SIRT3 restricts hepatitis B virus transcription and replication through epigenetic regulation of covalently closed circular DNA involving suppressor of variegation 3‐9 homolog 1 and SET domain containing 1A histone methyltransferases

Abstract

Hepatitis B virus (HBV) infection remains a major health problem worldwide. Maintenance of the covalently closed circular DNA (cccDNA), which serves as a template for HBV RNA transcription, is responsible for the failure of eradicating chronic HBV during current antiviral therapy. cccDNA is assembled with cellular histone proteins into chromatin, but little is known about the regulation of HBV chromatin by histone posttranslational modifications. In this study, we identified silent mating type information regulation 2 homolog 3 (SIRT3) as a host factor restricting HBV transcription and replication by screening seven members of the sirtuin family, which is the class III histone deacetylase. Ectopic SIRT3 expression significantly reduced total HBV RNAs, 3.5‐kb RNA, as well as replicative intermediate DNA in HBV‐infected HepG2‐Na⁺/taurocholate cotransporting polypeptide cells and primary human hepatocytes. In contrast, gene silencing of SIRT3 promoted HBV transcription and replication. A mechanistic study found that nuclear SIRT3 was recruited to the HBV cccDNA, where it deacetylated histone 3 lysine 9. Importantly, occupancy of SIRT3 on cccDNA could increase the recruitment of histone methyltransferase suppressor of variegation 3‐9 homolog 1 to cccDNA and decrease recruitment of SET domain containing 1A, leading to a marked increase of trimethyl‐histone H3 (Lys9) and a decrease of trimethyl‐histone H3 (Lys4) on cccDNA. Moreover, SIRT3‐mediated HBV cccDNA transcriptional repression involved decreased binding of host RNA polymerase II and transcription factor Yin Yang 1 to cccDNA. Finally, hepatitis B viral X protein could relieve SIRT3‐mediated cccDNA transcriptional repression by inhibiting both SIRT3 expression and its recruitment to cccDNA. Conclusion: SIRT3 is a host factor epigenetically restricting HBV cccDNA transcription by acting cooperatively with histone methyltransferase; these data provide a rationale for the use of SIRT3 activators in the prevention or treatment of HBV infection. (Hepatology 2018).

Potential conflict of interest: Nothing to report.

Supported by the National Natural Science Foundation of China (81672012 and 81571980), the National Science and Technology Major Project (2017ZX10202203), and the Chongqing Natural Science Foundation (cstc2016jcyjA0183).

Hepatitis B virus (HBV) infection remains a major global health problem, affecting more than 240 million individuals worldwide who are at high risk for developing liver cirrhosis and hepatocellular carcinoma.¹ HBV virions contain a 3.2‐kb partially double‐stranded circular DNA genome. Upon entry into hepatocytes, the capsid dissociates and the genomic relaxed circular DNA is converted into a covalently closed circular DNA (cccDNA) molecule. cccDNA serves as the template for transcription of all viral mRNAs. It accumulates in the nucleus of infected cells as a stable episome organized into minichromosomes by histones and nonhistone viral and cellular proteins.² During chronic HBV infection, cccDNA plays a key role in viral persistence, viral reactivation after treatment withdrawal, and drug resistance.³ Therefore, targeting cccDNA represents an attractive approach for future curative therapies of HBV infection.

The human silent mating type information regulation 2 homolog (sirtuin) family (SIRT1‐7) of oxidized nicotinamide adenine dinucleotide (NAD⁺)‐dependent histone deacetylase is implicated in the regulation of many cellular functions, such as the stress response,⁴ metabolism,⁵ DNA repair and apoptosis,⁶ carcinogenesis,⁸ and aging.¹⁰ SIRT1, the most widely studied member of the sirtuin family, is a nuclear deacetylase that is recruited onto the HBV cccDNA minichromosome during HBV replication.² SIRT1 regulates HBV replication and transcription by targeting the transcription factor activator protein‐1(AP‐1).¹² Unlike SIRT1, SIRT3 has been reported to be primarily localized within mitochondria and is widely expressed in adult and fetal tissues. SIRT3 is also reported to be a nuclear NAD⁺‐dependent histone deacetylase targeting histone 3 lysine 9 (H3K9) and histone 4 lysine 16 (H4K16) during cellular stress conditions.¹³ SIRT3 regulates a wide range of important biological functions including metabolic control,¹⁵ neuroprotection,¹⁶ cardiovascular diseases,¹⁸ aging,²⁰ and cancer,²¹ mainly through its deacetylase activity. However, its role in HBV replication remains unexplored.

In this study, we used models of HBV‐infected HepG2‐Na⁺/taurocholate cotransporting polypeptide (NTCP) and primary human hepatocytes (PHHs) to demonstrate that cccDNA transcription is regulated by SIRT3 through chromatin structure modulation. SIRT3 could be localized to the nucleus, where it deacetylates cccDNA‐bound H3K9, increases recruitment of histone methyltransferase suppressor of variegation 3‐9 homolog 1 (SUV39H1) and decreases SET domain containing 1A (SETD1A) recruitment. Additionally, SIRT3‐mediated HBV cccDNA transcriptional repression involves decreased binding of host RNA polymerase II (Pol II) and transcription factor Yin Yang 1 (YY1) to cccDNA. Finally, viral protein hepatitis B viral X protein (HBx) could relieve SIRT3‐mediated transcriptional repression of HBV cccDNA. Our data suggest a role for SIRT3 in HBV cccDNA transcription and identify it as a host factor in the HBV replication process.

Materials and Methods

CELL CULTURE

HepG2 and HepG2‐NTCP cells were maintained in modified Eagle's medium supplemented with 10% fetal bovine serum (FBS). PHHs were maintained in hepatocyte medium (catalog no. 5210; Sciencell). Huh‐7 cells were maintained in Dulbecco's modified Eagle medium (DMEM) containing 10% FBS. HepAD38 cells were maintained in DMEM supplemented with 10% FBS and 400 μg/mL G418. All cells were maintained in a humidified incubator at 37°C with 5% CO₂.

PATIENTS

Liver biopsies were acquired from 5 hepatitis B e antigen (HBeAg)–positive chronic hepatitis B (CHB) patients and 5 HBeAg‐negative inactive carriers and stored at –80°C before use. The clinical and virological parameters of the patients are summarized in Supporting Table S1. Written informed consent was obtained from all patients, and the study protocol was approved by the Ethical Committee of the Chongqing Medical University in accordance with the Declaration of Helsinki.

STATISTICAL ANALYSIS

Data are presented as the mean ± standard error of the mean of at least three independent experiments. Statistics were performed with the nonparametric Mann‐Whitney U test. A value of P < 0.05 was considered significant (^*P < 0.05, ^**P < 0.01). All statistical analyses were performed using SPSS 19.0 software.

Results

IDENTIFICATION OF SIRT3 AS A HOST FACTOR SUPPRESSING HBV RNA TRANSCRIPTION

To investigate whether histone deacetylases play an important role in HBV transcription and replication, we screened SIRT1‐7, the class III histone deacetylase, in derivatives of the human liver cell line HepG2. The HepAD38 cell line has stable integration of the HBV genome with HBV DNA replication under the control of tetracycline. HepG2‐NTCP cells stably express NTCP, the HBV receptor, and are susceptible to HBV infection. The mRNA and protein levels of SIRT1, SIRT3, SIRT6, and SIRT7 were significantly altered in response to HBV DNA replication in HepAD38 cells or HBV infection of HepG2‐NTCP cells (Fig. 1A,B). These data revealed that HBV replication regulated expression of several sirtuin members (SIRT1, SIRT3, SIRT6, and SIRT7). To investigate whether sirtuin members could also regulate HBV RNA transcription and DNA replication, we transfected expression constructs for FLAG‐tagged SIRT1‐7 into HBV‐infected HepG2‐NTCP cells to identify which sirtuin members could catalyze cccDNA‐bound histone deacetylation to result in transcription repression. Real‐time PCR found that SIRT3 or SIRT7 overexpression markedly decreased levels of total HBV RNAs and 3.5‐kb RNA (Fig. 1C,D), suggesting that SIRT3 and SIRT7 might epigenetically suppress cccDNA transcriptional activity. We focused on SIRT3 for further investigation.

Figure 1: Identification of SIRT3 as a host factor suppressing HBV RNA transcription. (A,B) Expression of sirtuin members in HepAD38 and HepG2‐NTCP cells. HepG2‐NTCP cells were infected with 5 × 10³ genome equivalents/cell of HBV particles in the presence of 4% PEG8000. mRNA and protein levels were analyzed in HepG2‐NTCP cells at day 5 postinfection and in HepAD38 cells with or without tetracycline treatment. (C,D) The effects of sirtuin members on total HBV RNAs and 3.5‐kb RNA. HepG2‐NTCP cells were infected with HBV and then transfected with the indicated plasmids. HBV RNAs levels were analyzed by real‐time PCR using specific primers at day 4 after transfection. (E) SIRT3 expression in HBV‐infected HepG2‐NTCP cells. mRNA and protein levels of SIRT3 were analyzed at the indicated days postinfection. (F) SIRT3 expression in HepG2‐NTCP cells infected with different doses of HBV particles. mRNA and protein levels of SIRT3 were analyzed 5 days postinfection. ^* P < 0.05, ^** P < 0.01. Abbreviation: Tet, tetracycline.

To substantiate the association between HBV and SIRT3, SIRT3 expression was analyzed in HBV‐infected HepG2‐NTCP cells. Both the mRNA and protein levels of SIRT3 were down‐regulated in HBV‐infected HepG2‐NTCP cells at day 4‐6 postinfection (Fig. 1E). Furthermore, the degrees of inhibition of SIRT3 mRNA and protein were dependent on the dose of the HBV inoculum (Fig. 1F). Collectively, these data suggest that SIRT3 and HBV might inhibit each other.

SIRT3 OVEREXPRESSION INHIBITS HBV TRANSCRIPTION AND REPLICATION, DEPENDENT ON ITS DEACETYLASE ACTIVITY

To clarify how SIRT3 inhibits HBV replication, HepG2‐NTCP cells or PHHs were infected with HBV particles before transduction with lentivirus expressing wild‐type SIRT3 or its deacetylase inactive form (SIRT3^H248Y). Ectopic SIRT3 expression inhibited total HBV RNAs and 3.5‐kb RNA levels in HBV‐infected HepG2‐NTCP cells (Fig. 2A), whereas expression of SIRT3^H248Y had no effect. Northern blot confirmed that SIRT3 overexpression resulted in markedly decreased 3.5‐kb, 2.4‐kb, and 2.1‐kb HBV RNAs levels (Fig. 2A). We treated HepG2‐NTCP cells with the RNA Pol II inhibitor 7‐aminoactinomycin D to block RNA transcription. The data showed that the half‐life of total RNAs and 3.5‐kb RNA remained unchanged in SIRT3‐expressing cells compared with control cells, suggesting that SIRT3 did not modulate HBV RNA stability (Supporting Fig. S1A,B). Next, we examined the effect of SIRT3 on the cccDNA level and its transcription activity. Real‐time PCR showed that SIRT3 slightly reduced the cccDNA level in HBV‐infected HepG2‐NTCP cells, although the difference did not reach statistical significance (Fig. 2B). The ratios of total RNA to cccDNA and 3.5‐kb RNA to cccDNA were calculated as indicators for HBV cccDNA transcriptional activity. Real‐time PCR found that SIRT3 overexpression significantly reduced the ratios of total RNA/cccDNA and 3.5‐kb RNA/cccDNA (Fig. 2B). Furthermore, real‐time PCR and Southern blot showed that ectopic SIRT3 expression significantly decreased the level of HBV DNA replicative intermediates (Supporting Fig. S1C). Similarly, SIRT3 also inhibited hepatitis B surface antigen (HBsAg) and HBeAg secretion (Supporting Fig. S1D). The inhibitory effect of SIRT3 on cccDNA transcriptional activity was fully confirmed in HBV‐infected PHHs, in which ectopic expression of SIRT3 resulted in an inhibition of total HBV RNAs, 3.5‐kb RNA, as well as the ratios of total RNA/cccDNA and 3.5‐kb RNA/cccDNA (Fig. 2C,D). SIRT3 also suppressed HBV DNA as evidenced by real‐time PCR (Supporting Fig. S1E,F). Finally, the correlation between SIRT3 and cccDNA transcriptional activity was examined during natural HBV infection. In HBV‐infected PHHs, HBV RNAs, cccDNA, and the cccDNA transcriptional activity (total RNA/cccDNA and 3.5‐kb RNA/cccDNA) fluctuated during natural HBV infection. Interestingly, the SIRT3 mRNA level was negatively correlated with cccDNA transcriptional activity (Fig. 2E,F). These findings support the notion that SIRT3 can regulate HBV cccDNA transcription, dependent on its deacetylase activity.

Figure 2: SIRT3 overexpression inhibits HBV transcription and replication, dependent on its deacetylase activity. HepG2‐NTCP cells or PHHs were infected with 5 × 10³ genome equivalents/cell of HBV particles in the presence of 4% PEG8000 and then transduced with lentivirus expressing the indicated plasmids. (A) SIRT3 overexpression reduced total HBV RNAs and 3.5‐kb RNA. HBV‐infected HepG2‐NTCP cells were transduced with lentivirus expressing wild‐type SIRT3, SIRT3^H248Y, and empty vector. HBV RNAs were analyzed by real‐time PCR using specific primers (left panel) and northern blotting analysis (right panel) at day 4 after lentiviral transduction. For northern blotting analysis, each panel was loaded with an equal amount of total RNA and probed with a 1‐kb single‐stranded HBV RNA probe (genome position 126‐1225). Ribosomal RNAs (28s and 18s) were used as loading control. (B) SIRT3 overexpression inhibited ratios of total RNA/cccDNA and 3.5‐kb RNA/cccDNA in HBV‐infected HepG2‐NTCP cells. HBV RNA levels were analyzed by real‐time PCR using specific primers at day 4 after lentiviral transduction. HBV cccDNA was extracted at day 4 post–lentiviral transduction using the Hirt method. Real‐time PCR analysis was performed using cccDNA‐specific primers. (C) SIRT3 overexpression in PHHs reduced total HBV RNAs and 3.5‐kb RNA. HBV‐infected PHHs were transduced with lentivirus expressing wild‐type SIRT3, SIRT3^H248Y, and empty vector. (D) SIRT3 overexpression inhibited ratios of total RNA/cccDNA and 3.5‐kb RNA/cccDNA in HBV‐infected PHHs. (E,F) Dynamic changes of HBV cccDNA, total HBV RNAs, and 3.5‐kb RNA in HBV‐infected PHHs. One‐quarter of HBV‐infected PHHs in 100‐mm dishes were used to extract cccDNA, and one‐quarter of the cells were used to extract RNA. HBV cccDNA, total RNAs, and 3.5‐kb mRNA in HBV‐infected PHHs were analyzed at the indicated days postinfection (E). Ratios of total RNA/cccDNA, 3.5‐kb RNA/cccDNA, and SIRT3 mRNAs were analyzed at the indicated days postinfection (F). ^* P < 0.05, ^** P < 0.01. Abbreviation: dpi, days postinfection.

SIRT3 SILENCING FACILITATES HBV REPLICATION

To elucidate whether endogenous SIRT3 exerts a similar antiviral ability, HepG2‐NTCP cells and PHHs were transduced with lentivirus expressing short hairpin RNA (shRNA) targeting SIRT3 (shSIRT3‐1 and shSIRT3‐2). HepG2‐NTCP cells expressing shSIRT3 showed an increase in total HBV RNAs and 3.5‐kb RNA as evidenced by real‐time PCR and northern blotting analysis (Fig. 3A,B). Importantly, SIRT3 knockdown increased the ratios of total RNA/cccDNA and 3.5‐kb RNA/cccDNA (Fig. 3C). SIRT3 suppression also resulted in increased HBV DNA replicative intermediates (Fig. 3D). In addition, SIRT3 down‐regulation resulted in increased HBsAg and HBeAg in culture supernatant (Supporting Fig. S2A,B). Similarly, SIRT3 knockdown in HBV‐infected PHH cells increased total HBV RNAs, 3.5‐kb RNA, total RNA/cccDNA, 3.5‐kb RNA/cccDNA, as well as HBV DNA replicative intermediates (Fig. 3E,F; Supporting Fig. S2C). These findings indicate that endogenous SIRT3 inhibits HBV transcription and replication.

Figure 3: SIRT3 silencing enhanced HBV transcription and replication. HepG2‐NTCP cells or PHHs were infected with 5 × 10³ genome equivalents/cell of HBV for 24 hours and then transduced with lentivirus expressing the indicated shRNA. Cells were harvested at day 4 after lentiviral transduction. (A,B) SIRT3 suppression increased total HBV RNAs and 3.5‐kb RNA expression in HBV‐infected HepG2‐NTCP cells. HBV RNA levels were analyzed by real‐time PCR (A) and northern blotting analysis (B). (C) SIRT3 silencing increased ratios of total RNA/cccDNA and 3.5‐kb RNA/cccDNA in HepG2‐NTCP cells. (D) SIRT3 suppression increased HBV DNA replicative intermediates in HBV‐infected HepG2‐NTCP cells. HBV DNA replicative intermediates were subjected to real‐time PCR and Southern blot. (E) SIRT3 suppression in HBV‐infected PHHs increased total HBV RNAs and 3.5‐kb RNA. (F) SIRT3 silencing enhanced the ratios of total RNA/cccDNA and 3.5‐kb RNA/cccDNA in HBV‐infected PHHs. ^* P < 0.05, ^** P < 0.01. Abbreviations: dsDNA, double‐stranded DNA; rcDNA, relaxed circular DNA; rRNA, ribosomal RNA; ssDNA, single‐stranded DNA.

NUCLEAR SIRT3 INHIBITS cccDNA TRANSCRIPTION THROUGH DEACETYLATION OF HBV cccDNA‐BOUND HISTONE 3

Growing evidence indicates that cccDNA transcription is controlled by posttranslational modifications of cccDNA‐bound histones.²⁴ The above findings encouraged us to investigate how SIRT3 regulates cccDNA transcription. Because cccDNA regulation is a nuclear event, we first examined the subcellular localization of SIRT3. Although SIRT3 is a mitochondrial deacetylase,²⁵ it has also been reported to serve as a nuclear NAD⁺‐dependent histone deacetylase upon cellular stress.¹⁴ Immunofluorescence analysis revealed that SIRT3 was localized in both the mitochondria and nucleus of HepAD38 cells (Supporting Fig. S3A). A subcellular fractionation experiment found that although the expression of SIRT3 in the mitochondria is much higher than that in the nucleus, SIRT3 was clearly detectable in the nuclear fraction (Supporting Fig. S3B). The nuclear localization of SIRT3 was further confirmed in HepG2‐NTCP and PHH cells (Supporting Fig. S3C,D). Moreover, ectopic expression of SIRT3 resulted in decreased acetylation of histone 3 (H3) and histone 4 (H4) in HepAD38 and HepG2‐NTCP cells, supporting its function as a nuclear NAD⁺‐dependent histone deacetylase (Supporting Fig. S3E). To further examine the effect of nuclear SIRT3 on cccDNA transcription, we added a nuclear localization signal to SIRT3 (NLS‐SIRT3). NLS‐SIRT3 expression in HepG2‐NTCP cells led to stronger inhibition of total RNA/cccDNA, 3.5‐kb RNA/cccDNA, as well as HBV DNA replicative intermediates relative to the parental wild‐type SIRT3 (Supporting Fig. S4A‐C), which confirmed the role of nuclear SIRT3 in the HBV transcriptional inhibition.

Next, we used a cccDNA chromatin immunoprecipitation (ChIP) assay, which couples chromatin immune precipitation with cccDNA‐specific PCR to selectively detect histones bound to cccDNA. The housekeeping gene glyceraldehyde‐3‐phosphate dehydrogenase (GAPDH) and myosin heavy chain 6 (MYH6) were chosen as the control genes (Supporting Fig. S5A‐C). The promoter of GAPDH was used as a control for signals activating transcription, while the MYH6 promoter was served as a control for signals suppressing transcription. The ChIP assay found that SIRT3 was recruited onto cccDNA (Fig. 4A). cccDNA‐bound H3 was significantly hypoacetylated in HepG2‐NTCP cells overexpressing SIRT3, whereas cccDNA‐bound H4 showed no change (Fig. 4B). Acetylation (Ac) of H3 and H4 associated with the promoter of GAPDH or MYH6 was similar in cells overexpressing SIRT3 or vector (Fig. 4B). We further performed cccDNA ChIP by using anti‐H3K9Ac and anti‐H4K16Ac to identify the acetylated lysine residue of histone that is preferentially targeted by SIRT3. The results showed that SIRT3 overexpression decreased the level of acetylated cccDNA‐bound H3K9 without affecting cccDNA‐bound H4K16Ac (Fig. 4C,D), suggesting that SIRT3 showed some preference for histone H3K9 at the cccDNA minichromosome.

Figure 4: Nuclear SIRT3 inhibits cccDNA transcription through deacetylation of HBV cccDNA‐bound H3. (A) SIRT3 was recruited to cccDNA. Cross‐linked chromatin from HBV‐infected HepG2‐NTCP cells transduced with lentivirus expressing Flag‐SIRT3 was immunoprecipitated with the relevant control immunoglobulin G or specific anti‐Flag antibodies, followed by PCR quantification of HBV cccDNA using specific primers. ChIP results are expressed as a percentage of input. (B‐D) Effect of SIRT3 on acetylation of H3 and H4 associated with cccDNA. Levels of H3Ac, H4Ac, H3K9Ac, and H4K16Ac associated with cccDNA or GAPDH or MYH6 promoter were examined by ChIP with anti‐acetyl‐H3, anti‐acetyl‐H4, anti‐acetyl‐histone H3 (Lys9), and anti‐histone acetyl‐H4 (Lys16) antibodies, respectively. ^** P < 0.01. Abbreviation: IgG, immunoglobulin G.

SIRT3‐MEDIATED HBV cccDNA TRANSCRIPTIONAL REPRESSION INVOLVES HISTONE METHYLTRANSFERASES, RNA Pol II, AND TRANSCRIPTION FACTOR YY1

Besides acetylation, H3 and H4 methylation is also associated with chromatin structure in the regulation of HBV cccDNA transcription. It has been reported that histone deacetylase and methyltransferase could act cooperatively and sequentially to promote heterochromatin formation.²⁸ We thus examined a panel of known histone methyltransferases, including SUV39H1, SETD1A, euchromatic histone lysine methyltransferase 2 (G9a), enhancer of zeste homolog 2 (EZH2), protein arginine methyltransferase 5 (PRMT5), and SETD2. The ChIP assay found that SIRT3 significantly increased the recruitment of SUV39H1 but decreased recruitment of SETD1A to cccDNA in HBV‐infected HepG2‐NTCP cells. SIRT3 has no effect on recruitment of either SUV39H1 or SETD1A to the promoter of GAPDH or MYH6 (Fig. 5A). SUV39H1 is reported to preferentially catalyze trimethyl‐histone H3 (Lys9) (H3K9me3), which is a chromatin repressive marker, and SETD1A preferentially targeted trimethyl‐histone H3 (Lys4) (H3K4me3), which is a chromatin active marker. We further confirmed that SIRT3 resulted in a strong increase of H3K9me3 and a decrease of H3K4me3 in HBV‐infected HepG2‐NTCP cells (Fig. 5B,C). However, ectopic expression of SIRT3 did not affect the recruitment of histone acetylases (such as p300 and cAMP response element binding protein binding protein) and histone deacetylase (such as HDAC1) to cccDNA (Supporting Fig. S6).

Figure 5: SIRT3‐mediated HBV cccDNA transcriptional repression involves histone methyltransferases SUV39H1 and SETD1A, RNA Pol II, and transcription factor YY1. (A) Effect of SIRT3 on the recruitment of histone methyltransferases to cccDNA. Levels of histone methyltransferases including SUV39H1, SETD1A, G9a, EZH2, PRMT5, and SETD2 associated with cccDNA or GAPDH or MYH6 promoter were examined by ChIP with the indicated antibodies in HBV‐infected HepG2‐NTCP cells. (B,C) SIRT3 overexpression correlated with increased H3K9me3 and decreased H3K4me3. Levels of H3K9me3 and H3K4me3 associated with HBV cccDNA or GAPDH or MYH6 promoter were determined by ChIP assay with the indicated antibodies. (D) Effect of SIRT3 on the binding of RNA Pol II to cccDNA. The level of RNA Pol II associated with HBV cccDNA or GAPDH or MYH6 promoter was analyzed by ChIP assay with the indicated antibodies. (E) Effect of SIRT3 on the binding of transcription factors to cccDNA. Levels of several transcription factors associated with cccDNA or GAPDH or MYH6 promoter, including COUP‐TF1, HNF4α, YY1, NF‐κB (p65), CREB1, and HNF3α, were examined by ChIP assay with the indicated antibodies. ^** P < 0.01. Abbreviation: IgG, immunoglobulin G.

In addition, we examined whether SIRT3 could regulate the binding of host RNA Pol II and transcription factors to cccDNA. cccDNA is transcribed by the host Pol II, whose occupancy on cccDNA is reported to be associated with enrichment of active posttranslational modifications on cccDNA chromatin. As expected, the enrichment of Pol II onto cccDNA was markedly decreased in HBV‐infected HepG2‐NTCP cells overexpressing SIRT3 (Fig. 5D). A panel of transcription factors such as nuclear receptor subfamily 2 group F member 1 (COUP‐TF1), hepatocyte nuclear factor 4 alpha (HNF4α), YY1, nuclear factor kappa B (NF‐κB; p65), cAMP responsive element binding protein 1 (CREB1), and HNF3α was screened by using the ChIP assay. The recruitment of transcription factor YY1 onto cccDNA was reduced in HBV‐infected HepG2‐NTCP cells ectopically expressing SIRT3 (Fig. 5E). The zinc finger–containing transcription factor YY1 can activate or repress the transcription of many viral and cellular genes. To elucidate the exact role of YY1 on HBV transcription, HepAD38 cells was transfected with a plasmid expressing YY1. Ectopic expression of YY1 increased HBV DNA replicative intermediates, total HBV RNAs, and 3.5‐kb RNA, suggesting that YY1 functions as an activator of HBV replication (Supporting Fig. S7A,B). Together, these data suggest that SIRT3 establishes a repressive chromatin structure on HBV cccDNA by regulating the recruitment of histone methyltransferases SUV39H1 and SETD1A to cccDNA and inhibiting the binding of Pol II and transcription factor YY1 to cccDNA, which may contribute to transcriptional repression of HBV cccDNA.

To validate the SIRT3‐mediated transcriptional repression of HBV cccDNA in vivo, SIRT3, the ratio of 3.5‐kb RNA to cccDNA, H3K9Ac, H3K4me3, and H3K9me3 were quantified in liver biopsies from 5 HBeAg‐positive CHB patients and 5 HBeAg‐negative inactive carriers. Consistently, SIRT3 mRNA was decreased in the livers of CHB patients compared with inactive carriers (Fig. 6A). Importantly, the level of SIRT3 on cccDNA was much lower in CHB patients than in the inactive carriers (Fig. 6B). The ratio of 3.5‐kb RNA/cccDNA was elevated in chronic HBV patients (Fig. 6C), which was accompanied by increased levels of H3K9Ac and H3K4me3 but decreased level of H3K9me3 on cccDNA (Fig. 6D‐F). These data suggest the existence of a transcriptional regulatory axis of SIRT3–HBV cccDNA in vivo.

Figure 6: Levels of SIRT3, HBV 3.5‐kb RNA, SIRT3‐cccDNA, H3K9Ac‐cccDNA, H3K4me3‐cccDNA, and H3K9me3‐cccDNA in liver tissue. (A) The mRNA level of SIRT3 in liver tissues was analyzed by real‐time PCR. Liver biopsies from 5 HBeAg‐positive CHB patients and 5 HBeAg‐negative inactive carriers were collected and subjected to RNA extraction. β‐Actin was used as a reference gene, and results are presented as 2^–▵Ct. (B) The level of SIRT3‐associated cccDNA in liver tissues. ChIP assay was performed using antibodies specific for SIRT3. (C) The ratio of 3.5‐kb RNA/cccDNA in liver tissue. HBV cccDNA was extracted from liver samples using the Hirt method. Both 3.5‐kb HBV RNA and cccDNA were analyzed by real‐time PCR. (D‐F) Levels of H3K9Ac, H3K4me3, and H3K9me3 associated cccDNA in liver tissues. ChIP assay was performed using antibodies specific for H3K9Ac, H3K4me3, and H3K9me3, respectively. ^* P < 0.05.

HBx RELIEVES SIRT3‐MEDIATED TRANSCRIPTIONAL REPRESSION OF HBV cccDNA

Our data showed that SIRT3 expression was reduced in HBV‐expressing cells and liver tissues of CHB patients. To determine which viral protein is responsible for SIRT3 down‐regulation, we analyzed SIRT3 protein levels in Huh‐7 cells transfected with expression constructs for different HBV proteins. SIRT3 expression was decreased in cells expressing HBx but not the other HBV proteins (Fig. 7A). To confirm the inhibitory effect of HBx on SIRT3, Huh‐7 or HepG2 cells were transfected with HBV 1.1‐mer replicons in which HBx expression was abrogated. As expected, HBx ablation abolished SIRT3 inhibition in both cell lines (Fig. 7B). Furthermore, ectopic expression of HBx inhibited both mRNA and protein levels of SIRT3 in HepG2 and Huh‐7 cells (Fig. 7C,D). The data suggested that HBx may regulate SIRT3 gene transcription. To test this hypothesis, we first identified the minimal SIRT3 promoter. Various lengths of the SIRT3 5′‐flanking region, including –992/+32(pGL3‐1025), –770/+32(pGL3‐803), –585/+32(pGL3‐618), –396/+32(pGL3‐429), and –166/+32(pGL3‐199), were cloned and transiently transfected into Huh‐7 and HepG2 cells. Construct pGL3‐1025 exhibited maximum luciferase activity, while pGL3‐199 exhibited much lower luciferase activity (Fig. 7E). Cotransfection of Flag‐HBx with various reporter constructs revealed that the HBx responsive element might be located between −770 and –585 bp of the SIRT3 promoter (Fig. 7E). pGL3‐186 (−770/–585) containing a fragment from −770 to –585 bp and pGL3‐1025▵−770/–585 with a fragment from −770 to –585 bp deleted were cloned and cotransfected with HBx in Huh‐7 cells. HBx drastically decreased promoter activity of pGL3‐186, while it had no effect on promoter activity of pGL3‐1025▵−770/–585 with the fragment from −770 to −585 bp deleted. These data further confirm that the HBx responsive element is located between −770 and –585 bp of the SIRT3 promoter (Fig. 7E).

Figure 7: HBx plays a major role in SIRT3 inhibition. (A) Effect of HBV viral proteins on SIRT3 expression. (B) Effect of wild‐type HBV and HBx mutant on SIRT3 expression. Mutant HBx has codon 8 of the HBx gene mutated to a stop codon without affecting the polymerase gene product. (C,D) HBx inhibited SIRT3 mRNA and protein levels. (E) HBx inhibited SIRT3 promoter activity. The promoter activity of the SIRT3 gene was measured using a dual‐luciferase reporter assay (left panel). Huh‐7 cells were cotransfected with pGL3‐1025, pGL3‐803, pGL3‐618, pGL3‐429, pGL3‐199, pGL3‐186 (–770/–585), or pGL3‐1025▵–770/–585 and Flag‐HBx (right panel). ^* P < 0.05, ^** P < 0.01. Abbreviations: Luc, luciferase; mt, mutant; wt, wild type.

Next, we examined whether HBx expression could relieve SIRT3‐mediated transcriptional repression of cccDNA. The ChIP assay revealed that HBx expression resulted in decreased recruitment of SIRT3 onto cccDNA and increased association of H3K9Ac with cccDNA (Fig. 8A). To further confirm the role of HBx in the binding of SIRT3 to cccDNA, we constructed an HBx‐mutant virus in which HBx expression was abrogated. Much more SIRT3 was bound to cccDNA in HepG2‐NTCP cells infected with HBx‐mutant virus than wild‐type HBV, which was accompanied by decreased H3K9Ac association with cccDNA (Fig. 8B). Moreover, HBx expression also led to decreased SUV39H1 binding on cccDNA but increased recruitment of SETD1A (Fig. 8C). A strong reduction of H3K9me3 but an increase of H3K4me3 association with cccDNA was observed in HBV‐infected HepG2‐NTCP cells expressing HBx (Fig. 8C). Importantly, HBx‐mediated regulation of SUV39H1, SETD1A, H3K9me3, and H3K4me3 associated with cccDNA was partially reversed by SIRT3 overexpression, implying that HBx allows the establishment of active chromatin at least in part by suppressing SIRT3. As expected, ectopically expressed HBx induced the binding of Pol II and transcription factor YY1 onto cccDNA. SIRT3 overexpression could partially offset the HBx‐mediated enhancement of Pol II and YY1 recruitment to cccDNA (Fig. 8D). Finally, we examined the relationship between recruitment of SIRT3, HBx, SUV39H1, and SETD1A to cccDNA throughout the period of viral infection. The ChIP assay found that both SIRT3 and HBx could bind to cccDNA 4 days after infection of PHHs with wild‐type HBV (Fig. 8E). Recruitment of SIRT3 to cccDNA was negatively correlated with recruitment of HBx to cccDNA. Moreover, recruitment of SUV39H1 and SETD1A to cccDNA also fluctuated with the binding of SIRT3 to cccDNA (Fig. 8E). Together, these data suggest that HBx expression can relieve SIRT3‐mediated transcriptional repression of HBV cccDNA by inhibiting SIRT3 expression as well as its recruitment to cccDNA.

Figure 8: HBx relieves SIRT3‐mediated transcriptional repression of HBV cccDNA. (A,B) HBx reduced the recruitment of SIRT3 to cccDNA but increased H3K9Ac associated with cccDNA. HepG2‐NTCP cells were infected with HBV particles for 24 hours and then transduced with lentivirus expressing HBx or vector. ChIP assay was performed 3 days after lentiviral transduction (A); HepG2‐NTCP cells were infected with HBx mutant or with wild‐type HBV, and ChIP assay was performed 4 days after HBV infection (B). (C) HBx regulated the recruitment of SUV39H1 and SETD1A to cccDNA. HepG2‐NTCP cells were infected with HBV for 24 hours and then transduced with lentivirus expressing the indicated genes. ChIP assay was performed using the indicated antibodies. (D) HBx regulated the binding of RNA Pol II and YY1 to cccDNA. ChIP assay was performed in HBV‐infected HepG2‐NTCP cells expressing the indicated gene. (E) Binding kinetics of SIRT3, HBx, SUV39H1, and SETD1A to cccDNA in HBV‐infected PHHs. Cells harvested at 4, 7, 10, 13, 16, and 19 days postinfection were used for ChIP assays with the indicated antibodies. ^** P < 0.01. Abbreviations: dpi, days postinfection; IgG, immunoglobulin G; mt, mutant; wt, wild type.

Discussion

SIRT3 is an NAD⁺‐dependent protein deacetylase belonging to the sirtuin family. SIRT3 can regulate many important biological activities such as metabolism, neuroprotection, cardiovascular diseases, cancer, aging, and the life span.²⁹ However, its role in viral infection remains elusive. In this study, we investigated the effect of SIRT3 on HBV replication and found the antiviral activity of SIRT3 on HBV replication through epigenetic regulation. We first demonstrated that SIRT3 overexpression or silencing inhibited or promoted HBV transcription and replication in both an HBV‐infected HepG2‐NTCP cell line and PHHs. Although the differentiation status or cell cycle of hepatocytes could affect HBV replication,³⁰ we found no evidence that SIRT3 significantly regulates the hepatocyte differentiation status or cell cycle (Supporting Fig. S8). Moreover, both real‐time PCR and northern blot revealed that SIRT3 overexpression reduced HBV transcripts. Using 7‐aminoactinomycin D to block RNA transcription, we could demonstrate that SIRT3 inhibited HBV RNA transcription rather than accelerating RNA degradation. HBV cccDNA is the template for transcription of all viral mRNAs and exists as a stable episome organized into a minichromosome by histone and nonhistone proteins. Taken together, these findings link SIRT3 deacetylation function with modulation of cccDNA chromatin.

The HBV cccDNA minichromosome structure provides numerous possibilities for epigenetic control of its transcriptional activity through repressive or activating posttranslational modifications of histones such as acetylation, methylation, and phosphorylation.²⁴ Accumulating evidence suggests that transcriptional activity of cccDNA is subject to epigenetic control. Belloni et al. reported that interferon‐α treatment exhibited a direct antiviral effect in HBV‐infected PHHs by inducing hypoacetylation of cccDNA‐bound histones to decrease HBV transcript levels.³³ Interleukin‐6 treatment in HBV‐infected HepG2‐NTCP cells led to a reduction of cccDNA‐bound histone acetylation paralleled by a rapid decrease in 3.5‐kb RNA and subgenomic RNAs.³⁴ Riviere et al. found that HBx recruitment to the cccDNA minichromosome serves to counteract chromatin‐mediated transcriptional repression established in part by histone methyltransferase SETDB, H3K9me3, and HP1.³⁵ Moreover, the tudor domain protein Spindlin 1 is recruited to cccDNA and inhibits its transcription by modulating histone H4K4 trimethylation at the cccDNA.³⁶ According to a recent study, histone methyltransferase PRMT5 restricts HBV replication through epigenetic repression of cccDNA transcription.³⁷ In the present study, we found that the histone deacetylase SIRT3 is recruited to HBV cccDNA, where it deacetylates H3K9 at the cccDNA without affecting the cccDNA level. Hypoacetylation of H3K9 at the cccDNA is a suppressive modification, correlating with inhibition of HBV transcription. However, SIRT3 overexpression or suppression had no effect on acetylation of histone H4 at the cccDNA.

Besides acetylation, methylation of H3 and H4 is also associated with chromatin structural change to regulate cccDNA transcription. It has been reported that histone deacetylase and methyltransferase could act cooperatively and sequentially to promote heterochromatin formation.²⁸ For instance, the sirtuin member SIRT1 is involved in facultative heterochromatin formation through an intimate functional relationship with the H3K9me3 methyltransferase SUV39H1, a chromatin organization protein. SIRT1 promotes facultative heterochromatin formation by deacetylating H4K16Ac, H3K9Ac, and H1K26Ac and by establishing heterochromatin markers H3K9me3 through interaction with the histone methyltransferase SUV39H1.²⁸ In this study, we also determined whether SIRT3 could affect other histone modifications at the cccDNA. The ChIP assay found that SIRT3 indeed affected the recruitment of histone methyltransferase SUV39H1 and SETD1A to cccDNA. SUV39H1 is reported to preferentially catalyze H3K9me3, while SETD1A preferentially induces H3K4me3. Consistently, the activating modification H3K4me3 is markedly decreased, whereas the suppressive modification H3K9me3 is increased in HBV‐infected cells overexpressing SIRT3. Together, these data reveal that SIRT3 is associated with the establishment of a repressive chromatin structure and transcriptional silencing of HBV cccDNA. This transcriptional repression could be mediated through the recruitment of SUV39H1 and SETD1A, leading to a concomitant increase of H3K9me3 and decrease of H3K4me3 on cccDNA (Supporting Fig. S9).

SIRT3 exists in two different forms: a full‐length protein (44 kDa) and a processed protein with the N‐terminal 142 amino acids removed (28 kDa). The subcellular localization and function of SIRT3's two forms have been the subject of some controversy. Full‐length SIRT3 is proposed to be inactive until it is translocated and proteolytically processed within the mitochondrion to the 28‐kDa form.²⁹ Early studies reported that processed SIRT3 was localized exclusively to the mitochondria, and this localization may have important implications in SIRT3 function.²⁵ It has also been reported that nuclear full‐length SIRT3 is subject to rapid degradation under conditions of cellular stress, whereas the mitochondrial processed form is unaffected.⁴² However, several other studies found that SIRT3 could reside in the nucleus to exert epigenetic control by deacetylating histone substrates.⁴¹ Scher et al. reported that both the full‐length and processed forms of SIRT3 in the nucleus have NAD⁺‐dependent histone deacetylase activity.¹⁴ They further found that SIRT3 could translocate from the nucleus to mitochondria upon cellular stress, such as ultraviolet irradiation, etoposide, and overexpression of SIRT3 itself. Nakamura et al. also found that coexpression of SIRT3 and SIRT5 resulted in translocation of SIRT3 protein from the mitochondria to the nucleus, suggesting its function in the nucleus.⁴³ HBV cccDNA regulation is a nuclear event, and we therefore examined the localization of SIRT3 in HepAD38 cells, HepG2‐NTCP cells, and PHHs. We found that although SIRT3 was much more abundant in mitochondria, it was clearly detectable in the nuclear fraction. Both the full‐length and processed forms of SIRT3 were detected in the nucleus, supporting its nuclear function in HBV‐expressing cells. Moreover, SIRT3 overexpression resulted in decreased acetylation of total histone H3 and H4. Importantly, SIRT3 with an extranuclear localization signal exhibited a stronger inhibitory effect on HBV transcription and replication than wild‐type SIRT3 in HBV‐infected HepG2‐NTCP cells. These data suggest that nucleus‐localized SIRT3 exhibits histone deacetylase activity and plays a major role in the regulation of HBV transcription and replication.

Having established the impact of SIRT3 on HBV RNA transcription from the cccDNA template, we further examined the effect of HBV replication on SIRT3 expression. We found that HBx protein could inhibit SIRT3 expression by targeting its promoter region. Overexpression of HBx decreased recruitment of SIRT3 and SUV39H1 onto cccDNA but increased recruitment of SETD1A, which led to a concomitant decrease of H3K9me3 and an increase of H3K4me3 and H3K9Ac on cccDNA. Interestingly, the time course ChIP assay found that both endogenous SIRT3 and HBx could bind to cccDNA 4 days after infection of PHHs with wild‐type HBV and that the recruitment of SIRT3 on cccDNA negatively correlated with recruitment of HBx. These data suggested that HBx expression could relieve SIRT3‐mediated transcriptional repression of HBV cccDNA by inhibiting SIRT3 expression as well as its recruitment to cccDNA. This finding is consistent with other groups' reports. It has been suggested that HBx is needed during infection to counteract a cellular repressive response that takes place after the start of viral transcription. Lucifora et al. reported that HBx is not required for the establishment of nuclear HBV cccDNA in the early phase of infection but is essential for initiating and maintaining transcription from cccDNA.²⁴ If HBV transcription from cccDNA starts shortly after infection, all HBV proteins including HBx would be translated. Subsequently, HBx could prevent inhibition of HBV transcription by a cellular mechanism, thus up‐regulating its own expression in a “positive‐feedback loop.” Based on the above findings, we hypothesized that host restriction factors might inhibit HBV transcription by binding to cccDNA before HBx, although no direct evidence exists. Under our experimental conditions, both SIRT3 and HBx could bind to cccDNA 4 days after infection of PHHs with wild‐type HBV. Therefore, current data do not establish whether HBx or SIRT3 binds to cccDNA first or if they bind to cccDNA simultaneously. Collectively, these data formed the basis for a model of HBV transcriptional activation, in which HBx recruitment to the HBV cccDNA minichromosome serves to counteract chromatin‐mediated transcriptional repression established in part by SIRT3, SUV39H1, and H3K9me3 by inhibiting both SIRT3 expression and its recruitment to cccDNA (Supporting Fig. S9).

The current study enhances our understanding of the mechanism of HBV replication, especially on HBV RNA transcription from the cccDNA template. The cccDNA is highly stable, and its persistence accounts for the difficulty in eradicating chronic HBV. As a result, there has been considerable interest in developing therapeutic approaches that directly target cccDNA for elimination. Further study is needed to determine whether SIRT3 could serve as an anti‐HBV target and to assess the possible therapeutic application of SIRT3 activators in anti‐HBV therapy.