Poly(ADP-ribose) polymerase-1 is a negative regulator of HIV-1 transcription through competitive binding to TAR RNA with Tat.positive transcription elongation factor b (p-TEFb) complex.

Human immunodeficiency virus, type 1 (HIV-1) transcription is regulated by a virus-encoded protein, Tat, which forms a complex with a host cellular factor, positive transcription elongation factor b (P-TEFb). When this complex binds to TAR RNA synthesized from the HIV-1 long terminal repeat promoter element, transcription is trans-activated. In this study we showed that, in host cells, HIV-1 transcription is negatively regulated by competition of poly(ADP-ribose) polymerase-1 (PARP-1) with Tat.P-TEFb for binding to TAR RNA. PARP-1, which has a high affinity for TAR RNA (K(D) = 1.35 x 10(-10) M), binds to the loop region of TAR RNA and displaces Tat or Tat.P-TEFb from the RNA. In vitro transcription assays showed that this displacement leads to suppression of Tat-mediated trans-activation of transcription. Furthermore in vivo expression of luciferase or destabilized enhanced green fluorescent protein genes under the control of the HIV-1 long terminal repeat promoter was suppressed by PARP-1. Thus, these results suggest that PARP-1 acts as a negative regulator of HIV-1 transcription through competitive binding with Tat or the Tat.P-TEFb complex to TAR RNA.

Human immunodeficiency virus, type 1 (HIV-1) 1 transcription is regulated by a virus-encoded protein, Tat, through its binding to an RNA stem-loop, TAR RNA (for reviews, see Refs. 1 and 2). This 59-bp RNA stem-loop is transcribed from a promoter located within the long terminal repeat (LTR) element by RNA polymerase II (Pol II). The synthesis of TAR RNA, however, stalls Pol II, resulting in the production of non-processive RNA transcripts (1,2). To rescue the stalled Pol II, Tat binds to a uracil-rich bulge in TAR RNA (ϩ22 to ϩ24) (3)(4)(5)(6). This binding is stabilized by the interaction of Tat with a host cellular factor, positive transcription elongation factor (P-TEFb) (3,4). The cyclin T1 subunit of P-TEFb is involved in the interaction (7)(8)(9)(10), and the other P-TEFb subunit, CDK9, hyperphosphorylates the C-terminal domain of the stalled Pol II to trans-activate transcription (11,12).
As HIV-1 can establish a latent infection, HIV-1 transcription is also thought to be negatively regulated in host cells, although the molecular mechanisms of the negative regulation remain to be elucidated. However, several factors that might be involved in this negative regulation have been identified, one being 7SK RNA, an abundant small nuclear RNA (13,14), which, in association with P-TEFb (15,16), inhibits CDK9 activity (15,16). Thus, the 7SK RNA is likely to have a negative regulatory role in HIV-1 gene expression (15,16). Another identified factor is 5,6-dichloro-1-␤-D-ribofuranosylbenzimidazole sensitivity-inducing factor (DSIF), comprising the two subunits hSpt5 and hSpt4 (17). 5,6-dichloro-1-␤-D-ribofuranosylbenzimidazole sensitivity-inducing factor acts in concert with negative elongation factor (NELF) to suppress transcription at the level of elongation (18) and is suggested to be involved in suppression of HIV-1 transcription (19).
In addition to these factors, we have previously found evidence (20,21) suggesting that the abundant nuclear enzyme poly(ADP-ribose) polymerase-1 (PARP-1) might act as a negative regulator of HIV-1 transcription. PARP-1 is an enzyme known to bind to DNA breaks and to subsequently catalyze a post-translational modification of various proteins with ADP-ribose polymers (22)(23)(24). PARP-1 molecules have three major domains: DNA binding, automodification, and catalytic domains (22,25). The DNA binding domain, located at the N-terminal end of PARP-1, contains two homologous zinc finger motifs (25), which possess a high binding affinity for DNA breaks (26). We have previously reported that the zinc finger motifs also bind to various RNA stem-loops, including those in TAR RNA (21). Although the specific TAR RNA substructure recognized by PARP-1 is not yet known, it is plausible that, if the binding site of PARP-1 on TAR RNA is in close proximity to that of Tat⅐P-TEFb, PARP-1 may compete with Tat⅐P-TEFb in binding to TAR RNA, leading to HIV-1 transcription suppression.
To test this hypothesis, we studied the effects of the binding of PARP-1 to TAR RNA on Tat-mediated trans-activation of transcription. We showed that PARP-1 binds to the loop region of TAR RNA with high affinity (K D of 1.35 ϫ 10 Ϫ10 M), even higher than the binding affinity of Tat⅐P-TEFb for TAR RNA (10). Thus, PARP-1 is able to displace Tat⅐P-TEFb from TAR RNA. In addition, PARP-1 suppressed Tat-mediated trans-activation of transcription in cell-free transcription assays. Furthermore, using in vivo luciferase (Luc) reporter assays, we observed a similar suppression in Tat-mediated trans-activation of transcription by PARP-1. Using a stable cell line expressing Tat and destabilized enhanced green fluorescent protein (d1EGFP) under the control of HIV-1 LTR, we found that PARP-1 was able to suppress the expression of d1EGFP in vivo. Thus, our results suggest that PARP-1 is, to the best of our knowledge, the first identified host cellular factor that negatively regulates HIV-1 transcription by directly competing with Tat⅐P-TEFb for binding to TAR RNA.
Surface Plasmon Resonance Biosensor-The binding affinities of Tat and PARP-1 to TAR RNA were determined using BIACORE 3000. A biotinylated single-stranded oligodeoxynucleotide (ssBI; 100 ng/ml, 20 bases in length), dissolved in 25 mM Tris-HCl, pH 8.0, 50 mM NaCl, 1 mM MgCl 2 , 5.0 mM EDTA, and 0.01% Nonidet P-40 (BI buffer), was immobilized onto streptavidin-coated biosensor tips (SA5) at a flow rate of 30 ml/min. Then a 53-base-long single-stranded RNA (ssRNA) or TAR RNA was synthesized with a complementary sequence to the ssBI by T3 RNA polymerase (21), dissolved in BI buffer, and immobilized onto the sensor tips at a flow rate of 1 ml/min. Tat or PARP-1 was dissolved in BI buffer, and the interaction of Tat or PARP-1 with TAR RNA was measured by applying the protein or enzyme to sensor tips at a flow rate of 1 ml/min. Association and dissociation constants were determined with BIA Evaluation Version 3.1 (BIACORE).
Competition of PARP-1 with Tat or Tat⅐P-TEFb in TAR RNA Binding-TAR RNA⅐Tat complexes were preformed by incubation of 300 pmol of Tat, 50 fmol of [ 32 P]TAR RNA, and 0.5 g of anti-Tat antibody (NT3 2D1.1 (29)) at 30°C for 20 min in a 15-l reaction containing 25 mM Tris-HCl, pH 8.0, 50 mM NaCl, 1 mM MgCl 2 , and 0.5 mM dithiothreitol (C buffer) and immunoprecipitated according to the method of Dingwall et al. (29) with minor modifications. Briefly 5 l of protein G-Sepharose was added, and the mixture was incubated at 4°C for 1 h. The protein G-Sepharose was then washed twice with 10 l of C buffer. Then 14 l of C buffer containing 0.2 mM ZnCl 2 was incubated with the protein G-Sepharose in the presence or absence of P-TEFb (20 ng), which was prepared as described previously (30), for 10 min at 30°C. Next PARP-1 was added and incubated at 30°C for 20 min. The protein G-Sepharose was centrifuged, and the 32 P activity remaining on and released from the protein G-Sepharose was measured by scintillation counting.
Cell-free Transcription Assay-pHIV-1/LTR/Luc (31, 32) and pCMV/ Luc, containing the Luc gene under the control of the HIV-1 LTR or CMV promoters, respectively, were linearized by EcoRI. These linear-ized templates were used in a run-off transcription assay with HeLa nuclear extracts (Promega). Reactions were carried out for 1 h at 30°C using 4 g/ml linearized pHIV-1/LTR/Luc or pCMV/Luc and 1000 transcription units of HeLa nuclear extract as described previously (21). The expected sizes of the run-off products from linearized pHIV-1/LTR/Luc and pCMV/Luc were 674 and 680 bases, respectively. After termination of the reaction and purification of transcripts, RNA was fractionated on a 5 or 10% polyacrylamide, 8 M urea gel. After gel drying, 32 P activity was visualized by autoradiography.
Northern Blot Analysis-Total RNA was extracted with a Promega RNA extraction kit. 32 P-Labeled probes for Luc mRNA were prepared from pHIV-1/LTR/Luc, and Northern blot analysis was carried out using a standard method.
RNA Protection Assay-pHIV-1/LTR/Luc was digested with XbaI, and the fragment containing the sequence for TAR RNA and the following 5Ј portion of the Luc gene were cloned into pBluescript K/S ϩ . A 32 P-labeled antisense probe complementary to the sequence from ϩ1 to ϩ156 was synthesized by T3 RNA polymerase. An RNA protection assay was carried out using the RPA III kit (Ambion). Protected fragments were fractionated on a 10% polyacrylamide, 8 M urea gel and visualized by autoradiography.
Establishment of Cells Constitutively Expressing d1EGFP-pd1EGFP-N1, containing d1EGFP under the control of a CMV promoter, was purchased from Clontech. The CMV promoter was then replaced with the HIV-1 LTR to construct the pd1EGFP-LTR. HeLatat-III cells were transfected with pd1EGFP-N1 or pd1EGFP-LTR to create HeLa-tat-III/CMV/d1EGFP and HeLa-tat-III/LTR/d1EGFP cells, respectively. The cells were grown in Dulbecco's modified Eagle's medium with high glucose containing 10% fetal bovine serum, antibiotics, and 1 mg/ml G418; cells were cloned from the resulting colonies. Expression of d1EGFP was confirmed by fluorescence microscopy.
Transfection of pPARP-1-DsRed and Confocal Microscopy-HeLatat-III/CMV/d1EGFP or HeLa-tat-III/LTR/d1EGFP cells were seeded in Delta T4 culture dishes (Bioptechs) in Dulbecco's modified Eagle's medium with high glucose containing 10% fetal bovine serum and antibiotics. Then either pPARP-1-DsRed (1.5 g/ml), a mammalian expression construct for PARP-1 tagged with red fluorescent protein (DsRed), or p24kDa-DsRed, a mammalian expression construct for the N-terminal 24-kDa PARP-1 fragment tagged with DsRed (35), was transfected by the standard calcium phosphate method. The cells were grown for another 48 h. Images taken through a 60ϫ 1.4 numerical aperture objective were captured by a FLUOVIEW FV300 confocal laser scanning unit (Olympus). For multiple channel imaging, fluorescence from each channel was imaged sequentially to eliminate cross-talk between the channels. A 488 nm argon ion laser line was used to excite d1EGFP with fluorescence being imaged using a 570 nm beamsplitter and the combination of a 510 -530 nm bandpass excitation filter and 510 nm long pass emission filter. DsRed was exited with a 543 nm helium-neon laser line, and the fluorescence was imaged using a 570 nm beamsplitter and a 575-630 nm bandpass emission filter. In this configuration, cross-talk between d1EGFP and DsRed was kept below detectable limits (35). Green and red colored images were created by FLUOVIEW 300 Version 3.3 software (Olympus). The expression level of d1EGFP was determined using Photoshop Version 6.0 software.

RESULTS
PARP-1 and RNA-PARP-1 has the ability to bind various types of RNA stem-loops, including TAR RNA (20,21), although the recognized site of this RNA stem-loop is not yet known. Thus, in this study, we first determined the binding site of PARP-1 on TAR RNA using gel retardation assays. Consistent with our previous observations (20, 21), PARP-1 reduced the mobility of TAR RNA on a native polyacrylamide Negative Regulation of HIV-1 Transcription by PARP-1 gel and formed a single retarding band (Fig. 1A, TAR RNA). Even when the loop sequence of TAR RNA was altered (M-TAR RNA), a single retarding band was still found (Fig. 1A, M-TAR RNA), suggesting that PARP-1 recognizes the stem-loop structure itself and not the loop sequence. To determine whether PARP-1 binds to the loop or stem region, we then prepared an RNA stem-loop that did not contain mismatches in the stem region ( Fig. 1B, Stem-loop). Then we carried out gel retardation assays with stem-loop RNA that was or was not treated with RNase T1, which digests ssRNA (Fig. 1B, Stem). As shown in Fig. 1B (Stem-loop), PARP-1 formed a single retarding band with the stem-loop RNA, while no obvious retardation was observed with the digested stem-loop (Fig. 1B, Stem). These results suggest that PARP-1 binds to the loop region of RNA stem-loops. Previously we demonstrated that PARP-1 forms two discrete retarding bands with RNA synthesized from the cystic fibrosis transmembrane conductance regulator gene (20). This RNA contains two stem-loops, and thus the formation of two retarding bands is likely related to the number of stemloops presented in the RNA. Thus, we then tested longer M-TAR RNA with an extra RNA stretch of 68 bases (M-TAR RNAϩ68). Based on the structure prediction using RNA Structure Version 3.1 (Scripps Institute), we found 10 putative ssRNA regions in M-TAR RNAϩ68. As described, only one retarding band was formed by incubating M-TAR RNA with PARP-1 (Fig. 1A), while the mobility of M-TAR RNAϩ68 further decreased with increasing amounts of PARP-1 (Fig. 1C). This retardation can be explained by the presence of multiple PARP-1 binding sites in M-TAR RNAϩ68. As PARP-1 also reduced the mobility of ssRNAs, including poly(A), poly(C), poly(U), and poly (G) (data not shown), these results suggest that PARP-1 has an affinity for ssRNA and thus recognizes loop regions of RNA stem-loops.
Binding Affinity of PARP-1 for TAR RNA-Next we deter-mined the equilibrium dissociation constant, K D , between PARP-1 and TAR RNA using a surface plasmon resonance biosensor. As a control, the K D between Tat and TAR RNA was also determined. Previously, using fluorescence resonance energy transfer, Zhang et al. (10) showed that the K D between Tat and TAR RNA is 8.2 ϫ 10 Ϫ9 M. The K D that we obtained using a surface plasmon resonance biosensor between Tat and TAR RNA was 3.01 ϫ 10 Ϫ9 M ( Fig. 2A). Thus, although our method showed a 2.7-fold lower K D than that obtained by fluorescence resonance energy transfer, our results are, overall, in agreement with the results obtained by Zhang et al. (10). We then determined the K D between PARP-1 and TAR RNA, which was 1.35 ϫ 10 Ϫ10 M (Fig. 2B), showing that PARP-1 has an affinity for TAR RNA 22 times stronger than Tat. Zhang et al. (10) reported that the affinity of Tat for TAR RNA increases about 10-fold after forming the Tat⅐P-TEFb complex. Even when taking this increase into account, PARP-1 still has a 2.2 times higher affinity for TAR RNA than does Tat⅐P-TEFb. Competition between PARP-1 and Tat or Tat⅐P-TEFb in Binding to TAR RNA-Binding of Tat⅐P-TEFb to TAR RNA primarily requires Tat recognition of a TAR RNA bulge region. Given that the bulge is located near the loop region (Fig. 1A), we next examined whether PARP-1 binding to TAR RNA inhibits Tat or Tat⅐P-TEFb binding to TAR RNA. As illustrated in Fig. 3A, [ 32 P]TAR RNA was incubated with Tat in the presence of anti-Tat antibody (29). Then the anti-Tat antibody⅐ Tat⅐[ 32 P]TAR RNA complexes were precipitated with protein G-Sepharose. PARP-1 was added to the complexes under the presumption that if PARP-1 displaces Tat from the complexes, [ 32 P]TAR RNA might be released. This predicted release of [ 32 P]TAR RNA by PARP-1 (Fig. 3A) did in fact occur, suggesting that PARP-1 was able to displace Tat from TAR RNA. We then added P-TEFb to anti-Tat antibody⅐Tat⅐[ 32 P]TAR RNA complexes to allow the formation of Tat⅐P-TEFb complexes and studied the effect of PARP-1 on the release of [ 32 P]TAR RNA from these complexes. P-TEFb was previously reported to stabilize Tat⅐TAR RNA complexes by increasing the binding affinity of Tat by about 10-fold (10). Accordingly the addition of 0.8 pmol of PARP-1, which released more than 60% of [ 32 P]TAR RNA from the anti-Tat antibody⅐Tat⅐[ 32 P]TAR RNA complexes (Fig. 3B, Tat, 0 versus 0.8 pmol of PARP-1) only released 10% of the [ 32 P]TAR RNA when the complexes were incubated with P-TEFb (Fig. 3B, TatϩP-TEFb, 0 versus 0.8 pmol of PARP-1).
Suppression of Tat-promoted trans-Activation of Transcription by PARP-1 in Vitro-In cell-free transcription assays with HeLa nuclear extracts, the addition of Tat has been shown to have a promoting effect on transcription initiated from the HIV-1 LTR because of the formation of Tat⅐P-TEFb and trans- activation of transcription through its binding to TAR RNA (36). In fact, in a run-off transcription assay with a DNA template containing the HIV-1 LTR, RNA synthesis was significantly promoted by addition of Tat to the HeLa nuclear extracts (Fig. 4A, lane 1 versus lanes 5-7). We found that PARP-1 suppressed this promoted transcription activity (Fig.  4A, lanes 8 -10). To study whether the suppression indeed occurred because of synthesis of TAR RNA from HIV-1 LTR, we then used a template DNA containing a CMV promoter instead of the HIV-1 LTR. We previously reported that PARP-1 reduces the rate of RNA synthesis (20,21), and addition of PARP-1 to transcription assays with the DNA template containing CMV promoter also resulted in the suppression of RNA synthesis (Fig. 4A, lane 11 versus lanes 12-14). However, the extent of this suppression was less significant compared with the effect of PARP-1 on the transcription activity promoted by Tat (Fig.  4A, lanes 8 -10 versus lanes 12-14). Thus, PARP-1 more significantly affected suppression of transcription mediated by Tat binding to TAR RNA than the CMV promoter-driven transcription in the assay.
If this suppression is in some way caused by inhibition of Tat⅐P-TEFb binding to TAR RNA by PARP-1, accumulation of TAR RNA would be expected. Thus, we analyzed RNA transcripts similar in length to TAR RNA (50 -80 bases in length). As shown in Fig. 4B, RNA showing a similar mobility to 32 P-labeled transcripts containing TAR RNA (76 bases in length, Fig. 4B, lane 1) were found after transcription reactions with a template DNA containing HIV-1 LTR but in the absence of Tat (Fig. 4B, lane 2). Thus, these results suggest that TAR RNA accumulates during the reaction. Adding Tat to the cell-free transcription assays reduced the amount of TAR RNA found in the reactions because of the trans-activation of transcription (Fig. 4B, lane 3). When PARP-1 was added to the reaction containing Tat, TAR RNA again accumulated (Fig. 4B, lanes  4 -6). These results indicate that trans-activation of transcription is suppressed by PARP-1 and that this suppression results from inhibition of the binding of Tat⅐P-TEFb to TAR RNA.
It has been proposed that phosphorylation of the hSpt5 subunit of 5,6-dichloro-1-␤-D-ribofuranosylbenzimidazole sensitivityinducing factor by CDK9 occurs during Tat-mediated trans-activation of transcription (19). If PARP-1 competes with Tat⅐P-TEFb in binding to TAR RNA, suppression of hSpt5 phosphorylation would also be expected. To further confirm whether PARP-1 suppresses Tat-mediated trans-activation of transcription, we decided to study the effect of PARP-1 on hSpt5 phosphorylation using Western blot. As a marker of both non-phosphorylated and phosphorylated hSpt5, we prepared whole cell extracts from HeLa S3 cells according to published methods (17), and two bands were visualized (Fig. 4C, lane 1). Nuclear extracts (Promega) used for the transcription assay contained only non-  -1 in vitro. A, a run-off transcription assay was carried out with a template DNA containing the HIV-1 LTR together with HeLa nuclear extracts (Promega). 32 P-Labeled transcripts were analyzed on a 5% polyacrylamide, 8 M urea gel and visualized by autoradiography. Tat and/or PARP-1 was also added to the reactions. Quantified data are shown. B, in vitro transcripts were fractionated on a 10% polyacrylamide, 8 M urea gel to analyze RNA of 50 -80 bases in length, which would encompass TAR RNA. [ 32 P]TAR RNA was loaded as a marker. C, after a run-off transcription reaction, the mixture was fractionated on a 7.5% SDS-polyacrylamide gel to separate the phosphorylated and nonphosphorylated forms of hSpt5, which were detected by Western blotting with anti-hSpt5 antibody. A HeLa S3 extract prepared by the method of Lavoie et al. (49) was used as marker of non-phosphorylated and phosphorylated hSpt5. phosphorylated hSpt5 (Fig. 4C, lane 2). Adding Tat to the transcription assay resulted in phosphorylated hSpt5 (Fig. 4C,  lane 3). Then we added both PARP-1 and Tat in the transcription reactions and found that PARP-1 reduced the formation of phosphorylated hSpt 5 (Fig. 4C, lanes 5-7). Thus, these results suggest that binding of Tat⅐P-TEFb to TAR RNA is inhibited by PARP-1.
Luc Reporter Assay-We then studied whether PARP-1 suppresses Tat-mediated trans-activation of transcription in vivo using a Luc reporter assay. pHIV-1/LTR/Luc, containing a Luc gene downstream of the HIV-1 LTR, was transfected into HeLa-tat-III cells, which constitutively express Tat (33,34), or HeLa S3 cells, and Luc activity was measured. In these cells, Luc activity was over 150-fold higher in the HeLa-tat-III cells compared with the HeLa S3 cells (Fig. 5A, HeLa S3,
We then measured the amount of Luc mRNA in HeLa-tat-III cells. As shown in Fig. 5B, p31 transfection reduced the amount of Luc mRNA in HeLa-tat-III cells, indicating that reduced Luc activity by overexpression of PARP-1 resulted from suppression of Luc mRNA transcription initiated from HIV-1 LTR.
Effect of PARP-1 Overexpression on Transcription from HIV-1 LTR and CMV Promoters-We have previously reported that PARP-1 reduces the rate of RNA synthesis by Pol II at the level of elongation (20,21). Thus, PARP-1 might suppress transcription of Luc gene under the control of HIV-1 LTR (Fig. 5B) by inhibiting Tat⅐P-TEFb binding to TAR RNA and/or by reducing the overall mRNA synthesis rate. To further elucidate these mechanisms, we utilized a CMV promoter in addition to the HIV-1 LTR. As shown in Fig. 6A, the Luc activity found in cells transfected with pCMV/Luc was reduced by PARP-1 overexpression. Thus, it appears that this reduction could be caused by a decreased rate of mRNA synthesis. However, a more prominent effect in reducing Luc activity was observed when pHIV-1/LTR/Luc was used. Furthermore similar results were obtained by expression of the 24-kDa PARP-1 fragment instead of PARP-1 (Fig. 6B). Thus, while PARP-1 can suppress HIV-1 gene expression by reducing the overall rate of mRNA synthesis, these results suggest that formation of TAR RNA leads to prominent suppression of transcription initiated from HIV-1 LTR.
Accumulation of TAR RNA in Vivo-In vitro, we found that addition of PARP-1 to the cell-free transcription reaction caused TAR RNA to accumulate by inhibiting Tat-mediated trans-activation of transcription (Fig. 4B). If, in vivo, PARP-1

FIG. 5. Suppression of transcription initiated from HIV-1 LTR by PARP-1 in vivo.
A, PARP-1 expression construct (p31) and/or pHIV-1/LTR/Luc containing the Luc gene under the control of HIV-1 LTR were transfected into HeLa S3 or HeLa-tat-III cells, and Luc activity was measured. As a control, pcDNA3.1/GFP (containing a non-related cDNA under the control of a CMV promoter), pcDNA3.1/89 kDa (coding for the 89-kDa PARP-1 fragment, p89 kDa), or pcDNA3.1/24 kDa (coding for the 24-kDa PARP-1 fragment, p24 kDa) was used instead of p31. pHIV-1/LTR ⌬TAR/Luc was also used instead of pHIV-1/LTR/Luc. Standard deviations from three independent experiments are shown. B, total RNA was extracted after the transfection, and a Northern blot was carried out with 32 P-labeled probes for Luc mRNA. Recovery of rRNA is also shown.
indeed suppresses mRNA synthesis initiated from the HIV-1 LTR by inhibiting trans-activation, TAR RNA would be expected to accumulate in cells that overexpress PARP-1. We thus used an RNA protection assay to test this in vivo. We designed a probe that has a sequence complementary to the LTR-Luc mRNA from ϩ1 to ϩ156 that covers the R region coding for TAR RNA and part of a Luc coding sequence as illustrated in Fig. 7. Thus, mRNA transcribed through the R region can be detected as a 156-base-long protected fragment. In fact, the protected fragments were found in RNA extracted from HeLa-tat-III cells transfected with the pHIV/LTR/Luc (Fig. 7, 156 bases, lane 3), while these fragments were not found in HeLa S3 cells (Fig. 7, 156 bases, lane 2) because of transcription terminating the R region prematurely, as reported previously (37). The amount of protected 156-base-long fragments was also reduced by overexpression of PARP-1 in Hela-tat-III cells that were transfected with pHIV-1/LTR/Luc (Fig. 7, 156 bases, lane 3 versus lane 4), suggesting that RNA synthesis is terminated by PARP-1 at an upstream location of ϩ156. We then analyzed RNA fragments that have a length corresponding to TAR RNA. In HeLa S3 cells transfected with pHIV-1/LTR/Luc, an accumulation of TAR RNA (or its digested products (37)) was observed (Fig. 7, TAR RNA, lane 2), while in HeLa-tat-III cells, the amount of fragments was reduced due to Tat-mediated trans-activation of transcription (Fig. 7, TAR RNA, lane 3) as reported by Laspia et al. (37). Interestingly overexpression of PARP-1 in HeLa-tat-III cells transfected with pHIV-1/LTR/Luc led to the accumulation of these fragments (Fig. 7, TAR RNA, lane 4), suggesting that Tat-mediated trans-activation of transcription is inhibited by PARP-1.
Inhibition of Tat-mediated trans-Activation of Transcription by PARP-1 in Live Cells-We then investigated the effect of PARP-1 expression on Tat-mediated trans-activation of transcription in live cells. For this purpose, we established HeLatat-III cell lines stably expressing d1EGFP under the control of either the CMV promoter (HeLa-tat-III/CMV/d1EGFP cells) or HIV-1 LTR (HeLa-tat-III/LTR/d1EGFP cells). Then an expression construct for PARP-1 tagged with DsRed was created (pPARP-1-DsRed), and both d1EGFP and DsRed emissions from these cells were captured after sequential excitations of these fluorescent proteins by laser as described previously (35). As shown in Fig. 8A (HeLa-tat-III/CMV/d1EGFP), d1EGFP was expressed in HeLa-tat-III/CMV/d1EGFP cells, and PARP-1-DsRed-expressing cells could be identified by monitoring fluorescence emission from DsRed. The level of d1EGFP in over 100 cells was then quantified to determine the effect of PARP-1-DsRed expression on d1EGFP transcription. On the basis of this quantitation, we found that the average intensity of d1EGFP in non-PARP-1-DsRed-expressing cells (Fig. 8B, Control) was 61 arbitrary units, while the intensity was reduced to 39 arbitrary units by PARP-1-DsRed expression (Fig. 8B, PARP-1-DsRed expressed). Thus, consistent with the results obtained from the Luc reporter assay (Fig. 6A), PARP-1 expression leads to an inhibition of transcription initiated by the CMV promotor. Then when pPARP-1-DsRed was transfected into HeLa-tat-III/LTR/d1EGFP cells, we found a prominent suppression of d1EGFP expression by PARP-1-DsRed (Fig. 8A, HeLa-tat-III/LTR/d1EGFP). As show in Fig. 8C, the average intensity of d1EGFP was in fact reduced from 75 arbitrary units (Control) to 9 arbitrary units (PARP-1-DsRed expressed) by PARP-1-DsRed expression. Furthermore similar results were obtained by using an expression construct of the 24-kDa PARP-1 fragment tagged with DsRed (p24 kDa-DsRed) instead of pPARP-1-DsRed (Fig. 8, D, E, and F). Thus, taken together, these results consistently suggest that PARP-1 suppresses Tatmediated trans-activation of transcription and that its zinc finger motifs located within the 24-kDa fragment play a role in the suppression. DISCUSSION Possible links between PARP-1 and HIV-1 have been reported previously. In 1991, for example, Yamagoe et al. (38) observed a promotion of UV irradiation-induced HIV-1 gene expression by PARP-1 inhibitors, although it is not yet known whether PARP-1 per se is indeed involved in the promotion. More recently, Ha et al. (39) reported a role for PARP-1 in HIV-1 integration, although Siva and Bushman (40) later showed that PARP-1 is not strictly required for integration as they observed efficient HIV-1 integration in cells lacking PARP-1. Thus, the role of PARP-1 in HIV-1 infection has remained elusive. In this report, we show that PARP-1 is a host cellular factor that negatively regulates HIV-1 transcription by directly competing with Tat⅐P-TEFb for binding to TAR RNA.
Tat⅐P-TEFb-mediated Transcription Regulation and PARP-1-In infected cells, while HIV-1 actively transcribes its genes during the lytic phase, this transcription is suppressed in the latent phase. Thus, HIV-1 transcription seems to be controlled by opposing positive and negative regulators that control the existence of the virus in host cells. During the lytic phase, the effect of a negative regulator, PARP-1, on HIV-1 transcription can therefore be overcome by a positive regulator, Tat⅐P-TEFb. How HIV-1 overcomes PARP-1-mediated suppression of tran-scription can be explained by comparing the binding affinities of PARP-1 and Tat⅐P-TEFb for TAR RNA. As shown in Fig. 2, PARP-1 has a K D for TAR RNA in the order of 10 Ϫ10 M, which enables the formation of a highly stable complex with TAR RNA. Thus, factors that have significantly less affinity for TAR RNA may not be able to compete with PARP-1 for binding to TAR RNA. In fact, the viral protein Tat, having a 22-fold lower FIG. 8. Suppression of Tat-mediated trans-activation of transcription by PARP-1 in living cells. A, pPARP-1-DsRed was transfected into either HeLa-tat-III/CMV/d1EGFP or HeLa-tat-III/LTR/d1EGFP cells stably expressing d1EGFP under the control of the CMV promoter or HIV-1 LTR, respectively. Cells were mounted onto a microscope stage 24 h after transfection. d1EGFP and DsRed were sequentially excited by helium-neon and argon lasers, respectively, using a confocal laser scanning unit. Fluorescence emissions from d1EGFP and DsRed were captured sequentially to eliminate cross-talk. Images were prepared from one optical section. B and C, the level of d1EGFP fluorescence in PARP-1-DsRednon-expressing and PARP-1-DsRed-expressing HeLa-tat-III/CMV/d1EGFP (B) or HeLa-tat/LTR/d1EGFP (C) cells was determined. Results obtained from over 100 cells are shown. D, p24 kDa-DsRed was used instead of pPARP-1-DsRed. E and F, quantified results obtained from HeLa-tat-III/CMV/d1EGFP (E) or HeLa-tat-III/LTR/d1EGFP (F) cells are shown. affinity for TAR RNA than PARP-1 does, was displaced from TAR RNA by PARP-1 (Fig. 3A). Therefore, without mechanisms to promote the binding affinity of Tat for TAR RNA, HIV-1 may not be able to compete with PARP-1 in binding to TAR RNA. Interestingly, by forming a complex with P-TEFb, the affinity of Tat for TAR RNA increases 10-fold (10) and becomes only a fewfold less than that of PARP-1. In fact, Tat⅐P-TEFb was more resistant to displacement from TAR RNA by PARP-1 than Tat alone due to the stabilization of Tat⅐TAR RNA complexes (Fig. 3B). Because of this stabilization by P-TEFb, Tat possibly becomes capable of competing with PARP-1 in binding to TAR RNA. Although one of the major functions of P-TEFb on HIV-1 gene expression is to transactivate HIV-1 transcription by phosphorylating hSpt5 and the C-terminal domain of Pol II (11,12,19), the stabilization of Tat⅐TAR RNA complexes per se, through promotion of Tat binding affinity for TAR RNA, must therefore be another critical role of P-TEFb in overcoming the negative regulatory effect of PARP-1 on HIV-1 transcription.
If HIV-1 transcription is regulated by the relative levels of negative and positive regulators, any factor capable of reversing favorable situations for HIV-1 transcription could lead to the suppression of HIV-1 transcription. For example, P-TEFb sequestration by 7SK RNA (15,16) may increase the probability of PARP-1 binding to TAR RNA, leading to the suppression of HIV-1 transcription. Alternatively if greater amounts of PARP-1 become available to bind TAR RNA, HIV-1 transcription could also be suppressed. In fact, as shown in Fig. 3B, Tat⅐P-TEFb complexes, even if resistant to displacement from TAR RNA by 0.8 pmol of PARP-1, were eventually removed from the stem-loop by adding greater amounts of PARP-1 (Fig.  3B). Furthermore we observed reduced expression of Luc and d1EGFP genes under HIV-1 LTR control when PARP-1 was expressed (Figs. 5, 6, and 8). HIV-1 establishes latent infections in CD4 ϩ memory T cells capable of survival for a significantly long period of time, and latent HIV-1 infections are often found in cell populations that survived antiviral therapy (41). It is possible that favorable conditions for the suppression of HIV-1 transcription are established in these cells, and PARP-1 may only be one of many factors involved in the establishment of these conditions. PARP-1 and HIV-1 Transcription-The function of PARP-1 to suppress HIV-1 transcription through TAR RNA binding is apparently related to its role in the regulation of host cellular genes. We have previously shown that PARP-1 has the ability to regulate RNA synthesis by interacting with topoisomerase I (35), an enzyme required for the progression of Pol II during transcription (42,43), by regulating DNA superhelical tension with its topoisomerase 1-like activity (27) and by binding to nascent RNA (20,21). In particular, binding of PARP-1 to nascent RNA reduced the rate of RNA synthesis, and therefore it was proposed that PARP-1 could be a novel type of negative elongation regulator (20,21). Suppression of HIV-1 transcription through binding of PARP-1 to TAR RNA is thus likely to be related to the function of PARP-1 as a negative elongation regulator.
In addition to the function of PARP-1 as a negative transcription elongation regulator, PARP-1 may also be involved in HIV-1 transcription at the level of initiation as it interacts with NF-B (44,45), a factor regulating the initiation of HIV-1 transcription (46). In fact, Hassa et al. (44) suggested that NF-B promoted transcription by interacting with PARP-1. Furthermore Kameoka et al. (47) recently reported an inactivation of HIV-1 LTR in cells treated with small interfering RNA directed against PARP-1, suggesting that PARP-1 is required for HIV-1 transcription initiation. However, contrary to the report from Hassa et al. (44), Chang et al. (45) demonstrated that NF-B is unable to bind to its target sequence when NF-B is interacting with PARP-1, which suggests that PARP-1 negatively regulates NF-B-dependent transcription initiation. Of note, although the reason for this discrepancy is not clear, Gwack et al. (48) reported a similar negative regulation of transcription where PARP-1 forms a complex involving a transcription activator of ␥-2 herpesvirus. Although the role of PARP-1 in HIV-1 transcription initiation through its interaction with NF-B may need clarification, PARP-1 appears to be involved in regulation of HIV-1 transcription at both the levels of initiation and elongation.
Conclusion-HIV-1 transcription is regulated at the level of elongation via Tat and TAR RNA (1, 2). As reported here, Tat⅐P-TEFb plays a critical role as a positive regulator, and PARP-1 acts as a negative regulator of HIV-1 gene expression. Because these regulations occur through binding of PARP-1 or Tat⅐P-TEFb to TAR RNA, this unique stem-loop RNA plays a central role in both positive and negative regulation of HIV-1 transcription.
In infected cells, positive regulation of HIV-1 transcription leads to the lytic phase, while HIV-1 can enter the latent phase by suppressing its transcription. On the other hand, HIV-1 in latently infected cells can be reactivated by various viral stimuli. Interestingly P-TEFb and PARP-1 activities are controlled in cells by exposure to viral stimuli, including UV and reactive oxygen species. For example, P-TEFb, which is sequestered by 7SK RNA, is released upon cell exposure to UV (15,16), and PARP-1 is converted to its inactive form (automodified PARP-1) by cell exposure to reactive oxygen species (22,24). The P-TEFb release from 7SK RNA and PARP-1 inactivation primarily occur in host cells to protect cells from stresses induced by reactive oxygen species or UV. HIV-1 might perhaps monitor the host cellular response to these stresses as well as the relative activities of P-TEFb and PARP-1 as a measure to control its own life cycle and may use TAR RNA as a sensor to detect these activities by creating a binding site for both positive and negative regulators.