HIV co-opts a cellular antiviral mechanism, activation of stress kinase PKR by its RNA, to enable splicing of rev/tat mRNA

Inhibitors of PKR activation repress HIV mRNA expression

TNF-α, a cytokine pivotal for protective immunity, is expressed promptly during inflammatory responses; in human peripheral blood mononuclear cells, TNF-α mRNA becomes maximally expressed within 3 h [12]. Efficient TNF-α mRNA splicing is achieved through a 104-nucleotide RNA element in the 3’-UTR that activates PKR [13]. The TNF-α RNA activator of PKR causes mRNA splicing to be sensitive to 2-aminopurine, an eIF2α kinase inhibitor [12, 13]. The RNA element in TNF-α pre-mRNA that activates PKR renders splicing not only fully dependent on PKR activation but also highly efficient [10, 13]. The 59-nucleotide TAR stem-loop element that is present at the 5′ and 3′ termini of HIV transcripts (Fig. 1A, B) activates purified recombinant human PKR in vitro [18]. Potentially, therefore, TAR might, as for TNF-α, serve in the function of PKR activator to render splicing of HIV mRNA efficient.

Fig. 1figure 1

Production of HIV-1 mRNA species is sensitive to PKR antagonists. A Proviral DNA genome maps of HIV-1 LAI (TARwt), HIV-rtTA (TARm) and HIV-rtTA-ΔTAR (ΔTAR), with the long terminal repeat (LTR) subdivided into U3, R, and U5 domains. B TAR secondary structures followed by the first five nucleotides of the adjacent poly(A) hairpin (light gray). The Tat/TAR axis of transcriptional regulation was inactivated in TARm by nucleotide substitutions in the bulge and loop. The HIV-rtTA variant lacking TAR (ΔTAR) has a new transcription start site at the second nucleotide of the poly(A) hairpin [22]. C Human HEK-293 T cells were transfected with a vector carrying the HIV-1, TARm or ΔTAR genome, in the absence or presence of the indicated concentrations of PKR inhibitor (PKRi). Total RNA was isolated 48 h after transfection and analyzed by northern blot using a probe that detects all HIV-1 RNA variants. Unspliced (9 kb), singly spliced (4 kb) and multiply spliced (2 kb) RNA size classes [30] and a band resulting from transcriptional read-through on the vector (*)[57] are indicated. The transcripts observed for the different virus constructs vary in size due to the mutations introduced to generate the HIV-rtTA variants, TARm and ΔTAR, including deletions and insertions (rtTA replaced Nef, different 3’UTR). Bottom panel shows 18S and 28S ribosomal RNA loading controls (ethidium bromide staining). A representative of 3 experiments is shown. D HEK-293 T cells were transfected with vector carrying the HIV-1 genome, in the absence or presence of the indicated amounts of Vaccinia E3L expression vector (E3L, ng DNA per transfection). Total RNA was isolated and analyzed by Northern blotting as in C. A representative of 3 experiments is shown

HIV expresses a full length ~ 9 kb transcript, used as mRNA for Gag and Pol production and as viral genome, and a large variety of singly spliced (~ 4 kb) and multiply spliced (~ 2 kb) transcripts for all other viral proteins, including Rev [30, 31]. To examine whether the production of HIV mRNA might depend on PKR, we transfected human HEK-293 T cells with a plasmid carrying the complete wild type HIV-1 genome (LAI strain) and quantitated the expression of various HIV mRNA species in the presence of increasing doses of a small-molecule inhibitor of the kinase catalytic site in PKR, PKRi. Northern blot analysis showed that PKRi progressively inhibited production of all HIV mRNA species, including singly spliced and multiply spliced mRNAs (Fig. 1C and Additional file 1: Fig. S1). A similar pattern of inhibition of mRNA species encoded by HIV was observed when instead of PKRi, we co-expressed the Vaccinia E3L protein, a viral PKR antagonist [23] that strongly attenuates the action of the intragenic IFN-γ RNA activator of PKR [9] (Fig. 1D). E3L competes with PKR in binding to the activating RNA, forming an E3L-PKR-RNA complex in which the N-terminal half of E3L interacts physically with the protein kinase domain of PKR [23]. The independent results with PKRi and E3L support a role for PKR activation in the control of HIV mRNA expression.

To study the possible role of TAR in PKR activation, we used an HIV genomic construct (TARm) in which the Tat-TAR transcription mechanism is inactivated through mutations in Tat and TAR (nucleotide substitutions in the bulge and loop sequence) and functionally replaced by the integrated doxycycline-inducible Tet-On gene regulation system [21]. TARm does not depend on TAR for the activation of transcription, which makes it possible to study other functions of TAR in gene expression. As for the wild type virus, expression of the various mRNA classes by TARm was sensitive to PKRi (Fig. 1C and Additional file 1: Fig. S1). We also tested RNA production of a TARm derivative that lacks both 5’ and 3’ TAR elements and, like TARm, replicates efficiently in infected cell cultures [22]. This ∆TAR variant produced similar amounts of the different viral RNA classes as TARm and their production was inhibited likewise by PKRi (Fig. 1A, C and Additional file 1: Fig. S1). These results suggest that the HIV RNA genome may harbor, in addition to the TAR element, another activator of PKR.

Splicing of rev/tat mRNA is regulated by activation of PKR

The inhibition of the production of all size classes of HIV mRNA by PKRi and by E3L, including that of the unspliced RNA (Fig. 1C, D), did not resolve whether splicing or another expression step requires the activation of PKR. Splicing effects are difficult to interpret within the complete HIV genome context. For example, decreased splicing will lower the production of the regulatory Tat and Rev proteins, both encoded by multiply spliced mRNAs [32]. Tat activates viral transcription and influences splicing of the viral RNAs [33, 34], whereas Rev stimulates nuclear export of unspliced and singly spliced RNAs, which reduces splicing [35, 36]. Unspliced RNA precursors that are retained in the nucleus may be degraded.

Given the complexity of full-length HIV expression and its multiple splicing products, we created a vector expressing the 3’ half of the HIV genome, including the large rev/tat intron whose excision constitutes the sole splicing event for RNA encoded by this vector (pcDNA-3′HIV) (Fig. 2A). Compared to full-length HIV (Fig. 1A), the 3′HIV construct covers 40% of the nucleotide sequence, including the complete 3′ domain (Fig. 2A). To avoid confounding effects on splicing by the Tat and Rev proteins, the viral sequences in this vector start downstream of the rev and tat AUG translation start codons, so that no functional Tat and Rev can be produced. Moreover, the construct lacks the 5’ TAR element that enhances splicing at the major splice donor site via the Tat protein [33, 34]. Upon transfection of this vector into cells, we monitored splicing of the pre-mRNA transcript containing the rev/tat intron by quantitating unspliced and spliced RNA. Ribonuclease protection analysis (Fig. 2B, C; the ratio of spliced over unspliced RNA within each individual lane in Fig. 2B reflects splicing efficiency plotted in Fig. 2C) and quantitative real-time polymerase chain reaction (qRT-PCR) analysis (Fig. 2D) each showed that the transcript was spliced efficiently, resulting in a high mRNA/pre-mRNA ratio.

Fig. 2figure 2

Expression of viral PKR antagonist protein Vaccinia E3L or Ebola VP35 inhibits excision of HIV rev/tat intron. A Vector encoding the 3’ half of the HIV genome. Vector pcDNA-3’HIV carries the 3’ portion of the HIV-1 genome expressed under the constitutive cytomegalovirus (CMV) promoter. The viral sequences start downstream of the rev AUG translation start codon to prevent Tat and Rev production. The construct retains splice donor D4 (5’ss #4) and splice acceptor A7 (3’ss #7), allowing for a single splicing event of the rev/tat intron (bottom). B, C E3L and VP35 inhibit splicing of rev/tat mRNA. BHK-21 cells were cotransfected with 1 µg of pcDNA-3’HIV DNA together with 1 µg pBS empty vector (EV), E3L expression vector (E3L) or VP35 expression vector (VP35). Total RNA was isolated at 18 h post-transfection. Unspliced pre-mRNA (413 nt) and spliced rev/tat mRNA (140 nt) were determined by RNase protection analysis (B). The top autoradiogram (pre-mRNA) underwent a longer exposure. Band intensity was quantitated and the ratio of spliced over unspliced RNA within each lane, which reflects splicing efficiency, is plotted in bar graph (C) (error bars, SEM; n = 3). A representative experiment is shown. D E3L and VP35 inhibit splicing of rev/tat mRNA. In independent transfections, performed as in B, total RNA was isolated at 12 h; spliced and unspliced rev/tat transcripts were determined by qRT-PCR. Splicing efficiency is expressed as mRNA/pre-mRNA ratio (error bars, SEM; n = 3). E Whereas rev/tat intron excision is inhibited by co-expression of E3L, it is stimulated by co-expression of PKR. BHK-21 cells were cotransfected with 1 µg of pcDNA-3’HIV DNA together with 1 µg pBS empty vector (EV) or with vector expressing E3L or human PKR. Total RNA was isolated at 12 h post-transfection. Spliced and unspliced rev/tat transcripts were determined by qRT-PCR. Splicing efficiency is expressed as mRNA/pre-mRNA ratio (error bars, SEM; n = 3). F Splicing of rev/tat mRNA is blocked by expression of K296R trans-dominant negative mutant PKR. BHK-21 cells were cotransfected with 1 µg of pcDNA-3’HIV DNA together with 1 µg pBS empty vector (EV) or with vector expressing mutant K296R or human PKR. Total RNA was isolated at 12 h post-transfection. Spliced and unspliced rev/tat transcripts were determined by qRT-PCR. Splicing efficiency is expressed as mRNA/pre-mRNA ratio (error bars, SEM; n = 3)

Excision of the rev/tat intron was inhibited strongly by co-expression of the viral PKR antagonist proteins, Vaccinia E3L and Ebola VP35 [27, 37], resulting in accumulation of the unspliced pre-mRNA and reducing the mRNA/pre-mRNA ratio which denotes splicing efficiency (Fig. 2B–D). By contrast, overexpression of PKR in the cell enhanced rev/tat intron excision, reflected by a significant increase in mRNA/pre-mRNA ratio (Fig. 2E and F). Notably, as shown in Fig. 2F, splicing of rev/tat mRNA was abrogated by expression of a trans-dominant negative mutant of PKR, K296R, that blocks the activation of PKR in the cell [23].

These results demonstrate a positive and indispensable role for PKR activation in HIV rev/tat mRNA splicing.

Phosphorylation of eIF2α controls rev/tat mRNA splicing

Repression of mRNA translation by PKR depends strictly on phosphorylation of the translation initiation factor protein chain, eIF2α, at its sole phosphorylation site, Serine51 [3, 38]. Expression of non-phosphorylatable mutant eIF2α, eIF2αS51A, but not of wild type eIF2α, abrogates efficient splicing of TNF-α [10] and globin pre-mRNA [14], showing that splicing driven by the TNF-α and globin RNA activators of PKR depends tightly on eIF2α phosphorylation. eIF2αS51A inhibits eIF2α phosphorylation by activated PKR [10]. Indeed, expression of non-phosphorylatable mutant eIF2αS51A, but not of wild type eIF2α, abrogated efficient splicing of rev/tat mRNA encoded in pcDNA-3’HIV (Fig. 3). qRT-PCR analysis shows that pre-mRNA was converted effectively into mRNA when empty control vector or wild type eIF2α vector was expressed, whereas co-expression of eIF2αS51A led to a strong decline in splicing efficiency. By contrast, as for splicing of TNF-α [10] and globin pre-mRNA [14], expression of eIF2αS51D, a phosphomimetic mutant of eIF2α that inhibits translation [39], did not significantly affect rev/tat mRNA splicing, indicating a requirement for authentic phosphorylated eIF2α (Fig. 3). Unlike eIF2αS51A, eIF2αS51D does not inhibit eIF2α phosphorylation by activated PKR [10].

Fig. 3figure 3

PKR-dependent rev/tat mRNA splicing requires eIF2α phosphorylation. Expression of non-phosphorylatable eIF2αS51A, yet not of phosphomimetic eIF2αS51D, inhibits splicing. BHK-21 cells were cotransfected with 1.5 µg of pcDNA-3’HIV DNA together with 1.5 µg DNA of pBS empty vector (EV), eIF2αS51A expression vector (S51A), vector expressing wild type eIF2α (eIF2αwt) or vector expressing eIF2αS51D (S51D). Total RNA was isolated at 20 h post-transfection. Spliced and unspliced rev/tat transcripts were determined by qRT-PCR. Splicing efficiency is expressed as mRNA/pre-mRNA ratio (error bars, SEM; n = 3). A representative experiment is shown

These results reveal an essential function for eIF2α phosphorylation in HIV rev/tat mRNA splicing that accounts for the need for PKR activation.

Nature of the RNA activator of PKR in HIV pre-mRNA

The 59-nucleotide TAR stem-loop is present at the 3′ end of the transcript encoded by the pcDNA-3’HIV vector (Fig. 2A). The 123-nucleotide 3’-terminal fragment of rev/tat pre-mRNA, including TAR (Fig. 4A), can activate recombinant human holo-PKR in vitro [18] and this activation was impaired severely by the TAR3R mutation [18] that replaces 4 nucleotides in the TAR stem, thereby abolishing base-pairing (Fig. 4A, B). Quantitation of phosphorylated recombinant PKR band intensity in Fig. 4B is presented in Fig. 4C. To examine whether TAR supports rev/tat mRNA splicing, we introduced the TAR3R mutation into pcDNA-3’HIV and analyzed the production of spliced and unspliced transcripts in transfected cells. Indeed, qRT-PCR analysis showed that the TAR3R mutation causes a significant reduction in rev/tat intron excision (Fig. 4D).

Fig. 4figure 4

3’-Proximal RNA pseudoknot is essential for activation of PKR and splicing. A Secondary structure of the 123-nucleotide region in the HIV-1 3’-UTR that contains the 3’-terminal TAR element and upstream pseudoknot. Pseudoknot stems P1 and P2 are indicated. Boxed nucleotides in P1 were mutated to the nucleotide sequence in the complementary strand (P1b, UUGCC > AGCGG; P1a, AGCGG > UUGCC). The 3R mutation in TAR was CUAG > UGGC; the 3 mutation in TAR was CUGG > GCCA [18]. B, C Both intact 3′-terminal TAR and pseudoknot stem P1 are required for PKR activation. Activation of PKR was assayed using rPKR (85 ng per lane) in the absence of RNA (-) or in the presence of wild type (wt), TAR3R or P1b mutant transcript at 0.1 μg/ml RNA. Position of phosphorylated rPKR (68 kDa) is indicated. A representative experiment is shown (B). Band intensity was quantitated and is plotted in bar graph (C), subtracting the value in the absence of RNA (error bars, SEM; n = 3). D Mutation of TAR or pseudoknot stem P1 impairs rev/tat splicing efficiency. BHK-21 cells were transfected with 3 µg of pcDNA-3’HIV DNA wt, TAR3R or P1b. Total RNA was isolated at 20 h and spliced and unspliced rev/tat transcripts were determined by qRT-PCR. Splicing efficiency, expressed as mRNA/pre-mRNA ratio, was determined for each DNA construct and corrected for between-session variation [58] (error bars, SD). E Mutation of each strand within pseudoknot stem P1 affects rev/tat splicing efficiency. BHK-21 cells were transfected with 3 µg of pcDNA-3’HIV DNA wild type (wt) or mutant forms P1b, P1a or TAR3R. Total RNA was isolated at the indicated times post-transfection. Spliced and unspliced rev/tat transcripts were determined by qRT-PCR. Splicing efficiency is expressed as mRNA/pre-mRNA ratio (error bars, SEM; n = 3). A representative experiment is shown. F Reduction in rev/tat intron splicing within the cell by TAR3 and TAR3R mutations and partial restoration by double mutation TAR3R3. BHK-21 cells were transfected with 3 µg of pcDNA-3’HIV DNA wild type (wt) or mutant forms P1b, TAR3, TAR3R or TAR3R3. Total RNA was isolated at 18 h post-transfection. Spliced and unspliced rev/tat transcripts were determined by qRT-PCR. Splicing efficiency is expressed as mRNA/pre-mRNA ratio (error bars, SEM; n = 3)

As noted above, the finding that production of mRNA by TAR-deficient HIV ∆TAR is sensitive to PKRi (Fig. 1C and Additional file 1: Fig. S1), supports the concept that the HIV genome contains an additional activator of PKR. As shown by mutational analysis, the activators of PKR in TNF-α pre-mRNA and IFN-γ mRNA fold into compact RNA pseudoknots [9,10,11]. Using bioinformatic analysis, we observed that the sequence just upstream of TAR in the 3′-terminal region of HIV-1 mRNA has the potential to fold into an RNA pseudoknot (Fig. 4A). This small pseudoknot had not been detected by chemical probing analysis of full-length HIV-1 RNA [40]. Yet, within the sequence just upstream of TAR, chemical probing supports two pseudoknot stems having moderate stability, consistent with dynamic refolding of the compact pseudoknot (Additional file 1: Fig. S2). Within the cell, HIV-1 RNA is highly dynamic in its folding, evident from the widespread heterogeneous nature of HIV-1 RNA structure conformation [41]. Notably, the short AGU linker connecting pseudoknot stems P1 and P2 is unreactive to chemical probing (Additional file 1: Fig. S2), indicative of structural constraint that may reflect the properties of the compactly folded TNF-α pseudoknot that functions as activator of PKR in splicing [10].

To evaluate whether the HIV pseudoknot might function as an activator of PKR in splicing, we examined PKR activation by a mutant 123-nucleotide 3’-terminal transcript in which formation of the putative pseudoknot RNA helix P1 is abrogated by base substitutions in the following strand (P1b) that replaces 5 nucleotides by the nucleotide sequence in the complementary strand, thereby abolishing base-pairing (Fig. 4A). As seen in Fig. 4B and C, not only TAR3R but also the P1b mutation impaired the ability to activate recombinant human PKR. PKR was phosphorylated far less when the RNA was mutated, reflecting strongly reduced PKR activation (Fig. 4B, C). Moreover, introduction of the P1b mutation into pcDNA-3’HIV strongly reduced the splicing efficiency of HIV rev/tat intron, even exceeding the reduction in splicing by the TAR3R mutation (Fig. 4D, E). The clear splicing phenotype of mutant P1b motivated mutation of the leading strand in helix P1, to yield mutant P1a that also replaces 5 nucleotides by the nucleotide sequence in the complementary strand, to abolish base-pairing (Fig. 4A). The P1a mutation likewise impaired splicing efficiency of rev/tat mRNA, though less severely than did P1b (Fig. 4E). These mutations validate the positive role of the pseudoknot in promoting splicing.

In view of the pronounced splicing phenotypes of mutants P1b and P1a (Fig. 4D, E), we created double mutant P1ab in which 5 base pairs in helix P1 are restored, albeit in the inverse orientation. In this regard, it should be noted that the pseudoknotted RNA activators of PKR in human IFN-γ mRNA [11] and TNF-α pre-mRNA [10] each lost the ability to activate PKR when only a single base pair was inverted within their RNA structure. However, upon transfection of the P1ab double mutant vector, in which 5 base pairs were inverted, no expression of pre-mRNA or mRNA could be detected using qRT-PCR (data not shown), documenting the sensitivity of this stem and precluding further analysis.

The extensively linear double-stranded RNA motif within the TAR element (Fig. 4A) allows the TAR3R mutation to be compensated by the complementary TAR3 mutation for activating PKR in vitro [18]. Indeed, although the TAR3R and TAR3 mutations each diminished rev/tat intron excision, albeit less severely than did pseudoknot mutation P1b, splicing within the cell could be restored in part by the double mutation TAR3R3 (Fig. 4F). These results demonstrate a collaborative role for the pseudoknot and the 3’-terminal TAR stem-loop in mediating PKR-regulated splicing of the rev/tat intron, the pseudoknot being dominant (Fig. 4D–F).

We next examined whether the two pseudoknot stems P1 and P2 are conserved among different HIV-1 strains and related simian immunodeficiency viruses (SIV). In both TNF-α pre-mRNA and IFN-γ mRNA, the pseudoknot elements show phylogenetic conservation [9,10,11]. Indeed, the potential to form the HIV pseudoknots is conserved broadly among isolates belonging to different HIV group M subtypes (Fig. 5). HIV-1 comprises three major groups: M, N, and O. Group M (Major) viruses causes more than 90% of all HIV/AIDS cases and is divided into subtypes A-K. Its zoonotic origin is SIV chimpanzee (SIVcpz). Group N stands for “Non-M and Non-O” and its occurrence is very rare. Until recently, only 20 group N cases have been recorded [

留言 (0)

沒有登入
gif