Screening of an epigenetic compound library identifies BRD4 as a potential antiviral target for hepatitis B virus covalently closed circular DNA transcription

Abstract

HBV cccDNA is the persistent form of viral genome, which exists in host cell nucleus as an episomal minichromosome decorated with histone and non-histone proteins. cccDNA is the authentic viral transcription template and resistant to current antivirals. Growing evidence shows that the transcriptional activity of cccDNA minichromosome undergoes epigenetic regulations, suggesting a new perspective for anti-cccDNA drug development through targeting histone modifications. In this study, we screened an epigenetic compound library in the cccDNA reporter cell line HepBHAe82, which produces the HA-tagged HBeAg in a cccDNA-dependent manner. Among the obtained hits, a bromodomain-containing protein 4 (BRD4) inhibitor MS436 exhibited marked inhibition of cccDNA transcription in both HBV stable cell line HepAD38 and HepG2-NTCP or primary human hepatocyte infection system under noncytotoxic concentrations. Chromatin immunoprecipitation (ChIP) assay demonstrated that MS436 dramatically reduced the enrichment of H3K27ac, an activating histone modification pattern, on cccDNA minichromosome. RNAseq differential analysis showed that MS436 does not drastically change host transcriptome or induce any known anti-HBV factors/pathways, indicating a direct antiviral effect of MS436 on cccDNA minichromosome. Interestingly, the MS436-mediated inhibition of cccDNA transcription is accompanied by cccDNA destabilization in HBV infection and a recombinant cccDNA system, indicating that BRD4 activity may also play a role in cccDNA maintenance. Furthermore, depletion of BRD4 by siRNA knockdown or PROTAC degrader resulted in cccDNA inhibition in HBV-infected HepG2-NTCP cells, further validating BRD4 as an antiviral target. Taken together, our study has demonstrated the practicality of HepBHAe82-based anti-HBV drug screening system and provided a proof-of-concept for targeting HBV cccDNA with epigenetic compounds.

1. INTRODUCTION

Hepatitis B virus (HBV) infection remains a substantial public health burden with approximately 250 million chronic carriers worldwide (Revill et al., 2019). Chronic HBV infection (CHB) causes a wide spectrum of liver diseases, including chronic hepatitis, liver fibrosis, cirrhosis, and hepatocellular carcinoma, which result in nearly 1 million deaths per year (Alter et al., 2018). To reduce the possibility of severe liver disease progression, therapeutic treatments suppressing HBV replication and ultimately eradicating HBV infection are warranted.

HBV is a non-cytopathic, hepatotropic virus belonging to the Hepadnaviridae family (Seeger and Mason, 2000). The virion particle contains a 3.2kb partially double-stranded relaxed circular (rc) DNA, which has a complete minus strand and an incomplete plus strand (Block et al., 2007). Upon infection, HBV virion gets endocytosed through binding to the hepatic sodium taurocholate co-transporting polypeptide (NTCP) and the viral rcDNA is translocated into the nucleus and converted into the covalently closed circular (ccc) DNA (Marchetti and Guo, 2020; Yan et al., 2012). The cccDNA is decorated with histone and non-histone proteins to form the episomal HBV minichromosome (Bock et al., 1994; Bock et al., 2001; Hong et al., 2017). Using cccDNA as the authentic HBV transcription template, the host RNA polymerase II synthesizes five overlapping viral mRNAs by length, including the 3.5 kb pregenomic (pg) RNA encoding viral core antigen (HBcAg) and reverse transcriptase; 3.5 kb precore (pC) mRNA encoding HBV e antigen (HBeAg); 2.4/2.1 kb preS/S mRNA encoding HBV surface antigen (HBsAg); and 0.7 kb X mRNA encoding HBV x protein (HBx) (Block et al., 2007; Seeger and Mason, 2000). The pgRNA also serves as template for reverse transcription to synthesize the new viral genome. The newly formed rcDNA-containing mature capsid is either enveloped and secreted through the multivesicular bodies as virion or recycled to nucleus for cccDNA replenishment (Guo and Guo, 2015; Marchetti and Guo, 2020; Xia and Guo, 2020). Currently, antiviral treatments of CHB comprise of two strategies, including two forms of IFN-α that modulate the host antiviral immune responses and five nucleos(t)ide analogues (NUCs) that inhibit HBV reverse transcription (Fanning et al., 2019; Hu et al., 2019). However, neither treatment regimen is able to cure HBV infection due to their incapability to eliminate or inactivate cccDNA (Nassal, 2015; Xia and Guo, 2020). Therefore, the development of new antiviral strategies targeting cccDNA biogenesis, stability and transcription is urgently needed.

A growing body of evidence has demonstrated that the transcriptional activity of cccDNA minichromosome undergoes extensive epigenetic regulations, suggesting a new perspective for anti-HBV drug development through targeting cccDNA epigenetics (Hong et al., 2017; Xia and Guo, 2020). Among the known host and viral factors associated with cccDNA, the histones exhibit different kinds of posttranslational modifications (PTMs), which are directed by enzymes including histone acetyltransferases (HAT), histone deacetylases (HDAC) and lysine (K) methyltransferase (KMT) (Hong et al., 2017; Mohd-Ismail et al., 2019). The common histone PTM patterns leading to transcriptional activation includes H3K27ac, H3K4me3, H3K9ac and H3K122ac, whereas the H3K27me3 and H3K9me3 result in transcription repression (Allis et al., 2007; Hong et al., 2017; Kouzarides, 2007; Rodriguez-Paredes and Esteller, 2011). By manipulating cccDNA-associated histone modifications, anti-HBV effect has been demonstrated in recent cell culture studies. It has been reported that protein arginine methyltransferase (PRMT5) repressed cccDNA transcription by inducing H4R3me2 on cccDNA chromatin (Zhang et al., 2017). HDAC11 overexpression reduced HBV transcription by decreasing H3K9ac and H3K27ac of cccDNA (Yuan et al., 2019). In addition, HAT inhibitor C646 can block the access of histone acetyltransferases p300/CBP to cccDNA minichromosome, followed by a reduction of activating PTM H3K27ac and H3K122ac and subsequent inhibition of cccDNA transcription (Tropberger et al., 2015). It has also been demonstrated that cytokines, such as IFN-α and IL-6, suppressed cccDNA transcription by reducing the cccDNA-related histone acetylation (Belloni et al., 2012; Cheng et al., 2020; Liu et al., 2013; Palumbo et al., 2015). In light of the tight relationship between cccDNA transcription and epigenetic regulation, epi-drugs may hold promise to be used to treat HBV infection.

In this study, we screened an epigenetic compound library in HepBHAe82 cells, in which the secreted HA-tagged HBeAg serves as a surrogate marker of intranuclear cccDNA (Cai et al., 2016). Among the validated hits, a bromodomain-containing protein 4 (BRD4) inhibitor MS436 exhibited the most significant antiviral activity against cccDNA transcription in both HBV inducible cell line and infection system without detectable cytotoxicity at treatment concentrations. Chromatin immunoprecipitation (ChIP) assay demonstrated that MS436 dramatically reduced the association of H3K27ac with cccDNA. Interestingly, the MS436-mediated inhibition of cccDNA transcription is accompanied by cccDNA destabilization in HBV infection and Cre/loxP-based recombinant cccDNA systems, indicating that BRD4 activity may also play a role in cccDNA maintenance. Furthermore, depletion of BRD4 by siRNA knockdown or PROTAC degrader resulted in cccDNA inhibition in HBV-infected HepG2-NTCP cells, further validating BRD4 as an antiviral target. Taken together, our study has demonstrated the practicality of HepBHAe82-based anti-HBV drug screening system and provided a proof-of-concept for targeting HBV cccDNA with epigenetic compounds.

2. MATERIALS AND METHODS

2.1. Cell cultures

HepG2 cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM)/F-12 (Gibco) with 10% fetal bovine serum (FBS), 100 U/ml penicillin and 100 μg/ml streptomycin supplemented. The tetracycline (tet)-inducible HBV stable cell line HepBHAe82 and HepAD38 cells were cultured in the same condition as HepG2 cells with additional 1 μg/ml tet and 500 μg/ml G418 (Cai et al., 2016; Ladner et al., 1997). To initiate HBV replication and cccDNA formation in HepBHAe82 and HepAD38 cells, tet was withdrawn from the culture medium and the cells were cultured for the indicated time period. The HepHA-HBe4 stable cell line constitutively expressing HA-HBeAg was maintained in the same way as HepG2 cells with the addition of 500 μg/ml G418 (Cai et al., 2016). The HepG2-NTCP12 cell line supporting HBV infection was maintained in the same way as HepG2 cells but with the addition of 8 μg/ml blasticidin (Yan et al., 2015). The primary human hepatocytes (PHH) were obtained from the Biospecimen Repository and Processing Core of Pittsburgh Liver Research Center.

2.2. Compounds

The epigenetic screening compound Library (96-well format) was purchased from Cayman Chemical (item# 11076, batch# 0479267), each compound was provided in DMSO at a concentration of 10 mM. The library consists of 146 compounds from four major categories according to their enzyme targets, including 18 histone acetyltransferase (HAT) inhibitors, 55 histone deacetylase (HDAC) inhibitors, 32 histone lysine methyltransferase (KMT) inhibitors, 13 histone lysine demethylases (KDM) inhibitors, and 28 other epigenetic compounds. Compound hits UNC0642, CAY10433, and MS436 were repurchased from Cayman Chemical. BRD4 degrader dBET1 (cat# SML2687) and Furin inhibitor I (cat# 344930) were purchased from Sigma-Aldrich. BRD4 degrader MZ-1 (cat# S8889) and HBV replication inhibitor lamivudine (3TC) (cat# S1706) were purchased from Selleck Chemicals.

2.3. Epigenetic compounds library screening

HepBHAe82 cell were cultured in the presence of tet until confluent, then tet was withdrawn to induce HBV replication for 14 days. Then the cells were trypsinized and seeded into 96-well plates at a density of 1.25×10⁵ cells/well in the presence of 1 μg/ml tet and 10 μM 3TC to block de novo cccDNA synthesis. Immediately after cells attached to the plate, compounds were added to screening plates at a final concentration of 10 μM, DMSO (0.1%) served as a solvent control. The treatment was repeated every other day for a total of 4 days. After treatment, the supernatant was collected, clarified by centrifugation at 10,000 rpm for 2 min, and subjected to Chemiluminescence ELISA (CLIA) detection of HA-HBeAg as described previously (Cai et al., 2016). The remaining cells on assay plates were stained by crystal violet solution (cat# V5265, Sigma-Aldrich) according to manufacturer’s instruction.

For counter screen in HepHA-HBe4 cells, the cells were cultured in 96-well plate until confluent, followed by compound treatment at 10 μM for 4 days, 0.1% DMSO served as a solvent control. The treatment was repeated at day 2, the supernatant samples and cells on assay plates collected at day 4 post-treatment were subjected to HA-HBeAg CLIA and crystal violet staining as above described, respectively.

2.4. Cytotoxicity assay

The cytotoxic effects of screened compounds on HepBHAe82 cells and HepHA-HBe4 cells during the primary and counter screening assays, respectively, were assessed by crystal violet staining as above mentioned. The 50% cytotoxic concentration (CC50) of MS436 is determined by MTT assay. HepG2 cell were seeded in 96-well plates at 1×10⁴ cells/well. The treatment of MS436 in a serial dilution started after cells attached and repeated every other day for 4 days. Then, 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) (cat# M2003, Sigma-Aldrich) were added in the supernatant at a final concentration of 0.5 mg/ml and the plate was incubated at 37°C for 4 h. Next, the supernatant was gently removed by vacuum and 150 μl DMSO was add to dissolve reaction product with a 30 min incubation at 37°C. The colorimetric absorbance was measured at 570 nm with BioTek Synergy HTX multimode plate reader.

2.5. HBV infection

The infectious HBV particles were collected from the supernatant of induced HepAD38 cells and the virion genome equivalent (vge) was determined as described previously (Yan et al., 2015). HBV in vitro infection was conducted as previously published with minor modifications (Mitra et al., 2019). Briefly, HepG2-NTCP or PHH cells were seeded in collagen-coated 6-well plates at a density of 1.2×10⁶ cells per well, and culture medium was switched to Cellartis hepatocyte maintenance medium (HMM) (cat# Y30051, TaKaRa) on the second day. After 24 h, the cells were inoculated with HBV at indicated vge/cell in HMM containing 4% PEG-8000 for 24 h, then the infected cells were cultured with hepatocyte maintenance medium (PMM) for indicated time duration (Mitra et al., 2019).

2.6. Immunofluorescence assay (IFA)

HepG2-NTCP cells were infected by HBV for 3 days, followed by fixation with 4% paraformaldehyde for 20 min and permeabilized in 0.5% Triton X-100 1×PBS solution for 1 h at room temperature. Subsequently, cells were blocked with IFA blocking buffer (10% FBS plus 2% bovine serum albumin in 1×PBS) for 1 h at room temperature and incubated with rabbit anti-HBcAg (cat# B0586, Dako) at 4°C for overnight. After washing by 1×PBS and a second blocking for 30 min, cells were incubated with Alexa Fluor 488 dye-conjugated goat anti-rabbit secondary antibody (cat# A11070, Invitrogen), and cell nuclei were counterstained with DAPI (4’,6-diamidino-2-phenylindole) for 30 min at room temperature. Both primary and secondary antibodies were diluted in IFA blocking buffer. The cells were washed with 1×PBS and subjected to Olympus FV1000 microscopy analysis under 20× objective lens. Images were analyzed using Image J software.

2.7. Plasmids and siRNA transfection

The prcccDNA and pCMV-Cre plasmids were kindly provided by Dr. Qiang Deng (Fudan University, China) (Qi et al., 2014). Human BRD4 siRNA (a pool of 3 unique 27-mer siRNA oligo duplexes) was purchased from OriGene (cat# SR323604). Control siRNA was purchased from Santa Cruz Biotechnology (cat# sc-37007). Transfection of HepG2 cells with indicated plasmid(s) was performed by Lipofectamine 3000 (cat# L3000150, Invitrogen) according to the manufacturer’s directions. Transfection of HepG2-NTCP cells with siRNA was performed by Lipofectamine RNAiMAX (cat# 13778150, Invitrogen) according to the manufacturer’s directions.

2.8. HBV DNA and RNA analyses

Intracellular HBV RNA and DNA (cytoplasmic core DNA and Hirt DNA) were extracted and subjected to northern blot and Southern blot, respectively, as described previously (Cai et al., 2013; Liu et al., 2017; Mao et al., 2013). Hybridization signals were exposed to a phosphorimager screen and scanned by Typhoon FLA-7000 imager.

For quantitative PCR (qPCR) detection of HBV cccDNA, the extracted Hirt DNA were heat denatured, followed by Plasmid-safe ATP-dependent DNase (PSAD) (cat# E3101K, Epicentre) treatment to remove the deproteination rcDNA, according to our previous publication (Long et al., 2017). Then, the pre-cleaned Hirt DNA samples were subjected to qPCR detection of HBV cccDNA and cellular mitochondrial DNA as described previously (Long et al., 2017).

For quantitative reverse transcription PCR (qRT-PCR) detection of HBV preCore mRNA, the RNA samples were amplified using LightCycler 480 RNA Master Hydrolysis Probes (Roche) in a 20 μl reaction containing 0.5 μM forward primer (5’-GTAGGCATAAATTGGTCTG-3’ (nt 1785–1803)), 0.5 μM reverse primer (5’-GTGCAGTATGGTGAGGTGAACAAT-3’ (nt 2065–2042)), and 0.25 μM TaqMan probe (5’-/56-FAM/CTCAGGAGACTCTAAGGCTTCCCGATACAG/36-TAMSp/−3’ (nt 2040–2011)) (Kim et al., 2022). The cellular GAPDH mRNA, serving as an internal control, was quantified by qRT-PCR with specific primer-probe mixture (PrimeTime^Ⓡ XL qPCR Assay, Hs.PT.39a.22214836, Exon Location 2–3). The thermal cycling condition was set up as 10 sec at 95°C, 30 sec at 60°C, and 5 sec at 72°C, for 50 cycles.

2.9. Chromatin immunoprecipitation (ChIP)-qPCR

The ChIP assay was performed using the ChIP-IT^Ⓡ Express Chromatin Immunoprecipitation Kit (cat# 53008, Active Motif). First, the cells were fixed with 1% formaldehyde and collected, which were subsequently homogenized in a Dounce homogenizer containing lysis buffer, the cell lysate was then sonicated using EpiShear ultrasound sonicator (cat# 53051, Active Motif) at 20% amplitude with 10 sec impulse and 30 sec rest for 7 cycles in shearing buffer. After centrifugation to remove the cell debris, the chromatin-containing supernatant was divided into input and ChIP samples, the latter was subjected to immunoprecipitation with Protein G Magnetic Beads coated with nonimmune serum control IgG antibody (cat# I8765, Sigma) or ChIP-grade specific antibodies, specifically anti-H3K27ac (cat# ab4729, Abcam), anti-H3K27me3 (cat# ab6002, Abcam), anti-H3K4me3 (cat# ab8580, Abcam), anti-H3K9me3 (cat# 8898, Abcam), anti-BRD4 (cat# 39909, Active Motif). Next, the beads were washed, and the immunoprecipitated chromatin was eluted from the beads, reverse crosslinked, and deproteinized by proteinase K. The input samples were also treated with proteinase K. All the above experimental conditions were set up following the manufacturers’ directions. The final reaction was cleaned up by QIAquick PCR purification kit (cat# 28106, Qiagen), and the immunoprecipitated cccDNA were directly quantified by qPCR as above described. The occupancy of the specific protein on cccDNA was expressed in fold enrichment above IgG control using -ΔΔCq method.

2.10. Western blot assay

Whole cell lysate prepared with Laemmli buffer was resolved in a Novex WedgeWell 4–12%, Tris-Glycine Gel (cat# XP04122, Invitrogen), and proteins were transferred onto an Immobilon-FL polyvinylidene difluoride (PVDF) membrane (cat# IPFL00010, Millipore). The membrane was then blocked with WesternBreeze blocking buffer (cat# WB7050, Invitrogen) and probed with antibody against BRD4 carboxyl-terminus (cat# 39909, Active Motif) and β-actin (cat# sc-47778, Santa Cruz Biotechnology). Bound antibodies were revealed by IRDye secondary antibodies and the signals were visualized using the Odyssey Fc imager system (LI-COR Biosciences).

2.11. Differential transcriptome analyses

HepG2-NTCP cells were infected by HBV for 3 days and treated with solvent control DMSO or 10 μM MS436 every other day for a total of 4 days. After treatment, the total RNA was collected and cleaned up with the RNA Clean & Concentrator-5 (cat# R1015, Zymo Research) to reach OD260/280 ≥ 2.0 and OD260/230 ≥ 2.0. Then, the purified RNA samples in duplicate were submitted to Novogene for mRNA sequencing, and the delivered sequencing data were analyzed as follows: A) RNA-seq preprocessing and alignment: The mapping pipeline steps were: 1) Adapter sequences and low-quality bases were trimmed using Cutadapt. 2) HISAT2 was used to align pair-end reads against human genome reference assembly. 3) Read alignments were sorted and indexed using SAMtools. 4) The aligned mapping was quantified into gene counts against GRCh38 annotation using Rsubread. B) Differential gene expression analysis: Gene count matrix was normalized using median of ratios method and visualized by Principal component analysis (PCA). Outliers were filtered out by visualization. Then, DEseq2 was applied to perform differential gene expression and a gene was considered significant for adjusted p value < 0.05 and expression fold change > 2. C) Functional enrichment: Significant differentially expressed genes were tested against Gene Ontology Biological Process terms for functional enrichment using Webgestalt (webgestalt.org). Default parameters were chosen for the analysis.

2.12. Statistical analysis

Data were analyzed by using GraphPad Prism 9.0 and expressed as mean ± standard deviation (SD). Student’s t test was used to determine the statistical significance at a significant level of p value<0.05.

3. RESULTS

3.1. Compound screening workflow

We have previously established a cccDNA reporter cell line HepBHAe82 suitable for screening of compounds potentially targeting HBV cccDNA transcriptional activity or stability (Cai et al., 2016). In principle, the HepBHAe82 cell line contains an HBV transgene encoding the viral pgRNA sequence under the control of a tet-inducible promoter. The start codon of 5’ pC ORF is mutated and an in-frame HA epitope tag is inserted in to the 5’ precore domain of the split pC ORF on pgRNA, by which the HA-tagged pC (HA-pC) ORF will be reconstituted only when pgRNA is reverse transcribed into rcDNA and subsequently converted into cccDNA, giving rise to the production of intracellular HA-pC mRNA and supernatant HA-HBeAg in a cccDNA-dependent manner (Fig. 1A).

Fig. 1. The screening paradigm for identification of cccDNA epigenetic inhibitors.

(A) Schematic diagram of cccDNA-dependent HA-tagged HBeAg (HA-HBeAg) reporter cell line HepBHAe82. See text and previous publication for details (Cai et al., 2016). (B) Cayman epigenetic compound library consisting of 146 compound was subjected to screening in HepBHAe82. (C) HepBHAe82 cells in 96-well-plate were induced for HBV replication and cccDNA accumulation in Tet- medium for 14 days, followed by treatment with 3TC (10 μM) and each library compound (10 μM) in the presence of Tet, DMSO served as mock control. The treatment was refreshed once on day 2, and the supernatant was collected on day 4 for HA-HBeAg chemiluminescence ELISA. (D) HepHA-HBe4 cell line constitutively expressing HA-HBeAg served as a counter screen to exclude compounds directly targeting HBeAg metabolism or secretion. Cells were treated with DMSO solvent control or library compounds every other day for 4 days, followed by HA-HBeAg chemiluminescence ELISA. (E) Compounds that reduce HA-HBeAg in HepBHAe82 by more than 50% without inhibiting HA-HBeAg production in HepHA-HBe4 cells will be selected as primary hits.

In this study, we made use of the HepBHAe82 cell-based screening system to identify potential HBV cccDNA transcription inhibitors from a compound library comprised of 146 epigenetic compounds (Fig. 1B). To this end, HepBHAe82 cells were induced for HBV replication and cccDNA accumulation for 14 days, then switched to compound treatment for additional 4 days in the presence of tet and 3TC, by which to block de novo synthesis of HBV DNA and allow an exclusive expression of HA-HBeAg from the preexisting cccDNA; cell culture media were collected and subjected to HA-HBeAg CLIA assay (Fig. 1C). The HepHA-HBe4 cell line constitutively expressing HA-tagged HBeAg from the transgene (HA-pC ORF under the control of a CMV-IE promoter) served as counter screen to filter out cccDNA-independent HBeAg inhibitors (Fig. 1D) (Cai et al., 2016). Compounds that reduced HA-HBeAg by more than 50% in HepBHAe82 cells without significantly affecting HA-HBeAg level in HepHA-HBe4 cells were considered as primary hits (Fig. 1E).

3.2. Primary hits identification

Following the screening diagram mentioned above (Fig. 1), we first determined initial hits that reduced HA-HBeAg signal in HepBHAe82 cells by more than 50% compared to DMSO control after the primary screen, and the emerged hits were grouped according to their known enzymatic targets, including 15 HDAC inhibitors, 7 KMT inhibitors, and 4 HAT inhibitors (Fig. 2A). Next, we counter screened the initial hits in HepHA-HBe4 cells to exclude any hits that reduced HA-HBeAg in the absence of cccDNA (Fig. 2B). Then we went back to check the HepBHAe82 cell assay plates stained by crystal violate to further exclude hits with obvious cytotoxicity (the cut-off was set at 80%). It is worth noting that a handful of initial hits exhibited cytotoxicity in the primary screen but not in the counter screen (data not shown), which was likely due to the decreased viability of HepBHAe82 cells after a prolonged induction of HBV replication. After the three-tiered funnel testing, we prioritized the top two compounds from each hit category which exhibited strongest nontoxic inhibitory effect in the HepBHAe82 assay without any inhibitory effect in the HepHA-HBe4 assay, including HDAC inhibitors CAY10433 and CAY10683; KMT inhibitors 3-Deazaneplanocin A and UNC0642, and HAT inhibitors I-CBP112 and MS436, as primary hits of cccDNA transcription inhibitors for further validation.

Fig. 2. Primary hits determination.

The primary hits belonging to histone deacetylase (HDAC) inhibitors, histone lysine methyltransferase (KMT) inhibitors, and histone acetyltransferase (HAT) inhibitors, respectively, were determined based on their ability to reduce HA-HBeAg by more than 50% in HepBHAe82 screening assay (A) but without significantly reducing HA-HBeAg in HepHA-HBe4 counter-screening assay (B) compared to DMSO solvent as described in Fig. 1.

3.3. Hit validation

We further tested the primary hits in another inducible HBV stable cell line HepAD38, in which the authentic pC mRNA is transcribed in a cccDNA-dependent manner as we previously reported (Zhou et al., 2006). Similar to the compound screening in HepBHAe82 cells, the compound treatment was conducted on the pre-induced HepAD38 cells in the presence of tet and 3TC, and cccDNA and pC mRNA were quantified by qPCR. None of the six primary hits showed significant reduction of cccDNA level, while several compounds even upregulated cccDNA compared to DMSO control (Fig. 3A). The levels of pC mRNA were reduced to various extents by different compounds (Fig. 3B). One primary hit, 3-Deazaneplanocin A, failed to reduce the level of cccDNA-dependent pC mRNA in HepAD38 cells. While we did not further investigate this observation, a plausible explanation is that this compound behaved differently in two HBV cell lines (HepBHAe82 vs HepAD38) or the compound inhibited HA-HBeAg production through inhibiting the translation from the authentic viral pC mRNA in HepBHAe82 cells. After normalizing the levels of pC mRNA to cccDNA, compound CAY10433, CAY10683, UNC0642 and MS436 exhibited significant inhibitory effect on cccDNA-based transcription (Fig. 3C). Since both CAY10433 and CAY10683 are HDAC inhibitors and possess a similar antiviral activity against cccDNA, only CAY10433 was subjected to further validation together with KMT inhibitor UNC0642 and HAT inhibitor MS436. The repurchased individual compounds were then retested in HepBHAe82 cells for their effects on cccDNA-dependent HA-HBeAg production. As shown in Fig. 3D, all the three validated hits markedly reduced the HA-HBeAg level compared to DMSO control.

Fig. 3. Primary hits validation.

Primary hits generated from HepBHAe82 screening assay were further validated in HepAD38 cell line. HepAD38 cells were cultured in Tet- media to induce HBV replication and cccDNA synthesis for 10 days, and then treated with 10 μM of primary hit compounds every other day for 4 days in the presence of Tet and 3TC (10 μM), DMSO served as solvent mock treatment control. (A) HBV Hirt DNA was extracted and subjected to cccDNA quantification by qPCR. Relative cccDNA levels are plotted as fold change to DMSO control. (B) Total RNA was extracted and subjected to HBV preCore (pC) mRNA qPCR. Relative pC mRNA levels are plotted as fold change to DMSO control. (C) The relative levels of cccDNA-based transcription were further determined by normalizing pC mRNA level to cccDNA level in each treatment group compared to DMSO control group. (D) The validated hits UNC0642, CAY10433, and MS436 were repurchased and their antiviral effect on cccDNA transcription were further verified in HepBHAe82 cells by HA-HBeAg CLIA assay as described above. RLU: relative luminescence units. All data are shown as mean ± SD (n=3). *p<0.05; **p<0.01; ***p<0.001.

Furthermore, we assessed the antiviral effect of the validated hits in HBV-infected HepG2-NTCP cells, in which HBV cccDNA serves as the genuine viral transcription template. As shown in Fig. 4, CAY10433, UNC0642, and MS436 markedly reduced HBV core protein (HBc) expression compared to DMSO control, which was consistent with the compounds’ activities in HepBHAe82 and HepAD38 stable cell lines. In light of the fact that MS436 consistently exhibited the strongest antiviral activity in the screening and validation assays, it was prioritized as a lead compound for further characterization and mechanistic studies.

Fig. 4. Antiviral assessment of validated hits in HBV infection system.

HepG2-NTCP cell were infected with HBV at 250 vge/cell for 3 days, followed by treatment with DMSO control, compound UNC0642, CAY10433 or MS436 at 10 μM for 4 days in the presence of 10 μM 3TC. HBV cccDNA-based gene expression was assessed by HBc immunofluorescence, the average percentage of HBc-positive cells were calculated from five microscope field of view. Nuclei were stained with DAPI. Scale bar: 50 μm.

3.4. CC50/EC50 of MS436

MS436 is diazobenzene-based small-molecule compound developed as a potent BRD4 bromodomain inhibitor (Fig. 5A) (Zhang et al., 2013). Previous studies have shown that MS436 demonstrated micromolar-range inhibition of BRD4-dependent host functions in murine and human cells (Barrow et al., 2016; Di Micco et al., 2014; Wang et al., 2017; Zhang et al., 2013), but the antiviral activity of MS436 in hepatocyte-derived cells has not been reported. In this study, considering that HepG2 cell line is the parental cell for both HepBHAe82 and HepHA-HBe4 cells and subsequent studies will be conducted in other HepG2-derived cells, such as HepG2-NTCP-based HBV infection system, we thus conducted MTT assay in MS436-treated HepG2 cells and the CC50 of MS436 was determined at 80.62±4.89 μM (Fig. 5B). In addition, MS436 treatment reduced the production of HA-HBeAg in HepBHAe82 cell line in a dose dependent manner. Furin inhibitor I, a known inhibitor of HBeAg secretion (Cai et al., 2016; Ito et al., 2009; Messageot et al., 2003), was included as a background signal control for normalization purpose. Based on the progressive decreased level of HA-HBeAg with the increased MS436 concentrations, the EC50 of MS436 for inhibiting HBV transcription was calculated as 15.55±6.09 μM (Fig. 5C). Based on the relatively narrow window between cytotoxicity and antiviral activity of MS436, we chose 10 μM as the working concentration for MS436 in further studies.

Fig. 5. CC50/EC50 of MS436.

(A) The chemical structure of compound MS436. (B) CC50 measurement of MS436. HepG2 cell were treated with 2-fold serially diluted MS436 (0 μM to 100 μM) every other day for 4 days, cell survival rate (% survival) was detected by MTT assay and plotted as percentage of DMSO control. The CC50 value was calculated through plotting % survival versus log concentration of MS436 via variable slope linear regression. (C) Determining EC50 of MS436. A range of experimental concentrations (0 μM to 200 μM, 2-fold serial dilutions) of MS436 were used to treat HepBHAe82 cells after tet- induction for 14 days. The furin inhibitor 1 (1 μM) which can efficiently block the secretion of HA-HBeAg served as control for background signal. Treatment was repeated every other day for 4 days in the presence of Tet and 3TC (10 μM). Supernatant samples were collected at treatment endpoint and subjected to HA-HBeAg CLIA analysis. HA-HBeAg CLIA signals were normalized by background signal and plotted as percentage of DMSO control. The EC50 was calculated through plotting % HA-HBeAg CLIA versus log concentration of MS436 via variable slope linear regression.

3.5. Antiviral assessment of MS436 in HBV infection in vitro

As above mentioned, MS436 inhibited cccDNA transcription in HBV stable cell lines and reduced HBc expression in HBV-infected HepG2-NTCP cells (Fig. 3–4). In order to specifically study the effect of MS436 on the preexisting cccDNA pool in the context of HBV infection, HepG2-NTCP cells with established HBV infection were treated with MS436 or DMSO control in the presence of 3TC to block HBV replication-mediated cccDNA replenishment. As shown in Fig. 6, MS436 treatment resulted in a marked reduction of HBV total RNA as revealed by northern blot (Fig. 6A), and the qPCR analysis also demonstrated a significant decrease of pC mRNA by MS436 treatment (Fig. 6B); a modest reduction of cccDNA (~30%) upon MS436 treatment was observed (Fig. 6C); after normalizing pC mRNA to cccDNA, the MS436-mediated inhibition of cccDNA transcription activity remained significant (Fig. 6D). The slight to modest reduction of cccDNA in both HepAD38 and HepG2-NTCP cells upon MS436 and 3TC treatment might be due to a possibility that the stability and transcription activity of cccDNA minichromosomes are coupled (Fig. 3A and 6C). Next, MS436 was further tested in PHHs, the gold standard for hepatic in vitro culture models (Schulze-Bergkamen et al., 2003). As shown in Fig. 7, in PHHs with established HBV infection, MS436 mono treatment markedly reduced HBV RNA transcription and the subsequent viral core DNA replication, which in turn resulted in further reduction of cccDNA due to the decrease of rcDNA available to intracellular cccDNA amplification pathway. On the other hand, the level of mitochondrial DNA (mtDNA) was unchanged upon MS436 treatment, indicating a noncytotoxic antiviral effect.

Fig. 6. Antiviral assessment of MS436 in HBV-infected HepG2-NTCP cells.

HepG2-NTCP cells were infected with HBV (250 vge/cell) for 3 days, followed by treatment with DMSO solvent control and 10 μM MS436 every other day for 4 days in the presence of 10 μM 3TC. (A) HBV total RNA were extracted and detected by northern blot, 3.5 kb pC mRNA and pgRNA, 2.4/2.1 kb surface mRNAs were labeled. Ribosomal RNA (rRNA) served as loading control. (B) HBV pC mRNA was detected by qPCR. The relative pC mRNA levels after MS436 treatment were plotted as fold change to DMSO control. (C) Hirt DNA was extracted, heat denatured and digested by PSAD to remove non-cccDNA species, followed by cccDNA qPCR. The relative cccDNA levels after MS436 treatment were plotted as fold change to DMSO control. (D) The relative transcription activity of cccDNA was presented by normalizing pC mRNA level to cccDNA level and plotted as fold change to DMSO control group. Data are shown as mean ± SD (n=3). *p<0.05, ***p<0.001.

Fig. 7. Assessment of antiviral effect of MS436 in HBV-infected primary human hepatocytes (PHHs).

Primary human hepatocytes were infected with HBV at 500 vge/cell for 3 days, followed by treatment by solvent DMSO or MS436 (10 μM) every other day for 4 days. Total Hirt DNA was subjected to cccDNA Southern blotting, mitochondrial DNA (mtDNA) served as Southern blot loading control. Total RNA was subjected to HBV RNA northern blotting, 3.5kb pC mRNA and pgRNA and 2.4/2.1kb surface mRNA were labeled. 28S and 18S rRNA served as loading control. HBV cytoplasmic core DNA replicative intermediates were detected by Southern blot. RC: relaxed circular DNA; SS: single stranded DNA.

3.6. MS436 inhibits cccDNA transcription via epigenetic mechanism

BRD4 belongs to the bromodomain and extra-terminal domain (BET) protein family and functions as a histone acetylation reader via binding to the acetylated lysine residues of histone H3 and H4 on chromosome DNA, by which regulates gene transcription (Wu and Chiang, 2007). Two major forms of BRD4 protein have been identified, including the long isoform (BRD4-L) and the short isoform (BRD4-S), both of which contain two tandem bromodomains serving as binding domain of acetylated lysine and target of MS436 (Drumond-Bock and Bieniasz, 2021). We thus examined the effect of MS436 on HBV cccDNA-associated epigenetic histone posttranslational modification (PTM) markers by ChIP-qPCR assay. In HepAD38 cells with established cccDNA, MS436 treatment significantly reduced the association of active histone marker H3K27ac with cccDNA (Fig. 8A), whereas slightly increased the level of repressive histone repressive marker H3K9me3 on cccDNA (Fig. 8B), indicating that MS436 induced an epigenetically inactive stage of cccDNA transcription. However, MS436 treatment did not obviously alter the association of another active histone PTM H3K4me3 with cccDNA (Fig. 8C), indicating that the inhibitory effect of MS436 on cccDNA is histone acetylation-specific, which is consistent with the substrate specificity of BRD4 reader. Interestingly, MS436 treatment only slightly decreased the association of BRD4 with cccDNA (Fig. 8D), which indicated that the binding of BRD4 to cccDNA minichromosome might be largely independent of the interaction between bromodomains and acetylated histones. Furthermore, MS436 treatment did not reduce the steady state level of total BRD4 proteins (Fig. 8E), suggesting that MS436 inhibits the histone acetylation reader function of BRD4 without affecting BRD4 stability.

Fig. 8. MS436-mediated epigenetic change of cccDNA minichromosome.

(A-D) HepAD38 cells cultured in T75 flask were induced in Tet- medium for 10 days for cccDNA establishment, then treated with DMSO or MS436 (10 μM) every other day for 4 days in the presence of 3TC (10 μM). Cells were crosslinked and lysed for further chromatin fragmentation by sonication-based shearing. cccDNA was immunoprecipitated by ChIP-grade non-immune IgG isotype control or antibodies against H3K27ac, H3K9me3, H3K4me3 or BRD4 and detected by qPCR. The enrichment of aforementioned histone PTMs or BRD4 protein on cccDNA was plotted as fold change to IgG control, respectively. Data are shown as mean ± SD (n=3). **p<0.01, ***p<0.001, ns: not significant. (E) The intracellular total BRD4 proteins, including both the long and short isoforms, were detected by western blot. β-actin served as the loading control.

3.7. MS436 inhibits HBV transcription without drastically altering the host transcriptome

Considering that MS436 targets a host epigenetic factor BRD4, it is possible that the observed anti-cccDNA effect by MS436 is partially due to host function alterations. To assess this possibility, we conducted cell transcriptome analysis of HBV-infected HepG2-NTCP cells upon MS436 treatment. As shown in Fig. 9, no drastic alteration of cellular gene transcripts was induced by MS436 treatment, only 20 differentially expressed genes (DEGs) showed increased or reduced expression with fold change more than log2 compared to DMSO mock treatment, and none of them was liver-specific transcription factors or reported HBV transcription activator or suppressor (Fig. 9A–B). Furthermore, the functional enrichment analysis showed no significant pathway enrichment related to cell signaling pathways that have been reported to inhibit HBV transcription (Fig. 9C), to name a few, the Ras-MAPK (Zheng et al., 2003), PI3K-AKT (Guo et al., 2007b), and IFN signaling pathway (Belloni et al., 2012), etc. The above results indicate that the observed antiviral effect of MS436 on cccDNA transcription is largely through directly targeting cccDNA minichromosome rather than secondary effects from altering host gene expression, at least under the 4-day treatment duration in this study.

Fig. 9. Differential transcriptome analyses of MS436-treated HBV-infected cells.

HepG2-NTCP cells were infected with HBV (250 vge/cell) for 3 days, followed by treatment with DMSO solvent control and 10 μM MS436 every other day for 4 days in the presence of 10 μM 3TC. The experiment was performed in duplicate. Then, total RNA was extracted and subjected to RNA seq and the following Differential transcriptome analyses. (A) Volcano plot. The log2 Fold Change indicates the mean expression level change for each gene. Q-value indicates the FDR adjusted p-value for each comparison. Each dot represents one gene while red dots are significant DEGs (fold change >2 and Q-value < 0.05) and black dots are insignificant DEGs. (B) Expression heatmap of significant DEGs. Expressions are normalized by the median of ratios method and then log-transformed for visualization. (C) Functional enrichment. Each row shows one function from Gene Ontology Biological Process. “Hits” indicates the number of enriched genes in each term. Significance is converted from the p-value of each enrichment and colored from blue (less significant) to red (more significant).

3.8. MS436 reduces the transcription activity and stability of a recombinant cccDNA.

We further tested the inhibitory effect of MS436 on cccDNA in a Cre/loxP-based recombinant cccDNA (rcccDNA) system. As previously described (Qi et al., 2014), the rcccDNA system was created by inserting a loxP-chimeric intron into the monomeric 3.2 kb HBV genome in a precursor plasmid prcccDNA, which, in the presence of Cre recombinase, will undergo Cre/loxP-mediated DNA recombination to remove the plasmid backbone and form the 3.3 kb rcccDNA, and the remaining loxP-containing intron will be spliced from rcccDNA-derived RNA transcripts without interrupting the subsequent reverse transcription (Fig. 10A).

Fig. 10. Antiviral assessment of MS436 in the rcccDNA system.

(A) Schematic illustration of rcccDNA system. The 5.4 kb prcccDNA is a precursor plasmid containing a monomeric HBV genome with a loxP-chimeric intron. In the presence of pCMV-Cre, prcccDNA will be turned into a 3.3 kb rcccDNA episome by Cre/loxP-mediated DNA recombination. The loxP-containing intron in rcccDNA-derived transcripts will undergo splicing to produce authentic HBV mRNAs for completing viral replication cycle. See text and previous publication for details (Qi et al., 2014). (B) Demonstration of the synthesis of rcccDNA. HepG2 cells in 6-well-plate were transfected with prcccDNA (1.5 μg) and pCMV-Cre plasmid together (lane 2–4), followed by Hirt DNA extraction and Southern blot assay. The open circular (oc), double stranded linear (dsl), and covalently closed circular (ccc) forms of prcccDNA, and the Cre/loxP-mediated rcccDNA are indicated on the blot (lane 2). Hirt DNA sample was heat denatured at 95 °C for 5 min before gel loading, which resulted in the shift-down of oc and dsl forms of prcccNA to the single stranded (ss) position (lane 3). The heat denatured sample was further digested by EcoRI before gel loading, which resulted in the shift-up of ccc form of both prcccDNA and rcccDNA to their corresponding dsl position (lane 4). In a separate experiment, HepG2 cells were transfected by pCMV-Cre or prcccDNA alone for 3 days, Hirt DNA was extracted, heat denatured, then further digested by EcoRI as described above, and subjected to Southern blotting (lane 5–6). The dsl and ss forms of prcccDNA are indicated by arrow lines, the plausible undigested prcccDNA is indicated by an asterisk. 100 μg untransfected prcccDNA were left untreated (lane 7) or heat denatured and digested by EcoRI (lane 8), and then subjected to Southern blot assay as a DNA form/size reference. The oc, ccc, dsl, and ss forms of prcccDNA are indicated, the asterisk indicates a topological isomer of the ccc form of prcccDNA or degradation product (lane 7–8). (C) MS436 inhibited rcccDNA-based HBV RNA transcription and DNA replication. HepG2 cells were transfected with prcccDNA and pCMV-Cre for 3 days, and then treated by DMSO solvent or MS436 (10 μM) for additional 4 days, followed by HBV RNA northern blot and DNA Southern blot analyses. The hybridization signals of HBV total RNA and core DNA were quantified, and the relative levels were plotted as percentage of DMSO control. (D) MS436 selectively reduced rcccDNA stability. HepG2 cells were co-transfected with prcccDNA and pCMV-Cre plasmids, cultured for 3 days post transfection and then treated with DMSO or MS436 every other day for 4 days in the presence of 3TC. The heat denatured, EcoRI-digested Hirt DNA was subjected to Southern blot. mtDNA served as loading control. The hybridization signals of prcccDNA and rcccDNA were quantified and the relative levels were plotted as percentage of DMSO control. (E) MS436 promoted the decay of rcccDNA. HepG2 cells were co-transfected with prcccDNA and pCMV-Cre for 3 days, followed by treatment with 3TC alone or 3TC plus MS436 daily for 3 days. Cells were collected before the treatment (D0) and at day 1, 2, and 3 post-treatment (D1-D3). The heat denatured, EcoRI-digested Hirt DNA was subjected to Southern blot. mtDNA served as loading control. The hybridization data shown here is a representative of two experimental trials. The hybridization signals of prcccDNA and rcccDNA were quantified and the relative levels were plotted as percentage of the signal of D0 sample in each treatment groups, respectively.

To produce rcccDNA, HepG2 cells were transfected with prcccDNA and pCMV-Cre individually or in combination for 3 days, followed by Hirt DNA extraction and Southern blot analyses. As shown in Fig. 10B, rcccDNA was formed through Cre/loxP-mediated recombination of prcccDNA precursor, coexisting with the remaining prcccDNA input and intermediates, including the covalently closed circular (ccc), double-stranded linear (dsl), and open circular (oc) forms (lane 2, from bottom to top); the identities of the above DNA molecules were validated by 95°C heat denaturation, which the oc and dsl forms of prcccDNA were shifted down to the single stranded (ss) position with the ccc form of prcccDNA and rcccDNA being unaffected (lane 3); the latter two were linearized by EcoRI digestion and shifted up to their corresponding dsl positions (lane 4). As expected, the pCMV-Cre transfection did not generate any HBV DNA signal (lane 5); the prcccDNA alone did not produce rcccDNA upon transfection, and the recovered Hirt DNA was processed into ss and dsl forms of prcccDNA after 95°C heat denaturation and EcoRI digestion (lane 6, asterisk indicates the undigested prcccDNA). The original prcccDNA stock was untreated or subjected to 95°C heat denaturation and EcoRI digestion, serving as a reference for the transfected cells-derived DNA analyses (lane 7–8, asterisk indicates a topological isomer of prcccDNA or degradation product). Previous studies have also demonstrated that rcccDNA were epigenetically organized as minichromosomes like the genuine HBV cccDNA (Qi et al., 2014; Wang et al., 2021). Thus, the rapid accumulation of high amount of rcccDNA in transfected cells makes the rcccDNA system a suitable surrogate for studying cccDNA transcription and maintenance.

MS436 treatment of HepG2 cells harboring an established rcccDNA pool resulted in a significant drop of both HBV total RNA and core DNA (Fig. 10C), together with a reduction of rcccDNA level to a lesser extent (Fig. 10D), which were consistent to above observations from HBV stable cell lines and infection systems (Fig. 3, 6–7). Interestingly, the levels of prcccDNA species and cellular mtDNA were largely unchanged upon MS436 treatment, indicating a MS436-mediated rcccDNA-specific destabilization (Fig. 10D). To further assess this possibility, a time course experiment was conducted to measure the decay kinetics of rcccDNA and its prcccDNA precursor in the presence of 3TC with or without MS436 treatment. As shown in Fig. 10E, MS436 treatment did not obviously change the decay kinetics of prcccDNA or mtDNA calculated on a daily basis for the duration of 3 days, while MS436 accelerated the decay rate of rcccDNA by approximately 2-fold compared to the control treatment, such effect was more profound after treatment for 1 day (Fig. 10E, lane 6 vs lane 2), further indicating that MS436 treatment may preferentially promote rcccDNA degradation.

3.9. Depletion of BRD4 suppresses HBV replication

Since BRD4 protein is the known target of compound MS436, we further examined the role of BRD4 in HBV cccDNA transcription. Upon siRNA knockdown of cellular BRD4 in the context of HBV infection in HepG2-NTCP cells, a remarkable reduction of HBV total RNA was observed by northern blot assay (Fig. 11A), together with a decrease of cccDNA qPCR level, albeit at a lesser extent compared to HBV RNA reduction (Fig. 11B). Thus, the effects of BRD4 knockdown on cccDNA are consistent with MS436 treatment.

Fig. 11. siRNA knock down of BRD4 reduces HBV transcription.

HepG2-NTCP cells in 6-well-plate were transfected with control or BRD4 siRNA for 2 days, followed by HBV infection with 500 vge/cell for 24 h, then a second round of siRNA transfection was conducted in the virally infected cells. The cells were cultured for another 2 days, (A) cellular BRD4 and β-actin proteins were detected by western blot; the intracellular HBV RNA and cccDNA were analyzed by northern blot and qPCR, respectively. (B) The relative cccDNA levels are plotted as fold change to siControl. The relative levels of HBV mRNA normalized to cccDNA are plotted as fold change to that of siControl. Data are shown as mean ± SD, n=3; ***p<0.001.

In addition, the depletion of BRD4 was also achieved by the proteolysis targeting chimera (PROTAC)-based BRD4 degraders. Mechanistically, a PROTAC compound consists of two chemically conjugated protein-binding molecules: one binds to the target protein, and another engages an E3 ubiquitin ligase, by which recruits the latter to ubiquitinate the target protein for proteasomal degradation (Pettersson and Crews, 2019). We first tested BRD4 degrader compound dBET1, which is a modified bromodomain inhibitor JQ1 by conjugating with Cereblon E3 ubiquitin ligase complex (CRBN)-binding compound phthalimides (Fig. 12A) (Winter et al., 2015). In HepG2-NTCP cells with established HBV infection, dBET1 treatment resulted in a significant degradation of BRD4 protein, and the levels of HBV total RNA, cytoplasmic core DNA, and cccDNA were greatly decreased compared to DMSO control (Fig. 12B, lane 1–2). Interestingly, only ssDNA intermediate, but not the rcDNA or dslDNA, was reduced by dBET1 treatment, suggesting that the majority of rcDNA in HBV-infected HepG2-NTCP cells are derived from the inoculated rcDNA from HBV virions, while the ssDNA indicates de novo HBV DNA replication, which is consistent with our previous report (Mao et al., 2021). We also did dBET1 treatment in combination with 3TC to block de novo HBV DNA replication, the treatment outcome was similar to dBET1 mono treatment except that the core DNA level was further reduced (Fig. 12B, lane 3–4). Another BRD4 PROTAC compound, MZ-1 (Zengerle et al., 2015) (Fig. 12C), has also been tested in HBV-infected HepG2 cells in the presence of 3TC, which exhibited a potent depletion of BRD4 at 1 μM, accompanying with significant reduction of HBV RNA level (Fig. 12D).

Fig. 12. Depletion of BRD4 by PROTAC degraders inhibits HBV transcription.

(A, C) Chemical structure of dBET1 and MZ-1. (B, D) HepG2-NTCP cell were infected with HBV particles (500 vge/cell) for 6 days, followed by treatment with (B) dBET1 (10 μM) in the absence or presence of 3TC (10 μM) or (D) MZ-1 (1 μM) and 3TC (10 μM) every other day for 4 days. Cellular BRD4 was detected by western blot, β-actin served as loading control. HBV total RNA and/or cytoplasmic core DNA were detected by northern and Southern blot, respectively.

Collectively, depleting the BRD4 protein by siRNA or PROTAC degraders clearly validated the role of BRD4 in HBV cccDNA activity, and hence, the on-target antiviral effect of MS436.

4. Discussion

HBV cccDNA minichromosome is the major chronic factor of the hepatitis B disease due to its central role in viral life cycle by serving as the transcription template with a high stability. Currently, therapies for HBV are limited to the immunomodulatory PEG-IFNα and Nuc-based reverse transcriptase inhibitors (Yuen et al., 2018). However, neither of them could completely eliminate cccDNA, and only a few copies of cccDNA are sufficient for viral rebound after treatment cessation (Lai et al., 2020). Therefore, in order to eradicate or inactivate cccDNA in infected cells, novel antiviral approaches targeting cccDNA stability and/or transcription activity are considered second to none.

In order to have a specific and robust assay for screening potential HBV cccDNA inhibitors and meanwhile overcome the obstacle of measuring the low copy number of intracellular cccDNA directly, we have previously established a cccDNA-dependent HA-HBeAg reporter cell line HepBHAe82 to serve for such purposes (Cai et al., 2016). In the HepBHAe82 cell line, HA-HBeAg is expressed and secreted in a cccDNA-dependent manner, and with the HA epitope in the precore domain, it possesses high detection specificity by CLIA through avoiding cross reaction with HBcAg. Furthermore, the HepHA-HBe4 cell line constitutively expressing HA-HBeAg serves as a counter screen to filter out the preliminary hits that inhibit non-cccDNA (transgene)-dependent HA-HBeAg production (Fig.1) (Cai et al., 2016). In this study, we applied the HepBHAe82 system to a pilot screening of an epigenetic compound library for cccDNA transcription inhibitors. From this specialized epi-compound collection, 6 initial hits from 3 different categories of compounds were obtained based on reducing HA-HBeAg by more than 50% in HepBHAe82 cells but without reducing HA-HBeAg in counter screen cell line HepHA-HBe4 cells (Fig.2). During hit validation in another HBV stable cell line HepAD38 cells by measuring the intracellular HBV cccDNA and precore mRNA level using qPCR, as well as in HBV infection system HepG2-NTCP cells by core staining, compound CAY10433, UNC0642 and MS436 were confirmed hits with marked inhibitory antiviral effects on cccDNA-dependent transcription and antigen expression (Figs. 3–4). Amoung them, MS436, a BRD4 inhibitor, exhibited the most significant anti-HBV activity, and therefore was prioritized to further validations in HBV-infected HepG2-NTCP and PHH cells, where it markedly inhibited HBV transcription and replication as shown by conventional northern and Southern blotting (Fig. 5–6). The above pilot screening effort demonstrated that the HepBHAe82 system is able to score hits specifically targeting cccDNA and thus amenable to future high throughput screening (HTS) development.

BRD4 is well known as an epigenetic reader that recognizes histone acetylation through its N-terminal bromodomains (Wu and Chiang, 2007). It interacts with hyper-acetylated histone regions along the chromatin and functions as a scaffold protein for transcription factor enrichment at promoter and super-enhancers, resulting in chromatin de-compaction and gene transcription promotion at both initiation and elongation steps (Dey et al., 2003; Donati et al., 2018; Rahnamoun et al., 2018). In line with this, the treatment of HepAD38 cells by MS436 dramatically reduced the enrichment of known transcriptionally active histone pattern H3K27ac on cccDNA minichromosome as evidenced by cccDNA ChIP-qPCR assay (Fig. 8A). Meanwhile, the enrichment of transcriptionally-repressive modification H3K9me3 was observed but in a much lesser extent compared to the change of H3K27ac (Fig. 8B), which is consistent with the previous findings that the transcription activity of cccDNA is mainly determined by these active histone PTMs rather than repressive ones (Kim et al., 2022; Tropberger et al., 2015). Moreover, besides its epigenetic reader function, BRD4 has been reported to possess an intrinsic histone acetyltransferase activity (“writer”) that is able to acetylate the tail lysines of histone H3 and H4 at different positions, including the H3K27 site (Devaiah et al., 2016), which may also contribute to the observed reduction of H3K27ac on cccDNA upon MS436 treatment. Consistent with the lysine-specificity of BRD4, the association of another activating histone PTM H3K4me3 with cccDNA was not obviously affected by MS436 treatment (Fig. 8C). On the other hand, MS436 treatment did not notably affect the total protein level of BRD4 or its binding to cccDNA (Fig. 8D–E), indicating that the primary interaction of BRD4 with cccDNA minichromosome does not highly rely on the interaction between BRD4 bromodomain with the acetyllysines on cccDNA-bound histones. Besides the MS436-mediated direct epigenetic repression of cccDNA, we also examined whether the cellular global responses to MS436 contributed to the observed antiviral effects. Interestingly, the comparative transcriptomic analysis did not show a drastic alteration of host transcriptome by MS436 treatment, there was no significant changes of hepatic transcription factors required for HBV transcription or upregulation of known anti-HBV host factors/pathways (Fig. 9), suggesting that the short-term treatment by BRD4 inhibitor MS436 inhibits HBV infection more directly at the cccDNA epigenetic level, although a possible partial antiviral effect derived from minor host transcriptomic changes cannot be completely ruled out.

With rapid establishment of a sizable pool of chromatinized cccDNA-like molecules intracellularly, the Cre/loxP-launched rcccDNA system provides an alternative and convenient platform to study cccDNA-based HBV transcription and replication under compound treatment (Qi et al., 2014) (Fig. 10A–B). While the MS436-mediated inhibition of cccDNA transcription could be reproduced in rcccDNA system (Fig. 10C), a moderate reduction of rcccDNA level was also observed, and interestingly, MS436 did not appear to downregulate the level of prcccDNA (Fig. 10D). Further kinetics study demonstrated that MS436 treatment preferentially promoted the decay of rcccDNA compared to prcccDNA (Fig. 10E). These observations indicate that the stability of prcccDNA and rcccDNA is differentially regulated under transcriptional suppression by MS436. Such discrepancy is not well understood at this moment. Considering both the input prcccDNA plasmid and the prcccDNA-derived rcccDNA are covalently closed circular double-stranded DNA with largely shared sequence, it is possible that the backbone sequence and the segmented loxP sites in the prcccDNA plasmid result in different nucleosome deposition and/or histone PTM landscape with those on rcccDNA, giving rise to different responses to MS436 treatment. Nonetheless, the more interesting phenomenon here is the MS436-mediated destabilization of rcccDNA, which was observed earlier on genuine cccDNA in HBV-infected HepG2-NTCP cells and PHHs under MS436 treatment (Figs. 6–7). Furthermore, such observation was also obtained in HBV-infected HepG2-NTCP cells with siRNA knockdown of BRD4 (Fig. 11), indicating that, in addition to cccDNA transcription, BRD4 and/or BRD4-mediated epigenetics play a critical role in cccDNA maintenance, and likely the regulations of cccDNA epigenetics and stability are tightly coupled. If this notion holds up, certain cccDNA epigenetic inhibitors may potentially act beyond their known capability to destroy cccDNA template. In line with this, a recent study reported that the transcription repression of cccDNA by a NEDD8-activating enzyme inhibitor Pevonedistat is associated with a co-reduction of cccDNA copy numbers (Sekiba et al., 2019).

It is worth noting that MS436 caused more significant cccDNA destabilization in HBV infection and rcccDNA systems compared to the inducible HBV stable cell lines throughout this study. We reason that such discrepancy may be due to the different kinetics of cccDNA formation among these cccDNA-producing systems. In the HBV infection or rcccDNA system, cccDNA or rcccDNA is the first/early viral product rapidly made upon infection or transfection, which may be more transcriptionally active in a de-compaction configuration, and therefore, more sensitive to epigenetic compounds. The epigenetic inhibitors, such as MS436, may disrupt the normal process of cccDNA chromatinization by altering the recruitment of histones and/or histone PTM writers/reader/erasers, resulting in impairment of cccDNA stability. On the contrary, cccDNA formation is rather a time-consuming process in the inducible HBV cell lines (Cai et al., 2016; Guo et al., 2007a; Zhou et al., 2006). With the slow accumulation of cccDNA, the early synthesized cccDNA molecules may be already epigenetically silenced in a compact chromatin configuration. If this is the case, MS436 treatment will not significantly affect the stability of unsynchronized cccDNA but mainly prevent BRD4 from activating or re-activating cccDNA transcription. Another confounding factor is the cell conditions, which the inducible HBV stable cell lines are cultured much longer to establish cccDNA and reporter signals prior to compound treatment. In addition, while cccDNA formation in HBV stable cell lines solely goes through the intracellular rcDNA recycling pathway, the de novo cccDNA formation from the incoming virus is predominant in HBV infection systems, especially when the infection is performed under Nuc treatment (Marchetti and Guo, 2020; Marchetti et al., 2022); growing evidence has suggested that different rcDNA uncoating mechanisms and host DNA repair factors are involved in these two cccDNA formation pathways (Hu et al., 2021; Qi et al., 2016; Tang et al., 2019), which may lead to different cccDNA chromatinization and/or subnuclear localization, and thus different response to epigenetic inhibitors. In these regards, the identified cccDNA epigenetic silencer candidates should be cross validated in multiple available cccDNA-producing systems to fully assess their effects on cccDNA transcription as well as stability.

Various studies have shown that, due to its transcription activation property, BRD4 is closely related to many human diseases, including cancers, inflammations, CNS disorders, viral infections and cardiovascular diseases, and it has become an attractive therapeutic target for those diseases (Alamer et al., 2021; Chen and Ott, 2022; Liang et al., 2020). A previous study reported that a BRD4 inhibitor JQ1 enriched BRD4 occupancy on HBV genome and enhanced HBV transcription in HepG2.2.15 cells and HBV transiently transfected Huh7 cells (Francisco et al., 2017). However, the above two HBV experimental systems predominantly produce HBV transcripts from the integrated HBV transgene and transfected HBV plasmid, respectively, thus, the observed effect of JQ1 was not on the cccDNA minichromosome. In our current study, making use of several cell-based systems supporting cccDNA-dependent transcription, we have clearly demonstrated that BRD4 plays a crucial role in HBV cccDNA chromatin epigenetics and could serve as a potential antiviral target for HBV inactivation or elimination. Currently, there are two common strategies for BRD4 inhibition, including BRD4 bromodomain inhibitors and PROTAC-based BRD4 degraders (Duan et al., 2018; Li et al., 2020; Niu et al., 2019; Winter et al., 2015; Zengerle et al., 2015). Besides using MS436 as a BRD4 inhibitor, we have also tested two BRD4 PROTACs, specifically the pan-BET degrader dBET1 and BRD4-selective MZ-1, both of which exhibited dramatic depletion of BRD4 protein and potent antiviral effect against HBV infection in HepG2-NTCP cells (Fig. 12). To date, a large number of BRD4 inhibitors have been developed and about 20 drug candidates have advanced to various phases of clinical trials, where some have shown promising efficacy and safety for treatment of lymphoma, multiple myeloma, and non-small-cell lung cancer, respectively (Lu et al., 2020; Tang et al., 2021). It would be of interest to assess those most advanced BRD4 inhibitors for their antiviral activities against HBV in future study. In addition, it has been shown that BRD4 is upregulated in liver fibrosis (Wu et al., 2021), and inhibiting BRD4 can inhibit liver fibrosis and repress the migration and proliferation of hepatocellular carcinoma (HCC) (Ding et al., 2015; Tsang et al., 2019), both of which are CHB-related diseases. Thus, it is envisioned that a successful BRD4-targeted hepatitis B therapy will not only alleviate chronic HBV infection, but also may reduce or even prevent the development of HCC and other CHB complications. To meet such needs, the potential adverse effects of long-term epi-drug treatment and viral rebound upon off-treatment should be assessed in future studies.

Taken together, our study has demonstrated the practicality of the cccDNA reporter cell line HepBHAe82 working as a drug screening system and provided a proof-of-principle that use epigenetic compounds as chemical probe to identify epigenetic regulators of cccDNA and as potential therapeutic means to inhibit cccDNA. While elimination of cccDNA is considered as the “holy grail” for a complete cure of chronic hepatitis B but remains very challenging, silencing cccDNA by epi-drugs may represent an alternative and feasible approach to achieve a functional cure.

Highlights:

• cccDNA is the key player of HBV persistence and an undisputed antiviral target.

• Epi-drug screening identifies BRD4 inhibitor MS436 as a cccDNA epigenetic silencer.

• BRD4 plays a critical role in cccDNA transcription and stability.

• BRD4 is a potential antiviral target for developing cccDNA epigenetic inhibitors.