Silencing Dicer and Drosha enhances HIV-1 replication
We hypothesized that HIV-1 infection alters miRNA expression in host cells, and specific sets of upregulated miRNAs could modulate HIV-1 replication. Since Dicer and Drosha are the key components of the miRNA biogenesis machinery (Chu and Rana, 2007
), depleting these two enzymes lowers production of mature miRNAs. To directly determine the role of host miRNAs in HIV-1 production and infectivity, we depleted Dicer in 293T cells to downregulate production of mature miRNA (Kim, 2005
) and analyzed HIV-1 production and infectivity. Dicer expression was efficiently silenced by siRNA in cells used to produce HIV-1 (producer cells; , top panel) and in infected (target) cells (, top panel). Quantitative analysis of HIV-1 production showed that depleting Dicer in producer cells increased HIV-1 production (, bottom panel). When HIV-1 was produced in 293T cells without Dicer depletion, depleting Dicer in target cells enhanced viral infectivity (, bottom panel). Similarly, depleting Drosha in producer and target cells upregulated HIV-1 production and infectivity, respectively (). Although these results cannot rule out the possibility that knocking down Dicer and Drosha can influence HIV-1 replication by an indirect mechanism (Triboulet et al., 2007
), they strongly suggest that miRNAs are involved in negatively controlling HIV-1 replication.
Depleting Dicer or Drosha increases virus production and infectivity
miR-29a represses HIV-1 replication
Which miRNAs could be involved in repressing HIV-1 mRNA expression? Target prediction analysis suggested that the HIV-1 3’-UTR can be targeted by 11 miRNAs (). To examine miRNA expression after HIV-1 infection, we infected H9 T lymphocytes with HIV-1 and allowed the infections to proceed for 5 passages. Small RNAs were isolated from HIV-1-infected H9 T lymphocytes, and miRNA expression was quantified by miRNA microarray (Table S1
). Of the 11 miRNAs predicted to target HIV-1 3’-UTR (), miR-29a was most highly expressed in H9 T lymphocytes. HIV-1 infection of 293T cells also enhanced miR-29a expression (). Given the complementary sequences of miR-29a and HIV-1 3’-UTR (), we reasoned that miR-29a could repress HIV-1 gene expression. To test this possibility, we inhibited miR-29a function using 2’-O
-methyl oligonucleotides complementary to miR-29a (anti-miR) and evaluated HIV-1 production and infectivity (see Experimental Procedures). Inhibiting miR-29a in 293T cells by anti-miR before viral infection enhanced HIV-1 production over that in cells treated with control 2’-O
-methyl oligonucleotides (). Similarly, blocking miR-29a in 293T cells with anti-miR oligos increased HIV-1 infectivity (). To confirm the role of miR-29a in suppressing HIV-1 mRNA expression, we performed a reciprocal experiment using a miR-29a mimic and analyzed its effect on virus production. Our results show that mimicking miR-29a efficiently suppressed HIV-1 production (). Finally, we tested the effect of inhibiting miR-29 on HIV-1 infectivity by first transfecting H9 T lymphocytes with anti-miR and then by infecting with HIV-1. We found that HIV-1 infectivity was greater than that in lymphocytes treated with control 2’-O
-methyl oligonucleotides (). HIV-1 infectivity of H9 T cells was less affected by anti-miR than that of 293T cells (), likely due to either lower transfection efficiency of the miR inhibitor or to higher expression of miR-29a in H9 T cells. Taken together, these results identify miR-29a as a cellular repressor of HIV-1 mRNA expression.
Predicted targets of human miRNAs in the HIV 3’-UTRa
miR-29a modulates HIV-1 production and infectivity
miR-29a directly targets HIV-1 mRNA
Does miR-29a directly and specifically mediate repression of HIV-1 mRNA expression? To address these questions, we created a mutant HIV-1 reporter construct, pNL4-3-Luc-E-m29t, containing 4 mutations in the HIV-1 3’-UTR region targeted by miR-29a (). To match this target site in the mutant HIV-1 mRNA, we designed a mutant miRNA, miR-29a-mt (). The mutant HIV-1 vector, pNL4-3-Luc-E-m29t, was then transfected into 293T cells in the presence of 10 nM miR-29a or miR-29a-mt, and HIV-1 production was analyzed as described above. Viral production from the pNL4-3-Luc-E-m29t mRNA was suppressed by miR-29a-mt, but was not affected by miR-29a (). As a reciprocal experiment, wild-type pNL4-3-Luc-E- was transfected into 293T cells in the presence of 10 nM miR-29a or miR-29a-mt, and HIV-1 production was analyzed. Viral production from this wild-type vector mRNA was suppressed by miR-29a, but not by miR-29a-mt ().
miR-29a specifically targets HIV-1 mRNA for translation repression
To further analyze the specific effects of miR-29a on HIV-1 mRNA expression, we made another mutant HIV-1 reporter construct, pNL4-3-Luc-E-Δ29t, containing a 20-nt deletion at the miR-29a target site in the HIV-1 3’-UTR region. This mutant construct or a control HIV-1 construct, pNL4-3-Luc-E, was transfected into 293T cells in the presence of 10 nM miR-29a or an miR control, and HIV-1 production was analyzed. Similar to the results in , viral production was suppressed by miR-29a only when its complete target site was in the HIV-1 sequence, but not when the target site contained a deletion mutation (data not shown).
We next determined the effect of these miR-29a target-site mutations on viral fitness. First, we analyzed the viral production from various mutant viral constructs. We transfected 293 T cells with 10 nM miR-29a or control, then co-transfected with pNL4-3LucR−E− (wt), the HIV-1 construct lacking the miR-29a target site pNL4-3LucR−E−(Δ29t), or pNL4-3LucR−E−(m29t) and vesicular stomatitis virus glycoprotein (VSVG). At 48 h post-transfection with the miR-29a mutant viral constructs, viral production in cell supernatants (see Experimental Procedures) increased and was insensitive to repression by miR-29a (). Second, we determined the effect of miR-29 target-site mutations on viral infectivity. Target cells were transfected with 10 nM miR-29a or control and infected 18 h later with the virus equivalent of 10 ng pNL4-3LucR−E−(wt), pNL4-3LucR−E−(Δ29t), or pNL4-3LucR−E−(m29t). At 72 h post-infection, viral infectivity was analyzed as described above. As shown in , mutant viruses, pNL4-3LucR−E−(Δ29t) and pNL4-3LucR−E−(m29t), were more infectious than the pNL4-3LucR −E− (wt). Third, we investigated the infectivity of mutant and wild-type viruses in T lymphocytes. H9 T lymphocytes were infected by spinoculation with a virus equivalent of 0.3 µg pNL4-3LucR-E- virus. At 72 h post-infection, luciferase activity was measured in total cell lysates relative to their protein content and virus infectivity data were analyzed (). We found that mutant virus pNL4-3LucR−E−(m29t) was significantly more infectious in T lymphocytes than pNL4-3LucR−E− (wt). Collectively, these results show that miR-29a target-site mutations enhance viral fitness and lead to loss of viral sensitivity to repression by miR-29a.
Our observation that mutant virus pNL4-3LucR−
(m29t) was significantly more infectious than pNL4-3LucR−
(wt) in T lymphocytes () suggests that miR-29a, which is highly expressed in T lymphocytes (Table S1
), would suppress wild-type virus more potently than the mutant viruses. These results, taken with those shown in , indicate that miR-29a specifically targets the predicted site in the HIV-1 3’-UTR () and directly mediates miRNA suppression of HIV-1 mRNA expression.
As an alternative approach to determining the specificity of the interaction between miR-29a and its target, we analyzed the ability of miR-29a inhibitors to relieve HIV-1 mRNA repression as described above (). Inhibiting miR-29a de-repressed only HIV-1 expression of the wild-type pNL4-3-Luc-E construct, but not of the pNL4-3-Luc-E-m29t or pNL4-3-Luc-E-Δ29t reporters (). As an additional control, we determined the effect of miR-29a on HIV-1 production in Dicer-depleted cells (). Dicer depletion enhanced HIV-1 production as shown in , and miR-29a reversed this effect by suppressing HIV-1 production to that in normal cells (). Together, these results indicate that miR-29a directly targets a specific region in the HIV-1 3’-UTR.
miR-29a specifically enhances HIV-1 mRNA interactions with RISC
Since miR-29a directly targets HIV-1 mRNA (), we next addressed whether HIV-1 mRNA interacts with RISC where specificity is dictated by miR-29a interactions with viral mRNA. To that end, we analyzed the effect of wt and mutant miR-29a on the association of wt and mutant HIV-1 mRNA with the immunopurified Myc-tagged RISC component, Ago2. Cells were transfected sequentially with miR-29a or miR-mt29a and with pNL4-3LucR−
(wt) or pNL4-3LucR−
m29t, VSVG, and Myc-Ago2. Cell extracts were immunopurified for Ago2 (, top panel). miRNA expression was monitored (bottom panel) to confirm the presence of corresponding miRs in cells. The RISC and P-body protein-binding mRNA, connexin 43 (Beitzinger et al., 2007
), was used as a control. We found that in the presence of miR-29a, wild-type HIV-1 gag mRNA associated with the Ago2 immunoprecipitate, but HIV-1 mRNA containing miR-29a target-site mutations did not efficiently associate with immunopurified Ago2. In a reciprocal experiment, mutant miR-29a enhanced the association of mutant HIV-1 with Ago2 immunoprecipitates (, middle panel). Quantification of gag mRNA associated with Ago2 revealed that miRNAs specifically enhanced ~3-fold accumulation of the target mRNAs with Ago2 (, bar graph).
miR-29a directly targets HIV-1 mRNA for its association with RISC and P-body RNP assemblies
To determine the effect of miR-29a on HIV-1 gag mRNA association with endogenous RISC proteins, we transfected 293T cells with miR-29a or miR-mt29a followed by pNL4-3LucR−
(wt) or pNL4-3LucR−
m29t, and VSVG. Cell extracts were immunopurified for the P-body protein, RCK/p54 (, top panel). miRNA expression in cells was monitored (bottom panel) to confirm the presence of corresponding miRs. The RISC and P-body protein-binding mRNA, connexin 43 (Beitzinger et al., 2007
), was used as a control. We found that in the presence of miR-29a, wild-type HIV-1 gag mRNA eluted with immunopurified RCK/p54, and HIV-1 mRNA containing miR-29a target-site mutations did not efficiently associate with immunopurified RCK/p54. In a reciprocal experiment, mutant miR-29a enhanced the association of mutant HIV-1 with RCK/p54 immunoprecipitates. Quantification of gag mRNA associated with RCK/p54 showed that miRNAs specifically increased ~3-fold association of the target mRNAs with RCK/p54 (, bar graph). The specificity of these interactions was verified by two experiments: (a) RCK/p54 immunoprecipitation experiments were repeated with wt-HIV-1 in the absence of miRNA, (b) HA antibodies were used to immunoprecipitate RNA in the presence of miR-29a. Our results showed that miR-29a specifically enhanced association of HIV-1 mRNA with endogenous P-body proteins, and the HA-immunopurified complexes showed no interaction between HIV-1 mRNA and miR-29a (Figure S1
). Altogether, these results indicate that miR-29a specifically targets HIV-1 mRNA and enhances its association with ribonucleoprotein (RNP) complexes containing RISC proteins.
HIV-1 mRNA interacts with P-body proteins, and depleting P-bodies enhances HIV-1 replication
One mechanism proposed for miRNA-mediated gene silencing suggests that the miRNA in RISC provides the sequence specificity for target mRNA interactions, and RISC effector proteins shuttle the target mRNA toward the fate of storage or processing in P-bodies (Chu and Rana, 2006
; Rana, 2007
). We found that miR-29a specifically enhances HIV-1 mRNA interactions with RISC (). Therefore, we reasoned that these HIV-1 mRNA-RISC interactions lead to the accumulation of viral mRNA at P-bodies for translational suppression.
We asked whether HIV-1 mRNA assembles into large RNP complexes and associates with P-body proteins. To address this question, we first immunopurified a P-body marker from HIV-1-infected cells and analyzed the resulting RNP complex for HIV-1 mRNA. As a P-body marker, we used a DNA-editing enzyme, APOBEC3G (Sheehy et al., 2002
), which has been shown to form RNP complexes with P-body proteins that have established roles in cap-dependent translation (eIF4E and eIF4E-T), translation suppression (RCK/p54), RNAi-mediated post-transcriptional gene silencing (Ago2), and de-capping of mRNA (DCP2) (Wichroski et al., 2006
). In immunoprecipitation experiments with HIV-1-infected cells, we used Vif-deleted HIV-1 vectors because HIV-1 Vif protein degrades APOBEC3G (Sheehy et al., 2002
). HIV-1 gag mRNA was found only in HA-APOBEC3G immunoprecipitates from cells expressing both HA-APOBEC3G and HIV-1 (Figure S2A
). In similar experiments, we found that HIV-1 gag mRNA immunopurified with other P-body proteins such as Ago2 and endogenous RCK/p54 ( and Figure S2B–D
). No HIV-1 gag mRNA bands were observed in the absence of APOBEC3G or Ago2. Control experiments showed that HIV-1 mRNA did not co-purify with HA- or PTEN-HA (HA-tagged phosphatase and tensin homolog deleted on chromosome 10) proteins that did not localize to P-bodies (Figure S2E
). These results indicate that HIV-1 mRNA interacts with RNP complexes containing P-body proteins.
To determine the functional significance of HIV-1 mRNA association with P-bodies, we disrupted P-body structures in host cells by depleting various P-body proteins, infected cells with HIV-1, and analyzed HIV-1 production and infectivity. Three P-body proteins, RCK/p54, Lsm1, and Ago2, were depleted in 293T cells by siRNA-mediated RNAi in the presence of APOBEC3G. Specific knockdown was confirmed 24 h after siRNA transfection by immunoblot analysis showing that targeted protein levels decreased significantly without affecting the expression of other proteins (Figure S3
). Next, cell structure and function were monitored after disrupting P-bodies, as described (Chu and Rana, 2006
). To that end, we first depleted P-body proteins by transfecting cells with siRNAs and examined the localization of YFP-tagged APOBEC3G. As shown in , depleting RCK/p54 disrupted cellular P-body structures. In these P-body-deficient cells, APOBEC3G proteins were diffused throughout the cytoplasm, no longer accumulating at specific foci (, panels d-f). Similarly, depleting Lsm1 and Ago2 resulted in a diffuse cytoplasmic distribution of APOBEC3G (, panels d-f). Together, these results show that knockdown of RCK/p54, Lsm1, or Ago2 disrupts P-body structures and disperses APOBEC3G throughout the cytoplasm.
Depleting P-body proteins disrupts P-body formation and increases HIV-1 production and infectivity
To examine the effect of P-body disruption on HIV-1 production, we depleted 293T cells of RCK/p54, Lsm1, or Ago2 before exposure to HIV-1 and VSVG constructs. This treatment enhanced HIV-1 production over that in cells pretreated with mismatched siRNA (si-control, ). RCK/p54 knockdown increased viral production more than Ago2 or Lsm1 knockdown because RCK/p54 is a general translational suppressor. Therefore, its knockdown could have a dual or additive effect on viral production, i.e., release translational suppression plus disrupt P-bodies. Next, we produced HIV-1 in 293T cells without first depleting P-body proteins and infected target 293T cells in which these three P-body proteins had been depleted. In single-round infections, depleting RCK/p54, Lsm1, or Ago2 increased viral infectivity (). We observed that Lsm1 knockdown increased viral infectivity less than it increased viral production. These results show that depleting RCK/p54, Lsm1, and Ago2 disrupts P-body structures and enhances HIV-1 production and infectivity.
To visualize HIV-1 mRNA in P-bodies, we analyzed co-localization of a highly abundant miRNA in 293T cells, miR-18a, with various P-body proteins and HIV-1 mRNA. 293T cells were transfected with APOBEC3G-YFP (APO-YFP) vectors, fixed at 24 h post-transfection, immunostained for APO-YPF or RCK/p54, hybridized in situ for miR-18a with fluorescein-labeled locked nucleic acid (LNA) probe, and analyzed by confocal microscopy. APO-YFP and endogenous RCK/p54 were detected by immunofluorescence with anti-YFP and anti-RCK/p54, respectively (panels b, e). Cells were stained to visualize nuclei and images were digitally merged (panels c, f). As expected, miRNA-18a localized at P-bodies with APOBEC3G or RCK/p54 (arrowheads, ). To detect HIV-1 mRNA with APOBEC3G, 293T cells were transfected with APO-YFP and pNL4-3 constructs, fixed at 36 h post-transfection, immunostained, and hybridized in situ for HIV-1 mRNA as described above. LNA probes complementary to HIV-1 Nef mRNA were labeled with Cy3 and used for FISH. We found that HIV-1 mRNA co-localized with APO-YFP (). These results show that HIV-1 mRNA can be detected in P-bodies with endogenous miRNAs.
RCK/p54 depletion increases viral protein synthesis without affecting its mRNA levels
Based on the above findings, we hypothesized that the repression of viral protein translation is modulated by miRNAs that enhance the association between viral mRNAs and P-bodies. To test our hypothesis, we analyzed expression of HIV-1 protein and GFP (as a control) by quantifying p24 levels in producer cells depleted of RCK/p54 by RNAi (Figure S4A
). For control experiments, cells were treated with a mismatched siRNA duplex. Total cellular protein was extracted from equal numbers of cells and measured by ELISA or GFP fluorescence (Figure S4B and C
). GFP protein levels were measured as a control. HIV-1 protein expression increased more robustly ~ 3-fold in RCK/p54-depleted cells than in control cells treated with mismatched siRNA (Figure S4B
). HIV-1 mRNA levels did not differ in producer cells after treatment with RCK/p54 or control siRNA (Figure S4D
). Since RCK/p54 is a general translational repressor, its depletion led to a moderate increase in GFP protein with no significant change in GFP mRNA levels (Figure S4C and E
). Remarkably, RCK/p54 depletion released ~3-fold translation suppression by viral mRNA (Figure S4
), and miR-29 enhanced ~3-fold target viral mRNA association with P-bodies (). Taken together, these results strongly suggest that miR-29a facilitates suppression of HIV-1 translation by enhancing the association of viral mRNA with P-bodies and that disrupting P-bodies by RCK/p54 depletion releases translation suppression and enhances production of viral proteins.
A highly conserved miR-29 family modulates HIV-1 production
miR-29a is a member of an miRNA family whose mature miRNAs contain the same seed region, which is highly conserved across humans, mice, rats, dogs, and chickens (http://www.targetscan.org/
). Besides miR-29a, miR-29b and miR-29c were also expressed in H9 T cells although at much lower levels (). To determine whether these 3 miR family members all modulate HIV-1 mRNA expression, we analyzed the effects of miR-29a, b, and c inhibitors on viral production. Inhibiting miR-29a, b, or c increased HIV-1 production, but inhibiting miR-29a resulted in the highest viral production (). Similarly, using miRNA mimics showed that all 3 miR family members decreased viral production, with 29-a again showing the strongest effect (). These findings strongly suggest that that miR-29 family members influence virus replication. Further, our analysis of the miR-29a target site in various HIV-1 subtypes revealed that the sequence targeted by the seed region of miR 29a family is highly conserved (; see Discussion).
A highly conserved miR-29 family modulates HIV-1 production and infectivity