Quantification and epigenetic evaluation of the residual pool of hepatitis B covalently closed circular DNA in long-term nucleoside analogue-treated patients

Hepatitis B virus (HBV) covalently closed circular (ccc)DNA is the key genomic form responsible for viral persistence and virological relapse after treatment withdrawal. The assessment of residual intrahepatic cccDNA levels and activity after long-term nucleos(t)ide analogues therapy still represents a technical challenge. Quantitative (q)PCR, rolling circle amplification (RCA) and droplet digital (dd)PCR assays were used to quantify residual intrahepatic cccDNA in liver biopsies from 56 chronically HBV infected patients after 3 to 5 years of telbivudine treatment. Activity of residual cccDNA was evaluated by quantifying 3.5 kB HBV RNA (preC/pgRNA) and by assessing cccDNA-associated histone tails post-transcriptional modifications (PTMs) by micro-chromatin immunoprecipitation. Long-term telbivudine treatment resulted in serum HBV DNA suppression, with most of the patients reaching undetectable levels. Despite 38 out of 56 patients had undetectable cccDNA when assessed by qPCR, RCA and ddPCR assays detected cccDNA in all-but-one negative samples. Low preC/pgRNA level in telbivudine-treated samples was associated with enrichment for cccDNA histone PTMs related to repressed transcription. No difference in cccDNA levels was found according to serum viral markers evolution. This panel of cccDNA evaluation techniques should provide an added value for the new proof-of-concept clinical trials aiming at a functional cure of chronic hepatitis B.


Results
Patients' characteristics. At the time of enrolment in the trial ( Supplementary Fig. 1), all 56 patients (37 HBeAg( +) and 19 HBeAg(−)) had alanine aminotransferase (ALT) levels higher than the upper limit of normal (ULN) and were classified as chronic hepatitis B (CHB) according to the 2017 EASL clinical practice guidelines 16 . The median age was 29 years old, HBV e antigen-positive (HBeAg( +)) patients being significantly younger than HBV e antigen-negative (HBeAg(−)) ones. Forty-three patients were infected with HBV genotype C and 13 with genotype B (Table 1). Baseline serum HBV DNA and ALT levels tended to be higher for HBeAg( +) patients and there was no significant difference between the two groups in the duration of telbivudine treatment (Table 1). Before treatment, the majority of the patients exhibited low histological scores of necroinflammatory activity and fibrosis although HBeAg(−) group showed a significantly higher proportion of patients with advanced fibrosis (Supplementary Fig. 2).
At the time of liver biopsy, i.e. one week after entry in the CLDT600ACN04E1 study, corresponding to three to five years of telbivudine therapy (Supplementary Fig. 1 and Methods section), all the patients reached ALT levels under the ULN and only 3/56 patients exhibited a detectable serum HBV DNA (all 3 HBeAg( +) patients).
Quantification of intrahepatic viral markers after telbivudine treatment using a qPCR method. Intrahepatic total HBV DNA (tHBV-DNA), cccDNA and 3.5 kb RNA (preC/pgRNA) were first quantified using a qPCR method already described for intrahepatic HBV markers quantification 17 and results are listed in Table 2. tHBV-DNA levels after telbivudine therapy were less than 1 copy per cell in all patients.
Intrahepatic cccDNA detection using rolling circle amplification and droplet digital PCR. To further investigate the presence of cccDNA in the liver samples with undetectable cccDNA by qPCR, we decided to perform two supplementary approaches: RCA and ddPCR (Fig. 1). Among the 39 samples with undetectable cccDNA by qPCR, 29 showed a positive signal after RCA analysis (Figs. 1 and 2). Since the major limitation of RCA technique is that it is not quantitative, we decided to use ddPCR, which has been recently shown to be more sensitive than qPCR in detecting HBV genome in low-level HBV infected samples [18][19][20] . We showed that the ddPCR cccDNA assay specifically amplifies episomal HBV DNA over integrated sequences and preferentially recognizes cccDNA over rcDNA (Methods section and Supplementary Fig. 3). In particular, we could confirm that ddPCR is 2 logs more sensitive than qPCR for the quantification of an HBV plasmid, allowing HBV DNA quantification from samples containing as less as 1 copy/µL of HBV genome (Fig. 3a). We also showed that the circular nature of cccDNA did not affect ddPCR efficiency, since the same quantification results were obtained from of a circular and a linearized HBV-plasmid template (Fig. 3b). The analysis of HBV-negative liver biopsy samples confirmed the specificity of the technique and served as reference for the settings of thresholds discriminating positive from negative droplets (Fig. 3c).
cccDNA-positive liver samples with qPCR were confirmed to be positive also by ddPCR. Twenty-seven out of 34 samples with undetectable cccDNA by qPCR had a positive cccDNA signal with both RCA and ddPCR (Fig. 4). ddPCR allowed the detection of cccDNA in 6 more patients than RCA while one sample was positive Table 1. Demographical and virological patients' characteristics at baseline. a Data are expressed as median (1st quartile-3 rd quartile) b Mann Whitney U test was used for comparison between HBeAg( +) and (−) groups. c χ 2 test was performed between HBeAg( +) and (−) groups. www.nature.com/scientificreports/ only by RCA analysis (Fig. 4). Altogether, after combining qPCR and ddPCR results, intrahepatic cccDNA was detectable in 50 out of 51 telbivudine-treated patients tested with both techniques (5 liver samples were not assessed by ddPCR due to insufficient residual material) ( Figs. 1 and 4).
At the time of liver biopsy, 28/37 (76%) HBeAg( +) patients had lost HBeAg and 15/37 (40%) had experienced HBe seroconversion (Table 2). No differences in baseline age, serum HBV DNA and ALT levels were found between patients who experienced or not HBeAg loss and/or seroconversion (Supplementary Table 1). No differences in intrahepatic tHBV-DNA, cccDNA or 3.5 kb RNA levels after telbivudine therapy were reported according to HBeAg loss and/or seroconversion at the time of liver biopsy (Supplementary Table 1 and 2). Two out of 3 patients with detectable serum HBV DNA at the time of liver biopsy showed positive cccDNA when assessed with qPCR, while all of them were positive when tested by ddPCR. No difference of tHBV-DNA or 3.5 kb RNA Step 1: Frozen liver biopsy samples were first tested for the presence and the level of intrahepatic cccDNA using quantitative PCR (qPCR) method 17 ; Step 2: Rolling Circle Amplification (RCA) 35,36 was performed on the 39 qPCR-negative samples and detected cccDNA in 29 of them; Step 3: digital droplet PCR (ddPCR) was performed on qPCRnegative liver samples with available remaining material (28 out of 29 RCA-positive samples and 6 out of 10 RCA-negative samples). ddPCR detected cccDNA for all-but-one samples tested. Overall, intrahepatic cccDNA was detectable in 50 out 51 patients tested when assessed by qPCR or ddPCR. cccDNA covalently closed circular DNA, RCA rolling circle amplification, ddPCR droplet digital PCR. www.nature.com/scientificreports/ was found for these patients compared to those with undetectable serum HBV DNA (data not shown). Only two patients (HBeAg( +) at baseline) lost HBsAg after telbivudine therapy, therefore it was not possible to perform www.nature.com/scientificreports/ statistical analysis in this group of patients (Table 2). However, both had undetectable cccDNA by qPCR, but scored positive by ddPCR quantification.
cccDNA epigenetic status on telbivudine treatment. 3.5 kb RNAs are exclusively transcribed from its cccDNA template 3,5 . Given the low levels of 3.5 kb RNAs found in telbivudine-treated patients, we wondered if it could be explained not only by a low level of its transcriptional template, i.e. cccDNA, but also by a reduction of its transcriptional activity. To test this hypothesis, we performed ChIP analysis on liver samples derived from 10 telbivudine-treated patients with sufficient liver material (4 HBeAg( +), 6 HBeAg(−)). All the patients had qPCR-detectable cccDNA, but did not differ significantly from the cccDNA-qPCR negative www.nature.com/scientificreports/ Telbivudine-treated patients not included in ChIP analysis, except for having higher cccDNA levels (Supplementary Table 3). They were compared to 7 untreated CHB patients with active HBV transcription and replication (5 HBeAg( +), 2 HBeAg(−)) (see Supplementary Table 4 for patients' characteristics). We assessed the levels of cccDNA-associated histone PTMs associated to either active (H3K27Ac, H3K56Ac) or inactive (H3K9me3, H3K27me3) transcription, using GAPDH promoter amplification as a control for antibody specificity (Supplementary Fig. 4a) 21 . Untreated CHB patients presented a higher 3.5 Kb RNA/cccDNA ratio (median of 160 vs 6.03 in telbivudine-treated group, p = 0.0001, Supplementary Table 4) associated with a significant enrichment of positive histone PTMs on cccDNA (Fig. 5a), while telbivudine-treated patients showed no significant enrichment with respect to No Antibody control of positive histone PTMs, but a significant increase of histone PTMs associated with repressed transcriptional activity, independently from patients' HBeAg status ( Fig. 5b and Supplementary Fig. 4b-c).

Discussion
In the context of the global HBV cure research programs 15 , the evaluation of the intrahepatic cccDNA amount and its transcriptional activity will be instrumental for the assessment of the efficacy of the novel antiviral strategies under clinical investigation and for the development of non-invasive biomarkers reflecting the pool of cccDNA 15 .
Here, we propose the use of a panel of molecular assays to robustly assess both cccDNA amount and transcriptional activity in vivo, in liver samples derived from long-term NUC-treated patients. The implementation of ddPCR, together with the adaptation of classic ChIP to small size liver biopsies represent an advent by providing technical tools to evaluate the residual pool of cccDNA and its epigenetic status in the infected liver.
Similar to the study of Lai et al., we found that the majority of the patients had undetectable levels of intrahepatic cccDNA after long course of NUC treatment, when assessed with qPCR technique 14 . However, in our study, using a more sensitive ddPCR assay, we could detect and quantify intrahepatic cccDNA in all but one tested liver biopsies of telbivudine-treated patients. Importantly, the assay used in this study for cccDNA quantification specifically recognize episomal cccDNA over HBV integrated sequences (Supplementary Fig. 3). The relevance of ddPCR analysis in CHB has already been reported for the investigation of CHB patients who received interferon therapy 19    www.nature.com/scientificreports/ also consistent with previous mathematical modelling of cccDNA kinetics that showed low rates of clearance in NUC-treated patients 13 and provide insight into the mechanism of viral rebound after treatment withdrawal even in long-term virally suppressed patients 15,22 . Similar to Lai et al., we found very low levels of intrahepatic 3.5 kb RNA and 3.5 Kb RNA/cccDNA in this cohort of long-term telbivudine-treated patients with respect to previously published cohorts of untreated patients 14,[23][24][25] . Therefore, we were interested in investigating the epigenetic status of the residual intrahepatic cccDNA.
An adaptation of the cccDNA-ChIP technique 5 to small biopsy samples (micro-ChIP) 26 allowed the identification of an enrichment of histone PTMs associated with inactive transcription on cccDNA derived from telbivudine-treated patients. Our observation of decreased histone H3 acetylation is in accordance with previous data obtained in vivo in the context of reduced cccDNA transcriptional activity in untreated patients 5,8,9 . Ren et al. 9 also found an increased enrichment for H3K9me3 in inactive carrier CHB patients having low 3.5 kb RNA levels, while Flecken et al. 10 failed to find such a difference for trimethylation on both H3 lysine 9 and 27 across patients belonging to different CHB phases. In the same study from Flecken et al., three NUC-treated patients were also analysed, but could not be differentiated from non-treated ones on the basis of cccDNA-associated histone PTMs enrichment profile 10 . Our data showed that in this telbivudine-treated patient population with low 3.5 kb RNA/cccDNA ratio, H3K9me3 and H3K27me3 are enriched on histones associated with cccDNA.
Whether these changes in cccDNA-associated histone PTMs profile could be due to a direct effect of telbivudine-treatment remains to be investigated. The primary mode of action of NUCs, i.e. inhibition of HBV polymerase reverse transcriptase activity, would exclude any direct effect on cccDNA epigenome. However, one might hypothesize that a prolonged suppression of pgRNA to rcDNA conversion might trigger a negative feedback regulation of cccDNA transcription. On the other hand, we cannot rule out the hypothesis that the inhibition of cccDNA activity could be due to an additional property of telbivudine. Indeed, telbivudine was shown to modify host histone PTMs at specific loci in HepG2.2.15 cells, by restoring H3K4me3 and H3K27me3 to levels comparable to those detected in HepG2 cells 27 . Moreover, it has been shown that HBx protein levels may be reduced after telbivudine treatment in vitro 28 . Since HBx is essential for full cccDNA transcription and HBx-deficient HBV mutant strains show enrichment in cccDNA-associated histone PTMs correlated with inactive transcription 29 , this could represent an additional antiviral mechanism of telbivudine. If this effect on cccDNA epigenome is shared with other NUCs, or with emerging new antivirals, remains to be investigated. In this respect, recent data obtained by Balagopal et al. using single-cell laser capture microdissection and ddPCR in liver biopsies from HIV/HBV co-infected patients under antiviral therapy (DAART or ART) confirmed the decline of cccDNA amount along with the duration of the antiviral treatment and also observed a sharp decrease in preC/pgRNA levels in samples derived from long term-treated patients 30 . They also showed an increased number of cccDNA-positive hepatocytes without a detectable preC/pgRNA, suggesting an inactivation of cccDNA transcriptional activity correlating with long-term antiviral therapy 30 . Overall, their results could be consistent with our observation of an epigenetic repression of cccDNA during long-term NUC therapy. Further singlemolecule studies would be required to better clarify the relative contribution to the 3.5 Kb RNA level decrease of cccDNA template decline vs epigenetic repression.
Of course, it cannot be excluded that other liver microenvironment factors such as changes in the inflammatory status and in cytokine expression during therapy may also be involved in this phenomenon.
Altogether, using a highly sensitive cccDNA quantification method together with the analysis of the cccDNAassociated epigenome, our data showed the persistence of low levels and epigenetically-modified intrahepatic www.nature.com/scientificreports/ cccDNA in patients with long-term telbivudine-induced viral suppression. Whilst the use of liver biopsies does not prevent a sampling bias due to potential focal differences in the infected tissue, our results underline the importance of implementing more sensitive techniques to quantify cccDNA in patients and to assess the epigenetic status and transcriptional activity of the viral minichromosome. This panel of cccDNA evaluation techniques should provide an added value for the new proof of concept clinical trials aiming at a functional cure of chronic hepatitis B.

Patients and methods
Patients. Fifty-six patients with chronic hepatitis B (CHB) (46 males and 10 females) previously enrolled in the clinical studies NV-02B-007 (NCT00057265) or NV-02B-015 (NCT00131742) and who entered the CLD-T600ACN04E1 (NCT00877149) study (Novartis Pharma AG) without discontinuation of telbivudine treatment were included in this study (Supplementary Fig. 1) [31][32][33] . The studies were approved by each local independent ethics committee and were in accordance with the ethical standards of the Helsinki Declaration of 1975, as revised in 2000 and 2008. Written informed consent was obtained from each patient and/or their legal guardians. Histological, clinical and virological data reported in this study were collected in studies NV-02B-007 or NV-02B-015 at baseline and after five years of antiviral treatment (Supplementary Fig. 1). Part of the liver specimens collected at CLDT600ACN04E1 study entry was snap frozen at − 80 °C for molecular biology analysis. These legacy clinical studies have been published with ethics committee approvals previously disclosed and publicly available [31][32][33][34] .
For the analysis of cccDNA histone post-translational modifications, 7 untreated CHB patients from a historical cohort collected at Hospices Civils de Lyon (Lyon's University, France) were used as comparative samples. These patients underwent liver biopsy as part of their clinical follow-up, a fragment was preserved for research purposes and stored at − 80 °C. The protocol was approved by the competent Institutional Ethics Committee (CPP Sud est IV 11/040, authorization number DC-2008-235). Written informed consent was obtained from all patients and/or their legal guardians to underwent liver biopsy. No patients were co-infected with HIV, hepatitis C virus or hepatitis delta virus.
This submitted article is based on data from the previously submitted studies and does not involve any new studies or new human subjects. All methods were carried out in accordance with relevant guidelines/regulations. No animal experimentation and no use of primary human cells was employed in this study. Viral load and serological assessment. Serum HBV DNA, alanine aminotransferase (ALT) levels, HBe antigen (HBeAg), HBs antigen (HBsAg) and HBe and HBs serology were analyzed in the frame of the clinical trials by the study sponsor. Serum HBV DNA levels were quantified with the Roche COBAS Amplicor PCR assay (Roche Molecular Systems, Branchburg, NJ, USA), with an estimated lower limit of detection and quantification of 300 HBV genome copies/mL (around 60 IU/mL). Other HBV serologic markers (HBsAg/anti-HBs, HBeAg/ anti-HBe) were assessed using standard commercially available assays 31,32 . Intrahepatic total HBV DNA, cccDNA and preC/pgRNA quantification using qPCR method. DNA was extracted from snap-frozen liver biopsies using the Master Pure DNA Purification Kit Patients' samples were run in duplicate and in two independent experiments. Quantitative PCR were performed using specific primers and fluorescence hybridization probes previously described 17 and a Light Cycler 480 Real Time PCR System (Roche Diagnostics, Manheim, Germany) 24,25 . Before cccDNA quantification, total DNA was treated with Plasmid-safe DNAse (LUCIGEN, Middleton, WI, USA), to limit relaxed circular DNA (rcDNA) contamination. Serial dilutions of an HBV monomer plasmid (pHBV-EcoR1) were used as standard for quantification. Beta-globin quantification was performed in parallel to assess the amount of HBV DNA copies/cell 17 . Intrahepatic 3.5kB RNA (preC/pgRNA) quantification, consisting of both preCore and pgRNA transcripts, was performed with specific primers and TaqMan fluorescence hybridization probes previously described by Volz et al. 23 . Results were normalized over the housekeeping gene GUSb (Hs00939627_m1, Thermofischer Scientific, Waltham, MA, USA). All the primers and probes used are listed in Supplementary Table 5 and visualized on HBV genome in Supplementary Fig. 5. cccDNA detection using rolling circle amplification. Rolling Circle Amplification (RCA) was shown to be highly specific for complete circular forms of HBV genome and particularly adapted for the amplification of cccDNA in liver tissue with low levels of viral replication 35 . We assessed the presence of intrahepatic cccDNA in qPCR-negative samples using this one-step PCR method without additional steps of completion and ligation, as previously described using primers RCA1-8 listed in Supplementary Table 5 and Supplementary Fig. 6  (Fig. 2a) 35,36 . The amplification of circular HBV DNA was performed using Phi29 polymerase (New England Biolabs, Ipswich, MA, USA) for 21 h at 30 °C on total DNA extracted from frozen liver samples. RCA products were analysed according to two different methods to assess the presence of intrahepatic cccDNA in the samples. The first one, Southern blot after digestion of RCA products allowed us to control for RCA specificity but was not sufficiently sensitive for samples with small amount of intrahepatic cccDNA. The second one, full length-HBV genomic PCR on RCA products allowed us to detect the presence of cccDNA even at very low levels (Fig. 2 www.nature.com/scientificreports/ smart Buffer 1.1 µL (New England Biolabs, Ipswich, MA, USA). Digestion products were run on 1% agarose gel and then transferred to perform hybridization with HBV-specific cold probes (Fig. 2b) 37 . Full-length HBV genomic PCR amplification used primers P1 and P2 38 (Supplementary Table 5 and Supplementary Fig. 6) to enhance cccDNA detection sensitivity in samples with low HBV DNA levels (Fig. 2c). The final PCR reaction included 2µL of RCA products, 0.25 µL of each sense and anti-sense primers previously described (100 µM) and 25 µL of PrimeSTAR HS (Premix) (Takara Bio Inc. Shiga, Japan) in a 50 µL final volume. The amplification protocol was adapted from Günther et al. 38 and was performed for 40 cycles as follow: denaturation at 94 °C for 40 s; 1 min annealing at 55 °C for 10 cycles, then at 60 °C for 10 cycles and finally at 62 °C for the last 20 cycles; elongation at 68 °C for 3 min with an increment of 2 min after each 10 cycles. PCR products were run on 1% agarose gel.
Intrahepatic total HBV DNA and cccDNA quantification by droplet digital PCR. Intrahepatic total HBV DNA and cccDNA were quantified by the QX100 Droplet Digital PCR System (Bio-Rad, Hercules, CA, USA) according to the manufacturer's instructions. Before cccDNA quantification, total DNA was treated with Plasmid-safe DNAse (Lucigen, Middleton, WI, USA), to limit rcDNA contamination. Briefly, the 25 μL ddPCR reaction was comprised of 2× ddPCR Supermix for probes (Bio-Rad, Hercules, CA, USA), 900 nM HBV cccDNA forward and reverse primers, 250 nM HBV probe, and 5 μL of DNA sample. The plates were loaded with required consumables into the Automated Droplet Generator to partition the sample around 20,000 droplets in 20 µL. PCR amplification was performed in a C1000 Touch thermal cycler (Bio-Rad, Hercules, CA, USA) with the following amplification program: 10 min at 95 °C, 40 cycles of denaturation for 30 s at 94 °C and annealing for 60 s at 60 °C (ramping rate set to 2 °C/s), final incubation step for 10 min at 98 °C. After reading in a QX100 Droplet Reader (Bio-Rad, Hercules, CA, USA), the data were analyzed using the QuantaSoft analysis software (Bio-Rad, Hercules, CA, USA), which automatically calculated absolute sample concentration after applying Poisson's distribution. Fluorescence amplitude threshold to distinguish positive from negative droplets was based on amplification of negative controls (water and non-HBV infected samples, Fig. 3). Concomitant quantification of HBV DNA and beta-globin in the same sample allowed the estimation of HBV DNA or cccDNA copies/cell. All primers and probes used are listed in Supplementary Table 5 and Supplementary Fig. 5. The pBR322-HBV construct was used to assess linearity of ddPCR quantification and to compare it to qPCR. Limit of detection for cccDNA quantification in ddPCR was 4.8 copies/10 6 cell (3.3-7, 95% confidence interval), calculated by Probit analysis on a sample composed of serial dilution of a minicircle HBV into genomic DNA.
Specificity of cccDNA assay for episomal HBV DNA vs integrated sequences was analyzed in Hep3B cells, which harbor nearly 1 HBV integration/cell, but no cccDNA (Supplementary Methods and Supplementary  Fig. 3a,b). cccDNA assay was specific for cccHBV DNA vs rcDNA if the rcDNA:cccDNA ratio was < 1000 without the need of prior PSD digestion ( Supplementary Fig. 3c). In HBV-infected HepG2-NTCP cells, NUC treatment by lamivudine (3TC) lowers the rcDNA:cccDNA to less than 100 and addition of PSD digestion ensures a rcDNA:cccDNA around 1, thus highly incrementing the specificity of our assay for cccDNA (Supplementary Methods and Supplementary Fig. 3d). Thus, these data supported the use of this ddPCR cccDNA protocol for specific cccDNA quantification over other HBV genomic forms in the long-term NUC-treated patients analyzed in this study.
Chromatin immunoprecipitation from frozen liver biopsies. ChIPs from liver biopsies were performed as described by Testoni et al. 26 . Briefly, frozen biopsies were removed from the freezer and immediately added with PBS/1% formaldehyde and incubated 10 min at room temperature. Crosslinking reaction was quenched with 0.125 M glycine. After centrifugation, the cross-linked biopsy was resuspended in lysis buffer (50 mM Tris-HCl pH 8, 10 mM EDTA, 1% SDS, protease inhibitors) and homogenized with the help of a pestle. Supernatant was then sonicated in a Bioruptor (DIAGENODE, Liege, Belgium) for 2 × 30 s cycles, diluted 1:10 in Ripa Buffer (10 mM Tris-HCl pH 7.5, 140 mM NaCl, 1 mM EDTA, 0.5 mM EGTA, 1% Triton X-100, 0.1% SDS, 0.1% Na-deoxycolate, protease inhibitors) and incubated with Dynabeads Protein G and 3 μg of each antibody (ChIP-grade anti-H3K27Ac, H3K56Ac, H3K9me3, H3K27me3, DIAGENODE, Liege, Belgium) overnight at 4 °C rotating. After immunoprecipitation, washes and reverse crosslinking, the samples were extracted twice with phenol/chloroform, once with chloroform and ethanol precipitated in the presence of 30 μg of glycogen (SIGMA ALDRICH-MERCK, Darmstadt, Germany). Quantitative PCR was performed using cccDNA-specific primers and probes. Glyceraldehyde-3-phosphate Dehydrogenase (GAPDH) promoter region amplification served as control for specific enrichment of antibodies (see Supplementary Table 5 and Supplementary Fig. 4a).