Identification of a Specific Region in the E1 Fusion Protein Involved in Zinc Inhibition of Semliki Forest Virus Fusion

Abstract

The enveloped alphaviruses infect cells via a low-pH-triggered membrane fusion reaction mediated by the viral transmembrane protein E1. During fusion, E1 inserts into the target membrane and refolds to a hairpin-like postfusion conformation in which domain III (DIII) and the juxtamembrane stem pack against a central core trimer. Although zinc has previously been shown to cause a striking block in alphavirus fusion with liposome target membranes, the mechanism of zinc's effect on the E1 fusion protein is not understood. Here we developed a cell culture system to study zinc inhibition of fusion and infection of the alphavirus Semliki Forest virus (SFV). Inclusion of 2 mM ZnCl₂ in the pH 5.75 fusion buffer caused a decrease of ∼5 logs in SFV fusion at the plasma membrane. Fusion was also inhibited by nickel, a chemically related transition metal. Selection for SFV zinc resistance identified a key histidine residue, H333 on E1 DIII, while other conserved E1 histidine residues were not involved. An H333N mutation conferred resistance to both zinc and nickel, with properties in keeping with the known pH-dependent chelation of these metals by histidine. Biochemical studies demonstrated that zinc strongly inhibits formation of the postfusion E1 trimer in wild-type SFV but not in an H333 mutant. Together our results suggest that zinc acts by blocking the fold-back of DIII via its interaction with H333.

INTRODUCTION

Enveloped viruses infect cells by membrane fusion reactions that are triggered at the plasma membrane or in the endocytic pathway (reviewed in references 16, 28, and 43). Virus membrane fusion is mediated by conformational changes in specialized transmembrane proteins on the virus surface. During fusion, these viral proteins interact with the target cell membrane via a hydrophobic fusion peptide and refold to a hairpin-like conformation that brings the fusion peptide and transmembrane anchor into close proximity. Inhibition of these conformational changes blocks virus fusion and infection and is a potent antiviral strategy (11, 29). In some cases, peptides or domains of the fusion protein can act as dominant negative inhibitors of the refolding reaction (10, 20, 23, 36). In other cases, small molecules that block fusion protein conformational changes have been identified by in silico studies or screens for inhibitors (6, 12, 35). Definition of the detailed molecular mechanism of inhibition is critical to improve such inhibitors and to develop them into usable antiviral therapies.

Alphaviruses are small enveloped plus-sense RNA viruses that include important human pathogens such as Chikungunya virus and eastern equine encephalitis virus and the well-characterized Semliki Forest virus (SFV) (reviewed in reference 18). Alphaviruses infect cells by endocytic uptake and a membrane fusion reaction triggered by endosomal low pH (reviewed in reference 17). Fusion is mediated by the transmembrane E1 protein, an elongated molecule containing three domains (DI to DIII), arranged with the central DI connecting on one side to DII and on the other side with DIII, which has an immunoglobulin-like fold (21, 34). The hydrophobic fusion peptide is found as a loop at the distal tip of DII, and the stem and transmembrane domain are at the E1 C terminus, connecting to DIII. E1 is synthesized together with a companion protein, E2, which remains associated with E1 on the virus surface and covers much of the E1 protein, including the fusion loop (22, 30, 39). At low pH, the E2-E1 dimer dissociates, exposing the E1 fusion loop, which can then interact with the target membrane (14, 22, 40). E1 trimerizes DIII and the stem pack against the core trimer of DI/DII, thus forming the postfusion E1 hairpin (15). Exogenous DIII proteins can bind to the core trimer if present during low-pH-induced E1 refolding, thus blocking the fold-back of endogenous DIII, the formation of the hairpin, and virus fusion and infection (23). Mutations that inhibit E2-E1 dimer dissociation, E1-membrane insertion, or core trimer formation can also block virus fusion and infection (reviewed in references 17 and 33). These studies define specific regions or interactions of E1 that play an important role in the alphavirus fusion reaction.

Zinc has been shown to inhibit alphavirus fusion with liposomes (7) and the fusion of alphavirus-infected cells with red blood cell targets (45). Zinc does not block E2-E1 dimer dissociation or E1-membrane insertion, but it decreased the formation of the postfusion homotrimer (7). Zinc is a metal that has a high affinity for binding to histidine, cysteine, aspartate, and glutamate residues in biological systems (reviewed in references 1 and 19). Zinc is an essential element for growth and development, and many enzymes have evolved networks of residues that optimize electrostatic interaction with zinc for use in catalysis. Zinc's stereochemical flexibility allows it to be coordinated by 2 to 8 ligands, with 4-, 5-, and 6-liganded forms seen most often in nature. The mechanism of zinc's effect on alphavirus E1 has not been defined. While divalent cations such as Ca²⁺, Mg²⁺, Ba²⁺, and Sr²⁺ do not inhibit alphavirus fusion (7), the effect of transition metals that have chemical properties similar to those of zinc has not been addressed.

Here we have taken advantage of viral genetics and the chemistry of zinc and related transition metals to explore the mechanism by which zinc inhibits SFV fusion. In keeping with zinc's known chelation properties, inhibition was strongly affected by pH and was also observed with the transition metal nickel. We isolated SFV mutants that escape zinc inhibition. The mutants identified a specific histidine residue on E1 DIII, H333, as a ligand for zinc interaction. Mutation of H333 increased the zinc resistance of both SFV fusion and E1 trimer formation. Together these studies define molecular details of the mechanism by which zinc inhibits the refolding and function of the alphavirus fusion protein.

(The data in this paper are from a thesis submitted by C.Y.L. in partial fulfillment of the requirements for the degree of Doctor of Philosophy in the Graduate Division of Medical Sciences, Albert Einstein College of Medicine, Yeshiva University.)

MATERIALS AND METHODS

Cells and viruses.

BHK-21 cells were cultured in BHK medium (Dulbecco's modified Eagle's medium containing 5% fetal bovine serum, 10% tryptose phosphate broth, 100 U penicillin/ml, and 100 μg streptomycin/ml).

Wild-type (wt) SFV was derived from the SFV infectious clone pSP6-SFV4 (25) as previously described (27). The SFV mutants H331A, H331A/H333A, H3A/D284A, H125A, and H230A/srf5 were generated by site-directed mutagenesis of the SFV infectious clone as previously described (4, 5, 31). Sindbis virus was derived from the Toto1101 infectious clone (32). The Chikungunya virus 181/25 vaccine strain was kindly provided by Robert Tesh (World Reference Center for Emerging Viruses and Arboviruses, University of Texas Medical Branch, Galveston, TX). [³⁵S]methionine-cysteine-labeled virus was prepared by infecting BHK-21 cells with wt SFV by electroporation or with H331/H333A by infection at 10 PFU/cell, followed by radiolabeling and gradient purification as previously described (27).

Fusion-infection assay with zinc.

Our standard fusion-infection assay (27, 37) was adjusted for use with zinc. Virus binding steps on ice were carried out as usual in RPMI medium without bicarbonate, 0.2% bovine serum albumin (BSA), 10 mM HEPES (pH 7.0). Since zinc precipitates in phosphate-based solutions, we used a buffer system, termed “buffer Z,” consisting of 1 mM CaCl₂, 0.5 mM MgCl₂, 2.5 mM KCl, 125 mM NaCl, 5 mM glucose, 20 mM MES (morpholinoethanesulfonic acid), 10 mM HEPES, 0.1% BSA. ZnCl₂ was added to this buffer at the indicated concentrations by dilution from a 50 mM stock solution. Since ZnCl₂ lowers the pH of the solution (∼0.2 pH units for the concentrations used in this study), the pH of the solution was adjusted after the addition of ZnCl₂.

To determine the effect of zinc on virus fusion, serial dilutions of the indicated viruses were bound to BHK cells on ice for 90 min. The cells were then washed once with buffer Z and treated for 1 min at 37°C with buffer Z at the indicated pH and zinc concentration. To assess whether zinc inhibits “prefusion” or “postfusion” steps, pH 7 buffer Z containing no zinc or 2 mM zinc was added to cells for 1 min at 37°C before or after low-pH treatment, respectively. After all zinc treatments, cells were washed with buffer Z and incubated at 28°C overnight in BHK medium containing 20 mM NH₄Cl to prevent secondary infection. Infection was visualized by immunofluorescence.

Selection for zinc-resistant mutants.

Zinc-resistant mutants were selected by binding wt virus to BHK cells on ice in 9 individual wells of 6-well plates and then treated for 1 min at 37°C with pH 5.75 buffer Z containing 0.5 to 2 mM zinc. The cells were washed and incubated overnight at 37°C in the presence of 20 mM NH₄Cl to prevent secondary infection and allow progeny virus release. Primary infection was noted by immunofluorescence and compared to parallel samples treated in the absence of zinc. The culture medium containing progeny virus was collected and used for the next round of selection by similar binding, fusion, and growth steps. We observed that viral titers decreased with every round of selection, and the amount of zinc used for the selection was adjusted accordingly. In addition, the virus stocks were amplified every 5 to 6 passages by treatment as described above, but the NH₄Cl was removed from the overnight incubation step. Potential zinc-resistant mutants were plaque purified and retested for zinc resistance, and the E1 protein sequence was determined by reverse transcription, PCR amplification, and sequence analysis, all as previously described (26).

E1 homotrimer assays.

The ability of virus to form E1 homotrimers in the presence of zinc was tested by trypsin resistance (24). Similar to the fusion-infection assay described above, radiolabeled virus was bound to BHK cells on ice for 90 min, washed once in ice-cold buffer Z at pH 7.0, and treated at low pH for 1 min at 37°C with or without 2 mM zinc. Cells were then washed once with buffer Z at pH 7.0 to remove zinc and solubilized in lysis buffer (1% Triton X-100, 5 mM EDTA, 1 μg pepstatin/ml, and 0.1% BSA in phosphate-buffered saline [PBS]). E1 trypsin resistance was assessed by digestion with 200 μg l(tosylamido-2-phenyl)ethyl chloromethyl ketone (TPCK)-trypsin/ml for 10 min at 37°C, followed by quenching with a 3-fold excess of soybean trypsin inhibitor. Parallel control samples were treated with premixed trypsin and soybean trypsin inhibitor and were used to determine the total E1 present. Samples were analyzed by SDS-PAGE and phosphorimaging. The low level of undigested E1 observed in parallel samples treated at pH 7.0 was subtracted for each virus.

RESULTS

Zinc inhibition of fusion in a cell culture system.

To obtain further insights into the mechanism of zinc inhibition, we developed a cell culture system amenable to testing the effects of zinc on virus fusion and infection. At high concentrations, zinc is toxic to cells (19), and thus it was important to ensure that cells were not adversely affected by zinc and that any effects on virus replication could be separated from effects on virus-membrane fusion. We therefore took advantage of our previously described fusion-infection assay (23), a system in which virus is prebound to cells on ice, the bound virus pulsed briefly with low-pH buffer at 37°C to trigger fusion with the plasma membrane, and the infected cells subsequently scored by immunofluorescence. Using this system, we found that inclusion of 2 mM zinc during a 1-min pulse at pH 5.75 inhibited infection by ∼5 logs (Fig. 1). In contrast, a 1-min treatment with 2 mM zinc before or after the low-pH pulse did not affect virus infection. Thus, zinc was not acting via effects on prefusion events such as virus-receptor interaction or postfusion events, such as RNA transcription or translation, and was affecting events during the 1-min low-pH treatment. In keeping with prior liposome and cell-cell fusion results (7, 45), zinc specifically blocked SFV infection through effects on low-pH-triggered virus-membrane fusion.

Fig 1.

Zinc specifically inhibits SFV fusion with cells. wt SFV was bound to BHK cells on ice and treated for 1 min with buffer Z containing 2 mM ZnCl₂ either before (prefusion), during (fusion), or after (postfusion) a 1-min pulse at pH 5.75 at 37°C. Controls were treated in parallel with buffer Z that did not contain ZnCl₂. Infected cells (IC) were scored by immunofluorescence. Data shown are the average and range of 2 experiments.

Concentration and pH dependence of zinc inhibition.

We determined the concentration dependence of zinc inhibition in the fusion-infection assay. Maximal inhibition was observed when 2 mM zinc was present during a pulse at pH 5.75, resulting in a decrease of ∼5 logs in virus infection (Fig. 2). Inhibition did not significantly increase at zinc concentrations of 3 mM or higher (Fig. 2 and data not shown). A decrease of ∼1 log was observed when 100 μM zinc was present during the low-pH pulse. By comparison, using the SFV-liposome fusion assay, a detectable decrease in fusion at pH 5.5 is observed at concentrations of 25 μM zinc, and full inhibition is observed at concentrations of 1 to 2 mM (7).

Fig 2.

Concentration dependence of zinc inhibition of SFV fusion. Serial dilutions of wt SFV were bound to BHK cells on ice and treated with buffer Z at pH 5.75 containing the indicated concentrations of ZnCl₂. Infected cells were scored by immunofluorescence. Data shown are the average and range of 2 to 4 experiments for each condition.

Proton binding triggers the refolding of E1 to the homotrimer and drives the fusion reaction. Zinc is known to compete with protons for binding to histidine, aspartate, and glutamate residues (38). We tested the ability of 2 mM zinc to inhibit fusion at different pH values. While ∼5 to 6 log inhibition of infection occurred at pH 5.75, ∼4 log inhibition was observed at pH 5.5, and almost no inhibition at pH 4.75 (Fig. 3A and data not shown). The decrease in inhibition at lower pH values did not appear to be due to faster fusion kinetics (3) as fusion was inhibited to the same extent when the rate was decreased by carrying out the fusion reaction at 20°C (data not shown). Rather, it appears that at this increased concentration protons compete more effectively for sites of zinc binding. Conversely, as the concentration of zinc is increased at a set pH (Fig. 2), the more likely it is to bind to titratable side chains and inhibit the usual fusion pathway. Thus, the relationship between pH and zinc inhibition probably reflects the overall ionizability of the residue(s) or region(s) involved in the zinc blocking mechanism, and is in keeping with the titration of the histidine side chain in this pH range (e.g., see reference 41).

Fig 3.

Zinc sensitivity and pH dependence of fusion of wt and mutant SFV. Serial dilutions of wt (A) and mutant (B and C) viruses were bound to BHK-21 cells on ice and then treated for 1 min at 37°C in buffer Z at the indicated pH and zinc concentration. Infected cells were scored by immunofluorescence. Data shown are the average and standard deviation of 4 experiments (wt SFV) or the average and range of 2 experiments (H333N, H333Y). The mutant virus titers were normalized to that of wt SFV at pH 5.75 in the absence of zinc.

Selection for zinc escape mutants.

To identify the E1 region(s) responsible for zinc inhibition, we selected for zinc-resistant SFV mutants by triggering virus fusion with the plasma membrane in the presence of zinc. Infected cells were then incubated overnight, and progeny viruses were collected and used for the next round of selection. After 20 to 26 cycles of selection at pH 5.75, mutants were plaque purified and tested for zinc inhibition at different pH values. All 9 mutant stocks showed increased resistance to zinc at pH 5.5 (Fig. 3 and data not shown). Sequence analysis of 5 of the mutants revealed that all carried a single amino acid change at E1 His333, with 2 mutants containing a change of His333 to asparagine (H333N) and 3 to tyrosine (H333Y). The majority of alphaviruses contain His at this position, although Leu, Ala, Pro, Met, and Glu are also found.

Characterization of H333N and H333Y mutants.

Since zinc inhibition of wt virus was less effective at lower pH values as discussed above, it was important to determine whether the mechanism of escape in mutant viruses was an indirect effect of a change in the pH dependence of mutant virus fusion. In the absence of zinc, fusion of the H333N and H333Y mutants showed pH dependence comparable to that of wt SFV, with a threshold of fusion at ∼pH 6.2 and maximal fusion at pH 5.75 (Fig. 3A to C, light gray bars). This is in keeping with prior results which showed that alanine substitution of H333 did not change the pH-dependence of SFV fusion (31). While zinc strongly inhibited wt, H333N, and H333Y fusion at pH 6.2, the mutants showed increased zinc resistance when fusion was tested at lower pH values. At pH 5.75, fusion of wt virus was inhibited by ∼6 logs by 2 mM zinc while the H333Y and H333N mutants were inhibited by ∼3 logs. At pH 5.5, fusion of wt virus was inhibited by ∼4 logs while neither mutant showed significant inhibition (Fig. 3A to C). Thus, mutation of H333 to asparagine or tyrosine directly affected the zinc sensitivity of virus fusion.

Nickel inhibition of SFV fusion.

Zn²⁺ is a metal with a natural affinity for binding histidine and the free sulfhydryl group of cysteine and can also interact with carboxylates on aspartate and glutamate (reviewed in reference 1). While the cysteines in the E1 ectodomain are in disulfide bonds (34), H333 and/or other ligands could be involved in the zinc-E1 interaction. Nickel is a transition metal with similar ligand affinity as zinc except that it rarely interacts with aspartate or glutamate (2). Therefore, we used nickel to address the role of E1 histidine residues in inhibition. As shown in Fig. 4 (gray bars), nickel inhibited wt virus fusion at pH 5.75, with a decrease of ∼2 logs at 5 mM NiCl₂ and ∼3 logs at 10 mM NiCl₂. Inhibition by nickel required higher concentrations and was less efficient than inhibition by zinc (Fig. 4 versus Fig. 2). Nickel inhibition showed a similar competitive effect of proton concentration as that observed with zinc, with 5 mM NiCl₂ causing a decrease in infection of ∼4 logs at pH 6.0 versus only ∼0.5 logs at pH 5.5 (data not shown). The zinc-resistant H333Y mutant was still sensitive to inhibition by nickel (Fig. 4, black bars), suggesting differences in the interaction of nickel and zinc with E1 333Y. However, the H333N mutant was resistant to NiCl₂ at concentrations of 3 to 10 mM (Fig. 4, white bars). The fact that the H333N mutation caused reduced inhibition by both zinc and nickel suggested that these two metals block fusion by a similar mechanism. Harder metals such as calcium and magnesium are known to bind to carboxylate groups of aspartate and glutamate rather than histidine (2), and these metals do not inhibit alphavirus fusion (7, 42). Thus, together the data support a role for E1 histidine residue(s) in the mechanism of zinc inhibition and suggest that H333 is involved in zinc interaction.

Fig 4.

Effect of nickel on fusion of wt and mutant SFV. Serial dilutions of wt and mutant viruses were bound to BHK-21 cells on ice and treated for 1 min at 37°C with buffer Z at pH 5.75 containing the indicated concentration of NiCl₂. Infected cells were scored by immunofluorescence. Data shown are the average and range of 2 experiments, with the mutant virus titers in the absence of nickel normalized to that of wt SFV.

Role of conserved E1 histidines in zinc inhibition.

We used a panel of SFV alanine substitution mutants to test the contribution of all of the conserved E1 histidine residues to zinc inhibition (Fig. 5A). These viruses were generated by site-directed mutagenesis of the SFV infectious clone and previously sequenced and characterized in detail. H230 is located at the tip of DII adjacent to the fusion loop. As the H230A mutation is lethal, we used a virus containing this mutation plus the second site rescue mutation V178A (4, 5). H125 is located in the hinge region between DI and DII. The H125A mutation has a slightly lower pH threshold for fusion but shows minimal effects on fusion and infection (31). H3 is located in DI and has important interactions with the DI-DIII linker region during trimerization (46). The H3A mutant shows decreased growth and fusion but is rescued by the second site D284A mutation (31). H331 is located on DIII and is part of a cluster of 3 histidine residues (H333, H331, and H18) within the interface of DIII and DI in the prefusion conformation of SFV E1 (Fig. 5A). The H331A mutation does not detectably affect virus fusion, pH dependence, or infection, either alone or in combination with an H333A mutation (31).

Fig 5.

Role of conserved E1 histidine residues in zinc inhibition of fusion. (A) Structure of the prefusion conformation of E1 (Protein Data Bank, 2ALA). E1 regions are colored as follows: DI in red, DII in yellow, DIII in blue, the fusion loop in orange, the DI-DIII linker in purple, and the stem in gray. The conserved histidine residues (H3, H125, H230, H331) and the histidine mutated in the zinc-resistant isolates (H333) are indicated as stick structures in green. Note the cluster of 3 histidines (H331, H333, H18) predicted to be a zinc-binding site, as discussed in the text. Figure prepared using PyMOL (8). (B) wt SFV and the indicated SFV mutants were bound to BHK cells on ice and treated for 1 min at 37°C with buffer Z at pH 5.5 containing 2 mM ZnCl₂ as indicated. Infected cells were scored by immunofluorescence. Data shown are the average and range of 2 experiments, with the mutant virus titers in the absence of zinc normalized to that of wt SFV. Infection by H230A/V178A in the presence of 2 mM zinc was below the limit of detection and thus was inhibited by at least 4 logs.

wt and mutant viruses were tested in fusion-infection assays at pH 5.5 in the presence or absence of 2 mM zinc (Fig. 5B). wt SFV was inhibited by ∼3 logs under these conditions. SFV mutants containing alanine substitutions of any of the conserved histidines were also strongly inhibited, indicating that these histidine residues did not play a key role in mediating zinc inhibition. The only mutant that showed increased resistance to zinc was the H331A/H333A double mutant. As the H331A single mutant was strongly inhibited by zinc, zinc resistance in the double mutant was due to the H333A mutation. Thus, replacement of H333 with A, Y, or N causes zinc resistance, while the other conserved E1 histidines do not significantly contribute to inhibition by zinc.

Zinc inhibition of E1 trimerization.

How do mutations at E1 H333 decrease SFV's sensitivity to zinc during virus membrane fusion? Previous studies of virus-liposome fusion showed that zinc does not affect E2-E1 heterodimer dissociation or E1 membrane insertion and suggested that zinc inhibited formation of the SDS- and trypsin-resistant homotrimer (7). We tested the effect of zinc on the formation of the homotrimer during virus-liposome fusion. In our hands, the presence of zinc destabilized the homotrimer in SDS even when zinc was added after generation of the homotrimer at low pH. Zinc also had nonspecific effects on the trypsin-resistance assay (data not shown). We therefore used radiolabeled SFV in a fusion-infection assay, which allowed removal of zinc by washing the cells after the low-pH pulse. E1 homotrimer formation was then assayed by trypsin digestion in 1% Triton X-100 at 37°C, conditions that completely digest the prefusion E1 monomer (13, 33). Radiolabeled wt and H331A/H333A mutant viruses were prepared from the SFV infectious clone, bound to cells, and treated at pH 5.5 in the presence or absence of 2 mM zinc. The trypsin-resistant E1 homotrimer was quantitated by SDS-PAGE and phosphorimaging and expressed as a percentage of the total E1 present in the fusion-infection reaction (Fig. 6). In the absence of zinc, trypsin-resistant trimer formation by mutant and wt SFV was similar (30% versus 33%), confirming that the mutations did not affect the efficiency or biochemical properties of the E1 homotrimer. wt virus homotrimer formation was significantly decreased by zinc treatment, being ∼3.7-fold lower than control conditions. In contrast, homotrimer formation by the mutant was relatively resistant to zinc, with a level ∼1.4-fold lower than control.

Fig 6.

Effect of zinc on wt and mutant E1 homotrimer formation. Radiolabeled wt and H331A/H333A mutant SFV were prebound to BHK cells and treated for 1 min at 37°C with pH 5.5 buffer Z containing 2 mM ZnCl₂ as indicated. Cells were washed and lysed, and the samples were digested with trypsin and evaluated by SDS-PAGE and phosphorimaging. The amount of trypsin-resistant E1 homotrimer is shown as a percentage of the total E1 in the sample. Data shown are the average and range of 2 experiments for wt SFV and 3 experiments for H331A/H333A. A two-tailed Student t test showed a significant difference between wt and mutant SFV in the presence of 2 mM zinc (P = 0.006) and no significant difference between wt and mutant SFV in the absence of zinc (P = 0.30).

DISCUSSION

Here we used a virus-cell fusion system to analyze the mechanism by which zinc inhibits low-pH-triggered SFV fusion and to select for viral zinc resistance. Zinc was a very efficient inhibitor of the fusion-infection reaction, producing ∼5-log inhibition at a concentration of 2 mM zinc and pH of 5.75. The pH dependence of zinc inhibition suggested that histidine side chain(s) could be involved in zinc interaction. Out of 5 independently selected virus mutants and 5 engineered histidine-alanine substitutions, only mutations at E1 H333 produced significant SFV zinc resistance. While H333 mutants remained sensitive to zinc at pH 6.2, they escaped inhibition completely at pH 5.5, conditions under which wt viruses were still inhibited by ∼4 logs. Zinc blocked formation of the trypsin-resistant E1 homotrimer in wt SFV, while an alanine substitution at H333 significantly reversed zinc's effect on trimerization. Thus, zinc's inhibition of SFV homotrimer formation and fusion involves its interaction with E1 H333 on DIII.

Role of histidines in inhibition of fusion by zinc and nickel.

The binding properties of several transition metals, including Ni²⁺, Cu²⁺, and Cd²⁺ are similar to those of Zn²⁺ and comparison with these transition metals has been used to study zinc interactions in proteins (reviewed in reference 1). Although tests of Cu²⁺ and Cd² were precluded due to solubility and toxicity issues (data not shown), SFV fusion was inhibited by both zinc and nickel (this study) and is resistant to calcium and magnesium (7, 42). As the E1 mutation H333N bypassed inhibition by both zinc and nickel and both metals inhibited wt SFV fusion in a pH-dependent manner, zinc and nickel likely target the same E1 region.

Subtle differences do exist between the zinc and nickel inhibition mechanisms. The binding of a metal to ligands in a matrix is a tug-of-war between fulfilling the most optimal coordination geometry for the metal and the most stable conformation of the protein (reviewed in reference 44). Thus, the interactions between a specific metal and ligands in a protein are unique. While zinc was a potent inhibitor of fusion, nickel was less efficient and showed a more marked pH dependence, with little inhibition observed at pH 5.5 (data not shown). In addition, while the E1 H333Y mutant was zinc-resistant, it remained sensitive to nickel. The differences in the coordination properties between zinc and nickel likely contribute to these differences in inhibition, even though their overall mechanisms appear similar.

Effects of zinc on fusion of other alphaviruses.

Fusion of the alphaviruses Sindbis virus and Chikungunya virus was also inhibited by zinc (data not shown) (45). While Sindbis virus E1 contains a histidine at position 333, Chikungunya virus E1 333 is methionine, a residue not commonly observed to bind zinc in nature (1). Thus, at least in the context of the Chikungunya virus E1 sequence, H333 is not strictly required for inhibition. Computer-based modeling, as discussed below, did not predict a strong zinc-binding site on Chikungunya virus E1. We also found that Chikungunya virus fusion is inhibited by nickel (data not shown). We have not further explored the interactions by which these metals inhibit Chikungunya virus fusion.

Model of the zinc-inhibited state.

Inhibition of alphavirus fusion requires the presence of zinc during low-pH triggering. The block occurs after E2-E1 dimer dissociation and E1-membrane insertion but prior to stable homotrimer formation (7). Does zinc act by blocking E1 at a step along the pathway of productive trimerization or does it shift E1 toward a nonproductive refolding pathway? When target membranes are present during low-pH treatment, zinc traps E1 in a membrane-inserted conformation that can still proceed to mediate fusion when zinc is removed and a second low-pH treatment is applied (45). Zinc inhibition appears consistent with the capture of either the membrane-inserted E1 monomer or an extended E1 trimer intermediate. Both of these E1 fusion intermediates are trypsin sensitive and occur prior to the fold-back of DIII (33).

Although DIII H333 does not play a significant role in the normal process of trimerization and fusion or in virus pH dependence (31), our data demonstrate that substitutions at H333 bypass the zinc-inhibited E1 intermediate and allow fusion to proceed. In the prefusion structure of E1, H333 is located close to H331 and H18 in the interface between DIII and DI (Fig. 5A). During fusion, the movement of DIII toward the target membrane disrupts this interface. We used a program developed by Ebert and Altman (9) to model potential zinc-binding sites in the SFV E1 prefusion monomer and postfusion trimer (data not shown). Zinc interactions were predicted in the monomer within the H333, H331, and H18 cluster (Fig. 5A), while no zinc interactions were predicted in the postfusion trimer. This suggests a model in which zinc binding to H333 and other ligands prevents the movement of DIII and thus the formation of stable trimers. Mutation of H333 would remove one of the predicted zinc-binding ligands, causing the mutants to be less inhibited by zinc and allowing DIII movement and fusion. This model is clearly oversimplified since while H331 was predicted to form part of the zinc interaction site, the E1 H331A mutant was as zinc sensitive as wt SFV. Thus, other presumably nonhistidine ligands on E1 must also play a role in interacting with zinc and preventing the critical refolding of E1 during SFV fusion.

Insights into antiviral therapy.

Zinc is a strong inhibitor of the fusion step of several alphaviruses. Is there potential for its use as an inhibitor during normal virus endocytic entry and infection? Intracellular zinc concentrations are tightly regulated since small changes in the free zinc concentration can have pleiotropic effects on many cellular processes and since excess zinc is toxic to cells (reviewed in reference 19). Excess free intracellular zinc is rapidly buffered by binding to cellular proteins such as the metallothioneins or is removed via zinc transporters (19). Thus, the use of zinc as an inhibitor would presumably require its release only at the site of alphavirus fusion; such a delivery method would be technically challenging. In addition, as shown here, alphaviruses can mutate to become relatively zinc resistant. Given these issues, our results are important in providing insights into the mechanism of zinc inhibition rather than in suggesting zinc as an antiviral therapy per se. However, it is possible that a small molecule that targeted the DIII H333 region could be a feasible strategy to recapitulate the inhibitory mechanism of zinc.