| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
J. Biol. Chem., Vol. 279, Issue 24, 24965-24975, June 11, 2004
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
||
From the
Unité de Génétique Moléculaire des Virus Respiratoires, CNRS URA 1966, Institut Pasteur, 75724 Paris Cedex 15, France and ¶CNRS UMR 5086, IFR 128 BioSciences, Lyon-Gerland, Institut de Biologie et Chimie des Protéines, 69367 Lyon Cedex 07, France
Received for publication, February 2, 2004 , and in revised form, April 1, 2004.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
3000 amino acid residues. This precursor is co- and post-translationally cleaved by host cellular signal peptidases within the region comprising the structural proteins (C-p7) and by viral proteinases within the nonstructural region (NS2-NS5B) to release 10 major viral proteins (1, 2).
Although subgenomic HCV RNA replicons that undergo autonomous amplification in cell cultures have recently made it possible to study RNA replication at the molecular level (for review, see Ref. 3), the lack of an efficient, fully permissive cell culture system for HCV has severely hindered the study of virion morphogenesis as well as virus entry into susceptible cells. The capsid protein (C) and two envelope glycoproteins (E1 and E2) are believed to be components of the virion. Studies of these structural proteins have for the most part relied on heterologous expression systems, such as recombinant baculoviruses in insect cells (4), or recombinant vaccinia or Sindbis viruses (5) and Semliki Forest virus-derived vectors (6) in eukaryotic cells. The results of these studies support the hypothesis that the HCV glycoproteins are retained in the endoplasmic reticulum. More recently, however, the development of pseudo-typed lentivirus particles bearing HCV envelope glycoproteins has provided new ways of studying virus entry into susceptible cells and generated evidence arguing against the retention of all HCV glycoproteins in the endoplasmic reticulum (7–9). A small protein of 7 kDa, p7, has been identified in the HCV polyprotein between the E2 glycoprotein and the first nonstructural protein, NS2 (10). Recently, p7 has been suggested to be a membrane-spanning protein localized to the endoplasmic reticulum (11) and to function as an ion channel (12–14). Pestiviruses (e.g. bovine viral diarrhea virus), other members of the family Flaviviridae, also encode a hydrophobic p7 protein at the junction between the structural and nonstructural proteins in the polyprotein (15). However, the relevance and functional role of these p7 proteins in the virus life cycle remain largely unknown.
An aim of this study was to determine whether other members of the Flaviviridae family encode such a small hydrophobic protein with potential ion channel activity that may be critical to the virus life cycle. We have, therefore, examined whether a p7-like protein is expressed by GB virus B (GBV-B), the most closely related of all animal viruses to HCV. GBV-B was identified in 1995 in a tamarin (Saguinus species) that developed acute hepatitis after inoculation with serum from the 11th tamarin passage of the original, human GB inoculum (16, 17). GBV-B infection is typically self-limited and associated with acute hepatitis (18, 19), although persistent infection can lead to chronic hepatitis in tamarins (20). The natural host for GBV-B remains unknown since this virus has never been isolated directly from tamarins or humans (21, 22), but the fact that GBV-B appears to be incapable of replicating in chimpanzees argues against a human origin (23). GBV-B has been classified within the Flaviviridae family but not assigned to any genus. However, GBV-B shares 27–33% nucleotide sequence identity over the entire open reading frame with HCV (24–26) and has a very similar genome organization. In addition, the serine protease and helicase, NS3/4A (27–29), as well as the RNA-dependent RNA polymerase, NS5B (30), of these viruses share common enzymatic specificities. These molecular characteristics in addition to the fact that this virus replicates robustly within the liver of small, New World primates (tamarins, marmosets) as well as in the primary cultures of hepatocytes from these species make GBV-B an attractive surrogate model for HCV studies.
Despite its obvious relevance to HCV and other flaviviruses, relatively little is known about the structural proteins of GBV-B. To date, there have been no reports of the direct sequencing of GBV-B proteins. Thus, predictions of the sites of polyprotein processing and GBV-B protein boundaries have relied on amino acid sequence alignments with the HCV polyprotein and computer-based algorithms for host signal peptidase recognition within the N-terminal third of the GBV-B polyprotein that contains the putative viral structural proteins. Importantly, published predictions do not report the existence of a polypeptide equivalent to the HCV p7 protein. We, therefore, carried out a detailed characterization of the proteolytic processing of the GBV-B polyprotein segment containing the viral structural proteins. We identified the N termini of the mature GBV-B envelope proteins, E1 and E2, and the nonstructural protein, NS2, by amino acid sequence analysis. These studies convincingly demonstrate that the GBV-B E2 protein is much shorter than HCV E2. Consistently, we document the existence of a heretofore unrecognized protein (p13) located within the GBV-B polyprotein between E2 and NS2 and with a calculated molecular mass of 13.1 kDa and a topology involving four predicted transmembrane segments. The C-terminal part of this novel protein exhibits clear structural homology with the HCV p7 protein, but its substantially larger size suggests a significantly different structural organization and/or additional functions.
|
EXPERIMENTAL PROCEDURES |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
pTM/C-NS2-V5 was constructed from a derivative of pTM/C-NS2 containing an engineered, unique AgeI restriction site overlapping the last codon of NS2, upstream of the existing BamHI site (pTM/C-NS2-AgeI-FLAG). A pair of complementary 54-nucleotide-long, 5'-phosphorylated oligonucleotides containing two glycine codons followed by the V5 tag sequence and a stop codon and designed to create protruding ends compatible with an AgeI site at the 5' end and BamHI site at the 3' end were hybridized together and cloned into pTM/C-NS2-AgeI-FLAG at the AgeI and BamHI sites. The resulting plasmid, pTM/C-NS2-V5, thus encodes a GBV-B C-E1-E2-NS2 precursor that is C-terminally fused to the V5 tag. pTM/C-NS2-V5 was used as a template to introduce substitutions at nucleotides 2656 (A 
G) and 2666 (C 3 A) of GBV-B cDNA, altering selected codons specifying Ile (ATA) and Leu (CTG), respectively, to Met (ATG) within the NS2 coding sequence. For this purpose, a PCR-based fusion strategy was used to produce an FseI-HpaI fragment (nucleotides 2547–2749 of the GBV-B cDNA) containing the indicated mutations that was inserted into pTM/C-NS2-V5 to generate pTM/C-NS2-V5-mut5,9.
To construct the plasmid pTM/C-E2701-V5, a pair of complementary, 5'-phosphorylated oligonucleotides containing two glycine codons followed by the V5 tag sequence and a stop codon and designed to create protruding ends compatible with an FseI site at the 5' end and a BamHI site at the 3' end were hybridized together and cloned into pTM/C-NS2 at the FseI (nucleotide 2547 of GBV-B cDNA) and BamHI sites. This resulted in a cDNA encoding GBV-B amino acid residues 1–701 fused in-frame at the C terminus to the V5 tag. Plasmid pTM/C-E2673-V5 was similarly constructed by inserting the same oligonucleotide pair into pTM/C-E2673-FseI-FLAG, which contains an engineered, unique FseI restriction site immediately downstream of nucleotide 2463 of GBV-B cDNA followed by a peptide sequence and a BamHI site. The resulting plasmid, thus, encodes GBV-B amino acid residues 1–673 fused at the C terminus to the V5 tag. The sequence of amplified fragments was verified in all plasmids.
Plasmids pTM/C-E2701-VP1 and pTM/C-E2673-VP1 were designed to encode GBV-B amino acid residues 1–673 or 1–701 fused at the C terminus to a truncated form of the poliovirus VP1 capsid protein for reporter purposes. To construct these plasmids, a cDNA segment representing nucleotides 2480–3241 of poliovirus was amplified by PCR using pTM/*2A2B* as a template (plasmid described in Ref. 32) and primers hybridizing to 5' and 3' VP1 sequences. The 5' primer also contained a 5' extension consisting of an engineered FseI site followed by a Gly-Ser-codon hinge, and the 3' primer contained a 3' extension consisting of a stop codon followed by a BamHI site. After restriction with FseI and BamHI, this PCR fragment was inserted into plasmids C-E2701-V5 and C-E2673-V5, respectively, at the corresponding sites.
Cell Culture and Viruses—Human 143B thymidine kinase-deficient cells were used for the isolation of recombinant vaccinia viruses and monkey kidney cells (CV1) for propagation of these viruses as well as for expression experiments. These cell lines were maintained in Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 5% fetal calf serum (Invitrogen), penicillin (50 units/ml), and streptomycin (50 µg/ml). vTF7–3, a recombinant vaccinia virus expressing T7 DNA-dependent RNA polymerase (33), was kindly provided by B. Moss (National Institutes of Health, Bethesda, MD).
Generation of Recombinant Vaccinia Viruses and Expression Assays in CV1 Cells—Recombinant vaccinia viruses vv-C-E2701-V5 and vv-C-E2673-V5 were generated by homologous recombination in CV1 cells infected with the temperature-sensitive ts7 strain of vaccinia virus and co-transfected with pTM/C-E2701-V5 or pTM/C-E2673-V5, respectively, as well as wild-type Copenhagen vaccinia virus DNA, essentially as described previously (34). Recombinant viruses were plaque-purified twice on 143B thymidine kinase-deficient cells under selective conditions. Vaccinia virus vv-TM1 was similarly derived from plasmid pTM1, which contains no cDNA insert. Large scale stocks of recombinant vaccinia viruses were produced on CV1 cells.
Expression of GBV-B proteins was accomplished by co-infection of CV1 cells with either of the recombinant vaccinia viruses and vTF7–3, each at a multiplicity of infection of 10 plaque-forming units per cell. Cytoplasmic extracts were prepared at various times post-infection by lysis of cells on ice in 0.2 ml of 10 mM Tris-HCl, pH 7.5, 1 mM EDTA, 100 mM NaCl, and 1% Nonidet P-40 containing 10 µl of Protease Inhibitor Cocktail (Sigma). Samples were mixed with Laemmli buffer, and polypeptides were separated by SDS-PAGE, then transferred to a polyvinylidene difluoride (PVDF) membrane (Amersham Biosciences) and detected by immunoblotting with an anti-V5 monoclonal antibody diluted 1:10,000 (Invitrogen) using enhanced chemiluminescence as a visualization method (ECL Plus, Amersham Biosciences) as detailed previously (35).
In Vitro Transcription—pGBV-B/2 plasmid DNA was linearized at restriction sites as indicated in the text ("Results" section, Figs. 1 and 4). All pTM1-derived plasmid DNAs were linearized at the unique BamHI restriction site. Linearized DNAs were purified by phenolchloroform extractions then ethanol-precipitated before transcription with T7 RNA polymerase for 4 h at 37 °C using either the RiboMAX kit (Promega) or MEGAscript kit (Ambion). RNAs were phenol-chloroformextracted, ethanol-precipitated, and quantified by electrophoresis in 1% agarose gels.
|
|
-mercaptoethanol loading buffer containing 9 M urea when indicated in the text (in the "Results" section, Fig. 3) before separation of GBV-B polypeptides by SDS-PAGE.
|
Immunoprecipitations—For immunoprecipitation of in vitro translated polypeptides, 6 µl of the translation reaction was diluted into 0.5 ml of radioimmune precipitation assay buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA, 0.5% Nonidet P-40) containing 0.1% SDS and incubated overnight at 4 °C with protein A-Sepharose (Amersham Biosciences) and either 1 µg of anti-V5 monoclonal antibody (Invitrogen) or 0.5 µl of rabbit polyclonal anti-poliovirus VP1 antibodies (a gift of B. Blondel, Institut Pasteur, Paris). Immune complexes were eluted from protein A-Sepharose with Laemmli buffer and separated by SDS-PAGE.
N-terminal Amino Acid Sequence Analysis—RNA transcripts derived from BamHI-linearized recombinant pTM plasmids were translated in rabbit reticulocyte lysates in the presence of canine pancreatic microsomal membranes as described above but in preparative 30–50-µl reactions and in the presence of appropriate [3H]-labeled amino acids or [35S]Met (Amersham Biosciences) depending on the protein analyzed. The resulting processed polypeptides were separated by SDS-PAGE either directly after translation or after immunoprecipitation of the translated products with the anti-V5 antibody as described above in the case of pTM/C-NS2-V5-mut5,9. Radiolabeled proteins were transferred to a PVDF membrane, excised from the membrane after identification of their location by autoradiography (Eastman Kodak Co. BioMax MR films) and subjected to automated Edman degradation on an Applied Biosystem 494 liquid-phase sequencer. The isotope content of fractions collected after each cycle of degradation was determined by liquid scintillation counting.
Sequence Analyses and Structural Predictions—Amino acid sequence homologies between GBV-B p13 and HCV p7 and within p13 were searched with the FASTA homology search program (36) using all reported HCV sequences from the GenBankTM/EMBL databases (HCV data base, hepatitis.ibcp.fr). Multiple sequence alignments were carried out with the ClustalW program (37). Internal amino acid sequence similarities within p13 were searched using the Local Similarity Program SIM (38). Peptidase cleavage sites were predicted using the SignalP program (39). Various methods were combined for the prediction of transmembrane sequences: PHDhtm (40), TMHMM (41), DAS (42), and TopPred2 (43). Theoretical 
-helix models of the predicted transmembrane segments were constructed using SwissPdbViewer program (44), and their space-filling representation was generated with Rasmol 2.7 (45). These models were manually positioned in a phospholipid bilayer that was built from the coordinates of phospholipids reported in the 1BCC
[PDB]
entry of the Protein Data Bank (46).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
180 kDa was detected in the absence of membranes (lane 2) that is likely to correspond to the full-length precursor encoded by the RNA (C-NS4B
). This protein was not produced in a control translation reaction carried out in the absence of GBV-B RNA (Fig. 1B, lane 1). An additional product of 55 kDa was observed in the absence of membranes (Fig. 1B, lane 2). This is likely to be identical to an irrelevant product produced in other reactions (see lanes 1–3), although it cannot be ruled out that a co-migrating product resulted from aberrant initiation or premature termination of translation within the precursor sequence.
In the presence of membranes, the GBV-B precursor was fully processed, producing three polypeptides with apparent molecular masses of 31, 44, and 66 kDa, respectively (Fig. 1B, lane 3). The 66-kDa polypeptide was identified as NS3, whereas the 31- and 44-kDa proteins were assumed to represent the envelope glycoproteins, E1 and E2, respectively. Neither NS4A-4B nor the core protein, C, was expected to be visualized, since the former has a very small size (7.4 kDa) and the latter lacks internal methionine residues in its sequence. Several minor, diffuse bands with molecular masses in the range of 20–28 kDa were also observed (Fig. 1B, lane 3). These products may represent incompletely glycosylated forms of E1 or the NS2 protein, which has a predicted molecular mass of 23 kDa. However, NS2 could not be conclusively identified under those conditions. A comparison of lanes 2 and 3 clearly shows that the C-NS4B precursor was not proteolytically processed in the absence of membranes. This indicates that the GBV-B proteases contained within this segment of the polyprotein (putatively NS2–3, and NS3/4A) are unable to direct cleavage at the NS2/NS3 or NS3/4A sites in the absence of prior cleavage of the structural protein precursor by signal peptidases. This suggests a fundamental difference from HCV in which NS2-3- and NS3/4A-directed cleavages have been shown to occur under similar conditions (our data and Ref. 47).
The predicted GBV-B E1 sequence contains 192 amino acid residues (residues 157–348 of the polyprotein, Ref. 25) and 3 putative N-linked glycosylation sites. Although somewhat greater than expected, the observed molecular mass of this protein (31 kDa) is consistent with this size prediction and the use of all three predicted glycosylation sites. In contrast, with 384 predicted amino acids (residues 349–732 of the polyprotein, Ref. 25), E2 would be expected to exhibit a 42-kDa molecular mass in the absence of glycosylation. Because there are 6 potential N-linked glycosylation sites in this predicted sequence, the observed molecular mass of 44 kDa suggests that the actual E2 sequence might be shorter or that E2 might be only partially glycosylated. To test this hypothesis, we treated the in vitro translation products with peptide N-glycosidase F. After removal of N-linked sugars, the electrophoretic mobilities of E1 and E2 shifted to 22 and 31 kDa, respectively (Fig. 1C, compare lanes 1 and 2). These results are consistent with the predicted sequence for E1, but they suggest that E2 is actually much shorter than predicted.
Determination of N-terminal Sequences of GBV-B Envelope Proteins—To map the boundaries of both E1 and E2 proteins, we used a heterologous expression system in which viral polypeptides were translated in vitro after internal initiation on the EMCV IRES. cDNA sequences coding for GBV-B C-E1-E2-NS2 were cloned into the plasmid pTM1 downstream of the EMCV IRES and the T7 RNA polymerase promoter (see "Experimental Procedures"), resulting in plasmid pTM/C-NS2. Such a strategy was expected to give higher yields of polypeptides than when expression was driven from the GBV-B IRES while also facilitating future generation of the corresponding recombinant vaccinia viruses. For N-terminal sequencing of proteins, RNA transcribed from the BamHI-linearized pTM/C-NS2 cDNA was translated in rabbit reticulocyte lysates in the presence of canine pancreatic microsomal membranes and either of two different tritiated amino acids, selected with respect to their occurrence and position in the predicted 10 N-terminal amino acids of the protein of interest.
For the E1 protein, labeling was with [3H]Val and [3H]Thr. Processed, in vitro translated products were separated by SDS-PAGE and transferred to a PVDF membrane. Radiolabeled E1 proteins were excised from the membranes and subjected to 10 cycles of Edman degradation. The isotope content of fractions collected after each cycle was determined. [3H]Val was clearly recovered at the third cycle, and [3H]Thr was recovered at the fourth cycle, a profile that was reproduced in two separate experiments (Supplemental Fig. 1A). These data map the N terminus of E1 to the Ala residue located at position 157 of the GBV-B polyprotein. This is in good agreement with the SignalP program (39), which predicts the most likely cleavage in this region to be at residues 156/157.
A similar analysis of [3H]Pro- and [3H]Val-labeled E2 proteins revealed Pro residues at positions 2 and 6 and a Val residue at position 5 of the protein, with marked increases in isotope recovery in these fractions that were reproduced in separate experiments (Fig. 2). This succession of amino acids is consistent with the N terminus of E2 being located at the Asn residue at position 350 of the polyprotein. Therefore, these results map the E1/E2 junction to residues 349/350, 1 amino acid residue downstream of the previously predicted site (25). However, this 349/350 site also conforms to predicted signal peptidase cleavage requirements and was also identified as a likely cleavage site by the SignalP program.
|
Much like the HCV NS2 protein, the predicted sequence of GBV-B NS2 is mostly hydrophobic, with several putative transmembrane domains. To facilitate detection of this polypeptide, we constructed a pTM/C-NS2 derivative coding for a C-terminal fusion of a V5 tag to NS2, including three glycine residues as a flexible hinge between the protein domains (Fig. 3A; see "Experimental Procedures"). In vitro translations were carried out with or without the addition of microsomal membranes, and aliquots of the resulting products were immunoprecipitated with a monoclonal anti-V5 antibody. The resulting [35S]Met-labeled products were diluted in a loading buffer containing 9 M urea in an effort to avoid aggregation of the membrane-spanning polypeptide and separated by SDS-PAGE. A precursor polypeptide of 100 kDa, corresponding to the full-length C-NS2-V5 precursor, was observed in the absence of membranes (Fig. 3B, lane 1). Upon the addition of membranes, the precursor protein was processed into E1 and E2 as well as a polypeptide with an apparent molecular mass of 22 kDa (Fig. 3B, lane 2). This latter product was the only one to be subsequently immunoprecipitated with an anti-V5 monoclonal antibody (Fig. 3B, lane 3) and was, therefore, identified as NS2-V5. The slightly different electrophoretic mobility of this polypeptide in lane 2 as compared with lane 3 is likely due to differences in total protein loading and the presence of rabbit reticulocyte lysate proteins in the crude translation reaction. This result confirms that a cellular signal peptidase is responsible for the release of NS2, as is the case for HCV.
Preliminary data obtained by sequencing an in vitro translated, [3H]Leu-labeled NS2-V5 polypeptide supported the initial prediction for the NS2 N terminus (Phe733) but were not definitive due to very low amounts of product recovered from the membranes (data not shown). To enhance the radioactive labeling of the product, we substituted codons 737 (ATA, Ile) and 741 (CTG, Leu), putatively located at positions 5 and 9 of NS2 in pTM/C-NS2-V5, respectively, with ATG (Met) codons, thus generating the plasmid pTM/C-NS2-V5-mut5,9. These substitutions are conservative in terms of the hydrophobicity of the side chains; positions 5 and 9 were chosen in an effort to reduce the potential impact of substitutions on the efficiency of cleavage by cellular signal peptidase. The in vitro translated NS2-V5-mut5,9 polypeptide, synthesized in the presence of [35S]Met, was immunoprecipitated with anti-V5 antibodies and subjected to 10 cycles of Edman degradation. Radiolabeled Met residues were recovered in fractions collected after the fifth and, to a lesser extent, ninth cycles (Supplemental Fig. 1B), a result that was confirmed in a repeat experiment. These data demonstrated that the N-terminal residue of NS2-V5 is the Phe residue at position 733, a result that is in agreement with the SignalP prediction.
The E2 Glycoprotein Is Smaller Than Predicted—To map the approximate position of the C terminus of the envelope glycoprotein E2, we expressed proteins from pGBV-B/2 RNA transcripts that were terminated at various sites within the predicted E2-encoded sequence (Fig. 4A). RNAs transcribed in vitro from BsaAI-, EciI-, BssHII-, BstEII-, SmlI-, PvuII-, or FseI-linearized cDNA templates were subsequently translated in rabbit reticulocyte lysates in the presence of microsomal membranes and [35S]Met. These RNAs generated various C-terminal-truncated polypeptide precursors that we refer to with respect to the position of the C-terminal amino acid within the GBV-B polyprotein. To produce a marker for the mature E2 protein, we used an AflIII-linearized cDNA template that resulted in synthesis of a C-NS3 precursor spanning amino acid residues 1–991 of the polyprotein. All of these precursor polypeptides gave rise to the expected, fully processed E1 protein (Fig. 4B). A comparison of the electrophoretic mobilities of the glycosylated forms of E2 released from the various C-terminal-truncated precursors revealed that several products of various intensities and molecular masses, all much lower than those of full-length E2, were generated upon processing of the 1–531 (BsaAI) and 1–563 (EciI) precursors (Fig. 4B, compare lanes 1–2 with lane 8). These truncations were further shown to alter the E2 glycosylation pattern, although they do not remove any predicted glycosylation sites. In contrast, the 1–637 (SmlI), 1–678 (PvuII), and 1–699 (FseI) products, when processed, produced glycosylated polypeptides that co-migrated with mature E2 (Fig. 5B, compare lanes 5–7 with lane 8). The 1–610 (BssHII) and 1–614 (BstEII) precursors released glycosylated proteins that migrated slightly faster than the mature E2 (Fig. 4B, compare lanes 3–4 with lane 8).
|
Demonstration of the Existence of a GBV-B Polypeptide between E2 and NS2—Next, we focused on the processing of the viral polyprotein in the region between E2 and NS2. As indicated above, the SignalP program suggests two possible cleavage sites in the 610–732 amino acid region, one at positions 681/682 and a second one, less likely, at positions 613/614. We selected an expression strategy based on the use of recombinant pTM plasmids permitting both in vitro translation as well as recombinant vaccinia virus-driven expression in eukaryotic cells and involving C-terminal fusion of structural precursors to a V5 tag. This C-terminal tag was placed either at positions 673 or 701 within the polyprotein, each downstream with respect to one of the two putative cleavage sites at positions 613/614 and 681/682. Two plasmids, pTM/C-E2673-V5 and pTM/C-E2701-V5, were thus generated, providing expression of precursor proteins spanning amino acids 1–673 or 1–701 of the GBV-B polyprotein, respectively, each fused at its C terminus to a Gly-Gly-Gly hinge followed by the V5 peptide.
We expressed these V5 fusion proteins in eukaryotic cells using recombinant vaccinia viruses generated by homologous recombination between pTM/C-E2673-V5 and pTM/C-E2701-V5 DNAs and the vaccinia virus genome, generating vv-C-E2673-V5 or vv-C-E2701-V5, respectively (see "Experimental Procedures"; Fig. 5A). CV1 cells were co-infected with vTF7–3 (expressing T7 RNA polymerase) and either vv-C-E2673-V5 or vv-C-E2701-V5. V5-tagged proteins present in cytoplasmic extracts prepared at various times (7–17 h) post-infection were detected by immunoblotting with anti-V5 antibodies. A protein of 8.5 kDa (pX701-V5) was detected in vv-C-E2701-V5-infected cells as well as another minor species 1-kDa smaller (Fig. 5B, left panel). Protein expression was optimal at the times shown in this figure, 9 and 11 h post-infection. Two V5 fusion proteins (pX673-V5) were observed in vv-C-E2673-V5 infected cells, with apparent molecular masses of 6.5 and 7.5 kDa, respectively (Fig. 5B, right panel). With each precursor, the second, smaller polypeptide species never became predominant with increasing time and even was not observed at some later time points. Although the precise molecular mass of very hydrophobic proteins may be difficult to estimate and although the difference in mass that was apparent between the pX701-V5 and pX673-V5 products is somewhat less than expected, the size of the products shown in Fig. 5B is most consistent with cleavage occurring between residues 613 and 614. The co-existence of two forms of each resulting polypeptide might reflect the use of an alternate, non-predicted cleavage site or possible post-translational modification of the polypeptides (see "Discussion"). Taken together, the results of these experiments demonstrate the existence of a previously unrecognized polypeptide located within the polyprotein between E2 and NS2.
Determination of the N-terminal Sequence of the Newly Identified Polypeptide—The expression levels and small sizes of the pX673-V5 and pX701-V5 products did not permit sufficient recovery for N-terminal sequencing of metabolically labeled proteins. Thus, for sequencing purposes, we fused a larger reporter sequence at the C terminus of the C-E2701 and C-E2673 precursors, allowing the use of in vitro translation as an expression system. pTM/C-E2673-VP1 and pTM/C-E2701-VP1 were constructed by replacing the V5 tag sequence in pTM/C-E2673-V5 and pTM/C-E2701-V5, respectively, with a sequence encoding a truncated form of the poliovirus VP1 capsid protein (aa 1–254) as a reporter protein of 28 kDa preceded by a "Gly-Ser-Gly" flexible hinge (Fig. 6A).
|
40.5 and 42 kDa with pTM/C-E2673-VP1 and 42 and 43 kDa with pTM/C-E2701-VP1, respectively (Fig. 6B, lanes 1 and 3). Both the larger and smaller polypeptide species produced from each transcript were immunoprecipitated with anti-VP1 antibodies (Fig. 6B, lanes 2 and 4). These results are similar to those obtained with V5-tagged expression products with respect to the presence of a doublet and the differences in the electrophoretic mobilities of pX701-VP1 and pX673-VP1. Taken together, these data provide strong support for a cleavage event between residues 613 and 614. To confirm this, a [3H]Leu-labeled pX673-VP1 protein doublet was transferred to PVDF membranes and subjected together to N-terminal Edman degradation, as described above. Radiolabeled Leu residues were recovered at cycles 3 and 7 (Fig. 7A), compatible with the SignalP prediction of a cleavage occurring between Gly613 and Tyr614 residues of the polyprotein. We then focused on pX701-VP1, separately recovering the large and small [3H]Leu- or [3H]Pro-labeled proteins from the corresponding doublet and subjecting each separately to 5 cycles of Edman degradation. Recovery of a Pro residue in the second cycle and a Leu residue in the third cycle from each species clearly demonstrated that the two proteins in the pX701-VP1 doublet have identical N termini corresponding to Tyr614 in the polyprotein (Fig. 7B and data not shown). This N-terminal boundary was identical to that found in pX673-VP1. The data, thus, demonstrate that the doublet bands observed with both the pX701-VP1 and pX673-VP1 products do not reflect the existence of an alternative N-terminal cleavage site. Because they appear to be similar in sequence, the different electrophoretic mobilities of these doublet bands might arise from differences in their conformation or, more likely, a post-translational modification (see "Discussion").
|
36–37 kDa was also observed at a low abundance among the in vitro translated products from reactions programmed with pTM/C-E2701-VP1 RNA; these were immunoprecipitated by anti-VP1 antibodies (pY701-VP1, lanes 3–4, Fig. 6B). We suspect that they may arise from an alternative cleavage event occurring at the putative 681/682 site, but the very low level of expression of these polypeptides did not permit their direct amino acid sequencing. This putative alternative cleavage event could not be observed with V5-tagged expression products (Fig. 5), since it would have given rise to products that were too small to detect. To determine whether the C-terminal truncation of p13 at amino acid 701 (Fig. 6) might have biased putative usage of the 681/682 cleavage site, we also expressed the full-length p13 sequence (amino acids 614–732) as a fusion with poliovirus VP1. We found that the N terminus of the major VP1-reactive processing product was located at Tyr614 (data not shown), in clear agreement with the existence of p13 in GBV-B. Sequence Analysis and Structural Predictions of the Newly Identified p13 Protein—Examination of the p13 sequence revealed that it contains four long hydrophobic stretches. Regardless of the method used (PHDhtm, TMHMM, DAS, or Top-Pred2, see "Experimental Procedures"), these hydrophobic stretches were predicted to be transmembrane (TM) helices and were, thus, denoted TM1-TM4 (gray boxes, Fig. 8). Because the p13 protein is located between the structural and nonstructural proteins in the GBV-B polyprotein, it was expected to share potential homologies with the p7 protein of HCV (10), a 63-amino acid-long protein containing two transmembrane segments that is similarly positioned within the HCV polyprotein (11). Although much lengthier, a comparison of the p13 sequence with that of p7 showed that the two C-terminal transmembrane segments (TM3-TM4) of GBV-B p13 possess rather clear similarities with the two transmembrane segments (TM1-TM2) of HCV p7 (Fig. 8A). In addition, the p13 TM3 and TM4 segments are connected with a short, basic segment that includes several Arg residues, similar to the short sequence located between the TM1 and TM2 segments of p7 (Fig. 8A). The C-terminal part of p13 including TM4 displays structural features typical of signal peptides (Fig. 8B), indicating that it is likely to represent the signal for the reinitialization of translocation of the downstream NS2 protein. These predictions suggest that the topology of the C-terminal part of p13 is similar to that of TM1-TM2 in p7. Interestingly, we also found strong internal sequence similarities between TM4 and TM2 of GBV-B p13, with 42% amino acid residue identities (Fig. 8B). The segments that share strong similarities end at residues 681 and 732, respectively, which correspond to the predicted and confirmed signal peptidase cleavage sites. The TM2 segment also exhibits structural features that are characteristic of a signal peptide (Fig. 8B) and is, thus, likely to act as a signal for the reinitialization of translocation of the C-terminal part of p13. From these analyses a putative topology can be proposed for GBV-B p13 protein wherein the N and C termini are oriented toward the ER lumen, with the short segments that connect the four transmembrane helices oriented alternatively toward the cytosol and the ER lumen (Fig. 8C).
|
-helical molecular models for both TMs (Fig. 8D). Examination of the spatial arrangement of the side chains of the amino acids residues that are common to both the TM2 and TM4 helices revealed that most of them (residues are in gray in Fig. 8D) are localized on one side of each helix and form a twisted groove extending throughout the helix. The other identical Arg and Leu residues shared by TM2 and TM4 are located on the opposite face of these helices (Arg656, Leu660, and Leu670 in TM2; Arg707, Leu711, and Leu721 in TM4; residues are in black in Fig. 8D). These observations provide circumstantial evidence that TM2 and TM4 share a similar structure, raising the hypothesis that GBV-B p13 might be derived from a gene duplication. Two amino acid residues in TM4, His725 and Trp715 (Fig. 8D), may have a functional role in ion channel activity (see "Discussion"). The fact that these residues have no homologs in TM2 suggests that TM2 and TM4 may possess different functions. |
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
The C/E1 junction was found to match the initial prediction by Muerhoff et al. (25) at amino acid residues Gly156/Ala157 of the polyprotein. In contrast, we found that the E1/E2 junction is actually located one amino acid downstream of the initial prediction (25), at the dipeptide Gly349/Asn350. Therefore, with 193 residues, the size of GBV-B E1 protein is very close to that of the HCV E1 protein (192 residues in a genotype 1 strain of HCV, Ref. 48). In addition, our data suggest that all three putative N-linked glycosylation sites are most probably used (Fig. 1). This also brings the molecular mass of the GBV-B E1 glycoprotein close to that of HCV E1, in which 4 N-glycosylation sites are used (49).
In contrast, our analysis of the C-terminal domain of the GBV-B structural precursor revealed previously unreported features that diverge substantially from the original predictions. We demonstrated that the location of the junction between the structural proteins and the first nonstructural protein NS2 conforms to the prediction and that this junction (Ala732/Phe733) is cleaved by a cellular signal peptidase, as in HCV (Fig. 3). Unexpectedly, we found that E2 is considerably smaller than originally predicted, since we demonstrated the efficient cleavage of the polyprotein at Gly613/Tyr614 by N-terminal sequencing of the downstream cleavage product (Figs. 6, 7). GBV-B E2, thus, consists of only 264 residues, resulting in a clear size difference compared with HCV E2 that contains 363 residues (e.g. genotype 1, Refs. 10 and 48). In addition, whereas our data indicate that most if not all of the six putative N-linked glycosylation sites of GBV-B E2 are utilized (Figs. 1 and 4), the HCV E2 is considerably more heavily glycosylated with 11 putative N-glycosylation sites (HCV genotype 1a). Despite these differences, structure predictions indicate the presence of a putative transmembrane domain at the actual C terminus of GBV-B E2 which is similar to that observed in HCV (50).
Consistent with the demonstration that cleavage events take place at dipeptides Gly613/Tyr614 and Ala732/Phe733, we documented the existence of a previously unrecognized GBV-B protein, p13, composed of 119 amino acid residues spanning amino acid residues 614–732 of the GBV-B polyprotein and with a calculated molecular mass of 13.1 kDa. Whether C-terminally fused to a small 17-amino acid peptide tag or to a larger, 28-kDa reporter protein, the GBV-B 614–673 (pX673-V5, pX673-VP1) and 614–701 (pX701-V5, pX701-VP1) polypeptides migrated as protein doublets with apparent molecular masses differing by 1–1.5 kDa in SDS-PAGE (Figs. 5, 6). These two species of each fusion protein appear to have identical sequences, since they were immunoprecipitated by antibodies to the reporter peptide sequence fused at their C termini while at the same time containing identical N termini (Fig. 7). We suspect that they may represent different conformations of the protein that co-exist under the conditions of SDS-PAGE or, more likely, two forms of the same protein differing by a post-translational modification. Such a modification would be incomplete in either of the expression systems we used, rabbit reticulocyte lysates programmed for translation with synthetic RNA or eukaryotic cells infected with recombinant vaccinia viruses (Figs. 5, 6). However, we observed no kinetic relationship in the relative proportions of these two forms. One intriguing possibility is that there might be S-acylation of the Cys residues located within or at the connection between the putative transmembrane regions of the protein (at positions 648, 649, 653) in some molecules. Interestingly, such S-acylation is frequent in membrane-spanning proteins and has been found in the 6K protein of Alphavirus (51) and the M2 protein of influenza virus (52), two viral membrane proteins members of a group of small proteins known as viroporins (53). However, further experiments will be required to refute or confirm this hypothesis.
Given its position within the viral polyprotein, this newly discovered p13 protein was expected to share homologies with the p7 protein of HCV (10), a protein that has not been intensively investigated until recently. HCV p7 is composed of 63 amino acids and has been suggested to be a membrane-spanning protein located in the endoplasmic reticulum and containing two transmembrane helices (11). Like HCV p7, prediction analyses of the topology of p13 revealed that it is a polytopic membrane protein that is likely to have its N and C termini oriented toward the ER lumen environment (Fig. 8). In contrast, with 119 amino acids and four predicted transmembrane helices, GBV-B p13 appears to have a structurally different organization than that of HCV p7. We demonstrated, however, that the TM3-TM4 segments of GBV-B p13 share clear homologies with the two transmembrane segments of HCV p7 (Fig. 8). This suggests the possibility that the p13 segment spanning the TM3-TM4 domains might be functionally equivalent to the p7 protein of HCV. Such a hypothesis would be consistent with the prediction of a putative signal peptidase cleavage site at Ala681/Gln682, immediately upstream of TM3 within GBV-B p13. However, our experimental data indicate that such a cleavage occurs with very low efficiency, if at all, regardless of whether p13 is C-terminally fused to a short or a long tag sequence or whether it is expressed as a truncated product (see Fig. 5, 6) or a full-length product (data not shown). Moreover, we have no indication that the faint additional products that we observed in processing reactions (pY701-VP1, Fig. 6) have an N terminus located at residue Gln682. Regardless of whether GBV-B p13 is further processed into two proteins or remains intact, the functional role of the N-terminal half of p13 remains to be determined.
The homologies evident in the sequences and structures of the TM2 and TM4 transmembrane helices of p13 (Fig. 8) are remarkable. The nature of the residues, particularly those with short side chains (Gly, Ala, Pro), that are present within identical grooves in TM2 and TM4 suggests that the corresponding helical faces might be involved in specific helix-helix interactions with a common partner, either p13 itself or another viral or cellular protein that is essential for virus replication. Three recent publications suggest that the HCV p7 protein may function as an ion channel (12–14). For it to have such activity, HCV p7 would need to multimerize, and p7 was indeed observed to form putative hexameric ring structures in membranes (12). By analogy with the p7 protein, p13 oligomerization might occur through intra- and/or intermolecular interactions involving the clearly homologous grooves in the TM2 and TM4 domains. Typically, a trimer of p13 could theoretically reproduce the arrangement of transmembrane helices present in a hexamer of p7. However, the presence of the His725 and Trp715 residues in TM4, which may have a functional role in ion channel activity but have no homologs in TM2, suggests that the amino and carboxyl halves of p13 may possess different functions. Although additional experimental data will be needed to determine the oligomeric state and the three-dimensional structure of p13, our analyses strongly suggest that p13, at least through its C-terminal domain, is likely to share ion channel activity with HCV p7.
The p7 ion channel activity has been reported to be inhibited by amantadine (12), a drug that is used in the treatment of influenza A virus infections and known to act by inhibiting the ion channel activity of the M2 viral protein (54). Therefore, HCV p7 is presumed to be a member of the viroporin family. Our data with GBV-B strengthen the hypothesis that numerous members of the Flaviviridae family, including Pestiviruses (e.g. bovine viral diarrhea virus, Ref. 55), HCV, and GBV-B) encode such small membrane-spanning proteins. These proteins are, therefore, likely to have a key role in the Flaviviruses life cycle and may through their ion channel activity contribute to virion maturation, release, and/or entry into cells. In a reverse genetics study of bovine viral diarrhea virus, it was shown that a large in-frame deletion of p7 does not affect RNA replication but prevents the formation of infectious virions (15). The role of HCV p7 in the virus life cycle is unclear at present, but a genome-length HCV RNA in which the p7 sequence had been deleted was not infectious upon intrahepatic inoculation in a chimpanzee (56). A subsequent analysis of genome-length molecular clones of HCV containing chimeric intertypic p7 sequences revealed that there was a genotype-specific p7 requirement localized to the N- and/or C-terminal regions of this polypeptide, suggesting that it may interact with other proteins encoded by HCV in a genotype-specific manner (56). It would be interesting to determine whether part or all of the GBV-B p13 sequence could be replaced by HCV p7 sequence without ablating GBV-B infectivity in tamarins. In addition to providing information on the structure and function of this novel polypeptide, such a chimeric GBV-B/HCV virus would be a valuable tool to evaluate the potential of antiviral candidates targeting HCV p7 sequences in vivo in small primates. GBV-B, as a surrogate model, also offers invaluable opportunities to look at the role of these small membrane-spanning proteins in the viral life cycle in primary cultures of hepatocytes and in a small primate animal model.
|
FOOTNOTES |
|---|
The on-line version of this article (available at http://www.jbc.org) contains Supplemental Fig. 1.
Supported by a French Ministère de la Recherche fellowship.
|| To whom correspondence should be addressed: Unité de Génétique Moléculaire des Virus Respiratoires, CNRS URA 1966, Institut Pasteur, 25 rue du Dr. Roux, 75724 Paris Cedex 15, France. Tel.: 33-1-40-61-33-60; Fax: 33-1-40-61-30-45; E-mail: annettem{at}pasteur.fr.
1 The abbreviations used are: HCV, hepatitis C virus; GBV-B, GB virus B; EMCV, encephalomyocarditis virus; IRES, internal ribosome entry site; PVDF, polyvinylidene difluoride; TM, transmembrane.
|
ACKNOWLEDGMENTS |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
