa-mirR-142-3p was not detected in our assay, but a trivial explanation might be because there are many different lines of Jurkat cells used in various laboratories. We note that our routinely included "spike-in" oligo (ath-miR-157a), used as a control for the success of the enzymatic reaction, behaved reproducibly from experiment to experiment. We also chose a subset of polymorphic miRNA (hsa-let-7 family) in order to verify the specificity of hybridization detected by our RAKE. Using small RNAs isolated from Jurkat cells for hybridization, RAKE was able to distinguish a single nucleotide difference (hsa-let-7a from hsa-let-7c and hsa-let-7f; hsa-let-7c from hsa-let-7b), suggesting the conditions used by us are highly stringent (Fig 2B).

4The sensitivity of RAKE was evaluated by hybridizing microarray slides with varying amounts of ath-miR-157a. As shown in Fig 2C, RAKE provided robust signals when challenged with as low as 10-7 M of substrate, and offered linear readouts in log2 scale for substrates in the 10-8 to 10-6 M range. We defined our signal as the median of foreground spot fluorescence at 532 nm wavelength minus background (defined by surrounding pixel intensity); negative values were reset as zero.

5After optimization of conditions in initial small scale tests, we next printed microarray slides which contained 312 individual probes based on published sequences of all-known mature human miRNAs at time of slide production. We separately hybridized individual slides with small RNAs (20 μg per slide) isolated from mock-transfected HeLa or HeLa cells transfected with infectious HIV-1 molecular clone, pNL4-3 (see Fig 3A for actual examples of typical results). The results from cell plot analysis of repeated hybridizations indicated that large numbers of miRNAs in pNL4-3-transfected HeLa cells, when compared to mock-transfected HeLa cells, were significantly downregulated (Fig 3B). Clear differences were revealed in comparisons of mock-transfected HeLa cells to pNL4-3-transfected HeLa cells using scatterplot analysis (Fig 4). Although many miRNAs were reduced in expression in the HeLa-pNL4-3 sample (e.g. ~43% of all of the miRNAs were more than two-fold downregulated), the majority of miRNAs remained unchanged, suggesting that the observed results are not due to non-specific generalized cellular toxicity. Interestingly, in our assays, miRNAs upregulated by transfected pNL4-3 were exceedingly rare. Pending further understanding of mechanisms, it is conceivable that the downregulation of mature miRNAs as detected by our RAKE assay may be due to the Dicer-suppressive effect exerted by HIV-1 Tat protein and/or TAR RNA.

6To confirm our RAKE assays, we tested selected results using real time PCR as described by Shi and Chiang. Using these assays, we checked the RAKE results in HeLa cells for four HIV-1 downregulated miRNAs (miR-93, miR-148b, miR-221 and miR-16) (Fig 5B, C, D and 5E). We used two normalization controls, a miRNA (miR-526c) whose expression was found empirically to be reproducibly unchanged in our assays, and a miRNA-unrelated small cellular RNA, the small nuclear U6 RNA (Fig 5A). Real time PCR results confirmed the findings from RAKE.

7In conclusion, we describe here a rapid assay that monitors reproducible changes in cells transfected with HIV-1 infectious molecular clone, pNL4-3. We find that a dominant pattern of response in HeLa cells to pNL4-3 transfection is the downregulated expression of many miRNAs. Studies are ongoing to examine changes in miRNA expression patterns in human cells (primary and T cell lines) after infection with HIV-1.

8The author(s) declare that they have no competing interests.