Trimethylguanosine capping selectively promotes expression of Rev-dependent HIV-1 RNAs

Abstract

5′-mRNA capping is an early modification that affects pre-mRNA synthesis/splicing, RNA cytoplasmic transport, and mRNA translation and turnover. In eukaryotes, a 7-methylguanosine (m7G) cap is added to newly transcribed RNA polymerase II (RNAP II) transcripts. A subset of RNAP II-transcribed cellular RNAs, including small nuclear RNA (snRNA), small nucleolar RNA (snoRNA), and telomerase RNA, is further hypermethylated at the exocyclic N2 of the guanosine to create a trimethylguanosine (TMG)-capped RNA. Some of these TMG-capped RNAs are transported within the nucleus and from the nucleus to the cytoplasm by the CRM-1 (required for chromosome region maintenance) protein. CRM-1 is also used to export Rev/RRE-dependent unspliced/ partially spliced HIV-1 RNAs. Here we report that like snRNAs and snoRNAs, some Rev/RRE-dependent HIV-1 RNAs are TMG-capped. The methyltransferase responsible for TMG modification of HIV-1 RNAs is the human PIMT (peroxisome proliferator-activated receptor-interacting protein with methyltransferase) protein. TMG capping of unspliced/partially spliced HIV-1 RNAs represents a new regulatory mechanism for selective expression.

Posttranscriptional processing of RNA plays a critical role in gene expression (1 –3). An early mRNA modification is the formation of a 7-methylguanosine (m7G) cap (4 –6). In mammals, the acquisition of an m7G cap requires two enzymes (7 –9), a capping enzyme (triphosphatase and guanylyltransferase activity) and an RNA-(guanine 7)-methyltransferase. These proteins are essential for cell growth, and mutations in the triphosphatase, guanylyltransferase, or methyltransferase components of the capping apparatus are lethal in vivo (10).

Cellular RNAP II transcripts are mostly m7G-capped (1 –3, 11). An m7G cap facilitates the initiation of translation in mammalian cells, and a failure to cap premRNAs results in their accelerated decay by 5′ exoribonuclease degradation (2, 12 –14). Capping of viral RNAs has been less extensively investigated. It has been generally assumed that like cellular RNAs, many viral RNAs have a 5′ m7G cap, and that viruses either use the cellular capping machinery (e.g., viruses that replicate in the nucleus) or encode their own capping enzymes (e.g., viruses that replicate in the cytoplasm) (15, 16). Interestingly, there is a surprisingly diverse range of cap modifications; for instance, Mimivirus, a large DNA virus of amoeba, encodes an RNA cap guanine-N2 methyltransferase that hypermethylates the viral RNA cap (17); influenza virus “snatches” caps from cellular mRNAs (18, 19); West Nile fever virus has two methyl additions on its RNA cap (20); Sindbis virus produces mRNAs that have dimethylguanosine and trimethylguanosine caps [hypermethylated caps/m^2,2,7G caps (21)]; and Semliki forest virus late mRNAs also have hypermethylated caps (22). On the other hand, poliovirus, encephalomyelitis virus, foot and mouth disease virus, and Calici virus are examples of viral RNAs lacking 5′ caps (11, 23), and there is limited in vitro evidence that HIV-1 RNAs are capped (24, 25).

The peroxisome proliferator-activated receptor-interacting protein with methyltransferase domain (PIMT) is a human homolog of the yeast RNA cap hypermethylase TGS1 (26 –32). In yeast, TGS1 hypermethylates the m7G RNA cap to a trimethylguanosine (TMG; m^2,2,7 guanosine) cap (28, 33 –37). A nuclear isoform of human PIMT has been shown to have methyl transferase activity (30 –32). Human PIMT is an 853-aa protein that has a 9-aa methyltransferase motif I (VVDAFCGVG) and an invariant segment (GXXGXXI) found in the K-homology motifs of many RNA-binding proteins (26, 27). PIMT is ubiquitously expressed in all human tissues and binds 5-adenosyl-L-methionine (a methyl donor) and RNA (26, 27).

The exact functions served by monomethylated, dimethylated, and trimethylated caps are incompletely understood. Here we report that unspliced/partially spliced HIV-1 RRE-containing RNAs have a PIMT-mediated TMG cap, and that TMG capping promotes the selective expression of these viral RNAs.

Results

PIMT Modulates HIV-1 Gene Expression.

We originally characterized PIMT as an HIV-1 RNA-binding protein (31). Studying PIMT in more detail revealed noted that it has a methyltransferase domain and an m7GpppG hypermethylase activity (Fig. S1) (30 –32). We were intrigued to investigate whether PIMT regulates the expression of HIV-1 RNAs, perhaps at the level of RNA-cap hypermethylation. To check whether PIMT influences HIV-1 gene expression, HeLa cells were transfected with the HIV-1 NL4-3 molecular clone with an HA-PIMT plasmid (Fig. 1A, lane 2) or vector control (Fig. 1A, lane 1). The expression of HIV-1 proteins was monitored by Western blot analysis, and PIMT was seen to enhance the expression of viral p55 and p24 by ≈10-fold (Fig. 1A). Viral reverse-transcriptase (RT) activity (Fig. 1A, Bottom; RT assay) released into the cell culture supernatant also was increased by cotransfection of PIMT plasmid.

Fig. 1.

PIMT enhances HIV-1 gene expression. (A) HeLa cells were co-transfected with HIV-1 NL4-3 molecular clone and HA-tagged PIMT (HA PIMT-WT) or HA-alone (vector). Cell lysates were prepared 24 hours post transfection and analyzed by Western blotting using anti-HIV-1 antibody. Arrows point to p55 and p24 Gag proteins. Loadings were normalized to β-actin. RT assays (Fig 1A, lower panel) of the cell culture supernatants are shown at the bottom and are consistent with increased viral gene expression and virus release. (B) Expression of HA-PIMT-WT, but not HA-PIMT-N nor HA-PIMT-C, activates HIV-1 gene expression. Lanes 2 and 3 were run in duplicate. (C) PIMT knockdown using siRNA efficiently decreased cell endogenous PIMT (lanes 3 and 4). Controls were scrambled irrelevant siRNAs. (D) siRNA-mediated PIMT knockdown reduced HIV-1 expression (compare lanes 7 and 8 with lanes 5 and 6; results are from the same gel from the same experiment with some intervening lanes removed for clarity). Loadings were normalized to β-actin.

HA-tagged deletion mutants of PIMT [PIMT-N (1–299 aa) and PIMT-C (504–853 aa)] (Fig S1A) were used to map the region(s) required to enhance HIV-1 expression. PIMT-WT (Fig. 1B, lanes 2 and 3) increased HIV-1 expression, but PIMT-N and PIMT-C (Fig. 1B, lanes 4 and 5) did not. The enhancing activity of PIMT on HIV-1 expression was investigated by knocking down cell-endogenous PIMT using siRNAs. PIMT in cells was knocked down by >75% using two independent siRNAs, PIMT si-1 and PIMT si-2 (Fig. 1C, lanes 3 and 4). Transfection of HIV-1 NL4-3 into cells that were depleted for PIMT using siRNAs showed significantly reduced HIV-1 protein expression (Fig. 1D; compare lanes 7 and 8 with lanes 5 and 6). Collectively, the PIMT overexpression and knockdown findings are consistent with a role for this protein in modulating HIV-1 expression.

PIMT Is a Nucleocytoplasmic Shuttling Protein That Binds the RNA Cap.

The human PIMT protein was initially described as a nuclear receptor transcriptional coactivator interacting protein (26). Later, others suggested that PIMT is the human homolog of yeast TGS1 and has methyl transferase activity (30 –32). In vitro, PIMT can efficiently transfer the methyl-group from S-adenosylmethionine (SAM) to m⁷G(5′)ppp(5′)G, but not to ATP, CTP, GTP, UTP, or m^2,2,7G(5′)ppp(5′)G (Fig S1C, lanes 12–15, and 17), consistent with its reported function as an m7G RNA cap hypermethylase.

We also verified that PIMT is a bona fide human RNA cap methyltransferase by asking whether m7G Sepharose beads would pull down PIMT from cell lysates. In that assay, the m7G beads efficiently captured PIMT from HeLa cells transfected with a PIMT WT expression vector (Fig. S2A, lane 5). The beads also captured two PIMT deletion mutant proteins, PIMT-MC and PIMT-C (Fig. S2A, lanes 7 and 8), which retained the C terminus of the protein; however, the beads did not capture a third PIMT mutant (PIMT-N) that lacked the C terminus (Fig. S2A, lane 6). These findings indicate that the PIMT C terminus contains its methyltransferase domain (26) and its RNA cap-binding property.

RNA-binding proteins often traffic with their substrate RNAs. We next sought to clarify PIMT localization inside the cell. HeLa cells were stained with anti-PIMT, and PIMT was found to localize in coilin-rich nuclear Cajal (38) bodies (Fig. S2B, Bottom; anti-coilin and anti-PIMT). Cajal bodies are sites involved in snRNP and telomerase RNA biogenesis and in the trafficking of snoRNPs and snRNPs (39, 40). Consistent with PIMT's role as a cap hypermethylase, TMG RNA-capped RNAs (e.g., snRNAs, snoRNAs, and telomerase RNAs) were abundantly stained by anti–TMG cap antibody (Fig. S2B, Top Right) in the same PIMT-localized, coilin-rich bodies (Fig. S2B, Top; anti-coilin).

Cell-endogenous PIMT is prominent in Cajal bodies. However, when we overexpressed the protein using GFP-PIMT, we observed a large proportion of GFP-PIMT in the cytoplasm (Fig. S2C, Top Right; no LMB). Interestingly, if GFP-PIMT–transfected HeLa cells were treated with leptomycin B (LMB), then virtually all GFP-PIMT protein became nuclear-restricted (Fig. S2C; LMB-treated). The overexpression experiments revealed a property of PIMT that is otherwise difficult to discern from the cell-endogenous protein—that PIMT can be shuttled from the nucleus into the cytoplasm via an LMB-sensitive pathway.

We further investigated PIMT methyltransferase activity by introducing two single-aa point changes into the SAM-binding domain of PIMT (PIMT-Mut5; Fig. S3A). This dual mutation abrogated the protein's methyltransferase activity (Fig. S3A, lane 3), and the loss of RNA methyltransferase activity in PIMT-Mut5 (Fig. S3B, lane 3) abolished the enhancing ability for HIV-1 expression seen for WT PIMT (Fig. S3B; compare lanes 2 and 3).

Intracellular HIV-1 RNAs Are m7G- and TMG-Capped.

The finding that PIMT can influence HIV-1 gene expression suggests that HIV-1 RNAs should be m7G- and/or TMG-capped. Previous in vitro transcription assays have indicated that HIV-1 mRNAs can be capped (24, 25). To the best of our knowledge, experimental characterization of the intracellular status of HIV-1 RNA capping has not yet been performed. To address the state of viral RNA capping, we characterized HIV-1 RNAs using two discrete antibodies directed against cap structures (36, 41, 42). One of the antibodies recognizes the m7G cap, whereas the other recognizes the hypermethylated TMG cap. Using known cellular RNAs that are either m7G-capped or TMG-capped, we performed characterizations that verifed the antibody specificities (Fig. S4). Next, we used the antibodies to immunoprecipitate RNAs from HIV-1 latently infected ACH2 cells activated for viral expression by treatment with phorbol myristic acid (PMA). The immunoprecipitated RNAs were extracted and analyzed by RT-PCR using primers complementary to sequences located directly before and after the major HIV-1 5′ splice junction. Thus, primers A and B (primer set B; Fig. 2A) would amplify all HIV-1 transcripts, whereas primers A and C (primer set C; Fig. 2A) would amplify only unspliced HIV-1 transcripts. HIV-1 RNA was precipitated by anti-TMG and anti-m7G antibodies (Fig. 2A, lanes 2, 3, 6, and 7), indicating that intracellular HIV-1 transcripts are indeed variously m7G- or TMG-capped.

Fig. 2.

Intracellular HIV-1 RNAs are capped, and PIMT affects the cytoplasmic distribution of HIV-1 unspliced RNA. (A) Total RNA isolated from ACH2 cells after PMA treatment was immunoprecipitated with either anti-TMG or anti-m7G. Two different primer sets (sets B and C) were used for RT-PCR. HIV-1–specific RNA was captured by both anti-TMG and anti-m7G antibodies (RT-PCR; lanes 2, 3, 6, and 7). (Upper) A schematic representation of the proviral RNA with primers. (B) PIMT expression increases TMG capping of HIV-1 RNAs. Total RNA was isolated from HeLa cells cotransfected with pNL4-3 and HA-PIMT. RNAs were captured as described previously using anti-TMG and anti-m7G. RT-PCR analyses of the antibody-captured RNAs demonstrated that PIMT increased the amount of HIV-1 RNA pulled down with anti-TMG, but not of that pulled down by anti-m7G (RT-PCR; lanes 2 and 5). (C) PIMT expression increased the relative cytoplasmic versus nuclear distribution of TMG-capped unspliced HIV-1 RNA. qRT-PCR analyses were performed on HIV-1–unspliced RNAs isolated from the nuclear (Left) and cytoplasmic (Right) fractions. Quantifications of TMG-capped HIV-1 RNAs expressed from transfected pNL4-3 in the presence of control vector (pCMV HA; red triangle) and in the presence of transfected PIMT (pCMV HA PIMT; blue circle) are shown. Compared with results from the expression of control pCMV HA vector, the overexpression of PIMT increased the relative abundance of cytoplasmic versus nuclear TMG-capped unspliced HIV-1 RNA (i.e., a leftward shift of the blue curve).

We wondered whether the amount of TMG-capped HIV-1 RNA is influenced by PIMT expression; thus, we investigated whether the deliberate overexpression of PIMT increased the amount of TMG HIV-1 RNA. The coexpression of HA-PIMT with HIV-1 NL4-3 indeed significantly increased the amount of viral TMG RNA over that seen with a cotransfection of vector CMV- HA-alone + pNL4-3 (Fig. 2B, RT-PCR; compare lanes 5 and 2). As a control for specificity, the level of m7G-capped HIV-1 RNA was unchanged by HA-PIMT expression (Fig. 2B, RT-PCR; compare lanes 6 and 3). Going forward, we plan to also test a Rev-independent CTE-dependent HIV-1 (43) to see if its requirement for PIMT in viral protein expression is reduced. However, past experience with the HIV-1 Rev-independent CTE-dependent molecular clone (43) suggests that it replicates substantially less well than parental pNL4-3; thus, a direct comparison between the two viruses is difficult and requires careful and thoughtful normalizations of the respective readouts.

To understand how PIMT/hypermethylation of viral RNA cap contributes to increased HIV-1 gene expression, we next examined the effect of cap hypermethylation on the cellular distribution of unspliced HIV-1 RNA. PIMT expression was observed to increase the relative cytoplasmic versus nuclear distribution of TMG-capped unspliced HIV-1 RNA (Fig. 2C). qRT-PCR analysis of unspliced HIV-1 RNA showed that compared with cells transfected with control pCMV HA vector cells with overexpressed PIMT had increased relative abundance of cytoplasmic versus nuclear TMG-capped unspliced HIV-1 RNA (i.e., a leftward shift of the blue curve; Fig. 2C). Analysis of the effect of PIMT expression on the distribution of cellular GAPDH demonstrated that PIMT expression compared with control vector (pCMV HA) expression had no effect on the relative cytoplasmic versus nuclear profiles of the m7G-capped GAPDH transcript (Fig. S5). In contrast, PIMT overexpression compared with control vector expression did change the relative cytoplasmic versus nuclear profiles of TMG-capped U2 snRNA (compare the relative position of the blue relative to the red curve in the nucleus vs. the cytoplasm; Fig. S5). snRNAs have nuclear export and nuclear reimport steps; in this instance, which steps might be affected by PIMT remain to be determined.

PIMT Selectively Modulates the Expression of Rev/RRE-Dependent RNAs.

p55 and p24 HIV-1 proteins are encoded from Rev/RRE-dependent RNAs (44 –46). Next, the specificity of PIMT for the expression of Rev/RRE-dependent RNAs was tested. We cotransfected a Rev/RRE-dependent Gag-Pol-RRE plasmid with either PIMT alone or PIMT + Rev (Fig. 3A). In the absence of Rev no enhanced Gag expression was observed, while in the presence of Rev, PIMT substantially increased Gag expression (Fig. 3A; compare lanes 2 and 4). We also tested PIMT on a Gag construct that was Rev/RRE-independent, but CTE-dependent (Gag-Pol-CTE), for expression (47, 48). Unlike a Rev/RRE RNA, a CTE-transcript uses TAP instead of CRM-1 for nuclear export (49 –51). Interestingly, the expression of Gag CTE RNA was not influenced by PIMT with or without the coexpression of Rev (Fig. 3A; compare lanes 5 and 7 and lanes 6 and 8).

Fig. 3.

PIMT activity is selective for Rev/RRE-dependent transcripts. (A) HeLa cells were transfected with either RRE- or CTE-HIV-1 Gag expression constructs along with PIMT alone or PIMT + Rev. In the presence of Rev, PIMT increased the expression of RRE-containing transcripts but had no effect on the expression of CTE-containing transcripts. (B) PIMT interacts with HIV-1 Rev. HeLa cells were cotransfected with either GFP-Rev or GFP-Tat and HA-tagged PIMT. Immunoprecipitations performed with anti-HA showed that Rev (lane 2) interacted with PIMT, but Tat did not. (C) PIMT is recruited to Rev/RRE RNAs. HeLa cells were cotransfected with Rev and either pCMV-GagPol-RRE or pCMV-GagPol-CTE with HA-PIMT WT or HA vector. Lysates were then immunoprecipitated with anti-HA. RNA was isolated from the immunoprecipitates and analyzed by RT-PCR as indicated. (D) PIMT mediates TMG-capping of RRE-containing transcripts. HeLa cells were transfected with pCMV Gag-Pol-RRE or pCMV Gag-Pol-CTE plasmids and pRSV-Rev with or without PIMT. Immunoprecipitations were performed using anti-TMG or anti-m7G. RT-PCR analysis of RNAs pulled down by the antibodies showed increased capture of RRE-containing transcripts, but not of CTE-containing transcripts, by anti-TMG (GAG; lanes 3, 6, 9, and 12). GAPDH RNA is known to be only m7G-capped and serves as a specificity control for antibody immunoprecipitation. (E) PIMT does not affect the expression of spliced HIV-1 transcripts. The expression of spliced HIV-1 transcript was measured by the synthesis of Tat from the pNL4-3 molecular clone. The activity of Tat expressed from pNL4-3 was measured using a standard HIV-1 LTR-luc reporter plasmid. HeLa cells were cotransfected with pCMV HA + pNL4-3 + pLTR-luc or with pCMV HA PIMT + pNL4-3 + pLTR-luc. Three independent transfections were performed, and the results verified that expression of PIMT did not affect the activity of Tat expressed from pNL4-3 in a statistically significant manner. Activation of LTR-luc by Tat expressed from transfection of pNL4-3 was reproducibly >50 fold. (F) Knockdown of cell-endogenous PIMT by siRNA did not significantly affect Tat expression for pNL4-3. Knockdowns of PIMT using siRNAs were as shown in Fig. 1. There were small, statistically insignificant differences in Tat activity as measured by LTR-luc activation when cell-endogenous PIMT expression was reduced using two independent siRNAs. The experiments were performed three independent times. (G) PIMT expression does not affect the expression of HIV-1 spliced transcripts. HeLa cells were transfected with pNL4-3 with (red) or without (blue) PIMT, and cytoplasmic RNAs were analyzed by qRT-PCR using primers specific for spliced (primer sj4.7A and primer A) and unspliced viral transcripts. Spliced viral RNAs (Upper) were captured by anti-m7G antibody, and unspliced viral RNAs were captured by anti-TMG antibody. Analyses were performed three times, as shown. Note the lack of change in the blue (without PIMT) and red (with PIMT) curves for the spliced HIV-1 RNAs (Upper), and the distinct shift of the red (with PIMT) curves from the blue (without PIMT) curves for the unspliced HIV-1 RNAs (Lower).

To understand how Rev contributes to effect of PIMT on the expression of HIV-1 RRE-containing RNAs, we wondered if PIMT might be recruited to RRE RNAs by Rev. We tested if Rev could bind PIMT directly. Indeed, in RNA-free coimmunoprecipitations, PIMT was bound by Rev (Fig. 3B, lane 2), but not by the control Tat (Fig. 3B, lane 4) protein. We next asked whether PIMT could be found in a Rev-RRE–containing ribonucleoprotein complex. We coexpressed HA-PIMT and Rev with either Gag-Pol-RRE or Gag-Pol-CTE plasmid and immunoprecipitated PIMT using monoclonal anti-HA. The immunoprecipitations revealed PIMT association in the presence of Rev with Gag-Pol-RRE RNA (Fig. 3C, lane 2), but not Gag-Pol-CTE RNA (Fig. 3C, lane 4). Consistent with a Rev-dependent recruitment, in the absence of Rev, PIMT did not coimmunoprecipitate with either Gag-Pol-RRE- or Gag-Pol-CTE- RNA.

The recruitment of PIMT by Rev to RRE-containing RNAs might be expected to increase TMG-capping of these RNAs. Indeed, the coexpression of PIMT and Rev increased intracellular TMG-capped Gag-Pol-RRE RNAs (Fig. 3D, lanes 3 and 6). In contrast, in the same assay, PIMT did not change the amount of TMG-capped Gag-Pol-CTE transcripts (Fig. 3D, lanes 9 and 12). The collective results are consistent with a model in which PIMT is recruited to Rev-bound RRE viral transcripts and not to Rev-unbound CTE viral transcripts.

We next examined whether PIMT affects the expression of spliced HIV-1 transcripts. We measured the expression of spliced HIV-1 transcript by the synthesis of Tat from the pNL4-3 molecular clone using an HIV-1 LTR-luciferase (LTR-luc) reporter (Fig. 3E). We found that PIMT expression did not affect the expression of Tat from pNL4-3 (Fig. 3E). We further investigated whether the knockdown of cell-endogenous PIMT by siRNA affected Tat expression from pNL4-3 (Fig. 3F). The siRNAs used in this experiment are the same as si-1 (Fig. 3F, Middle) and si-2 (Fig. 3F, Right) used in Fig. 1. No statistically significant effect on Tat expression as measured by LTR-luc activity was observed when cell-endogenous PIMT expression was reduced using the two independent siRNAs (Fig. 3F). Finally, to more directly measure the effect of PIMT on spliced transcripts from pNL4-3, we used specific primers that would measure spliced HIV-1 RNAs and performed qRT-PCR analyses on anti–m7G-captured viral transcripts. The results show that PIMT overexpression had no effect on the cytoplasmic expression of spliced viral transcripts (Fig. 3G, Upper). In contrast, control parallel qRT-PCR analyses showed that, as expected, PIMT overexpression significantly affected the cytoplasmic expression of anti–TMG-captured unspliced HIV-1 RNAs (Fig. 3G, Lower).

RNA Cap Hypermethylation Influences HIV-1 Replication in Primary T Cells.

The foregoing experiments do not address how RNA cap hypermethylation affects HIV-1 replication in authentically infected primary T cells. Activation of freshly isolated quiescent human peripheral blood mononuclear cells (PBMCs) is a well-known requirement for these cells to become permissive for HIV-1 replication. This requirement suggests that such agents as PHA/IL-2 increase the expression of certain cellular factors needed by the virus. Conceivably, PIMT is an “activation”-dependent factor required in PBMCs for HIV-1 replication. To address this possibility, we collected fresh PBMCs from several donors and cultured them with or without PHA for 2 d, followed by 2 additional d in IL-2 (Fig. 4A). Separately, we also enriched the PBMCs for CD4⁺ T cells by first depleting them of other cell types and then treating the cells with PHA and IL-2 (Fig. 4B). After PHA/IL-2 treatment, we evaluated the cells for PIMT by Western blot analysis. As shown in Figs. 4 A and B, activation of PBMCs or primary CD4⁺ T cells significantly increased the expression of PIMT, indicating that PIMT might be a limiting factor in quiescent cells.

Fig. 4.

PIMT modulates HIV-1 replication in primary T cells. PIMT expression is influenced by the activation state of primary cells. (A and B) PIMT expression in stimulated (PHA/IL-2) and unstimulated PBMCs (A) and CD4⁺ T cells (B) was assessed by Western blot analysis. PIMT expression was increased by activation of cells with PHA/IL-2. D1, D2, D3, and D4 represent different donors. US, mock stimulated; S, stimulated with PHA/IL-2. (C and D) TLC results showing that sinefungin (lane 3), but not amino-cyclo-pentane (lane 2), inhibits in vitro methyltransfrease activity of PIMT (C) and HIV-1 replication in PBMCs (D). HIV-1 replication was measured by RT activity in the culture supernatants. Results are averages of two experiments. (E) Sinefungin treatment decreased the expression of Rev/RRE dependent HIV-1 p24 transcript. (Left) Western blot analysis. (Right) Quantification of p24 based on the Western blot. (F) Sinefungin treatment did not affect the expression of spliced HIV-1 Tat transcript. HeLa cells transfected with NL4-3 were treated with sinefungin (10 μM) for 24 h. Cell lysates were collected and analyzed by Western blot for p24 expression (E), and in parallel, Tat expression was quantified by LTR-luc assay (F). Sinefungin decreased p24 expression by approximately 4-fold (E) without affecting Tat expression (F).

Is PIMT's methyltansferase activity required for productive HIV-1 replication in PBMCs? Although the knockdown of PIMT in cell line experiments affected the expression of a transfected HIV-1 molecular clone (Fig. 1 C and D), efficient knockdowns are technically untenable for PBMCs or primary CD4⁺ T cells. It is possible to treat PBMCs with small molecule methyltransferase inhibitors, however. We surveyed several known methyltransferase inhibitors and found one that inhibited m7G to TMG cap hypermethylation (sinefugin) and another that did not (amino-cyclo-pentane) (Fig. 4C, lanes 2 and 3). In assays for inhibition of productive HIV-1 replication, sinefungin, but not amino-cyclo-pentane, suppressed HIV-1 replication (Fig. 4D) in PBMCs at concentrations that did not exhibit cellular cytotoxicity.

We next investigated how sinefungin treatment of transfected HeLa cells affected viral expression. In treated cells, we observed that the compound decreased the expression of Rev/RRE-dependent HIV-1 p24 protein by approximately 4-fold (Fig. 4E). Interestingly, singefungin treatment did not affect the expression of Tat protein synthesized from spliced Rev-independent HIV-1 transcripts (Fig. 4F).

Discussion

Certain viruses use the cellular capping machinery, whereas other viruses encode their own RNA capping enzymes (11). For HIV-1, how RNA capping and cap methylation influence viral replication has not been investigated extensively. Here we show that HIV-1 RNAs have m7G or TMG caps, and that the latter occurs through PIMT hypermethylation of the former. Our experimental results are consistent with the following scenario. Initially, transcribed HIV-1 transcripts are m7G-capped. A subset of viral transcripts containing RRE (i.e., unspliced or partially spliced HIV-1 RNAs) are bound by Rev, which then recruits PIMT to these RNAs. PIMT brought in this way to the viral RNA hypermethylates the m7G cap to a TMG cap. Acquisition of the PIMT-mediated TMG cap facilitates the optimal expression of these RNAs and likely serves to direct them to the CRM-1 pathway (Fig. S6). In accordance with the foregoing, the expression of Rev- and CRM-1–dependent Gag-Pol-RRE RNA is regulated by PIMT, whereas the otherwise identical Rev- and CRM-1–independent Gag-Pol-CTE RNA is insensitive to PIMT regulation (Fig. 3A). We also observed that expression of viral protein from spliced HIV-1 transcripts (e.g., Tat; Fig. 3 E and F) is not affected by PIMT. Moreover, because cellular RNAs that are transported by CRM-1 (e.g., snRNAs, snoRNAs, and telomerase TLC1 RNAs) are TMG-capped (45, 52 –54), and because only RRE-containing RNA bound by Rev can recruit PIMT for TMG capping, our results suggest that the TMG cap found on snRNA, snoRNA, and telomerase RNA, as well as on unspliced/partially spliced HIV-1 RNAs, might be a conserved CRM-1 recognition marker (Fig. S6).

Further experiments are needed to fully clarify the role of TMG capping in RNA transport. For snRNAs, there is evidence in some systems that the RNAs are exported from the nucleus with a monomethyl cap and that cap hypermethylation is required for nuclear reimport, although there may be additional complexities in yeast cells (55). Interestingly, Gallardo et al. (36) reported that yeast telomerase RNA (TLC-1), which shuttles between the nucleus and cytoplasm similar to snRNA, acquires its TMG cap modification in the nucleus, not in the cytoplasm. Others have reported that small nucleolar RNAs that are not exported from nucleus are cap-hypermethylated in the nucleus (30, 56), and that this TMG-modification likely guides the interaction of snoRNA with CRM1, which then directs the snoRNA to the nucleolus (56). Thus, some TMG hypermethylation events likely occur in the nucleus, whereas others occur in the cytoplasm. The hypermethylated TMG cap could then dictate interactions with CRM1 for intranuclear or extranuclear RNA transport, although complete details remain to be clarified.

Once an mRNA is in the cytoplasm, the binding of m7G cap by eIF-4E is the first step mRNA ribosome association and translation. There is some concern that if an RNA has a TMG cap instead of an m7G cap, this could have repercussions for the initiation of translation. The proportion of hypermethylated guanosine-capped RNAs was found to be lower in polysomal RNA than in nonpolysomal RNA, suggesting that TMG RNAs may not be translated efficiently (57, 58). How these in vitro findings reflect in vivo translation is not entirely clear, however. Our findings suggest that PIMT expression increases the amount of TMG-containing unspliced/partially spliced HIV-1 RNAs that are then exported through the CRM-1 pathway and expressed into the corresponding proteins in the cytoplasm. Compatible with our results, in Caenorhabditis elegans and Ascaris lumbricoides, nematode mRNAs can acquire TMG caps by transsplicing, and in these settings, virtually all actin and ribosomal protein mRNAs are TMG-capped and loaded onto polysomes and are functionally translated with efficiency (59 –61). Similarly, during the replication of togaviruses (e.g., Semliki Forest virus and Sindbis virus), late viral mRNAs acquire hypermethylated guanosine caps, and the expression of late proteins is proficient (21, 22).

In summary, intracellular HIV-1 RNAs are characterized here to be m7G- and TMG-capped. Genesis of the TMG cap RRE RNA requires the PIMT methyltransferase, which is recruited by the Rev protein. PIMT-mediated TMG capping increases the expression of HIV-1 proteins by the TMG cap apparently serving as a marker for the CRM-1 pathway. Our work suggests the possibility that small-molecule inhibitors of RNA methyltransferases might be a new class of anti–HIV-1 drugs. Indeed, we and others (62 –65) have found that treating cells with RNA methylation inhibitors can suppress HIV-1 replication (Fig. 4).

Experimental Procedures

Detailed information on cells and plasmids, antibodies, Western blot analysis, confocal imaging, and coimmunoprecipitation protocols are provided in SI Experimental Procedures.

Methyltransferase Assays.

Methyltransferase assays were performed as described previously (66). Reaction mixtures (30 μL) containing 50 mM Tris-HCl (pH 8.0), 5 mM DTT, 50 μM [₁₄C-CH3] AdoMet, 2.5 mM of the specified nucleotide or CAP analogs, and PIMT were incubated for 30 min at 37 °C. Aliquots (10 μL) were spotted onto polyethylenimine cellulose TLC plates, which were developed with 0.05 M ammonium sulfate. ¹⁴C-labeled material was visualized by phosphorimaging.

Isolation and Immunoprecipitation of RNA.

Total RNA from cells was isolated with Tri-Reagent (Sigma-Aldrich). Nuclear and cytoplasmic fractions of cells were prepared by treating the cells with cell fractionation buffer (Paris Kit; Applied Biosystems), and RNA was extracted with Tri-Reagent. Extracted RNA was resuspended in radioimmunoprecipitation assay (RIPA) buffer containing 10 mM Ribonucleoside Vanadyl complex (New England Biologicals). The cap structure of HIV-1 RNA was analyzed by incubating purified RNA with H20 or K121 antibody immobilized on protein G–agarose beads in RIPA buffer at 4 °C for 4 h. RNA bound to the antibodies was extracted using Tri-Reagent and analyzed using the One-Step RT-PCR Kit (Qiagen) or by qRT-PCR using the iScript One-Step RT-PCR Kit with SYBR Green (Bio-Rad) in accordance with the manufacturer's instructions. Samples were reverse-transcribed at 50 °C for 30 min, and amplification was performed after an initial step at 95 °C for 10 min, followed by 20–40 cycles of 95 °C for 30 s, 55 °C for 30 s, and 72 °C for 60 s. The primers and sequences used in the analyses are listed in Table S1.