Pre-P Is a Secreted Glycoprotein Encoded as an N-Terminal Extension of the Duck Hepatitis B Virus Polymerase Gene

Abstract

The duck hepatitis B virus (DHBV) pregenomic RNA is a bicistronic mRNA encoding the core and polymerase proteins. Thirteen AUGs (C2 to C14) and 10 stop codons (S1 to S10) are located between the C1 AUG for the core protein and the P1 AUG that initiates polymerase translation. We previously found that the translation of the DHBV polymerase is initiated by ribosomal shunting. Here, we assessed the biosynthetic events after shunting. Translation of the polymerase open reading frame was found to initiate at the C13, C14, and P1 AUGs. Initiation at the C13 AUG occurred through ribosomal shunting because translation from this codon was cap dependent but was insensitive to blocking ribosomal scanning internally in the message. C13 and C14 are in frame with P1, and translation from these upstream start codons led to the production of larger isoforms of P. We named these isoforms “pre-P” by analogy to the pre-C and pre-S regions of the core and surface antigen open reading frames. Pre-P was produced in DHBV16 and AusDHBV-infected duck liver and was predicted to exist in 80% of avian hepadnavirus strains. Pre-P was not encapsidated into DHBV core particles, and the viable strain DHBV3 cannot make pre-P, so it is not essential for viral replication. Surprisingly, we found that pre-P is an N-linked glycoprotein that is secreted into the medium of cultured cells. These data indicate that DHBV produces an additional protein that has not been previously reported. Identifying the role of pre-P may improve our understanding of the biology of DHBV infection.

Hepadnaviruses are small DNA-containing viruses that replicate by reverse transcription (39). Hepadnaviruses have been found in birds, rodents, and primates (22, 38). Human hepatitis B virus (HBV) chronically infects over 350 million people worldwide and is a major cause of liver disease and liver cancer (23). Duck hepatitis B virus (DHBV) is a common model for HBV (25, 46). Avian hepadnaviruses have also been detected in heron (heron HBV [HHBV]) (41), Ross goose (Ross goose HBV [RGHBV]), snow goose (snow goose HBV [SGHBV]) (5), sheldgoose (12), stork (stork HBV [STHBV]) (32), and crane (crane HBV [CHBV]) (31).

The organization of the 3,021-nucleotide (nt)-long DHBV genome (Fig. 1) is very compact. All nucleotides are within at least one of three open reading frames (ORFs), and the expression of multiple proteins from one ORF via initiation at multiple in-frame AUG codons is common. The first ORF encodes the core protein (C) and e antigen (e-Ag) (37), with the e-Ag being encoded as an N-terminal extension of the C ORF. The second ORF encodes the envelope proteins L and S, and like the organization of C and e-Ag, the L protein is an N-terminal extension of the S protein. The third ORF encodes the polymerase/reverse transcriptase protein (P). In mammalian viruses, a fourth ORF encodes the X protein, a multifunctional regulatory protein (2, 3). DHBV lacks an apparent X ORF, but a potential cryptic X-like ORF has been reported (6). In vivo experiments revealed no functional role for this protein in short-term infection (26).

FIG. 1.

DHBV3 pgRNA genomic organization. (Top) DHBV3 pgRNA showing the location of the ORFs, ɛ, the cap, and the poly(A) tail. The pgRNA is 3.3-kb polyadenylated RNA with a terminal redundancy of approximately 270 nt. (Bottom) Enlarged view of the 5′ end of the DHBV3 pgRNA. C1 and P1 are the AUGs for C and P. CO2, C2-C14, and P2 are AUG codons. S1 to S10 are stop codons. The open boxes are small ORFs upstream of P. NsiI and EcoRI are sites for the insertion of the BamHI-SL. The lines below the genomic diagram represent the shunting mechanism employed for the initiation of P translation. Dashed line, scanning ribosomes; thin lines, shunting ribosomes; black arrows, ribosomes translating C or P. The nucleotide positions of key sites examined in this study are shown.

The products of all of the hepadnaviral ORFs possess regulatory functions in addition to their structural and enzymatic roles. The S ORF encodes the viral surface glycoproteins, but these proteins are also secreted at high levels into the blood in subviral particles, where they assist immune evasion by acting as an immunoglobulin trap (23). In addition, the S-Ags may regulate cellular transcription (14). The e-Ag is a secreted protein that appears to be a neonatal tolerogen (27), and its absence may confer a growth advantage of precore-minus mutants over wild-type virus late in chronic infection (52). The X protein in mammalian hepadnaviruses is exclusively a regulatory protein that controls host signal transduction and transcription (2). Finally, we have found that DHBV P can suppress mRNA accumulation in cells, including its own message, the pregenomic RNA (pgRNA) (4).

The hepadnaviral pgRNA is the RNA template for reverse transcription, and it is also a bicistronic mRNA encoding the C and P proteins (Fig. 1). In DHBV, the pgRNA has a ∼118-nt-long 5′ leader upstream of the C ORF that contains a stem-loop (ɛ) (1), which is an essential signal for encapsidation and reverse transcription (15, 19, 30, 45, 47). In DHBV strain 3, the P ORF starts 544 nt downstream of the start site for C and ∼662 nt downstream of the cap. Thirteen AUGs (C2 to C14) are between the C1 AUG and the P1 AUG that initiates P translation, and all of them except C10 are in frame with the P1 AUG. Ten stop codons (S1 to S10) are also between the C1 and P1 AUGs. These stop codons terminate all upstream translation products and produce seven small ORFs, which have coding potentials of 2 to 29 amino acids (aa).

Despite being located in a very unfavorable position on the pgRNA, DHBV P is translated 10% as rapidly as C on a molar level, and the majority of DHBV P in cells accumulates in a nonencapsidated form (48, 49). Our analysis of the translation mechanism of P revealed that P is translated by ribosomal shunting (40) (Fig. 1). Ribosomal shunting is the discontinuous transfer of ribosomes along the mRNA, in which the ribosomes bind to the cap and then transfer from a donor site to an acceptor site without linear scanning of the intervening region of the mRNA. Shunting has been described for a few viral and cellular mRNAs (8, 11, 35, 50), but its mechanism is poorly understood. Here, we report the discovery of a novel isoform of P that was found as we were investigating the mechanism of DHBV P initiation following ribosomal shunting.

MATERIALS AND METHODS

Plasmids.

D1.5G is a wild-type overlength DHBV3 (42) expression construct containing a 5′ duplication of nucleotides 1658 to 3021 in pBluescript(−) (Stratagene). D1.5G-C1^AUG− is a mutant in which the C1 AUG is destroyed by mutating A2647T. D1.5G-C1^AUG− was used as the “wild-type” construct in this study to eliminate the production of mutant C proteins that could confound the analysis of P translation or that could have unexpected effects on cell viability. A series of mutants was constructed based on D1.5G-C1^AUG− (Table 1).

TABLE 1.

Plasmids employed

Plasmid	Description
D1.5G-C1AUG−	Overlength expression vector for DHBV3 in which the C1 AUG was destroyed by mutating A2647T; this is the wild-type construct in these studies
P1loop−	Collapse of the P1 loop by changing nt 96-108 from AGAAGCTAATGTA to CATCTCT
P1loop−/C13−	Knockout of the C13 AUG by mutating T21C on the P1loop− background
S10−	Knockout of the S10 stop codon by mutating A108T
S10−/C14−	Knockout of the C14 AUG by mutating T105C on the S10− background
S10−/C13−	Knockout of the C13 AUG by mutating T21C on the S10− background
S10−/C13−/C14−	Knockout of the C13 and C14 AUGs by mutating T21C and T105C on the S10− background
C+/S10−	Knockout of the S10 stop codon on the D1.5G background
S10−/HA	Insertion of HA tag after nt 43 on the S10− background; the HA tag is in frame with the C13 AUG and the P ORF
S10−/C13−/HA	Insertion of HA tag after nt 43 on the S10−/C13− background
S10−/del43−94	Deletion of nt 43-94 on the S10− background
S10−/del110−163	Deletion of nt 110-163 on the S10− background
3′-HA	Insertion of the HA tag at the 3′ end of the P gene on the D1.5G background
C−/HA	Insertion of HA tag after nt 43
P1loop−/SL5−	Insertion of the BamHI-SL after nt 2539 adjacent to the pgRNA site on the P1loop− background
P1loop−/SLEcoRI	Insertion of the BamHI-SL at the EcoRI site on the P1loop− background
P1loop−/SLNsiI	Insertion of the BamHI-SL at the NsiI site on the P1loop− background
P1loop−/P1−	Knockout of the P1 AUG by mutation of A170T on the P1loop− background
DHBV16	Overlength expression vector for DHBV16 driven by cytomegalovirus immediate-early promoter
DHBV16/S10+	S10 stop codon was added to DHBV16 by mutating T108A

Cell culture, transfection, and harvesting.

LMH cells were used for most experiments. LMH cells are chicken hepatoma cells that produce infectious DHBV when transfected with pgRNA expression vectors such as D1.5G (7). Some experiments employed LMH-D2 cells, in which a dimer of the DHBV16 genome is stably integrated into the genome of LMH cells (28). The cells were maintained in 1:1 Dulbecco's modified Eagle's medium-F12 medium with 10% fetal bovine serum. LMH cells were seeded onto 60-mm dishes at a density of 1.2 × 10⁶ cells per plate 18 h prior to transfection. Transfections employed FuGENE (Roche) according to the manufacturer's instructions. Cell lysates and medium were harvested 1 or 4 days posttransfection by lysis in 0.75× radioimmunoprecipitation assay (RIPA) buffer (1× RIPA buffer is 20 mM Tris [pH 7.2], 150 mM NaCl, 1% Triton X-100, 1% sodium deoxycholate, and 0.1% sodium dodecyl sulfate [SDS]) on ice for 10 min, followed by clarification at 12,000 × g for 10 min at 4°C.

Western blotting.

Proteins were resolved by SDS-polyacrylamide gel electrophoresis (PAGE) and transferred onto Immobilon-P (Millipore) membranes. P was detected with anti-DHBV P monoclonal antibody (MAb) 11 (epitope aa 53 to 61) (48), and hemagglutinin (HA)-tagged P was detected by MAb 3F10 (Invitrogen) following incubation with the appropriate immunoglobulin G-alkaline phosphatase conjugate (Promega) or immunoglobulin G-horseradish peroxidase conjugate (GE Healthcare). Proteins were visualized by incubation with nitro blue tetrazolium and 5-bromo-4-chloro-3-indolyl phosphate (Promega) or ECL Plus (GE Healthcare).

Duck liver tissue lysate preparation.

A total of 0.2 g of duck liver was added to 4 ml prechilled 1× RIPA buffer plus 2 mM phenylmethylsulfonyl fluoride (Sigma) and 1 μg/μl leupeptin in an ice-cold Dounce homogenizer and was then homogenized on ice until the tissue was disrupted. After incubation on ice for an additional 5 min, the extract was clarified at 12,000 × g for 15 min, and the supernatant was saved.

Immunoprecipitation.

Polyclonal anti-P antibody (R2B2) was bound to protein A/G beads (Calbiochem), and the antibody-bead complexes were incubated with duck liver tissue lysate or DHBV-transfected medium overnight at 4°C. Immunocomplexes were washed four times with 1 ml 1× RIPA buffer. Bound protein was released by boiling in Laemmli buffer, followed by SDS-PAGE and detection of P by using Western blotting.

Glycosidase treatment.

Cell lysates or immunocomplexes were boiled for 10 min in denaturing buffer (0.5% SDS, 40 mM dithiothreitol). For N-glycosidase F (PNGase F) treatment, sodium phosphate (pH 7.5) and NP-40 were added to give final concentrations of 50 mM and 1%, respectively; for endoglycosidase H (Endo H), sodium citrate (pH 5.5) was added to a final concentration of 50 mM. The samples were divided into two portions and incubated for 1 h at 37°C with or without 1,000 U PNGase F (New England Biolabs) or Endo H (New England Biolabs). The reaction mixtures were dissolved in Laemmli buffer, followed by SDS-PAGE and detection of P by Western blotting.

Isolation of core particles.

Extracellular core particles were isolated from transfected LMH cells 5 days posttransfection by clarifying the supernatant at 3,000 × g for 10 min and then layering it over a 30% sucrose cushion and centrifugation at 192,000 × g overnight. Pellets containing core particles were dissolved in 50 μl/100-mm plate of cells of B/EDTA (10 mM HEPES [pH 7.8], 15 mM KCl, 5 mM EDTA) containing 5% sucrose (44). Duck serum was diluted twofold with 1× B/EDTA containing 0.5% NP-40 before layering it over a 30% sucrose cushion and centrifugation as described above.

Bioinformatic analysis of the pre-P region.

The avihepadnavirus sequences used in pre-P region analysis were obtained from GenBank: DHBV isolates included those reported under GenBank accession numbers X58567, X12798, X74623, M60677, K01834, AF493986, AF047045, X58569, X58568, M21953, AF404406, M32990, X60213, M32991, DQ195079, AJ006350, NC_001344, EU429326, EU429325, EU429324, DQ276978, AY494850, AY494851, AY250904, AY250903, AY250902, AY250901, AY294029, AY294028, AY521227, AY521226, AY536371, AY433937, AY392760, AY294656, and AF505512; sheldgoose HBV isolates included those reported under accession numbers AY494852 and AY494853; RGHBV isolates included those reported under accession numbers M95589, AY494848, AY494849, and NC_005888; STHBV isolates included those reported under accession numbers AJ251934, AJ251935, AJ251936, AJ251937, and NC_003325; HHBV isolates included those reported under accession numbers M22056 and NC_001486; CHBV isolates included those reported under accession numbers AJ441111, AJ441112, and AJ441113; and SGHBV isolates included those reported under accession numbers AF110999, AF111000, AF110996, AF110997, and AF110998. The ORFs were identified manually, and the deduced amino acid sequences were aligned using Clustal W. Protein distances were determined using the P distance algorithm in the MEGA DNA sequence analysis package (21). Statistical analyses were performed using SPSS v.13 (SPSS Inc.).

RESULTS

Ribosomes can initiate translation at the C13, C14, and P1 AUGs.

The P1 AUG that initiates the translation of the previously identified P isoform is located at the 3′ end of a predicted bulge on the pgRNA. To determine whether the predicted bulge is part of the shunting acceptor, we collapsed it by changing the sequence of nt 96 to 108 on the back side of the bulge from AGAAGCTAATGTA to CATCTCT. This mutation was built into the overlength genomic expression vector D1.5G-C1^AUG− to create D1.5G-P1loop⁻. In this vector, P is translated from the pgRNA. The only change in the pgRNA beyond the P1loop⁻ mutation was to ablate the expression of the C protein by mutating the C1 AUG. This was done to prevent ribosomal interference at the P1 AUG from ribosomes translating the overlapping C ORF and to prevent the synthesis of mutant C proteins from the lesions within the C ORF. For the purposes of this study, D1.5G-C1^AUG− (“DHBV C⁻”) was considered to be the wild-type DHBV expression vector.

The wild-type and P1loop⁻ mutant genomic expression vectors were transfected into LMH cells, which are chicken hepatoma cells that reverse transcribe DHBV and secrete infectious virus (7). One day posttransfection, whole-cell lysates were collected, and P was detected by Western analysis using MAb 11, which recognizes DHBV P aa 53 to 61. Collapsing the loop had no effect on initiation at P1 (Fig. 2B and C, lane 3), indicating that the loop is not an essential part of the shunt acceptor. However, a diffuse cluster of larger P isoforms appeared when the P1 loop was collapsed. This mutation fortuitously ablated the S10 stop codon and the C14 AUG that are present in this strain of DHBV (DHBV3), while keeping the C13 AUG in frame with the P ORF (Fig. 2A). The size of these larger forms was consistent with initiation from the upstream in-frame C13 AUG. Initiation from C13 was confirmed because mutating the C13 AUG in the P1loop⁻ mutant eliminated the larger forms of P (Fig. 2B, lane 4).

FIG. 2.

Translation of the P ORF can initiate at the C13, C14, and P1 AUGs. LMH cells were transfected with DHBV C⁻ derivatives, and P was detected by Western analysis of lysates on day 1 posttransfection. (A) Mutation sites on the DHBV3 pgRNA. (B) Accumulation of P from the C13, C14, and P1 AUGs in the P1loop⁻ background. (C) Accumulation of P from the C13, C14, and P1 AUGs in the S10⁻ background. PBS, pBluescript (empty vector control); WT, wild type (DHBV C⁻). The codons from which the various P isoforms initiate are indicated. (D) Accumulation of pre-P from genomes in which C1 is intact (C⁺/S10⁻) or ablated (S10⁻).

To eliminate the possibility that the P1loop⁻ mutation may have had effects on P translation through mechanisms other than ablating the S10 stop codon, we mutated the S10 stop codon in D1.5G C⁻ by changing A108 to T. Larger P forms indistinguishable from those expressed by the P1loop⁻ construct were detected in lysates of cells transfected with the S10⁻ mutant (Fig. 2C, lane 4). Because there are two in-frame AUGs upstream of S10 (C13 and C14) (Fig. 1) that could produce P products predicted to be 5.5 kDa and 2.4 kDa larger than P translated from the P1 AUG, we asked if the larger P products produced by the S10⁻ mutant arose from initiation at C13, C14, or both. Deleting C13 and C14 individually in the S10⁻ background each reduced the amount of the larger P isoforms (Fig. 2C, lanes 5 and 6), and mutating both C13 and C14 ablated the production of the larger isoforms (Fig. 2C, lane 7). Therefore, the diffuse cluster of larger P isoforms is comprised of products initiated from both the C13 and C14 AUGs.

To address the possibility that the production of pre-P was induced by the ablation of the C1 AUG, we restored the C1 AUG in C⁺/S10⁻ and evaluated the P isoforms produced following the transfection of LMH cells. Figure 2D (lanes 3 and 4) reveals that the production of pre-P was unaffected by the presence or absence of the C1 AUG.

Translation of the sequences immediately downstream of C13 was confirmed using the mutant S10⁻/HA, in which an in-frame HA epitope tag (20) was added 21 nt downstream of the C13 AUG but upstream of the C14 AUG. S10⁻/C13⁻/HA is a negative control in which the C13 AUG was ablated in the S10⁻/HA construct (Fig. 3A). 3′ HA is the positive control for the detection of HA-tagged P; it has the HA tag added to the 3′ end of P in the wild-type background. LMH cells were transfected with the S10⁻/HA, S10⁻/C13⁻/HA, and 3′ HA expression vectors, and cellular lysates were analyzed by Western analysis employing anti-HA antibodies. The larger P isoforms were found in lysates from S10⁻/HA-transfected cells (Fig. 3B, lane 3), and the standard P isoform was found in the 3′ HA lysates (Fig. 3B, lane 5), but no bands were detected in lysates from cells transfected with the mutant S10⁻/C13⁻/HA (Fig. 3B, lane 4). As expected, both the standard and larger P isoforms were detected when the blots were probed with an anti-P antibody (Fig. 3C, lanes 2 to 5).

FIG. 3.

The pre-P region is translated in the S10⁻ background. An HA epitope tag was inserted 21 nt downstream of C13 in frame with P, the expression vectors were transfected into LMH cells, and P expression was measured by Western analysis of cell lysates. (A) Diagram of mutation sites on the DHBV3 pgRNA. (B) Detection of pre-P by anti-HA antibody 3F10. (C) Detection of P and pre-P by anti-P antibody MAb 11. PBS, pBluescript (empty vector control); WT, wild type (DHBV C⁻); HA, HA epitope tag. The codons from which the various P isoforms initiate are indicated.

Therefore, the translation of the DHBV P ORF initiates at C13 and C14 in addition to P1, but in DHBV3, translation from C13 and C14 is terminated by the S10 codon. We named these larger P isoforms present when S10 is ablated “pre-P” by analogy to the pre-C and pre-S proteins.

Pre-P is produced naturally by DHBV.

Initiation at C13 cannot be detected in wild-type DHBV3 due to the S10 stop codon, but S10 is not present in most other DHBV strains. Therefore, to determine whether the pre-P products are synthesized naturally, we immunoprecipitated P from liver lysates from ducks infected with DHBV3, DHBV16, or AusDHBV using a polyclonal antibody (R2B2) and detected P with a MAb (MAb 11). As shown in Fig. 4, larger P isoforms of the same size as pre-P produced in LMH cells were detected in DHBV16- and AusDHBV-infected duck liver (Fig. 4B, lanes 4, 8, and 9). However, the larger P isoforms could not be detected in DHBV3-infected duck liver, consistent with the presence of the S10 stop codon (Fig. 4B, lane 7).

FIG. 4.

Pre-P is produced during infection of ducks. P was immunoprecipitated from liver lysates from ducks infected with DHBV3, DHBV16, or AusDHBV and detected by Western blotting. (A) Diagram of the DHBV16 and AusDHBV pgRNAs. (B) Detection of pre-P in DHBV-infected duck livers by Western blotting. PBS, pBluescript (empty vector control); WT, wild type (DHBV C⁻).

We then examined the predicted pre-P region in 57 different avian hepadnavirus sequences in GenBank. There were three types of potential pre-P products (Fig. 5A). Type I could initiate translation from an AUG within four codons of the position of C13 in DHBV3, and S10 is absent, producing a 46- to 50-codon-long 5′ extension of the P ORF. Virus isolates from STHBV (5/5), HHBV (2/2), CHBV (3/3), sheldgoose HBV (2/2), RGHBV (1/4), and DHBV (33/36, including DHBV16 and AusDHBV) are type I. Type II could also initiate from at or near the position of the DHBV3 C13 AUG, but translation would be terminated by the S10 codon, producing a ∼29-aa-long pre-P peptide. DHBV3 and all SGHBV sequences (5/5) investigated belong to this type. Type III includes some of the RGHBV (3/4) and DHBV (2/36) isolates, and these genomes could produce one or more short peptides (≤14 aa) from the pre-P region. Therefore, 46 of 57 (80%) reported avihepadnavirus sequences have the genetic potential to produce a full-length pre-P protein, and 6 more could produce a substantial peptide from the pre-P region.

FIG. 5.

Genomic structure of the pre-P region in avian hepadnaviral sequences. (A) Diagrams showing the structures of the pre-P region (open box) in 57 different avian hepadnavirus sequences in GenBank. Pre-P was predicted to exist in 46 of the 57 isolates. (B) The pre-P region is under selective pressures. The genetic distances among the pre-P terminal protein domains and the spacer domains of P were compared, and the statistical significance was evaluated using one-way analysis of variance with a Games-Howell post hoc test. Pre-P, pre-P sequences from C13 to P1; TP-Core, overlap between the terminal protein domain of P and the C protein; Spacer-SAg, overlap between the spacer domain of P and the S-Ag.

We next analyzed the pre-P region for evidence of selective pressures that would be present if the region were functional. The mean pairwise genetic distance for the pre-P region was calculated for each of the 46 avian hepadnavirus sequences encoding an intact pre-P protein. The genetic distances among genes in this region were then compared to those in the terminal protein domain (which is under strong selection) and to the distances within the spacer domain of P (which is under minimal selective pressure). The analysis was limited to P sequences within the overlap between the P and C ORFs (aa 1 to 82) and between P and the L antigen (aa 212 to 382) because the pre-P region overlaps the C ORF, and hence, its genetic distances must be compared to those of sequences that are also translated in two reading frames. As expected, the mean genetic distance among the viruses in the spacer region was much higher than the mean distance in the terminal protein domain (Fig. 5B). The mean genetic distance in the pre-P region was intermediate between that of the terminal protein and spacer domains, with the majority of the isolates having distances significantly closer to the terminal protein domain than to the spacer domain. Therefore, the pre-P region appears to be under moderate purifying selective pressures.

Initiation at C13 occurs by ribosomal shunting.

Initiation of translation at the P1 AUG occurs by the discontinuous transfer of ribosomes from near the cap of the pgRNA to near the P1 AUG in a process called ribosomal shunting (40), so we next asked if translation from the C13 AUG also occurs by shunting. Two observations must be made to demonstrate shunting: (i) translation must be cap dependent, and (ii) ribosomes must not scan continuously from the cap to the target AUG. To test these points, we assessed the P and pre-P translation from pgRNA derivatives carrying insertions of a stem-loop (BamHI-SL) (ΔG = −69.2 kCal/mol) that is considerably more stable than is required to block scanning ribosomes (ΔG = −40 kCal/mol) (29). When this stem-loop is located near the 5′ cap of an mRNA, translation is dramatically reduced by interference with the assembly of the preinitiation complex or by blocking the initial scanning of the 40S ribosomal subunits. When the stem-loop is located further downstream, it blocks scanning of the 40S subunits. Therefore, we inserted the BamHI-SL within 10 nucleotides of the cap (P1loop⁻/SL_5′) and at the NsiI and EcoRI sites between the cap and C13 AUG (P1loop⁻/SL_NsiI and P1loop⁻/SL_EcoRI) (Fig. 6A). LMH cells were transfected with genomic expression constructs carrying these lesions, and P was detected in whole-cell lysates by Western blotting. Inserting the BamHI-SL near the cap (Fig. 6, lane 4) nearly eliminated the accumulation of all forms of P, indicating the cap dependence of translation from both C13 and P1. Inserting the BamHI-SL at the NsiI (Fig. 6, lane 6) or EcoRI (Fig. 6, lane 5) sites upstream of C13 had little effect on either P or pre-P synthesis, indicating that ribosomes do not scan over this region of the pgRNA during the translation of either P isoform. Northern blots revealed that the insertion of the BamHI-SL into the pgRNA did not cause the production of a novel mRNA in which the P AUGs were adjacent to the cap (data not shown). Therefore, pre-P is translated by ribosomal shunting.

FIG. 6.

Initiation at C13 occurs by ribosomal shunting. The BamHI-SL was inserted into the 5′ end of the pgRNA coding sequences or at the NsiI and EcoRI sites between C1 and P1 on the P1loop⁻ background, the genomic expression plasmids were transfected into LMH cells, and P levels were measured by Western analysis of lysates on day 1 posttransfection. (A) Diagram of BamHI-SL insertion sites on DHBV3 pgRNA. (B) Representative experiment. PBS, pBluescript (empty vector control); WT, wild type (DHBV C⁻).

Pre-P is not encapsidated.

We next asked whether pre-P was encapsidated into core particles. We isolated DHBV cores from DHBV16- and DHBV3-transfected LMH cells as well as from DHBV16-infected duck serum. The cores were permeabilized by a low-pH pulse (33) and treated with micrococcal nuclease to degrade the viral genome prior to the detection of P by Western blotting. P was detected in core particles, but pre-P was not (Fig. 7, lanes 4, 5, and 10), although pre-P could easily be detected in lysates from which the cores were derived (Fig. 7, lanes 3 and 9). Therefore, pre-P is not encapsidated into cores made by either LMH cells or naturally infected duck liver.

FIG. 7.

Pre-P is not encapsidated. Cell lysates and subviral cores were isolated from transfected LMH cells and liver lysates, and serum-derived virions were isolated from infected and noninfected ducks. Cores were permeabilized and treated with micrococcal nuclease prior to the detection of P by Western blotting. Lanes 4, 5, 10, and 11, DHBV cores; lanes 1 to 3 and 6 to 9, cell lysates. Dpol, P expression vector; CDNA 3.1, empty vector control; WT, wild type (DHBV C⁻); LMH, untransfected LMH cells.

To determine if we could detect the pre-P peptide that could be made by DHBV3, we transfected a genomic expression vector that contained an intact S10 stop codon plus an in-frame HA tag inserted 21 nt downstream of the C13 AUG (C⁻/HA) into LMH cells and attempted to detect the peptide by Western blotting and immunofluorescence using anti-HA antibody 3F10. No positive bands in the Western blot or positive fluorescence signals were detected from the C⁻/HA-transfected cells, while pre-P made by the S10⁻/HA mutant (C⁻/HA in which the S10 codon has been ablated) was easily detected by both methods (Fig. 3 and data not shown). Therefore, either the DHBV3 pre-P peptide does not accumulate in cells or technical limitations prevented its detection by standard methods.

Pre-P is N-glycosylated.

Initiation at C13 and C14 is predicted to produce 50-aa or 22-aa N-terminal extensions of P, which would be 5.5 kDa and 2.4 kDa larger than P, respectively. However, pre-P appears as a diffuse cluster of bands on Western blots instead of two discrete bands, and ablating C13 or C14 lightened the cluster, rather than deleting specific bands within it (Fig. 2C, lanes 5 to 7). Therefore, these isoforms appear to be posttranslationally modified and/or proteolytically trimmed. Because the size difference between pre-P and P is consistent with the shift produced by glycosylation, we evaluated the potential N-glycosylation of pre-P. LMH cells were transfected with DHBV C⁻ derivatives carrying the S10⁻, S10⁻/C14⁻, S10⁻/C13⁻, or S10⁻/C13⁻/C14⁻ mutations, and the cells were lysed 1 day posttransfection. The lysates were treated with PNGase F, an enzyme that cleaves high-mannose, hybrid, and complex oligosaccharides from N-linked glycoproteins, and with Endo H, which cleaves N-glycans from the high-mannose type present on glycoproteins in the endoplasmic reticulum. As shown in Fig. 8A, the pre-P bands in the S10⁻-, S10⁻/C14⁻-, and S10⁻/C13⁻-transfected cell lysates were converted to a single isoform whose size was similar to that of P after treatment with either Endo H or PNGase F. In contrast, no mobility changes were observed for P in the S10⁻/C13⁻/C14⁻-transfected cell lysates. These data indicate that intracellular pre-P is N-glycosylated.

FIG. 8.

Pre-P is N-glycosylated. Genomic expression plasmids carrying the DHBV derivatives were transfected into LMH cells. One day posttransfection, cell lysates were harvested and treated or mock treated with Endo H or PNGase F, and P and pre-P were detected by Western blotting. The mobilities of pre-P and P are indicated. (A) Pre-P is N-glycosylated. Odd-numbered lanes, mock treated; even-numbered lanes, Endo H or PNGase F treated, as indicated. (B) Deglycosylated pre-P and P are similar in size. Lanes 1 and 3, mock treated; lane 2, Endo H treated. The mobilities of pre-P initiating from C13, P initiating from P1, and N-terminally truncated P initiating from P2 are indicated. WT, wild type. (C) DHBV16 pre-P is glycosylated. LMH cells were transfected with the indicated genomes (lanes 1 to 4 and 7 to 12), or LMH D2 cells that have been stably transfected with DHBV16 were employed; DHBV16 is in lanes 1 to 6, and DHBV3 is in lanes 7 to 10. The even-numbered lanes were treated with PNGase F, and the odd-numbered lanes were mock treated. The mobility difference between pre-P and deglycosylated pre-P is smaller for DHBV16 than for DHBV3 because DHBV16 pre-P does not appear to be proteolytically trimmed. PBS, pBluescript. (D) Mapping the signal sequence. Odd-numbered lanes, mock treated; even-numbered lanes, Endo H or PNGase F treated, as indicated.

To more carefully characterize the size of pre-P following glycosidase treatment, we transfected cells with the mutant expression vector D1.5G-P1loop⁻/P1⁻, in which the S10, C14, and P1 codons have been ablated, resulting in the initiation of pre-P from the C13 AUG and a truncated version of P from the P2 AUG (Fig. 8B, lane 1) (40). Lysates from these transfected cells were treated with Endo H, and the P isoforms were detected by Western analysis. Glycosidase treatment of these extracts did not alter the mobility of P translated from P2, but it reduced the mobility of pre-P to a single band that migrated very close to the position of P (Fig. 8B, lane 2). This deglycosylated pre-P product is smaller than the 95-kDa mass predicted for the translation product from the C13 AUG, and hence, its more rapid mobility indicates that DHBV3 pre-P may be proteolytically trimmed.

To determine if pre-P made by a DHBV strain that naturally produces pre-P is also glycosylated, genomic expression constructs for DHBV16 (DHBV16 and DHBV16/S10⁺) were introduced into LMH cells. We also examined LMH-D2 cells, in which a dimer of the DHBV16 genome is stably integrated into the cellular genome of LMH cells. Lysates from these cells were treated with PNGase F, and P was detected by Western analysis. The mobility of DHBV16 pre-P was slightly increased (Fig. 8C, lanes 3 and 5), and the diffuse pre-P band became tighter following glycosidase treatment (Fig. 8C, lanes 4 and 6); hence, DHBV16 pre-P is also a glycoprotein. As expected, no pre-P was detected in DHBV16/S10⁺-transfected LMH cell lysates because this mutant genome carries the S10 stop codon (Fig. 8C, lane 1). However, there were two differences in pre-P derived from DHBV3 and that derived from DHBV16. First, DHBV16 pre-P accumulated to lower levels, in part due to a lower level of expression from a cytomegalovirus-driven genomic construct than from the DHBV promoter-driven DHBV3 construct. Second, deglycosylated DHBV16 pre-P migrated in SDS-PAGE gels at its predicted mass (∼95 kDa) (Fig. 8C, lanes 4 and 6) rather than at the mass of P (90 kDa) (Fig. 8C, lane 8), implying that in contrast to the DHBV3 pre-P, the DHBV16 protein was apparently not proteolytically trimmed.

Proteins that become N-glycosylated must be translocated into the endoplasmic reticulum, a process mediated by a signal sequence which is usually located at the very N terminus. Signal sequences vary greatly in amino acid sequence, but a hydrophobic core region of ∼7 to 13 aa is the essential feature, and a basic “N domain” and a slightly polar “C domain” are also common (13). Sequences downstream of DHBV3 C14 must be able to act as a signal sequence because glycosylated pre-P isoforms are initiated from both the C13 and C14 AUGs. The sequence of pre-P starting at C14 is MLLTIFLDCVLGYQLLRDIEVEM(P)PQPLKQSLDQSKWLREA, in which seven continuous hydrophobic amino acids (MLLTIFL) are followed by two negatively charged amino acids (E) and a positively charged amino acid (K). Therefore, the C14 pre-P region has a predicted signal sequence.

The presence of this putative signal sequence was mapped by deleting nt 43 to 94 (downstream of C13) and nt 110 to 163 (downstream of C14) in the S10⁻/del43-94 and S10⁻/del110-163 mutants. C13 and C14 were intact, and S10 was ablated in both constructs. LMH cells were transfected with S10⁻, S10⁻/del43−94, and S10⁻/del110−163, and the cells were lysed in RIPA buffer 1 day posttransfection. The cell lysates were treated with Endo H and PNGase F or were mock treated, and pre-P was then detected by Western blotting. Larger pre-P bands were detected in the S10⁻ control and in the S10⁻/del43−94- and S10⁻/del110−163-transfected LMH cell lysates (Fig. 8D, lanes 1, 3, 5, 7, 9, and 11). After glycosidase treatment, the size of pre-P in S10⁻/del43-94 was reduced, while no changes in the mobility of the P isoform from S10⁻/del110-163 were observed (Fig. 8D, lanes 2, 4, 6, 8, 10, and 12). These data indicate that pre-P in S10⁻/del43-94 is N-glycosylated, while in S10⁻/del110-163, it is not. Therefore, sequences downstream of C14 encode a signal sequence that leads to the translocation of pre-P that was initiated from either C13 or C14.

Pre-P is secreted.

Finally, we asked whether pre-P was secreted from cells because glycosylated pre-P accumulated inside cells, and such proteins are often intermediates in the secretion process. We transfected genomic expression vectors for the S10⁻ and S10⁻/C13⁻/C14⁻ mutants into LMH cells and collected cell lysates and medium from the cultures 4 days posttransfection. The medium was clarified by centrifugation, and pre-P was then immunoprecipitated using an anti-P polyclonal antibody. The immunocomplexes were treated with Endo H and PNGase F or were mock treated, and pre-P was then detected by Western blotting. Pre-P was easily detected in the medium from cultures transfected with the S10⁻ construct, whereas P was not (Fig. 9A, lanes 3 and 11). As expected, pre-P was not detected in medium from cells transfected with S10⁻/C13⁻/C14⁻ (Fig. 9A, lane 7 and 15). Most of the pre-P from S10-transfected medium was not sensitive to Endo H (Fig. 9A, lanes 3 and 4). In contrast, PNGase F treatment reduced the mobility of medium-derived pre-P, producing a more tightly concentrated band that migrated near the mobility of P (Fig. 9A, lanes 11 and 12). Endo H cannot digest glycans that have been matured by trimming during passage through the Golgi apparatus, whereas PNGase F can digest both mature and immature glycoproteins. Therefore, medium-derived pre-P must be a mature glycoprotein, which is secreted to the medium after passing through the endoplasmic reticulum and Golgi, and it cannot be cytoplasmic pre-P released from damaged cells.

FIG. 9.

Pre-P is secreted. DHBV genomic expression plasmids were transfected into LMH cells, or stably transfected LMH D2 cells were employed. Cell lysates were harvested, the medium was collected, and P and pre-P were then immunoprecipitated. The lysates and medium-derived immunocomplexes were treated with Endo H or PNGase F or were mock treated, and P was detected by Western blotting. The mobilities of pre-P and P are indicated. Odd-numbered lanes, mock treated; even-numbered lanes, Endo H or PNGase F treated, as indicated. (A) DHBV3 pre-P is secreted. (B) DHBV16 pre-P is secreted. Pre-P was immunoprecipitated from medium from LMH cells, LMH D2 cells, or LMH cells transfected with the wild-type DHBV16 genome (lanes 2 to 4) and detected by Western analysis. A lysate from cells transfected with wild-type DHBV16 (lane 1) was employed as a mobility control for pre-P.

Finally, we asked whether pre-P made by DHBV16, which naturally lacks the S10 codon, was also secreted. Medium from stably transfected LMH D2 cells which constitutively produce infectious DHBV16 or from LMH cells transiently transfected with the DHBV16 genome was immunoprecipitated with polyclonal anti-P antibodies, and the immunoprecipitated P isoforms were detected by Western analysis. DHBV16 pre-P was found in the culture medium in both cases (Fig. 9B, lanes 3 and 4), and consequently, DBV16 secretes pre-P.

DISCUSSION

In this study, we report that the translation of the DHBV P ORF initiates at two upstream in-frame AUGs, C13 and C14, in addition to starting at the classical P1 AUG (Fig. 2 and 3). Initiation at the upstream AUGs produced a cluster of larger P isoforms, which we named “pre-P” because they are translated as N-terminal extensions of the P ORF, similar to the pre-C and pre-S extensions of the C and S ORFs. Pre-P was detectable in both DHBV16- and AusDHBV-infected duck liver and DHBV16-transfected tissue culture cells due to the absence of the S10 stop codon in these two DHBV strains, but it was undetectable in DHBV3 due to the S10 codon (Fig. 4). We previously reported that nonencapsidated P from DHBV16-infected liver tissue migrates as two major sets of bands in SDS-PAGE gels, whereas nonencapsidated P from DHBV3-transfected LMH cells migrates as only a single major band, equivalent to the lowest form observed in liver tissue (48). We initially interpreted the upper bands in liver as being due to the posttranslational modification of P; here, we found that they are derived from the use of different initiation sites in addition to posttranslational modification.

Our results reveal that all three avian hepadnaviral ORFs have a parallel structure in which short and long protein isoforms are made by translation from multiple in-frame AUGs (pre-C/C, pre-S/S, and pre-P/P). Translational initiation for both P and pre-P is done through ribosomal shunting because in both cases, translation was cap dependent, yet blocking scanning ribosomes internally in the message did not appreciably affect translation (Fig. 6). Therefore, the upstream AUGs in the DHBV genome can be accessed by either differential transcription or translation: pre-C/C and pre-S/S are translated from different mRNAs, while pre-P/P is translated from the same mRNA.

HBV P uses a different translational initiation mechanism than DHBV P. The majority of HBV P appears to be translated by reinitiation at the P1 AUG after translation termination of a minicistron that is initiated from the second AUG between the C and P ORFs (10, 16). The remaining P translation appears to be initiated via a combination of leaky scanning and backward scanning to the P AUG after translation termination of the C ORF. Therefore, ribosomes pass through the sequences upstream of the HBV P1 AUG, but like in DHBV3, there is a stop codon upstream of P1 AUG that terminates all upstream translation products in all HBV genotypes (genotypes A to H). Therefore, HBV cannot make a pre-P protein. This indicates that an additional difference exists in the types of proteins produced by the avian and mammalian hepadnaviruses. The mammalian viruses make C, e-Ag, L-Ag, M-Ag, S-Ag, X, and P, whereas the avian viruses make C, e-Ag, L-Ag, S-Ag, P, usually pre-P, and perhaps an X-like protein.

Intracellular pre-P migrated as a diffuse cluster of bands, and unexpectedly, these bands were sensitive to Endo H and PNGase F cleavage (Fig. 8A), indicating that pre-P is N-glycosylated. There are six potential glycosylation sites (Asn-X-Ser/Thr) in P but none within the pre-P region, and hence, the oligosaccharides must be attached to P itself. In support of this observation, deletion studies revealed that sequences downstream of C14 could function as a signal sequence. Even more surprising, pre-P could be immunoprecipitated from the medium of cells transfected with the DHBV genome, and medium-derived pre-P had a mature glycosylation pattern that was sensitive to PNGase F but not Endo H (Fig. 9). Therefore, pre-P is a secreted glycoprotein. To our knowledge, pre-P is the only product of a viral nucleic acid polymerase gene to be secreted from cells independently of virions.

The function of pre-P is unknown. Pre-P is not encapsidated, so it is not used for viral genomic replication. Furthermore, only 46 of 57 avian hepadnaviral sequences in GenBank could make pre-P (Fig. 5). It is not clear how many of the sequences that could not make pre-P represent viable viruses because few of these sequences have been tested for growth in vitro or in vivo (12, 32, 43). However, DHBV3, one of the viruses that cannot make pre-P, grows well in cultured cells and can infect ducks (26). When the growth characteristics of viruses within serum from ducks congenitally infected with AusDHBV (pre-P⁺) and DHBV3 (pre-P⁻) were compared, DHBV3-infected ducks had higher S-Ag concentrations (110 versus 50 μg/ml), and the two strains had similar levels of DNA content (3.1 × 10¹⁰ versus 2.3 × 10¹⁰ viral genome equivalents/ml) and similar infectivity titers (2.5 × 10⁹ versus 2.8 × 10⁹). The spread of DHBV3 and the spread of AusDHBV in the livers of a small group of ducks were compared in two different studies (18, 26). When inoculated with the same amount of virus-containing serum, DHBV3 spread in duck liver faster than AusDHBV. At day 5 postinfection (p.i.), DHBV3 S-Ag could be detected in 1.5% of cells, whereas only 0.02% of AusDHBV-infected hepatocytes were S-Ag positive. At day 7 p.i., >95% of DHBV3-infected cells were S-Ag positive, whereas it took until day 13 p.i. before AusDHBV-infected hepatocytes were >95% S-Ag positive. Finally, all 14-day-old ducks inoculated with high doses of DHBV3 became persistently infected, as was expected from results with AusDHBV (17, 26). Therefore, DHBV3 is clearly a viable strain, and hence, pre-P is not required for efficient DHBV growth. However, it is possible that pre-P could provide a selective advantage to DHBV under conditions of direct competition between pre-P⁺ and pre-P⁻ viruses or during conditions of unusual stress.

Although pre-P is not essential for viral replication, morphogenesis, or infectivity, several observations support a role for pre-P in DHBV biology when it is present. First, the ability to produce pre-P is conserved among 80% of available avian hepadnaviral sequences (Fig. 5), and this phylogenetic conservation implies a conserved function. Second, when the full pre-P protein can be made, it appears to be under purifying selective pressures due to its relatively low average genetic distance. Third, much more pre-P than P is made (∼5.5-fold in DHBV3/S10⁻), and both humoral and cellular immune responses to HBV and woodchuck hepatitis virus P develop (9, 24, 34, 51). Although no studies have examined anti-DHBV P immune responses, they are presumably present, so there is almost certainly an immunological cost from making pre-P. Fourth, intracellular DHBV P can modestly suppress mRNA accumulation, presumably creating a negative-feedback loop to limit viral antigen accumulation and replication (4), and intracellular pre-P may have a similar activity. Fifth, glycosylation can influence the range of antigenic peptides generated in the endosomal pathway for presentation by major histocompatibility complex class II (36), and hence, the sugars on pre-P may change the antigenic epitope repertoire of P and possibly help suppress the development of T-cell responses. Finally, secreted pre-P in the bloodstream may be picked up by antigen-presenting cells, where it could be a decoy antigen that distorts host immune responses. Therefore, we feel that although pre-P is not essential for replication in vitro or in vivo, it may provide a selective advantage to the virus under certain conditions. However, determining whether pre-P contributes to viral fitness in animals will require head-to-head competition experiments between isogenic pre-P⁺ and pre-P⁻ viruses in ducks.