Characterization and Application of Precore/Core-Related Antigens in Animal Models of Hepatitis B Virus Infection

Abstract

The hepatitis B core-related antigen (HBcrAg), a composite antigen of precore/core gene including classical hepatitis B core antigen (HBc) and e antigen (HBeAg), and additionally, the precore-related antigen PreC, retaining the N-terminal signal peptide, has emerged as a surrogate marker to monitor the intrahepatic hepatitis B virus (HBV) covalently closed circular DNA (cccDNA) and to define new meaningful treatment endpoints. Here, we found that the woodchuck hepatitis virus (WHV) precore/core gene products (i.e., WHV core-related antigen, WHcrAg) include the WHV core protein (WHc), WHV e antigen (WHeAg), as well as the WHV PreC protein (WPreC) in infected woodchucks. Unlike in HBV infection, WHeAg and WPreC proteins were N-glycosylated, and no significant amounts of WHV empty virions were detected in WHV-infected woodchuck serum. WHeAg was the predominant form of WHcrAg, and a positive correlation was found between the serum WHeAg and intrahepatic cccDNA. Both WHeAg and WPreC antigens displayed heterogeneous proteolytic processing at their C-termini resulting in multiple species. Analysis of the kinetics of each component of the precore/core-related antigen, along with serum viral DNA and surface antigens, in HBV-infected chimpanzees and WHV-infected woodchucks revealed multiple distinct phases of viral decline during natural resolution and in response to antiviral treatments. A positive correlation was found between HBc and intrahepatic cccDNA, but not between HBeAg or HBcrAg and cccDNA in HBV-infected chimpanzees, suggesting that HBc can be a better marker for intrahepatic cccDNA. In conclusion, careful monitoring of each component of HBcrAg along with other classical markers will help understand intrahepatic viral activities to elucidate natural resolution mechanisms as well as guide antiviral development.

Introduction

Chronic hepatitis B virus (HBV) infection affects an estimated 257 million individuals worldwide and carries high risks of developing hepatocellular carcinoma (1). Nucleos(t)ide analogues (NUCs) and interferon alpha (IFN-α) are two classes of approved therapies to treat chronic hepatitis B. However, current therapies can suppress viral replication, resulting in improvement of liver histology, but rarely cure HBV infection due to the inability to eliminate the viral covalently closed circular DNA (cccDNA) in infected hepatocytes (2).

HBV is the prototype of Orthohepadnaviruses, within Hepadnaviridae, with a partial double-stranded relaxed circular DNA (rcDNA) genome (2). Upon infection, the rcDNA is delivered into the nucleus and converted into the cccDNA, which is the transcriptional template for all viral mRNAs and thus the molecular basis of viral persistence (2). The pregenomic RNA (pgRNA) encodes the HBV core protein (HBc) and reverse transcriptase (RT). In the cytoplasm, RT and pgRNA are packaged together into a nucleocapsid (NC) formed by HBc. In the NC, pgRNA is reversely transcribed by RT to form rcDNA during NC maturation. Mature NCs can be secreted as complete virions after coating by viral surface (envelope) proteins (HBsAg) or recycled, intracellularly, back to the nucleus to replenish the cccDNA pool. In HBV-infected patients, HBV genome-free (empty) virions assembled from cytoplasmic empty capsids and HBsAg but devoid of pgRNA or DNA are also secreted in large excess (3–6). HBsAg alone forms subviral particles, which are found at 1,000-fold excess over virions in the blood (6). HBV e antigen (HBeAg) is the secretory product of precore (p25), which is translated from the precore mRNA. p25 is identical to HBc except it is extended by a N-terminal precore-specific 29-amino-acid (aa), which directs p25 into the endoplasmic reticulum (ER). The N-terminal 19-aa signal peptide of p25 is removed by the signal peptidase during ER translocation. Subsequently, ca. 30-aa at the C-terminal domain (CTD) are removed by proprotein convertases (e.g., furin) in the secretory pathway to generate multiple species of HBeAg, due to heterogeneity in the C-terminal cleavage sites (7).

Serum HBV DNA (i.e., viremia), HBsAg, and HBeAg are classical markers to monitor HBV replication during the natural course of infection or antiviral therapies. The hepatitis B core-related antigen (HBcrAg) is an emerging surrogate marker for intrahepatic cccDNA levels and activities (8). The current HBcrAg assay detects a combination of HBc, HBeAg, and a 22KDa precore protein (p22cr, renamed to PreC), which retains the entire N-terminal signal peptide but is processed at the C-terminus similarly to HBeAg (9–11). Serum HBc, contributed predominantly by HBV empty virions and secreted despite effective inhibition of viral DNA synthesis, can itself be used as a marker to monitor the intrahepatic cccDNA during NUC treatments (3–5). The major component of HBcrAg is not HBc but the precore-derived HBeAg (11).

With strict host species specificity, few experimental animal models are available to study HBV infection (12). Chimpanzees are susceptible to HBV infection and develop antiviral immune responses, but these studies have been restricted due to ethical concerns (12). Woodchuck hepatitis virus (WHV), another orthohepadnavirus closely related to HBV, infects its natural host, the Eastern woodchuck, Marmota monax (12, 13). Experimental infection of woodchucks with WHV is a well-accepted model for studying HBV pathogenesis and evaluation of anti-HBV therapies. Thus, the woodchuck model provides an excellent opportunity to evaluate HBcrAg as a potential biomarker. Building on our recent characterization of the HBcrAg components (11), we have here characterized the composition of WHV core-related antigen (WHcrAg) in WHV-infected woodchucks, taking advantage of a panel of monoclonal antibodies (mAbs) that specifically recognize WHV core protein (WHc) and precore-related proteins. Our results shed light on the WHcrAg composition and characteristics of each component. Kinetic analysis of each precore/core-related antigen component in WHV-infected woodchucks and HBV-infected chimpanzees, along with other serum viral markers and intrahepatic cccDNA, revealed important insights into the mechanisms of viral clearance during natural infection and in response to different therapeutic strategies.

Materials and Methods

Serum samples and liver biopsies

Serum samples and liver biopsies from five HBV (Genotype D, Genbank accession number: V01460)-infected chimpanzees (1603, 1616, 1618, A2A007, and AOA006) and WHV strain 7 infected woodchucks were described previously (14–17).

Additional materials and methods are available in Supplementary Materials.

Results

Screen for antibodies cross-reactive with WHV precore/core antigens

Inspection of the precore/core open reading frame (ORF) sequence revealed a 70.28% identity at the amino acid sequence level between the WHV strain 7 (Genbank accession number: M18752) and our laboratory HBV strain (genotype D, ayw) (11) (Fig. 1A). We thus screened a panel of mAbs for the HBV precore/core proteins for their cross-reactivities to the WHV proteins and found that three mAbs targeting the precore/core NTD (T2221, 10E11, and 19C18), two mAbs targeting the precore/core CTD (A701 and 14–2), and two precore-specific mAbs (1A11 and 7E9) cross-reacted with WHV precore/core-derived proteins. Indeed, the 1A11, 7E9, 10E11, 19C18, A701, and 14–2 epitopes are 100% conserved and are thus expected to react equally well with both HBV and WHV precore/core-derived proteins (Fig. 1A, S1). On the other hand, there are potentially two mismatches in the T2221 epitope between the HBc and WHc NTD sequence, explaining the ca. 5-fold lower sensitivity of mAb T2221 in detecting WHc vs. HBc in quantitative western blots (Fig. S1).

Figure 1. Screen for antibodies cross-reactive with WHV precore/core antigens.

(A) Amino acid sequence alignment of the HBV genotype D (gt D) and WHV strain 7 (WHV7) precore/core proteins. The epitopes for each mAb are indicated by the dashed boxes. mAbs 1A11 and 7E9 are specific for the HBV and WHV eAg and PreC proteins. mAbs 19C18, 10E11, and T2221 react with the NTD shared by the HBV and WHV core and precore proteins. mAb A701 is selective for the non-phosphorylated (HBc and WHc) CTD whereas 14–2 for the phosphorylated (HBc and WHc) CTD. mAbs 366–2 (relatively non-selective for phosphorylation state) and 25–7 (selective for non-phosphorylation) are specific to HBc CTD without cross-reactivity to WHV proteins. The arrows indicate the putative cleavage sites in the production of WHeAg (WHe0 and WHe1), WHV PreC protein (WPreC0 and WPreC1), and HBeAg and HBV PreC (e1 and PreC1) as identified in this study. Two N-glycosylation motifs, NIT and NDT, in WHV (but not HBV) precore are underlined. (B) Immunoblot analysis of HBcrAg components in the serum of the genotype D HBV-infected chimpanzee (#1616, Wk22) and a WHV7-infected woodchuck by SDS-PAGE in a regular gel (12.5 cm in height). WHV-infected woodchuck serum samples from M1002 prior to wIFN-α treatment (Wk-3) and after recovery following treatment (Wk15) were included here as the WHeAg-positive and WHeAg-negative controls. (C) Schematic representations of each WHcrAg or HBcrAg species detected in the serum of WHV-infected woodchucks or HBV-infected chimpanzees shown in B. SP, signal peptide. g, glycosylated.

WHV e antigen (WHeAg) was N-glycosylated in WHV-infected woodchuck sera

SDS-PAGE and western blot analysis of woodchuck serum samples revealed that the predominant species of WHcrAg, detected by both 7E9 and T2221, was presumably WHeAg and displayed much slower mobility compared to HBeAg, suggesting a post-translational modification(s) on WHeAg but not HBeAg (Fig. 1B). Additional minor species were also detected, with the fastest migrating species running just above HBeAg (e1), as expected of unmodified WHeAg. WHeAg secreted from human HEK293 cells transfected with a WHV-precore expression construct was shown to be N-glycosylated (18). To determine whether WHeAg was N-glycosylated in WHV-infected woodchuck sera, we treated the serum samples with PNGase F, which removes N-linked glycans from glycoproteins. This treatment caused the predominant WHeAg species to migrate much faster on SDS-PAGE, like the unmodified HBeAg (Fig. 2A), indicating that most of the serum WHeAg was N-glycosylated, but a minor proportion was not, and WHeAg was the predominant component of WHcrAg, as in HBV-infected patients (11). We confirmed that complete deglycosylation was achieved in one hour, and digestion as long as overnight did not lead to additional changes in the WHeAg/PreC protein species detected (Fig. S2A). Digestion with another glycosylase, PNGase A, which removes different N-linked glycans resistant to PNGase F, did not cause any mobility shift of WHV precore-derived proteins, suggesting there was no alternative N-glycosylation resistant to PNGase F (Fig. S2B). Furthermore, we demonstrated that the WHeAg secreted from WC3 woodchuck hepatoma cells transfected with the WHV-precore expression construct was also N-glycosylated (Fig. S2C), as reported in human HEK293 cells (18). Altogether, our results indicated that WHeAg was the predominant form of WHcrAg and was N-glycosylated.

Figure 2. WHeAg was N-glycosylated in WHV-infected woodchuck sera.

(A) Serum from a chronically infected woodchuck F6005 or an acutely infected woodchuck M7392 was treated with PNGase F followed by SDS-PAGE (12.5 cm gel) and immunoblotting using the anti-precore mAb 7E9 and the anti-NTD mAb T2221. (B) Immunoblot analysis of WHcrAg components in WHV-infected woodchuck sera following SDS-PAGE in a 32 cm height, high-resolution gel. Sera from acute WHV-infected woodchucks (M7392, F7386, and F7394; weeks refer to the time after WHV inoculation) and chronic WHV-infected woodchucks (F1018 and M7994; weeks refer to time after the beginning of antiviral treatments - these woodchucks had been inoculated as neonates and chronic WHV infection lasted for over a year before the start of the treatments), with or without PNGase F treatment, were analyzed. M7392 Pre (lane 1, 2, 15, 16), serum from the M7392 before infection, and M7994 Wk9 (lane 13, 14, 27, 28), serum from chronically infected M7994 after recovery following GS-9620 treatment, were included as negative controls. (C) Schematic representations of the various WHcrAg species detected by immunoblots shown in A & B. *, unknown woodchuck serum proteins cross-reactive with the anti-precore mAbs, 7E9 and 1A11, which co-migrated with the putative glycosylated WPreC proteins (g-WPreC0 and g-WPreC1).

Multiple species of WHV precore-related proteins were released into the woodchuck blood

Serum WHV precore-derived proteins remained heterogeneous after complete deglycosylation (Fig. 2, S2), indicating multiple species of WHV precore-related proteins were released into the woodchuck blood, similar to HBeAg and PreC proteins in human sera (11). Analogous to HBeAg, we called the WHeAg species, WHe0 or WHe1 (and g-WHe0 or g-WHe1 for the glycosylated form), based on the putative cleavage sites in their CTD (at position 149 and 159, respectively) (7, 19), as deduced from the mobilities on SDS-PAGE (Fig. 1A, 1C, 2C, S2, S3), with WHe1 running slightly above HBV e1 due to the five-residue insertion in WHeAg just N-terminal to the putative cleavage site (Fig. 1B).

The existence of WHV PreC (WPreC) proteins, other than the two WHeAg species, could be deduced from the detection of a minor species, WPreC1 (non-glycosylated), corresponding to WHe1 that retained the N-terminal signal peptide and running slower than WHe1 and just above the HBV PreC1 protein, and its putative glycosylated isoform, g-WPreC1, running above g-WHe1 (Fig. 1B, lane 2, 5). As expected, PNGase F treatment caused the WPreC1 signal to increase and the g-WPreC1 signal to decline or become absent (Fig. 2A, lanes 2, 4; 2B, lanes 9–12, 23–26; S2; S3). A putative WPreC0 species, migrating just below WPreC1, was also detected in those woodchucks showing the WHe0 species, esp. following deglycosylation (Fig. 2A, lanes 6, 8; 2B, lanes 3–6, 17–20; S2; S3), suggesting that WPreC0 corresponded to WHe0 in its C-terminal cleavage site selection but retained the N-terminal signal peptide, as the HBV PreC proteins (11). The g-WPreC0, corresponding to the glycosylated WPreC0, was readily detected in some but not all woodchuck serum samples, owing to comigrating background signals (Fig. 2A, 2B, S2, S3). We further verified the lack of the CTD sequence in WHeAg and WPreC by using two mAbs that are selective for either the non-phosphorylated (A701) or phosphorylated (14–2) CTD (Fig. 1A, S3C). We attempted to detect the predicted N-terminal signal peptide sequence in WPreC by mass spectrometry following tryptic digestion of the gel-purified g-WPreC1 band. Although that has not yet been successful, likely due to the limited amounts of the WPreC protein available and low ionization efficiency of the N-terminal tryptic peptide owing to its high hydrophobicity, we detected the C-terminal peptides spanning up to position 151 (Table S1; Fig. S4), consistent with the previous studies of the HBV PreC (p22cr) protein (10, 11). Thus, these results indicated that WPreC lacked the CTD sequence, but most likely retained the N-terminal signal peptide accounting for its slower mobility on SDS-PAGE, as reported for HBV PreC (10, 11)

We estimated that serum WHeAg, mostly glycosylated, was ca. 10–20 μg/mL, and WPreC was ca. 2 μg/mL in the sera of chronic WHV-infected woodchucks. Serum concentrations of WHeAg and WPreC were ca. 10-fold higher than those of HBeAg and PreC in the sera of chronic HBV-infected humans respectively that we reported (11), suggesting that N-glycosylation might enhance the secretion and stability of WHeAg (20). Interestingly, we noticed the heterogeneity of the WHeAg/WPreC species appeared to change during the course of infection (Fig. S2B, S5A), suggesting that CTD processing of the WHeAg/WPreC might change dynamically during the course of infection. However, the biological significance, if any, of this heterogeneity and their dynamic change during infection remains unclear at present.

No significant amounts of WHV empty virions were found in WHV-infected woodchuck sera

Given the large excess of empty virions to complete virions in HBV-infected patients (4–6, 11), we were interested in determining if any WHV empty virions were secreted in WHV-infected woodchucks. We defined WHc here as detectable by the mAb T2221 or 19C18 but not by 1A11 or 7E9 (Fig. 1). We only detected WHc by SDS-PAGE and western blotting from several animals with high viremia titers of ca. 10¹¹ genome of equivalent (GE)/ml (Fig. 2B, lanes 23, 24; 5B, lanes 12, 13, 36, 37, S2B, lanes 25–27; S3A and S3B, lanes 2, 3; S5A, lane 4; Table S2). When detected, WHc co-migrated with WPreC0 (if present) and migrated just below WPreC1, adding to the difficulty in WHc detection (Fig. 2B, lanes 3–6, 17–20; S3A and S3B, lanes 5–7). The specificity of mAb T2221, 19C18, 7E9, and 1A11 was verified by the lack of any cross-reactive serum proteins in uninfected or resolved woodchucks (Fig. 2B, S2A, S2B, S3A and S3B). We estimated the total serum WHc (monomer) concentration from samples with detectable WHc was ca. 2 μg/ml (Table S2), which was actually close to HBc (monomer) level in HBV-infected human and chimpanzees. Nevertheless, with the much higher viremia levels in these woodchucks, the levels of detected WHc amount were equal to or only slightly over (by ca. 2-fold) that predicted from complete virions (assuming 240 copies of WHc and one copy of WHV rcDNA per complete virion), suggesting that no significant quantities of empty WHV virions were present in WHV-infected woodchucks (Table S2). This was also consistent with the results that WHc was predominantly dephosphorylated in WHV virions (Fig. S3C), indicative of complete (DNA-containing) virions, as reported for HBV (5, 11).

Figure 5. Analysis of serum WHeAg and WHV DNA kinetics in woodchucks treated with different therapies.

(A) Changes of serum WHV DNA, WHeAg, WHsAg, and intrahepatic WHV cccDNA from chronically infected woodchucks treated with wIFN-α relative to week −3 (pre-treatment baseline). Chronically infected woodchucks were treated three times a week with wIFN-α at the indicated doses for 15 weeks total, as described (15). (B) Changes of serum WHV DNA, WHeAg, and WHsAg from chronically infected woodchucks treated with the TLR7 agonist GS-9620 relative to week −2 (pre-treatment baseline). Chronic WHV-infected woodchucks were treated with GS-9620 at the indicated doses three times a week for four weeks total (M7883 and F7979) or once a week for 8 weeks (M7969 and M7994) as described (17). Woodchuck sera samples resolved by SDS-PAGE and analyzed by immunoblotting using mAbs 7E9, or 1A11, and 19C18, are shown as the examples. (C) Changes of serum WHV DNA, WHeAg, and WHsAg from chronically infected woodchucks treated with telbivudine relative to week −1 or 0 (pre-treatment baseline). Chronic WHV-infected woodchucks were treated with telbivudine at a daily dose of 10 mg/kg for 12 weeks. WHc was detected by SDS-PAGE and immunoblotting using 19C18 but not 7E9 or 1A11 whereas WHeAg and PreC were detected by all those mAbs, as annotated in B. Dashed horizontal lines indicate the lower limits of detection. Dashed vertical boxes indicate the different phases of viral clearance: black, clearance of viremia alone; grey, clearance of viremia and antigenemia, as well as intrahepatic cccDNA.

Serum WHeAg showed different characteristics during acute vs. chronic infection

As we found that serum HBeAg and PreC showed two distinct density populations in chronic HBV-infected patients (11), we were interested in determining the density profiles of the serum WHeAg and WPreC. Following CsCl isopycnic gradient ultracentrifugation of four woodchuck serum samples, we detected WHc, by SDS-PAGE and western blot analysis, in two chronic WHV-infected woodchucks (F6004 and F6005) with high viremia (ca. 10¹¹ GE/ml) (Fig. 3, lane 7; S6A, lane 7), in the same peak fraction (#18) as WHV DNA (i.e., complete virions) (Fig. S6B, lane 7), but not in the other two acutely WHV-infected woodchucks (F7394 and F7386) with lower viremia (ca. 10⁹-10¹⁰ GE/ml) (Fig. 3, lane 18; S6A and S6B, lane 19), consistent with the lack of significant amounts of empty WHV virions. WHeAg and WPreC from the two chronically-infected woodchucks also sedimented into two distinct density populations, with density either higher or lower than WHV virions (Fig. 3, S6), suggesting that WHeAg/WPreC, like HBeAg/PreC (11), might also associate with low-density serum factors, e.g., lipids or lipoprotein, to generate the low-density population. In line with that, our sucrose gradient ultracentrifugation analysis indicated that most of g-WHe1 and the WPreC1 proteins had similarly small sizes but some of them sedimented further down into multiple fractions (Fig. S7). In contrast, WHeAg/WPreC in both of the two acutely-infected woodchucks sedimented mostly to the denser fractions at the expected density of pure proteins (ca. 1.3 g/cm³) (Fig. 3, S6).

Figure 3. Different density profiles of WHeAg between chronically and acutely infected woodchucks.

Analysis of WHV virions and antigens by CsCl density gradient fractionation of the chronically and acutely WHV-infected woodchuck serum, F6005 and F7386, respectively. Gradient fractions were resolved by SDS-PAGE (regular height gel) and analyzed by immunoblotting using mAbs 7E9 and T2221. *, unknown woodchuck serum proteins cross-reactive with the anti-precore mAb 7E9, which co-migrated with the putative g-WPreC0 and/or g-WPreC1.

WHeAg was positively correlated with intrahepatic cccDNA.

Since WHc and WPreC were difficult to detect reliably by immunoblotting in the serum of most WHV-infected woodchucks, we quantified serum WHeAg (g-WHe and WHe) levels by immunoblots and correlated those with intrahepatic WHV cccDNA and other serum markers. We found a positive correlation between serum WHeAg and intrahepatic cccDNA (Fig. 4A), suggesting WHeAg could be used as a marker for intrahepatic WHV cccDNA. Furthermore, serum WHeAg was also positively correlated serum WHV DNA (Fig. 4B), and WHsAg (Fig. 4C), further supporting that serum WHeAg could faithfully reflect intrahepatic viral gene expression and replication activities in the woodchuck model.

Figure 4. Correlations between WHeAg and cccDNA, serum WHV DNA, and WHsAg.

Correlations of serum WHeAg with intrahepatic cccDNA (A), serum WHV DNA (B), and serum WHsAg (C). Serum samples and paired liver biopsies (n = 46) from 7 chronic WHV-infected woodchucks from the wIFN-α treatment study were included (15). Serum WHV DNA was measured by qPCR, WHsAg was measured by ELISA, and WHeAg was quantified by immunoblotting of the N-glycosylated WHeAg (g-WHe) and WHeAg (WHe) bands in the current study. The correlation coefficient was calculated by Spearman’s correlation test. Two-tailed p-value was calculated for a 95% confidence interval. The dashed grey line represents the detection limit of WHeAg (120 ng/ml) by our assay.

Kinetics of serum WHeAg vs. intrahepatic viral activity during natural resolution and antiviral therapies.

To explore the utility of serum WHeAg as a marker to understand the mechanisms of viral clearance during natural recovery from acute infection or in response to antiviral treatment from chronic infection, we measured serum WHeAg, together with serum WHV DNA (viremia), WHsAg, and intrahepatic cccDNA during natural resolution of acute WHV infection and in chronically-infected woodchucks treated with IFN-α, the Toll-like receptor (TLR-7) agonist GS-9620, and NUC treatments (15–17). During the resolution of acute WHV infection, viremia clearly decreased while WHeAg remained unchanged at the early phase (from week 5 to 13); subsequently, a further drop in viremia was accompanied by a drop in WHeAg, following a decrease in intrahepatic cccDNA (Fig. S5A). These results suggested that host antiviral cytokines mediated the initial clearance by inhibiting pgRNA packaging or DNA synthesis leading to the drop in viremia, non-cytolytically (16, 21, 22). The second phase of clearance likely involved cytolytic mechanisms leading to loss of both viremia and antigenemia as well as intrahepatic cccDNA, consistent with elevated liver enzyme in the serum, WHV-specific T cell response, as well as liver inflammation (16).

Similar to the kinetics of WHeAg, viremia, and intrahepatic cccDNA during recovery from acute WHV infection, we observed two similar phases of viral clearance in chronically infected woodchucks that responded well to IFN-α treatment (Fig. 5A). In M1002, the early drop in viremia from week 7 to 11 was not associated with any significant change in serum WHeAg, WHsAg, or intrahepatic cccDNA, consistent with the inhibitory effect of IFN-α on pgRNA packaging (21). Subsequently, a second drop in viremia was accompanied by a decrease in serum WHeAg, WHsAg and cccDNA from week 11 to 15, suggesting cytolytic clearance (Fig. 5A). In F1022, viremia, as well as WHeAg and WHsAg dropped slightly from week 3 to 7, with another more dramatic drop from week 7 to 11. Thus, in this animal, the clearance of viremia and antigenemia didn’t show any obvious kinetic difference. In woodchuck (F1018) that did not respond well to IFN-α treatment, we observed only a decrease in viremia, but WHeAg, WHsAg, and cccDNA did not change significantly (Fig. S5B), consistent with the notion that the direct antiviral effect of IFN-α in blocking pgRNA packaging still occurred but the subsequent cytolytic phase of clearance failed to be induced. Interestingly, in woodchuck F1013, there was an early drop (week 1 and 3 after treatment) in serum WHeAg alone, in the absence of any drop in viremia, WHsAg, or cccDNA (Fig. 5A), suggesting that the peripheral WHeAg clearance probably was mediated by WHeAg-specific antibodies, potentially induced rapidly by IFN treatment, as we observed in some of the HBV-infected chimpanzees during natural resolution (see below). This clearance was transient and serum WHeAg rebounded at week 15, which was followed by another drop in WHeAg at week 17, when accompanied by drops in viremia, WHsAg, and cccDNA (Fig. 5A). In this partial responder, the second phase of viral clearance (presumably cytolytic) again was transient, and viremia, antigenemia, and cccDNA rebounded again by week 23.

Similar to the IFN-α responders, four responders of the TLR7-agonist GS-9620, known to induce endogenous IFNs strongly (17), all showed a two-phase recovery, with the first phase featuring the drop mainly in viremia only and the subsequent drop in both WHeAg and viremia, possibly reflecting the agonist-induced non-cytolytic and cytolytic mechanisms respectively, similar to IFN-α treatment (Fig. 5B). In the two NUC (telbivudine)-treated woodchucks, the direct drug inhibition on viral DNA synthesis led to a rapid drop of viremia (within 2 weeks of treatment), while both serum WHeAg and WHsAg remained mainly unchanged as anticipated (Fig. 5C). Subsequently, NUC treatment caused a further drop in viremia as well as a decrease of serum WHeAg and WHsAg, likely due to a transient drop in intrahepatic cccDNA, known to occur with NUC treatment at relatively high dosage and/or prolonged treatment duration in the woodchuck model (23). As expected, after cession of NUC treatment, both viremia and antigenemia (WHeAg and WHsAg) rebounded (Fig. 5C).

Serum HBcrAg components as markers for intrahepatic viral activity in HBV-infected chimpanzees.

To further validate our observations on the two-phase recovery in the woodchuck model, we analyzed the kinetics of serum HBcrAg components in HBV-infected chimpanzees during the recovery phase of acute infection (14). We detected HBeAg, HBc, and PreC in the serum of HBV-infected chimpanzees by SDS-PAGE and western blot analysis (Fig. 1B, 6A), as in the genotype D HBV-infected patient (11). PreC consistently accounted for ca. 10% of total HBcrAg, with HBeAg being usually the predominant form of HBcrAg (Fig. 6A). Similar to human sera (11), HBeAg level in the chimpanzees was ca. 10 μg/mL in serum, equivalent to 10¹⁴ molecules of HBeAg, and HBc and PreC levels each were ca. 1.5 μg/mL and 1 μg/mL in serum, respectively, equivalent to 10¹³ to 10¹⁴ molecules of HBc or PreC monomer (Fig. S8A). Paired-correlation analysis of HBcrAg components indicated positive correlations between HBcrAg and HBeAg, HBcrAg and PreC, PreC and HBeAg, but not between HBcrAg and HBc (Fig. S8B–E), as we found in the human sera (11). These results thus indicated that whereas the two precore-derived proteins, PreC and HBeAg, followed similar kinetics during natural infection, serum HBc (mostly empty virions) did not. We also analyzed the viral particles in the chimpanzee serum samples by NAGE assay followed by immunoblots using different antibodies against HBc and HBsAg. We found that total HBc, representing empty virions mostly, had different kinetics than serum HBV DNA (complete virions or viremia), as revealed by Southern blotting and qPCR (Fig. 6B, S9; Table S3). On the other hand, CTD-dephosphorylated HBc, as detected by mAb 25–7, followed similar kinetics as HBV DNA, indicating dephosphorylated HBc was mostly associated with complete virions in the infected chimpanzees, as in HBV-infected patients (5, 11). Furthermore, serum HBc (empty virions), but not HBV DNA (complete virions), followed similar kinetics as HBsAg (Fig. 6B). Quantification of cccDNA from liver biopsies showed a significant positive correlation to HBc, but not to HBcrAg (Fig. 6C, 6D).

Figure 6. HBcrAg species in the sera of HBV-infected chimpanzees.

(A) Immunoblot analysis of HBcrAg components in the genotype D HBV-infected chimpanzee sera, resolved by SDS-PAGE in a high-resolution gel, using mAbs 1A11 (eAg/PreC specific), T2221 (NTD-specific), and 366–2 (CTD-specific) (Fig. 1A). (B) Analysis of serum HBV particles in infected chimpanzees by NAGE. Serum HBV particles from infected chimpanzees at the indicated time points (weeks or wk post-infection) were resolved by NAGE and transferred to nitrocellulose membrane. HBV DNA, capsid, envelope proteins (HBsAg particles co-migrating with virions) in the virions or subviral particles were detected sequentially using a ³²P-labeled HBV DNA probe, indicated HBc mAbs, and anti-HBs antibody, respectively, on the same membrane. V, virions containing either rcDNA or empty; S, viral surface or envelop proteins. Dashed boxes indicate the different phases of viral clearance involving distinct mechanisms during the resolution: black, viremia drop alone; grey, drop of viremia and antigenemia. (C) and (D) Correlations between intrahepatic cccDNA with serum HBc and HBcrAg. Serum samples and paired liver biopsies were from chimpanzees 1603 (Week 22), 1616 (Weeks 10, 22, and 38), 1618 (Weeks 12 and 18), A2A007 (Weeks 13 and 21), AOA006 (Week 5). Serum HBc and HBcrAg was quantified by immunoblotting. cccDNA quantification was normalized to mitochondrial DNA of each samples. The correlation coefficient was calculated by Spearman’s correlation test. Two-tailed p-value was calculated for a 95% confidence interval.

Analysis of individual chimpanzees with regard to serum HBcrAg components, along with viremia, serum HBsAg, and intrahepatic cccDNA, revealed interesting patterns of peripheral and intrahepatic viral activities during the acute phase of infection. In chimpanzee 1603, both viremia and HBeAg/PreC decreased (by ca. 5-fold and 3-fold, respectively), but HBc (empty virions) and HBsAg remained unchanged, from week 16 to 22 (Fig. 6A, 6B, S9A). These results together indicated that the decrease in serum HBV DNA, but not HBc or HBsAg, might be caused mainly by the immune-mediated, non-cytolytic blocking of pgRNA packaging leading to a drop in viral DNA synthesis (21, 22), as above for the first phase viremia decline in the woodchucks. On the other hand, the decrease in serum HBeAg/PreC could be due to the peripheral clearance, possibly antibody-mediated, as we observed in woodchuck F1013 above, rather than a loss of hepatic HBeAg production, given that both HBc and HBsAg remained unchanged.

In chimpanzee 1616, we observed an initial drop in viremia from week 10 to 20, without concomitant decrease in HBc or cccDNA and an increase in HBeAg/PreC and HBsAg (Fig. 6A, 6B, S9B). As in 1603 above, this initial decrease in viremia, but not antigenemia or cccDNA, was most likely caused by the non-cytolytic blocking of pgRNA packaging (21, 22). After a plateau period from week 20 to 23, there was another drop in viremia from week 23 to 38, this time accompanied by a drop in HBc (empty virions), HBeAg/PreC), as well as HBsAg (Fig. 6A, 6B, S9B). Intriguingly, intrahepatic cccDNA was increased by 5-fold over the same timeframe (Fig. S9B), suggesting viral clearance at this stage might be predominantly due to transcriptional silencing of cccDNA. Remarkably, we noticed a dramatic HBeAg loss from week 23 to 38 based on ELISA (by over 10,000-fold; Table S3), but only a very modest drop (by ca. 5-fold) based on western blotting (Fig. 6A, lane 8), likely due to immune complex formation between HBeAg and HBeAb interfering with ELISA detection of HBeAg.

In chimpanzee 1618, viremia and HBeAg decreased by ca. 5-fold and 3-fold, respectively, but PreC1 did not decrease from week 12 to 17 (Fig. 6A, 6B, S9C). However, HBc increased ca. 3-fold and HBsAg also increased ca. 6-fold (Fig. 6A, 6B, S9C), implying that non-cytolytic mechanisms, as shown in 1603 and 1616, contributed to this initial viremia drop. From week 17 to 18, there was another drop in viremia (ca. 6-fold), which was accompanied by a drop in HBc (empty virions) (ca. 3-fold) and HBeAg (ca. 3-fold), as well as cccDNA (4-fold) (Fig. 6A, 6B, S9C), suggesting cytolytic clearance.

In chimpanzee A2A007, we observed again a significant decrease (ca. 4-fold) in viremia (complete virions) with little reduction in HBc (empty virions), HBcrAg (HBc, e1, and PreC1), or HBsAg from week 13 to 21 (Fig. 6A, 6B, S9D). Over the same timeframe, there was little change of cccDNA (Fig. S9D, S9F), indicating that the substantial drop in viremia, but not antigenemia, was also caused mainly by selective blocking of pgRNA packaging caused by the non-cytolytic antiviral immune response (21, 22).

In chimpanzee AOA006, total serum HBcrAg did not change from week 5 to 7; while HBeAg and PreC1 were decreasing, HBc was increasing, thus maintaining the total HBcrAg level (Fig. 6A, 6B, S9E). Consequently, HBc became the predominant component of HBcrAg at week 7, in contrast to all other chimpanzee sera tested here. Also, while the HBc level increased by ca. 6-fold from week 5 to 7, the viremia level was steady (Fig. 6A, 6B, S9E), indicating that empty virions peaked later than DNA-containing virions during the expansion phase of acute infection. At week 9, we could no longer detect any HBcrAg components by immunoblots or HBV DNA by Southern blotting (DNA decreased by 230-fold based on qPCR) (Fig. 6A, 6B, Table S3) (14). However, intrahepatic cccDNA levels remained unchanged from week 5 to week 9, suggesting that the loss of viremia and antigenemia from week 7 to 9 was non-cytolytic and likely due to transcriptional silencing of cccDNA.

Discussion

HBcrAg is an emerging HBV marker that is being widely explored for monitoring intrahepatic cccDNA (8). To better understand this marker in the woodchuck model of HBV infection, we characterized the composition of WHcrAg in detail. We found multiple species of WHeAg and WPreC proteins derived from the precore ORF, besides WHc encoded by the core ORF, in sera of WHV-infected woodchucks. Unlike HBV, WHeAg was N-glycosylated and no significant amount of empty WHV virions was apparently detected. Kinetic analysis of WHeAg in WHV-infected woodchucks and HBcrAg components in HBV-infected chimpanzees during natural infection and in response to different antiviral treatments, together with other serum viral markers and intrahepatic cccDNA revealed multiple distinct mechanisms of viral clearance. Moreover, serum HBc, representing predominantly empty HBV virions, displayed different kinetics from that of HBeAg, PreC, and viral DNA, and was better correlated with intrahepatic cccDNA than HBeAg or HBcrAg.

The two glycosylation sites in the WHV precore/core ORF (Fig. 1A) are retained in avian hepadnaviruses but missing in HBV and other primate hepadnaviruses (7, 18). Whether N-glycosylation of WHeAg and WPreC has any roles in viral infection, and how this modification is lost in primate hepadnaviruses, are interesting questions that remain to be explored. N-glycosylation can enhance protein secretion and stabilize the modified proteins (20), possibly accounting for the high serum WHeAg/WPreC levels in woodchucks observed here (generally an order of magnitude higher than those of HBeAg/PreC in human serum (11)) and related to the apparent lack of spontaneous WHeAg seroconversion (13). The detection of glycosylated WHeAg/WPreC also strongly supports the involvement of the ER-Golgi secretory pathway, where protein glycosylation occurs, during the release of both classes of precore-derived proteins. The CTD cleavage sites of WHeAg in our study were assigned putatively, based on previous studies on HBeAg (7, 19); further experimental validation on the assignment of the cleavage sites is needed.

The secretion of PreC, in addition to the classical HBeAg (WHeAg), by both HBV and WHV suggests an important function(s) for the secreted PreC proteins, distinct from that of HBeAg/WHeAg, which is also supported by the high degree of conservation of the precore-specific region as noted earlier (24). The signal peptide, universally hydrophobic, retained in the secreted apolipoprotein M (ApoM) was shown to help anchoring ApoM to lipoprotein particles and prevent its rapid clearance from the circulation by filtration in the kidney (25). As in chronically HBV-infected patients (11), WHeAg/WPreC in chronically WHV-infected woodchucks were fractionated into a density population much lighter than virions and displayed size heterogeneity, suggesting that these WHV proteins probably associate with other factors having low density, such as lipoproteins, in the blood, which may affect their clearance and/or functions. Intriguingly, we observed predominantly a single high-density population of WHeAg/WPreC, with the expected density of pure proteins (ca. 1.3 g/cm³), in the acutely WHV-infected woodchucks, suggesting that the differential density distributions of WHeAg/WPreC in acute vs. chronic infections may indeed be associated with infection outcomes.

Kinetic analysis of individual components of serum HBcrAg in both the woodchuck and chimpanzee models, together with other traditional serum HBV markers, such as viremia and HBsAg, during natural infection and in response to antiviral therapies revealed at least three distinct phases of viral clearance, which in turn suggest at least three different mechanisms that could be involved in the resolution of HBV infection (Fig. 7). The initial phase was characterized by the drop of viremia (complete virions), alone, without concomitant decrease in HBc (empty virions), HBeAg/PreC, HBsAg, or intrahepatic cccDNA. This was observed during the initial phase of resolution of acute infection in chimpanzees and woodchucks, as well as during the initial response to IFN-α or TLR-7 agonist, which induces endogenous IFN-α. As IFNs are known to block HBV pgRNA packaging (21), which would lead to a reduction of viral DNA synthesis and a drop in viremia, the initial phase can be fully explained by the non-cytolytic cytokine-mediated clearance of viremia (Fig. 7A) (22). Similarly, the initial decrease in WHV viremia caused by NUC treatment in woodchucks was not accompanied by any decrease in antigenemia, consistent with the known mechanism of NUC to block viral DNA synthesis (Fig. 7A). Following this initial phase, viremia continued to drop, which may be subsequently accompanied by a drop of serum HBeAg/PreC. Conversely, serum HBc or HBsAg may not decrease at this time. This pattern was observed during natural resolution of HBV-infected chimpanzees and a woodchuck treated with IFN-α (F1013), suggesting that HBeAg/WHeAg can be cleared, likely by specific antibodies in the periphery (Fig. 7B). Indeed, we have observed at least one case in the infected chimpanzees (1616, week 38) where the traditional HBeAg ELISA assay detected only ca. 0.1% of the HBeAg detected by SDS-PAGE and immunoblotting analysis, suggesting that the failure to efficiently detect HBeAg by ELISA in this case was due to immune-complex formation in which serum HBe-antibodies masked the HBeAg epitopes needed for detection by ELISA that were liberated following SDS-PAGE. Finally, the third pattern of viral clearance we have observed here featured a drop in all serum viral markers, including viremia, HBc, HBeAg/PreC, and HBsAg (Fig. 7B). This was observed during natural clearance of acute infection in chimpanzees and woodchucks, as well as in response to antiviral treatments in chronically-infected woodchucks. This loss of all peripheral viral markers, usually occurring late during viral clearance, was sometimes accompanied by a loss of cccDNA in the liver, indicative of a cytolytic mechanism; but more frequently, at least in chimpanzees, was not associated with cccDNA loss, consistent with transcriptional or post-transcriptional suppression on cccDNA.

Figure 7. HBcrAg species, biogenesis, and application as biomarkers to monitor HBV gene expression and replication during natural resolution.

(A) The HBV replication cycle, from virus entry to secretion, is shown schematically. Following virus entry into hepatocytes and cccDNA formation in the nucleus, pgRNA and subgenomic viral mRNAs are transcribed from the cccDNA. During translation of the 3.5 kb precore mRNA, the N-terminal signal peptide of the precursor protein p25 is cleaved by the ER-associated signal peptidase during translocation to the ER lumen, resulting in the production of p22 in the lumen, which then traffics through the secretory pathway, where the CTD of p22 is cleaved off by preprotein convertases such as furin, to generate HBeAg secreted into the blood. Other than HBeAg, the PreC protein, retaining the signal peptide but having the CTD cleaved, is also secreted into the blood. Due to rapid processing of the precore precursor p25 to the intracellular p22 or the secreted PreC proteins, p25 may not actually exist or be detectable in the cell. On the other hand, pgRNA is the template for translation of HBc and viral polymerase. pgRNA binds to viral polymerase and both are packaged into the assembling capsid formed by HBc, a process known to be blocked by IFNα. A delayed silencing effect on HBV gene expression by IFNα was also observed in some cases (dashed line). pgRNA is subsequently reverse transcribed into rcDNA and the resulting mature nucleocapsid is enveloped by the viral envelop proteins (L-HBs, M-HBs, and S-HBs) to form virions for secretion. The reverse transcription step is targeted by NUC treatment. Empty virions, which composes HBs and HBc, are secreted to the blood at ca. 100-fold excess over DNA virions. The current HBcrAg assay detects HBeAg, PreC protein, and HBc from complete and empty virions. (B) Summary of potential clearance mechanisms that likely contributed to peripheral viral marker clearance, as characterized in the current study. *, HBsAg may continue to be expressed from the integrated HBV DNA.

The current HBcrAg assay measures a composite antigen, including HBc, HBeAg, and PreC (9–11). Here, we found that WHeAg/HBeAg was the predominant form of serum WHcrAg/HBcrAg in the serum of WHV-infected woodchucks and HBV-infected chimpanzees, as in patients (11). Thus, any factors that affect HBeAg levels, such as core promoter mutations which reduces HBeAg secretion (and presumably also PreC secretion) (26), will affect the serum HBcrAg level (Fig. 7B), rendering it unreliable as a marker for HBV cccDNA. In the case of HBV precore ORF mutations, such as the frequent G1896A, leading to premature termination and loss of precore production, no HBeAg or PreC will be secreted; thus, only HBc will be detected by the current HBcrAg assay in these cases. Whereas PreC usually displayed the same kinetics as HBeAg during natural infection, the serum HBc level, reflecting empty virions, displayed a kinetics that was distinct from that of HBeAg/PreC, suggesting that monitoring HBc, separately from HBeAg/PreC, will provide additional insights into viral activities and clearance mechanism. Indeed, we found a positive correlation between HBc, but not HBcrAg, and intrahepatic cccDNA in HBV-infected chimpanzees. However, we found a positive correlation between WHeAg and intrahepatic WHV cccDNA, suggesting WHeAg could be a faithful marker for WHV cccDNA, likely due to the apparent lack of naturally occurring WHeAg seroconversion in chronically infected woodchucks (13).

As a marker for intrahepatic cccDNA, serum HBc (mostly empty virions), has clear advantages over HBcrAg (which is mostly HBeAg) or HBsAg, since (1) HBc is essential for viral replication and persistence whereas HBeAg/PreC is not and can be reduced or eliminated by cccDNA mutations that may not impair viral replication and persistence (26); (2) integrated HBV DNA, in addition to cccDNA, can produce HBsAg but not HBc (27); and (3) unlike HBeAg/PreC, HBc is enclosed by the envelope in virions and so not subject to antibody-mediated peripheral clearance. As discussed above, comparison of HBc levels to other viral markers such as viral DNA, HBeAg/PreC, HBsAg can further help to monitor viral replication and clearance, such as the loss of complete virions without loss of empty virions suggesting selective block of pgRNA packaging and/or DNA synthesis. Interestingly, whereas HBc was usually a minor component (ca. 10–20%) of the total serum HBcrAg, HBc levels could reach levels as high as, or even higher than, HBeAg as we observed in some infected chimpanzees. The potential biological significance of this observation is currently unclear but warrants further investigation. Unlike HBV, which secretes 10- to 100-fold more empty virions than complete virions (6) (Fig. 7), no significant amounts of WHV empty virions were found. A possibly related phenomenon is that WHV secretes complete virions at 10- to 100-fold higher levels than HBV in vivo (13). Also, our observation of little to no empty virions in WHV infection is consistent with the predominantly cytoplasmic WHc staining in infected hepatocytes – representing immature and mature NCs containing pgRNA and DNA, as opposed to the predominantly nuclear HBc staining representing empty capsids (3, 13). The reason and significance for these interesting differences remain to be explored.

Our studies are limited by the relatively small number of serum and particularly, liver biopsy samples. The findings here will need to be validated by larger cohort studies in patients. Also, the sensitivity and throughput of the current western blot assay for each component of HBcrAg are fairly low, and assay development to enhance detection sensitivity and throughput will be required for eventual clinal application. Our results do indicate that careful monitoring of individual components of the composite HBcrAg, including HBc (empty virions), two types of precore-related proteins (HBeAg and PreC), along with traditional serum viral markers such as viral DNA and HBsAg, could help assess intrahepatic viral gene expression and replication including cccDNA, in order to understand mechanisms underlying natural resolution of HBV infection as well as to guide antiviral treatment.

List of abbreviation

ApoM

Apolipoprotein M

cccDNA

covalently closed circular DNA

CTD

C-terminal domain

ER

endoplasmic reticulum

HBV

hepatitis B virus

HBc

HBV core protein

HBcrAg

hepatitis B core-related antigen

HBeAg

HBV e antigen

HBsAg

HBV surface antigen

GE

genome of equivalent

IFN

interferon

NAGE

native agarose gel electrophoresis

NC

nucleocapsid

NTD

N-terminal domain

NUC

nucleos(t)ide analogue

ORF

open reading frame

pgRNA

pregenomic RNA

PreC

HBV PreC protein

rcDNA

relaxed circular DNA

RT

reverse transcriptase

SDS-PAGE

sodium dodecyl sulfate polyacrylamide gel electrophoresis

TLR

Toll-like receptor

WHV

woodchuck hepatitis virus

WHc

WHV core protein

WHcrAg

WHV core-related antigen

WHeAg

WHV e antigen

WHsAg

WHV surface antigen

WPreC

WHV PreC protein