|Home | About | Journals | Submit | Contact Us | Français|
Hepatitis B virus (HBV) contains three coterminal envelope proteins on the virion surface: large (L), middle (M), and small (S). The M and S proteins are also secreted as empty “subviral particles,” which exceed virions by at least 1,000-fold. The S protein serves as the morphogenic factor for both types of particles, while the L protein is required only for virion formation. We found that cotransfecting replication constructs with a small dose of the expression construct for the missing L, M, and S proteins reconstituted efficient virion secretion but only 5 to 10% of subviral particles. The L protein inhibited secretion of subviral particles in a dose-dependent manner, whereas a too-high or too-low L/S protein ratio inhibited virion secretion. Consistent with the results of cotransfection experiments, a point mutation at the −3 position of the S gene AUG codon reduced HBsAg secretion by 60 to 70% but maintained efficient virion secretion. Surprisingly, ablating M protein expression reduced virion secretion but markedly increased the maturity of virion-associated genomes, which could be reversed by providing in trans both L and M proteins but not just M protein. M protein stability was dependent on the coexpression of S protein. Our findings suggest that efficient HBV virion secretion could be maintained despite drastic reduction in subviral particle production, which supports the recent demonstration of separate secretion pathways adopted by the two types of particles. The M protein appears to facilitate core particle envelopment, thus shortening the window of plus strand DNA elongation.
Hepatitis B virus (HBV) is a small enveloped DNA virus that infects only humans and higher primates (17). It is estimated that 2 billion people worldwide have been exposed to HBV and more than 350 million are chronically infected. Inside the 42-nm virion is a nucleocapsid or core particle that shields the relaxed circular double stranded DNA genome. The minus-strand DNA is full length (3.2-kb) and terminally redundant, whereas the plus-strand DNA is incomplete. Upon infection of hepatocytes, the genome is repaired to form the unit-length, fully double-stranded covalently closed circular DNA in the nucleus (54), which serves as the template for transcription. The four major transcripts of 3.5, 2.4, 2.1, and 0.7 kb are termed pre-C/C, pre-S1, preS2/S, and X, respectively, all sharing the same 3′ end but different 5′ initiation sites (4). The mRNAs are exported to cytoplasm for protein translation and genome replication. The 3.5-kb transcript is over the genome length and therefore contains all of the genetic information of the viral genome. The longer and less abundant version (preC RNA) directs the expression of hepatitis B e antigen (HBeAg), a secreted viral protein with immunomodulatory function (37). The shorter one or pregenomic RNA (pgRNA) serves as the messenger for core protein and viral polymerase, as well as the template for DNA synthesis. To this end, the nascent core protein assembles into core particle, packaging both the pgRNA and the DNA polymerase together with host heat shock proteins (23, 24). Inside the core particle the polymerase carries out a series of enzymatic reactions to convert the pgRNA into partially double-stranded DNA as found in virions. Along this process the polymerase serves as the protein primer for the synthesis of the minus-strand DNA, the reverse transcriptase, RNase H, and DNA-dependent DNA polymerase (49, 51). Envelopment of core particle leads to virion formation. In this regard, core particles with mature (double-stranded DNA) genomes are preferentially enveloped and secreted, although preventing plus strand DNA synthesis through mutation at the RNase H domain of DNA polymerase did not abolish virion secretion (9, 18). It has been suggested that genome maturation alters the phosphorylation status of the core protein, thus triggering envelopment (3, 58).
HBV expresses large (L), middle (M), and small (S) envelope proteins. They differ in length by their N-terminal region as a result of alternative translation initiation (see Fig. Fig.1).1). Consequently, the three envelope proteins contain preS1/preS2/S, preS2/S, and the S domain, respectively. The L protein is translated from the 2.4-kb preS1 RNA, whereas the M and S proteins are expressed from the shorter 2.1-kb RNA with a heterogeneous 5′ end. The envelope proteins are inserted into endoplasmic reticulum (ER) membrane cotranslationally. An asparagine in the S domain serves as a facultative N-linked glycosylation site, thus generating two size forms for all of the three envelope proteins. The M protein contains an extra constitutive N-linked glycosylation site at its N terminus. Although all of the three envelope proteins are incorporated into the virions of the wild-type virus, M protein is dispensable for virion secretion and M-minus mutants frequently arise during the later stage of chronic HBV infection (10, 14, 15, 47). Moreover, avian homologues of HBV, such as duck hepatitis B virus (DHBV), do not express the M protein (44, 48). Recent studies suggest that budding of HBV virions is similar to some enveloped RNA viruses through the formation of multivesicular bodies in the late endosome (31, 46, 55).
HBV is unique in that the envelope proteins can be secreted independent of nucleocapsid in the form of 20-nm “subviral particles.” They are formed in the ER or post-ER, pre-Golgi apparatus (26, 42). The subviral particles contain host-derived lipids embedded with M and S envelope proteins but little L protein (13, 39, 43). The S protein alone is sufficient for the formation and secretion of subviral particles as demonstrated by transfection experiments. The subviral particles can reach up to 10,000- to 1,000,000-fold higher concentration than virions in the blood of infected individuals (16). Due to their lack of nucleocapsid, the subviral particles are noninfectious. Why HBV secretes an overabundance of such noninfectious envelope protein particles has been a longstanding enigma. One possibility is that S protein is incorporated into virions inefficiently relative to its auto assembly, thus necessitating overexpression of S protein to maintain virion secretion. In the present study we tested this hypothesis by altering the envelope protein expression level or the L/S protein ratio and clarified the role of M protein on virion formation.
(This study was presented in part at the 2006 Molecular Biology of Hepatitis B Viruses Meeting in Vancouver, Canada, and also at the 2007 Gordon Conference on Viruses and Cells in Tilton, NH.)
The HBV clones used in the present study all belong to genotype A. Two size forms were generated: 1.5x genome lengths (1.5mer) and 0.7x genome length (0.7mer) (Fig. (Fig.1).1). A 1.5mer genome has the 4.8-kb EcoRV-ApaI fragment (nucleotides 1044 to 3221 and nucleotides 1 to 2600) cloned, after blunt ending of the ApaI site, into the SmaI and XhoI sites of pBluescript vector. Such genomes can transcribe all of the HBV RNAs under the endogenous promoters and enhancers, leading to expression of all of the viral proteins. They are replication competent and can release virions into culture supernatant. Of the two parental 1.5mer genomes used, clone 6.2 was directly isolated from a patient and has wild-type sequence in the entire genome; N16 is a chimeric construct with the backbone of clone 4B, a high replicating core promoter mutant, and the envelope gene of clone 6.2 (27, 41). The 0.7mer construct of N16 (0.7xN16L+M+S+) was created by inserting a 2.3-kb fragment covering nucleotides 2721 to 3221 and nucleotides 1 to 1770 into the SacI and HindIII sites of pBluescript vector. The 0.7mer construct expresses all three envelope proteins under endogenous L and M/S promoters, as well as the two enhancers. They do not express core protein, DNA polymerase, or HBx protein and are deficient in genome replication.
Most mutations were introduced into the 1.5mer and 0.7mer constructs by overlap extension PCR, followed by restriction fragment exchange. A single mutagenic PCR was used to generate the M-minus mutants because of the proximity of the mutations to the EcoRI site. The replaced restriction fragment was fully sequenced to confirm the intended mutations and lack of PCR errors. Figure Figure11 illustrates the single or multiple mutations used to abolish the expression of one or more envelope proteins, and Table Table11 summarizes the impact of these mutations on the coding capacities of the envelope and polymerase genes. Expression of full-length L protein was abolished by a G3028T mutation converting the 59th codon of the preS1 region into TGA. M protein expression was prevented by conversion of its initiation codon to GTG (A3211G), ACG (T3212C), or ATA (G3213A). A C117A mutation converts codon 43 in the preS2 region to TAG, thereby abolishing the expression of intact L and M proteins. The A155G/T156C double mutation converts the S gene AUG into GCG to prevent S protein expression. The G261A substitution truncates all of the three envelope proteins by converting codon 36 of the S gene into UAG.
In addition to mutations that prevent envelope protein expression, we introduced mutations into the Kozak sequence of the S gene to reduce S protein expression. Mutations targeting the −3, −1, and +4 positions of the S gene AUG were initially introduced into the 0.7xN16L−M−S+ construct (Table (Table2)2) . Mutations that greatly reduced S protein expression were subsequently tested in the 1.5mer constructs of N16 or 6.2. Table Table11 shows the amino acid substitutions rendered to the envelope proteins and, for those introduced into the 1.5mer construct, the polymerase protein as well.
Huh7 cells were maintained in Dulbecco modified Eagle medium supplemented with 10% fetal bovine serum. Transient transfection was performed on cells seeded overnight at a density of 7.0 × 105/well in six-well plates, with pcDNA3.1zeo(−) DNA to bring the total amount of DNA to 2 μg/well. In addition, 5 ng of cDNA encoding secreted alkaline phosphatase was cotransfected to check for possible variations in the transfection efficiency. TransIT-LT1 reagent (Mirus) was used as the reagent for transfection (2, 33). The medium was replaced 15 to 20 h later, and cells were harvested at day 5 posttransfection.
The details for DNA analysis have been described (21, 27, 33, 53). Cells were scraped off the six-well plates and lysed in 80 μl of lysis buffer [10 mM HEPES (pH 7.5), 100 mM NaCl, 1 mM EDTA, and 1% NP-40]. An aliquot of the cell lysate (40 μl) was digested at 37°C for 20 min with 7.5 U of mung bean nuclease and 0.5 U of DNase I. Next, core particles were precipitated by polyethylene glycol solution and further treated with DNase I (2 U) and mung bean nuclease (3 U). After proteinase K digestion, DNA was extracted with phenol, precipitated with ethanol, and dissolved in Tris-EDTA buffer. DNA was separated in 1.2% agarose gel in the presence of 1 μg of ethidium bromide/ml, transferred to GeneScreen Plus membranes (Perkin-Elmer), and hybridized with 32P-labeled full-length HBV DNA probe. The blots were washed with 2× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate)-0.1% sodium dodecyl sulfate (SDS) solution at 65°C for 1 h and then 0.5× SSC-0.1% SDS for 3 h. Signals were revealed by exposure to X-ray films.
Protein G-agarose beads (Roche) were incubated at 4°C overnight with a mixture of horse polyclonal HBs antibody (anti-Ad/Ay, Abcam) and rabbit polyclonal preS1 antibody (R271) at a ratio of 10-μl bed volume of beads with 1.4 μl of HBs antibody and 2.8 μl of preS1 antibody. The R271 antibody was raised against synthetic peptide 1-35 of the preS1 region of genotype D (MGQNLSTSNPLGFFPDHQLDPAFRANTANPDWDFN). Next, a 10-μl bed volume of loaded beads was incubated at 4°C 48 h with 1.4 ml of precleared culture supernatant. The beads were brought down by low-speed centrifugation and washed twice with phosphate-buffered saline (PBS). The precipitated virions were digested at 37°C for 15 min with 1 U of DNase I and 1.5 U of mung bean nuclease, followed by proteinase K digestion, phenol extraction, and DNA precipitation using 20 μg of glycogen as a carrier. DNA was separated in agarose gels and subject to Southern blot analysis. Histograms of virion DNA quantification were calculated by the NIH ImageJ 1.37v software. To detect naked core particles, 1.4 ml of precleared culture supernatant was incubated at 4°C overnight with 1.4 μl of a rabbit antibody against core protein (Dako). Protein G-beads (10 μl) were added, followed by 5 additional hours of incubation. The immune complex was collected by low-speed centrifugation, washed, and subjected to Southern blot analysis.
As previously described (41, 53), virions and naked core particles were precipitated from culture supernatant by ultracentrifugation at 39,000 rpm in Sorvall rotor and further centrifuged for 72 h in 0.33-g/ml CsCl solution in TEN buffer (10 mM Tris, 1 mM EDTA, 100 mM NaCl). Fractions of 400 μl were collected from the top, weighed, and dialyzed against TEN solution. Samples were treated with DNase I and mung bean nuclease, followed by proteinase K digestion and DNA precipitation. To verify the nature of virion-associated HBV DNA, DNA samples were digested with EcoRI (25 U) at 37°C for 7 h or heated at 100°C for 10 min, followed by cooling on ice. DNA was separated in agarose gels for Southern blot analysis.
Proteins from 20 μl of cell lysate were separated on a 0.1% SDS-12% polyacrylamide gels and transferred to polyvinylidene difluoride membranes. The membranes were blocked at room temperature for 30 min with 3% nonfat milk dissolved in PBS-0.05% Tween 20 (PBST) and subsequently incubated at 4°C overnight with horse polyclonal anti-HBs antibody (anti-Ad/Ay; Abcam) diluted 1:4,000 in 3% milk-PBST. Blots were washed for 40 min in PBST and incubated at room temperature for 1 h with 1:100,000 dilution of rabbit anti-horse antibody conjugated with horseradish peroxidase (HRP; Abcam). The blots were washed again, and signals were detected by using Western Lightning chemiluminescence reagent (Perkin-Elmer). The bound antibodies were stripped from the blots by shaking at 37°C for 25 min with Restore Western blot stripping buffer (Pierce). The blots were rinsed with water and washed with PBST buffer for 40 min and blocked again. A similar procedure was performed for other antibodies. The dilutions of the primary and secondary antibodies were as follows: 1:5,000 for the anti-preS1 antibody (R271), followed by 1:60,000 of anti-rabbit-HRP; 1:1,000 for anti-preS2 antibody (clone S26; Virogen), followed by 1:10,000 anti-mouse-HRP; and 1:5,000 for anti-GAPDH (Millipore), followed by 1:10,000 of anti-mouse-HRP. For the preS2 and GAPDH antibodies, 3% bovine serum albumin (BSA) rather than 3% milk was used in the blocking solution and for antibody dilution.
Hepatitis B surface antigen (HBsAg) was measured from 5 μl of precleared culture supernatant by the Auszyme monoclonal HBsAg kit (Abbott Laboratories).
Our goals were to determine the impact of envelope protein expression levels and L/S protein ratios on virion secretion. To this end, transcomplementation assay was used as our initial approach. We used DNA constructs of 1.5x genome length (1.5mers) to drive genome replication. They were rendered defective in the expression of one or more envelope proteins through ablation of the translation initiation site or introduction of a premature stop codon (Fig. (Fig.11 and Table Table1).1). The missing envelope proteins were provided in trans by an expression construct of 0.7x genome length (0.7mer), which was deficient in genome replication (Fig. (Fig.1).1). By transfecting a fixed amount of the replication construct with variable amounts of the expression construct, we could easily alter the ratio between genome replication and envelope protein expression or the ratio between different envelope proteins. Since M protein has been reported to be dispensable for virion secretion (10, 15, 52), we reconstituted just the L and S proteins in some cotransfection experiments. We measured HBsAg in culture supernatant by a commercial enzyme-linked immunosorbent assay (ELISA)as a surrogate and quantitative marker for subviral particles, although the exact ratio between virions and subviral particles remains to be validated by other methodologies.
Most of the study was based on N16, a chimera between clones 4B and 6.2 (27). It has a high replication capacity due to core promoter mutations (41). Huh7 cells were transfected with 1 μg of 1.5xN16L−M−S− (unable to express functional L, M, or S protein) together with 0 −1 μg of 0.7xN16L+M+S+ for the missing envelope proteins (Fig. (Fig.2,2, lanes 1 to 8). As a positive control, 1 μg of the parental N16 construct (1.5xN16L+M+S+) was transfected alone (lane 9). The total amount of DNA transfected was kept at 2 μg using pcDNA3.1/zeo(−) DNA. Not surprisingly, increasing the expression construct gradually augmented both intracellular L and S proteins (Fig. (Fig.2D),2D), as well as secreted HBsAg (Fig. (Fig.2F).2F). However, compared to the parental 1.5mer construct, the 0.7xN16L+M+S+ construct was much more efficient at retaining the L protein and, to a lesser extent, the S protein. Thus, 0.05 and 0.2 μg of 0.7xN16L+M+S+ produced amounts of intracellular L and S proteins, respectively, comparable to that produced by 1 μg of 1.5xN16L+M+S+ (Fig. (Fig.2D,2D, lanes 3, 5, and 9). Remarkably, as little as 0.025 μg of the expression construct efficiently rescued virion secretion (Fig. 2B and E, lane 2), when HBsAg secretion reached only 5% of the parental construct. Highest level of virion secretion was achieved at 0.05 or 0.1 μg (lanes 4 and 5), when HBsAg secretion was 10 and 30% of the parental construct, respectively. Further increase of the expression construct reduced virion secretion instead (lanes 7 and 8). Thus, at least in the setting of the cotransfection experiments between 1.5mer and 0.7mer constructs, very little expression construct was needed to drive efficient L and S protein retention, as well as virion secretion, suggesting preferential incorporation of envelope proteins into virions.
One may wonder whether the DNA detected in the Southern blot actually represented naked core particles, which are also released from transfected Huh7 cells (2, 27, 41). Immunoprecipitation with a polyclonal core antibody did demonstrate secretion of naked core particles, but that was independent of envelope protein expression (Fig. (Fig.2C,2C, lanes 1 to 9). Moreover, HBV DNA inside core particles was predominantly single stranded (Fig. (Fig.2C),2C), in contrast to the double-stranded DNA associated with virions (Fig. (Fig.2B).2B). Ultracentrifugation through CsCl gradient enabled separation of virions from naked core particles without the need for antibodies. This approach confirmed more efficient virion secretion by a low dose (0.025 μg) of the expression construct than a high dose of 1 μg (Fig. (Fig.3A,3A, compare panels 2 and 3). The nature of virion-associated DNA was analyzed by EcoRI digestion, which should convert relaxed circular DNA (which migrates slower than the 3.2-kb marker) into 3.2-kb linear form, and the 3.2-kb partially double stranded DNA (which migrates faster than the 3.2-kb marker) into two fragments of 1.8 and 1.4 kb. The result confirmed higher content of the relaxed circular DNA by the low dose of the expression construct than the high dose (Fig. (Fig.3B,3B, left panel, lanes 2 and 3). As further support for the assignment of the two DNA forms, heat treatment converted both the relaxed circular and the partially double stranded DNA into faster-migrating forms indicative of single-stranded DNA (Fig. (Fig.3B,3B, right panel).
Clone N16 contains a quadruple core promoter mutation that markedly increases HBV genome replication (41, 53). Such a high concentration of core particles may favor virion formation. To test this hypothesis, we repeated the cotransfection experiment of 0.7xN16L+M+S+ with the L−M−S− mutant of clone 6.2 as a 1.5mer construct, which has a much lower replication capacity than N16 (27, 41). As shown in Fig. 4B and D, efficient virion secretion (relative to the parental 1.5mer of clone 6.2) was also achieved at 0.025 μg of the expression construct, when HBsAg secretion was only 10% of the parental clone (lanes 2 and 7). Increasing the expression construct to 0.8 μg rather reduced virion secretion (lane 6). In contrast to virion secretion, the accumulation of intracellular L, M, and S proteins and secretion of HBsAg were rather dependent on the dose of the 0.7mer expression construct (Fig. 4C and E). Therefore, the ability of a very low dose of the envelope protein expression construct to rescue virion secretion does not depend on high replication capacity.
Having established the effect of altered expression of all of the three envelope proteins on virion secretion, we next decided to alter the expression of individual envelope proteins. It is well established that HBsAg secretion is driven by the S protein but can be suppressed by L protein in a dose-dependent manner (13, 39, 43). Virion secretion requires L protein, but too high an L/S ratio will also block virion secretion. In the first set of experiments, increasing doses of an S protein construct (0.7xN16L−M−S+) was cotransfected with constant amount of a replication construct capable of L protein expression (1.5xN16L+M−S−). This permitted L protein expression at the physiological level. Increasing S protein construct from 0.03 to 1 μg augmented both intracellular S protein (Fig. (Fig.5C)5C) and secreted HBsAg (Fig. (Fig.5E).5E). Virion secretion was not only inefficient at the lower end of the S protein construct, but also at the high end of 1 μg. It peaked at 0.25 or 0.5 μg of the S protein construct, which produced similar intracellular levels of S and L proteins as the N16 parental construct (Fig. (Fig.5C,5C, lanes 5, 6, and 9).
In the reverse experiments, we cotransfected a replication construct capable of S protein expression (1.5xN16L−M−S+) with various doses of L/M expression construct (0.7xN16L+M+S−) or an expression construct for just the L protein (0.7xN16L+M−S−). Increasing the dose of the L/M or L construct caused a modest increase in intracellular level of the S protein (Fig. (Fig.6C6C and Fig. Fig.7C)7C) but a striking and dose-dependent reduction in HBsAg secretion (Fig. (Fig.6E6E and Fig. Fig.7E).7E). Thus, at the highest expression level of the L/M or L protein a significant amount of retained S protein was probably degraded. Virion secretion was inefficient at both the low and high ends of L+M or L protein construct. The highest level of virion secretion was achieved at 0.1 μg of the 0.7xN16L+M+S− construct (Fig. 6B and D, lane 5) and 0.2 μg of the 0.7xN16L+M−S− construct (Fig. 7B and D, lane 5). It is noteworthy that 0.1 μg of the 0.7xN16L+M+S− construct produced similar amounts of the intracellular L and S proteins as 1 μg of the N16 parental construct (Fig. (Fig.6C,6C, lanes 5 and 9).
We further performed triple transfection of a replication construct with defective envelope protein expression (1.5xN16L−M−S−) with an expression construct for L protein (0.7xN16L+M−S−) and an construct for S protein (0.7xN16L−M−S+). The replication construct was kept at 1 μg as before. In part of the experiment, the S protein construct was kept at 0.6 μg, whereas the L construct varied from 0.012 to 0.4 μg. Highest level of virion secretion (about half of the efficiency of N16) was achieved at 0.10 and 0.20 μg of the L protein construct (Fig. 8B and D). In the other part, the L construct was kept at 0.25 μg, whereas the S construct changed from 0.015 to 0.75 μg. In that case highest level of virion secretion (ca. 70% of N16) was achieved at 0.5 and 0.75 μg of the S expression construct. At such doses the levels of intracellular L and S proteins exceeded those of the N16 parental construct (Fig. (Fig.8C8C).
All of the observations made thus far were based on cotransfection of two or three plasmids. To determine the feasibility of generating HBV variants with reduced secretion of subviral particles, we attempted to reduce S protein expression at the translational level. In this regard, a purine (guanine or adenine) at the −3 position and a guanine at the +4 position of the translation initiation codon are the most important elements for efficient protein expression (30). For the S gene, these positions correspond to A152 and G158, respectively. Experiments with the 0.7xN16L−M−S+ construct revealed that A152G/G158C and A152T reduced S protein secretion by 55 and 83%, respectively, whereas the A152T/G158C and A152C/G158C mutations completely abolished S protein secretion (Table (Table2).2). When introduced into the 1.5mer of clone N16 (1.5xN16 L+M+S+), the A152G/G158C double mutation reduced HBsAg secretion by only 30% and maintained wild-type level of virion secretion (Fig. (Fig.9B,9B, D, and E, lane 3 versus lane 1). The A152T mutation still caused an impressive 70% reduction of HBsAg secretion but only slightly increased virion secretion (lane 2). In addition, we introduced the A152T mutation into the 1.5mer of clone 6.2, a wild-type clone of much lower replication capacity. In that case the A152T mutation reduced HBsAg secretion by 60% and also increased virion secretion (Fig. (Fig.4B,4B, D, and E, lanes 9 and 10).
To address the role of M protein in virion secretion, we generated M-minus mutants by converting the preS2 AUG into ACG. The mutation was introduced into N16, as well as its A152T single mutant and A152G/G158C double mutant. Western blot analysis confirmed loss of M protein expression (Fig. (Fig.9C,9C, top panel). The M-minus mutants secreted fewer virus particles than the parental constructs, but these virions had much higher ratio of the relaxed circular DNA than the partially double-stranded linear DNA. Moreover, the M-minus mutants showed slower migration of both forms, suggestive of more complete elongation of the plus-strand DNA (Fig. (Fig.9B,9B, compare lanes 1 to 3 with lanes 4 to 6). Analysis of two additional M-minus mutants of N16, for which the preS2 AUG was mutated to AUA and GUG, respectively, revealed a phenotype similar to that of the ACG mutant (Fig. 10B). Lack of M protein expression also increased intracellular level of the S protein for the A152T mutant, which had a low level to begin with (Fig. (Fig.9C,9C, lanes 2 and 5). Consequently, HBsAg secretion of the A152T mutant was further reduced from 30% of 1.5xN16L+M+S+ to a mere 10% (Fig. (Fig.9E,9E, lanes 2 and 5). This observation suggests that the M protein accelerates S protein-mediated particle release, a notion already proposed by others (42).
Curiously, the A152T mutation not only markedly reduced intracellular S protein but also the M protein (Fig. (Fig.4C,4C, lanes 9 and 10; Fig. Fig.9C,9C, lanes 1 and 2). To study this issue further, we compared expression of the three envelope proteins among five 0.7mer constructs: LMS, LM, L, M, and S (Fig. 11A, lanes 1 to 5). The same amount (2 μg) of each DNA construct was transfected, and a preS2 antibody was used to simultaneously detect L and M proteins. The M construct and also the LM construct produced negligible amounts of the M protein in comparison with the LMS construct (Fig. 11A, top panel, compare lanes 2 and 4 with lane 1). Cell lysis in the presence of 0.1% SDS did not improve M protein detection (data not shown). Since the low intracellular levels of the M protein could not be accounted for by increased secretion (Fig. 11B), this result suggests that M protein stability requires coexpression of the S protein. To confirm this hypothesis, 1 μg of the M protein construct was cotransfected with increasing doses of L or S protein construct or with just pcDNA vector. Only cotransfection with the S construct rescued both the intracellular and secreted M protein, with the maximum effect achieved by 0.18 μg of the S construct (Fig. 11A and B, lanes 9 to 11).
To restore virion secretion efficiency of the M-minus mutant, we cotransfected 1.3 μg of an M-minus mutant of N16 (the ACG mutant) with 0.01 to 0.64 μg of the M protein construct (0.7xN16L−M+S−). Surprisingly, neither the efficiency of virion secretion nor the genome maturity was reverted to that of the M-expressing construct (data not shown). The experiment was repeated using a double expression construct for L/M proteins (0.7xN16L+M+S−) or M/S proteins (0.7xN16L−M+S+). Whereas the M/S coexpression construct did not affect genome maturity (Fig. 12B, lanes 4 to 7), the L/M construct reduced genome maturity in a dose-dependent manner (lanes 8 to 11). At the dose of 0.16 μg it also increased the efficiency of virion secretion (lane 10). The reduced virion secretion at 0.64 μg (lane 11) is most likely due to too high expression level of L protein (Fig. 12C), which also markedly reduced HBsAg secretion (Fig. 12E).
Of the three envelope proteins produced by HBV, the L and S proteins are essential for virion secretion, whereas the M protein is dispensable (10, 15, 52). The S protein is required for particle assembly because it can be secreted alone as 22-nm subviral particles. The L protein serves as the matrix protein by physical contact with core particles in the cytosol. Both the preS1 and the preS2 domains of the L protein are cytosolically disposed, and a continuous 22-amino-acid sequence at the preS1/preS2 junction is essential for virion formation (8, 32). It is not clear whether this sequence directly contacts the core particle or is brought closer to the core particle by an adaptor protein such as γ2-adaptin (22, 46). Although residues 25 to 79 of the S protein are also exposed in the cytosol, they do not harbor a determinant for virion formation (5).
The major objective of the present study was to determine whether markedly reducing the secretion of subviral particles also impairs virion secretion. Two types of approach were taken: transcomplementation assay between a replication construct and one or two expression constructs for the missing envelope proteins(s), a single plasmid for which expression of the S protein was reduced. We confirmed the ability of the L protein to inhibit HBsAg secretion in a dose-dependent manner (Fig. (Fig.77 and and8)8) (13, 39, 43, 50). We also found that virion secretion was suboptimal at both the low and high ends of the L/S ratio (Fig. (Fig.55 to to7).7). Most importantly, cotransfection of replication constructs defective in L, M, and S protein expression with an expression construct for the missing envelope proteins led to efficient reconstitution of virion secretion at a very low dose of the expression construct, when HBsAg secretion was only 5 to 10% of the parental construct based on a commercial ELISA (Fig. (Fig.22 and and4).4). It is possible that the actual reduction in the secretion of subviral particles was much greater than suggested by the ELISA. In addition to the transcomplementation assay, an A152T substitution at the −3 position of the S gene AUG codon reduced HBsAg secretion by 60 to 70% but only slightly increased virion secretion (Fig. (Fig.44 and and9).9). Although it will be of interest to further reduce S protein expression (or the expression of all of the three envelope proteins), our observations thus far indicate that efficient HBV virion secretion does not require a marked overproduction of subviral particles.
Consistent with our observations, Xu and Yen found that a 129-nucelotide deletion in the preS1 region (a 43-amino-acid deletion in the L protein), the promoter for the 2.1-kb M/S transcript, greatly reduced intracellular S protein and secreted HBsAg by 90 and 99%, respectively (57). However, the deletion mutant was still capable of virion secretion. In another study, cloning of 1.1 copies of the HBV genome downstream of the cytomegalovirus promoter resulted in much greater transcription of the pgRNA relative to the subgenomic RNAs and consequently a much higher ratio of virions to subviral particles (11). These reports and our findings suggest that envelope proteins are preferentially routed for virion formation when expressed at low levels. Whereas subviral particles bud into a post-ER, pre-Golgi compartment and are transported by the constitutive secretory pathway (26, 42), the ESCRT (endosomal sorting complex required for transport) complex mediates virion secretion (31, 46, 55). This complex sorts monoubiquitinated proteins for degradation, lysosomal functions, or release through the formation of multivesicular bodies in the endosome. Both the L and the core proteins have been found in the late endosomal compartment, and a possible ubiquitination site and late motif have been identified in core protein (46). Importantly, perturbation of this system by gene silencing or dominant-negative mutants selectively impaired the secretion of virions but not subviral particles.
For both clones N16 and 6.2, the A152T mutation increased virion secretion despite reducing intracellular levels of the S protein and secreted HBsAg (Fig. (Fig.44 and and9).9). This result suggests that the L/S protein ratio achieved by the wild-type virus, at least the 1.5mer construct, is suboptimal for virion secretion. Similarly, the ability of very low doses of the 0.7xL+M+S+ construct to rescue efficient virion secretion is apparently linked to a higher L/S protein ratio achieved by such a short construct and intracellular retention of both proteins. Thus, as little as 0.05 μg of 0.7xN16L+M+S+ produced similar level of intracellular L protein as 1 μg of 1.5xN16L+M+S+, whereas 0.2 μg of 0.7xN16L+M+S+ was needed to produce amounts of intracellular S protein similar to those produced by the 1.5mer construct (Fig. (Fig.2D).2D). In this regard, all of the HBV transcripts share the same 3′ end, making it possible for the upstream transcription unit to interfere with the downstream transcription units through promoter interference (40). In the 1.5mer constructs, the transcriptional unit for pgRNA lies upstream of the transcription units of the 2.4- and 2.1-kb mRNAs for envelope proteins. Such an arrangement may reduce the expression of envelope proteins at the transcriptional level. If so, inhibition by the pgRNA should be relieved in the 0.7mer constructs, but the 2.4-kb L transcript may begin to interfere with the transcription of the 2.1-kb mRNA. In DHBV, ablation of pgRNA transcription from a tandem dimer construct has been shown to increase the transcription of mRNAs for envelope proteins (25). Further experiments are needed to test the relevance of promoter interference in the regulation of HBV gene expression.
Several unexpected findings were made with regard to the M protein. First, M protein expression or stability requires coexpression of the S protein (although not necessarily from the same plasmid). In most previous studies of M protein expression that we are aware of, the internal S gene AUG codon was either intact or changed by a single nucleotide change. We converted the AUG codon of the S gene into GCG, thus making it unlikely for the S protein to be expressed from a noncanonical initiation site. The S protein could also increase intracellular level of the L protein (Fig. (Fig.5C5C and and8C),8C), especially in cis (Fig. 11A, compare lanes 1 and 2). The M protein is probably degraded unless associated with the S protein. Second, the lack of M protein expression led to less-efficient virion secretion but more relaxed circular genome than linear genome and higher genome maturity (slower migration) of both forms (Fig. (Fig.99 and and10).10). This observation is interesting in that DHBV, which expresses only two envelope proteins (L and S), secretes virions with majority having full-length plus-strand DNA (38). Third, virion secretion phenotypes of the M-minus mutant could be reversed by coexpression of L and M proteins at a certain dose but not by M protein alone or coexpression of the M and S proteins (Fig. (Fig.1212).
The ability of a viral factor to simultaneously affect the efficiency of HBV virion secretion and genome maturity is a recurrent theme. The myristoylation signal in the L protein was reported to reduce the efficiency of virion secretion but enhance the genome maturity (20). A 12-amino-acid insertion unique to the core protein of genotype G had a similar effect and also greatly increased relaxed circular DNA content (33). Genome maturation has been found to increase the nuclease sensitivity and reduce the stability of core particles (29), possibly leading to particle degradation in late endosomes in light of the current understanding of HBV virion morphogenesis (31, 46, 55). Core particle association with the L protein, or the L protein together with the M and S proteins, is necessary to avoid the fate of endosomal destruction. Thus, envelopment of core particles at an earlier stage of DNA synthesis will increase virion yield but reduce genome maturity due to the shutoff of DNA synthesis. At present we do not know how the M protein exerts its effect on virion formation. The transmembrane topology of the M protein is similar to the S protein rather than the L protein. Thus, the entire preS2 domain faces the ER lumen and is unable to interact with core particles. Considering that the S protein is required for M protein stability and yet the effect of M protein on virion secretion requires its coexpression with the L protein, we propose that the M protein increases the efficiency of virion formation by simultaneously contacting both the L and the S proteins. One feature unique to the M protein is the N-linked glycosylation site at position 4 of the preS2 domain, which is absent in the S protein and not used in the L protein due to a cytosolic disposition. Prevention of glycosylation or subsequent trimming of the oligosaccharides blocked virion secretion (6, 34), and mutation of this site was reported to impair virion or subviral particle secretion (36, 56). We found that virion secretion defects caused by naturally occurring missense mutations in the S protein could be overcome by another mutation in the S protein creating a novel N-linked glycosylation site (K. Ito et al., unpublished data).
If overproduction of subviral particles is not a by-product of efficient virion secretion, what are the true functions of subviral particles? They could affect HBV infectivity at the single-cell level (a direct impact on viral entry) or at the organism level (most likely through modulation of the host immune clearance mechanism). Since the HBV subviral particles contain very little L protein, the ligand to viral receptor (7, 28), the large excess of subviral particles probably does not directly compete for receptor binding. Rather, the subviral particles may induce immune tolerance during perinatal infection, thus delaying the rise of neutralizing antibodies against the S domain. A similar immunotolerant role has been attributed to HBeAg, a secreted version of the core protein (12, 37). Alternatively or additionally, the vast access of subviral particles can soak the rising anti-HBs antibodies to delay the clearance of infection. Indeed, immune complexes have been reported for both HBV and related woodchuck hepatitis virus (1, 19, 35, 45). It will be most interesting to determine whether drastically reducing the production of subviral particles curtails the duration of woodchuck hepatitis virus infection in neonatal woodchucks.
This study was supported by NIH predoctoral fellowship F31CA119941 (T.G.), an American Cancer Society Research Scholar Award RSG 06-059-01-MBC (S.T.), and NIH grants CA109733 to J.L. and AA08169 and CA123544 to J.W.
Published ahead of print on 12 August 2009.
†This paper is dedicated to the memory of T. S. Benedict Yen for his contribution to HBV research.