Journal of Virological Methods 103 (2002) 67–74

Overexpression and purification of the hepatitis B e antigenprecursor

Sebastien Laine a, Samia Salhi b, Jean-Michel Rossignol a,*a Laboratoire de Genetique des Virus, UPR 9053-CNRS, A�enue de la Terrasse, 91198 Gif sur Y�ette, France

b Unite de Biochimie Cellulaire, FRE 2219-CNRS, Uni�ersite Pierre et Marie Curie, 9 quai St Bernard,75252 Paris Cedex 05, France

Received 6 September 2001; received in revised form 10 January 2002; accepted 11 January 2002

Abstract

Circumstantial evidence suggests that the secreted hepatitis B virus (HBV) e antigen (HBeAg) and/or its 22 kDaprecursor (P22) have an essential role in the establishment of persistent infection. In order to identify cellular proteinsthat could interact with P22, large amounts of this protein are required to perform pull-down assays. A plasmid wasconstructed encoding a recombinant P22 with a Histidine-tag at its N-terminal extremity (P22r). The initial attemptsto overexpress P22r in a conventional Escherichia coli strain failed, most likely due to the presence of rare AGA/AGGcodon clusters in the 3� part of the gene. To overcome this difficulty, P22r was overexpressed in the Epicurian coliBL21-codonplus™ (DE3)-RIL strain, which possesses extra copies of the ArgU gene that encodes the tRNAAGA/AGG.In this strain, P22r was overexpressed successfully and then purified in milligram quantities by metal affinitychromatography on Ni2+-chelated His-Bind resin. The purified recombinant protein P22r was able to interact witha cellular protein (P32), which had previously been shown to co-immunoprecipitate with native P22, indicating thatat least some of the P22r molecules were folded correctly. © 2002 Elsevier Science B.V. All rights reserved.

Keywords: HBV; P22; HBeAg; Precore protein; Overexpression; E. coli epicurian

www.elsevier.com/locate/jviromet

1. Introduction

The multiplication cycle of the hepatitis B virus(HBV) is now well understood (Ganem, 1996).However, the function of the HBV e antigen(HBeAg), a protein found in the serum of patientssuffering from acute hepatitis (Hollinger, 1996),remains enigmatic. It derives from a 25 kDaprecursor (the precore protein), which is directed

to the secretory pathway by a 19 amino acid-longsignal sequence that is cleaved during transloca-tion into the lumen of the endoplasmic reticulum(Junker et al., 1987; Garcia et al., 1988). Theresulting protein (P22) has a molecular mass of 22kDa and is processed further in a post-endoplas-mic reticulum compartment by removal of a 34amino acid-long arginine-rich domain located atits C-terminus, leading to the mature and secretedHBeAg (Wang et al., 1991).

Although HBeAg is not required for HBV in-fection (Tong et al., 1990), circumstantial evidence

* Corresponding author. Tel./fax: +33-1-69-823847.E-mail address: jmrossi@gv.cnrs-gif.fr (J.-M. Rossignol).

0166-0934/02/$ - see front matter © 2002 Elsevier Science B.V. All rights reserved.

PII: S0166 -0934 (02 )00019 -8

S. Laine et al. / Journal of Virological Methods 103 (2002) 67–7468

(Ou, 1997 for review) suggests that P22 and/orHBeAg could play an important role in establish-ing persistent infection. To gain new insights intothe biological function of P22, the characterisa-tion of cellular proteins that interact with thisprotein is required and the availability of largeamounts of purified P22 is thus necessary to un-dertake this study. To avoid a possible matura-tion of P22 by a cellular protease, it was decidedto overexpress P22 in bacteria rather than in thebaculovirus system.

The first attempts to overexpress P22 in Es-cherichia coli all failed, as full-length P22 was notdetectable by SDS-polyacrylamide gel elec-trophoresis (SDS-PAGE) after Coomassie stain-ing. One explanation for this could be that, in thelast 34 amino acids, P22 contains nine arginineresidues encoded by AGA or AGG, codons thatare used rarely in E. coli (Wada et al., 1992).Indeed AGA and AGG codons are known toaffect negatively the translation rate of recombi-nant proteins in E. coli (Spanjaard et al., 1990;Kane, 1995). To overcome this problem, a new E.coli strain that possesses extra copies of ArgU,LeuY and IleW genes was used. Thus, amounts oftRNAAGA/AGA, tRNACTA and tRNAATA in theEpicurian coli BL21-codonplus™ (DE3)-RIL arehigher than in E. coli.

In this paper, the successful overexpression ofthe His-tagged P22 protein (P22r) in the Epicurianbacteria and its purification by a single chro-matography step are described. It is also demon-strated that at least some molecules of P22r arefolded correctly as it interacts with a 32 kDacellular protein (P32), the interaction betweenP22r and P32 being dependent on the spatialconformation of P22 (Salhi et al., 2001).

2. Materials and methods

2.1. Construction of plasmid pHP22

The DNA sequence corresponding to P22 wasamplified from plasmid pHPC by the polymerasechain reaction (PCR). Plasmid pHPC (formerlynamed pMLP-PC) has been described previously(Jean-Jean et al., 1989a). Primers (P22N1: CG-

GAATTCCATATGTCCAAGCTGTGCCTTG-GGTGGC and P22C1: CGGAATTCTTATG-AGTCCAAGGAATACTAAC) were derivedfrom the published HBV C-gene sequence (Galib-ert et al., 1982). The PCR mixes were heated at92 °C for 5 min, then the DNA polymeraseGoldstar (Eurogentec) was added. The PCR wascarried out in a thermal cycler, for 30 cycles, eachconsisting of 1 min at 92 °C, 1 min at 60 °C and1 min at 72 °C. At the end, an extra cycle of 1min at 92 °C and 10 min at 72 °C was per-formed. The amplified product was digested byappropriate enzymes and then inserted betweenthe NdeI and EcoRI sites of pET28a (Novagen).The resulting plasmid (pHP22) was sequenced bythe Sanger method.

2.2. Expression of P22r in Epicurian coliBL21-codonplus™ (DE3)-RIL

Plasmid pHP22 was used to transform Epicu-rian coli BL21-codonplus™ (DE3)-RIL competentcells (Stratagene). Bacteria were grown at 37 °Cin 100 ml of Luria Broth (LB) medium supple-mented with 10 �g/ml kanamycin and 40 �g/mlchloramphenicol, until an absorbance A600=0.6was reached. At this time, isopropyl-�-D-thiogalactopyranoside (IPTG) (Roche) was addedto the culture to a final concentration of 1 mM.Two hours after induction, bacteria were cen-trifuged at 5000×g for 15 min and resuspendedin 10 ml of 50 mM Tris–HCl pH 8.0 containing 2mM EDTA, 10 �g/ml lysozyme and 0.1% TritonX-100. The lysate was incubated for 15 min at37 °C, then briefly sonicated and centrifuged at12,000×g for 15 min. A 100 �l aliquot of thesupernatant was used to study the solubleproteins while the pellet (insoluble proteins) wasresuspended in 500 �l of Laemmli buffer(Laemmli, 1970). Proteins were then separated by12.5% SDS-PAGE and detected by Coomassieblue staining.

2.3. Purification of P22r

Epicurian coli BL21-codonplus™ (DE3)-RILcells were transformed with pHP22 and grown asdescribed in Section 2.2. Following IPTG induc-

S. Laine et al. / Journal of Virological Methods 103 (2002) 67–74 69

tion, the pellet from a 500 ml culture (�1.5 g ofbacteria) was resuspended in 50 ml of the Bug-Buster Protein Extraction reagent (Novagen),which contains a mixture of non-ionic detergentsand is thus capable of cell wall perforation. Bacte-ria were lysed following the manufacturer’s in-structions and nucleic acids were digested by agenetically engineered non-specific endonuclease(Benzonase from Novagen). The resulting extracthas a low viscosity and contains both soluble andinsoluble proteins. Inclusion bodies were collectedby centrifugation at 16,000×g for 20 min at 4 °Cand after complete dispersion of the pellet in theBugBuster reagent, lysozyme was added to a finalconcentration of 200 �g/ml. After 5 min incubationat room temperature, the suspension was cen-trifuged at 16,000×g for 15 min at 4 °C to collectthe inclusion bodies, which were then resuspendedin 50 ml of BugBuster reagent diluted 1:10. Theinclusion bodies were then collected by centrifuga-tion for 15 min at 16,000×g and resuspended in10 ml of loading buffer (50 mM Na2HPO4/NaH2PO4, pH 7.7, 300 mM NaCl, 8 M urea). Theremaining insoluble debris was then eliminated bycentrifugation for 30 min at 10,000×g.

The resulting supernatant was loaded, at a flowrate of 100 �l/min, onto a Sepharose CL-4Bcolumn (Amersham Pharmacia Biotech)precharged with Ni2+-chelated His-Bind resin(Qiagen). The column was washed with the washbuffer (50 mM Na2HPO4/NaH2PO4, 300 mMNaCl, 8 M urea, 20 mM imidazole) until the A280

returned to the baseline (usually about 20 ml ofbuffer). Proteins were then eluted with the elutionbuffer (50 mM Na2HPO4/NaH2PO4, 300 mMNaCl, 8 M urea, 500 mM imidazole). Urea wasremoved from the purified protein fractions bydialysis for 16 h at 4 °C against buffer 1 (50 mMNa2HPO4/NaH2PO4, 300 mM NaCl, 3 M urea, 5mM dithiothreitol) followed by dialysis againstbuffer 2 (50 mM Na2HPO4/NaH2PO4, 300 mMNaCl, 5 mM dithiothreitol) for 16 h at 4 °C. Thepurified protein was stored at −20 °C in buffer 2containing 10% (w/v) glycerol.

2.4. Cell labeling and immunoprecipitation

Simian COS-7 cells grown under standard con-

ditions in 10 mm dishes were transfected with 30�g of plasmid pHPC, using the calcium phosphateco-precipitation procedure (Graham and Van derEb, 1973). Forty-eight hours after transfection,cells were grown for 1 h in 10 ml of methionine-freecysteine-free Eagle’s minimal essential medium(ICN) followed by 3 h in 6 ml of Eagle’s minimalessential medium containing 400 �Ci of [35S]-me-thionine/[35S]-cysteine (Promix, Amersham Phar-macia Biotech). The cells were then harvested,centrifuged for 10 min at 1600×g and lysed inphosphate buffer saline (PBS)–NP40 buffer (137mM NaCl, 2 mM KCl, 10 mM Na2HPO4, 1.8 mMKH2PO4, pH 7.4, 2% NP40). The cell extract wascentrifuged for 15 min at 12,000×g to separate thecytoplasm from the nucleus and the membranedebris. An aliquot corresponding to 5×107 cpm ofcytoplasmic extract was immunoprecipitated withan anti-HBc (Dako) antiserum and analyzed aspreviously described (Carlier et al., 1994).

3. Results

3.1. Construction of plasmid pHP22 encoding theHis-tagged P22 protein (P22r)

The pET28a vector was used for the construc-tion of plasmid pHP22. This vector possesses anIPTG-inducible T7 promoter, a kanamycin resis-tance gene, a sequence coding for both a His-tag(six histidine residues) and a thrombin cleavagesite. The P22 coding sequence was cloned asdescribed in Section 2. As shown previously, theC-terminal region of P22 is important for theinteraction with cellular proteins (Messageot et al.,1998; Salhi et al., 2001), the P22 DNA was inserteddownstream of the His-tag coding sequence. Thus,the recombinant protein (P22r) contained 20 addi-tional amino acid residues at its N-terminus includ-ing a cluster of six histidine residues and athrombin cleavage site that could be used toremove the His-tag (Fig. 1).

3.2. O�erexpression of P22r

As a first step, the IPTG induction time givingthe best overexpression of P22r was determined.

S. Laine et al. / Journal of Virological Methods 103 (2002) 67–7470

Fig. 1. Nucleotide sequence of the 5� part of pHP22 and amino-acid sequence of the N-terminal part of P22r. T7 promoter and Lacoperator are represented by arrows as the Ribosome entry site (rbs) and the His-tag. The thrombine cleavage site is underlined. TheNdeI restriction site is upperlined and the amino acids from P22r are in bold.

To this end, Epicurian coli BL21-codonplus™(DE3)-RIL cells were transformed with plasmidpHP22 and an IPTG induction time-course wascarried out (not shown). The best overexpressionof pHP22 was obtained 2 h after the addition ofIPTG and consequently this induction time wasused in all subsequent experiments. After IPTGinduction, the cells were treated to separate theinsoluble and soluble proteins, as described inSection 2. The corresponding samples were ana-lyzed on SDS-PAGE (Fig. 2). Two polypeptidesin the range of the expected molecular mass ofP22r (24.5 kDa) were observed both in the insolu-ble and soluble fractions of transformed bacteria(Fig. 2, lanes 3 and 4). However, the smaller band(24 kDa) was also present in the insoluble andsoluble fractions of non-transformed bacteria(Fig. 2, lanes 1 and 2) and therefore cannotcorrespond to P22r. On the other hand, the 24.5kDa species was observed only in the pHP22transformed bacteria and most likely corre-sponded to P22r. This assumption was confirmedby a Western blot analysis of insoluble proteinsfrom pHP22 transformed bacteria, using an anti-HBc antibody, which is used routinely to im-munoprecipitate P22 (data not shown). As P22rwas found predominantly in the insoluble proteins(Fig. 2, compare lanes 3 and 4), inclusion bodiesfrom pHP22 transformed Epicurian coli BL21-codonplus™ (DE3)-RIL were used in furtherpurification experiments. Different buffers were

tested in order to solubilize P22r. In our experi-ence, the use of a buffer containing at least 8 Murea was required to solubilize the majority of theP22r.

Fig. 2. Overexpression of P22r in Epicurian coli BL21-codon-plus™ (DE3)-RIL. Non-transformed cells (− ) and pHP22transformed cells (+ ) were treated with 1 mM IPTG for 2 h.Insoluble (I) (lanes 2 and 4) and soluble (S) (lanes 1 and 3)proteins were separated as described in Section 2 and analyzedon 12.5% SDS-PAGE stained with Coomassie Blue. On theright are indicated migration of P22r and on the left, themolecular mass standards in kDa.

S. Laine et al. / Journal of Virological Methods 103 (2002) 67–74 71

Fig. 3. Chromatogram of the P22r purification. Each step of purification is presented. The loading step was applied with 0 mMimidazole and the unbounded proteins were collected (peak 1). The washing step was realized with 20 mM imidazole andnon-specifically bounded proteins were collected (peak 2). The elution was done with 500 mM imidazole and the eluted proteins werecollected (peak 3). The measurement of the absorbance in mAU was indicated on the left. On the right, is indicated theconcentration of imidazole used in mM. Elution volume was indicated in ml.

3.3. Purification of P22r

Purified inclusion bodies were loaded at flowrate of 100 �l/min onto Sepharose CL-4B chargedwith Ni2+ chelated His-Bind resin. The result of atypical chromatogram is shown in Fig. 3. Threepeaks of absorbance at 280 nm were detected andthe corresponding fractions were pooled. Peak 1corresponded to proteins that were not retainedon the column (flow-through fractions) and peak2 to proteins bound non-specifically (wash frac-tions). Specifically retained proteins (eluted frac-tions) were most likely present in peak 3. AnSDS-PAGE analysis of an aliquot of these differ-ent fractions was performed to confirm this as-sumption. As expected, the elution fraction (Fig.4, lane E) contained a substantial amount ofhighly purified P22r. Based on the 280 nm ab-

sorbance, the total recovery of P22r was deter-mined to be 3 mg/l culture of bacteria, while P22rpurity was estimated on SDS-PAGE to be over95%. However, a significant amount of P22r wasnot retained on the column (Fig. 5, lane FT) evenif a 50 �l/min flow rate for the loading step wasused instead of a 100 �l/min.

3.4. Interaction of P22r with a cellular protein

The goal of this work was to obtain largeamounts of correctly folded P22r to determine thecellular proteins interacting with P22 in the cell.To establish that the purified P22r was indeedfolded correctly, its interaction was establishedwith a cellular protein (P32) that was reportedrecently to interact specifically with P22 (Salhi etal., 2001). Thus, a cell extract containing 35S-la-

S. Laine et al. / Journal of Virological Methods 103 (2002) 67–7472

beled P32 was prepared as described in Section 2and mixed with about 50 �g of purified P22r.After addition of anti-HBc antibodies, which rec-ognize P22, immunoprecipitated proteins were an-alyzed by SDS-PAGE followed byautoradiography. As shown in Fig. 5, lane 2, a 32kDa protein was specifically observed when P22rwas added (compare lanes 1 and 2). To confirmthat the 32 kDa protein corresponded to P32, anextract from labeled cells expressing P22 was im-munoprecipitated with anti-HBc antibodies. Asshown in Fig. 5, lane 3, P32 was co-immunopre-cipitated with P22 as expected and migrated toexactly the same place on the gel as the 32 kDaspecies (lane 2).

This experiment showed that P32 co-immuno-precipitated in vitro, with the recombinant P22r.However, as we cannot compare the amounts of

Fig. 5. Co-immunoprecipitation of P32 with P22 or P22r.pHPC-transfected cells (+ ) (lane 3) and non-transfected cells(− ) (lanes 1 and 2) were metabolically labeled for 3 h. Afterlabeling, proteins from cell extracts were immunoprecipitatedwith anti-HBc antiserum in the presence (+ ) (lane 2) orabsence (− ) (lane 1) of the recombinant protein P22r. Theimmunoprecipitated proteins were separated on 12.5% SDS-PAGE and revealed by autoradiography. On the right areindicated migrations of the cellular P32 and viral P22 proteins.On the left are indicated migrations of molecular mass stan-dards in kDa.

Fig. 4. Analysis of proteins from pHP22 transformed Epicu-rian coli BL21-codonplus™ (DE3)-RIL after affinity chro-matography. Insoluble proteins from pHP22-transformedEpicurian coli BL21-codonplus™ (DE3)-RIL were loaded on aHis-Bind resin. The proteins that were not retained on thecolumn (flow through, FT), unspecifically bounded (wash, W)and specifically retained (eluate, E) were concentrated. Equalfractions were analyzed on 12.5% SDS-PAGE stained withCoomassie blue. On the right migration of P22r is indicated.Molecular mass standards in kDa are shown on the left.

P22r and P22 present in the two assays, (lanes 2and 3), we can conclude only that at least someP22r molecules are folded correctly and able tointeract with P32.

4. Discussion

In this paper, the overexpression and purifica-tion of the recombinant precursor of HbeAg aredescribed. The overexpression of P22r wasachieved by the use of the Epicurian coli BL21-codonplus™ (DE3)-RIL strain. To our knowl-edge, this is the first time that the HBV P22protein has been overexpressed successfully inbacteria and obtained in milligram quantities.

At the beginning of the study, it was speculatedthat our failure to overexpress P22 in E. coli wasdue to the presence in the C-terminal part of P22,of arginine codons that are used rarely in conven-tional E. coli strains. Although this hypothesis

S. Laine et al. / Journal of Virological Methods 103 (2002) 67–74 73

was not demonstrated fully in this paper, it issupported by two complementary results. Firstly,P22 could be overexpressed when the Epicurianstrain, which contains extra copies of ArgU, LeuYand IleW genes, was used. Secondly, the overex-pression of C-terminally truncated P22 (deleted ofits arginine-rich domain) can been achieved in aconventional E. coli strain (not shown). Further-more, the capsid protein (P21) of HBV, which hasexactly the same C-terminal domain as P22 (Jean-Jean et al., 1989b), has been overexpressed in E.coli only as a C-terminally truncated protein(Bottcher et al., 1997; Wynne et al., 1999; Wize-mann and Von Brunn, 1999). However, overex-pression of the full-length P21 using the Epicurianstrain was achieved recently (data not shown).Taken together these results confirm that the pres-ence of AGA/AGG codons is detrimental for theoverexpression of P21 and P22.

Thus, the use of this newly available bacterialstrain avoided the ‘rare codon’ effect on proteinexpression and consequently significantly in-creased the yield of P22r expression. A similarobservation has been reported for the expressionof peanut allergens that have an AGG/AGAcodon content of 8–10%. In the conventionalBL21 (DE3) E. coli, these proteins are synthesizedwith a low yield, whereas they are overexpressedin milligram quantities in the Epicurian coli BL21-codonplus™ (DE3)-RIL (Kleber-Janke andBecker, 2000).

The second difficulty that was encountered wasto resolubilize P22r. As mentioned above, a highconcentration of urea was required for P22r. Thisis in apparent contradiction with the result re-ported recently by Wizemann and Von Brunn(1999). These authors have overexpressed theHBV capsid protein in E. coli, which shares thesame amino acid sequence as P22 with the excep-tion of the first 10 amino acids. Despite thisstrong similarity between the two proteins, Wize-mann and Von Brunn (1999) were able to solubi-lize the HBV capsid protein with a 2 M ureasolution. It is possible that this difference betweenthe solubility of capsid protein and P22r was dueto their different spatial conformation. This hy-pothesis is supported by the fact that the capsidprotein does not interact with the cellular proteinP32 (Salhi et al., 2001) in contrast to P22.

As P22r was histidine tagged at the N-terminalend, its purification on a Ni2+-chelated His-Bindresin was relatively simple. As described, thepurification protocol consists of a single chro-matography step and was sufficient to prepareP22r with a high purity (�95%) in milligramquantities. However, the purified fraction contains8 M urea that must be removed before any assayof interaction. Consequently, a substantial frac-tion of P22r (estimated to 30–40%) was precipi-tated during the dialysis.

In addition, it was also shown that recombinantP22r, like native P22, is able to interact with thecellular protein P32. As P22 is known to bepresent in both the cytosol and the secretorypathway (Garcia et al., 1988; Ou et al., 1989;Guidotti et al., 1996) it may play an importantrole during HBV infection through interactionswith specific cellular proteins in either one or bothcompartments. Thus, the availability of this re-combinant P22 protein should be useful for theidentification of these novel cellular targets).

Acknowledgements

This work was supported by a grant from theAssociation pour la Recherche sur le Cancer. Wethank Fabienne Messageot and Rhoderick H. El-der for their helpful comments. We are grateful toMarie-Therese Bidoyen and Anne Thouard fortheir excellent technical assistance.

References

Bottcher, B., Wynne, S.A., Crowther, R.A., 1997. Determina-tion of the fold of the core protein of hepatitis B virus byelectron cryomiscroscopy. Nature 386, 88–91.

Carlier, D., Jean-Jean, O., Rossignol, J.M., 1994. Characteri-zation and biosynthesis of the Woodchuck hepatitis virus eantigen. J. Gen. Virol. 75, 171–175.

Galibert, F., Chen, T.N., Mandart, E., 1982. Nucleotide se-quence of a cloned Woodchuck hepatitis virus genome:comparison with the hepatitis B virus sequence. J. Virol.41, 51–65.

Ganem, D., 1996. Hepadnaviridae: the viruses and their repli-cation. In: Fields, B.N., Knipe, D.M., Howley, P.M.(Eds.), Virology, 3rd ed. Lippincott-Raven, Philadelphia,pp. 2703–2737.

S. Laine et al. / Journal of Virological Methods 103 (2002) 67–7474

Garcia, P.D., Ou, J.H., Rutter, W.J., Walter, P., 1988. Target-ing of the hepatitis B virus precore protein to the endoplas-mic reticulum membrane: after signal peptide cleavagetranslocation can be aborted and the product released intothe cytoplasm. J. Cell Biol. 106, 1093–1104.

Graham, F.L., Van der Eb, A.J., 1973. A new technique forthe assay of infectivity of human adenovirus 5 DNA.Virology 52, 456–467.

Guidotti, L.G., Matzke, B., Pasquinelli, C., Shoenberger,J.M., Rogler, C.E., Chisari, F.V., 1996. The hepatitis Bvirus (HBV) precore protein inhibits HBV replication intransgenic mice. J. Virol. 70, 7056–7061.

Hollinger, F.B., 1996. Hepatitis B virus. In: Fields, B.N.,Knipe, D.M., Howley, P.M. (Eds.), Virology, 3rd ed.Lippincott-Raven, Philadelphia, pp. 2738–2807.

Jean-Jean, O., Levrero, M., Will, H., Perricaudet, M., Rossig-nol, J.M., 1989a. Expression mechanism of the hepatitis Bvirus (HBV) C gene and biosynthesis of HBeAg antigen.Virology 170, 99–106.

Jean-Jean, O., Salhi, S., Carlier, D., Elie, C., De Recondo,A.-M., Rossignol, J.M., 1989b. Biosynthesis of hepatitis Bvirus e antigen: directed mutagenesis of the putative as-partyl protease site. J. Virol. 63, 5497–5500.

Junker, M., Galle, P., Schaller, H., 1987. Expression andreplication of the hepatitis B virus genome under foreignpromoter control. Nucl. Acid Res. 15, 10 117–10 132.

Kane, J.F., 1995. Effect of rare codon clusters on high-levelexpression of heterologous proteins in Escherichia coli.Curr. Opin. Biotechnol. 6, 494–500.

Kleber-Janke, T., Becker, W.M., 2000. Use of modifiedBL21(DE3) Escherichia coli cells for high-level expressionof recombinant peanut allergens affected by poor codonusage. Protein Expr. Purif. 19, 419–424.

Laemmli, U.K., 1970. Cleavage of structural protein duringthe assembly of the head of bacteriophage T4. Nature 227,680–685.

Messageot, F., Carlier, D., Rossignol, J.M., 1998. The Cterminus of the Hepatitis B virus e antigen precursor isrequired for a tunicamycin-sensitive step that promotesefficient secretion of the antigen. J. Biol. Chem. 273,18 594–18 598.

Ou, J.H., Yeh, C.T., Yen, T.S.B., 1989. Transport of hepatitisB virus precore protein into the nucleus after cleavage ofits signal peptide. J. Virol. 63, 5238–5243.

Ou, J.-H., 1997. Molecular biology of hepatitis B virus eantigen. J. Gastroen. Hepatol. 12 (Suppl.), S178–S187.

Salhi, S., Messageot, F., Carlier, D., Jean-Jean, O., Rossignol,J.M., 2001. Identification of a cellular protein specificallyinteracting with the precursor of the hepatitis B e antigen.J. Viral. Hepat. 8, 169–173.

Spanjaard, R.A., Chen, K., Walker, J.R., Van Duin, J., 1990.Frameshift suppression at tandem AGA and AGG codonsby cloned tRNA genes: assigning a codon to argU tRNAand T4 tRNAARG. Nucl. Acid Res. 18, 5031–5036.

Tong, S., Li, J., Vitvitski, L., Trepo, C., 1990. Active hepatitisB virus replication in the presence of anti-HBeAg is associ-ated with viral variants containing an inactive pre-C re-gion. Virology 176, 596–603.

Wada, K., Wada, Y., Ishibashi, F., Gojobori, T., Ikemura, T.,1992. Codon usage tabulated from the genebank geneticsequence data. Nucl. Acid Res. 20, 2111–2118.

Wang, J., Lee, A.S., Ou, J.H., 1991. Proteolytic conversion ofhepatitis B virus e antigen precursor to end product occursin a postendoplasmic reticulum compartment. J. Virol. 65,5080–5083.

Wynne, S.A., Crowther, R.A., Leslie, A.G., 1999. The crystalstructure of the human hepatitis B virus capsid. Mol. Cell3, 771–780.

Wizemann, H., Von Brunn, A., 1999. Purification of E. coli-expressed HIS-tagged hepatitis B core antigen by Ni2+-chelate affinity chromatography. J. Virol. Methods 77,189–197.