Suppression of HIV-1 replication by microRNA effectors

The rate of HIV-1 gene expression is a key step that determines the kinetics of virus spread and AIDS progression. Viral entry and gene expression were described to be the key determinants for cell permissiveness to HIV. Recent reports highlighted the involvement of miRNA in regulating HIV-1 replication post-transcriptionally. In this study we explored the role of cellular factors required for miRNA-mediated mRNA translational inhibition in regulating HIV-1 gene expression. Here we show that HIV-1 mRNAs associate and co-localize with components of the RNA Induced Silencing Complex (RISC), and we characterize some of the proteins required for miRNA-mediated silencing (miRNA effectors). RCK/p54, GW182, LSm-1 and XRN1 negatively regulate HIV-1 gene expression by preventing viral mRNA association with polysomes. Interestingly, knockdown of RCK/p54 or DGCR8 resulted in virus reactivation in PBMCs isolated from HIV infected patients treated with suppressive HAART.


Background
RNA silencing (RNAi) is a new gene regulatory mechanism conserved from plants to humans. RNAi mediators are small non-coding RNAs (sncRNAs) that function through sequence specific mRNA targeting to either induce their degradation and/or inhibit translation [1,2]. In mammals, RNAi is mediated by different classes of small non-coding RNAs including piRNAs, microRNAs and siRNAs [3][4][5]. MicroRNAs are produced from a primary transcript (pri-miRNA) which is processed in the nucleus by the microprocessor complex containing RNase Drosha and DGCR8. The resulting product or pre-miRNA is exported to the cytoplasm through the exportin-5 pathway. Cytoplasmic pre-miRNA is processed by typeIII RNase Dicer to miRNA/miRNA* duplex of 19 to 25 nucleotides. miRNA/miRNA* is incorporated into the RNA-Induced Silencing Complex (RISC) where miRNA* is degraded while miRNA serves as a guide for mRNA targeting [2]. Key components of miRISC are proteins of the Argonaute family (Ago1 to Ago4) that are required for miRNA-mediated silencing [6,7]. To ensure mRNA translational inhibition and decay, miRISC, loaded with miRNA and its mRNA targets, associate with proteins involved in mRNA processing [2]. A key factor in this process is the GW182 protein that interacts directly with Argonaute1 (Ago1) [8], and the human homologs of GW182 that interact with Ago1-4 [9]. GW182 orchestrates both mRNA decapping, through the recruitment of p54/RCK that regulates the activity of the decapping enzymes DCP1/DCP2 [10], and mRNA deadenylation by recruiting the CCR4-NOT1 complex [11]. mRNA decapping and deadenylation leads to mRNA decay through the action of XRN1, a 5'-3' exonuclease [10]. Interestingly, RNAi effectors, including miRNAs and their target mRNAs, Ago proteins, GW182, RCK/p54, LSm-1 and DCP proteins co-localize in cytoplasmic structures called GWbodies or P-bodies suggesting that miRNA-mediated silencing occurs at these sites [11][12][13][14][15]. Emerging evidence suggests that miRNA-mediated gene regulation serves as a defence mechanism against both RNA and DNA viruses in mammals [16][17][18][19][20]. The present study was designed to explore physical and functional interaction between effectors of miRNA-mediated silencing and HIV-1 replication.

Results and discussion
To investigate whether RNAi effectors regulate HIV-1 replication, we analyzed virus replication in cells where expression of RNAi effectors was reduced using specific siRNA. HeLa cells were transfected with siRNA specific to RCK/p54, GW182, LSm-1 or XRN1. As controls, HeLa cells were transfected with scrambled siRNA (Scr) or CDK9 specific siRNA and subsequently infected with HIV-1 ( Figure 1a). Knockdown of RCK/p54, GW182, LSm-1 and XRN1 enhanced virus replication by up to 10 fold ( Figure 1b). As we have previously shown, knockdown of Drosha [21] and DGCR8 (Figure 1b), the two subunits of the microprocessor complex, increased virus production while knockdown of the CDK9 subunit of the PTEFb complex that is required for viral gene expression, reduced HIV-1 production ( Figure 1b). Interestingly, analysis of HIV-1 cytoplasmic mRNA distribution on glycerol gradient showed that knockdown of RCK/p54 shifted HIV-1 mRNA from the non-polysomal fraction to polysomes as compared to control siRNA transfected cells (Figure 2, upper panel). As control, we analyzed the distribution of endogenous mRNA expressed from a gene encoding Hdm2. Knockdown of RCK/p54 did not affect Hdm2 mRNA distribution (Figure 2, lower panel). These experiments show that GW182, RCK/p54, LSm-1 and XRN1, factors required for RNAi, are repressors of HIV-1 gene expression that act by preventing HIV-1 mRNA translation.
We next investigated the physical interaction between RNAi effectors and HIV-1 mRNA. 293 cells were mock transfected or transfected with combinations of pNL4-3, Myc-Ago2, a central component of the RISC complex, or its RNA-binding mutant Myc-Ago2PAZ9 constructs as indicated in figures 3 and 4. First, we verified that Myc-Ago2 and Myc-Ago2PAZ9 were equally expressed ( Figure  3a). Second, cytoplasmic extracts were prepared, and a fraction was used for total RNA extraction while the rest was subjected to immunoprecipitation using anti-Myc antibody to purify Myc-Ago2 associated mRNP. Both total RNA (Figure 3b, left panels) and Myc-Ago2 associated RNA (Figure 3b, right panels) were reverse transcribed and subjected to PCR amplification using oligonucleotides specific for HIV-1 TAR RNA (a structured motif associated with all HIV-1 mRNAs) or HIV-1 unspliced mRNA, Hdm2 mRNA or GAPDH mRNA. PCR analysis of total RNA showed that equal amounts of HIV-1, Hdm2 and GAPDH mRNAs were present in all samples (Figure 3b, left panels). HIV-1 mRNAs (both TAR and unspliced) were associated with Myc-Ago2, but not with Myc-Ago2PAZ9 mutant (Figure 3b, right panels). In agreement with the results shown in figure 2, Hdm2 mRNA was not detected in Myc-Ago2 mRNPs, suggesting that under these conditions Hdm2 is not regulated by RNAi. A similar experiment was performed to analyze the association of HIV-1 multispliced mRNA with Myc-Ago2 mRNPs. The RT-PCR reactions were performed in the presence of 32P-α ATP and were analyzed by autoradiography (Figure 3c). HIV-1 multispliced mRNAs associated with Myc-Ago2 (compare lane 3 to 2) and weakly with Myc-Ago2PAZ9 (compare lane 4 to lanes 3 and 2). Co-localization of HIV-1 mRNA and effectors of RNAi such as Ago2 and RCK/p54 within the P-bodies was also observed by immunofluorescence using HIV-1 containing MS2 binding sites and MS2-GFP constructs ( Figure 4). Indeed, HIV-1 mRNAs visualized through their binding to MS2-GFP colocalized with endogenous RCK/p54 and ectopically expressed Myc-Ago2 ( Figure 4). Our results show that HIV-1 mRNAs physically associate with Ago2, a central component of RISC, and co-localize with cellular proteins required for miRNA-mediated silencing such as RCK/p54 and Ago2 in P-bodies. We observed that all HIV-1 mRNA species associated with RISC. Accordingly, Huang et al. had identified 5 cellular miRNAs able to target the 3'UTR sequence present in all HIV-1 mRNAs [22]. Additionally, other cellular miRNAs able to target regions out side the 3'UTR may also participate [23].
Emerging evidence suggests the physical and functional interactions between P-bodies and the viral life cycles [24]. Viral mRNA trafficking through P-bodies may repre-sent a pool of translationally repressed viral transcripts otherwise used for efficient packaging or formation of viral-replication complexes. Indeed, yeast retrotransposons Ty1 and Ty3 mRNA associate with P-bodies, and this association is required for efficient retrotransposition [25][26][27]. In the case of BMV (Brome Mosaic Virus), formation of the virus replication complex occurs in P-bodies [28]. In addition, P-bodies may also function in host defences against viruses and transposable elements. Indeed, the cellular factors APOBEC 3G (A3G) and 3F (A3F), which are viral restriction factors, are found to accumulate in P-bodies [29,30]. It has been suggested that A3G and A3F mediated HIV-1 restriction may involve viral mRNA targeting to P-bodies leading to their translational inhibition [30]. We, therefore, asked whether Pbodies are positive or negative regulators of HIV-1 replication. Thus, we analyzed HIV-1 replication in cells where Pbodies were disrupted by knocking down RCK/p54 or LSm-1 [31]. HeLa CD4+ cells were transfected with RCK/ p54 or LSm-1 specific siRNA or control siRNA. Forty eight miRNA effectors are repressors of HIV-1 replication Taken together, our results show a physically repressive interaction between RNAi effectors and HIV-1 mRNA.
Since cellular miRNAs were shown to play a role in HIV-1 latency [22], we asked whether RCK/p54, which is HIV-1 mRNAs associate with Argonaute 2 required for miRNA-mediated mRNA translational inhibition, contributes to HIV-1 silencing in vivo. Thus, PBMCs isolated from 3 HAART-treated HIV-1-infected patients with undetectable viremia were transfected with control siRNA or with siRNA specific for Drosha, DGCR8 or RCK/p54. Transfected cells were co-cultured with PHA/ IL2-activated PBMCs isolated from healthy donors. Virus production was monitored every 3 days by measuring p24 antigen in the culture supernatant ( Figure 7). As we have previously shown, knockdown of Drosha resulted in virus reactivation in PBMCs isolated from 3 HAART-treated HIV-1-infected patients [21]. Remarkably, viral replication from its natural reservoir resumed also when DGCR8 or RCK/p54 was silenced. No virus was isolated from control siRNA transfected PBMCs suggesting that virus production observed in Drosha, DGCR8 and RCK/p54 knock down was not due to actively infected PBMCs relieved from drug pressure. These results show that endogenous levels of Drosha, DGCR8 and RCK/p54 contribute to HIV-1 latency and/or its maintenance in infected patients.

Conclusion
The outcome of HIV-1 infection results from complex interactions between viral components and host cell factors [32][33][34][35]. In most cases, HIV-1 successfully hijacks cellular pathways and bypasses restriction factors for optimal replication leading to continuous rounds of infection, replication, and cell death. Continuous viral replication causes the loss of CD4+T cells and progression to immunodeficiency in infected individuals. HAART treatment revealed the existence of a pool of resting memory CD4+ T cells harbouring integrated, but silent HIV-1 provirus [36,37]. This latent reservoir is believed to be the major obstacle for virus eradication by HAART. Therefore, it is critical to understand how HIV-1 latency is established and maintained [38]. Post-integration latency takes place HIV-1 mRNA co-localizes with RCK/p54 and Ago2 at both transcriptional and post-transcriptional levels [39]. Transcriptional latency involves different mechanisms ranging from integration position effect [40][41][42], limitation in transcription factors [43][44][45][46], establishment of chromatin repressive marks and recruitment of chromatin silencers [47][48][49][50][51]. Post-transcriptional silencing involves defects in mRNA export and translation [52][53][54].
All together, these studies show that HIV-1 post-integration latency is a multi-factorial process. In the present study, we show that HIV-1 gene expression is additionally regulated by the miRNA pathway. HIV-1 mRNA associates with components of the RISC complex by a mechanism that does not involve APOBEC3G, but does need sncRNAs. Accordingly, it has been recently shown that the suppressor of RNAi P19 from tomato bushy stunt virus, known to bind and sequester sncRNAs including miRNA, enhances HIV-1 replication [55]. Additionally, the RNAi suppressor function of HIV-1 Tat [56] could be complemented by VP35 from Ebola virus [57] and the NS3 protein of rice hoja blanca virus through sequesteration of small non-coding RNAs [58]. HIV-1 mRNAs associated with RISC are sequestered in the non-polysomal fraction, thereby preventing translation. In agreement with two previous reports [19,21,22], we show that knockdown of RCK/p54, a protein required for miRNA-mediated silencing, led to virus reactivation from PBMCs isolated from HIV-1 infected patients who were undergoing suppressive HAART.
A challenge in AIDS treatment is the need to activate latent viral reservoirs in order to eradicate these viruses through HAART. In this respect, targeting the miRNA processing pathway could offer a strategy that could be exploited to activate latent viral reservoirs, for instance, during HAART. Several molecules have been used to reactivate viral reservoirs [59]. However, none of these approaches provides the sequence specific targeting that can be achieved using siRNA. Recent data suggest that siRNA can be used therapeutically in vivo in certain mouse disease models [60] and more recently in non-human primates [61,62]. It remains to be explored whether, as suggested here, the in vivo targeting of miRNA-effectors using siRNA can assist in activating latent HIV-1 reservoirs for eradication by HAART.

Transfections
PBMCs were transfected with siRNA or miRNA using the Nucleofector II Device with the appropriate Nucleofection Disruption of P-bodies through knockdown of RCK/p54 and LSm-1 lead to enhanced production of infectious HIV-1 virions

PBMC isolation and co-culture assay for virus production
Peripheral blood mononuclear cells of HIV-1 infected patients were isolated by lymphocyte separation medium density centrifugation (Lonza). PBMCs from healthy donors were pre-activated using 5 μg/ml PHA (phytohemagglutinin-P, DIFCO)/10 U/ml IL-2 (interleukin-2, Roche) for 72 hours. They were then washed once with PBS and once with RPMI medium before co-culture assay. siRNA transfected HIV-infected PBMC (10 6 cells/ml) were co-cultured with pre-activated PBMC (10 6 cells/ml) from the same healthy donor in the presence of 10 U/ml IL-2. The culture medium was collected every 3 or 4 days. Fresh pre-activated healthy PBMCs were added to the culture every 7 days. Viral production was measured by quantifying the amounts of p24 in the culture medium using an ELISA kit (Ingen).

Pseudotyped virion production and single-round infections
The plasmid pNL4-3-env -Luc + harboring a luciferase gene (obtained from the NIAID AIDS Reagent Program) was co-transfected with the envelope plasmid pMD.2G encod-

Cytoplasmic extracts analysis on sucrose gradients
To isolate cytoplasmic extracts, cells were lysed for 10 minutes in buffer B (5 mM Tris-HCl pH 7.4, 1.5 mM KCl, 2.5 mM MgCl 2 , 0.5% NP40 and protease inhibitor). Nuclei were pelleted by centrifugation for 10 minutes at 10,000 rpm. 2 mg of cytoplasmic extracts were loaded on a 7-47% sucrose gradient. Briefly, 5 layers of 7 to 47% sucrose were prepared in sucrose buffer (20 mM Tris-HCl pH7.4, 80 mM NaCl, 5 mM MgCl 2 , 1 mM DTT and protease inhibitors) and diffused at 4°C for 16 hours to obtain a linear sucrose gradient. 2 mg of cytoplasmic extracts were loaded on the top of the column, and centrifuged for 3 hours at 36,000 rpm in a SW41Ti rotor. After ultracentrifugation, 28 fractions were collected and OD at 254 nm was measured in each fraction using a Nanodrop apparatus (Labtech).

Competing interests
The authors declare that they have no competing interest.