INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

The envelope glycoproteins of a number of virus types are produced first as precursors and then are cleaved during virion maturation at short, positively charged sequences (reviewed in references 19 and 26). With several viruses, including alphaviruses, the subtilisin-like host cell protease furin has been identified as mediating this cleavage (6, 19, 23, 42). In mammalian cells, furin is localized to the protein secretory pathway between the trans-Golgi and cell surface (27, 45). The consensus recognition sequence for furin proteases is X-Arg-X-Lys/Arg-Arg-X (i.e., XBXBBX, where B is a basic amino acid and X is a hydrophobic amino acid, with the protein cleaved between the final Arg and X residues); the final X is Ser with many viruses (19, 27, 33, 45).

The Sindbis virus attachment protein E2 is synthesized as a precursor, PE2, which is cleaved at amino acid 64, yielding the mature E2 spike protein. The amino terminus of PE2, E3, contains the XBXBB portion of the cleavage sequence but is not itself retained in virus particles (42). Glycoprotein spikes on mature virus particles are composed of heterodimers of E2 and E1, another viral structural glycoprotein (reviewed in reference 42). Mutagenesis and mutant selection studies have indicated that the efficiency of cleavage at furin recognition sites can be decreased by substitution of amino acids other than Ser at the final X position (9, 14, 35). With Sindbis virus PE2, substitution of Arg at this site results in partial cleavage compared to that with the wild-type Ser, while substitution of Asn, Leu, or Val can result in PE2 that is predominantly cleavage defective (9). Uncleaved PE2 is incorporated into PE2-E1 heterodimers on mutant Sindbis virus particles that are released from infected cells (9, 35).

Surprisingly, the furin recognition sequence is identical (although in opposite orientation) to one of two heparin-heparan sulfate (HS) interaction consensus sequences (XBBXBX or XBXXBBBX) deduced by Cardin and Weintraub through comparison of protein domains known to interact with heparin (4). Two additional heparin-HS consensus sequences have been described more recently (13, 40). In the context of heparin-HS binding proteins, such as vitronectin, fibronectin, and lipoprotein lipase, these sequences promote attachment primarily through ionic interaction with carboxylate and sulfate groups on heparin-HS chains (reviewed in references 16 and 18). It has been proposed that protein-HS interactions can be mediated by single linear consensus sequences as well as binding sites determined by the protein tertiary structure and composed of several linear heparin-HS interaction sequences (13). Biological processes involving protein-heparin-HS interaction include cell adhesion, maintenance of tissue structure, anticoagulation activity, sequestration and concentration of cell signaling factors, and possibly regulation of transcription factor activity (16, 44). In addition, HS attachment plays a role in the in vitro infection of a number of viruses (3, 5, 20, 31, 34, 43, 46).

We have recently shown that Sindbis virus strains adapted to baby hamster kidney (BHK) cells attach to HS during the infection of cultured cells (20). In contrast, the Sindbis virus strain AR339 consensus sequence virus (22) infects cultured cells by a primarily HS-independent mechanism (20). We have identified three loci in the E2 attachment protein (E2 1, E2 70, and E2 114) that mutate to positive charge during the adaptation of Sindbis virus to BHK cells and can independently confer the ability to bind cell surface HS. However, the exact composition of HS binding sites on the Sindbis virus glycoprotein spike remains to be identified.

Several lines of evidence suggest that the furin protease cleavage signal of Sindbis virus PE2 may be involved in HS binding: (i) the E2 Arg 1 mutation occupies the carboxy-terminal position of the XBXBBX PE2 furin protease cleavage site; (ii) as indicated above, Arg at E2 1 results in a reduction in the cleavage efficiency of PE2, resulting in virus particles that contain some PE2 and bind HS (20); and (iii) completely cleavage-defective viruses that maintain the furin protease cleavage signal by incorporating PE2 into virions attach more efficiently to cultured cells than the E2 Arg 1 mutant (10). These observations, in addition to the resemblance of the furin cleavage site to a heparin-HS interaction consensus, have prompted an investigation of whether the XBXBBX furin protease sequence of PE2 could mediate virion attachment to cell surface HS and whether decreased PE2 cleavage efficiency as found with the E2 Arg 1 mutation might be one mechanism by which Sindbis viruses adapt to growth in cultured cells.

In the studies reported here, we have compared the cell and heparin bead binding of viruses with various PE2 cleavage efficiencies (determined by the residue at E2 position 1) as well as those of noncleaving viruses with all or part of the furin protease cleavage site deleted. The results of these studies indicate (i) that furin protease cleavage sequences can mediate attachment of Sindbis viruses containing PE2 to cellular HS and (ii) that the magnitude of cell binding is correlated with the abundance of PE2 and the associated furin cleavage sequence in virus particles. This is in contrast with binding mediated by positive-charge mutations at E2 70 (Lys) or E2 114 (Arg) loci, which is independent of the PE2 furin protease cleavage sequence.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Cell culture. Chinese hamster ovary (CHO K1) cells were maintained at 37°C in Ham's F-12 medium supplemented with 10% fetal bovine serum, 100 U of penicillin per ml, and 0.05 mg of streptomycin per ml. BHK cells were maintained at 37°C in alpha minimal essential medium supplemented with 10% donor calf serum, 10% tryptose phosphate broth, 0.29 mg of L-glutamine per ml, and antibiotics as described above. Mosquito cells (C6/36) were maintained in alpha minimal essential medium supplemented with 10% fetal bovine serum, tryptose phosphate broth, L-glutamine, and antibiotics as described above and were incubated at 28°C.

Viruses. The construction of the consensus Sindbis virus AR339 clone (pTR339) and clone p3070 has been described previously (20, 22). The "p" prefix indicates the cDNA form of the virus clone. Additional full-length cDNA clones, p39R1, p39N1, p39L1, and p39V1, were constructed by substitution into pTR339 of a StuI-to-BssHII fragment from pTRSB (22), pTRSB-N (9), pTRSB-E2L1 (9), and pTRSB-E2V1 (9), respectively. The cDNA clones pFD, pFD1, pFD2, and pFDN1 were constructed by "megaprimer" PCR mutagenesis (39) of pTR339. Briefly, mutagenesis primers were designed to create the desired PE2 cleavage sequence of each virus. Each mutagenesis primer was used separately, along with a nonmutagenic primer, in a first round of PCR amplification with pTR339 DNA as a template. The nonmutagenic primer was chosen so that the resultant amplicon (megaprimer) would span the StuI restriction site (nucleotide 8571) of pTR339 and be ~150 to 300 bp in length. After agarose gel purification, each mutant megaprimer was used separately as a primer for a second round of PCR with another nonmutagenic primer. This yielded an amplicon of ~1.4 kb that spanned from upstream of the StuI restriction site to downstream of the BssHII (nucleotide 9804) restriction site of pTR339. Gel-purified amplicons were then digested with StuI and BssHII and cloned into similarly digested pTR339. p39K70 and pFDK70 were constructed by a similar method but with nonmutagenic primers spanning the same regions and switching template DNA (p3070 in the first round for both and, in the second round, pTR339 for p39K70 and pFD for pFDK70) between the rounds of PCR amplification. This resulted in the combining of mutations present in separate clones. All genetic manipulations were confirmed by DNA sequencing at the University of North Carolina at Chapel Hill Automated DNA Sequencing Facility with a model 373A DNA sequencer (Applied Biosystems) with the Taq DyeDeoxy terminator cycle sequencing kit (Applied Biosystems). Viruses were produced by high-efficiency electroporation of BHK cells with in vitro transcripts of linearized cDNA clones as described previously (20).

Virus purification. BHK or C6/36 cell monolayers were infected either with TR339, 39R1, 39L1, 39V1, or 39K70 at a multiplicity of infection of >5, or the cells were harvested and electroporated with in vitro transcripts of pFD, pFD1, pFD2, p39N1, pFDN1, or pFDK70, followed by radiolabeling and sucrose gradient purification as previously described (20). Viruses generated by infection were purified on discontinuous 20 to 60% (wt/wt) sucrose gradients in TNE (0.05 M Tris-HCl [pH 7.2], 0.1 M NaCl, 0.001 M EDTA) buffer, followed by continuous 20 to 60% sucrose gradients, and pelleted through 20% sucrose. Due to lower yields from electroporated cells than from infected cells, viruses generated from electroporation were purified on discontinuous sucrose gradients as described above, followed by pelleting through 20% sucrose. The binding phenotypes of virus particles prepared by the two methods were compared by using FD1 (cleavage defective; weak cell binding), 39L1 (cleavage defective; strong cell binding), and TR339 (cleavage competent; weak cell binding) viruses and found not to differ significantly. Specific infectivity (PFU/counts per minute) analysis of radiolabeled virus preparations was performed on BHK cell monolayers as previously described (20).

Virus attachment assay. CHO and BHK cell binding assays and heparin-agarose and bovine serum albumin (BSA)-agarose bead (both from Sigma) binding assays were done as previously described (20). Virus was allowed to attach to cells or beads for 1 h at 4°C. Heparin-agarose bead binding of virus particles routinely varied between 70 and 90% of added counts per minute between reactions, most likely due to variable loss of beads during the wash steps. For this reason, a positive result from bead binding is reported if >70% of radiolabeled virus particles bound in repeated assays. A negative result is indicated if heparin- or BSA-agarose beads bound <10% of added counts per minute in repeated assays. Heparinase I (Sigma) digestion of CHO cells was done as previously described for BHK cells (20). All binding assays were repeated at least twice.

Polyacrylamide gel analysis of [³⁵S]methionine-labeled viral proteins. Individual structural proteins from radiolabeled virus particles prepared for binding assays were resolved by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) (10% acrylamide) under reducing conditions (50 mM 2-mercaptoethanol). The radiolabeled protein bands were visualized with a Storm model PhosphorImager and ImageQuant software (Molecular Dynamics). The relative PE2 cleavage efficiency was determined by dividing the image densities of bands corresponding to PE2 by the total image density of PE2 plus E1 plus E2 plus capsid proteins. The results slightly underestimate the difference between cleaving and noncleaving viruses due to the loss of E3, which contains one methionine (of a total of 27 in PE2), from the PE2-cleaving-virus lanes. PE2 cleavage calculations were performed on at least two gels for each virus with similar results.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

To directly evaluate the role of the furin protease cleavage recognition sequence in cell binding, a panel of mutants was created containing substitutions that had previously been shown to affect PE2 cleavage efficiency or that had portions of the XBXBBX sequence deleted (Table 1). Substitution of Arg for Ser at E2 1 is associated with an increase in HS-dependent attachment to cells (20) and partial inhibition of PE2 cleavage (9). Substitution of Val or Leu at E2 1 greatly reduces PE2 cleavage, perhaps due to interference of the aliphatic R group of these amino acids with the furin protease (9). Likewise, substitution of Asn at E2 1, which confers a rapid-penetration phenotype on the related alphavirus S.A.AR86, inhibits PE2 cleavage in the contexts of both Sindbis virus AR339 and S.A.AR86 (9, 35). This residue creates an N-linked glycosylation signal, and the glycosylation of E2 Asn 1 is proposed to interfere with furin activity. For these mutants, uncleaved PE2 would contain an intact BXBB furin recognition and HS binding consensus sequence. In addition, mutants were created with either the entire BXBB sequence or individual positively charged amino acids deleted. With these mutant viruses, the correlation between retention of PE2 in virions, the presence of an intact furin protease cleavage signal, and cell binding could be evaluated.

PE2 cleavage of viruses produced in BHK cells. SDS-PAGE analysis of virions with different PE2 cleavage phenotypes is shown in Fig. 1, and the PE2 content of purified virus particles is quantitated in Table 1. A completely noncleaving virus (100% PE2) should give a ratio of PE2 to PE2 plus E2 plus E1 plus capsid of 0.259 (7 methionine residues in PE2 of 27 total methionine residues). Consistent with previous reports (9), Ser at E2 1 (TR339 and 39K70) resulted in nearly complete cleavage of PE2 by the BHK cell furin protease (2.3 and 1.5% PE2, respectively), resulting in virions containing little or no PE2. Also, consistent with previous reports, substitution of Arg at E2 1 (39R1) resulted in significant reduction of cleavage efficiency (~14% PE2 content in virions). Substitution of Asn at E2 1 (39N1) resulted in near-complete inhibition of PE2 cleavage, suggesting that mature particles of this virus contained spikes composed entirely of PE2-E1 heterodimers. Altered migration of PE2 from 39N1 and FDN1 preparations reflects the additional glycosylation of PE2 with Asn at E2 1 (Fig. 1B) (35). Viruses with the BXBB portion of the cleavage signal deleted (FD, FDN1, or FDK70) or the 1 (FD1) or 1 and 2 (FD2) amino acids of the cleavage signal deleted also failed to cleave PE2. Viruses with Leu (39L1) or Val (39V1) at E2 1 were predominantly cleavage defective; however, Phosphor- Imager analysis suggested that these viruses contained less PE2 (and consequently, more E2) than the deletion mutants or 39N1 (Table 1).

View larger version (112K): [in this window] [in a new window]   FIG. 1.   SDS-PAGE analysis of purified radiolabeled virus particles; 5.0 × 104 cpm of virus was loaded in each lane. Molecular mass markers (kDa) are shown in the extreme right-hand and left-hand lanes.

Binding of viruses to CHO cells. Binding studies with CHO K1 cells indicated that the presence of an intact furin protease cleavage sequence was correlated with efficient attachment to cells (Fig. 2). Consistent with the near-complete cleavage of PE2 and the lack of E2 internal mutations associated with HS binding, TR339 exhibited barely measurable binding to CHO cells. Substitution of Arg for Ser at E2 1 (39R1) resulted in a significant increase in cell attachment (P = <0.01; Student's t test). As noted above, the enhanced attachment of 39R1 was correlated with partial inhibition of PE2 cleavage and increased retention of PE2 in virions. Further increase in virion PE2 content conferred by substitution of Val (39V1), Leu (39L1), or Asn (39N1) at E2 1 resulted in viruses with four- to fivefold-greater binding to the cells than 39R1, suggesting that the PE2 content of virus particles and, consequently, the presence of the furin cleavage signal was correlated with attachment efficiency. These data also indicate that the individual differences in PE2 content between 39L1, 39V1, and 39N1 virions is not correlated with a difference in attachment, suggesting a threshold of PE2 content above which binding is not increased.

View larger version (15K): [in this window] [in a new window]   FIG. 2.   Binding of radiolabeled purified viruses to CHO cells. The bars represent averages of triplicate binding assays with 105 cpm of virus and ~106 cells per reaction. Each set of triplicates was repeated at least twice. The error bars represent standard deviations.

Effect of heparinase I digestion on CHO cell binding. Previously, we showed that the attachment to cells of cell culture-adapted Sindbis virus strains was due to virus interaction with cell surface HS (20). This phenotype was demonstrated by soluble-heparin competition of virus binding and infectivity, reduction of binding and infectivity after heparinase digestion of cell surfaces, evaluation of virus binding and infectivity with HS- or GAG-deficient CHO cells, and direct virus attachment to heparin-agarose beads. In the present studies, we have used digestion of cell surface HS with heparinase I and heparin-agarose bead binding to determine the requirement for HS in cell attachment.

View larger version (12K): [in this window] [in a new window]   FIG. 3.   Binding of viruses to CHO cells without (solid bars) or with (hatched bars) previous digestion with heparinase I. The cells were either digested with 8 U of heparinase I (in phosphate-buffered saline with 0.1% BSA/ml) or mock-digested (phosphate-buffered saline with 0.1% BSA only) for 1 h at 37°C followed by processing for attachment assays as described in Materials and Methods. The bars represent averages of triplicate binding assays with 105 cpm of virus and ~106 cells per reaction. Each set of triplicates was repeated at least twice. The error bars represent standard deviations.

Binding of viruses to heparin-agarose and BSA-agarose beads. Consistent with a direct interaction of efficiently binding viruses with HS, all of these viruses (39R1, 39L1, 39N1, and 39K70) bound to heparin agarose beads (Table 2). Binding of all viruses to BSA-agarose beads ranged from 0.25 to 5% of added counts per minute (data not shown). This result, in combination with our previous report (20), indicates that mutations associated with the PE2 cleavage site (e.g., Arg, Leu, Val, and Asn at the +1 position) and mutations internal to E2 (e.g., E2 Lys 70 and E2 Arg 114) can mediate interaction with heparin. Consistent with our previous report, TR339 attached very poorly to the beads. Similarly, FD exhibited little attachment, suggesting that, similar to cellular HS binding, E3 residues outside of the BXBB signal region are not responsible for the heparin interaction. Deletion of the 1 and 2 basic residues together (FD2) also eliminated the heparin bead interaction; however, deletion of only the 1 basic residue resulted in a virus that bound poorly to cells yet was capable of interaction with the heparin beads. This result indicates that, in the context of the furin cleavage sequence region of the Sindbis virus spike, the XBXBBX or XBXBX sequences are sufficient for interaction with heparin while the full furin site is required for efficient attachment to cellular HS.

Correlation of PE2 cleavage efficiency with cell attachment. The studies described above correlated the attachment of genetically different viruses to HS on cells and heparin-agarose beads with retention of an intact furin protease cleavage sequence in virus particles. Studies in this section evaluated the effect of altered PE2 cleavage efficiency on binding of genetically identical viruses. We initially attempted to produce radiolabeled purified virus in furin protease-deficient CHO cells (23, 24) and BHK cells treated with monensin, which inhibits PE2 cleavage (17, 32). However, virus yields following purification were insufficient (data not shown). Heidner et al. (11) demonstrated that PE2 containing Arg, Leu, or Val at E2 1 was cleaved more efficiently when virus was produced in the mosquito cell line C6/36 than in BHK cells. In these studies, PE2 with Val or Arg at E2 1 was efficiently cleaved in C6/36 cells while PE2 with Leu at E2 1 exhibited partial cleavage. The authors suggested that the host cell protease responsible for PE2 cleavage in arthropod cells was more tolerant of amino acid sequence variation at the +1 position of the furin recognition sequence. Therefore, depending on the cell of origin, stocks of genetically identical viruses could be prepared that differed in PE2 content.

View larger version (18K): [in this window] [in a new window]   FIG. 4.   Binding of radiolabeled viruses prepared in BHK cells or C6/36 cells to CHO cells. The bars represent averages of triplicate binding assays with 105 cpm of virus and ~106 cells per reaction. Each set of triplicates was repeated at least twice. The error bars represent standard deviations.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

HS attachment can be mediated by furin protease cleavage sequences. The results of these studies indicate that the XBXBBX furin protease cleavage signal is directly involved in attachment of partially cleaving or noncleaving Sindbis viruses to cellular HS and that the efficiency of attachment correlates with the content of PE2 in virus particles. While heparin bead attachment can apparently be mediated by XBXBX or XBXBBX sequences, high-affinity attachment to cellular HS requires the XBXBBX motif. This mechanism of HS attachment is distinct from that involving internal E2 residues, such as E2 Lys 70 and E2 Arg 114.

Relationship of binding to infectivity. Cleavage of exterior viral spike glycoproteins is normally associated with activation of the viral spike complex for infection (19). During virus particle assembly, uncleaved spike proteins are presumed to be "protected" from premature fusion with infected host cell membranes. After cleavage, which usually occurs as a late event in morphogenesis, virus particles are released from the cell in a fusion-competent state. This also appears to be the case with Sindbis virus strain AR339. Completely cleavage defective Sindbis virus AR339 particles, while attaching very efficiently to cells, are poorly infectious, perhaps due to inhibition of fusion or other uncoating processes (36). In contrast, the Sindbis virus-like alphavirus S.A.AR86 (in which E2 Asn 1 was selected by adaptation to BHK cells [35]) exhibits enhanced infectivity when cleavage is inhibited by glycosylation at E2 Asn 1. With this virus, uncoating and fusion may be allowed in the presence of uncleaved PE2, with greater infectivity potentially arising from HS attachment mediated by the furin cleavage sequence. With Sindbis virus strain AR339, partial inhibition of cleavage (e.g., 39R1), may allow the virus to increase binding affinity due to the interaction of the furin site with HS while maintaining uncoating and fusion competence, resulting in an overall increase in infection efficiency. Similarly, internal mutations in E2 (Lys 70 and Arg 114) increase cell attachment while potentially maintaining or increasing uncoating and fusion efficiency, resulting in a large infectivity increase. Indeed, second-site-reverting mutations (e.g., Glu to Gly at E2 216), that promote enhanced infection efficiency can be selected in the cleavage-defective background (12). The E2 Gly 216 mutation does not alter the efficiency of binding mediated by the furin cleavage sequence (data not shown). Therefore, this mutation may act by improving the efficiency of uncoating and entry (36). In addition, the significant difference in infectivity among 39L1, 39V1, and 39N1 (Table 1), although these viruses bind similarly to cells (Fig. 2), may also result from uncoating and entry efficiency. While there are very likely different cellular HS structures bound by different Sindbis virus mutants (20), our data suggest that cell culture-adaptive mutations that increase virus attachment efficiency while maintaining uncoating and fusion competence result in enhanced infection efficiency for fibroblast cells such as BHK.

Can furin protease sequences of other virus types bind HS? The results of these studies raise the possibility that furin protease cleavage sequences could participate in an HS-dependent attachment by other virus types. Spike proteins of the Togaviridae, Orthomyxoviridae, Paramyxoviridae, Flaviviridae, Coronaviridae, Toroviridae, Retroviridae, and Herpesviridae families are cleaved by furin-like host proteases at XBXBBX sequences during maturation (reviewed in reference 19). Several studies with human immunodeficiency virus (HIV) have suggested that some T-cell-tropic strains attach to HS during infection of cultured cells (29, 31, 34). At least one of the HS attachment domains has been mapped to the V3 loop of HIV gp120 (34). Interestingly, this region of the glycoprotein contains what has been termed an "occult" furin protease cleavage sequence that has been proposed to be cleaved during entry into cells (25). Additionally, two colocalized furin consensus sites in the immature gp160 can be cleaved during virus maturation to yield the mature gp120 (2). Short, branched-chain synthetic peptides comprising these gp160 furin protease cleavage sequences inhibit HIV replication in a dose-dependent manner (1). The antiviral effect is associated with peptide attachment to and internalization by cells; however, the mechanism of attachment and internalization of these peptides is unknown (1). It is possible that these positively charged peptides bind to cellular HS and are internalized as proteoglycan-peptide complexes.