Membrane binding mechanism of an RNA virus-capping enzyme.

The RNA replication complex of Semliki Forest virus is bound to cytoplasmic membranes via the mRNA-capping enzyme Nsp1. Here we have studied the structure and liposome interactions of a synthetic peptide (245)GSTLYTESRKLLRSWHLPSV(264) corresponding to the membrane binding domain of Nsp1. The peptide interacted with liposomes only if negatively charged lipids were present that induced a structural change in the peptide from a random coil to a partially alpha-helical conformation. NMR structure shows that the alpha-helix is amphipathic, the hydrophobic surface consisting of several leucines, a valine, and a tryptophan moiety (Trp-259). Fluorescence studies revealed that this tryptophan intercalates in the bilayer to the depth of the ninth and tenth carbons of lipid acyl chains. Mutation W259A altered the mode of bilayer association of the peptide and abolished its ability to compete for membrane association of intact Nsp1, demonstrating its crucial role in the membrane association and function of Nsp1.

The RNA replication complex of Semliki Forest virus is bound to cytoplasmic membranes via the mRNA-capping enzyme Nsp1. Here we have studied the structure and liposome interactions of a synthetic peptide 245 GSTLYTESRKLLRSWHLPSV 264 corresponding to the membrane binding domain of Nsp1. The peptide interacted with liposomes only if negatively charged lipids were present that induced a structural change in the peptide from a random coil to a partially ␣-helical conformation. NMR structure shows that the ␣-helix is amphipathic, the hydrophobic surface consisting of several leucines, a valine, and a tryptophan moiety (Trp-259). Fluorescence studies revealed that this tryptophan intercalates in the bilayer to the depth of the ninth and tenth carbons of lipid acyl chains. Mutation W259A altered the mode of bilayer association of the peptide and abolished its ability to compete for membrane association of intact Nsp1, demonstrating its crucial role in the membrane association and function of Nsp1.
Electron microscopic studies of Alphavirus-infected cells have revealed specific cytoplasmic structures designated as cytoplasmic vacuoles (1,2). Electron microscopic autoradiography suggested that cytoplasmic vacuoles might be the sites of virus-specific RNA synthesis (2). Later it was shown that cytoplasmic vacuoles are modified endosomes and lysosomes (3,4). Similar structures have been detected in rubella virus-infected cells (5,6). The following two obvious questions arose. How was the replication complex bound to the membrane, and how was it targeted specifically to the endo/lysosomal membranes? To answer these questions, we have expressed the virus-specific RNA replicase components, i.e. the nonstructural proteins Nsp1-4 of Semliki Forest virus (SFV) 1 in different cells (7,8). Of these proteins only Nsp1 (537 amino acids) was found to attach to the plasma membrane and to some extent also to endosomal/lysosomal membranes (8 -10). The strength of membrane association of Nsp1 was as strong as that of integral membrane proteins due to post-translational palmitoylation of the cysteine residues 418 -420. Mutation of these cysteines to alanines resulted in peripheral-type membrane association (10). Surprisingly, infection of cells with SFV coding for nonacylated Nsp1 resulted in a normal virus production appearance of typical cytoplasmic vacuole structures (11).
Membrane binding of Nsp1 in the absence of acylation was studied in more detail by producing the protein either in Escherichia coli or by translation in vitro (12). Flotation tests in discontinuous sucrose gradients indicated that the synthesized protein was associated with the plasma membrane of E. coli. When the membrane was solubilized with detergents, Nsp1 lost its methyltransferase and guanylyltransferase activities (13)(14)(15). Both activities were restored upon the addition of either liposomes or detergent micelles containing phosphatidylserine or other anionic phospholipids. To map the membrane binding domain within Nsp1, we used deletion and point mutagenesis and in vitro translation in the presence of phosphatidylserine-containing liposomes. By these means, a specific lipid binding region in the middle of Nsp1 was identified. A synthetic peptide corresponding to residues 245-264 was able to compete for the binding of in vitro translated Nsp1 to liposomes (12).
Here we have studied the mechanism of membrane association of the putative membrane binding peptide of Nsp1 and its mutant derivatives using fluorescence and circular dichroism spectrometry. The three-dimensional structure of the peptide was determined by NMR spectroscopy. The results suggest that the membrane association of the peptide and, thus, that of Nsp1 is mediated by polar interactions between positively charged amino acid residues and the negatively charged head groups of the anionic phospholipids. Also, hydrophobic interactions, particularly those between the single tryptophan residue and the lipid acyl chains, were critical. We expect that other positive-strand RNA viruses have similar mechanisms for membrane binding of their replication complexes (see 12).
Peptide Synthesis-The peptides were synthesized at the Division of Biochemistry, Department of Biosciences and at the Haartman Insti-* This work was supported by Academy of Finland Grant 8397 and by the Technology Development Center (TEKES). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The tute Peptide and Protein Laboratory of the University of Helsinki using the Applied Biosystems model 433A peptide synthesizer and 9-fluorenylmethoxycarbonyl chemistry. The peptides were purified on a octadecylsilica reverse-phase column, and their purity and mass were checked by matrix-assisted laser desorption-ionization time-of-flight mass spectrometry.
Preparation of Liposomes-The lipids dissolved in chloroform/methanol (9:1) stocks were mixed in small glass tubes and dried under nitrogen flow. Any residual solvent was removed by keeping the residue in a vacuum desiccator for 1 h. The lipids were hydrated by adding 5 mM Tris, pH 7.5, and sonicated on ice with a Branson sonicator (50 W output) equipped with a microtip until the solution clarified (ϳ10 min). The resulting small vesicles were stored at 4°C and used the following day. Multilamellar liposomes for the flotation assays were prepared by vortexing the dried lipids into 50 mM Tris buffer, pH 7.5, containing 50 mM NaCl.
Protein Expression and Flotation Analysis-Combined in vitro transcription-translation was performed with the TnT T7 kit (Promega) using an appropriate circular plasmid containing the Nsp1 gene under the control of the T7 promoter. Nsp1 was translated in vitro in the presence of multilamellar liposomes (1.67 mM lipids), and the translation mixtures were subjected to flotation in discontinuous sucrose gradients (12). The sucrose solutions were prepared in a buffer containing 100 mM NaCl and 50 mM Tris, pH 7.5. When a synthetic peptide (0.5 mM) was present in the translation mixtures, it was also included in the 67, 60, and 50% sucrose layers at 50 M concentration. The gradients were centrifuged overnight at 35,000 rpm in a SW50.1 rotor (Beckman) at 4°C. The vesicles and the associated proteins floated to the top of the 50% sucrose layer, whereas the soluble proteins remained in the 60% layer. Fractions were collected from the top and analyzed by SDSpolyacrylamide gel electrophoresis followed by autoradiography.
Fluorescence and CD Spectroscopy-The fluorescence spectra were recorded on a PTI QuantaMaster fluorescence spectrophotometer with the emission and excitation slits set to 2 nm. The samples were kept in a 1-cm quartz cuvettes thermostatted to 25°C, and they contained 5.0 M peptide and 100 M lipid in 5 mM Tris buffer, pH 7.5, in a total volume of 2 ml. CD spectra were obtained at 25°C with a Jasco 720 spectropolarimeter equipped with a 1-mm quartz cell. The scan rate was 50 nm/min, and 10 spectra were averaged. The samples were in a total volume of 200 l in 5 mM Tris buffer, and the peptide and lipid concentrations were 50 M and 1 mM, respectively, unless otherwise indicated.
NMR Spectroscopy-Samples of freeze-dried peptide were dissolved in D 3 -trifluoroethanol/H 2 O (7:3, v/v) to obtain 1 mM solution. The samples were loaded to 220 l Shigemi micro cells and pH (not corrected for the deuteron or solvent effects) was adjusted to 6.0 by adding dilute HCl. The NMR spectra were acquired with 600 and 800 MHz Varian Unity NMR spectrometers. Homonuclear two-dimensional spectra, correlation spectroscopy (COSY), relayed COSY, total correlation spectroscopy (TOCSY, with mixing times of 30, 55, and 180 ms), and nuclear Overhauser enhancement (NOESY, with mixing times of 50, 100, 200, and 400 ms) data were collected at 10°C and processed using the Felix 98 software.
Structure Generation-Spin systems were assigned from the through-bond correlation spectra, and the sequence-specific assignments were deduced from the NOEs between the adjacent spin systems. Distance restraints were extracted from the homonuclear NOE data by fitting a second-order polynomial to integrated cross-peak volumes of the NOE series. The intra-methylene and helical NH i -NH i ϩ1 NOEs were used for the calibration. Distances were given with ϩ30 and Ϫ40% uncertainties. When a distance could not be extracted from the build-up curve, either owing to an overlap, a poor signal-to-noise ratio, or other disturbances, the distance was restrained to be not more than 5.0 Å. The upper bounds were extended by 1.0 Å for each pseudo-atom in methyl groups and 1.5 Å for pseudo-atoms in aromatic rings. Backbone -dihedral angles characterized by small JH N H␣, measured from the correlation (COSY) spectra, were restrained to the helical conformation (Ϯ30 degrees) on the basis of the Karplus relation. Dihedral angles characterized by intermediate JH N H ␣ or hydrogen bonds were not constrained. The spectra were assigned, and the distance restraints were generated using the Felix 98 software. Structures were generated by DYANA 1.5 using the torsion angle dynamics approach (18). A final set of structures was analyzed for root mean square density, energy, and backbone dihedrals. The coordinates and NMR restraint files of Semliki Forest virus Nsp1 membrane binding peptide will be deposited to the Protein Data Bank (code 1FW5).

Inhibition of Membrane Binding of Nsp1 by Synthetic
Peptides-The peptide corresponding to the previously assigned membrane binding segment (G 245 STLYTE-SRKLLR-SWHLPSV 264 ) of SFV Nsp1 (12) and three variant peptides with single amino acid replacements (R253E, K254E, and W259A) were synthesized. In addition, a peptide containing the same amino acids as the wild type, but in a random sequence (Fig. 1B), was prepared. Secondary structure prediction (19,20) suggested an ␣-helical conformation for residues Thr-250 -Ser-258 of the wild type peptide. At pH 7.5, the net charge of the wild type, the W259A variant, and the random peptide should be ϩ2, and that of R253E and K254E variants should be close to zero.
The wild type peptide, variants W259A and R253E, and the random peptide were tested for their ability to compete with Nsp1 for binding to negatively charged multilamellar liposomes. Nsp1 was translated in vitro in the presence of PS:PC (1:1) liposomes and one of the peptides. The fraction of Nsp1 associating with the vesicles was then determined by subjecting the samples to centrifugation in discontinuous sucrose gradients (12). The wild type peptide inhibited the association of Nsp1 with liposomes by 64% (Fig. 2C), whereas the random peptide (Fig. 2B), W259A (Fig. 2D), or R253E (not shown) variant peptides did not inhibit the binding of Nsp1 to liposomes.
As can be seen in Fig. 2, a relatively large amount of wild  Rost and Sander (19,20). The bold region of SFV wild type peptide represents ␣-helix in the presence of trifluoroethanol, as indicated in Fig. 8. SAG, Sagiyma; EEE, eastern equine encephalitis; WEE, western equine encephalitis; Igbo, Igbo Ora; SIN, Sindbis; Aura, Barmah, Barmah Forest; RRV, Ross River; ONN, O'nyon-nyong. B, previously identified mutations in the binding peptide region of SFV Nsp1 protein that affect the liposome binding of in vitro synthesized Nsp1 protein and its methyltransferase activity, as determined previously (12). The bold peptide sequences have been synthesized chemically and used in this study. The random peptide consisted of the same amino acids as the SFV wild type (wt) peptide. type peptide was required to successfully compete with the binding of Nsp1 to the liposomes. In these experiments the surface of the liposomes must be covered by the peptide to obtain inhibition, and the liposomes were in excess to the in vitro translation product.
Binding of the Peptides to Lipid Vesicles with Varying Charge-To study association of the peptides with membranes of varying charge, we made use of tryptophan fluorescence properties (21,22). The addition of PC vesicles containing 30 mol % of a negatively charged phospholipid, i.e. phosphatidylglycerol, PS, or phosphatidic acid, caused a major blue shift of tryptophan emission and a marked increase in the fluorescence intensity (Fig. 3). In contrast, a minor shift of tryptophan emission was observed upon the addition of vesicles consisting of zwitterionic phospholipids only. Since a blue shift and an increase in fluorescence intensity are considered to indicate penetration of a tryptophan residue into a lipid bilayer (e.g. Refs. 21 and 23), these data provide strong evidence that the wild type peptide binds avidly to negatively charged membranes but not to uncharged membranes. The variant peptides behaved similarly, except that the effects on the fluorescence parameters were somewhat less pronounced (see below).
To study further the effect of the density of negative charge on membranes for peptide association, the PS content of the vesicles was varied from zero to 50 mol % (Fig. 4A). With the wild-type peptide, the blue shift increased until PS content reached 35-40 mol % and then leveled off. With the variants, a threshold concentration of about 10 mol % of PS was necessary before any significant shift was observed. In addition, the blue shift was markedly smaller than with the wild type at all PS concentrations (Fig. 4A). The effect of PS content on the intensity of tryptophan fluorescence is shown in Fig. 4B. In case of the wild-type peptide, the intensity increased linearly with PS content until 20 -30 mol % and then declined. With the vari-ants, a minimum of ϳ10 mol % of PS was again required to obtain a significant effect, and the maximal effect was much smaller than with the wild type (Fig. 4B). As with the wild type peptide, the intensity decreased above 30 mol % of PS. The reason for this is not clear, but obviously it does not indicate that less peptide would be binding to the vesicles, since the blue shift did not level off until 40 -50 mol % (Fig. 4A). This decrease could result from self-quenching tryptophan-tryptophan interactions that become significant when the surface concentration of the peptide increases with increasing PS content. The addition of increasing amounts of NaCl in the peptide-PC/PS-vesicle solution diminished the blue shift and abolished it completely when the concentration of NaCl reached 100 mM (Fig.  4C). The effect of NaCl concentration was similar with the wild type and the variant peptides. These data support the importance of ionic interactions for the binding of the peptides to negatively charged membranes.
Depth of Tryptophan Penetration to Membrane-The blue shift observed upon interaction of the peptides with the negatively charged vesicles indicates that the tryptophan moiety penetrates into the vesicle bilayer (23). To estimate the depth of penetration, we made use of PC species containing a brominated fatty acid in the sn-2 position. Bromine quenches tryptophan fluorescence by a collision mechanism, and therefore, it is possible to obtain information on the depth of tryptophan penetration by using a set of phospholipids with the bromines attached to different acyl carbons (23)(24)(25)(26)(27)(28). Accordingly, we prepared vesicles consisting of POPS (50 mol %) and a PC species (50 mol %) with bromines in carbons 6 and 7 (6,7dibromo-PC), 9 and 10 (9,10-dibromo-PC), or 11 and 12 (11,12dibromo-PC) of the sn-2 acyl chain, mixed them with the peptide, and then recorded the tryptophan emission spectrum. As shown in Fig. 5A, the tryptophan fluorescence of the wild-type peptide was quenched most effectively with 9,10-dibromo-PC. Quenching by the 6,7-and 11,12-dibromo-PC species was clearly less efficient. These findings indicate that the tryptophan of the wild-type peptide penetrates to a depth of the bromines attached to carbons 9 and 10 of the PC sn-2 acyl chain. The 9,10-dibromo-PC species was also the most efficient quencher of K254E (Fig. 5B) and R253E (Fig. 5C) variants. However, the fluorescence of the variants was quenched less efficiently with each dibromo-PC species than that of the wild type peptide. This finding implies that the tryptophan residue in the variants penetrates to a similar depth as that of the wild type, but the fraction of the peptide bound to the vesicles was smaller than in the case of the wild type peptide.

Effect of Peptides on the Lateral Mobility of PS and PC-To
study the effect of the peptides on the mobility/lateral organization of the vesicle lipids, we included 5 mol % of either pyrene-labeled PS (PyrPS) or PC (PyrPC) in PS/PC (3:7) vesicles and then determined the effect of the peptides on the pyrene excimer to monomer fluorescence intensity ratio (E/M). The E/M ratio is an indicator of the lateral mobility of the pyrene phospholipid (16,17). Mixing of the wild type peptide with vesicles containing PyrPS caused a significant (10%) decrease in the E/M ratio (Fig. 6, A and B). This phenomenon was not seen with the variant peptides (Fig. 6B). In contrast, a slight increase in the E/M ratio was observed when the wild type peptide or the variant peptides were mixed with vesicles containing PyrPC (Fig. 6C). These results suggest that the wild type peptide interacts preferentially with PS molecules, thus selectively retarding their lateral mobility, whereas interaction of the variant peptides with PS-containing vesicles was weaker. It is intriguing that the W259A variant, even if it contains the same positively charged amino acid residues as the wild type, had no effect on the E/M ratio of PyrPS-containing vesicles.
Conformational Changes-CD spectroscopy was used to determine whether the peptides undergo a conformational change upon interaction with the lipid vesicles. In buffer or in the presence of PC vesicles, all peptides gave spectra with a minimum close to 200 nm and an overall shape suggesting a random coil conformation, as shown for the wild type peptide in Fig. 7A. When PS was included in the vesicles, the spectrum changed remarkably, with new minima appearing at 208 and 220 nm (Fig. 7A). These changes imply that the peptide adopts predominantly an ␣-helical conformation in the presence of PS-containing vesicles. A maximal effect was obtained when the concentration of PS reached 20 -30 mol % (Fig. 7A). Identical results were obtained when PS was replaced by another acidic phospholipid such as phosphatidic acid or phosphatidylglycerol but not when it was replaced by the zwitterionic phospholipid PE. Similar results were obtained for the variant peptides R253E, K254E, and W259A (data not shown), whereas the CD spectrum of the random peptide was not significantly affected in the presence of 50 mol % of either PS or PC in the vesicles (Fig. 7C).
The CD spectroscopy of the peptides was done also in trifluoroethanol, which had to be used for the NMR studies (29). A gradual change from a random coil to an ␣-helical conformation was observed when increasing amounts of trifluoroethanol was added to the wild type peptide in buffer (Fig. 7B). The maximal effect was obtained already with 30% trifluoroethanol. With the variant peptides, similar results were obtained (data not shown). These results indicate that the conformation of the peptides in 30% trifluoroethanol is very similar to that of the liposome-bound peptides.
Structure of the Wild Type Peptide-NMR spectroscopy was used to assess the structure of the wild type peptide. In water, the wild type peptide was predominantly in a random-coil conformation as indicated by the typical crowding of the NH resonances in fast exchange with water (30). This finding agrees with the CD data (see above). We attempted to determine the structure of the wild type peptide in the presence of lipid vesicles, but this was not feasible due to excessive spectral broadening. This broadening, observed also in the presence of SDS micelles, probably resulted from aggregation of the peptide in the presence of the vesicles or with vesicles. Such vesicle-induced peptide aggregation has been observed previously (31).
Because of these difficulties, we decided to perform the NMR measurements in trifluoroethanol/water (3:7 v/v) solution. According to CD measurements (see above) the peptide adopts a similar, i.e. a largely ␣-helical conformation in trifluoroethanol/ water solutions, as in the presence of negatively charged liposomes. The three-dimensional structure of the wild type pep-tide in 3/7 (v/v) trifluoroethanol/water is shown in Fig. 8. The overall structure is compatible with the CD data, indicating that in the presence of trifluoroethanol the peptide forms a slightly bent ␣-helical structure. This helix has an obvious amphiphilic character. One face consist of the hydrophobic residues Leu-248, Leu-255, Leu-256, Leu-261, Val-264 and residues Ser-252 and Trp-259 (Fig. 8B). The other face consists of polar residues mainly, i.e. the positively charged residues Arg-253, Lys-254, Arg-257 and the negatively residue Glu-251. The positively charged residue Arg-253, shown to be important for the interaction with negatively charged membranes (see above), lies close above the hydrophobic cluster consisting of residues Leu-55, Leu-256, and Trp-259. The hydrophobic surface of the peptide is well defined, whereas the hydrophilic surface is less so due to considerable mobility of the amino acid side chains (Fig. 8C). DISCUSSION Recently, we identified a short sequence of the Semliki Forest virus RNA-capping protein Nsp1 that appeared to be responsible for the binding of the protein to the E. coli plasma membrane and to anionic liposomes (Fig. 1B) (12). Here, we have studied the mechanism of membrane binding of a synthetic peptide corresponding to the binding sequence of Nsp1. Parallel studies were carried out with variant peptides with single amino acid replacements corresponding to Nsp1 mutant proteins that lacked enzymatic activity and had a reduced membrane affinity (Fig. 1). All the peptides occurred as a random coil in buffer solutions but attained a partly ␣-helical conformation in the presence of anionic liposomes as well as in 30% trifluoroethanol. This conformational change may reflect the behavior of the wild type Nsp1 protein, which is enzymatically active in the presence of membranes or mixed micelles of detergent and anionic phospholipids but loses activity in the presence of detergents (12). The structure of the wild type binding peptide, determined by NMR spectroscopy in 30% trifluoroethanol, revealed an amphiphilic ␣-helix in which hydrophobic residues concentrate on one side and the polar residues residues concentrate on the other (Fig. 8). As similar CD spectra were obtained both in trifluoroethanol and in the presence of liposomes, we suppose that the NMR structure of the binding peptide is similar in both environments, with the possible exception of polar residues as discussed below.
The tryptophan fluorescence data provide strong evidence that the wild type peptide binds avidly to vesicles containing negatively charged phospholipids but not to vesicles consisting of zwitterionic phospholipids only. The positively charged amino acid residues of the peptide obviously play an important role in this interaction, since the R253E and K254E variants appeared to interact differently with negatively charged vesicles. Other findings supported the importance of ionic interactions in the association of the peptide with membranes. First, the peptide is released from the vesicles upon the addition of salt (Fig. 4C). Second, it adopts an ␣-helical conformation only in the presence of negatively charged but not uncharged vesicles (Fig. 7). Third, the wild type peptide reduces lateral diffusion (excimer formation) of PyrPS but not that of PyrPC (Fig.  6). NMR analysis of the wild type peptide indicated that the side chains of Arg-253 and Lys-254 are not well localized (Fig.  8C), suggesting that they are quite mobile and thus capable of interacting with acidic phospholipids in vivo. We assume that Arg-257 would also interact with anionic phospholipids, since mutation R257E inhibits the enzymatic activities of Nsp1 protein and its binding to liposomes (12).
A marked blue shift and enhancement of tryptophan fluorescence of the wild type peptide was observed upon addition of negatively charged vesicles, indicating that the tryptophan moves to a less polar milieu (22,32). This conclusion is supported by the efficient quenching of the tryptophan fluorescence by bromines attached to a PC acyl chain. The quenching data indicated that tryptophan 259 of the wild type peptide penetrates approximately to the level of 9 -10 carbon atoms of the acyl chains. Placing tryptophan 259 to this level would bring leucines 248, 255, 256, and 261 as well as Val-264 in contact with the acyl chains. Assuming that the helix of wild type peptide lies nearly parallel to the bilayer surface, one can estimate that the maximum distance from the charged side chain groups of Arg-253 to the ␣-carbon of Leu-256 is about 10 Å and about 12-15 Å to the aromatic carbons of Trp-259. These estimates comply with the tryptophan quenching data (Fig. 5) placing the tryptophan moiety close to the acyl carbons 9 and 10, i.e. 12 and 13.5 Å below the phospholipid polar head groups.
This type of membrane association of proteins, i.e. only one leaflet of a lipid bilayer has been designated as monotopic (33). Alignment of Nsp1 s of the Alphavirus family suggests that a similar amphipathic ␣-helix is present in each protein (34). The amino acid residues that form the hydrophobic surface of the Nsp1 wild type peptide are conserved, as is the critical arginine residue 253 (Fig. 1). Thus, all Alphavirus Nsp1 proteins most probably have a similar monotopic membrane binding mechanism. Several other proteins seem to associate with membranes monotopically. One of the best characterized among these is prostaglandin H synthase, which binds to the endoplasmic reticulum membrane via four short amphipathic helices with hydrophobic surfaces interacting with the outer leaflet of the lipid bilayer (35,36). CTP:phosphocholine cytidyltransferase (EC2.7.7.15) associates with membranes via an amphipathic ␣-helical peptide (23,37,38). The binding of human coagulation factors V and VIII to phosphatidylserine-rich platelet membranes is mediated by a C2 domain that is homologous in both proteins. The crystal structures of the C2 domains of factors V and VIII determined recently revealed finger-like loop structures that enable hydrophobic interactions within the outer leaflet of the lipid bilayer as well as polar interactions with the phosphatidylserine head groups (39,40).
The importance of tryptophan 259 for the membrane associ-FIG. 8. Solution structure of wild type peptide. A, energetically most preferred conformation from an ensemble of 15 out of 20 calculated structures. B, surface presentation of the wild type peptide in the amino acid residues are color-coded according to their hydrophobicity; blue represents hydrophobic residues and red represents hydrophilic residues. C, conformational ensemble of 15 out of 20 calculated structures. The backbone root mean square deviation for the structure ensemble is 0.62 Å and the overall heavy atom root mean square deviation is 1.19 Å. 304 NOE distances and 17 angle restraints were used in the calculation.
ation of Nsp1 was confirmed by the finding that its replacement by alanine diminished the ability of the peptide to retard lateral diffusion of PyrPS and abolished binding of intact Nsp1 to membranes (12). Variant peptide W259A also failed to compete with binding of wild type Nsp1 protein to liposomes (Fig. 2D). It is becoming increasingly apparent that tryptophan plays an important role in membrane association of many cellular proteins, such as phospholipases (41)(42)(43)(44). Notably, replacement of valine by tryptophan (V3W) in human phospholipase PLA 2 resulted in 250-fold enhancement of the activity due to enhanced membrane association (41). Two of the amphipathic helices of prostaglandin H synthases have tryptophans in the hydrophobic surface (36). The hydrophobic loops of factors V and VIII described above also have tryptophans. Tryptophan appears to possess several characteristics (including aromaticity) that make its disposition at the membrane interphase particularly favorable (for a detailed discussion, see Ref. 45).
In conclusion, we have shown that the binding peptide of Nsp1 attains an ␣-helical conformation in the presence of anionic liposomes. Its structure in the presence of trifluoroethanol reveals an amphipathic ␣-helix with a protruding tryptophan residue and a hydrophobic surface, which interact with acyl chains in the liposomes. Tryptophan 259 is necessary for proper binding of the peptide to lipids. Residues Arg-253, Lys-254, and probably also Arg-257 are needed for the interaction with head groups of anionic phospholipids, indicating that lipid binding of the peptide and most probably also of the entire Nsp1 protein is a finely tuned process perhaps initiated by polar interactions that lead to conformational change, allowing hydrophobic interactions to take place (cf. Ref. 45). Our intention is to introduce the present mutations of the binding peptide to the infectious clone of SFV to analyze the localization of the replication complex, putative infectivity, and pathogenicity of the transcribed RNAs.
In virus-infected cells, Nsp1 protein is found attached to the cytoplasmic side of the plasma membrane (8, 10) containing high concentrations of phosphatidylserine (46). High concentrations of phosphatidylserine are also needed to restore the methyltransferase activity of detergent-inactivated isolated Nsp1 (12) and for the optimal binding of the synthetic peptide to phospholipid bilayers (this study). Thus, we propose that membrane binding, enzymatic activation and intracellular targeting of Nsp1, and consequently, the entire RNA polymerase complex is regulated by this monotopic binding peptide. This proposition is supported by the RNA synthesis of Sindbis virus, another Alphavirus, is severely inhibited in the Chinese hamster ovary cell mutants with a lowered PS content (47).