Novel poly (ADP-ribose) polymerase 1 binding motif in hepatitis B virus core promoter impairs DNA damage repair


  • Potential conflict of interest: Nothing to report.

  • This work was supported by funding from A*STAR.


It is well-established that hepatitis B virus (HBV) infection is associated with the development of hepatocellular carcinoma (HCC), but patients with high viral DNA load have significantly higher risk. As host factors are required for efficient viral replication and may, therefore, contribute to high viral DNA load, we screened for host factors that can transcriptionally activate the HBV core promoter (HBVCP). We report here that poly (ADP-ribose) polymerase 1 (PARP1), which is known for its DNA repair activity, binds prominently to an octamer motif in the HBVCP and increases transcriptional efficiency. By utilizing a series of single base substitutions at each nucleotide position of the octamer, the PARP1 binding motif can be defined as “RNNWCAAA.” Intriguingly, introduction of a vector construct bearing tandem repeats of the octamer motif was able to impair the DNA repair function of PARP1. This finding suggests that HBV viral DNA contains specific sequence motifs that may play a role in disrupting the DNA repair pathways of infected hepatocytes. Conclusion: This study has identified a novel octamer motif in the HBVCP that binds PARP1, and this interaction increases the replication efficiency of HBV. The presence of this octamer motif in hepatocytes was shown to inhibit the DNA repair capacity of PARP1, potentially contributing to the development of HCC. (HEPATOLOGY 2011;)

Chronic infection with the hepatitis B virus (HBV) contributes to more than half the world's cases of hepatocellular carcinoma (HCC).1 Several mechanisms have been proposed to account for HBV-associated HCC, including chronic inflammation and constant liver regeneration, oncogenic effects of viral proteins, such as hepatitis B virus X (HBx) and truncated pre-S2/S, as well as insertional mutagenesis of HBV genomes.2-4 However, occult HBV infection, characterized by the presence of HBV genomes in the absence of hepatitis B virus surface antigen (HBs) expression, can also lead to the development of HCC when chronic liver inflammation and viral DNA integration is minimal.5, 6 Furthermore, covalently closed circular DNA (cccDNA), a persistent replicative intermediate required for HBV replication, is found in higher quantities in tumor tissues of HCC patients, when compared to nontumor tissues.7 Moreover, high HBV DNA load is a strong risk factor for the development of HCC.8-11 These suggest the possibility that HBV DNA itself may actively contribute to HCC development.

Chronic infection with HBV also leads to accumulation of genotoxic lesions, such as oxidative DNA damage and DNA strand breaks.12 Many of these DNA lesions are repaired via pathways involving the poly (ADP-ribose) polymerase 1 (PARP1),13, 14 where recognition of DNA strand breaks trigger its enzymatic activation, adding poly-ADP ribose (PAR) to protein acceptors required for the recruitment of DNA repair enzymes.15, 16 Consistent with dependence on PAR for DNA repair, loss of PARP1 expression or enzymatic activity results in hypersensitivity to DNA damage inducers17, 18 and spontaneous development of HCC.19-21 Interestingly, inhibition of PARP1 enzymatic activity has also been reported to increase HBV DNA integration,22 contributing further to the risk of developing HCC.

Because high HBV DNA load leads to increased chance of HCC development, which is, in turn, associated with impaired DNA repair, we investigated whether the hepatitis B virus core promoter (HBVCP)-host interaction that regulates HBV genomic replication23, 24 can alter the properties and function of nuclear proteins involved in the DNA repair pathways. Using a series of deletion mutants along the HBVCP to map host factor binding sites, PARP1 was uncovered to bind in a sequence-specific manner, exerting transcriptional activation effects to regulate HBV replication. Furthermore, by binding its recognition motif, its enzymatic activity was reduced, compromising cellular DNA repair.


bp, base pair; cccDNA, covalently closed circular DNA; DMSO, dimethyl sulfoxide; EMSA, electrophoretic mobility shift assay; HBs, hepatitis B virus surface antigen; HBV, hepatitis B virus; HBVCP, hepatitis B virus core promoter; HBx, hepatitis B virus X; HCC, hepatocellular carcinoma; HTLV Tax RE, human T-cell leukemia virus Tax responsive element; MALDI-TOF/TOF, matrix-assisted laser desorption/ionization time of flight; PAR, poly (ADP-ribose); PARP1, poly (ADP-ribose) polymerase 1; pgRNA, pregenomic RNA; RFP, red fluorescent protein; SE, standard error; si, short interfering RNA.

Patients and Methods

Cell Culture.

HepG2 was maintained at 37°C and 5% (v/v) CO2 in a humidified incubator with complete Dulbecco's modified Eagle's medium, supplemented with 10% (v/v) fetal bovine serum without antibiotics.

Plasmids and Short Interfering RNA.

HBVCP deletions were generated using a multistep strategy (Supporting Fig. 1A). Wild-type HBVCP (nt 1600-1860, genotype A) was inserted via KpnI and HindIII restriction sites into the PGL3 basic vector (Promega, Madison, WI). Using this as a template, sequences flanking either ends of the deletion (denoted “X”) were generated with primers PGLF and BX or primers PGLR and CX (Supporting Fig. 2B). Resultant products were annealed by complementary base pairing, amplified with PGLF and PGLR, and then inserted into PGL3 basic vector via KpnI and HindIII restriction sites. Single base substitutions were generated with the QuickChange® II Site-Directed Mutagenesis Kit (Stratagene, Santa Clara, CA). Full-length replicative HBV genotype A (1.1X HBV genome, nt 1535-1937) was amplified from another construct25, 26 using the primers HBVF and HBVR, inserted into pcDNA3.1+ upstream of the cytomegalovirus promoter via MfeI and MluI restriction sites. The coding sequence for RFP was cut from pTurboFP635N plasmid (Evrogen, Moscow, Russia) and inserted via KpnI and NotI sites. The PARP1 motif construct was synthesized by annealing oligomers PARP1motif-F and PARP1motif-R and was inserted into the pcDNA3.1+ vector via MfeI and MluI restriction sites. To synthesize the PARP1 overexpression vector, total RNA was extracted using the NucleoSpin® RNA II kit (Machery Nagel, Germany) and reverse transcribed using the Accuscript® High Fidelity 1st Strand cDNA Synthesis Kit (Stratagene). The coding sequence was amplified with primers PARP1-F and PARP1-R and inserted into pcDNA3.1+ via NheI and XhoI restriction sites. PARP1 specific knockdown was achieved with 10 nM of Silencer® Select Validated short interfering (si)RNA #s1098 (Ambion, Austin, TX), whereas the Silencer Select Negative Control #2 siRNA (Ambion, Austin, TX) was used as a nonspecific control. Primer sequences are provided in Supporting Table 1.

Protein Extraction, Immunoblotting, and Immunofluorescence.

Lysates were extracted with the NE-PER kit (Thermo Scientific, Rockford, IL). Reducing sodium dodecyl sulfate polyacrylamide gel electrophoresis and immunofluorescence were performed under standard reducing conditions. Cells were fixed with 4% paraformaldehyde (Sigma-Aldrich, St. Louis, MO). Primary antibodies (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) for lamin B1 (sc-56145; 1:400), PARP1 (sc-74469X; 1:1,000), and HBs (sc-52411; 1:200) were used.

Luciferase Assays.

The Dual-Luciferase® Reporter Assay System (Promega) was used. First, 1.5 × 105 cells were seeded in 24-well plates and transfected with 2.2 μL of Lipofectamine 2000 (Invitrogen, San Diego, CA), 1 μg of PGL3 basic construct plasmid, and 15 ng of Renilla-luciferase construct, then lysed 30 hours later with 60 μL of buffer. Luminescence was measured using 50 μL of lysate with the GloMax® Multi-Microplate Multimode Reader (Promega) and normalized to that of the empty vector.

Electrophoretic Mobility Shift Assay, Streptavidin Pull-Down, and Matrix-Associated Laser Desorption/Ionization Time of Flight.

The electrophoretic mobility shift assay (EMSA) was performed using the LightShift Chemiluminescent kit (Thermo Scientific) and 2-μg nuclear lysates and 1-ng biotinylated probes (sense-strand sequence: 5'-TTGAGGCCTACTTCAAAGACTGTGTG-3'). The biotinylated EBNA probes that were provided were used as the negative control. Binding was performed at 37°C for 45 minutes in 20-μL reactions with EMSA buffer (12.5% glycerol, 0.5 mM of ethylenediaminetetraacetic acid, 0.3 mg of bovine serum albumin, 0.05% Nonidet P40, and 1 μg of poly-dIdC). Also, 1 μL of PARP1 antibody (sc-74469X; Santa Cruz Biotechnology) was used. Streptavidin pull-down was performed with 10 μL of Dynabeads® M-280 streptavidin (Invitrogen), 1 μg of biotinylated EMSA probe, and 70-μg nuclear lysates in 100-μL reactions in EMSA buffer. Bound proteins were eluted by boiling and sent to the Protein and Proteomics Center in the National University of Singapore for matrix-assisted laser desorption/ionization time of flight (MALDI-TOF/TOF) analysis.

cccDNA Assay.

DNA was extracted using the DNeasy Blood and Tissue kit (Qiagen, Hilda, Germany). Quantitative real-time polymerase chain reaction was performed using LightCycler® FastStart DNA MasterPLUS SYBR Green I (Roche, Basel, Switzerland) in 10-μL reactions containing 1 ng of total DNA. cccDNA was amplified with primers cccF and cccR (Supporting Table 1) and normalized to the relative amount of pcDNA3.1+, amplified by primers pcDNA-F1 and pcDNA-R1 (Supporting Table 1).

PARP1 Function, DNA Damage, and Cell Death Assays.

Histone H1 modification assay was performed with the PARP Universal Colorimetric Assay Kit (R&D Systems) and 5-μg nuclear lysates. Also, 1-μL DNA duplexes (Supporting Table 2), formed by annealing equal amounts of 100-μM DNA oligomers, were used. Alkaline comet assays were performed with the CometAssay® kit (Trevigen, Gaithersburg, MD) and scored using TriTek CometScore™ version 1.5 software (TriTek Corporation, Sumerduck, VA). Annexin V staining was performed with Annexin V-Fluos (Roche). Apoptosis was measured using the Caspase-Glo® 3/7 Assay (Promega).

Statistical Analysis.

The F-test for equal variance, followed by the one-tailed Student's t-test with equal or unequal variance, were performed.


PARP1 Is a Novel Transcriptional Activator at the HBVCP.

To determine the host factors interacting specifically with the HBVCP that may be involved in transcriptional activation, DNA probes spanning the HBVCP were biotinylated and subjected to affinity pull-down assays. A strong band of approximately 120 kDa was selectively enriched from HepG2 nuclear lysate by the probe nt 1696-1722 of enhancer II23, 24 within the HBVCP (Fig. 1A). MALDI-TOF/TOF analysis revealed that the bound protein was PARP1 (Fig. 1A; Supporting Fig. 2). In agreement, nuclear lysates derived from PARP1 siRNA-treated cells were EMSA negative when tested with biotinylated nt 1696-1722 probe (Fig. 1B). The HBVCP-PARP1 interaction was further affirmed when both PARP1-specific antibody and excess unlabeled competitor probes significantly diminished complex formation. It is important to demonstrate that the HBVCP-PARP1 interaction was not the result of binding of PARP1 to the free ends of the DNA probes. The addition of a 1,000-fold excess of poly-dIdC failed to abolish complex formation, whereas 100-fold excess of unlabeled HBVCP was sufficient to do so (Supporting Fig 3), providing confirmation for the sequence-specific nature of PARP1 binding.

Figure 1.

PARP1 is a novel transcriptional activator at the HBVCP. (A) Selective enrichment of an approximately 120-kDa protein using 1 μg of the biotinylated HBVCP probe (5'-TTGAGGCCTACTTCAAAGACTGTGTGT-3') in affinity pull-down assays from 70-μg HepG2 nuclear lysates. The MALDI-TOF/TOF result identifies PARP1 as the specific binder. Sequenced peptides are underlined and in bold. (B) One nanogram of biotinylated HBVCP probe was incubated with 2-μg HepG2 nuclear lysates treated with nonspecific siRNA. The binding complex formed in the presence of 1 μg of poly-dIdC could be removed with a 100-fold excess of unlabeled probe, PARP1-specific antibody, or with the use of nuclear lysates with PARP1 specific knockdown (K/D). The relative amounts of PARP1 in the nuclear lysates used are shown. (C) Deduction of the putative PARP1 binding site by luciferase reporter assays. The HBVCP fragment (nt 1600-1860) was cloned upstream of a luciferase reporter construct; five constructs were generated, in which each contained a 15-bp deletion that overlapped to span nt 1689-1727. Results show normalized mean luminescence ± standard error (SE) relative to the wild-type promoter of three independent experiments performed in triplicates. The exact nucleotides deleted in each construct are represented by gray lines.

PARP1 is also an important transcriptional regulator,27, 28 as studies of fibroblasts from PARP1−/− mice have altered the expression of a large number of genes.29 To determine whether the novel PARP1 binding site would be transcriptionally functional, the effect of its deletion on HBVCP activity was investigated by a luciferase reporter assay in HepG2 cells (Fig. 1C). Consistent with enhancer II function,23, 24 all deletions resulted in the loss of luciferase expression. Of these, two overlapping deletions, covering nt 1701-1721 that share the “TTCAAA” sequence, had significantly reduced luciferase expression, indicating that this is the minimal motif required for PARP1-dependent transcriptional activation.

PARP1 Motif Is Well Conserved in HBVCP.

To define the PARP1 recognition motif and map its precise site on the HBVCP, we generated scanning mutations of the “TTCAAA” sequence and three flanking nucleotide positions at either ends. All four base substitutions were tested at each position. The results indicate an absolute requirement for the “CAAA” sequence, as any change would cause significant (>75%) reduction in luciferase expression (Fig. 2). The effect of nucleotide substitutions was observed to extend two positions 5' of the “TTCAAA” motif, such that an eight-nucleotide sequence “ACTTCAAA” was defined by the boundary where nucleotide substitutions flanking it had little effect on luciferase expression. Interestingly, only substitutions at position 3 of the octamer motif resulted in increased luciferase expression, whereas all other substitutions were either neutral or deleterious. The PARP1 sequence-dependent transcription motif can, therefore, be described as “RNNWCAAA,” where “R” is either “A” or “G,” “N” is any nucleotide, and “W” is either “A” or “T,” and the optimal sequence for PARP1 sequence-dependent transcription is “ACATCAAA.” The data also suggest that wild-type HBVCP PARP1 binding motif “ACTTCAAA” is a near-optimal PARP1 recognition motif. Curiously, HBV genome alignments revealed that the HBV PARP1 site is highly conserved (Supporting Fig. 4). Most HBV genotypes possess the “ACTTCAAA” PARP1 motif, whereas genotypes F and H possess the optimal “ACATCAAA” motif. This high degree of functional PARP1 motif conservation in the HBVCP reflects the importance of PARP1 to HBV replication.

Figure 2.

PARP1 recognition motif. The effect of all possible single base substitutions on the predicted PARP1 core motif “TTCAAA” and the three bases flanking either of its sides were determined by examining altered luciferase expression. Results show mean normalized luminescence ± SE of triplicates relative to that of the wild-type promoter. Relative positions of each nucleotide with respect to the defined PARP1 binding motif are indicated.

HBV Replication Is Dependent on PARP1.

The HBVCP transcriptionally regulates the synthesis of pregenomic (pg)RNA, which converts to cccDNA, required for persistent HBV replication.30-32 To determine whether HBV replication would be dependent on PARP1, the effects of reduced PARP1 expression on cccDNA and HBs expression were investigated. HBV replication was established with a full-length genomic replicon (HBV-RFP)25, 26 driven by native HBV promoters (Supporting Fig. 5), which enables HBs and cccDNA accumulation in transfected HepG2 cells (Fig. 3A). The effects of the loss of PARP1 expression was then tested in HepG2 cells pretreated with PARP1-specific siRNA 24 hours before HBV-RFP transfection, when PARP1 expression was significantly reduced (Supporting Fig. 6). As anticipated, the loss of PARP1 resulted in the failure to accumulate cccDNA, whereas cells treated with control siRNA were still able to do so (Fig. 3B). Furthermore, the expression of HBs was also significantly diminished in transfected cells pretreated with PARP1-specific siRNA (Fig. 3C). These results concur with the loss of transcriptional activity by deletion of the PARP1 motif (Fig. 1C), providing evidence that HBV replication is dependent on HBVCP-PARP1 interaction.

Figure 3.

HBV replication is dependent on PARP1. (A) HBV replication in HepG2 is indicated by the production of HBV surface antigens in cytoplasmic lysates and the accumulation of cccDNA with time. (B) Effect of PARP1 specific knockdown on HBV replication, as indicated by cccDNA fold-change relative to 6 hours post-transfection. Values represent mean fold-change ± SE of independent experiments performed in triplicates. ***P < 0.001 with the one-tailed Student's t-test. (C) Effect of PARP1-specific knockdown on enhancer II function, as determined by immunofluorescence (IF) staining for HBs expression 60 hours after siRNA treatment. HBV-RFP-transfected cells fluoresce red.

PARP1 Motif Inhibits PARP1-Dependent ADP-Ribosylation.

As PARP1 enzymatic activity is known to be activated by binding DNA strand breaks,15, 33 we investigated whether the same could be induced by the PARP1 binding motif. Using an in vitro histone H1 modification assay, we detected the amount of ADP-ribosylation activity in the presence of damaged DNA and 20-base-pair (bp) DNA duplexes bearing the “ACATCAAA” motif with endogenous PARP1 from HepG2 nuclear lysates (Fig 4). Surprisingly, instead of increasing the amount of ADP-ribosylated histone H1, motif addition reduced the amount of ADP-ribosylated histone H1, when compared to buffer control. The effect of the PARP1 motif was sequence dependent, as mutations within the octamer core “ACATCAAA” sequence significantly diminished the capacity to block PARP1-dependent histone H1 modification. Furthermore, mutations to sequences flanking the motif showed no difference from the wild-type sequence in ability to ADP-ribosylate histone H1, validating the PARP1 binding properties of the defined motif. These results suggest that, in contrast to damaged DNA, which activates PARP1, binding the “ACATCAAA” sequence results in PARP1 inhibition. It is not clear, at this point, whether the PARP1 binding motif competes with damaged DNA for the same PARP1 binding site, but it appears that upon binding an optimal motif sequence, the PARP1-motif complex is stable and negates the activation of PARP1 to ADP-ribosylate targets. To demonstrate the relative potency of motif-mediated PARP1 inhibition, nuclear lysates from HepG2 cells treated with PARP1-specific siRNA was shown to reduce histone H1 modification by 40%, when compared with lysates from nonspecific siRNA controls (Fig. 4). This indicates that small amounts of active PARP1 is sufficient for histone H1 modification in vitro; thus, the 25% reduction of ADP-ribosylated histone H1 achieved with the addition of wild-type PARP1 motif reflects that the bulk of PARP1 molecules was inhibited.

Figure 4.

PARP1 motif reduces PARP1 enzymatic activity in vitro. The effect of PARP1 binding to 20-bp DNA duplexes containing the PARP1 motif “ACATCAAA” (highlighted) on PARP1 enzymatic activity was investigated by histone H1 modification assays. Mutations to the duplex sequence are shown in bold. The effect of PARP1 specific knockdown (K/D) on PARP1 activity is also shown for comparison to demonstrate that the near absence of PARP1 expression reduces in vitro histone H1 modification by 40%. Means of an experiment ± SE performed in triplicates are shown. **P < 0.01 using the one-tailed Student's t-test.

Exogenous PARP1 Motif Impairs DNA Repair.

Consistent with its ability to bind PARP1 for transcriptional activation, the “ACTTCAAA” HBVCP PARP1 binding motif could also interfere with histone H1 ADP-ribosylation (Fig. 5A). This raises the possibility that HBVCP-PARP1 interaction not only supports HBV replication, but also impairs PARP1 enzyme-dependent functions, such as DNA repair in vivo. If this were true, the ability of cells to effectively carry out DNA strand-break repair when challenged by DNA-damaging agents would be compromised. To verify this, a construct bearing the HBV-PARP1 binding motif in three tandem copies (Fig. 5B) was tested for its capacity to inhibit cellular PARP1 enzymatic activity by determining the degree of DNA damage induced with etoposide (DNA single- and double-strand break inducer) or bleomycin (DNA double-strand break inducer). Alkaline comet assays revealed that HepG2 cells transfected with the PARP1 motif had significantly more DNA in comet tails than cells treated with dimethyl sulfoxide (DMSO) or the control vector (Fig. 5C), reflecting enhanced DNA damage. This suggests that the ability of PARP1 to ADP-ribosylate protein targets required in DNA damage-repair pathways was reduced, supporting the inhibitory role of HBVCP-PARP1 motif expression on nuclear PARP1 enzymatic activity. The effect of the PARP1 motif was further assessed for its ability to sensitize cells to induced cytotoxicity caused by DNA-damaging agents. Consistent with accumulation of damaged DNA, etoposide or bleomycin treated HepG2 cells transfected with the PARP1 motif had a significantly larger population of Annexin V–positive cells (Fig. 5D). In contrast, DMSO treatment or vector control did not show significant changes in Annexin V staining. The enhanced cytotoxicity toward sublethal amounts of etoposide and bleomycin in cells transfected with the motif is reminiscent of the hypersensitivity of PARP1 knockout and haploinsufficient mice toward DNA-damaging agents,18, 20 reflecting compromised DNA repair with the loss of PARP1 enzymatic function. The ability of the PARP1 motif to specifically disrupt cellular PARP1 function was also demonstrated by diminished HBs expression in HepG2 cells cotransfected with HBV-RFP (red fluorescent protein) (Supporting Fig. 7).

Figure 5.

Exogenous PARP1 motif impairs cellular DNA repair. (A) Inhibition of PARP1-dependent histone H1 modification by short DNA duplexes containing the “ACTTCAAA” HBVCP PARP1 binding motif. The mean ± SE reduction in histone H1 modification of triplicates with the viral PARP1 motif relative to buffer control was analyzed using the one-tailed Student's t-test with equal variance. ***P < 0.001. (B) Construct containing the HBVCP PARP1 binding motif “ACTTCAAA” in tandem repeats. (C) The extent of DNA damage visualized by alkaline comet assays induced with 50 nM of etoposide and 20 ng/mL of bleomycin in HepG2 cells transfected with PARP1 motif construct of independent experiments is as plotted and analyzed using the one-tailed Student's t-test. Cells transfected with empty vector were used as controls. Bars indicate the median extent of DNA damage. ***P < 0.001 with the one-tailed Student's t-test. (D) HBVCP PARP1 motif sensitizes cells to apoptosis by DNA-damaging agents. Dying cells stain positive by Annexin V staining resulting from irreparable DNA damage induced by 50 nM of etoposide or 20 ng/mL of bleomycin. DMSO (0.01%) and empty vectors were used as controls.

To confirm that the effects of the HBVCP-PARP1 motif are specific to PARP1, rescue experiments were performed, in which PARP1 was overexpressed to compensate for the loss of DNA repair. Excess PARP1 cannot avert the accumulation of cytotoxic DNA lesions if alternative DNA repair pathways were instead compromised. Using apoptotic cell death as the end-point of extensive irreparable DNA damage, the effect of etoposide or bleomycin on HepG2 cells cotransfected with the HBVCP-PARP1 binding motif and PARP1 or RFP expression vectors was determined by apoptotic caspase-dependent cleavage of luminogenic substrates. It was observed that increased sensitization to apoptosis was overcome by specific overexpression of PARP1 (Fig. 6), greatly reducing the enhanced caspase activities of cells transfected with the PARP1 binding motif, when compared to vector controls. Conversely, overexpression of RFP had little effect on the sensitivity of transfected cells toward DNA damage-induced apoptosis, demonstrating that the reduction in apoptosis toward induced DNA damage was PARP1 specific. Therefore, the HBVCP-PARP1 binding motif is a specific inhibitor of cellular PARP1 activity, compromising the capacity of a cell to carry out DNA repair.

Figure 6.

PARP1 motif-dependent enhanced cytotoxicity to induced DNA damage is PARP1 specific. Exogenous PARP1 was readily detectable in 10 μg of HepG2 nuclear lysates within 48 hours of transfection with 1 μg of its expression plasmid. One microgram of the plasmid containing the PARP1 binding motif and 500 ng of the PARP1 overexpression plasmid were cotransfected into HepG2 cells, and the mean percentage ± SE increase in caspase activity associated with the PARP1 motif relative to the empty vector was compared to that of RFP overexpression. DNA damage was induced by 100 nM of etoposide or 10 ng/mL of bleomycin. ***P < 0.001 obtained using the one-tailed Student's t-test. O/E, overexpression.


A number of clinical epidemiological studies have demonstrated that patients with high viral DNA loads have significantly enhanced risk of developing HCC,8-11 although the reason for this remains unclear. To understand the regulation of HBV viral replication, we focused on the interaction of host transcription factors that influence HBVCP activity. A surprising finding is the specific recognition of a DNA binding motif in the HBVCP by the PARP1 DNA repair enzyme. Interestingly, HBV is not the only oncogenic DNA virus with a functional PARP1 binding motif. Similar PARP1 binding sequences have been found on the human T-cell leukemia virus Tax responsive element (HTLV Tax RE) and have been shown to be required for transcriptional activation.34 Kaposi's sarcoma-associated virus has also been shown to bind PARP1 via its DNA for genomic maintenance and replication.35 Consistent with the suppression of PARP1-dependent poly-ADP ribosylation by motif recognition, inhibition of PARP1 has been shown to enhance PARP1 motif-dependent transcription, resulting in increased viral transcripts and genome copies.34, 35 The inhibition of PARP1 thus could be a prerequisite for motif recognition and transcriptional activation, as the inhibition of automodification prevents PAR-mediated electrostatic repulsion from DNA16, 36 to enable PARP1 retention on its recognition motif for the transcriptional apparatus to assemble. Besides transcriptional regulation, viruses such as the human immunodeficiency virus and HBV itself also make use of PARP1 for genomic integration, the process that is also enhanced by enzymatic inhibition.22, 37 The requirement for enzymatically inactive PARP1 for transcription and genomic integration suggests that PARP1 inhibition may be a common mechanism utilized by viruses for replication and, in doing so, impairs DNA repair, leading to enhanced risk of developing malignancy. This is supported by evidence that individuals with decreased PARP1 enzymatic activity have increased risks of developing cancers.38, 39 Perhaps, low PARP1 enzymatic activity is also a risk factor for chronic infections and renders chronic carriers of PARP1-dependent viruses susceptible to cancer development.

Even though PARP1 can bind DNA in a sequence-dependent manner to carry out transcription,27 its consensus recognition motif has not been agreed upon.34, 40 By studying the effects of single base substitutions on PARP1-dependent transcription, the octamer, “RNNWCAAA,” was shown to be the minimal motif required for transcriptional activity (Fig. 2). Importantly, in agreement with our data, substituting “A” with “C” at position 1 of the SMARCB1 promoter had been associated with decreased PARP1-dependent transcriptional activity.41 Within the PARP1 binding site of the HTLV Tax RE, mutating nucleotides 5 and 6 from “CA” to “AC” abolished PARP1 binding, whereas substitutions at positions 2 and 3 were less important for the functional specificity of the motif.34 Furthermore, in line with the deleterious effects of nucleotide substitution, changing position 5 from “C” to “T” at the Bcl-6 PARP1 binding site resulted in abrogation of PARP1-dependent transcription.42 These data are congruent with our findings that nucleotide positions 5 and 6 are critical for PARP1-dependent transcriptional activation, whereas nucleotide positions 2 and 3 of the octamer are less so. Thus, the “RNNWCAAA” octamer may be used to describe the PARP1 motif that also reflects the relative contribution of each nucleotide position to bind PARP1 required for transcription.

PARP1 hyperactivity has been associated with various disease states, such as cancer.43, 44 A survey of 37 HCC patient tumor samples with its matched nontumor tissue showed that PARP1 mRNA is, on average, 21.11-fold (range, 20.98 to 21.21) above the mean of nontumor tissues (Supporting Table 3) and, therefore, indicates that the PARP1 levels in the HepG2 liver cell line is moderately elevated from physiological levels. Suppression of PARP1 enzymatic activity by general PARP inhibitors is thought to have therapeutic potential, as they have been shown to enhance the cytotoxic potential of DNA-damaging agents in clinical trials.43, 44 In contrast to binding DNA strand breaks for DNA repair, the capacity for PARP1 to ADP-ribosylate histone H1 surprisingly decreased when bound by the PARP1 motif (Fig. 4). Similar to how HBV DNA impairs cellular PARP1 functions, we propose that exogenous DNA bearing the PARP1 binding motif can function as a cognate ligand for PARP1 that interferes with its ability to carry out DNA repair, enhancing synthetic lethality of chemotherapeutic agents. Indeed, transfection of a synthetic construct bearing tandem repeats of the HBVCP PARP1 binding motif was able to increase cytotoxicity of HepG2 HCC cells induced by etoposide and bleomycin (Fig. 5). Because affinity pull-down with the PARP1 binding motif produced PARP1 as the only interacting (Fig. 1), this specificity of the PARP1 binding motif for PARP1 would be advantageous over current PARP inhibitors, potentially reducing adverse effects associated with inhibition of other PARP family members targeted by general PARP inhibitors.28, 45 Understanding how PARP1 inhibition is achieved by engaging a specific DNA binding motif would also shed light on how the enzyme is allosterically regulated.

In conclusion, this study describes the identification of a specific PARP1 binding site in the HBVCP and its role as a recognition motif for PARP1-dependent transcription. The PARP1 motif also possesses enzymatic inhibitory properties, resulting in impaired DNA repair and the accumulation of damaged DNA when exogenously expressed in cells. This finding suggests that HBV DNA impairs PARP1 cellular functions, which may contribute to genomic instability over time. Taken together, the results indicate that the HBV PARP1 binding motif is not only important for HBV replication, but also suppresses PARP1-dependent DNA repair, providing a novel mechanism to explain the association between high HBV DNA loads and the increased risk of HCC development.


The authors thank Prof. W.N. Chen (Nanyang Technological University) for the kind gift of the HBV replicon. The authors also thank M.K. Sng for technical assistance and B. Wang, Z. Xiao, and members of the E.C.R. lab and the Protein and Proteomics Center (National University of Singapore) for technical help and advice.