Abstract
Lassa virus glycoprotein 2 (LASV GP-2) belongs to the class I fusion protein family. Its N terminus contains two stretches of highly conserved hydrophobic amino acids (residues 260–266 and 276–298) that have been proposed as N-terminal or internal fusion peptide segments (N-FPS, I-FPS) by analogy with similar sequences of other viral glycoproteins or based on experimental data obtained with synthetic peptides, respectively. By using a pH-dependent, recombinant LASV glycoprotein mediated cell–cell fusion assay and a retroviral pseudotype infectivity assay, an alanine scan of all hydrophobic amino acids within both proposed FPSs was performed. Fusogenicity and infectivity were correlated, both requiring correct processing of the glycoprotein precursor. Most point mutations in either FPS accounted for reduced or abolished fusion or infection, respectively. Some mutations also had an effect on pre-fusion steps of virus entry, possibly by inducing structural changes in the glycoprotein. The data demonstrate that several amino acids from both hydrophobic regions of the N terminus, some of which (W264, G277, Y278 and L280) are 100 % conserved in all arenaviruses, are involved in fusogenicity and infectivity of LASV GP-2.
INTRODUCTION
The entry of enveloped viruses into host cells requires the fusion of viral and cellular membranes, which is mediated by viral envelope glycoproteins. For the human pathogen Lassa virus (LASV), a biosafety level 4 Old World arenavirus that causes haemorrhagic fever in West Africa, this process is poorly characterized. The LASV membrane glycoprotein is synthesized in infected cells as the inactive precursor GP-C, which is cleaved in the endoplasmic reticulum by the endoprotease SKI-1/S1P into two subunits, GP-1 and GP-2, requiring participation of the unusually long signal peptide (Lenz et al., 2001; Eichler et al., 2003). Both cleavage fragments are transported to the cell surface, where probably homotrimeric complexes of GP-1 are linked non-covalently to homotrimeric complexes of membrane-anchored GP-2 (Eschli et al., 2006). LASV interacts via GP-1 with the α subunit of the transmembrane dystroglycan complex, thereby initiating uptake of the virus into the cell (Cao et al., 1998; Kunz et al., 2002). Fusion of viral and cellular membranes is thought to be mediated by the low pH of the endosomal compartment, because entry of the related lymphocytic choriomeningitis virus (LCMV) is inhibited by lysosomotropic agents and mixing of LCMV virions with fluorescently labelled liposomes led to their dequenching at low pH (Di Simone et al., 1994; Di Simone & Buchmeier, 1995). GP-2 of LASV was recently shown to belong structurally to the class I viral fusion protein family, which includes fusion proteins of retroviruses, orthomyxoviruses, paramyxoviruses, filoviruses and coronaviruses (Eschli et al., 2006). They are characterized by a hydrophobic N terminus and two antiparallel helices separated by a glycosylated antigenic apex. After insertion of a short hydrophobic peptide located at the N terminus of the fusion protein into the membrane of the target cell, the helices form thermodynamically favoured six-helix bundles that bring the viral and cellular membranes into apposition, leading ultimately to their fusion (Earp et al., 2005). Whilst the principal features of LASV and LCMV GP-2 correspond to those of other class I fusion proteins, their highly conserved N-terminal regions (residues 260–266) with the canonical fusion tripeptide sequence Gly–X–Phe are only weakly hydrophobic and it was therefore proposed that their function may differ from that of classical fusion peptides (Gallaher et al., 2001). Experimentally, an internal hydrophobic sequence of LASV GP-2 comprising residues 276–298, which is also highly conserved among all arenaviruses, has been shown to fuse liposomal membranes under acidic conditions (Glushakova et al., 1992). For the New World arenavirus Junín virus (JUNV) and very recently also for LCMV, it was shown that expression of the homologous signal peptide of GP-C is required for cell–cell fusion in a recombinant assay, possibly by stabilising the six-helix bundle and thereby being involved directly in the overall fusion process (York & Nunberg, 2006; Saunders et al., 2007). In order to map the region in the N terminus of LASV GP-2 required for fusion and infectivity, we established two highly sensitive assays based on reporter-gene activation by cell–cell fusion or transduction of target cells with LASV GP-C pseudotyped murine leukemia virus (MLV) particles. All hydrophobic amino acids in the two postulated fusion peptide segments (FPSs) of LASV GP-C were substituted individually by alanine, and the mutant glycoproteins were investigated for intracellular cleavage, cell-surface expression, fusogenicity and infectivity.
METHODS
Cells.
Vero cells (ATCC CCL-81), 293 cells (ATCC CRL-1573) and the 293 subclone cell line ΦNXgp, constitutively expressing MLVgagpol (kindly provided by G. P. Nolan, Department of Microbiology and Immunology, Stanford University School of Medicine, CA, USA), were grown in Dulbecco's modified Eagle medium (DMEM) with 1 mM pyruvate and supplemented with 10 % heat-inactivated fetal calf serum (FCS), 100 units penicillin ml−1, 0.1 mg streptomycin ml−1 and 2 mM l-glutamine. Chinese hamster ovary (CHO)-K1 cells (ATCC CRL-61) were grown in DMEM nutrient mixture F12 Ham with 1 mM pyruvate supplemented with 10 % heat-inactivated FCS, 100 units penicillin ml−1, 0.1 mg streptomycin ml−1 and 2 mM l-glutamine, whereas the SKI-1/S1P-deficient CHO subclone SRD-12B (kindly provided by J. L. Goldstein, Department of Molecular Genetics, University of Texas Southwestern Medical Center, Dallas, TX, USA) was maintained in the same medium as the CHO cells except for the addition of 5 μg cholesterol ml−1, 1 mM sodium mevalonate and 20 μM sodium oleate (Rawson et al., 1999).
Expression vectors, antibodies and construction of GP-C mutants.
Synthetic codon-optimized LASV GP-C (strain Josiah, GenBank accession no. AAA46286) was purchased from GeneArt and cloned into the eukaryotic expression vectors pAdApt (long cytomegalovirus promoter) and pCAGGS (chicken β-actin promoter), which were described previously (Havenga et al., 2001; Niwa et al., 1991). Construction of pCAGGS vectors for the expression of LASV nucleoprotein (NP) and LASV matrix protein (Z) and generation of anti-peptide rabbit serum for detection of NP, Z and GP-2 were described previously (Lenz et al., 2001; Strecker et al., 2003). Eighteen single amino acid exchanges to alanine (Fig. 1⇓) were introduced into the proposed FPS of GP-2 as well as at position G271 by recombinant PCR and confirmed by DNA sequencing (Higuchi et al., 1988). G260 was also changed to arginine and leucine. The 100 % conserved proline residue at position 275 was changed to the hydrophilic amino acid arginine, because substitution of prolines in internal fusion peptides has been shown to affect the fusogenicity of Ebola virus and avian sarcoma/leukosis virus significantly (Delos et al., 2000; Gómara et al., 2004). The plasmid pCDNA-SKI-1/S1P for the expression of SKI-1/S1P was kindly provided by N. G. Seidah, Laboratory of Biochemical Neuroendocrinology, Clinical Research Institute of Montreal, Quebec, Canada (Seidah et al., 1999). The reporter-gene expression plasmids for the transient cell–cell fusion assay, encoding the human immunodeficiency virus 1 (HIV-1) tat protein (pL3tat) or the HIV-1 long terminal repeat linked to the β-galactosidase gene (HIV-1-LTR-β-gal, pJK2), were a kind gift of V. Bosch (DKFZ Heidelberg, Germany) and have been described previously (Schwartz et al., 1990; Kimpton & Emerman, 1992). As internal positive controls for the fusion assay, the GP-C of the New World arenavirus JUNV and the haemagglutinin (HA) of influenza virus were used, because they have been evaluated in similar assays (Huang et al., 1981; York & Nunberg, 2006; Saunders et al., 2007). Construction of the pCAGGS plasmid for expression of LCMV GP-C (strain WE) was described previously (Beyer et al., 2003). The expression plasmid and antibody for detection of influenza virus HA, subtype H7 (H7-HA), were kindly donated by R. Wagner, Paul-Ehrlich-Institut, Langen, Germany, and described before (Wagner et al., 2005). The GP-C of JUNV, strain MC2, cloned into the expression vector pCDNA-3.1+, was a kind gift of J. Nunberg (Montana Biotechnology Center, The University of Montana-Missoula, MT, USA) (York et al., 2004). The pHIT-derived packageable MLV genome vector pCnBg, encoding the β-gal reporter gene, which was used for the production of retroviral MLV pseudotypes (rPT) expressing foreign viral surface proteins, and the polyclonal goat anti-MLV p30 antibody (Quality Biotech), for titration of the pseudotyped particles, were generous gifts from P. M. Cannon (University of Southern California, Los Angeles, CA, USA) and described previously (Soneoka et al., 1995; Bruett & Clements, 2001). The input quantity of equal amounts of cells in the individual assays was tested with a mouse monoclonal anti-β-actin antibody (Abcam).
Alignment of the GP-2 N terminus of Old and New World arenaviruses. The N termini of GP2 of LASV strain Josiah (GenBank accession no. AAA46286), LCMV strain WE (P07399), Mopeia virus (MOPV, AAC08700), JUNV strain MC2 (BAA00964), Tacaribe virus (TACV, NP694849), and Machupo virus (MACV, AAS77879) are shown. The numbering of LASV GP-C (aa 260–298) is given. Amino acids printed in bold were changed individually to alanine, leucine or arginine to generate 22 different LASV GP-C mutants. The putative N-FPS and I-FPS are underlined. Asterisks denote 100 % conserved residues in Old World (OWA) and New World (NWA) arenaviruses. The arrow indicates the cleavage site and start of the N terminus of GP2. †N terminus determined experimentally (Lenz et al., 2001; Buchmeier, 2002; Beyer et al., 2003); ‡N terminus and SKI-1/S1P cleavage site predicted (York et al., 2004).
Recombinant cell–cell fusion assay.
All transfections for this assay were done in 6 cm dishes using Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions. Vero or SRD cells were co-transfected with viral glycoproteins (H7-HA, LCMV-GP-C, JUNV GP-C, LASV GP-C or LASV GP-C mutants) and pJK2, trypsinized 6 h post-transfection, resuspended in DMEM/10 %FCS and co-cultured overnight in 12-well dishes with identically treated pL3tat-transfected 293 cells at a ratio of 1 : 1. The next morning, cell mixtures were exposed to pre-warmed (37 °C) RPMI 1640 medium (Gibco-BRL) supplemented with 2 % FCS and adjusted to different pH values with citric acid, and then cultured for 15 min at 37 °C. The cell cultures were then washed twice with PBS and incubated with DMEM/10 %FCS for a further 24 h to allow the transactivation of pJK2 by HIV-1 tat. As controls, Vero cells transfected with pJK2 and empty pAdApt vector were mixed with pL3tat-transfected 293 cells and treated as above. Cell–cell fusion was quantified by lysing cells 24 h post-pH shift in 100 μl β-gal reporter gene lysis buffer (Roche) according to the manufacturer's instructions. A 100 μlvolume of 1 mM chlorophenol red β-d-galactopyranoside (CPRG; Roche) was added to the cleared lysate supernatants as a β-gal substrate and the colorimetric β-gal-induced conversion of CPRG was measured as A570 in an Dynatech ELISA plate MR7000 reader. CHO and SRD-12B cells were treated as described above for Vero cells, except for the modification of a triple transfection with pAdApt-LASV-GP-C, pJK2 and pCDNA-SKI-1/S1P. In all experiments, the amount of transfected plasmid DNA was equalized by co-transfection of the respective amounts of empty vector. All assays were performed at least three times in triplicate (i.e. at least nine times). Fusion activities of all FPS mutants were normalized with respect to the level of processed, cell-surface-expressed GP-2 as measured by quantitative Western blotting (WB), resulting in the normalized fusion activity (NFA). Mutants showing ≥10 % of wild-type (wt) activity after normalization were regarded as positive for fusion.
WB analysis.
After electrophoretic transfer of proteins onto nitrocellulose membranes, the membranes were blocked at 4 °C overnight with 10 % milk powder in PBS. The blots were incubated for 1 h with the respective primary antibody diluted in PBS supplemented with 1 % milk powder and 0.1 % Tween 20, followed by incubation with a secondary antibody coupled with horseradish peroxidase (Dianova). Bound antibodies were visualized by using the SuperSignal chemoluminescence substrate as described by the supplier (Pierce). Quantitative WB analysis was done using the tina 2.09g software (Raytest Isotopenmessgeräte).
Cell-surface biotinylation and immunoprecipitation.
Vero cells in six-well plates were transfected with pAdApt-LASV-GP-C (wt or FPS mutants) using Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions. Control cells were transfected with the empty pAdApt vector. After a 24 h period following lipofection, the cells were incubated on ice for 30 min, washed twice with cold PBS containing 1 % NP-40 and centrifuged in a Heraeus Biofuge at 13 000 r.p.m. (16 060 g) for 30 min; the biotinylated proteins in the supernatant were precipitated with streptavidin–Sepharose beads (Amersham Biosciences) overnight at 4 °C. The beads were washed four times with lysis buffer and bound biotinylated proteins were eluted with reducing Laemmli SDS sample buffer, boiled and resolved by SDS-PAGE (12 % gel) followed by transfer to a nitrocellulose membrane. LASV GP-C (uncleaved) and GP-2 (processed subunit) were detected with polyclonal rabbit anti-GP-2 antibody (anti-GP477).
Infectivity assay with rPT harbouring LASV GP-C.
MLV was pseudotyped with recombinantly expressed LASV GP-C (wt or FPS mutants) by using a modified transient plasmid expression system described previously (Soneoka et al., 1995). Briefly, ΦNXgp cells were co-transfected in 10 cm dishes with equal amounts of pCnBg and pAdApt-LASV-GP-C (wt or FPS mutants) or empty pAdApt vector. rPT titration was performed by quantitative WB analysis, comparing protein-band intensities of the MLV p30 capsid protein 2 days post-transfection in the supernatants of ΦNXgp cells cleared of cellular debris. The presence of LASV GP-C (wt and mutants) on rPT was tested by WB analysis using the anti-GP477 antibody. Vero cells were infected with rPT-containing cell-culture supernatant and transduction efficacy was measured 2 days post-infection as described above for the cell–cell fusion assay. To enhance adsorption of rPT to target cells, polybrene (Sigma-Aldrich) was added to the rPT-containing supernatants during infection at a final concentration of 8 μg ml−1 (Sena-Esteves et al., 2004). Infectivity of all FPS mutants was normalized according to rPT titre and incorporation of processed GP-2 in the rPT, resulting in normalized infectivity (NINF). Mutants showing ≥10 % of wt activity after normalization were regarded as being positive for infectivity.
RESULTS
LASV GP-C induces cell–cell fusion in a recombinant assay at an unusually acidic pH
Cell–cell fusion by recombinantly expressed viral GP-C was tested in a reporter gene assay, whereby Vero cells were co-transfected with a viral glycoprotein construct and pJK2 (HIV-1-LTR-β-gal), overlaid with 293 cells transfected with pL3tat (HIV-1-tat), and exposed to different pHs. If the resulting conformational changes in the cell-surface-expressed viral glycoproteins result in fusion of the two different cell populations, the HIV-1 tat protein will transactivate the HIV-1 LTR. This leads to expression of β-gal, which is detected by conversion of the chromogenic substrate CPRG. Fig. 2⇓ shows that recombinant influenza virus H7-HA and recombinant JUNV GP-C as positive controls induced cell–cell fusion at a pH optimum of 5.0, as described previously (Huang et al., 1981; York & Nunberg, 2007). Vero cells expressing codon-optimized LASV GP-C fused with 293 cells only at pH≤4.5, with maximal fusion occurring at pH 4.0. LCMV GP-C induced cell–cell fusion over the same range and with the same pH optimum as LASV GP-C.
Recombinant cell–cell fusion assay. Vero cells co-transfected with plasmids expressing one of the indicated viral glycoproteins (□, LASV GP-C; ▪, LCMV GP-C; ▵, JUNV GP-C; ○, influenza virus HA) and a plasmid expressing HIV-1-LTR-β-gal were overlaid with 293 cells transfected with a plasmid expressing HIV-1 tat and exposed to a range of acidic pH values as indicated. The β-gal activity was measured by colorimetric conversion of CPRG (A570). Plasmid-expressed influenza virus A subtype 7 HA (○) and the empty expression vector pAdApt (•) served as positive and negative controls, respectively. One experiment, representative of three experiments performed independently and in triplicate, is shown. Bars represent sd.
Cleavage of LASV GP-C by SKI-1/S1P is required for fusogenicity
To test whether processing of LASV GP-C into the GP-1 and GP-2 subunits is required for fusogenicity, SRD cells lacking the LASV GP-C activating protease SKI-1/S1P were transfected with the respective constructs and tested in the cell–cell fusion assay with 293 cells. Cleavage of LASV GP-C was detected by WB analysis. Fig. 3⇓ shows that fusion did not occur in the absence of cleavage, whereas co-transfection of LASV GP-C with SKI-1/S1P restored cleavage and resulted in fusion at pH 4.0.
(a) Recombinant cell–cell fusion assay with SRD cells lacking the endoprotease SKI-1/S1P. Without cleavage of GP-C, no fusion occurs, whereas transfection of a plasmid expressing SKI-1/S1P restores cleavage and fusogenicity. Positive control, CHO wt cells; negative control, SRD cells transfected with empty vector. Bars represent sd of one representative experiment performed in triplicate. Empty bars, pH 7; filled bars, pH 4. (b) LASV GP-C cleavage in SRD and CHO cells. Uncleaved GP-C and the GP-2 subunit were detected by WB analysis with rabbit anti-GP-2 serum.
The pH of optimal fusion activity of LASV GP-C is not influenced by co-expression of LASV Z and/or LASV NP
Electron microscopy studies of LCMV particles have previously revealed a spatial proximity of GP-C to the matrix protein Z and the nucleoprotein NP, which supported the idea of a possible structural and functional interaction (Neuman et al., 2005). We therefore investigated their influence on cell–cell fusion by co-transfection experiments. Fig. 4⇓ shows that, compared with cell–cell fusion induced by GP-C alone, co-transfection with Z and NP did not change the pH requirements for fusion, and reduced fusion activity slightly.
Recombinant cell–cell fusion assay with co-expression of LASV GP-C and LASV Z and/or LASV NP. (a) Vero cells were co-transfected with the indicated expression vectors. Cell–cell fusogenicity was measured by reporter-gene activation after shift of the medium to the indicated pH. The mean±sd of three independent measurements are shown. (b) WB detection of LASV protein expression by rabbit antisera in the cells used for the assay.
Mutations in the N-terminal (N)-FPS and internal (I)-FPS of LASV GP-C affect cleavage and cell-surface expression
To investigate the involvement in fusion of the two conserved hydrophobic regions in the N terminus of LASV GP-2, an alanine scan was performed. Because cleavage of GP-C into GP-1 and GP-2 was shown to be a prerequisite for fusogenicity, we first tested whether the 22 mutants of LASV GP-C (Fig. 1⇑) were processed correctly (Fig. 5a⇓). WB analysis revealed that all mutants, except the G260L mutant, were cleaved, but four were cleaved with a much lower efficiency than that of the wt molecule (T261A, M284A, L285A and F293A; Fig. 5a⇓). We next determined whether mutated GP-2 was expressed on the surface of transiently transfected Vero cells by cell-surface biotinylation and immunoprecipitation. All 22 mutants were detected as uncleaved GP-C on the cell surface, whereas only 14 mutants were also transported as cleaved molecules to the cell surface (Fig. 5b⇓). For all mutants with reduced GP-C cleavage and four mutants with wt-like cleavage no cell-surface expression of GP-2 was detected (Fig. 5b⇓).
Cleavage and cell-surface expression of recombinantly expressed LASV GP-C mutants. (a) Cleavage of GP-C mutants was detected by WB analysis in whole Vero cell lysates 24 h post-transfection by using rabbit anti-GP-2 serum. (b) Cell-surface expression of GP-C mutants was detected by biotinylation and immunoprecipation as described in Methods. (c) WB detection of β-actin expression in the unbiotinylated fractions in each sample from (b).
Fusogenicity of LASV GP-C mutants depends on cleavage and cell-surface expression
All GP-C alanine mutants were tested in the cell–cell fusion assay and their fusogenic activity was compared to the wt molecule after normalization to the level of wt, processed, cell-surface-expressed GP-2. Fig. 6⇓ shows that no GP-C alanine mutant was fusogenic above 30–40 % of wt GP-C activity. Fusogenicity was abolished for all mutants without cell-surface expression of subunit GP-2 (G260L, T261A, L266A, M284A, L285A, I286A, C292A and F293A). For mutants that showed GP-2 cell-surface expression, fusogenicity was either abolished completely (F262A, W264A, G277A, Y278A and L280A) or reduced to 10–40 % of wt activity (G260R, G260A, G271A, P275R, G276A, W283A, L290A, G294A and V298A). This includes three of the five alanine mutations introduced in the proposed N-FPS (at amino acid positions 260, 262 and 264) and eight of the 13 alanine mutations introduced in the proposed I-FPS (at amino acid positions 276, 277, 278, 280, 283, 290, 294 and 298). All fusion-active mutants were tested over a range from pH 7 to 3, but no change in pH optimum compared with wt GP-C was observed (data not shown). These results are compatible with an involvement of both the proposed N-FPS and I-FPS in fusogenicity of GP-2.
Normalized fusion activity (NFA) of LASV GP-C mutants. Vero cells were co-transfected with reporter-gene plasmids and the indicated GP-C mutants, and subsequently used in the recombinant cell–cell fusion assay described in Methods. Fusion activities of all FPS mutants were measured by chromogenic CPRG conversion and normalized to the different cell-surface expression intensities of each GP-2 mutant. (cf. Fig. 5b). Each experiment was performed in triplicate and mean±sd are shown. The wt LASV GP-C expression vector and the empty pAdApt vector served as positive and negative controls, respectively. GP-2 mutants that are expressed at levels as high as or higher than wt and that show reduced or abolished fusogenicity are shown in bold.
Fusogenicity of LASV GP-C mutants is required for infectivity of rPT
The infectivity of all LASV GP-C mutants was tested by using a reporter-gene assay based on transduction of target cells with pseudotyped murine retroviral particles carrying the β-gal gene (rPT). The results were normalized according to the amount of processed GP-2 incorporated into the rPT. Uncleaved GP-C was always incorporated into rPT, but only 11 of the 22 mutant GP-2 molecules were detected as subunits on the particle surface (data not shown). Six of these (G260R, G260A, G271A, P275R, W283A and L290A) transduced Vero cells in the absence of polybrene, whereas all other mutants did not (Fig. 7⇓). Addition of polybrene to the assay resulted in additional infectivity in three of the remaining mutants (G276A, G294A and V298A), which had also shown activity in the fusion assay. The data suggest that these mutations influence the secondary structure of GP-2 such that activity in the fusion assay is retained, but attachment or other pre-fusion steps are impaired. Of the GP-2 mutants that show both cell-surface expression and incorporation into rPT, six of 11 clearly correlate in the cell–cell fusion and infectivity assay (nine of 11 taking the polybrene results into account). The results therefore demonstrate that fusogenicity is a prerequisite for infectivity.
Normalized infectivity (NINF) of MLV particles pseudotyped with LASV GP-C (rPT). rPT were produced and titrated as described in Methods. LASV GP-C and GP-2 were detected by WB analysis in cell-free supernatants of ΦNXgp cells used for production of rPT. The infectivity of LASV GP-C mutants. The β-gal reporter-gene activity in rPT-transduced cells was measured by a colorimetric reaction 2 days post-transfection. Infectivity was normalized to the rPT titre and the different incorporation level of each GP-2 mutant. Infection was performed in the presence (filled bars) or absence (empty bars) of polybrene. Experiments were performed three times independently and in triplicate, with the mean±sd of one representative experiment shown. GP-2 mutants that are incorporated at levels as high as or higher than wt and that show reduced or abolished fusogenicity are shown in bold.
Interestingly, the infectivity of the rPT carrying the G260A or G271A mutation, the latter located between the N-FPS and I-FPS was 25–50 % higher than that of the wt GP-C rPT, despite a reduction in fusogenicity of 80 and 70 %, respectively (Fig. 7⇑).
DISCUSSION
Recent experimental data have provided structural evidence that GP-2 of LASV belongs to the class I fusion protein family (Eschli et al., 2006). In this study, we demonstrate that GP-2 acts as the pH-dependent fusion protein of both LASV and LCMV by testing the functionality of wt and mutated viral glycoproteins in a recombinant cell–cell fusion and a pseudovirus infectivity assay. Unexpectedly, cell–cell fusion mediated by LASV and LCMV GP-2 required a pH of <5.0 with an optimum at pH 4.0, which is much lower than that reported for other class I fusion proteins and not observed within the endosomal compartment. However, HA of influenza virus A and GP-C of JUNV mediated fusion in our assay at a pH optimum close to 5.0, which corresponds to previously reported data from similar assays (York & Nunberg, 2006; Huang et al., 1981). As fusion of the murine astrocytoma cell line DBT by recombinantly expressed GP-C of LCMV was recently reported to occur at pH 5.0 (Saunders et al., 2007), cell-line-specific differences may account for the observed disparities. As reported for the fusion glycoproteins of other enveloped viruses, cleavage of LASV GP-C and cell-surface expression of GP-2 were found to be required for fusion activity in our assay.
The highly conserved hydrophobic N terminus (aa 260–266) of LASV and LCMV GP-2, comprising the canonical fusion tripeptide Gly–X–Phe, was proposed as an N-FPS (Gallaher et al., 2001). However, the in vitro fusion of liposomes at pH 4.5–5.5 with a peptide corresponding to an internal hydrophobic sequence (aa 276–298) suggested the existence of an additional I-FPS (Glushakova et al., 1992). Because functional viral FPSs are sensitive to mutations, we performed an alanine scan of all hydrophobic amino acids of both predicted FPSs. Fusogenicity was reduced or abolished in all 14 mutants that were expressed as GP-2 on the cell surface. Four of these mutations were located in the N-FPS, eight in the I-FPS and two in between (G271 and P275), thus clearly demonstrating involvement of the whole N terminus of GP-2 in the fusion process. Conservative glycine to alanine changes tended to have the least pronounced effect, reducing fusogenicity compared with wt by 65–80 %, including the N-terminal G260A substitution. Interestingly, all New World arenaviruses have an alanine at this position (Fig. 1⇑). The most pronounced effect on fusogenicity was observed for alanine substitutions of aromatic amino acids. Based on 100 % conservation among all Old and New World arenaviruses and on the loss of fusogenicity in the assay when substituted with alanine, we identified four critical hydrophobic amino acid positions in the N terminus of GP-2: W264, located in the proposed N-FPS, and G277, Y278 and L280, located in the proposed I-FPS. Interestingly, hydrophilic substitution of P275 reduced but did not abolish fusogenicity.
In the recombinant pseudotype assay, nine of 11 GP-C mutants with processed GP-2 detectable on the surface of the particles showed reduced or abolished rPT infectivity; two mutants were hyperinfectious (G260A and G271A). Three mutants with significant, albeit reduced, fusogenicity (G276A, G294A and V298A) showed infectivity only in the presence of polybrene, which non-specifically enhances adsorption of viral particles to cells (Davis et al., 2002). We therefore speculate that the alanine substitutions resulted in structural changes in GP-2 that abolished attachment (or internalization) and led to loss of epitopes.
Taken together, the data demonstrate that (i) GP-2 is the functional fusion protein of LASV, (ii) processing of GP-C is required for fusogenicity, which in turn is a prerequisite for infectivity, and (iii) amino acids from both highly conserved hydrophobic regions in the N terminus of GP-2 are critical for fusion and infectivity. This finding can probably be generalized to all arenaviruses, because several amino acid positions are 100 % conserved within the family. Similar findings of a hierarchical correlation of processing of LCMV GP-C, fusogenicity and infectivity were published recently (Saunders et al., 2007).
Exactly how the N terminus of GP-2 interacts with the cellular membrane remains to be investigated. The canonical fusion tripeptide sequence Gly–X–Phe is clearly important, as shown by the mutational analysis; however, the glycine tolerated an exchange to a polar amino acid, as opposed to the N-FPS of influenza HA2, for example. All New World arenaviruses have an alanine at this position and, interestingly, the alanine exchange, whilst reducing fusogenicity, increased the infectivity of Lassa GP-C pseudotypes. Based on its comparatively weak hydrophobicity and the fact that post-translational cleavage of GP-C occurs not at dibasic amino acids, but within the hydrophobic site (LL/GTFTWTL) (Lenz et al., 2001), its candidacy as a functional fusion peptide analogous to those of influenza and HIV-1 has been questioned (Gallaher et al., 2001). The primary role of the arenavirus N-FPS may therefore not be as a definitive anchor for GP-2 in the endosomal target membrane, but rather to enable contact with the internal hydrophobic sequence. This sequence, however, differs structurally from previously identified I-FPSs, which are segmented into two ordered regions with an intervening turn or loop, usually containing one or more proline residues (Delos et al., 2000). The Robson–Garnier algorithm predicts an order–turn–helix structure for the N terminus of GP-2, with the turn at P275, which is located between the putative N-FPS and I-FPS. The predicted turn region would be approximately 10 aa, which falls within the range of the predicted turn regions for other viral I-FPSs (Delos et al., 2000). However, with respect to fusogenicity and infectivity, P275 was less sensitive to hydrophilic substitution than the prolines located in the I-FPSs of other viruses, e.g. Ebola virus (Ito et al., 1999).
We therefore propose that the two hydrophobic regions within the N terminus of GP-2 resemble N-terminal and internal fusion peptides, respectively, but that both need to interact with the cellular membrane to initiate fusion.
Acknowledgments
We thank O. Lenz, P. M. Cannon, R. Wagner, N. G. Seidah, V. Bosch and J. Nunberg for kindly providing plasmids and antibodies. We thank M. Matrosovich, T. Strecker, S. Daffis and W. Garten for helpful discussion and critical reading of the manuscript. C. K. was supported by the Jung-Stiftung für Wissenschaft und Forschung, Hamburg, Germany, and performed this work in partial fulfilment of the requirements for a PhD degree from the Philipps University Marburg, Germany.