Membrane Binding and Rearrangement by Chikungunya Virus Capping Enzyme nsP1

Abstract

Alphavirus genome replication is carried out by the viral replication complex inside modified membrane structures called spherules. The viral nonstructural protein 1 (nsP1) is the only membrane-associated protein that anchors the replication complex to the cellular membranes. Although an internal amphipathic helix of nsP1 is critical for membrane association, the mechanism of nsP1 interaction with membranes and subsequent membrane reorganization is not well understood. We studied the membrane interaction of chikungunya virus (CHIKV) nsP1 and show that both the CHIKV nsP1 protein and the amphipathic peptide specifically bind to negatively charged phospholipid vesicles. Using cryo-electron microscopy, we further show that nsP1 forms a contiguous coat on lipid vesicles and induces structural reorganization, while the amphipathic peptide alone failed to deform the membrane bilayer. This suggests that although amphipathic helix of nsP1 is required for initial membrane binding, the remaining cytoplasmic domain of nsP1 is involved in the subsequent membrane reorganization.

Introduction

Replication of positive strand RNA viruses occurs inside membranous compartments in the cytoplasm of infected cells. Newly synthesized RNA and replication intermediates inside these membranous compartments avoid detection by the host innate immune system and are readily available for viral genome packaging (Harak and Lohmann, 2015; Paul and Bartenschlager, 2013). The viral replication machinery rearranges the cellular membranes into predominantly two types of membrane structures, the double-membranous vesicles seen in hepaciviruses, Nidovirales, and Picornaviridae, and the spherule-like invaginations seen in flaviviruses, Togaviridae, Bromoviridae and Nodaviridae (Belov and van Kuppeveld, 2012; den Boon et al., 2010; Harak and Lohmann, 2015; Miller and Krijnse-Locker, 2008; Salonen et al., 2005). Alphavirus, a member of Togaviridae, replicates in spherules formed on early endosomes, lysosomes, and on the cytoplasmic side of the plasma membrane (Kujala et al., 2001; Peränen and Kääriäinen, 1991; Peränen et al., 1995).

Alphaviruses are arthropod-borne infectious agents that cause viral disease in a wide range of vertebrates and invertebrates. Members of the alphavirus genus include chikungunya virus (CHIKV), the Venezuelan equine encephalitis virus (VEEV), Sindbis virus (SINV) and Semliki forest virus (SFV) among others (Gould et al., 2010; Kuhn, 2007; Schwartz and Albert, 2010). Alphavirus infections typically result in fever, rashes, arthralgia as in the case of CHIKV infection or neurological symptoms such as viral encephalitis as in the case of VEEV. The alphavirus genome is a single strand, positive sense RNA of ~11.8 Kb that contains two open reading frames (ORFs). The genome is protected on either end by the 5’ cap structure and the 3’ poly-A tail (Strauss and Strauss, 1986; Strauss and Strauss, 1994). Translation of the first ORF of alphavirus genome yields two polyproteins containing either three or four non-structural proteins (nsP), P123 or P1234, which are post-translationally processed in cis or trans by the viral protease nsP2 (de Groot et al., 1990; Griffin, 2007). Structural proteins are translated from the second ORF near the 3’ end of the genome. RNA synthesis in alphaviruses requires all four nsPs. NsP1 is a peripheral membrane protein that has methyltransferase (MTase) and guanylyltransferase (GTase) activities. NsP1 first methylates GTP (MTase function), and then transfers the methylated GTP to the 5’ end of the viral RNA to form the type 0 cap (GTase function) (Ahola and Kääriäinen, 1995; Ahola et al., 1997; Li et al., 2015; Tomar et al., 2011). NsP2 is a protease required for processing of the polyprotein P123 and P1234. In addition, nsP2 carries the nucleoside triphosphatase (NTPase) and RNA helicase activities required for capping and genome replication (Gomez de Cedrón et al., 1999; Rikkonen et al., 1994; ten Dam et al., 1999; Vasiljeva et al., 2000). NsP3 is a heavily phosphorylated protein whose function is not well understood (Li et al., 1990; Vihinen and Saarinen, 2000; Wang et al., 1994). NsP4 is the viral RNA-dependent RNA polymerase that catalyzes the synthesis of both plus- and minus-strand RNA (Rubach et al., 2009; Tomar et al., 2006). In infected cells, synthesis of plus- and minus-sense RNA is carried out by membrane-associated viral replication complexes containing all four nsPs (nsP1–4) as well as unknown host factors (Kääriäinen and Ahola, 2002; Kujala et al., 2001; Pietilä et al., 2017; Rupp et al., 2015). However, little is known about the membrane association of the replication complex and the formation of spherules inside infected cells.

Among the four viral non-structural proteins, nsP1 is the only membrane-associated protein, and hence is solely responsible for the localization and anchoring of the viral replication complex to the site of replication on the membranes (Ahola et al., 1999; Lampio et al., 2000; Peränen et al., 1995; Spuul et al., 2007). The membrane binding ability of nsP1 is critical for viral RNA replication since mutations destroying membrane interaction resulted in no detectable RNA replication in cells (Kallio et al., 2015). Although the methyltransferase and guanylyl transferase activities are retained in recombinantly expressed nsP1, membrane interaction and the RNA capping functions of nsP1 seem to be coupled in infected cells, indicating that the protein is adapted to function in a membranous environment (Ahola et al., 1999; Li et al., 2015; Tomar et al., 2011). Membrane binding of nsP1 was shown to be mediated by a putative amphipathic helix formed by amino acids 244–263 and the covalent palmitoylation of cysteine residues (⁴¹⁸CCC⁴²⁰) (Ahola et al., 2000; Laakkonen et al., 1996; Spuul et al., 2007). However, nonpalmitoylated mutants of nsP1 retained their membrane binding capacity (Ahola et al., 1999; Laakkonen et al., 1996) and did not disrupt viral replication (Ahola et al., 2000). On the other hand, mutations in the amphipathic helix such as R253E and W259A resulted in loss of membrane interaction in SFV nsP1, and in turn were found to be deleterious to viral replication in the cell (Spuul et al., 2007). In vitro lipid interaction studies with the synthetic membrane-binding peptide from SFV nsP1 showed that the peptide was unstructured in solution, but had an increase in α-helical content in the presence of anionic phospholipids such as POPS (1-palmitoyl-2-oleoyl-sn-glycero-3-phospho-L-serine), suggesting interactions between the peptide and the membrane bilayer of the liposomes (Lampio et al., 2000). However, whether nsP1 or the amphipathic peptide alone can induce rearrangement of the constituent membrane phospholipids is not known. We studied membrane association property of CHIKV nsP1 and its membrane-binding amphipathic peptide. Recombinant CHIKV nsP1 exists as oligomers and monomers, and both forms are well folded in the absence of membrane. Using circular dichroism (CD) and membrane flotation assays, we show that both full-length CHIKV nsP1 and the individual peptide have an affinity for negatively charged phospholipids. This interaction between the peptide and phospholipids results in a change in the secondary structure of the peptide. Furthermore, using cryo-electron microscopy (cryo-EM), we show that CHIKV nsP1 protein assembles on the outer leaflet of liposomes and lipid nanotubes. Longer incubation with nsP1 resulted in deformation of liposomes into elongated tubes, suggesting that nsP1 alone can rearrange membranes.

Materials and Methods

Peptide synthesis

The amphipathic helix peptide from CHIKV nsP1, ²⁴⁴GSTLYPESRKLLKSWHLPSV²⁶³ and its mutants Pro249Thr (P249T), Trp258Ala (W258A), Arg252Glu (R252E) and the triple mutant Arg252Glu-Lys253Glu-Lys256Glu (RKK_E) were custom synthesized at ThermoFisher Scientific peptide synthesis facility. The peptides were obtained as trifluoroacetic acid (TFA) salts that were greater than 95% pure.

Expression and purification of Chikungunya virus nsP1

The codon-optimized full-length recombinant gene coding for CHIKV nsP1 (La Reunion strain, Genbank: FR717337.1, aa 1–535) was obtained from Dr. Brian Geiss at Colorado State University. The gene was subsequently cloned into pET28b vector that carries the T7 polymerase promoter and kanamycin resistance. Recombinant CHIKV nsP1 protein was expressed with an N-terminal 6xHis tag in E.coli Rosetta cells (Novagen) grown in Terrific Broth supplemented with 30 μg/mL of kanamycin and 37 μg/ml of chloramphenicol. Cells were grown at 37 °C to an O.D.₆₀₀ of 0.8, and protein expression was induced by the addition of 0.5 mM IPTG with growth continued overnight at 18 °C. For protein purification, the cell pellet from a 2 L culture was resuspended in 30 mL of lysis buffer (25 mM Tris HCl, pH 8.0, 200 mM NaCl, 5 mM β-mercaptoethanol and cOmplete™, EDTA-free protease inhibitor cocktail (Roche)) and lysed by sonication. Protein in the soluble fraction of the lysate was loaded onto TALON™ (Clontech) metal-affinity chromatography resin pre-equilibrated in lysis buffer (without the protease inhibitor cocktail). Bound CHIKV nsP1 was eluted using a gradient of 5–150 mM imidazole in elution buffer (25 mM Tris HCl, pH 7.0, 200 mM NaCl, and 5 mM β-mercaptoethanol). Fractions containing nsP1 were pooled and concentrated to a final volume of 1 mL using Amicon® Ultra centrifugal filters (Millipore). The sample was then loaded onto a HiLoad 16/60 Superdex 200 (prep-grade) size-exclusion column (GE Healthcare) pre-equilibrated with buffer containing 20 mM Tris–HCl, pH 7.5, 200 mM NaCl and 1 mM TCEP (tris-2-carboxyethylphosphine) in order to separate nsP1 from its degradation products and contaminants. CHIKV nsP1 eluted from Superdex 200 in two separate elution peaks. The first elution peak corresponds to the void volume of the column (> 600 kDa), and the second, later elution peak corresponds to the molecular size of nsP1 monomer (60 kDa) based on the elution profile of protein standards (Bio-Rad).

Generation and purification of GFP-fusion constructs:

A plasmid containing GFPuv, the “cycle 3” variant of green fluorescent protein from Aequorea victoria (jellyfish) described by Crameri et al. (1996), was obtained from Dr. Jose Barral at UTMB (Crameri et al., 1996; Fukuda et al., 2000). GFPuv sequence was subcloned into pET28b vector to generate an N-terminal 6xHis tagged construct of GFPuv alone. To generate GFP-fused CHIKV amphipathic peptide constructs, Minigene™ plasmid containing two tandem copies of the amphipathic peptide sequence (WT2x) of CHKV nsP1 separated by the linker Pro-Gly-Ala, ²⁴⁴GSTLYPESRKLLKSWHLPSV²⁶³-PGA-²⁴⁴GSTLYPESRKLLKSWHLPSV²⁶³, was synthesized from IDT. The amphipathic peptide sequence, either a single copy or two tandem copies along with the short linker separating the two copies, was PCR amplified from the Minigene™ plasmid and subcloned into pET28b between the NdeI and BamH1 restriction sites. GFPuv sequence was then subcloned onto the C-terminus of the amphipathic peptide sequences between BamH1 and XhoI restriction sites of pET28b vector to generate the final N-terminal 6xHis tagged amphipathic peptide-GFP fusion constructs, WT1x-GFP (MGSS-6xHis-SSG-LVPRGSH-²⁴⁴GSTLYPESRKLLKSWHLPSV^263-EF-GFPuv) and WT2x-GFP (MGSS-6xHis-SSG-LVPRGSH-²⁴⁴GSTLYPESRKLLKSWHLPSV²⁶³-PGA-²⁴⁴GSTLYPESRKLLKSWHLPSV²⁶³-EF -GFPuv). Trp258 was mutated to Ala in both copies of the amphipathic helix in WT2x-GFP using site-directed mutagenesis to generate W258A_2x-GFP mutant. 6xHis-tagged GFP and GFP fusion constructs were purified similarly as the CHKV nsP1 protein described above.

Preparation of liposomes and lipid nanotubes

The phospholipids for liposome generation, 1-palmitoyl-2-oleoyl-sn-glycero-3-phospho-L-serine (PS), 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (PC), 1-palmitoyl-2-oleoyl-snglycero-3-phosphoethanolamine (PE), 1-palmitoyl-2-oleoyl-sn-glycero-3-phospho-(1′-racglycerol) (PG), and Sphingomyelin (SM) were purchased as chloroform solutions from Avanti Polar Lipids Inc. Cholesterol powder was purchased from Sigma and dissolved in chloroform. The phospholipids for lipid nanotubes, C24:1 β-D-galactosyl ceramide (GC) and 1,2-dioleoyl-snglycero-3-phospho-L-serine (DOPS) were purchased from Avanti Polar Lipids Inc.

To prepare liposomes for circular dichroism (CD) and sucrose gradient flotation experiments, phospholipid mixtures were constituted in chloroform, following which chloroform was evaporated under nitrogen gas. The mixtures were left under vacuum for one hour to completely remove all traces of chloroform. The dried phospholipids were hydrated either in water or Tris buffer (pH 7.0) to a final lipid concentration of 2.5 mM. The mixtures were then sonicated until the lipid suspension clarified. The liposome solutions were stored at 4 °C and used the following day for CD and flotation experiments. For cryo-EM analysis, giant unilamellar vesicles (GUVs) were generated using a modified electroformation method (Dimitrov and Angelova, 1987; Dimitrov and Angelova, 1988). Briefly, POPC and POPS lipids were mixed in a glass tube at a 1:1 ratio and applied on the conductive side of indium tin oxide (ITO) coated glass plates at 1mM lipid concentration. Chloroform was evaporated and the plates were placed in vacuum for half hour. GUV swelling step was carried out by a function generator applied in 20 mM Tris, 100 mM NaCl buffer and 300 mM sucrose for 2h at 10Hz and 54 °C. The resultant GUVs were stored at 4 °C and used the following day for cryo-EM analysis. Lipid nanotubes used in cryo-EM analysis were prepared by mixing GC and DOPS in 1:4 (w/w) ratio in chloroform as previously described (Stoilova-McPhie et al., 2014). Chloroform was evaporated under nitrogen and the dried lipid mixture was rehydrated in 25 mM Tris, pH 7.0 to a final concentration of 1 mg/ml and sonicated briefly. The nanotubes were stored at 4 °C and used the next day.

Sucrose gradient flotation

Sucrose gradient flotation experiments were adapted for CHIKV nsP1 from those described for SFV nsP1 (Ahola et al., 1999). Sucrose solutions of 67% (w/v), 50% (w/v) and 10% (w/v) were prepared in 25 mM Tris, pH 7.5 or water. Purified nsP1, GFP, WT1x-GFP, WT2x-GFP, and W258A_2x-GFP proteins (~240 μM each) were separately mixed with 10-fold molar excess of liposomes and added to 67% (w/v) sucrose to constitute 500 μl of 60% sucrose sample. The sucrose solutions were layered in 5ml Ultra-Clear™ Beckman-Coulter ultracentrifuge tubes to form discontinuous gradient with the highest concentration of sucrose at the bottom. The layers from the bottom were as follows: 67% sucrose (500 μl), 60% sucrose and protein-liposome mixture (500 μl), 50% sucrose (3 ml) and finally the top layer of 10% sucrose (1 ml). The gradient was centrifuged at 150,000g at 4 °C for 4 hours in a Beckman Coulter Optima™ Max-XP ultracentrifuge fitted with MLS-50 swinging-bucket rotor. Following centrifugation, samples for SDS-PAGE analysis were collected by serially pipetting 500 μl from top to bottom of the resulting gradient. The SDS-PAGE gel was stained with Coomassie blue, and relative amount of protein in each of the flotation fractions was measured using the 1D gel electrophoresis density analysis software GelAnalyzer 19.1 (http://www.gelanalyzer.com). The raw volume of each of the protein bands was calculated and expressed as a percentage of the total volume of the protein bands in all lanes.

Circular dichroism (CD) spectroscopy

Far-UV CD spectroscopy data for nsP1 peptides and protein were collected with a Jasco J-815 spectropolarimeter. The CD spectra for synthetic peptides (WT, P249T, W258A, R252E, and RKK_E) were collected in either water or liposome solution of final concentration of 1.25 mM. To measure CD spectra with liposomes in high salt condition, the liposomes were constituted in Tris buffer, pH 7.0, containing 200 mM NaCl. The spectra for the purified nsP1 monomer and oligomer were collected in 25 mM Tris, pH 7.5 and 200 mM NaCl in the presence and absence of liposomes at 20 °C. CD spectra were collected in a 1 mm quartz cuvette at 20 °C from 195 nm to 260 nm using a data integration time of 8 seconds and were averaged over 2 measurements. Data were converted from millidegrees to mean residue ellipticity (MRE) using the formula:

$$[\theta ]=(\theta \times {10}^6)/(c\times l\times n)$$

where [θ] is the mean residue ellipticity, θ is the CD signal in millidegrees, c is the protein concentration in μM, l is the path length in mm, and n is the number of amino acid residues. The CD data analysis program BeStSel was used to analyze the data and estimate the percentage of secondary structure in each sample. BeStSel fits the experimental CD curve with a linear combination of fixed basis components to get the proportion of the eight structural elements (Micsonai et al., 2015).

Cryo-electron microscopy

CHIKV nsP1 monomer (10 mg/ml) was mixed with equal volume of GUVs (1 mM total lipid concentration) or lipid nanotubes (1 mg/ml final lipid concentration) and incubated between 0 and 15 min. Three microliters of prepared lipid-protein suspension were pipetted onto a Quantifoil grid (Electron Microscopy Sciences) and flash frozen in liquid ethane. The vitrified samples were imaged at 40,000X using a JEM-2200FS electron microscope (JEOL), and images recorded on a DE-20 camera (Direct Electron) in movie mode. Individual movie frames were aligned to correct image movement due to stage drift or electrical charging of the sample. GUVs and lipid nanotubes were also incubated with either GFP or GFP-fusion constructs of nsP1 amphipathic peptide (WT2x-GFP and W258A_2x-GFP) between 0 and 15 min and imaged as described above.

Results

CHIKV nsP1 membrane-binding peptides undergo a structural change from random coil to α-helix upon interaction with phosphatidylserine

A key mediator of the membrane interaction of alphavirus nsP1 was shown to be an internal amphipathic α-helix (Spuul et al., 2007). In CHIKV nsP1, this region corresponds to the amino acids ²⁴⁴GSTLYPESRKLLKSWHLPSV²⁶³. Fig 1A shows a sequence alignment of the putative membrane-binding peptide from CHIKV nsP1 with those of other alphaviral nsP1 proteins. The SFV nsP1 peptide corresponding to this region acquired an α-helical conformation in the presence of phospholipid vesicles, and this membrane-induced conformational change was proposed to promote binding and anchoring of nsP1 to cellular phospholipid membranes (Lampio et al., 2000). The peptide sequence of CHIKV nsP1 differs from SFV nsP1 by the presence of a proline residue at the amino acid position 249 (Fig 1A–B). This proline conceivably disrupts the secondary structure of the predicted α-helix, and thus it is not clear whether CHIKV nsP1 peptide readily forms an α-helix like SFV nsP1. To test if the CHIKV nsP1 membrane-binding peptide forms an α-helix upon interaction with phospholipid membranes, we synthesized the membrane-binding peptides of CHIKV nsP1, wild-type (WT) and two mutants, Pro249Thr (P249T) and Trp258Ala (W258A). The substitution of Pro249 to Thr will create a peptide identical to the SFV nsP1 sequence. Trp258 to Ala substitution was introduced to test if Trp258 is critical for membrane interaction. Trp258 is conserved in all nsP1 sequences (Fig 1A) and the Trp to Ala mutation in SFV nsP1 was shown to significantly disrupt the membrane interaction of nsP1 and its methyltransferase activity (Ahola et al., 1999).

Figure 1. Phosphatidylserine induces formation of α-helix in the membrane-binding peptide of CHIKV nsP1.

(A) Sequence alignment of the membrane-binding peptides in alphavirus nsP1. Peptides belong to chikungunya virus (CHIKV, FR717337.1), Semliki Forest virus (SFV, NC_003215.1), Venezuelan Equine Encephalitis virus (VEEV, KR260736.1), Sindbis virus (SINV, NC_001547.1), and O’nyong-nyong virus (ONNV, NC_001512.1). Pro249, Arg252, Lys253 and Lys256, and Trp258 that are mutated in the CHIKV nsP1 peptides are shown in bold and underlined, of which Arg252 and Trp258 are conserved in all alphaviral nsp1 sequences. (B) Helical wheel representation of the CHIKV nsP1 membrane binding peptide. Amino acids from Ser245 to Trp258 were used to create the helical wheel based on the NMR structure of SFV nsP1 peptide (PDB code 1FW5) (Lampio et al., 2000). Amino acids are color coded according to the side chain characteristics; yellow, nonpolar; red, acidic; purple, polar-uncharged; blue, basic. The dotted line separates the hydrophobic cluster from the hydrophilic side of the α-helix, highlighting the amphipathic nature of the α-helix. Pro249 that disrupts the α-helix in CHIKV nsP1 is shown in red. (C) CD spectra of CHIKV nsP1 membrane-binding peptides in the presence of liposomes. CD spectra were collected for the WT (left), P249T (center) and W258A (right) peptides in water, PC-only, and PC:PS liposomes. All three peptides acquired an α-helical secondary structure in PC:PS liposomes. (D) CD spectra of CHIKV nsP1 membrane binding peptides in liposomes containing 200 mM NaCl. Addition of salt to the PC:PS liposomes deterred the formation of α-helix.

The CHIKV peptide conformation in the presence and absence of liposomes was analyzed using CD spectroscopy. Far UV spectra for all three peptides (WT, P249T, and W258A) measured in water had absorption minima at 200 nm, indicative of random coil structure (Fig 1C). In the presence of phosphatidylcholine (PC)-only liposomes, the peptides retained the random coil conformation indicating that the neutral phospholipid PC had no effect on the secondary structure of the amphipathic peptides. However, in the presence of liposomes containing a 1:1 mixture of PC and the negatively charged phosphatidylserine (PS) the absorption minima shifted dramatically to 208 and 222 nm, characteristic of α-helices for all three peptides (Fig 1C). Secondary structure estimation of the CD data by BeStSel (Micsonai et al., 2015) indicated that the P249T peptide, which is identical to the SFV nsP1 peptide, has an α-helical content of ~19 %, while CHIKV WT nsP1 peptide has ~8%. Thus, the presence of Pro249 in CHIKV nsP1 decreases the α-helical content compared to SFV nsP1. Unexpectedly, the W258A peptide also bound PC:PS liposomes and folded into an α-helix with significant helical content of ~13 %, slightly higher than that of the WT peptide (Fig 1B). Thus, Trp258 is not required for membrane interaction in the context of the peptide. The amphipathic peptides of CHIKV nsP1 likely bind the negatively charged PS through electrostatic interactions. We thus tested if high salt concentration would interfere with peptide binding with PC:PS membranes. We measured secondary structure content of CHIKV nsP1 peptides in the PC:PS (1:1) liposomes containing 200 mM NaCl. As expected, the α-helix formation of CHIKV nsP1 peptides in the PC:PS liposomes was abolished (Fig 1D), confirming an electrostatic interaction between the peptides and liposomes.

Negatively charged phospholipids induce α-helix formation in CHIKV nsP1 membrane-binding peptides

We next tested if lipids other than PS could induce the formation of α-helix in CHIKV nsP1 peptides, WT and W258A. First, the secondary structure of the peptides was measured in PC liposomes containing 40% (w/v) of either zwitterionic phosphatidylethanolamine (PE) or the negatively charged phosphatidylglycerol (PG) (Fig 2A). PC liposomes containing 40% (w/v) PS were also included as a control. For both WT and W258A peptides, PE did not have any effect on the peptide conformation. In contrast, in the presence of negatively charged PG, the absorption minima of both peptides shifted to those of an α-helix, suggesting an interaction between the peptides and the liposome membranes. Thus, the membrane-binding peptides of CHIKV nsP1 preferentially interact with negatively charged phospholipids such as PS and PG.

Figure 2. Membrane lipids influence the secondary structure of the CHIKV nsP1 peptides differently.

(A) CD spectra of CHIKV nsP1 membrane binding peptides WT and W258A collected in water (dashed grey), PC-only (solid grey), PC:PS (red), PC:PE (purple) or PC:PG (green). Anionic phospholipids PS and PG rendered the peptides α-helical, while the neutral phospholipid PE had no effect on the secondary structure of the peptides. (B) Effect of sphingomyelin (SM) on the secondary structure of CHIKV nsP1 WT and W258A peptides. CD spectra were collected in PC:PS liposomes containing 10%, 20%, and 30% (w/w) sphingomyelin. Increasing concentration of sphingomyelin seemed to increase disorder in the peptides without altering α-helical content. (C) Effect of cholesterol (CL) on the secondary structure of CHIKV nsP1 WT and W258A peptides. CD spectra were collected in PC:PS liposomes supplemented with 10%, 20%, and 30% (w/w) cholesterol. Cholesterol did not have any effect on the secondary structure of either peptide.

We next tested the interaction of the nsP1 peptides with liposomes containing sphingomyelin or cholesterol, which alter membrane curvature and fluidity. Sphingomyelin, a sphingosine based phospholipid tends to disrupt the lipid bilayer by rigidifying membranes and forms domains within the bilayer (Goñi and Alonso, 2006). Increasing amount of sphingomyelin (0–30%) in the PC:PS liposomes resulted in an increased absorbance minimum in the range between 200 and 210 nm, while the absorbance at 220 nm remained unperturbed. This indicates that the peptides were unfolding in the presence of sphingomyelin without changing the α-helical content (Fig 2B). On the other hand, cholesterol tends to increase membrane fluidity and induces a negative curvature in the lipid monolayer (Wang et al., 2007). Increasing amounts of cholesterol (0–30%) in the PC:PS liposomes did not alter the α-helix content in the nsP1 peptides. Thus, cholesterol did not affect the membrane association of CHIKV nsP1 peptides (Fig 2C).

Positively charged residues in the nsP1 membrane-binding peptide are required for electrostatic interactions with negatively charged phospholipids

It was previously shown that the helical content of the SFV nsP1 membrane-binding peptide increases in proportion to the amount of negatively charged PS in the membrane (Lampio et al., 2000). We also tested if the α-helical content of CHIKV nsP1 membrane-binding peptide increases with the relative amount of PS in the liposomes. With increasing proportion of PS from 10 to 50% (w/v), the measured ellipticity at 220 nm of the WT peptide, indicative of an α-helix, gradually increased (Fig 3). Increase of PS content from 40 to 50% did not increase the α-helical content any further. Based on the highly electrostatic nature of the interactions between the CHIKV nsP1 membrane-binding peptide and the negatively charged PS or PG, we hypothesized that mutation of positively charged amino acids in the amphipathic peptide would disrupt membrane binding and subsequent helix formation. In order to test this, we synthesized two additional mutants, a conserved Arg252 to Glu substitution (R252E) and a triple mutant Arg252, Lys253, and Lys256 to Glu (RKK_E), which eliminated all positively charged amino acids in the amphipathic peptide (Fig 1A). We compared the helix formation of the three mutant peptides (W258A, R252E and RKK_E) in PC:PS liposomes with increasing proportion of PS from 10% to 50% (w/v). W258A and R252E behaved similarly to the WT peptide with an increase in α-helical content proportional to an increase in PS concentration (Fig 3). The triple mutant RKK_E, however, did not acquire an α-helical conformation even in 50% PS-containing liposomes (Fig 3). The inability to form α-helix of the RKK_E peptide is not due to its amino acid substitutions, because RKK_E peptide was able to form an α-helix in trifluoroethanol (TFE), similar to other peptides (Sonnichsen et al., 1992) (Fig 3). Thus, the positively charged residues (R252, K253, and K256) in the membrane-binding CHIKV nsP1 peptide are necessary for interaction with negatively charged lipids in the membrane.

Figure 3. CHIKV nsP1 peptide binds membranes via electrostatic interactions between phosphatidylserine and basic amino acids.

CD spectra for WT, W258A, R252E and RKK_E peptides were measured in PC:PS liposomes containing 0 – 50 % PS (w/v). CD spectra of the peptides were also collected in 40% TFE, an α-helix inducer, as a positive control for helix formation (magenta). The α-helical content of peptides as indicated by the ellipticity at 220 nm increased proportionally with the fraction of negatively charged PS up to 40 % (w/v). The RKK_E mutant did not acquire α-helix conformation even in 50 % PS (w/v), indicating that the positively charged amino acids are required to interact with negatively charged PS membrane.

CHIKV nsP1 is well folded without lipids and exists as a monomer and oligomer in solution

We next tested membrane association of recombinant full-length CHIKV nsP1. Although nsP1 is shown to associate with membranes in infected cells, recombinant CHIKV nsP1 is mostly soluble, and could be purified from the soluble fraction of the E.coli cell lysate (Bullard-Feibelman et al., 2016). Size exclusion chromatography of CHIKV nsP1 showed two separate elution peaks. The first elution peak corresponds to the void volume of the column suggesting the presence of higher order oligomeric species of nsP1. The second, later elution peak corresponds to the molecular size of nsP1 (60 kDa), suggesting the presence of monomeric nsP1 in solution (Fig 4A). The oligomerization is not due to interaction with nucleic acids, because the ratio of UV absorbance at 260 nm and 280 nm was 0.66 and 0.7 for nsP1 oligomer and monomer, respectively. To determine the size of the oligomer, the first peak was analyzed by sedimentation velocity ultracentrifugation. The data show that the protein does not exist as a discrete oligomer but forms large soluble aggregates (not shown). To determine if the monomer and oligomer of nsP1 are well folded in the absence of lipids, CD spectra were collected for both peaks (Fig 4B). There was no difference in the CD spectra of the nsP1 monomers and oligomers. BeStSel program estimated that nsP1 contains 46% α-helix, 24% β-sheet and 24% disordered regions. JPred, the protein secondary structure prediction program predicted CHIKV nsP1 to be a mixed α/β protein with 34% α-helix and 22% β-sheet content (Drozdetskiy et al., 2015). Thus, the protein secondary structure content estimated by CD and predicted by bioinformatics tools are consistent. Since the membrane-binding amphipathic peptide folds into α-helix in the presence of PS lipid (Fig 1C), we tested if the addition of lipids can also increase the α-helical content of CHIKV nsP1. The CD spectra of CHIKV nsP1 in the presence of PC-only or PC:PS liposomes are not significantly different from the spectrum in the absence of liposomes (Fig 4C). This suggests that most of nsP1 protein is well folded even in the absence of lipids.

Figure 4. Recombinant CHIKV nsp1 is a soluble and predominantly α-helical protein.

(A) Size exclusion chromatography of CHIKV nsP1. The protein elutes in two peaks, the early peak corresponds to large oligomeric aggregates of nsP1, and the second peak corresponds to the ~60 kDa monomer of nsP1. Protein molecular weight standards are shown on top of the elution profile of CHIKV nsP1. (B) CD spectra of CHKV nsP1 monomer and oligomer. Both the monomer and oligomer forms of CHIKV nsP1 are well folded in solution with a characteristic, predominantly α-helical protein spectrum. (C) CD spectra of nsP1 monomer in the absence and presence of liposomes. CD spectra were collected in Tris buffer, pH 7.0, with 200 mM NaCl (dashed grey), PC-only liposomes (solid grey) and PC:PS liposomes (red). No significant change in the overall α-helical content of the protein was seen upon addition of liposomes to the protein solution.

CHIKV nsP1 binds to phospholipid vesicles via the membrane-binding peptide

We next studied the propensity for membrane interaction of recombinant CHIKV nsP1 (both monomer and oligomer) by testing its ability to bind to phospholipid vesicles. Small unilamellar vesicles, either PC-only or PC:PS liposomes, were incubated with purified nsP1 and subjected to ultracentrifugation in a discontinuous sucrose gradient. The protein bound to the phospholipid vesicles would float to the top, at the interface of 10% (w/v) and 50% (w/v) sucrose, whereas the protein not interacting with the phospholipid membrane would sediment near the bottom of the sucrose gradient (60% (w/v) sucrose) (Ahola et al., 1999). Fractions were taken from the sucrose gradient following ultracentrifugation of the protein-liposome mixtures and analyzed using SDS-PAGE. In the absence of liposomes, both monomers and oligomers of CHIKV nsP1 migrated to the bottom of the sucrose gradient (Fig 5A). In the presence of liposomes, both forms of full-length nsP1 protein floated to the top of the gradient in varying amounts. Both oligomer and monomer forms showed preference for PC:PS liposomes over PC-only liposomes (Fig 5A). However, oligomeric CHIKV nsP1 floated completely to the top of the gradient, whereas monomeric CHIKV nsp1 partially floated in the presence of PC:PS liposomes (Fig 5A). Hence, oligomeric CHIKV nsp1 shows preferable binding to phospholipid vesicles over the monomeric form. Bovine serum albumin (a negative control) sedimented at the bottom of the gradient even in the presence of liposomes, indicating that interaction between CHIKV nsP1 and liposomes was specific.

Figure 5. CHIKV nsP1 binds negatively charged membranes via the amphipathic α-helix.

(A) Membrane binding capacity of CHIKV nsP1 protein. Flotation of CHIKV nsP1 monomer and oligomer on a discontinuous sucrose gradient was measured in the presence of PC-only or PC:PS liposomes. Following ultracentrifugation, fractions were collected from top to bottom and analyzed by SDS-PAGE. The protein amount in each lane was measured by 1D gel density analysis and listed as percentage of the total protein measured from all fractions. Both monomer and oligomer show preferential binding to PC:PS liposomes, although the oligomer binds liposomes with greater affinity than the monomer. BSA, the negative control, sedimented at the bottom of the sucrose gradient in the presence of PC:PS liposomes. (B) Membrane binding capacity of CHIKV nsP1 amphipathic peptides. Flotation of GFP alone and GFP tagged with CHIKV nsP1 amphipathic peptides (WT1x, WT2x, and W258A_2x) was measured with PC-only and PC:PS liposomes. GFP alone or WT1x-GFP does not interact with either PC or PC:PS liposomes and sinks to the bottom of the sucrose gradient. WT2x- and W258A_2x-GFP interact with PC:PS liposomes (but not PC-only liposomes) and migrate partially to the top of the sucrose gradient.

Since the amphipathic membrane-binding peptide of CHIKV nsP1 was shown to interact with liposomes, we tested if the peptide alone was sufficient to mediate membrane interaction of nsP1. Since the small size (20 aa) of the peptide could not be well resolved in an SDS-PAGE, we attached the green fluorescent protein (GFP) to the membrane-binding peptide. Previously, GFP tagged with two copies of the membrane binding peptide from SFV nsP1 in tandem was shown to successfully localize to the plasma membrane inside cells (Spuul et al., 2007). We thus designed GFP fusion constructs with a single copy (WT1x) or two copies (WT2x) of the CHKV nsP1 membrane-binding peptide. Both WT1x-GFP and WT2x-GFP fusion proteins eluted as a monomer in size exclusion chromatography, similar to GFP alone, suggesting that the membrane-binding peptide does not cause oligomerization of full-length nsP1. In flotation assays, GFP alone or WT1x-GFP did not bind either PC or PC:PS liposomes (Fig 5B). In contrast, a measurable amount of WT2x-GFP floated to the top of the sucrose gradient along with PC:PS liposomes, but not in the presence of PC-only liposomes (Fig 5B). This indicates that the localization of WT2x-GFP to the liposomes was mediated by the specific interaction of the amphipathic peptide with the PS lipids as in the case of the full-length CHIKV nsP1. However, despite the presence of two copies of the amphipathic peptide, the amount of the WT2x-GFP bound to the liposomes was significantly lower than the full-length protein, suggesting that additional elements in the full-length CHIKV nsP1 are involved in membrane association. Next, we introduced the W258A mutation in both repeats of the amphipathic peptide in WT2x-GFP (W258A_2x-GFP) and tested if the mutation affected the co-migration of the fusion protein with liposomes. In the flotation assay, small amount of W258A_2x-GFP comigrated with PC:PS liposomes to the top of the sucrose gradient, but not with PC-only liposomes, similar to the wild type construct WT2x-GFP (Fig 5B). Thus, Trp258 is not critical for direct membrane interaction of the amphipathic peptide in CHIKV nsP1. This is consistent with the CD experiment showing that the W258A peptide was able to form an α-helix in the presence of PC:PS liposomes.

CHIKV nsP1 associates with and deforms the outer layer of the lipid bilayer

Since viral RNA replication occurs in specialized membrane compartments or spherules that require extensive rearrangement of the constituent membranes, we tested whether nsP1 could induce membrane rearrangement. We used cryo-EM to image liposomes and lipid nanotubes in the presence of recombinant CHIKV nsP1. To visualize liposomes by cryo-EM, we used giant unilamellar vesicles (GUVs) containing PC and PS (1:1). Lipid nanotubes were assembled by mixing a 1:4 (w/w) ratio of galactoceramide that tends to form long tubular bilayers and dioleoyl phosphoserine (DOPS) that carries the negatively charged phosphoserine head group. Cryo-EM images of liposomes and lipid nanotubes alone showed a well-formed lipid bilayer (Fig 6). In the presence of CHIKV nsP1, the outer layer of the lipid bilayer in both liposomes and lipid nanotubes was clearly decorated with protein. CHIKV nsP1 forms a ~40Å thick layer extending from the outer leaflet of the liposome membranes and did not seem to penetrate the inner lipid leaflet. The size of nsP1 density would be consistent with that of a 60 kDa globular protein. Thus, CHIKV nsP1 has a large cytoplasmic domain and a relatively small membrane association region. Upon longer incubation (> 10 min) of CHIKV nsP1 with liposomes, spherical membrane curvature was lost, and elongated tubular vesicles were formed (Fig 6A). Lipid nanotubes were also similarly distorted upon longer incubation and showed uneven bilayer surfaces (Fig 6B). Hence, CHIKV nsP1 induces structural changes in the phospholipid membrane.

Figure 6. CHIKV nsP1 binds to the outer leaflet of the lipid bilayer and deforms membranes.

Cryo-electron microscopy images of CHIKV nsP1 in liposomes (A) and lipid nanotubes (B). Liposomes/lipid nanotubes alone (top), immediately after mixing with CHIKV nsP1 monomer (middle), and following > 10 min incubation with CHIKV nsP1 (bottom) are shown. Right column shows zoomed-in views. CHIKV nsP1 forms a contiguous protein coat on the outer leaflet of the lipid bilayer. Longer incubation with CHIKV nsP1 results in the deformation of the liposomes and lipid nanotubes into featureless tubular structures (bottom).

We next tested if the membrane-binding peptide alone is sufficient to bind and distort the curvature of the lipid bilayer. We incubated GFP, WT2x-GFP, and W258A_2x-GFP with liposomes (GUVs) and lipid nanotubes (Fig 7). GFP alone did not bind liposomes or lipid nanotubes, as expected. When WT2x- or W258A_2x-GFP were incubated with liposomes and lipid nanotubes, clear decoration of the proteins on the outer leaflet of the lipid bilayer was visible (Fig 7). However, compared to the full-length nsP1, binding of WT2x-GFP and W258A_2x-GFP showed a discontinuous protein layer with some lipid vesicles and nanotubes either partially coated or not bound to protein at all. This is consistent with the flotation results since the GFP fusion constructs did not show a robust binding to liposomes. Further, longer incubation with WT2x-GFP or W258A_2x-GFP did not result in any membrane distortion (Fig 7). This suggests that the amphipathic helix by itself cannot alter membrane curvature and distort the lipid bilayer. Thus, additional factors present in the full-length nsP1 are likely involved in membrane rearrangement.

Figure 7. Membrane binding peptide of CHKV nsP1 alone cannot induce membrane rearrangement.

Cryo-electron microscopy images of liposomes (A) lipid nanotubes (B) mixed with GFP alone (top) or GFP tagged with two copies of CHIKV nsP1 amphipathic peptide, WT2x-GFP (middle), and W258A_2x-GFP (bottom). GFP alone does not bind liposomes and lipid nanotubes. WT2x-GFP and W258A_2x-GFP bind to the outer layer of the liposome and lipid nanotubes, although binding was not continuous and the proteins were not uniformly distributed on all the available lipid bilayers of liposomes and nanotubes. No membrane distortion was observed after longer incubation with the GFP fusion constructs.

Discussion

Alphaviruses form membrane invaginations called spherules in infected cells and assemble the viral replication complex inside these spherules. Although the mechanism of spherule formation is not well understood, it requires bending of the phospholipid bilayer resulting in localized induction of curvature in the membrane. The alphavirus nsP1 is the only viral nonstructural protein that associates with membranes to anchor the viral replication complex (Lampio et al., 2000; Peränen et al., 1995). Thus, we investigated membrane binding properties of CHIKV nsP1. In alphavirus nsP1, the amphipathic moiety responsible for membrane interaction was shown to be the peptide spanning residues ²⁴⁴GSTLYPESRKLLKSWHLPSV²⁶³ (CHIKV nsP1 sequence). The capacity for membrane binding by CHIKV nsP1 peptide would be dependent on local phospholipid composition and the intrinsic rigidity and curvature of the cellular membranes. Thus, we first tested affinity of the membrane-binding peptide of CHIKV nsP1 for various phospholipids (PC, PS, PG and PE) and lipids that influence membrane rigidity and curvature (sphingomyelin and cholesterol). Upon binding to liposomes containing negatively charged phospholipids PS and PG, CHIKV nsP1 peptide folds into an α-helix from a disordered structure in solution (Figs 1C and 2A). The neutral phospholipids PC and PE did not induce any α-helix formation in the peptide. The preferential interaction of CHIKV nsP1 peptides with negatively charged PS or PG lipids is consistent with previous results reported for SFV nsP1 peptides (Ahola et al., 1999). The effects of sphingomyelin or cholesterol were minimal in inducing a conformational change in the amphipathic peptides (Fig 2B–C).

Next, we investigated the sequence-specific effects of CHIKV nsP1 peptide on membrane association. The CHIKV nsP1 sequence differs from the well-studied SFV membrane-binding peptide in that Pro249 is present in the middle of the peptide, instead of Thr249 in the SFV peptide (Fig 1A). Since Pro is a helix breaker, we investigated whether the Pro to Thr substitution increases lipid association of the peptide. The P249T mutant peptide indeed showed higher α-helical content than the wild-type CHIKV nsP1 peptide (19% vs. 8%), suggesting that Pro moderately reduces α-helix formation of the CHIKV nsP1 peptide. The conserved Trp259 in the membrane-binding peptide was shown to be essential for membrane association in the context of the full-length SFV nsP1, and W259A mutation in SFV nsP1 disrupted membrane localization of nsP1 and was lethal for viral replication (Ahola et al., 1999; Spuul et al., 2007). We thus originally designed the corresponding CHIKV W258A peptide as a negative control for membrane binding. Surprisingly, the W258A peptide shows essentially identical membrane-binding preferences and a lipid-induced structural change to an α-helix as the wild-type peptide (Fig 3). Thus, Trp258 is not required for the amphipathic peptide to interact with phospholipids, although it may be essential for membrane association in the context of the full-length SFV nsP1 (Ahola et al., 1999). It has recently been shown that CHIKV containing the W258A substitution in nsP1 results in a temperature-sensitive phenotype. CHIKV nsP1 mutant replicates similarly to the wild-type virus at the permissive temperature (28 °C), but shows ~300-fold reduction in viral replication at 39 °C (Bartholomeeusen et al., 2018; Utt et al., 2019). The membrane-binding amphipathic peptides are proposed to fold into an α-helix in the interfacial region between the aqueous and hydrophobic membrane core by inserting a large hydrophobic residue into the membrane (Gimenez-Andres et al., 2018; Lampio et al., 2000). The CHIKV nsP1 peptide contains several hydrophobic residues (L247, Y248, L254, L255, L260) (Fig 1A), and thus other hydrophobic residues in the W258A peptide may compensate for initial interaction with membranes at the permissive temperature. At the higher nonpermissive temperature, however, W258A may destabilize hydrophobic interactions between the peptide and membrane, leading to decreased viral replication. We also tested if the positively charged residues in the amphipathic peptide are required for the liposome interaction and α-helix formation. Arg252 is the only basic amino acid conserved in all alphavirus nsP1 peptides (Fig 1A). However, a single mutation of the conserved Arg252 to Glu did not disrupt its membrane interaction and α-helix formation (Fig 3). In contrast, a triple mutant RKK_E, which abolished all basic amino acids, failed to acquire an α-helical fold (Fig 3), suggesting that the charge-charge interactions between the peptide and the negatively charged phospholipids are required for membrane binding. The R252E substitution in nsP1 was shown to severely reduce viral replication in CHIKV and SFV, and abolish membrane localization of nsP1 protein in SFV (Ahola et al., 1999; Spuul et al., 2007; Utt et al., 2019). The dramatic effect of the R252E substitution in the context of the protein, compared to the peptide result, may result from the packing of the amphipathic peptide within the protein structure. In our liposome-binding assays, the peptide is completely accessible to phospholipids in liposomes and hence other basic amino acids could compensate for the role of R252 in membrane binding. In the nsP1 protein, however, the orientation of the amphipathic peptide is likely fixed by steric hindrance and R252 may predominantly interact with membranes. This could lead to loss of membrane interaction and in turn viral replication in R252E mutant.

Membrane association of full-length CHIKV nsP1 protein was next examined using flotation assays. Recombinant CHIKV nsP1 protein exists as monomers and oligomers in solution, and both forms preferably bind PC:PS liposomes over PC-only liposomes, consistent with the peptide results observed in CD. CHIKV nsP1 oligomer showed greater interaction with liposomes than the monomer, suggesting that nsP1 may bind membranes as oligomers (Fig 5). To test if the amphipathic peptide of CHIKV nsP1 alone was sufficient for membrane interaction, we linked one or two copies of the membrane-binding nsP1 peptide to GFP and measured membrane association in flotation assays. The GFP construct with two copies of the amphipathic peptide, either wild-type or the W258A mutant (WT2x- and W258A_2x-GFP) comigrated with liposomes, but not GFP alone or GFP with a single peptide copy (WT1x-GFP). This result is similar to what is reported for SFV nsp1 peptide (Spuul et al., 2007). However, even with two copies of the CHIKV nsP1 peptide, the amount of WT2x-GFP that migrated to the top was much smaller than the full-length nsP1 protein (Fig 5B). Thus, full-length nsP1 seems to have additional regions interacting with membranes. It has been shown that CHIKV nsP1 is palmitoylated at Cys417, Cys418, and Cys419 and the palmitoylation-deficient nsP1 greatly reduces viral replication (Utt et al., 2019; Zhang et al., 2019). Although our CHIKV nsP1, recombinantly expressed in E. coli, lacks palmitoylation at the cysteine residues, protein regions containing the palmitoylation sites may stabilize nsP1 further in the membrane.

We next tested if CHIKV nsP1 alone is capable of inducing membrane deformation by cryo-EM. CHIKV nsP1 incubated with liposomes or lipid nanotubes associates with the outer leaflet of the lipid bilayer. Approximately 40 Å thick protein layer protrudes outward (to cytoplasmic side), suggesting that membrane-association region in nsP1 is relatively small. Interestingly, longer incubation of liposomes or nanotubes with CHIKV nsP1 showed deformed membrane structures, indicating that nsP1 alone is sufficient to induce structural changes in membranes. However, the GFP fused amphipathic peptides of CHIKV nsP1, WT2x-GFP and W258A_2x-GFP, did not result in the distortion of the lipid bilayer, although they assemble on the outer leaflet of the membrane (Fig 7). This suggests that the amphipathic peptide alone is not capable of membrane deformation. Membrane-binding proteins can deform a membrane bilayer in several ways. In addition to embedding an amphipathic moiety in the bilayer, known as the local spontaneous curvature mechanism, oligomerization of proteins could stabilize and propagate the induced membrane curvature by forming a scaffolding coat on the membrane (McMahon and Gallop, 2005; Zimmerberg and Kozlov, 2006). The flotation and cryo-EM analyses suggest that alphavirus nsP1 may use both the insertion of α-helix and oligomerization strategies to deform membranes. This would be similar to brome mosaic virus (BMV) replication protein 1a, the sole protein required to induce a membrane-invaginating spherule in BMV (den Boon et al., 2001; Restrepo-Hartwig and Ahlquist, 1999). Multimerization of BMV 1a is proposed to induce spherule formation by initiating ER membrane invagination away from the cytoplasm, and coating the interior surface of spherules (Diaz et al., 2015). Oligomerization of CHIKV nsP1 on liposomes suggests that nsP1 may similarly decorate the internal surface of spherules. It should be noted that although nsP1 protein alone is sufficient to distort membranes, formation of spherules in alphavirus may require additional viral and/or cellular factors to conform membranes into a spherical invagination. It has been recently shown that uncleaved P123 and cleaved nsP4 protein (viral polymerase) of SFV form spherules in the absence of RNA synthesis (Hellstrom et al., 2017). Thus, uncleaved form of nsP1 in the P123 may play an additional role in the spherule formation. Future studies on the structure and membrane interaction of alphavirus nsP1 should provide the mechanism of membrane remodeling and strategies to develop antiviral therapeutics to disrupt the assembly of the viral replication complex.