Journal of General Virology (1992), 73, 139-148. Printedin Great Britain

139

Translational modulation in hepatitis B virus preS-S open reading frame expression A. Gallina, 1 A. De Koning, 1 F. Rossi, 1 R. Calogero, 2 R. Manservigi 3 and G. Milanesi 1. t lstituto di Genetica Biochimica ed Evoluzionistica, Consiglio Nazionale delle Ricerche, Via Abbiategrasso 207, 1-27100 Pavia, z SORIN Biomedica, Strada Crescentino, 1-13040 Saluggia ( VC) and 3Istituto di Microbiologia, Facolth di Medicina, Universith di Ferrara, Via Luigi Borsari 46, 1-44100 Ferrara, Italy

A series of hepatitis B virus open reading frame (ORF) preS-S variants, including mutants in which the relative

order of the in-frame start codons (AUG1, AUG2 and AUG3) and nearby sequences had been altered, was expressed both in vivo (in HepG2 hepatoblastoma cells) and in vitro (by T7 promoter-driven transcription followed by translation in a rabbit reticulocyte lysate). The ratio of the synthesis of the large, middle (M) and major (S) proteins or their chimeric counterparts was analysed to study the translational regulation of ORF expression. As expected on the basis of the ribosome scanningmodel, the AUG sequence context was found to be a prominent factor in determining the different translational behaviour of the two preS-S-specific mRNAs of 2-4 kb (predominantly translated from AUG1) and 2.1 kb (which includes AUG2 and/or AUG3 and can be translated from either). Results from

Introduction Hepatitis B virus (HBV) is an enveloped, partially double-stranded circular DNA virus which causes acute and chronic liver disease in humans (Tiollais et al., 1985; Ganem & Varmus, 1987; Blum et al., 1989). Its envelope (env) gene, the open reading frame .(ORF)preS-S (Valenzuela et al., 1979; Galibert et al., 1979; Pasek et al., 1979; Ono et al., 1983; Kobayashi & Koike, 1984), encodes a set of related glycoproteins which are expressed by initiation at three in-frame start codons, AUG1, AUG2 and AUG3. Initiation at AUG3 generates the major (S) protein (p24 and its glycosylated form, gp27); initiation at AUG2 adds 55 amino acids (the socalled preS2 region) to the N terminus, generating the middle (M) protein (p30 and its mono- and diglycosylated forms, gp33 and gp36, respectively); initiation at AUG1 adds a further 108 to 119 amino acid extension (the preS1 region) to the N terminus of the M protein, generating the large (L) protein (p39 and its monoglyco-

0001-0466 © 1992SGM

both experimental systems suggested that initiation at internal AUGs in the 2.4 kb RNA is possible. In experiments in vitro, p r e S S mutants bearing lesions in a region 5' to AUG2, which has been implicated in AUG2/AUG3 cis repression, showed no increase in the utilization of internal AUGs. In addition, the chimeric envelope polypeptides produced in transfected HepG2 cells in this study were informative with respect to preS-mediated endoplasmic retention: replacement of the preS2 N terminus with that from preS1 generated a chimeric M protein that was glycosylated within the putative preS 1 retention sequence ad was not secreted. Thus, the preS1 retention sequence most likely acts inside the lumen of endoplasmic reticulum and its function is insensitive to glycosylation. A similar element might be active at the N terminus of M protein.

sylated form, gp41) (Heermann et al., 1984) (Fig. la). The rates of synthesis of L protein, and M and S proteins appear to be regulated separately at the transcriptional level. A hepato-specific promoter, SPI, distally located upstream from ATG1 (Chang & Ting, 1989; Chang et al., 1989; Raney et al., 1990), directs transcription of a 2-4 kb mRNA species which encodes the complete ORF and is believed to be translated predominantly into L protein (Cheng et al., 1986; Standring et al., 1986). A proximal, simian virus 40 (SV40)-like promoter, SPII, located within the preS1 coding region (Malpiece et al., 1983), directs transcription of 2.1 kb mRNAs starting near (either upstream or downstream of) AUG2 (Cattaneo et al., 1983; Standring et al., 1984) (Fig. la). The 2.1 kb subspecies including AUG2 is believed to be translated into both M and S proteins (Cheng et al., 1986; Standring et al., 1986: McLachlan et al., 1987; Ou & Rutter, 1987) (Fig. 1a). The independent expression of L, and M and S proteins might reflect the functional specialization of the former. Indeed, L protein has been

140

A. Gallina and others

found to be necessary, like S and in contrast to M protein, for HBV virion assembly and secretion (Bruss & Ganem, 1991). Moreover, it is retained within the rough endoplasmic reticulum (RER) and is able to trans-inhibit the constitutive secretion of the shortest envelope forms (Cheng et al., 1986; Chisari et al., 1986: Persing et al., 1986; Standring et al., 1986, McLachlan et al., 1987; Ou & Rutter, 1987). A short amino acid stretch at the N terminus of the preS1 region is involved in this retention process (Kuroki et al., 1989). This property is not directly related to L protein function in virogenesis, because deletion of the L protein retention sequence does not hamper virion secretion (Bruss & Ganem, 1991). The transcriptional programme described is accompanied at the translational level by differential control of start codon usage. Both 2.4 kb mRNA and the 2.1 kb subspecies covering AUG2 offer ribosomes multiple start sites, but ribosomes apparently prefer AUG 1 in the 2.4 kb mRNA, whereas they are capable of initiating at high frequency also at AUG3 in AUG2containing 2.1 kb templates. This behaviour can be explained in terms of AUG sequence context, according to Kozak's 'modified scanning' model (Kozak, 1988a, 1989 a). The sequence around AUG 1 [(A/G)GCAUGG] closely resembles the consensus sequence for an efficiently recognized start site [(A/G)CCAUGG, with a purine at position -3 and guanine at position + 4 being of particular importance] (Kozak, 1987a). In contrast, AUG2 lies in a relatively 'weak' context (GCCAUGC), which could allow for 'leaky scanning' in favour of AUG3 [which is within a strong context, A(A/U)CAUGG]. On the other hand, a recent report (Masuda et al., 1990) suggested that a cis-acting sequence located upstream from AUG2 might prevent initiation at 2.4 kb mRNA internal AUG codons, in a manner independent of the context of AUG1. The present report deals with these translational aspects of ORF preS-S regulation. A collection of deleted or rearranged ORF preS-S variants was compared with wild-type (wt) variants for the relative usage of initiation codons in two distinct experimental systems, cell transfection with an episomal vector and in vitro translation. As a by-product of this study, observation of the secretion and glycosylation phenotypes of chimeric envelope proteins expressed in transfected cells allowed us to gain some insight into the endoplasmic retention phenomenon.

Methods Construction of plasmids. Vector pRPRSV has been described previously (Manservigi et al., 1990). It contains a defective BK virus genome which expresses large T antigen and enables the vector to replicate as a high copy number episome in human cells (Milanesi et al.,

1984). In its expression cassette, a unique BarnHI cloning site is preceded by the Rous sarcoma virus (RSV) long terminal repeat (LTR), which promotes transcription initiation (Stoltzfus, 1988), 44 nucleotides upstream from the cloning site. LTR-driven mRNAs thus bear a short heterologous 5' untranslated region (UTR), the first 34 residues of which correspond to nucleotides 1 to 34 of the RSV mRNA leader [hence excluding RSV 5' UTR minicistrons (Hensel et al., 1989)]. pRPRSV derivatives carrying various HBV genome fragments in the correct orientation for surface protein expression were constructed by standard cloning techniques (Sambrook et al., 1989) (Fig. 1 b, c). pRPAUG3 bears the 2 kb BamHI (map position 30)-BglII (map position 1986) fragment of HBV adw2 DNA (map position zero is the unique EcoRl site in HBV DNA) including only the S protein coding region. For construction of pRPAUG2.3, the 2 kb EcoRI (map position 0/3221)-Bglli (map position 1986) fragment from HBV adw2 DNA, the EcoRI extremity of which is five nucleotides downstream from AUG2, was ligated into the homologous sites of plasmid pKSV-10 (Pharmacia). The product was recut at the EcoRI site, at which a headto-head dimer of a synthetic BgllI-EcoRI adaptor reproducing the HBV sequence from 10 bases upstream of AUG2 to the EcoRI site TCCTCAGGCCATGCAGTGG ( 5' P - G A T C ~66A6TCC66TACGTCAccrrAA-~ 5") was inserted (plasmid pKSVpreSSlink). The complete preS2-S coding sequence was then extracted as a 2 kb BgllI fragment and cloned into pRPRSV to generate the construct of interest, pRPAUG1.2.3 ayw and pRPAUG1.2.3 adw2 contain, respectively, the 2-3 kb BgllI (map position 2839)-Bglli (map position 1986) HBV ayw genomic segment and the 2-3 kb AluI (map position 2847)-Bg1II (map position 1986) HBV adw2 genomic segment, spanning the entire preS-S coding region of either serotype. Plasmids pRPAUG2. 2. 3, pRPAUG1.3 ayw and pRPAUG1.3 adw2 express rearranged versions of ORF preS-S. In RPAUG2.2.3, the 23 bp BgllI-BamHI fragment from pKSVpreS-Slink, covering the AUG2 site and surrounding sequences, and the first 12 preS2 codons, replaces the 5' terminus of the preS1 coding region to the Sau3AI site (map position 2933) in the HBV adw2 genomic insert, but is otherwise identical to pRPAUG1.2.3 adw2. In pRPAUG1.3 ayw and pRPAUG1.3 adw2, the insert is an internal deletion mutant of ORF preS-S, generated by joining either the ayw BgllI (map position 2839)-BamHI (map position 2906) fragment or the adw2 AIM (map position 2847)-Sau3AI (map position 2933) fragment, spanning the A U G I codon and surrounding sequences and the first 18 to 25 preS1 codons, to the adw2 BamHI (map position 30)BgllI (map position 1986) fragment, thus recreating a continuous ORF in which most of the preS1 coding region, AUG2 and 11 preS2 codons downstream from the latter have been eliminated. (Constructs are designated ayw or adw2 owing to the Y-terminal preS1 codons, in which the serotypes differ in number and in the presence of an additional in-phase A U G in adw2. The preS2-S portion in both constructs was contributed by serotype adw2.) In all the plasmids described above, the viral sequences downstream from ORF preS-S constantly cover the HBV enhancer [enh (Shaul et al., 1985)], ORF X and the polyadenylation (pA) signal (Simonsen & Levinson, 1983). Preliminary experiments had shown that the inclusion of both enh and ORF X augmented LTR activity between two- and fivefold compared to that of constructs lacking either element or both, most probably by virtue of the trans-activating effect of the ORF X product on both enh and the LTR (Spandau & Lee, 1988). The same set of ORF preS-S variants was expressed in vitro by cloning them downstream from the T7 promoter into the plasmid BlueScribe (BS; Stratagene) polylinker (Fig. I b, c). In this case, inserts spanned HBV sequences downstream from ORF preS-S to the SphI (map position 1265) site. For construction of deletions within the preS 1 coding region, 5 gg of plasmid BSAUG1.2.3 adw2 was cleaved within the preS2 coding region with BamHI and then treated with nuclease Bal 31 (Bethesda Research Laboratories) for between 30 s and 5 min at 25 °C in the

Hepatitis B virus surface gene translation

appropriate buffer (20 mM-Tris-HCl pH 8.1, 12 mM-MgC12, 12 mMCaCI2, 200 mM-NaC1, 1 mM-NazEDTA).After phenol extraction and ethanol precipitation, DNA was resuspended in 25 ~tl Klenow buffer (50 mM-Tris-HC1pH 7.2, 10 mM-MgSO4, 0.1 mM-DTT, 0.5 mM each dNTP, 5 lxg/ml acetylated bovine serum albumin) and exposed to Klenow polymerase (2 units; Bethesda Research Laboratories) for 30 min at 22 °C, to blunt staggered DNA extremities. After another ethanol precipitation, HBV sequences downstream from the original BamHI site were resected from the plasmid by cutting DNA with SphI. Linear plasmid containing unidirectionally eroded preS1 coding sequences was finally separated by electrophoresis on a 0.7% low melting point agarose gel (Seaplaque; FMC) and ligated 'in gel' to the 1.2 kb BgllI SphI fragment from pKSVpreS-Slink, the Bglll 5'protruding terminus of which had been blunt-ended using mung bean nuclease. The ligation products bore ORF preS-S mutants with deletions starting 10 nucleotides upstream from AUG2 and extending into the preS1 coding region for various distances. After amplification in Escherichia coti DH5-~ cells, plasmids BSpreS1A8 and BSpreS1A18, bearing in-frame deletions of interest (Fig. 1c), were identified by DNA sequencing.

TransJectionof HepG2 cellsfor in vivo translation experiments. HepG2 cells (Knowles et al., 1980) were grown in 25 cm2 plastic culture flasks (Nunc) in MEM with 10% foetal calf serum (FCS). One day before transfection, they were plated at 2 x 105 cells/flask. Transfection was by the calcium phosphate method (Gorman, 1985), using 20 ~g of the indicated plasmid, for 12 h. After a 20 min shock with 10% dimethylsulphoxide, fresh MEM and 10% FCS were added and cells were incubated for 48 h at 37 °C in a 5% CO2 atmosphere. RNA analysis. Cellular RNA was harvested according to Chomczynski & Sacchi (1987). For Northern blot analysis, 20 ~tg RNA/transfectant was resolved through a 1.4% agarose-formaldehyde gel, blotted onto activated nylon membranes (GeneScreen plus; DuPont) and probed with the 1.4 kb BamHI (map position 30)-BamHI (map position 1403) fragment from HBV adw2 DNA, labelled to > 1 × 109 c.p.m.Atg by the random priming method. After autoradiography, blots were stripped of HBV probe and rehybridized with a probe for human dehydrofolate reductase (hDHFR) mRNA, which is expressed from an SV40 promoter in the pRPRSV vector (Manservigi et al., 1990). Autoradiographic bands from both hybridizations were quantified by densitometry; normalization for transfection efficiency was obtained by calculating the HBV/hDHFR mRNA signal ratio for each lane. Immune precipitations. Two days post-transfection, cell monolayers were washed four times with PBS and then incubated in 800 ~tl methionine-free MEM for 90 min at 37 °C, after which [35S]methionine (100 ~tCi, >1000 Ci/mmol; Amersham) was added. For pulse experiments, incubation was continued for another 3 h, then cells were harvested. For pulse-chase experiments, after 2 h of labelling, cells were washed again and incubated in non-radioactive MEM and 10% FCS for 18 h, then processed. To prepare cytoplasmic extracts, cells were washed in PBS, resuspended in 800 ~tl lysis buffer (10 mM-Tris-HC1 pH 8, 150 mMNaC1, 1% Triton X-100, 10 mM-NazEDTA), subjected to a double cycle of freezing (5 rain in liquid nitrogen) and thawing (5 rain at 37 °C) and then centrifuged in an Eppendorf minifuge to pellet cell debris. Cytoplasmic supernatants, in parallel with cleared culture media, were mixed with 10 ~tl of sheep anti-subviral HBV particles (HBSAg) IgGSepharose conjugate slurry (20 ~tg antibody per immune precipitation) and shaken overnight at 4 °C. Sepharose-antibod~surface protein immune complexes were washed four times in lysis buffer plus 0-2% Tween 20 and twice in 10 mM-Tris-HC1 pH 8, 150 mM-NaCI, and resuspended in 50 mM-Tris-HC1 pH 6-8. An equal volume of 2 x sample buffer (25 mM-Tris-HC1 pH 6.8, 2% SDS, 6% v/v 2mercaptoethanol, 0.02% bromophenol blue) was added and the

141

suspension was heated at 95 °C for 5 min. After discarding Sepharose beads, samples were electrophoresed on 12% polyacrylamide gels containing 0-1% SDS. Fixed gels were soaked in En3Hance (New England Nuclear), dried and exposed for 24 to 48 h to X-ray film before development. Protein bands were quantified by densitometry. In vitro transcription and translation of BS-ORF preS-S constructs. BlueScribe derivatives were linearized with HindIII, cutting into the plasmid polylinker just downstream from the HBV insert. Run-off transcription with phage T7 RNA polymerase (Promega) and translation of mTGpppG-capped synthetic mRNAs in rabbit reticulocyte lysates (Promega) were performed exactly as described by Kozak (1989b). Salt conditions in the translation mixture were adjusted to 2.2 mM-Mg2÷, 45 mM-KCI, 90 mM-potassium acetate to ensure that the results would be comparable to in vivo data (Kozak, 1990a).

Results T h e strategy adopted in this study is s u m m a r i z e d in Fig. 1. W e reasoned that to analyse the effects of sequence context and relative position on O R F p r e S - S A U G codon usage, a simple a p p r o a c h would be to change the relative order of the A U G codons. I n s p e c t i o n of the O R F p r e S - S nucleotide sequence revealed the presence of c o m p a t i b l e restriction sites, positioned equally with respect to the r e a d i n g frame, d o w n s t r e a m from both A U G 1 and A U G 2 (Fig. lb). These were exploited t o construct r e c o m b i n a n t v a r i a n t s of O R F p r e S - S (see Methods). I n v a r i a n t p R P A U G 2 . 2 . 3 (Fig. l c), a s e g m e n t starting u p s t r e a m of A U G 2 to codon 12 of the preS2 coding region (thus i n c l u d i n g all of the A U G 2 codon a n d s u r r o u n d i n g sequences) was fused in-phase to c o d o n 27 of the preS1 coding region (adw2), s u b s t i t u t i n g the (theoretically) ' w e a k ' A U G 2 for the ' s t r o n g ' A U G 1 c o d o n at the b e g i n n i n g of O R F preS-S. I n v a r i a n t s p R P A U G 1 . 3 ayw a n d p R P A U G 1 . 3 adw2 (Fig. 1 c), o n the other h a n d , deletion a r o u n d A U G 2 fused preS1 codons 19 or 26, respectively, in phase to preS2 codon 13, p u t t i n g A U G 1 ad A U G 3 in direct succession. C o m p a r i son of the two subtype versions is of interest owing to the presence of a n 11 codon extension in subtype adw2 w h i c h generates a serial d u p l i c a t i o n of A U G 1 a n d s u r r o u n d i n g sequences. T h e i n t e r n a l d u p l i c a t i o n of A U G 1 (herein referred to as A U G 1-bis) in subtype adw2 lies in a ' s t r o n g ' sequence context a n d has b e e n s h o w n to be utilized, at low efficiency, for i n i t i a t i o n of 2.4 kb m R N A . R e c o m b i n a n t v a r i a n t s were expected to synthesize, from their distal start codons, chimeric c o u n t e r p a r t s of n a t u r a l envelope proteins: a large c h i m e r a (c-L) b e a r i n g a d u p l i c a t i o n of preS2 N - t e r m i n a l a m i n o acids at its N t e r m i n u s in place of the preS1 N t e r m i n u s , t r a n s l a t e d from the p R P A U G 2 . 2 . 3 v a r i a n t ; a n d proteins o f the size of M p r o t e i n (c-M, ayw or adw2 versions) s u b s t i t u t i n g preS1 N - t e r m i n a l a m i n o acids for the N - t e r m i n a l tip o f M protein, t r a n s l a t e d from p R P A U G t . 3 variants.

A. Gallina and others

142

(a) AUG1

s~ I ~ I

SPII

AUG2

~ - - - I -~-

preSl

AUG3

pA

~"

I

I preS2 I

J

S 2-1 kb m R N A

(A).

2.4 kb m R N A

(A). N* I N* I N* I

N* I

(N) m-I

(b)

s M L

pA ,~I'~B d

IJgYSd

(c)

I~:B-v///,-,~

pA

pRPAUG3 BSAUG3

t(

pA pRPAUG2.3 BSAUG2.3 pA pRPAUG1.2.3 (adw2 & ayw) BSAUG1.2.3 (adw2 & ayw)

A - 186

pRPAUG2.2.3

=~

~l///~

..............................

I]]3333~r////f ~

t( /(

~

pA [

BSAUG2.2.3 pRPAUG1.3 (adw2 & ayw) BSAUG1.3 (adw2 & ayw)

Fig. 1. Plasmids encoding normal and chimeric HBV surface proteins. (a) ORF preS-S (above) subdivided into preS1, preS2 and S regions by in-phase AUG codons (angled arrows). The sequence implicated in repression of translation from AUG2 and AUG3 is depicted as a horizontal bar above the ORF. SPI and SPII promoters direct the transcription of 2.4 kb and 2.1 kb mRNAs, respectively (centre), both polyadenylated[(A),]just downstream from the HBV pA. ORF protein products, L, M and S (below), are translated from AUGI, AUG2 and AUG3, respectively. N*, asparagine residues bearing N-linked glycans; (N), normally unused glycosylationsite at the beginning of the L protein; m, myristate attached to L protein N-terminal glycine. (b) DNA segments exploited in this study and (c) expression constructs assembled from them. The precise location of the indicated restriction sites in the HBV genome is given in Methods. A, Alul; B, BamHI; Bg, BgllI; RI, EcoRI, S, Sau3AI, Sp, SphI; d and y indicate that the restriction site is specific for the adw2 or ayw serotype, respectively; the asterisk marks the position, with respect to the ORF, of the 5"end of the synthetic adaptor used to reconstruct the preS2 coding region and AUG2 context sequence upstream of the EcoRI site (see Methods); ~, RSV LTR; II, T7 promoter. The horizontal bars below the preS1 region of construct BSAUG1.2.3 indicate in-frame deletions within this region in plasmids BS preSIA8 (A-83/-11) and BSpreS1AI8 (A- 186/-11), which are otherwise identical to BSAUG1.2.3 adw2.

The translational behaviour of the re-assembled O R F preS-S variants was compared with that of variants p R P A U G 3 , p R P A U G 2 . 3 , p R P A U G 1 . 2 . 3 ayw and p R P A U G 1 . 2 . 3 adw2 (Fig. 1 c) spanning, respectively, the S region (subtype adw2), the preS2-S region (subtype adw2) and the entire O R F preS-S of either subtype adw2 or ayw. All of them could express only normal envelope proteins.

Translation of ORF preS-S variants transfected into HepG2 cells Variants were cloned into the expression cassette of vector p R P R S V (Manservigi et al., 1990), downstream from the RSV L T R (Fig. l b, c). Cloned sequences spanned the HBV genome from O R F preS-S through the HBV pA site (Simonsen & Levinson, 1983). Thus, L T R directed m R N A s were expected to share the following characteristics: (i) to span O R F preS-S variants completely, (ii) to bear a short heterologous 5' U T R upstream of HBV sequences and (iii) to terminate at HBV pA. Thus, the L T R was expected to drive the synthesis of m R N A s mimicking either the 2.1 kb m R N A (constructs p R P A U G 2 . 3 , p R P A U G 1 . 3 ayw, p R P A U G 1 . 3 adw2 and p R P A U G 3 ) or 2.4 kb m R N A (constructs pRPAUG1.2.3 ayw, p R P A U G 1 . 2 . 3 adw2 and p R P A U G 2 . 2. 3) in length, and regularly initiating upstream from the distal A U G , allowing direct evaluation of the relative use of A U G codons from the abundance of expressed proteins. A potential difficulty in our experiments was the presence in the latter group of constructs of HBV promoter SPII, embedded within the preS1 coding sequence (Fig. 1 a, b, c), which could enable transcription of some natural 2.1 kb m R N A , contributing to the synthesis of M and S proteins and interfering with the analysis of translational events. p R P R S V derivatives were transfected into H e p G 2 cells, a differentiated h u m a n hepatoblastoma cell line (Knowles et al., 1980) permissive for HBV replication (Sureau et al., 1986; Sells et al., 1988), offering a cellular milieu closely resembling that of hepatocytes. Total R N A was extracted 48 h later. Alternatively, cells were labelled with [35S]methionine and cell lysates prepared for immunoprecipitation. The structure of plasmid-directed m R N A s was analysed in Northern blots. All of the transfectants expressed m R N A s whose size was consistent with initiation from the L T R and polyadenylation at the H B V pA site (Fig. 2). That the cap site was at the + 1 nucleotide of the RSV m R N A leader was verified by R N a s e protection experiments (data not shown). In addition, the three constructs harbouring SPII produced natural 2.1 kb m R N A (Fig. 2, lanes 1 to 3). It was noteworthy that with these constructs both the overall

Hepatitis B virus surface gene translation (a)

143

(a) 1

2

3

4

1

2

3

4

5

6

7

8

4

5

6

7

8

~28S S

2-4

(b)

2-1

1

2

3

~18S

(~

Fig. 2. (a) Transcription in vivo of ORF preS-S mRNAs. Total RNA was extracted from HepG2 ceils transfected with different plasmids 48 h post-transfection and subjected to Northern analysis using an S RNA-specific probe. To the left, the approximate size (in kbp) of transcripts is shown; 2.4 kb mRNA bands detected in cells transfected with pRPAUG1.2.3 adw2 (lane 3), p R P A U G 1.2.3 ayw (lane 2) or p R P A U G 2 . 2 . 3 (lane 1) correspond to LTR-directed transcripts, as does the 2.1 kb transcript of pRPAUG2.3 (lane 4). In contrast, minor 2.1 kb transcripts in lanes 1 to 3 represent HBV SPII-directed mRNAs. To the right, the position of ribosomal RNAs is shown. Reprobing for hDHFR (b) enabled comparison of the amount of ORF preS-S-specific mRNAs between samples.

Fig. 3. Synthesis in vivo of surface protein variants. HepG2 cells were transfected with pRPAUG3, -2.3, -2.2.3, -1.2.3 ayw, -1.2.3 adw2, -1.3 adw2, -1.3 ayw or pRPRSV (lanes 1 to 8) and 48 h later were labelled with [3sS]methionine for 3 h. HBV surface polypeptides in cytoplasmic lysates (a) and in culture media (b) were immunoprecipitated with anti-HBsAg antibody and analysed by SDSPAGE and fluorography, as described in Methods. Lane 8 (pRPRSV) is a negative control of cells transfected with the vector lacking an HBV insert. The position of wt surface proteins is indicated by arrows to the left.

Table 1. Expression ratio* of HBV surface protein variants in transfected HepG2 cells Surface variant synthesized (%) Construct

level of synthesis and the ratio of recombinant 2.4 kb m R N A to natural 2-1 kb m R N A (approximately 5:1) were very similar. Not unexpectedly, the ratio of 2-1 kb mRNA subspecies initiating on either side of AUG2 was also constant (two upstream :one downsteam, as determined by RNase protection). Thus, constructs p R P A U G 1 . 2 . 3 ayw, p R P A U G 1 . 2 . 3 adw2 and pRPAUG2.2.'3 were equivalent transcriptionally. Cytoplasmic extracts and culture supernatants from pulse-labelled cells were immunoprecipitated with antisera to HBsAg. A series of polypeptide bands of the sizes expected for unglycosylated and glycosylated forms of L, M and S proteins or their chimeric counterparts was observed (Fig. 3) and quantified by densitometry (Table 1). Construct pRPAUG3 synthesized only S polypeptides, p24 and its monoglycosylated form gp27. Both

pRPAUG3 pRPAUG2.3 p R P A U G 1.2.3 adw2 pRPAUG 1.2.3 ayw pRPAUG2.2.3 pRPAUG1.3 adw2 pRPAUG1.3 ayw

L

M

S

65"5t 60.1 37-3:~ -

54.5 24.1 28.4 49.4 91-2t:~ 88.0~

100 45-5 10.4~: 11-5 13-3 8-8 12.0

* Ratios (percent) were calculated from densitometric profiles by summing peak areas of each of the S, M and L proteins or their chimeric counterparts (non-glycosylated plus glycosylated forms) and dividing the result by the cumulative lane area. t This value is the total of initiation events at A U G 1 and AUG 1-bis. Chimeric polypeptide.

were abundantly secreted into the culture medium during labelling. (Fig. 3a, b). p R P A U G 2 . 3 synthesized equivalent amounts of S (p24 and gp27) and M proteins (p30, and its mono- and

144

A. Gallina and others

diglycosylated forms, gp33 and gp36, the latter bearing the second N-linked oligosaccharide at asparagine 4 in the preS2 region) (Fig. 3a). The envelope forms were again released into the medium, but at an appreciably lower rate; only glycosylated forms of M protein (which showed the characteristic 34K to 36K smeared band resulting from Golgi oligosaccharide trimming) were secreted (Fig. 3b). Chimeric 2-1 kb m R N A s transcribed from pRPAUG1.3 ayw and p R P A U G 1 . 3 adw2 were translated predominately as c-M polypeptides, with minor amounts of p24 and gp27 being synthesized (Fig. 3 a). All envelope forms were absent from the medium (Fig. 3 b). Of note, in this case, was the presence in both c-M subtype variants of a diglycosylated form. [This interpretation of the slowest migrating band was supported by the observation that it shifted with the intermediate band to the fastest migrating c-M band after endoglycosidase F or H treatment. In experiments to be reported elsewhere (A. Gallina et al., unpublished results) we have now demonstrated by site-directed mutagenesis that the actual site of the second glycosylation is at asparagine 4 in the ayw subtype, or 15 in the adw2 subtype, within the recombinant preS1 N terminus of c-M.] Furthermore, adw2 c-M, besides the non-, and mono- and diglycosylated forms translated from AUG1 (expectedly with a slightly lower electrophoretic mobility than ayw forms), showed less abundant polypeptides initiated at AUG1bis (comigrating with ayw forms). Transcripts from p R P A U G 1.2.3 ayw, p R P A U G 1 . 2 . 3 adw2 and p R P A U G 2 . 2 . 3 expressed L, M and S polypeptides (Fig. 3a). No secretion was observed in any case (Fig. 3b). S protein forms were always the least abundant. L protein (p39 and gp41, glycosylated at the only S protein site) prevailed over M protein in transfectants expressing wt ORF preS-S; utilization of AUGl-bis in the adw2 variant was not apparent. In contrast, in p R P A U G 2 . 2 . 3 transfectants M protein was over-expressed with respect to chimeric L protein (chimeric p39 and gp41). (Interestingly, the preS2 N terminus, transplanted ahead of preS1 sequences, apparently failed to be glycosylated.) Control pulse-chase experiments (not shown) demonstrated that all envelope proteins, including the chimeric forms, were stable during labelling and after, so that distinct half-lives cannot be used to explain differences in their abundance. Thus, such differences should be attributable to translational effects. Our results suggest, in particular, that the distal A U G sequence context is always a determinant in deciding the synthesis ratio (Table 1). LTR-directed 2.1 kb transcripts, including AUG2 and surrounding sequences, and mimicking natural 2.1 kb mRNA, were translated as both M and S proteins, as previously reported. This was attributable to

(a)

(b) 1

2

3

4

5

6

7

12

L--

MS-

Fig. 4. Synthesis in vitro of surface protein variants. (a) T7 transcripts produced from BSAUG3, -2.3, -1.2.3 ayw, -1.2.3 adw2, -2.2.3, -1.3 ayw or -1.3 adw2 (lanes 1 to 7) were translated in vitro in rabbit reticulocyte lysates. The [35S]methionine-labelled polypeptides generated were analysed by S D S - P A G E and fluorography. (b) Effect of inframe deletions upstream from A U G 2 on the translation pattern of ORF preS-S in vitro. Plasmids BSpreS1A18 (lane 3) and BSpreS1A8 (lane 2), bearing an O R F preS-S insert with deletions of nucleotides - 1 8 6 / - 11 and - 8 3 / - 11, respectively, upstream from A U G 2 were processed as described for (a); for comparison, plasmid BSAUG 1.2.3 adw2 (lane 1), bearing the wt ORF, was analysed in parallel. M and S proteins bands, clearly observed on the original autoradiograph, are only faintly reproduced on this photograph. In (a) and (b) the position of wt surface proteins is shown to the left.

Table 2. Expression ratio* of HBV surface protein variants expressed in vitro Surface variant synthesized (~) Construct pRPAUG3 pRPAUG2.3 p R P A U G 1.2.3 adw2 p R P A U G 1 . 2 . 3 ayw pRPAUG2.2.3 p R P A U G 1 . 3 adw2 p R P A U G 1 . 3 ayw

L

M

S

62.41 70.3 36.2:~ -

54.2 17.6 15.9 30.6 85.41-~ 85.0~

100 45-8 20.0 13.8 33.2 14.6 14.0

* Ratios (percent) were calculated from densitometric profiles by summing peak areas of each of the S, M and L proteins or their chimeric counterparts and dividing the result by the cumulative lane area. t,:~ See footnotes to Table 1.

the 'weakness' of AUG2, as its substitution with the 'strong' AUG1 decreased initiation at AUG3. The context of AUG1 determined synthesis of L protein from 2.4 kb templates; its substitution with AUG2 and surrounding sequences derepressed the use of proximal AUGs (Table 1). This interpretation is not affected by overlapping transcription from an internal HBV promoter, because, owing to the homogeneous activity of LTR and SPII in all constructs, no variation can be attributed to transcriptional factors. Translation in vitro of ORF preS-S variants

To verify in vivo data, we resorted to a cell-free translation system, in which overlapping transcription

Hepatitis B virus surface gene translation

effects could be avoided. After cloning downstream from a phage T7 promoter, ORF preS-S variants (Fig. 1b, c) were transcribed in vitro and translated in a rabbit reticulocyte lysate (Fig. 4a), under buffer conditions known to reflect the in vivo initiation fidelity and context requirements (Kozak, 1989b, 1990a). The data, shown in Table 2, were in agreement with in vivo results. On templates mimicking 2.1 kb mRNA, AUG2 allowed for "leaky scanning' in favour of AUG3 in little more than half of the scanning events. By comparison, AUG 1 ayw and AUG 1 + AUG 1-bis captured more than 80% of initiation events. Similarly, on transcripts mimicking the 2.4 kb mRNA AUG arrangement, both AUG1 and AUG1 + AUG1bis captured most, but not all, initiation events (Table 2), which unambiguously proves, in this system, the imperfect dominance of the distal start codon and any possible cis-repressing element over internal AUGs. The possibility that cis-regulatory sequences lie upstream of AUG2 was addressed directly by analysing the translational performance of two ORF preS-S mutants bearing in-phase deletions within the preS1 coding sequence in vitro., Both A - 8 3 / - 11 and A - 186/- 11 (numbering from AUG2) mutants (plasmids BSpreS1A8 and BSpreSA18, respectively), deleting a 3' portion or all, respectively, of the putative - 102/-38 inhibitory sequence identified by Masuda et al. (1990), did not show an appreciable increase in internal AUG usage with respect to wt virus (Fig. 4b).

Discussion The mechanism whereby eukaryotic ribosomes choose a start codon on a capped mRNA molecule is currently accounted for by the 'modified scanning' model (Kozak, 1988a, 1989a), in which 40S ribosomal subunits bind to the 5' end of the mRNA and scan linearly toward the 3' end until they encounter an AUG within a sequence context acceptable for initiation (Kozak, 1987a). Aspects further influencing the use of a given AUG codon include its distance from the 5' cap (Johanson et al., 1984; Kozak, 1986, 1987b, 1988b; Sedman et al., 1990), mRNA secondary structure (Pavlakis et al., 1980; Pelletier & Sonenberg, 1985; Lawson et al., 1986; Kozak, 1988b, 1989b, c, 1990b; Fu et al., 1991) and the presence of cis-acting elements, which can increase or decrease translation of downstream regions in cooperation with trans-acting proteins (Hentzel et al., 1987; Aziz & Munro, 1987; Parkin et al., 1988; Rosen, 1991). The combination of these factors determines which of the potential cistrons On a translational template are actually expressed, and to what extent. Most cellular and viral mRNAs are translated from the AUG, in a good context,

145

closest to the 5' end. In a minority of transcripts both 5' and internal AUG(s) are exploited, with more than one protein being expressed (Kozak, 1988a). These alternatives coexist for HBV env gene expression strategy: 2.4 kb mRNA is translated mainly from the distal AUG (AUG 1), whereas in 2.1 kb mRNA, both the first and second AUG (AUG2 and AUG3, respectively) are utilized. The results presented here indicate that, in both cases, AUG sequence context is a determining factor for such behaviour. AUG1 context, in concordance with the consensus for a strong AUG, reduced initiation at downstream start codons not only when at its natural position, but also when transplanted directly upstream from AUG3. By contrast, the AUG2 'weak' sequence context, when substituted for AUG1 ahead of ORF preS-S, derepressed the use of downstream AUGs (Fig. 3 and 4; Tables 1 and 2). This effect of the context of the distal AUG on the overall expression ratio gives an experimental demonstration that the scanning process on 2.4 and 2.1 kb templates starts upstream from the first possible AUG, that is, at or near the 5' end, as predicted by Kozak's general model. Thus, our data render unlikely the possibility of internal entry of ribosomes into the preS coding region for expression from internal AUGs. This alternative mechanism of ribosome binding, well established for translation of uncapped picornavirus mRNAs (Jackson et al., 1990; Sonenberg, 1991), most likely acts in the expression of HBV ORF pol (Jean-Jean et al., 1989) and of the homologous gene of duck hepatitis B virus (Chang et al., 1990). Our results suggest, on the other hand, that the preclusion from using internal AUGs of the 2-4 kb mRNA was not as absolute in our experiments as previously reported for the Xenopus (Standring et al., 1986) and the vaccinia virus (Cheng et al., 1986) systems. Direct evidence came from in vitro experiments, in which internal initiation of transcripts spanning the entire ORF was observed, and approximately equal amounts of protein were translated from AUG2 and AUG3 (Fig. 4; Table 2). In HepG2 cells transfected with constructs pRPAUG1.2.3 adw2 and pRPAUG1.2.3 ayw, M protein was expressed to a relatively high extent, clearly prevailing over S protein (approximately 6 L : 3 M : 1 S) (Fig. 3; Table 1). These cells harboured both LTR-driven 2.4 kb mRNA, covering the entire ORF, and minor amounts of SPII-driven 2.1 kb mRNA, the larger (and more abundant) subspecies of which includes AUG2. Whether M protein is translated predominantly from natural transcripts (the translational efficiency of which, given the relative levels of transcription of 2.4 kb and 2.1 kb mRNA species, would then be postulated to be significantly higher), from LTR-driven transcripts (by internal initiation), or from both, one has to conclude

146

A. Gallina and others

that, in this case, AUG2 is capable of intercepting much more than 5 0 ~ of ribosome passages, contradicting previous observations (Cheng et al., 1986; Standring et al., 1986; McLachlan et al., 1987; Ou & Rutter, 1987) and our observations (this work) on recombinant 2.1 kb m R N A translation. Interestingly, in repeated transfections of 293 cells [a human embryonic kidney cell line expressing adenovirus E 1A/E 1B proteins (Graham et al., 1977)] with p R P A U G 1 . 2 . 3 adw2 and p R P A U G 1 . 2 . 3 ayw, even higher levels of M protein, close to those of L protein, were observed (A. Gallina, A. De Koning & G. Milanesi, unpublished results). Thus, a degree of modulation in the usage of initiation codons might exist in addition to the rigid effect of A U G sequence context, as revealed in cells differing in their tissue origin/differentiation state. Whether this plays a role in natural infection, in addition to transcriptional regulation, in adjusting the relative rate of envelope protein synthesis, can not be determined on the basis of our experiments. As a line of evidence, not only the L : S protein but also the M : S protein ratio, in the presence of ongoing viral replication, is markedly increased and is reflected in virion envelope composition (Tiollais et al., 1985). An example of a defined cis-regulating sequence which might participate in translational modulation has recently been reported by Masuda et al. (1990). Analysing the expression of ORF preS-S deletion variants in a hamster cell line, these authors identified a cis-repressing element, mapping in the region - 102 to - 38 nucleotides upstream from AUG2, which inhibits initiation at AUG2 and AUG3. Within this region, a potential hairpin structure of moderate stability is predicted. Data presented here suggest that, both in HepG2 cells and in rabbit reticulocyte lysates, the effect of such a sequence is relatively unimportant compared to the context of A U G 1 (Fig. 3). Moreover, two mutants bearing in-phase deletions within the preS1 region that damage or eliminate the putative cis-acting sequence were analysed by in vitro translation, and no significant increase in internal A U G utilization was observed in relation to the intact gene (Fig. 4b). A possible explanation of this discrepancy might be the absence from rabbit reticulocyte lysates of a factor which is essential in mediating the downstream repression effect in rodent cells. Alternatively, in hamster cells the 40S ribosomal subunits could be more sensitive to secondary structures in m R N A than in rabbit reticulocyte lysates. This issue deserves further investigation. Although primarily intended for the study of translational modulation, our experiments gave additional information about the mechanisms involved in HBV envelope protein RER retention. The ability of L protein to retain itself and the shorter envelope forms within a RER or early Golgi compartment, verified here in

HepG2 cells (Fig. 3), has been observed previously in various experimental systems, including transfected mammalian cells (Persing et al., 1986; McLachlan et al., 1987; Ou & Rutter, 1987), Xenopus oocytes (Persing et al., 1986; Standring et al., 1986) and transgenic mice (Chisari et al., 1986). A recent report from Kuroki et al. (1989) demonstrated that a specific sequence [endoplasmic retention sequence (ERS)], mapping between amino acids 5 and 19 of the L protein of subtype adw2 (corresponding to amino acids 1 to 8 of subtypes ayw and adw) is necessary and sufficient for RER retention. The non-secretory characteristics of the two c-M mutants described in the present work (Fig. 3) agree with those findings. These chimeric proteins, bearing amino acids 1 to 19 or 1 to 26 of the N terminus of L protein (subtype ayw or adw2, respectively) fused to amino acid 13 of the preS2 region, include the ERS. It is noteworthy that the N terminus of c-M proteins is N-glycosylated in both subtype variants at the asparagine of a target Asn-LeuSer sequence within the putative ERS (A. Gallina et al., unpublished results). This potential glycosylation site is normally unused in L protein (see Fig. l a), but constitutes an absolutely conserved sequence among all serotypes. As the attachment of an N-linked oligosaccharide follows translocation into the intracisternal space of the RER, our data suggest that (i) the ERS of cM proteins is transported into the lumen of the RER, where it mediates retention, and (ii) glycosylation at a position located centrally within ERS does not interfere with its function. These conclusions point to an interaction of ERS with a soluble or anchored component inside the RER, and counter the hypothesis that it can act as a cytoplasmic anchor. A puzzling aspect of our experiments was that the complex of surface proteins, including a chimeric L protein form devoid of the N-terminal ERS, synthesized by p A F A U G 2 . 2 . 3 transfectants also was retained within cells (Fig. 3). The idea that the preS2 N terminus replacing the natural preS1 N terminus introduces some gross misfolding in c-L impairing transportability, is contradicted by the absence of an absolute preclusion of c-L oligomerization and secretion, because high-level coexpression of S protein by cotransfection of p R P A U G 2 . 2 . 3 and pRPAUG3 allowed secretion of all envelope forms (data not shown). An attractive alternative is that the preS2 domain, similarly to preS1, harbours within amino acids 1 to 12 an ERS, possibly also active when transplanted to the N terminus of L protein. We note, in fact, that the most abundantly expressed envelope form in these transfectants was the M protein, with a ninefold molar excess of preS2 N termini compared to S protein. We are currently testing our hypothesis by constructing site-specific and deletion mutants of the preS2 coding region.

Hepatitis B virus surface gene translation

This work was supported in part by Progetto Finalizzato 'Biotecnologie e Biostrumentazione' and by Progetto Finalizzato 'Ingegneria Genetica' of Consiglio Nazionale delle Ricerche. A.G. was the recipient of a fellowship from Fondazione Camillo Golgi, Brescia. A D . K . was the recipient of a fellowship from the Commission of European Communities.

References AZlZ, N. & MUNRO, H. N. (1987). Iron regulates ferritin mRNA translation through a segment of its 5' untranslated region. Proceedings of the National Academy of Sciences, U.S.A. 84, 8478 8482. BLUM, H. E., GEROK, W. & VIYAS, G. N. (1989). The molecular biology of hepatitis B virus. Trends in Genetics 5, 154-158. BRUSS, V. & GANEM, D. (1991). The role of envelope proteins in hepatitis B virus assembly. Proceedings of the National Academy of Sciences, U.S.A. 88, 1059-1063. CATTANEO, R., WILL, H., HERNANDEZ, N. & SCHALLER, H. (1983). Signals regulating hepatitis B surface antigen transcription. Nature, London 305, 336-338. CHANG, H. K. & TING, L. P. (1989). The surface gene promoter of the human hepatitis B virus displays a preference for differentiated hepatocytes. Virology 170, 176-183. CHANG, H., WANG, B., YEn, C., WEI, C. & TING, L. (1989). A liverspecific nuclear factor interacts with the promoter region of the large surface protein gene of human hepatitis B virus. Molecular and Cellular Biology 9, 5189-5197. CHANG, L., GANEM, D. & VARMUS, H. E. (1990). Mechanism of translation of the hepadnaviral polymerase (P) gene. Proceedingsof the National Academy of Sciences, U.S.A. 87, 5158-5162. CHENG, K., SMITH, G. L. & MOSS, B. (1986). Hepatitis B virus large surface protein is not secreted but is immunogenic when selectively expressed by recombinant vaccinia virus. Journal of Virology 60, 337-344. CHISARI, F. V., FILIPPI, P., McLACHLAN, A., MILICH, D. R., RIGGS, M., LEE, S., PALMITER,R. D., PINKERT, C. A. & BRINSTER, R. L. (1986). Expression of hepatitis B virus large envelope polypeptide inhibits hepatitis B surface antigen secretion in transgenic mice. Journal of Virology 60, 880-887. CHOMCZYNSKI, P. & SACCm, N. (1987). Single-step method of RNA isolation by acid guanidinium thiocyanate-phenol-chloroform extraction. Analytical Biochemistry 162, 156-159. Fu, L., YE, R., BROWDER, L. W. & JOHNSTON, R. N. (1991). Translational potentiation of messenger RNA with secondary structure in Xenopus. Science 251, 807-810. GALIBERT, F., MANDART, E., FITOUSSI,F., TIOLLAIS, P. & CHARNAY,P. (1979). Nucleotide sequence of the hepatitis B virus genome (subtype ayw) cloned in E. coll. Nature, London 281, 646-650. GANEM, D. & VARMUS, H. E. (1987). The molecular biology of the hepatitis B virus. Annual Review of Biochemistry 56, 651-693. GORMAN, C. (1985). High efficiency gene transfer into mammalian cells. In DNA Cloning: A Practical Approach, vol. II, pp. 143-190. Edited by D. M. Glover. Oxford: IRL Press. GRAHAM, F. L., SMILEY, J., RUSSELL, W. C. & NAIRN, R. (1977). Characteristics of a human cell line transformed by DNA from human adenovirus type 5. Journal of General Virology 36, 59-72. HEERMANN, K. H., GOLDMANN, O., SCHWARTZ, W., SEYFFARTH, T., BAUMGARTEN,H. & GERLICH, W. H. (1984). Large surface proteins of hepatitis B virus containing the pre-s sequence. Journalof Virology 52, 396402. HENSEL, C. H., PETERSEN, R. B. & HACKETT, P. B. (1989). Effects of alterations in the leader sequence of Rous sarcoma virus RNA on initiation of translation. Journal of Virology 63, 4986-4990. HENTZEL, M. W., ROUAULT, T. A., CAUGHMAN, S. W., DANCIS, A., HARFORD, J. B. & KLAUSNER, R. D. (1987). A cis-acting element is necessary and sufficient for translational regulation of human ferritin expression in response to iron. Proceedings of the National Academy of Sciences, U.S.A. 84, 6730-6734.

147

JACKSON, R. J., HOWELL, M. T. & KAMINSKI, A. (1990). The novel mechanism of initiation of picornavirus RNA translation. Trends in Biochemical Sciences 15, 477-483. JEAN-JEAN, O., WEIMER, T., DE RECONDO, A., WILL, H. 8/. ROSSIGNOL, J. (1989). Internal entry of ribosomes and ribosomal scanning involved in hepatitis B virus P gene expression. Journal of Virology 63, 5451-5454. JOHANSON, H., SCHUMPERLI, D. & ROSENBERG, M. (1984). Affecting gene expression by altering the length and sequence of the 5' leader. Proceedings of the National Academy of Sciences, U.S.A. 81, 76987702. KNOWLES, B. B., HOWE, C. C. & ADEN, D. (1980). Human hepatocellular carcinoma cell lines secrete the major plasma proteins and hepatitis B surface antigen. Science 209, 497-499. KoBAYAsHI, M. & KOIKE, K. (1984). Complete nucleotide sequence of hepatitis B virus DNA of subtype adr and its conserved gene organization. Gene 30, 277-282. KOZAK, M. (1986). Influences of mRNA secondary structure on initiation by eukaryotic ribosomes. Proceedings of the National Academy of Sciences, U.S.A. 83, 2850-2854. KOZAK, M. (1987a). An analysis of 5"-noncoding sequences from 699 vetebrate messenger RNAs. Nucleic Acids Research 15, 8125-8148. KOZAK, M. (1987b). At least six nucleotides preceding the AUG initiator codon enhance translation in mammalian cells. Journal of Molecular Biology 196, 947-950. KOZAK, M. (1988a). A profusion of controls. Journal of Cell Biology 107, 1-7. KOZAK, M. (1988b). Leader length and secondary structure modulate mRNA function under conditions of stress. Molecular and Cellular Biology 8, 2737-2744. KOZAK, M. (1989a). The scanning model for translation: an update. Journal of Cell Biology 108, 229-241. KOZAK, M. (1989b). Context effects and inefficient initiation at nonA U G codons in eucaryotic cell-free translation systems. Molecular and Cellular Biology 9, 5073-5080. KOZAK, M. (1989c). Circumstances and mechanisms of inhibition of translation by secondary structure in eucaryotic mRNAs. Molecular and Cellular Biology 9, 5134-5142. KOZAK, M. (1990a). Evaluation of the fidelity of initiation of translation in reticulocyte lysates from commercial sources. Nucleic Acids Research 18, 2828. KOZAK, M. (1990b). Downstream secondary structure facilitates recognition of initiator codons by eukaryotic ribosomes. Proceedings of the National Academy of Sciences, U.S.A. 87, 8301-8305. KUROKI, K., RUSSNAK, R. & GANEM, D. (1989). Novel N-terminal amino acid sequence required for retention of a hepatitis B virus glycoprotein in the endoplasmic reticulum. Molecular and Cellular Biology 9, 4459-4466. LAWSON, T. G., RAY, B. K., DODDS, J. T., GRIFO, J. A., ABRAMSON, R. D., MERRICK, W. C., BETSCH,D. F., WEITH, H. L. & THACH, R. E. (1986). Influence of 5' proximal secondary structure on the translational efficiency of eukaryotic mRNAs and on their interaction with intiation factors. Journal of Biological Chemistry 261, 13979-13989. MCLACHLAN, A., MILICH, D. R., RANEY, A. K., RIGGS, M. G., HUGHES, J. L., SORGE, J. & CmsARI, F. V. (1987). Expression of hepatitis B virus surface and core antigens: influences of pre-S and precore sequences. Journal of Virology 51, 683-692. MALPIECE,Y., MICHEL, M. L., CARLONI, G., REVEL, M., TIOLLAIS,P. & WEISSENBACB, J. (1983). The gene S promoter of hepatitis B virus confers constitutive gene expression. Nucleic Acids Research 11, 4645-4654. MANSERVIGI, R., GROSSI, M. P., GUALANDRI, R., BALBONI, P. G., MARCHINI, A., ROTOLA, A., RIMESSI, P., DI LUCA, D., CASSAI,E. & BARBANTI-BRODANO, G. (1990). Protection from herpes simplex virus type 1 lethal and latent infections by secreted recombinant glycoprotein B constitutively expressed in human cells with a BK virus episomal vector. Journal of Virology 64, 431-436. MASUDA, M., YUASA, T. & YOSHIKURA, H. (1990). Effect of the preS1 RNA sequence on the efficiency of the hepatitis B virus preS2 and S protein translation. Virology 174, 320-324.

148

A . Gallina and others

MILANESI, G., BARBANTI-BRODANO,G., NEGRINI, M., LEE, D., CORALLINI, A., CAPUTO, A., GROSSI, M. P. & RICCIARDI, R. P. (1984). BK virus-plasmid expression vector that persists episomally in human cells and shuttles into Escherichia coll. Molecular and Cellular Biology 4, 1551 1560. ONO, Y., ONDA,H., SASADA,R., IGARASHI,K., SUGINO,Y. & NISHIOKA, K. (1983). The complete nucleotide sequences of the cloned hepatitis B virus DNA; subtype adr and adw. Nucleic Acids Research l l , 17471757. Ou, J. & ROarER, W. J. 0987). Regulation of the hepatitis B virus major surface antigen by the preS-1 protein. Journal of Virology 61, 782-786. PARKIN, N., DARVEAU,A., NICHOLSON,R. & SONENBERG,N. (1988). cis-Acting translational effects of the 5' noncoding region of c-myc mRNA. Molecular and Cellular Biology 8, 2875-2883. PASEK,M., GOTO,T., GILBERT,W., ZINK, B., SCHALLER,H., MACKAY, P., LEADBETTER,G. & MURRAY,K. (1979). Hepatitis B virus genes and their expression in E. coli. Nature, London 282, 575-579. PAVLAKIS,G. N., LOCKARD,R. E., VAMVAKOPOULOS,M., RIESER, L., RAJBHANDARY, U. L. & VOURNAKIS, J. M. (1980). Secondary structure of mouse and rabbit ct- and fl-globin mRNAs : differential accessibility of ct and fl initiator AUG codons towards nucleases. Cell 19, 91 102. PELLETIER, J. & SONENBERG, N. (1985). Insertion mutagenesis to increase secondary structure within the 5' noncoding region of a eukaryotic mRNA reduces translational efficiency. Cell40, 515-526. PERSING, D. H., VARMUS,H. E. & GANEM, D. (1986). Inhibition of secretion of hepatis B surface antigen by a related presurface polypeptide. Science 234, 1388-1391. RANEY,A. K., MILICH,D. R., EASTON,A. J. & MCLACHLAN,A. (1990). Differentiation-specific transcriptional regulation of the hepatitis B virus large surface antigen gene in human hepatoma cell lines. Journal of Virology 64, 2360-2368. ROSEN, C. A. (1991). Regulation of HIM gene expression by RNAprotein interactions. Trends in Genetics 7, 9-14. SAMBROOK, J., FRITSCH, E. F. & MANIATIS, T. (1989). Molecular Cloning: A Laboratory Manual. New York: Cold Spring Harbor Laboratory. SEDMAN,S. A., GELEMBIUK,G. W. & MERTZ, J. E. (1990). Translation

initiation at a downstream AUG occurs with increased efficiency when the upstream AUG is located very close to the 5' cap. Journal of Virology 64, 453-457. SELLS, M. A., ZELENT, A. Z., SHVARTSMAN,M. & Acs, G. (1988). Replicative intermediates of hepatitis B virus in HepG2 cells that produce infectious virions. Journal of Virology 62, 2836-2844. SHAUL,Y., RUTTER,W. J. & LAUB,O. (1985). A human hepatitis B viral enhancer element. EMBO Journal 4, 427-430. SIMONSEN,C. C. & LEVINSON,A. D. 0983). Analysis of processing and polyadenylation signals of the hepatitis B virus surface antigen gene by using simian virus 40-hepatitis B chimeric plasmids. Molecular and Cellular Biology 3, 2250-2258. SONENBERG, N. (1991). Picornavirus RNA translation continues to surprise. Trends in Genetics 7, 105-106. SPANDAU,n. t~ LEE, C. (1988). Trans-activation of viral enhancers by the hepatitis B virus X protein. Journal of Virology 62, 427-434. STANDRING, n. N., RUTTER, W. J., MARIUS, H. E. & GANEi, n. (1984). Transcription of the hepatitis B surface antigen gene in cultured murine cells initiates within the presurface region. Journal of Virology 50, 563-571. STANDRING,D. N., Ou, J. & REarER, W. J. (1986). Assembly of viral particles in Xenopus oocytes: pre-surface-antigens regulate secretion of the hepatitis B viral surface envelope particle. Proceedings of the National Academy of Sciences, U.S.A. 83, 9338-9342. STOLrZFUS, C. M. (1988). Synthesis and processing of avian sarcoma retrovirus RNA. Advances in Virus Research 35, 1-38. SUREAU, C., ROMET-LEMONNE,J., MULLINS, J. & ESSEX, M. (1986). Production of hepatitis B virus by a differentiated human hepatoma cell line after transfection with cloned circular HBV DNA. Cell 47, 37-47. TIOLLAIS,P., POURCEL,C. & DEJEAN,A. (1985). The hepatitis B virus. Nature, London 317, 489-495. VALENZUELA,P., GRAY, P., QUIROGA, M., ZALDIVAR,J., GOODMAN, H. M. & RUTTER,W. (1979). Nucleotide sequence of the gene coding for the major protein of hepatitis B virus surface antigen. Nature, London 280, 815 819. (Received 1 July 1991 ; Accepted 2 September 1991)