Functional Characterization of Naturally Occurring Variants of Human Hepatitis B Virus Containing the Core Internal Deletion Mutation

Abstract

Naturally occurring variants of human hepatitis B virus (HBV) containing the core internal deletion (CID) mutation have been found frequently in HBV carriers worldwide. Despite numerous sequence analysis reports of CID variants in patients, in the past decade, CID variants have not been characterized functionally, and thus their biological significance to HBV infection remains unclear. We report here two different CID variants identified from two patients that are replication defective, most likely due to the absence of detectable core protein. In addition, we were unable to detect the presence of the precore protein and e antigen from CID variants. However, the production of polymerase appeared to be normal. The replication defect of the CID variants can be rescued in trans by complementation with wild-type core protein. The rescued CID variant particles, which utilize the wild-type core protein, presumably are enveloped properly since they can be secreted into the medium and band at a position similar to that of mature wild-type Dane particles, as determined by gradient centrifugation analysis. Our results also provide an explanation for the association of CID variants with helper or wild-type HBV in nature. The significance of CID variants in HBV infection and pathogenesis is discussed.

Hepatitis B virus (HBV) is one of the most common infectious agents in humans. Chronic active hepatitis B often leads to the development of cirrhosis and liver cancer (56, 62). The molecular and cellular mechanisms of pathogenesis and chronicity of HBV infection remain to be elucidated. Hepatitis B core antigen (HBcAg or nucleocapsid protein) has been shown to be a major target of T-cell immunity (19, 40, 42). Recently, naturally occurring variants containing a heterogeneous population of core antigen internal deletions (CID) were found to be geographically ubiquitous and highly prevalent in chronic HBV carriers (Table 1), including 100% of asymptomatic carriers (45), 14 to 100% of patients with chronic hepatitis (1, 2, 20, 38, 60, 64, 65), and patients with hepatocellular carcinoma (HCC) (27). However, only 7% of chronic hepatitis patients in Hong Kong were found to contain CID variants (3). CID variants have also been found in immunosuppressed patients with renal transplantation (22, 23). Interestingly, CID variants have never been found in acute hepatitis patients (4, 18).

TABLE 1.

Ubiquitous occurrence of CID variants in HBV carriers

Prevalence (%) of CID variants	Clinical statusa	Geographic origin	Reference
4/4 (100)	ASC	Japan	45
7/11 (63)	CH	Japan	65
2/? (NA)b	CAH	United Kingdom	1
3/4 (75)	CAH	United Kingdom	2
1/2 (50)	CAH	Japan	60
2/? (NA)	CH	Italy	20
1/7 (14)	CAH	Japan	64
6/6 (100)	RT	Germany	22
17/263 (7)	CH	Hong Kong	3
2/18 (11)	HCC	Taiwan	27
17/67 (25)	CH	United Kingdom	38
7/? (NA)	RT	Germany	23

^aThe clinical status of HBV carriers is according to the original reports. ACS, asymptomatic carrier; CH, chronic hepatitis; CAH, chronic active hepatitis; RT, renal transplantation. 

^bNA, information not available. 

To date, CID variants have not been characterized functionally. Most of the CID variants share the following features (references are in Table 1). (i) Deletions often occur within core amino acids (aa) 80 to 120. The deletions usually do not extend into the partially overlapping polymerase (pol) open reading frame. (ii) The exact end points and sizes of deletions vary from variant to variant. (iii) Deletions appear to be more often in frame than out of frame. (iv) CID mutants are almost always found in the presence of HBV with an apparent full-length core gene. (v) Although the biological significance of CID mutations remains unclear, the internal deletions coincide with a potent T-cell epitope (32, 39, 40, 63), suggesting an immune escape function for this mutation.

The 28-nm HBV spherical nucleocapsid particle contains about 180 subunits of the 21-kDa core protein, which is organized into an icosahedral shell (10, 16). Assembly of the nucleocapsid involves the core protein, pol, and a 3.5-kb pregenomic RNA. The sequence elements necessary and sufficient for HBV pregenomic RNA encapsidation have been defined (15, 33) and characterized (37, 44, 50, 68). pol is required for pregenomic RNA encapsidation (5, 6, 13, 25, 51, 52). Furthermore, pol appears to act preferentially in cis to direct its own mRNA into the nucleocapsid (5, 25, 47) and initiate DNA synthesis via reverse transcription (31, 58, 66). In addition to viral factors, host factors have been hypothesized to be involved in hepadnaviral replication (28, 29, 49). The core protein has a nucleic acid binding domain rich in arginine residues near the C terminus (9, 21, 24, 43). Serial deletions from the C terminus of the core protein have shown that the arginine-rich domain is not required for multimerization of the core protein, although it might increase the stability of the nucleocapsid (21, 43). The deletion found in CID variants is located outside the nucleic acid binding domain. It is unclear, though, if the deletion affects multimerization of the core proteins.

In this study, we have characterized the functional properties of HBV CID variants at the molecular level.

MATERIALS AND METHODS

Plasmid constructs. (i) pDEL85 and pDEL109.

To construct pDEL85 and pDEL109, DNA fragments from nucleotides (nt) 1636 to 2688 containing the HBV deleted core gene were PCR amplified from total DNA of hepatoma samples T85 and T109 (27) and used to replace both copies of the wild-type counterpart in the HBV tandem dimer plasmid pWT (55). The two oligonucleotide primers used in PCR amplification for T85 and T109 are as follows: one primer is a 30-mer (5′-A AGG GCA AAT ATT TGG TAA GGT TAG GAT AG-3′) containing HBV minus-strand DNA sequences from nt 2659 to 2688 with an SspI cleavage site (underlined). The other primer is a 27-mer (5′-AGA AAT ATT GCC CAA GGT CTT ACA TAA-3′) containing HBV plus-strand DNA sequences from nt 1636 to 1659 with an SspI cleavage site (underlined). The locations of these two primers are separated by an HBV integration hot spot near nt 1820 (54). Upon integration into the host chromosomes, the HBV template for PCR will orient these two primers away from each other, thus resulting in no PCR product. Therefore, only the replicative, but not the integrated, forms of HBV DNA will be amplified. One microgram of tumor DNA and 100 ng of each primer were used in a 10-μl PCR consisting of a 94°C denaturing step (20 s) followed by 40-cycles of amplification at 94°C (1 s), 47°C (1 s), and 72°C (40 s). The amplified target sequence (0.9 kb) was subcloned into the pGEM-T vector (Promega Co.) and screened by DNA sequencing. The DNA fragments containing CID mutations were gel purified by digestion with SspI and used to replace the normal counterpart of the wild-type HBV genome carried on a pUC12-HBV plasmid. The 3.1-kb EcoRI fragment containing the CID mutation was ligated with the EcoRI-cleaved pUC12-HBV CID monomer. The resulting tandem dimer plasmids, pDEL85 and pDEL109, were then confirmed by restriction enzyme digestion and DNA sequencing.

(ii) pPOLCAT, p85POLCAT, and p109POLCAT.

To measure HBV pol expression, plasmid pPOLCAT was constructed. A 1.6-kb SmaI-digested DNA fragment containing the full-length chloramphenicol acetyltransferase (CAT) gene and a simian virus 40 (SV40) polyadenylation signal at its 3′ end was fused in frame and downstream of the wild-type HBV pol gene at the BstEII site (nt 2663) by blunt-end ligation. The resulting chimeric construct of HBV pol and CAT was subcloned into the BamHI site of pGEM4Z by using two BamHI sites at HBV nt 409 and 2907 (18a). The expression of the reporter CAT gene of pPOLCAT is under the transcriptional and translational control of native HBV sequences. To construct plasmid p85POLCAT, the wild-type HBV sequence from nt 682 to nt 2332 of pPOLCAT was replaced with the HBV counterpart from pDEL85 by using SpeI and BspMII sites. Plasmid p109POLCAT was constructed similarly, except that the wild-type HBV sequence from nt 905 to nt 2364 was replaced with the HBV counterpart of pDEL109 by using two PpuMI sites.

(iii) pSVC, pSVCflu, pSV85flu, and pSV109flu.

To construct pSVC, the PCR-amplified core gene fragments from nt 1877 to nt 2463 were digested with restriction enzymes HindIII and SacI and subcloned into the HindIII and SacI sites of plasmid pGCE, which contains the SV40 enhancer and early promoter (48). Two oligonucleotide primers were used for the PCR. One 30-mer (5′ AGA AAG CTT AGC TGT GCC TTG GGT GGC TTT 3′) contains HBV plus-strand DNA sequences from nt 1877 to nt 1897 with a HindIII cleavage site (underlined). The other primer is also a 30-mer (5′ AGA GAG CTC ATA CTA ACA TTG AGA TTC CCG 3′) containing a SacI cleavage site (underlined). One nanogram of pSV2ANeo-HBV monomer and 100 ng of each primer were used in a 10-μl PCR consisting of a 94°C denaturing cycle (20 s) followed by 40 cycles of amplification at 94°C (1 s), 53°C (1 s), and 72°C (40 s). The cloning strategy used for plasmids pSVCflu, pSV85flu, and pSV109flu is similar to that used for pSVC, except that the downstream primer was replaced with an oligonucleotide containing an influenza virus epitope (flu) sequence (underlined) with the sequence 5′-AGA GAG CTC CTA AGC ATA ATC TGG AAC ATC ATA TGG ATA ACA TTG AGA TTC CCG AGA TTG AGA-3′, and the annealing temperature for the PCR was changed to 61°C. pSV2ANeoHBV, pDEL85, and pDEL109 plasmid DNAs were used as templates for PCR amplification to construct pSVCflu, pSV85flu, and pSV109flu, respectively.

(iv) pSPC, pSP85, pSP109, and pSP109ATA.

To construct pSPC, the core gene DNA fragment from pSVC was subcloned into the HindIII and SacI sites of pSP64, which contains an SP6 RNA pol promoter. Plasmids pSP85 and pSP109 were constructed in the same way as pSPC, except that plasmids pDEL85 and pDEL109 were used as templates for PCR amplification, respectively. To construct pSP109ATA, the upstream primer (5′-AGA AAG CTT GCT TTG GGG CAT AGA CAT TGA C-3′) was used to eliminate the ATG translational start codon of the core gene. All plasmid constructs were confirmed by restriction enzyme digestion and DNA sequencing.

Preparation of core particles.

Preparation of intracellular core particles was done as described elsewhere (68). Extracellular core particles were collected 5 days posttransfection from a 10-cm-diameter dish of 48-h conditioned medium (10 ml). The medium was centrifuged at 3,000 rpm for 30 min at 4°C in an IEC Centra GP8 centrifuge. Particles from the clarified medium were pelleted through a 16-ml cushion of 20% sucrose by spinning at 26,000 rpm for 16 h at 4°C in a Beckman SW28 rotor. HBV DNA or RNA was purified from core particles as described previously (68).

Antibodies and immunoblot analysis.

Core-specific antisera obtained from R. Lanford (8) and purchased from Dako and Chemicon were all of rabbit origin and polyclonal. Antiserum was applied at a 1/3,000 to 1/500 dilution from the stock, adjusted according to the sensitivity of each antiserum. Influenza virus hemagglutinin-specific antiserum against the peptide sequence (5′-YPYDVPDYA-3′) was obtained from J. Giam. This antiserum was of mouse origin and monoclonal and was applied in a 1/1,000 dilution from the stock.

In vitro transcription and translation.

For in vitro transcription, 1 μg of plasmid DNA was linearized with HindIII and extracted first with Tris-EDTA-saturated phenol and then chloroform, followed by ethanol precipitation. One microgram of linearized DNA was mixed with 2 μl of 10× buffer, 4 μl of a nucleoside triphosphate mixture (10 mM), 1 μl of RNase inhibitor (40 U/μl), and 2 μl of SP6 pol (5 U/μl; Promega). The 20-μl reaction mixture was incubated for 1 h at 37°C. To remove the plasmid DNA after the reaction, 1 μl of RNase-free DNase I (2 U/μl) was added and the mixture was incubated at 37°C for 15 min. A Promega TNT coupled rabbit reticulocyte lysate system was used for in vitro protein synthesis. One microgram of plasmid DNA was used in a reaction mixture containing 25 μl of rabbit reticulocyte lysate, 2 μl of TNT reaction buffer, 1 μl of SP6 pol, 1 μl of an amino acid mixture minus methionine (1 mM), 40 U of RNase inhibitor, and 4 μl of [³⁵S]methionine (1,000 Ci/ml; Amersham Co.). The volume was increased to 50 μl with nuclease-free water, and then the mixture was incubated at 30°C for 90 min. A 10-μl aliquot of the reaction mixture was added to 20 μl of sodium dodecyl sulfate (SDS) sample buffer (2% SDS, 10% glycerol, 100 mM dithiothreitol, 60 mM Tris [pH 6.8], 0.0001% bromophenol blue) and analyzed on an SDS–12.5% polyacrylamide gel.

Gradient sedimentation analysis of virus particles.

Cell culture medium was collected 5 and 7 days posttransfection and centrifuged at 3,000 rpm and 4°C for 30 min in an IEC Centra GP8 centrifuge. The clarified medium was then layered on a 16-ml 20% sucrose cushion in 150 mM NaCl–20 mM Tris-HCl (pH 7.4) (TNE) and centrifuged at 25,000 rpm for 16 h at 4°C in an SW28 rotor (Beckman). The supernatant was discarded, and the pellet was resuspended in TNE buffer. Isopycnic centrifugation of the particles was performed in a gradient of 20 to 50% (wt/vol) cesium chloride in TNE buffer in an SW50.1 rotor (Beckman) at 35,000 rpm for 16 h at 4°C. Approximately 200-μl fractions were collected from the top of the tubes.

For the preparation of viral DNA, cesium chloride was removed from each fraction by dialysis against TNE buffer overnight (1 to 1,000 [vol/vol]). Viral DNAs in each pooled fraction were prepared by a standard procedure with proteinase K digestion, phenol-chloroform extraction, and ethanol precipitation (27). PCR analysis of the viral DNAs in each fraction was then conducted as described in the legend to Fig. 1B. In addition, Southern blot analysis using HBV DNAs extracted from every other fraction was conducted by using full-length vector-free HBV DNA as the probe.

FIG. 1.

Identification of replicating CID variants in tumor and nontumor liver tissues. (A) Demonstration of the presence of replicative forms of HBV DNA in liver samples by Southern blot analysis. Vector-free HBV DNA was used as the probe. The data from samples N109 and T109 are shown here as an example. A lambda HindIII size marker is shown in lane 3. (B and C) Core gene of HBV, from liver samples containing replicative HBV DNAs, amplified by PCR by using core gene-specific primers (27). After gel electrophoresis, the PCR-amplified DNA fragments were identified by Southern blotting with the HBV-specific oligonucleotide probe outside (HBC-7) (B) or inside (HBC-9) (C) the deletion hot spot of the core gene (65). HBC-7 (5′-CTG TGG AGT TAC TCT CTT TTT TGC-3′) contains the consensus HBV sequence in the core gene from nt 1937 to nt 1960. HBC-9 (5′-ATG TCA ATG TTA ATA TGG GCC TAA AAA TCA GA-3′) contains the HBV consensus sequence in the core gene from nt 2165 to nt 2196. The sequence of HBC-9 resides within the core internal deletion, and it does not hybridize with HBV sequences containing the core internal deletion. RC, relaxed-circle DNA; SS, single-stranded DNA.

HBeAg and HBsAg assay.

Abbott HBe (recombinant DNA) EIA and Auszyme Monoclonal kits were used for the immunoassay of hepatitis B e antigen (HBeAg) and hepatitis B surface antigen (HBsAg) respectively, as recommended by the manufacturer (Abbott Laboratory).

RESULTS

Replicating CID variants in liver samples from HCC patients.

HBV core gene sequences were amplified by PCR from tumor and nontumor liver samples (27). The majority of HCC samples contain only a few copies of integrated HBV DNA and no detectable HBV replication, as shown by Southern blot analysis. However, replicative forms of HBV DNA can be detected in 10 to 25% of HCC samples by Southern blot analysis (7 of 27 in reference 36) (27). In our previous studies, we have identified 14 samples containing HBV DNA replication from approximately 100 liver samples of HCC patients (27, 59). For example, replicative intermediates of HBV DNA can be detected in samples T109 and N109 (Fig. 1A). PCR-amplified products of HBV core gene sequences from some of these HBV replication-positive liver samples were analyzed by agarose gel electrophoresis and ethidium bromide staining (27). A 586-nt DNA fragment of the full-length core gene was present in all samples from all patients, except for sample T61. This is consistent with the fact that sample T61, which was included here as a negative control, contained only the integrated form and no replicative intermediates of HBV DNA, as shown by Southern blot analysis (see Discussion) (27). In addition to this full-size DNA fragment, several lower-molecular-weight DNA fragments were observed in samples T61, T85, T109, and N109 (Fig. 1B and C). To verify the identities of these lower-molecular-weight DNA fragments, we performed Southern blot analysis with oligonucleotide probes derived from outside (probe HBC-7) or within (probe HBC-9) the deleted core gene (45, 65). As predicted, the 586-nt DNA fragment hybridized with both probes (Fig. 1B and C), while the lower-molecular-weight fragments hybridized only with the HBC-7 probe (Fig. 1B) but not with the HBC-9 probe (Fig. 1C). The results reported here demonstrated that CID variants are present in approximately 25% of the liver samples from those HCC patients whose end stage livers still contain ongoing HBV DNA replication.

The CID mutation of liver-derived DEL85 is identical to that of a serum-derived CID variant frequently found at different geographical locations.

To isolate and functionally characterize CID variants, a 1-kb DNA fragment containing the core gene was amplified by PCR from total DNAs of samples T85 and T109. The PCR primers were designed for preferential amplification of replicative, unintegrated HBV DNA (see Materials and Methods). This 1-kb DNA fragment was used to replace the wild-type counterpart in an HBV tandem dimer. The resulting plasmids were named pDEL85 (or DEL85) and pDEL109 (or DEL109). DNA sequence analysis indicated that DEL85 and DEL109 contain 48- and 41-aa in-frame deletions in the core gene, respectively (Fig. 2). The deduced amino acid sequence of the core antigen from the CID variants of sample T109 has been published previously (27). In the case of DEL85, the deletion appears to end exactly before the start of the pol (P) gene. Surprisingly, both deletion end points of the DEL85 variant, which is derived from the liver of a Taiwanese patient, are identical to those of a CID variant in the sera of a British patient (FT2 clone in reference 38), a Chinese patient in Hong Kong (patient 5b clone in reference 3), and a Korean patient in Philadelphia, Pa. (69). The global occurrence of this same CID mutation as DEL85, found in both serum and liver samples, shows that DEL85 is a reasonably representative example of replicating CID variants for our further functional characterizations.

FIG. 2.

Sequence analysis of the 1-kb DNA fragments used for the construction of the HBV CID mutant dimers. The diagram illustrates the deletion regions of HBcAg of two different CID variants identified from two different hepatoma patients, T85 and T109. This deletion region does not overlap with any other HBV genes, including X, P (pol), and pre-S/S (envelope). DEL85 is missing aa 88 to 135, while DEL109 is missing aa 82 to 122.

Replication defects of HBV CID mutants.

To determine whether the CID variants are capable of replication, either pDEL85 or pDEL109 was transfected into a human hepatoma cell line, Huh7. Core particle-associated HBV DNA was then assayed for viral replication by Southern blot analysis. Both mutants were found to be replication defective, and no replication intermediate DNA could be detected in the cells (Fig. 3A). The deficiency in DNA synthesis is correlated with a defect in RNA encapsidation, since no encapsidated pregenomic RNA from the CID mutants was detected by primer extension analysis (Fig. 3B). Productive pregenomic RNA encapsidation requires at least three different viral components: nucleocapsid or core protein, pol, and the 3.5-kb pregenomic RNA. To identify the cause of the encapsidation defect in CID mutants, we first examined the steady-state level of total HBV RNA by Northern blot analysis (Fig. 3C). No significant difference in RNA level between the wild type and the CID mutants was detected.

FIG. 3.

HBV CID variants are replication defective upon transfection into human hepatoma cell line Huh7. (A) Five days after transfection with wild-type (pWT) (55) or mutant HBV, viral DNAs from intracellular core particles were harvested and subjected to Southern blot analysis with the 3.1-kb full-length vector-free HBV DNA probe. (B) Encapsidation activity was assayed by primer extension using core particle-associated viral RNA from transfected cultures and a 5′-end-labeled oligonucleotide primer (nt 1980 to nt 2001), which was described in a previous report (52). (C) Twenty-five micrograms of cellular RNA from transfected cells was subjected to Northern blot analysis and probed with a 3.1-kb full-length HBV probe (top). Similar amounts of cellular RNA were used in all of the lanes, as indicated by the similar intensities of ethidium bromide staining (bottom). (D) The CAT reporter gene was fused in frame with pol sequences originating from pWT, pDEL85, and pDEL109. CAT activities of the pol-CAT fusion proteins were measured 2 days after transfection as detailed elsewhere (48). (E) The core proteins produced from pWT, pDEL85, and pDEL109 were analyzed by immunoblot assay with a rabbit anti-core antibody.

Although the deletions in both CID mutants do not extend into the pol open reading frame, these deletions might affect its expression due to their proximity to the pol translation initiation codon. To measure any potential change in the normally low level of pol expression, we engineered an in-frame fusion construct between pol and CAT as a reporter. The CAT activity of the pol-CAT fusion protein is expected to correlate with the quantity of the pol-CAT fusion protein and, hence, pol translation. Since the 5′ upstream RNA sequences and the immediate sequence context around the translational initiation codon of the pol-CAT fusion protein are virtually identical to those of the parental CID constructs DEL85 and DEL109 (Fig. 2 and 3D), this pol-CAT reporter system should faithfully reflect the strength of translation of pol in the CID variants. As shown in Fig. 3D, no appreciable difference in CAT activity was detected between the variants and the wild type (Fig. 3D). Finally, we examined the production of nucleocapsid protein from the CID variants after transfection into Huh7 cells. As shown in Fig. 3E, no core protein from the CID variants was detected by immunoblot analysis. This defect correlates with the absence of RNA encapsidation and DNA synthesis (Fig. 3B).

CID mutants are rescuable by wild-type core protein in trans.

The fact that these replication-defective CID mutants can be detected by PCR in patients suggests that they might be able to survive in the presence of other HBV species. We tested this possibility by cotransfecting CID mutants with a wild-type HBcAg expression vector (see Materials and Methods). Both CID mutants were rescued, resulting in a replication level similar to that of the wild type (Fig. 4A). These rescued CID mutants were also secreted into the medium (Fig. 4B). The sedimentation profile of these rescued and secreted CID viral particles on gradient centrifugation by Southern blot analysis was almost indistinguishable from that of the wild type in the Dane particle fractions (Fig. 4C and D). The presence of the core gene deletion in the rescued and secreted Dane particle fractions was confirmed by PCR amplification (data not shown).

FIG. 4.

The replication-defective CID variants can be rescued by trans complementation with wild-type HBcAg and can be secreted into the medium with a buoyant density similar to that of wild-type HBV. (A) Increasing doses of a wild-type HBcAg expression vector (pSVC) were cotransfected with constant amounts of pWT, pDEL85, and pDEL109. Viral DNAs from intracellular core particles were analyzed by Southern blotting as described in the legend to Fig. 3. (B) Ten micrograms of pDEL85 or pDEL109 was transfected either alone or with 10 μg of pSVC. Extracellular HBV particles from 20 ml of conditioned medium were collected 5 days after transfection via centrifugation through a 20% sucrose cushion. HBV replication activity was assayed as described in the legend to Fig. 3. (C) Conditioned medium was collected from cells transfected with 10 μg of pWT. Viral particles in the medium were purified through a 20% sucrose cushion and subjected to isopycnic centrifugation in a gradient of 20 to 50% (wt/vol) cesium chloride. Fractions were collected for the assay of HBsAg with the Abbott Auszyme EIA kit (top). To locate the fractions containing HBV genomes, Southern blot analysis was performed as described in Materials and Methods (bottom). (D) Conditioned medium from cells transfected with 10 μg of DEL85 and pSVC was assayed for HBsAg and HBV DNA as described in the legend to Fig. 4B and C. RC, relaxed-circle DNA; SS, single-stranded DNA.

Undetectable HBcAg and HBeAg in CID variants.

Despite the use of several different anti-HBV core antibodies, we detected no core protein from the CID variants (Fig. 3E and data not shown). This negative result could be due to a number of possibilities, such as loss of a dominant antibody epitope (53) and/or instability of the mutant core protein product. To differentiate between these possibilities, we genetically tagged both the deleted and wild-type core proteins with an influenza virus epitope (flu) (see Materials and Methods). As shown in Fig. 5, wild-type core-flu fusion protein can be detected by either anti-core (Fig. 5A) or anti-flu (Fig. 5B) antibodies, while the wild-type core protein can only be detected by anti-core antibody. In contrast, CID core-flu fusion proteins cannot be detected by either antibody, despite the stable expression of CID mutant core-flu mRNAs from plasmids pSV85flu and pSV109flu (data not shown). These results indicate that the loss of a dominant anti-core antibody epitope alone is not a likely explanation for the undetectable CID-specific core protein in vivo. However, by using an in vitro transcription-and-translation system, we demonstrated that a CID mutant core protein with a reduced molecular weight can be produced in vitro, albeit at a considerably lower intensity relative to the wild-type core protein. Furthermore, when the ATG initiation codon of the CID core protein was ablated by changing into ATA, no translated protein was observed (Fig. 5C). Taken together, these results are consistent with the interpretation that the CID core protein might be unstable in vivo.

FIG. 5.

The internally deleted core proteins produced by the CID variants can be detected in vitro but not in vivo. The flu epitope peptide sequence (YPYDVPDYA) from the influenza virus hemagglutinin was introduced into the carboxyl termini of the wild-type and CID core proteins in an SV40 expression vector (see Materials and Methods). The wild-type core protein from pWT and the wild-type and deleted core flu fusion proteins from pSVCflu, pSV85flu, and pSV109flu were measured by immunoblot analysis with anti-core (A) or anti-hemagglutinin (B) antibody. (C) The wild-type and deleted core proteins were expressed in vitro from pSPC, pSP85, and pSP109 (see Materials and Methods) by using a rabbit reticulocyte lysate system (Promega Co.). The in vitro-synthesized proteins were analyzed on an SDS–12% polyacrylamide gel (top). pSP109ATA is a derivative of pSP109, except that the ATG initiation codon of the core gene has been changed to ATA. To control for equal amounts of RNAs used in the in vitro translation experiment, the in vitro-synthesized RNA transcripts from these plasmids were quantitated by gel electrophoresis (bottom). The beta-actin transcript, with a size of 360 nt, from pRT1 was used as a positive control and a size marker.

Because HBcAg and HBeAg share the same open reading frame, the CID mutation should affect both HBcAg and HBeAg. Indeed, we have not been able to detect any HBeAg production from CID variants by using an enzyme immunoassay (Abbott Laboratories) (Table 2). Thus, CID mutants DEL85 and DEL109 exhibit an HBeAg-negative phenotype, which may be associated with altered HBV pathogenesis (12, 41). On the other hand, the detection of similar amounts of secreted HBsAg from both CID variants and the wild type suggests that the core internal deletion has no effect on HBsAg expression (Table 2).

TABLE 2.

HBeAg and HBsAg assay of CID mutants^a

Plasmid	A492
	HBeAg	HBsAg
pWT	0.893	1.369
pDEL85	0.034	1.243
pDEL109	0.037	1.487

^aThe medium from a transfected culture was collected 2 days after transfection and assayed for HBeAg and HBsAg (see Materials and Methods). One representative result from two independent experiments is shown. The cutoffs of the enzyme immunoassays are 0.099 for HBeAg and 0.056 for HBsAg. 

DISCUSSION

We have characterized two different CID variants (DEL85 and DEL109) isolated from two different patients by using human hepatoma cell line Huh7. Despite differences in the sizes and end points of their deletions, as discussed below, both CID variants appeared to have some functional features in common, including a replication defect resulting from the absence of a stable core protein. The fact that the wild-type core antigen alone can rescue DEL85 and DEL109 in trans to wild-type replication levels suggests that the replication-defective phenotype is mainly due to the internal deletion of HBcAg, rather than the nucleotide substitutions in the core promoter region (data not shown; 11, 46).

Infectivity of CID variants?

As shown in Fig. 1A and B, tumor sample T109 and adjacent nontumor liver sample N109 appear to have been infected by the same CID variant, suggesting at least a certain degree of infectivity of DEL109 in vivo. However, the remote possibility of cross contamination between tumor and nontumor tissues during surgical removal cannot be excluded. Conversely, the absence of detectable CID variants in sample N85 does not necessarily mean that CID variants are not infectious in general (Fig. 1B). For example, an additional mutation of the pre-S envelope gene could impair the secretion of mature HBV particles (67).

It remains unclear if the CID variant is infectious or not. We believe that CID variants are most likely infectious because of the following reasons. First, CID variants can mature and be secreted into the medium (Fig. 4B). Second, the exported CID particles generated by providing the core gene in trans exhibit a buoyant density indistinguishable from that of Dane particles of wild-type origin (Fig. 4C and D). Third, since the core proteins which make up the nucleocapsid of CID variants are of wild-type origin, it is a logical prediction that the envelope formation of CID variants will be normal and consequently so will its associated infectivity. Perhaps the strongest argument that CID variants are infectious is that they are always found to be accompanied by wild-type HBV and can be rescued in trans with wild-type core protein. On the other hand, until coinfection or superinfection of CID variants with wild-type helper virus can be demonstrated, there is no direct experimental evidence that these CID variants are indeed infectious.

Mechanism of core internal deletion.

A certain degree of heterogeneity of the internal deletions found in T85, T109, and N109 was observed (Fig. 1B). It is unclear how these various deletions were generated. The CID mutation does not appear to result from the reverse transcription of spliced HBV-specific RNA (61). First, none of the deletional end points have the consensus sequences required for RNA splice donor and acceptor sites (i.e., GT and AG dinucleotides at the splice junction of introns). Second, the deletion end points are variable in position (2, 27, 45, 65). Third, the deletion end points do not coincide with any of the reported splice junctions of HBV (14, 57). We noted the existence of a 3-nt (ATG) junctional homology at both ends of the internal deletion in samples T85 and KP (69). This result is consistent with an HBV illegitimate DNA recombination mechanism (54). However, the possibility of a mechanism of replicative error by pol via template switch and copy choice cannot be excluded (34, 35).

Functional pol and CAT activities of pol-CAT fusion protein.

As described in Fig. 3D, the CAT activities of pol-CAT fusion proteins appear to be similar in the wild type and the CID variants. This result is consistent with the fact that hepadnaviral pol is known to be required in cis for RNA encapsidation and DNA replication (5, 25). It should be noted that the AUG codon for the translational initiation of wild-type HBV pol is in a poor Kozak context (the −3 position is a C). In the CID deletion, an improved Kozak context for the AUG initiator of DEL85 pol (the −3 position becomes a G) is seen, and this could probably lead to increased CAT activity. It is possible that our CAT assay in Fig. 3D is not sensitive enough to detect a small increase in DEL85 pol expression. However, increased production of pol, if any, may not necessarily constitute a replication advantage for CID variants. It has been estimated that each hepadnaviral particle contains approximately one pol molecule and one pregenomic RNA molecule (7). Therefore, even if more pol molecules can be translated from each 3.5-kb pregenomic mRNA template as a consequence of the CID mutation, unless the rate-limiting amount of the 3.5-kb pregenomic mRNA is also increased, neither RNA encapsidation nor DNA replication is likely to increase significantly.

Immune escape of CID variants?

It has been hypothesized that immune escape mutations of HBV core antigen might contribute to the chronicity of HBV infection (27). The internal deletion of CID variants around aa 80 to 120 coincides with a potent major histocompatibility complex class I- and class II-restricted T-cell epitope (32, 39, 40, 63). This observation certainly raises the possibility that CID variants are immune escape mutants. The fact that the deleted core protein appears to be highly unstable in vivo adds to the argument that the CID protein might be degraded soon after synthesis, before ever being presented to the immune system as a target antigen. On the other hand, CID variants depend on the wild-type core protein for replication (Fig. 4). Therefore, hepatocytes coinfected with a CID variant and the helper virus must also process and present the wild-type core antigen peptides. Further investigation is required to determine the potential role of CID mutation in immune escape, if any.

Parasitic or symbiotic relationship between the wild type and the CID variants?

Our results in Fig. 3 show that CID variants are replication defective at the RNA packaging level, probably because the deleted core protein is not stable in vivo (Fig. 5) and cannot form a functional nucleocapsid. However, the defective genome of CID variants is packaged and replicates with a supply of a wild-type core protein (Fig. 4), suggesting that the CID variants probably can replicate in the presence of wild-type virus (69). Indeed, in the literature (Table 1), CID variants always occur in association with an HBV containing a nondeleted HBcAg.

It should be noted that in sample T61 (Fig. 1B), the deleted DNA fragment is not found together with nondeleted wild-type DNA. This result seems to be at odds with the conclusion that CID variants are replication defective and require a helper virus for propagation. The reason for the unusual pattern of T61 is that the deleted DNA fragment observed in Fig. 1B was amplified from a single integrated copy of the CID genome on the host chromosome. Unlike other samples (e.g., N109 and T109) containing only or predominantly replicative unintegrated HBV DNA (Fig. 1A), no HBV DNA replication was detected in sample T61 by Southern blot analysis (data not shown).

Defective interfering (DI) viruses have been widely found in bacterial, plant, and animal viruses in the laboratory setting (17, 26, 30). DI viruses often contain a less-than-full-length genome and are replication defective. One important feature of DI virus is that it can replicate in the presence and at the expense of helper viruses (the so-called interference-and-enrichment phenomenon). It has been hypothesized that the attenuating effect of DI variants on the helper virus titer could provide the virus an advantage in establishing or maintaining a persistent viral infection (17, 26, 30). Several characteristics of the CID variants, including the deletion in the core gene, the functional defect in replication, and the rescuability by a helper virus, are reminiscent of DI particles. However, the interference-and-enrichment phenomenon remains to be demonstrated (69).