Hepatitis B virus core antigen determines viralpersistence in a C57BL/6 mouse modelYi-Jiun Lina,1, Li-Rung Huangb,1, Hung-Chih Yangc, Horng-Tay Tzengd, Ping-Ning Hsud, Hui-Lin Wue, Pei-Jer Chena,f,2,and Ding-Shinn Chene,2

aGraduate Institute of Microbiology, dGraduate Institute of Immunology, fGraduate Institute of Clinical Medicine, cDepartment of Medical Research, andeHepatitis Research Center, National Taiwan University College of Medicine and National Taiwan University Hospital, Taipei 10002, Taiwan; and bInstitute ofMolecular Medicine and Experimental Immunology, Bonn University Hospital, Sigmund Freud, 53105 Bonn, Germany

Contributed by Ding-Shinn Chen, April 8, 2010 (sent for review November 24, 2009)

We recently developed a mouse model of hepatitis B virus (HBV)persistence, in which a single i.v. hydrodynamic injection of HBVDNA to C57BL/6 mice allows HBV replication and induces a partialimmune response, so that about 20–30% of the mice carry HBV formore than 6 months. The model was used to identify the viral an-tigen crucial for HBV persistence. We knocked out individualHBV genes by introducing a premature termination codon to theHBV core, HBeAg, HBx, and polymerase ORFs. The specific-gene-deficient HBV mutants were hydrodynamically injected into miceand theHBVprofiles of themiceweremonitored. About 90%of themice that received the HBcAg-mutated HBV plasmid exhibited highlevels of hepatitis B surface antigenemia and maintained HBsAgexpression for more than 6 months after injection. To map the re-gion of HBcAg essential for viral clearance, we constructed a set ofserial HBcAg deletion mutants for hydrodynamic injection. We lo-calized the essential region of HBcAg to the carboxyl terminus,specifically to the 10 terminal amino acids (HBcAg176–185). Themajority of mice receiving this HBV mutant DNA did not elicita proper HBcAg-specific IFN-γ response and expressed HBV virionsfor 6 months. These results indicate that the immune response trig-gered in mice by HBcAg during exposure to HBV is important indetermining HBV persistence.

hepatitis B surface antigenemia | hydrodynamic injection

Persistent hepatitis B virus (HBV) infections affect about 350million people worldwide and are a major health problem.

Factors directing the infection toward chronicity have beenstudied extensively. The exposure of neonates or young childrento a high HBV viral load, together with hepatitis B e antigen(HBeAg), predicts a high rate of persistent HBV infection. Arecent genomewide association study identified HLA-DP poly-morphisms as another factor in the persistence of HBV infections(1). Nevertheless, the immune mechanisms that lead to HBVpersistence have not been resolved. To address this issue, a com-monly used mouse model, such as the HBV transgenic mouse, hasbeen used to study the possible mechanisms. However, the maindrawback of HBV transgenic mouse models is that they are im-munologically tolerant of viral antigens. Therefore, to explore theissue of HBV persistence, the adoptive transfer of HBV-primedimmune cells or other manipulations must be used to overcomethis tolerance. Another alternative is to introduce the HBV ge-nome into the mouse liver by hydrodynamic injection through thetail vein. With this approach, HBV was shown to replicate in themouse liver, and the immune responses against HBV proteins toclear the HBV infection could be documented in 1–2 weeks afterthe injection (2). Recently, we improved this approach by modi-fying the HBV DNA plasmid and injecting the plasmid intoC57BL/6 mice and succeeded in delaying the mouse immuneclearance of HBV (3). Clearance could be postponed for 6–8weeks and about 10–30% of the injected mice maintained HBVpersistence even up to 6 months after injection. This animalmodel provides a platform from which to study the mechanisms ofHBV persistence.

As a first step in a systematical investigation of the HBV genesthat influence viral persistence in this model, we knocked outeach of the ORFs in the HBV genome by site-directed muta-genesis. Each mutant HBV plasmid, deficient in the expressionof only one viral gene, was then injected into C57BL/6 mice andHBV persistence was determined by serum HBV surface antigen(HBsAg) and HBV DNA detection. We found that the mutantHBV DNA that did not express the HBV core (HBc) gene es-tablished HBsAg persistence for more than 6 months in about90% of the injected animals. More interestingly, the HBV mu-tant expressing a shortened core antigen of 175 amino acidsestablished not only HBsAg but also HBV virion persistence formore than 6 months in 70% of the injected animals. Here, wedemonstrate the accessibility of this model for investigating thecontribution of various HBV mutants to HBV persistence. Theseresults should allow us to understand and explore the mecha-nisms involved in HBV persistence.

ResultsInfluence of Individual HBV Antigens on Viral Persistence in C57BL/6Mice. Although some lines of evidence suggest that fetal or peri-natal exposure to HBeAg contributes to the persistence of HBV,a systematic investigation of the role of each individual HBV genehad not been conducted. Our mouse model provides a platformfor this reverse genetic analysis (3). Therefore, we constructed anarray of viral mutants that contained premature stop codons in thecore, HBe, HBx, or polymerase (pol) ORFs of a replication-competent HBV plasmid pAAV/HBV1.2 using in situ site-directed mutagenesis (Fig. S1 A and B and Table S1). Themutations were carefully selected and did not affect the trans-lation of other overlapping frames. Ten micrograms of wild-type(WT) or mutant pAAV/HBV1.2 were injected hydrodynamicallyinto the tail veins of male C57BL/6 mice, as described previously(3). After injection, the mice were regularly bled to monitor theserum levels of HBsAg, HBeAg, and HBVDNA or were killed toevaluate the intrahepatic viral transcription and replication, andHBcAg expression.Of the mice receiving WT pAAV/HBV1.2, ≈50% remained

HBsAg positive at 12 weeks postinjection (wpi) (Fig. 1 A–D).The serum levels and the persistence rates of HBsAg in micereceiving HBeAg-null or pol-null mutants were similar to thoseof the mice receiving WT DNA, from the first day to the 12thweek after hydrodynamic injection (Fig. 1 A and B). Theseresults indicate that the absence of HBeAg or polymerase did

Author contributions: Y.-J.L., L.-R.H., P.-N.H., P.-J.C., and D.-S.C. designed research; Y.-J.L.,L.-R.H., and H.-T.T. performed research; Y.-J.L., L.-R.H., and P.-J.C. analyzed data; and Y.-J.L., L.-R.H., H.-C.Y., H.-L.W., and P.-J.C. wrote the paper.

The authors declare no conflict of interest.1Y.-J.L. and L.-R.H. contributed equally to this work.2To whom correspondence may be addressed. E-mail: peijerchen@ntu.edu.tw or chends@ntu.edu.tw.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1004762107/-/DCSupplemental.

9340–9345 | PNAS | May 18, 2010 | vol. 107 | no. 20 www.pnas.org/cgi/doi/10.1073/pnas.1004762107

not alter the HBsAg persistence rate in mice after hydrodynamicinjection.In contrast, almost all of the mice receiving HBeAg/core-null

pAAV/HBV1.2 expressed high levels of HBsAg in their sera and93.3% of them even remainedHBsAg positive at 12 wpi (Fig. 1C).The mice receiving the HBx-null mutant also exhibited nearly90% persistence at 12 wpi, although the HBsAg levels in thesemice were lower in the beginning than those of the mice receivingWT HBV (Fig. 1D). The levels of HBsAg then increased in themice receiving the HBx-null mutant, whereas those of the micereceiving theWTHBVDNAdeclined rapidly. These data suggestthat the expression of HBcAg and HBx during HBV exposureinfluences viral persistence in C57BL/6 mice.The levels of serum HBV DNA in the hydrodynamically

injected mice were also quantified using real-time PCR. In themice receiving WT pAAV/HBV1.2, the average titer of serumHBV DNA in 15 injected animals was below 1 × 104 copies permilliliter at 1 day postinjection (dpi) and reached 2.16 × 106

copies per milliliter at 7 dpi (Fig. 2A). At later time points, somemice lost both serum HBsAg (Fig. 1D) and HBV DNA, whereasothers remained positive for both. The average titer of serumHBV DNA in all 15 injected mice dropped below 1 × 105 copiesper milliliter after 7 dpi (Fig. 2A). At all time points, the averagetiters of serum HBV DNA in the mice receiving HBeAg-nullpAAV/HBV1.2 were similar to those of the mice receiving WT

DNA. On the contrary, in the absence of HBcAg or pol, thereplication cycle of HBV was disrupted, and no HBV DNA wasdetected at any time point after the first two time points, whenthe presence of HBV DNA could be attributable to residualinjected DNA (Fig. 2 B and C). In the absence of HBx, thereplication efficiency of HBV was reduced. The serum HBVDNA of the mice receiving the HBx-null mutant was detectablebut lower than that of the mice receiving WT HBV (Fig. 2D).This observation is consistent with the previous study, whichshowed that HBx is required for HBV replication in vivo (4).

Negative Correlation Between Intrahepatic Expression of HBcAg andHBV Persistence. To further characterize the viral transcription,replication, and protein expression patterns of the variousmutants, we collected the liver tissues of the mice receiving WTpAAV/HBV1.2 or each individual mutant at 3 and 33 dpi andassayed for HBV replication intermediates and transcripts usingSouthern and Northern hybridization. The episomal input DNAwas detected in the livers of all mice receiving either WT ormutant pAAV/HBV1.2 at 3 dpi (Fig. 3A Top). However, HBVreplication intermediates, including relaxed circular (RC) DNAand single-stranded DNA (ssDNA), were only detected in thelivers of mice receiving WT, HBx-null, or HBeAg-null pAAV/HBV1.2. For unknown reasons, most of the replication inter-mediates in the mice receiving the HBeAg-null mutant weressDNA (Fig. 3A Top, lanes 3 and 4). In this hydrodynamics-basedHBV model, most of the HBV transcripts were derived from theinput plasmid. Therefore, as long as the input DNA remainedpresent, the HBV transcripts could be detected in the liver (Fig.3A Middle). However, only low levels of HBV transcripts wereobserved in the livers of the mice receiving mutants that did notexpress HBx protein, including the HBx-null mutant and triple-knockout (tko) pAAV/HBV1.2 (Fig. 3A Middle, lanes 7 and 8).We also observed a reduction in both the HBV transcription/replication efficiency (Fig. 3A Top, lanes 7 and 8) and HBcAgexpression (Fig. 3A Bottom, lanes 7 and 8). The intrahepatic ex-pression of HBcAg was only observed in the mice receiving WT,HBeAg-null, or pol-null pAAV/HBV1.2, but not in those re-ceiving the HBeAg/core-null, HBx-null, or tko mutant (Fig. 3ABottom). Serum HBeAg was undetectable in the mice receivingthe HBeAg/core-null, HBeAg-null, or tko mutant at 2 and 32 dpi(Fig. 3). All mice expressed high levels of serum HBsAg at 2 dpi(Figs. 1 and 3).

Fig. 1. HBsAg persistence rates inmice receivingmutants that do not expressHBcAg or HBx are higher than those in the mice receiving WT pAAV/HBV1.2.C57BL/6 mice were injected hydrodynamically with 10 μg of pAAV/HBV1.2 ora mutant construct. The serum HBsAg titers (Left) and the rates of positiveserum HBsAg (Right) in the mice receiving WT (●) pAAV/HBV1.2 were com-pared with those in mice receiving each individual mutant DNA (○), includingthe HBeAg-null (A), pol-null (B), HBeAg/core-null (C), and HBx-null mutants(D). The serum HBsAg titers were determined at the indicated time pointswith an enzyme immunoassay [calculated as signal/noise (S/N) ratios]. Posi-tivity for HBsAg was defined as S/N ≧ 2. N = number of mice in each experi-ment. Error bars indicate SD here and in the other figures. The statisticalP values were analyzed by Kaplan–Meier analysis and significant differenceswere observed in the Right panels of C (P = 0.0079) and D (P = 0.0224).

Fig. 2. HBV DNA in the sera of mice receiving WT or mutant pAAV/HBV1.2.The serum HBV DNA in the hydrodynamically injected mice at the indicatedtime points was quantified by real-time PCR. The titers of serum HBV DNAobserved in the mice receiving WT pAAV/HBV1.2 (●) were compared withthose of mice receiving each individual mutant DNA (○), including theHBeAg-null (A), HBeAg/core-null (B), pol-null (C), and HBx-null mutants (D).The detection limit for HBV DNA in our system was 100 copies per milliliter.

Lin et al. PNAS | May 18, 2010 | vol. 107 | no. 20 | 9341

MICRO

BIOLO

GY

The serumHBsAg levels and the intrahepatic levels of the inputDNA, HBV replication intermediates, transcripts, and HBcAg insome mice receiving WT, HBeAg-null, or pol-null pAAV/HBV1.2 declined to undetectable levels at 33 dpi (Fig. 3B). Incontrast, the input DNA persisted in the livers of all mice re-ceiving the HBeAg/core-null, HBx-null, or tko mutant at 33 dpi(Fig. 3B Top). Despite the lack of detectable HBV replication inthe livers of these mice, high to moderate levels of HBV tran-scripts and HBsAg were readily detected in either the livers orsera (Fig. 3B). Therefore, only the HBV mutants that failed toexpress a detectable amount of intrahepatic HBcAg could persistin hepatocytes and maintain a continuous expression of HBVproteins such as HBsAg.To further verify the inverse correlation between HBcAg and

HBV persistence, we coinjected the HBcAg-expressing plasmidwith HBeAg/core-null pAAV/HBV1.2, which promotes HBVclearance in mice (SI Text, and Fig. S2 A–C). These results to-gether indicate that HBcAg plays a vital role in intrahepatic HBVDNA persistence. The presence of HBcAg may trigger intra-hepatic antiviral responses and facilitate the clearance of bothinfected hepatocytes and the input HBV DNA in this hydrody-namics-based mouse model. Consistently, our previous studydemonstrated that the HBcAg-specific cellular immunity gener-ated by a DNA vaccine encoding HBcAg was crucial for theclearance of HBV during both the acute phase and the chroniccarriage phase in this model (3).

C-Terminal Domain of HBcAg Is Critical for HBV Clearance. BecauseHBcAgplays a crucial role in the persistence ofHBsAg, we tried toidentify the region of HBcAg that is important for HBsAg per-sistence. HBcAg consists of two domains: the assembly domain(amino acids 1–149), which forms the contiguous spherical shell,and the protamine-like domain (amino acids 150–183/185), whichis responsible for RNA packaging and DNA synthesis. We gen-erated a series of C-terminally truncated HBcAg mutants derivedfrompAAV/HBV1.2, which resulted in the differential deletion ofthese domains (Fig. 4A). Each of the HBcAg mutants containeda premature stop codon at a unique site within the HBc ORF,which did not alter the amino acid sequence of the overlapping polORF, although the mutations may have unavoidably affected theoverlapping precore protein sequence and subsequently expressedHBeAg (Fig. 4A, see legend for details). However, from the abovedata, viral HBeAg is not critical for HBsAg persistence in this

mouse model. Hence we mainly focused on the role of HBcAg inthe persistence of HBsAg. Each individual HBcAg-truncatedmutant, WT, and HBeAg/core-null pAAV/HBV1.2 was hydro-dynamically injected intomice and the serum levels of HBsAg andHBeAg were examined (Fig. 4 B–E). The significant difference inthe HBsAg persistence rates of the mice receiving WT DNA andthose receiving the HBeAg/core-null mutant is consistent with theabove data (Figs. 1C and 4 B and C). The mice receiving HBc118pAAV/HBV1.2 showed a pattern of HBsAg persistence verysimilar to that of the mice receiving the HBeAg/core-null mutant.It is noteworthy that the HBc118 mutant retained only a partialassembly domain of HBcAg and therefore could not form coreparticles. The mice receiving HBc150 pAAV/HBV1.2, whichcontains the intact assembly domain but no protamine-like do-main of HBcAg, and the HBeAg/core-null mutant exhibitedidentical patterns of HBsAg persistence. Furthermore, the per-sistence rates in themice receiving theHBc166 orHBc175 pAAV/

Fig. 4. pAAV/HBV1.2 mutants expressing C-terminal-truncated HBcAg en-hanced HBsAg persistence in mice. (A) A schematic map of the differentpAAV/HBV1.2 mutants with a variety of C-terminal truncations in HBcAg.Each individual HBcAg mutant contains a unique single-nucleotide mutation(marked as a match with a yellow head), which results in a premature stopcodon (*) within the HBc coding region. The integrity of HBeAg corre-sponding to each mutant is shown on the Right. The titers of serum HBsAg(B) and HBeAg (D) and the positive rates of serum HBsAg (C) and HBeAg (E)in the mice receiving WT (●), HBeAg/core-null (○), HBc118 (▲), HBc150 (△),HBc166 (▼), or HBc175 (□) pAAV/HBV1.2 are shown. The differences in theHBsAg persistence rates in the mice receiving WT DNA or the individualmutants were analyzed by Kaplan–Meier analysis and were shown to besignificant (P < 0.01).

Input DNA

3.5 kb

wt e-null e/c-null x-null tko pol-null

RC DNA

SS DNA

2.1/2.4 kb

mGAPDHHBcAg

β actin

HBsAgHBeAg

D2 Serum

HBsAgHBeAg

25481

20336

22102000

23602120

1541214314

17101620

2165422032

Input DNA

RC DNA

SS DNA

3.5 kb2.1/2.4 kb

mGAPDHHBcAgβ actin

HBsAgHBeAg

10310

55

D32 Serum101560

1440

HBsAgHBeAg

1440

12011071

11601180

1047

66

SB

NB

WB

A Bwt e-null e/c-null x-null tko pol-null

Fig. 3. HBV replication and transcription, and intrahepatic HBcAg expres-sion in mice receiving WT or mutant pAAV/HBV1.2 injections. Liver sampleswere collected at 3 dpi (A) or 33 dpi (B) after the hydrodynamic injection of10 μg of WT or mutant DNA. The intrahepatic levels of the input HBV DNA,RC, ssDNAs, and the 3.5-kb pregenomic and 2.1/2.4-kb surface mRNAs weredetermined using Southern and Northern blotting. The intrahepatic levels ofglyceraldehyde-3-phosphate dehydrogenase mRNA (mGAPDH) were used asthe loading control for Northern blotting. The levels of β-actin were used asthe loading control for Western blotting. One day before the mice werekilled, serum samples were collected from the mice and HBsAg and HBeAgwere measured. The titers of HBsAg (S/N) and HBeAg (signal/cutoff, S/CO)are shown beneath the figures.

9342 | www.pnas.org/cgi/doi/10.1073/pnas.1004762107 Lin et al.

HBV1.2 mutant, both of which result in truncated HBcAg witha partially impaired protamine-like domain, were slightly lowerthan those of mice receiving HBeAg/core-null mutant but signif-icantly higher than those of mice receiving WT DNA. At 12 wpi,84.5% of the mice receiving the HBc175 mutant DNA wereHBsAg positive, whereas only 20%of themice receivingWTHBVDNA remained HBsAg positive.In addition to HBsAg, HBe antigenemia was also present in the

mice receivingWT,HBc150, HBc166, orHBc175 pAAV/HBV1.2,with markedly higher levels of serumHBeAg in the mice receivingthe HBc150 mutant DNA (Fig. 4 D and E). The persistence ofHBeAg and HBsAg correlated well in the mice receiving theseDNAs. As expected (5), HBeAg could not be detected in the seraof either themice injected with theHBeAg/core-null mutant DNAor those receiving the HBc118 mutant DNA.

Deletion of the 10 Amino Acids at the C Terminus of HBcAg PromotesHBV Persistence. Because HBcAg is critical for the HBV life cycle,HBcAg mutations may affect HBV replication and subsequentlyinfluence HBV persistence rates. Therefore, we examined theintrahepatic viral replication, transcription, and protein expres-sion in mice receiving the WT or different HBcAg mutants.Previous studies have shown that the HBVmutant with truncatedHBcAg1–173, equivalent to HBcAg1–175 of genotype A used inour study, behaves like WT HBV, whereas mutant HBcAgs withless than 173 residues exhibit impairedDNA synthesis during viralreplication in vitro (6, 7). In the mice receiving HBc175 pAAV/HBV1.2, the intrahepatic levels of viral replication were similarto those of the mice receiving the WT HBV DNA at both 3 dpi(Fig. 5A) and 41 dpi (Fig. 5B). Consistent with intrahepatic viralreplication, serum viral DNA was readily detected in these mice[WT: (1.77 ± 0.33) × 106 and (2.89 ± 2.05) × 105; HBc175 pAAV/HBV1.2: (7.43 ± 1.09) × 106 and (3.43 ± 3.77) × 105 at 1 wpi and4 wpi, respectively]. All other HBc mutants, including HBeAg/core-null, HBc118, HBc150, and HBc166 pAAV/HBV1.2,showed impaired viral replication, which is in agreement with theresults of previous studies (Fig. 5 A and B). The levels of viraltranscription from the input DNA, which was used as the tem-plate, were fairly similar in the livers of the mice receiving theWTand HBc mutant DNAs. Despite the similar transcription levelsamong the mice receiving different mutants, the expression of theviral proteins was not equal. Among the truncated forms of

HBcAg, only HBcAg1–166 and HBcAg1–175 were detected witheither immunohistochemical staining (Fig. 5D xv and xviii) orWestern blotting analysis (Fig. 5C lanes 6 and 7). However, theexpression of HBsAg was similar in the mice receiving the WT ormutant DNA (Fig. 5DMiddle). Our failure to detect the truncatedHBcAgs was not attributable to the failure of the antibody torecognize them because this antibody has been proven functional.Therefore, even in the presence of HBcAg and HBV replication,the mice receiving HBc175 pAAV/HBV1.2 exhibited a dramati-cally elevated HBsAg persistence rate compared with that of micereceiving WT DNA (Fig. 4C). Taken together, our observationssuggest that a stretch of 10 amino acids at the C terminus ofHBcAg (HBcAg176–185) is critical for the clearance of not onlyHBsAg but also HBV.

Impaired Immune Response to HBV Core Antigen in Mice ReceivingHBc175 pAAV/HBV1.2. Because HBcAg functions as a potent im-munogen, inducing both the humoral and cellular immuneresponses, we first examined the capacity of WT HBV and thedifferent HBc mutants to induce anti-HBc antibodies in mice.We observed that all of the mice receiving WT pAAV/HBV1.2produced anti-HBc at 1 wpi (Fig. 6A), consistent with the factthat HBV-infected patients produce anti-HBc. In contrast, noneof the mice receiving the HBeAg/core-null, HBc118, or HBc150mutant generated anti-HBc even at 13 wpi. The absence of anti-HBc in these mice correlated well with their undetectableHBcAg expression (Fig. 5 C and D). Interestingly, the kinetics ofanti-HBc production were slow in the mice receiving the HBc166or HBc175 mutant. At 1 wpi, anti-HBc was detected in 33.3%and 53.8% of mice receiving the HBc166 and HBc175 mutants,respectively. However, anti-HBc positivity reached 100% in themice receiving the HBc175 mutant at 8 wpi, whereas 41.7% ofthe mice receiving the HBc166 mutant remained anti-HBc neg-ative at 13 wpi (Fig. 6A). Even though its major immunodo-minant region is located around residue 80 (8), HBcAg1-166 wasstill unable to elicit a strong B-cell response in some of the mice.These results indicate that the C terminus of HBcAg promotesthe proper B-cell response to HBcAg.It has been shown that IFN-γ production in T cells in response

to HBcAg stimulation correlates with the resolution of HBVinfections (9). Therefore, we investigated the ability of differentHBc mutants to elicit IFN-γ secretion using an enzyme-linked

wt e/c-null c118 c150 c166 c175A

B

C

Di ii

iv v

vii viii

x

xiii xiv xv

xvi xvii

iii

vi

ix

xi xii

xviii

Input DNA

RC DNA

SS DNA

3.5 kb2.1/2.4 kb

mGAPDH

Input DNA

RC DNA

SS DNA

3.5 kb2.1/2.4 kb

mGAPDH

1 2 3 4 5 6 720 kDa-15 kDa-

10 kDa-

β actin

SB

NB

SB

NB

NC HBs HBc

WT

e/c-null

c118

c150

c166

c175

wt e/c-null c118 c150 c166 c175 Fig. 5. HBV replication, transcription, and translation inmice receiving WT or mutant pAAV/HBV1.2. Liver sampleswere collected from mice receiving WT, HBeAg/core-null,HBc118, HBc150, HBc166, or HBc175 pAAV/HBV1.2 at 3 dpi(A) and 41 dpi (B) to examine viral replication and tran-scription, as described in Fig. 3. (C)Western blot analysis ofthe intrahepatic expression of HBcAg (arrows) in micereceiving PBS only (lane 1, the negative control), WT (lane2), HBeAg/core-null (lane 3), HBc118 (lane 4), HBc150 (lane5), HBc166 (lane 6), or HBc175 (lane 7) pAAV/HBV1.2 at 41dpi. Intrahepatic β-actin expression was used as the load-ing control. (D) Immunohistochemical staining at 3 dpi forHBsAg (ii, v, viii, xi, xiv, and xvii), HBcAg (iii, vi, ix, xii, xv,and xviii), and the negative control (i, iv, vii, x, xiii, and xvi)in the livers of mice receiving WT (i–iii), HBeAg/core-null(iv–vi), HBc118 (vii–ix), HBc150 (x–xii), HBc166 (xiii–xv), orHBc175 (xvi–xviii) pAAV/HBV1.2.

Lin et al. PNAS | May 18, 2010 | vol. 107 | no. 20 | 9343

MICRO

BIOLO

GY

immunospot (ELISPOT) assay. Splenocytes isolated from micereceiving WT HBV or the HBc mutants were cocultured withrecombinant HBcAg and the numbers of responding IFN-γ-secreting cells were measured. In the mice receiving WT HBVDNA, the frequency of HBcAg-specific IFN-γ-secreting cells was346/106 splenocytes (Fig. 6B). However, none of the mice re-ceiving the HBc mutants, including HBeAg/core-null, HBc150,and HBc175 pAAV/HBV1.2, induced IFN-γ-secreting cells ata similar level to that observed in mice receiving WT HBV DNA.Significant differences (P < 0.001) were noticed between themice injected with WT HBV and those receiving each individualHBc mutant DNA. Deletion of HBcAg residues 176–185 seemedsufficient to prevent HBcAg from triggering the IFN-γ responsein mice. Using an ELISPOT assay, we excluded the contributionof a critical T-cell epitope located in this region to this phe-nomenon (Fig. S3). Together, our data suggest that the increasedHBV persistence in mice receiving the HBc mutant DNAs (Fig.4) resulted from the inability of these truncated HBcAgs to in-duce an efficient protective immune response, including theproduction of IFN-γ. Further studies are required to investigatethe underlying mechanisms.

DiscussionIn this study, we systematically investigated the effect of each in-dividual HBV gene on viral persistence using a reverse geneticsstrategy and a hydrodynamics-based immunocompetent mousemodel. We found that knocking out HBcAg or HBx, but notHBeAg or pol, led to HBV persistence in mice. We further lo-calized the region of HBcAg critical for viral clearance to the 10residues at its C terminus, and deletion of a mere 10 residues at theHBcAg C terminus mitigated the host immune response to HBV.The outcome of HBV infection, recovery or persistence, is

controlled by the interplay between viral proteins and host fac-tors. A previous study revealed that the HBcAg-induced immuneresponse is associated with recovery from hepatitis B (10), andIFN-γ production is associated with the resolution of HBVinfections (11). Moreover, HBcAg-stimulated IFN-γ responseswere detected in patients with acute hepatitis rather than chronichepatitis (12). Our study consistently identified HBcAg as themost immunogenic viral antigen during the early course of HBVinfection. Knockout of HBcAg significantly promoted HBVpersistence in C57BL/6 mice receiving HBV DNAs. Whereasnonneutralizing anti-HBc antibodies do not appear to playa critical role in viral clearance, the robust IFN-γ response in-duced by WT HBcAg is linked to the resolution of HBV. Incontrast, all of the mice receiving HBcAg mutants, including theHBc175 mutant, only exhibited weak IFN-γ responses and failedto purge the HBV.

Interestingly, the 10 residues at the C terminus of HBcAg arecritical for inducing robust immune responses against HBV. Itremains unclear whether the primary sequence of HBcAg176–185or a possible conformational change resulting from the deletion ofthis region plays the key role. Capsid formation, the main featureof HBcAg, would not be disrupted in mice receiving HBc150,HBc166, or HBc175 pAAV/HBV1.2 because all of the mutantscontain the essential assembly domain (13). However, the pro-longed HBV persistence in mice receiving either the HBc166 orthe HBc175 mutant suggests that these two C-terminally trun-cated HBcAgs and WT HBcAg behave differently in the in-duction of the antiviral immune responses. We were unable todetect the intrahepatic expression of HBcAg1–150 probably be-cause this truncated protein was unstable and subsequently de-graded by proteasome. Alternatively, it might be rapidly secreted.The highly conserved C-terminal domain of HBcAg controlsHBV pregenomic RNA (pgRNA) binding and viral DNA syn-thesis. HBcAg1–166 failed to support viral replication, whereasHBcAg1–175 could. HBsAg persistence can occur irrespectiveof HBV replication, which suggests that the deletion ofHBcAg176–185 promotes HBV persistence but retains HBVreplication almost intact.So far no B-cell or T-cell epitopes have yet been identified in the

C terminus of HBcAg. Experimentally, we also did not find a can-didate T-cell epitope in this region (Fig. S3). Therefore, it is veryunlikely that these 10 residues directly induce strong adaptive im-mune responses againstHBV.Alternatively,HBcAgmay influenceHBV persistence by evoking innate immunity. Because it is likelythat HBcAg1–166 still has the ability to bind pgRNA without fur-ther reverse transcription, it is possible that the C-terminal trun-cation of HBcAg prevents the exposure of the pgRNA, which mayfunction as a ligand for intracellular pattern recognition receptors.The encapsidated ssRNA, rather than HBcAg itself, stimulatesToll-like receptor 7 signaling inmice (14).Moreover, thebindingofthe C-terminal domain of HBcAg to membrane heparin sulfate onthe surfaces of macrophages, B cells, and dendritic cells mediatesthe breakdown of the viral capsid (15–17). This interaction canstimulate proinflammatory cytokine production. Therefore, thelack of the 10 HBcAg C-terminal residues causes a failure to elicitsuccessful immune responses to clear the HBV infection.In addition to HBcAg, we also observed increased HBV per-

sistence in the absence of HBx. It has been demonstrated thatHBx transactivates the transcription of the major HBV genes,including HBcAg (4, 18). Furthermore, some patients in theinactive viral replication phase of chronic hepatitis B infectionsdo not express HBcAg in their liver (19), suggesting that the lackof HBcAg may contribute to HBV persistence. In this model,knockout of the HBx gene dramatically reduced the HBV tran-

Fig. 6. Humoral immune responses and IFN-γ responses stimulated by WT or individual mutant forms of HBcAg. (A) The percentage of anti-HBc-positive micereceiving WT or individual HBc mutant pAAV/HBV1.2 at different time points after hydrodynamic injection. The blood of the mice receiving WT (n = 10),HBeAg/core-null (n = 12), HBc118 (n = 12), HBc150 (n = 12), HBc166 (n = 12), or HBc175 (n = 13) DNA was collected at 1, 4, 8, and 13 wpi to examine theproduction of anti-HBc antibodies. The bars representing HBeAg/core-null, HBc118, and HBc150 mutants are absent because none of the corresponding miceproduced anti-HBc. (B) The IFN-γ responses of splenocytes from mice receiving a mock vector (negative control, NC), WT, HBeAg/core-null, HBc150, or HBc175pAAV/HBV1.2 (n = 3). Each sample was collected at 10 dpi and subjected to an ELISPOT assay to measure the frequency of HBcAg-specific IFN-γ-secreting cells.The results are expressed as spot-forming cells (SFCs) per million splenocytes. The asterisk symbolizes significant differences between the WT group and eachof the other groups (P < 0.001).

9344 | www.pnas.org/cgi/doi/10.1073/pnas.1004762107 Lin et al.

scripts. Therefore, we surmise that the increase in the HBsAgpersistence rate in mice receiving HBx-null pAAV/HBV1.2 wasattributable to the low expression of HBcAg in their liver. Thisindicates that the expression of HBcAgmight be tightly controlledby HBV itself, such as HBx, to ensure persistent infections.Although HBeAg and HBcAg share 149 residues and are cross-

reactive at theT-cell level during infection and immunization, theyapparently elicit different patterns of immune responses, as in-ferred from the comparison of the HBV persistence rates in micereceivingWTormutant pAAV/HBV1.2 (Fig. 1A andC) andothersystems (20). Previous studies have demonstrated that HBeAgexhibits some immunoregulatory functions (20, 21). However, wefailed to observe any immunoregulatory effect of HBeAg in ourmodel (Fig. 1A). There are two plausible explanations of thisdiscrepancy. First, the timing for HBeAg to execute its immuno-regulatory function may be critical. Secreted HBeAg may have animmunoregulatory function in utero and may induce a T-cell tol-erance of HBeAg/HBcAg in exposed neonates (22). Infants bornto HBeAg-seropositive mothers often develop chronic hepatitisrather than acute hepatitis after infection, supporting the aboveargument (23). Second, the levels of serum HBeAg in the micereceiving hydrodynamic injections of pAAV/HBV1.2 and theHBeAg-transgenic mice may not be equivalent. Indeed, it is clearthat the levels of serumHBeAg influence the potency ofHBeAg ininducing T-cell tolerance (24). Taken together, these data suggestthat some mechanisms other than the immunoregulatory func-tions of HBeAg contribute to the establishment of HBV persis-tence in this mouse model.In conclusion, HBcAg is a major viral determinant of HBV

persistence, and the region HBcAg176–185 is critical for thispersistence. Enhanced viral persistence resulted from the impaired

host immunity induced by the 10-residue-long C-terminal trunca-tion of HBcAg. The underlying mechanism responsible for thisphenomenon remains unclear and is currently under investigation.

Materials and MethodsPlasmids. Site-directed mutagenesis for generating all pAAV/HBV1.2 mutantswas performed by QuickChange II site-directed mutagenesis kit (Stratagene)following the manufacturer’s protocol. For more details, see SI Text.

Animal Study. C57BL/6 mice (male, 6 to 7 weeks old) were anesthetized withketamine and xylazine. Ten micrograms of HBV plasmid DNA were injectedinto the tail veins ofmicewithin 5 s in a volume of PBS equivalent to 8%of themouse body weight (SI Text).

Detection of HBV Antigen, Antibody, and HBV DNA. Serum levels of HBsAg,HBeAg, anti-HBc of the mice were determined using the AXSYM system kits(Abbott), and the reporting unit is S/N or S/CO ratio. The cutoff value fordetermining HBsAg-, HBeAg-, and anti-HBc-positivity is S/N ratio ≧ 2, S/CO ≧ 1,and S/CO < 1, respectively. For detection of serum HBV DNA, each serumsample was pretreated with 25 units of DNase I (Roche) at 37 °C for over-night and total DNA was extracted and detected for HBV DNA by real-timePCR as previously described (25) (SI Text)

Southern and Northern Hybridization. HBV viral RNA and replicative DNAintermediates were detected by Northern and Southern blot analysis of totalliver RNA and DNA, respectively, as previously described (26).

IFN-γ Enzyme-Linked Immunospot Assay. Splenocytes were prepared from themice receiving hydrodynamic injection at 10 dpi. The IFN-γ enzyme-linkedimmunospot (ELISPOT) assay was performed as described in SI Text.

ACKNOWLEDGMENTS. This work was supported by National Science CouncilGrants NSC97-2321-B-002-025 and NSC98-2321-B-002-003.

1. Kamatani Y, et al. (2009) A genome-wide association study identifies variants in theHLA-DP locus associated with chronic hepatitis B in Asians. Nat Genet 41:591–595.

2. Yang PL, Althage A, Chung J, Chisari FV (2002) Hydrodynamic injection of viral DNA: Amousemodel of acute hepatitis B virus infection. Proc Natl Acad Sci USA 99:13825–13830.

3. Huang LR, Wu HL, Chen PJ, Chen DS (2006) An immunocompetent mouse model forthe tolerance of human chronic hepatitis B virus infection. Proc Natl Acad Sci USA 103:17862–17867.

4. Keasler VV, Hodgson AJ, Madden CR, Slagle BL (2007) Enhancement of hepatitis Bvirus replication by the regulatory X protein in vitro and in vivo. J Virol 81:2656–2662.

5. Carlier D, Jean-Jean O, Fouillot N, Will H, Rossignol JM (1995) Importance of the Cterminus of the hepatitis B virus precore protein in secretion of HBe antigen. J GenVirol 76:1041–1045.

6. Nassal M (1992) The arginine-rich domain of the hepatitis B virus core protein isrequired for pregenome encapsidation and productive viral positive-strand DNAsynthesis but not for virus assembly. J Virol 66:4107–4116.

7. Le Pogam S, Chua PK, Newman M, Shih C (2005) Exposure of RNA templates andencapsidation of spliced viral RNA are influenced by the arginine-rich domain ofhuman hepatitis B virus core antigen (HBcAg 165-173). J Virol 79:1871–1887.

8. Salfeld J, Pfaff E, Noah M, Schaller H (1989) Antigenic determinants and functionaldomains in core antigen and e antigen from hepatitis B virus. J Virol 63:798–808.

9. Jung MC, et al. (1991) Hepatitis B virus antigen-specific T-cell activation in patientswith acute and chronic hepatitis B. J Hepatol 13:310–317.

10. Lau GK, et al. (2002) Resolution of chronic hepatitis B and anti-HBs seroconversion inhumansby adoptive transfer of immunity to hepatitis B core antigen.Gastroenterology122:614–624.

11. Wieland S, Thimme R, Purcell RH, Chisari FV (2004) Genomic analysis of the hostresponse to hepatitis B virus infection. Proc Natl Acad Sci USA 101:6669–6674.

12. Szkaradkiewicz A, et al. (2003) HBcAg-specific cytokine production by CD4 Tlymphocytes of children with acute and chronic hepatitis B. Virus Res 97:127–133.

13. Gallina A, et al. (1989) A recombinant hepatitis B core antigen polypeptide with theprotamine-like domain deleted self-assembles into capsid particles but fails to bindnucleic acids. J Virol 63:4645–4652.

14. Lee BO, et al. (2009) Interaction of the hepatitis B core antigen and the innate

immune system. J Immunol 182:6670–6681.15. Vanlandschoot P, Van Houtte F, Serruys B, Leroux-Roels G (2005) The arginine-rich

carboxy-terminal domain of the hepatitis B virus core protein mediates attachment of

nucleocapsids to cell-surface-expressed heparan sulfate. J Gen Virol 86:75–84.16. Cooper A, Tal G, Lider O, Shaul Y (2005) Cytokine induction by the hepatitis B virus

capsid in macrophages is facilitated by membrane heparan sulfate and involves TLR2.

J Immunol 175:3165–3176.17. Cooper A, Shaul Y (2006) Clathrin-mediated endocytosis and lysosomal cleavage of

hepatitis B virus capsid-like core particles. J Biol Chem 281:16563–16569.18. Reifenberg K, et al. (1999) The hepatitis B virus X protein transactivates viral core

gene expression in vivo. J Virol 73:10399–10405.19. Hsu HC, et al. (1987) Biologic and prognostic significance of hepatocyte hepatitis B

core antigen expressions in the natural course of chronic hepatitis B virus infection. J

Hepatol 5:45–50.20. Milich DR, et al. (1997) Role of B cells in antigen presentation of the hepatitis B core.

Proc Natl Acad Sci USA 94:14648–14653.21. Chen MT, et al. (2004) A function of the hepatitis B virus precore protein is to regulate

the immune response to the core antigen. Proc Natl Acad Sci USA 101:14913–14918.22. Milich DR, et al. (1990) Is a function of the secreted hepatitis B e antigen to induce

immunologic tolerance in utero? Proc Natl Acad Sci USA 87:6599–6603.23. Stevens CE, Neurath RA, Beasley RP, Szmuness W (1979) HBeAg and anti-HBe

detection by radioimmunoassay: Correlation with vertical transmission of hepatitis B

virus in Taiwan. J Med Virol 3:237–241.24. Chen M, et al. (2005) Immune tolerance split between hepatitis B virus precore and

core proteins. J Virol 79:3016–3027.25. Yeh SH, et al. (2004) Quantification and genotyping of hepatitis B virus in a single

reaction by real-time PCR and melting curve analysis. J Hepatol 41:659–666.26. Wu HL, et al. (2005) RNA interference-mediated control of hepatitis B virus and

emergence of resistant mutant. Gastroenterology 128:708–716.

Lin et al. PNAS | May 18, 2010 | vol. 107 | no. 20 | 9345

MICRO

BIOLO

GY