An Amphipathic Alpha-Helix Domain from Poliovirus 2C Protein Tubulate Lipid Vesicles

Positive-strand RNA viruses universally remodel host intracellular membranes to form membrane-bound viral replication complexes, where viral offspring RNAs are synthesized. In the majority of cases, viral replication proteins are targeted to and play critical roles in the modulation of the designated organelle membranes. Many viral replication proteins do not have transmembrane domains, but contain single or multiple amphipathic alpha-helices. It has been conventionally recognized that these helices serve as an anchor for viral replication protein to be associated with membranes. We report here that a peptide representing the amphipathic α-helix at the N-terminus of the poliovirus 2C protein not only binds to liposomes, but also remodels spherical liposomes into tubules. The membrane remodeling ability of this amphipathic alpha-helix is similar to that recognized in other amphipathic alpha-helices from cellular proteins involved in membrane remodeling, such as BAR domain proteins. Mutations affecting the hydrophobic face of the amphipathic alpha-helix severely compromised membrane remodeling of vesicles with physiologically relevant phospholipid composition. These mutations also affected the ability of poliovirus to form plaques indicative of reduced viral replication, further underscoring the importance of membrane remodeling by the amphipathic alpha-helix in possible relation to the formation of viral replication complexes.


Introduction
Positive-strand RNA viruses [(+)RNA viruses] are the largest among all viral classes infecting eukaryotes, and include numerous important pathogens infecting humans, animals, and plants. A hallmark of (+)RNA virus replication is its dependence on host intracellular membranes to assemble functional viral replication complexes (VRCs) or replication organelles [1,2]. Different viruses replicate in specific organelle membranes, such as the ER, mitochondrion, chloroplast, vacuole, or peroxisome,

Peptides
Lyophilized peptides were obtained from Peptide 2.0 at a purity greater than 95%. Protein concentration was estimated using the extinction coefficient of peptides at 280 nm based on the number of tryptophan molecules in the peptide. The peptides were dissolved in aqueous buffer.

Phospholipid Vesicle Clearance Assay
Peptides were tested for their ability to clear 1000 nm extruded phospholipid vesicles by measuring the change in light scattering as a function of time using a Jasco V-550 UV/Visible spectrophotometer. A wavelength of 500 nm with a slit width of 2 nm and medium response time was used to monitor the change in light scattering. Phospholipid vesicles were suspended in 20 mM of Hepes with a pH = 7.4 and 100 mM of NaCl at a final volume of 500 µL in a quartz cuvette. Large unilamellar vesicles (LUVs) composed of either 100% POPG, POPS/PI/POPC/POPE 15%/10%/50%/25% (termed as POPC/POPE 2:1 in the text and figures), or POPS/PI/POPC/POPE 15%/10%/37.5%/37.5% (termed as POPC/POPE 1:1 in the text and figures) were used.

Dye Leakage Assay
Leakage assay was modified from a previous method [25]. Large unilamellar vesicles (LUVs) composed of either 100% POPG, POPS/PI/POPC/POPE 15%/10%/50%/25% (termed as POPC/POPE 2:1 in the text and figures) or POPS/PI/POPC/POPE 15%/10%/37.5%/37.5% (termed as POPC/POPE 1:1 in the text and figures) were prepared by resuspending dried lipid in a mixture of 9 mM of ANTS (8-aminonaphthalene-1,3,6-trisulfonic acid, disodium salt) and 25 mM of DPX [p-xylene-bis (pyridinium bromide)] (Invitrogen, Thermo Fisher Scientific). This solution was then treated with 10 cycles of freeze/thaw. To prepare large 1000 nm diameter vesicles, the mixture was passed through a 1000 nm cutoff polycarbonate membrane (Avanti Polar Lipids Inc.) using a mini extruder. The unencapsulated dye was separated from vesicles by gel filtration using a PD-10 column (GE). One hundred percent leakage was attained using a final concentration of 0.1% Triton-X 100. All data were normalized to 100% leakage. Percent leakage was calculated using the formula y2 − y1/y3 − y1 × 100, where y1 is the fluorescence intensity of intact vesicles, y2 is the fluorescence intensity after the addition of peptide, and y3 is the fluorescence intensity after the addition of Triton-X. Fluorescence measurements were recorded using a JASCO fluorometer (FP-6500), setting excitation and emission at 380 nm and 520 nm with slits of 5 nm and 20 nm, respectively.

Circular Dichroism (CD)
CD spectra were collected using a Jasco J-810 spectropolarimeter with a 1 mm quartz cell at room temperature. The following parameters were used for measurement: a scan rate of 50 nm/minute, step resolution of 0.5 nm, 0.1 nm time response, and a bandwidth of 1 nm. Protein concentration was estimated using the extinction coefficient of peptides at 280 nm on the basis of tryptophan molecules in the peptides. The final spectra were obtained by measuring appropriate blanks under similar conditions and subtracting them from the sample spectra. A 10 mM sodium phosphate (pH = 7.4) buffer was used for all CD studies. Large unilamellar vesicles (LUVs) composed of either 100% POPG, POPS/PI/POPC/POPE 15%/10%/50%/25% (termed as POPC/POPE 2:1 in the text and figures), or POPS/PI/POPC/POPE 15%/10%/37.5%/37.5% (termed as POPC/POPE 1:1 in the text and figures) were used.

Plaque Assay for Testing RNA Infectivity of wt and L3A Mutant Polioviruses
To generate L3A mutant poliovirus (PV-L3A), the three mutations were induced into the pXpA-SH, a plasmid containing full-length cDNA of poliovirus (PV-wt), via mutagenesis. For the RNA infectivity assay, full-length polio RNA of PV-wt or PV-L3A was synthesized with Ambion T7 RNA MEGA Script transcription kit (Thermo Fisher Scientific) using corresponding plasmid as the template [29]. HeLa cells grown on six well plates were transfected with serial dilutions of the transfection mix prepared using transIT mRNA transfection reagent (Mirus) and polio RNA according to the manufacturer's protocol. Six hours post-transfection, the cells were overlaid with DMEM growth medium containing 2% FBS solidified with 1% agarose. To visualize plaques, the monolayer was stained with crystal violet at two days post RNA transfection.

A Helix Domain from the Poliovirus 2C Protein Displays Amphipathic Propensity
Previous studies have shown the requirement of the N-terminal region within the poliovirus 2C protein for its membrane association [17,18,21]. In this study, we evaluated the role of an 18 amino acid peptide (2C-wt) from the N-terminal region of the poliovirus 2C protein (amino acids [19][20][21][22][23][24][25][26][27][28][29][30][31][32][33][34][35][36] for its ability to interact with membranes ( Figure 1). The helical wheel projection showed that the peptide possessed the ability to form an amphipathic α-helix with a hydrophobicity (H) of 0.440 and a hydrophobic moment (µH) at 0.526. Hydrophobic moment is a measure of the amphipathicity perpendicular to the axis of any periodic peptide structure, like an α-helix or a β-sheet. It can be calculated for a peptide using the specific hydrophobicity values associated with each amino acid.
The hydrophobicity values are used to weight vectors for each amino acid residue as they are positioned around the helix. The hydrophobic moment is the summation of the vectors. A mutated peptide (2C-L3A) was designed to study the disruption of the hydrophobic face of the putative amphipathic peptide. Amino acids at positions 3 (V3), 7 (I7), and 14 (L14) were mutated to alanine, thereby affecting the hydrophobicity (H) of the peptide at 0.229 and a µH at 0.330.

The α-Helix Domain Transforms Large Vesicles into Small Structural Entities
The 2C protein has been reported to affect membrane architecture. To test whether the amphipathic α-helix is sufficient to cause membrane restructuring, we monitored the ability of both 2C-wt and 2C-L3A to transform large unilamellar vesicles using a vesicle clearance assay that has been previously employed in studies reporting membrane remodeling by amphipathic α-helices [25]. In this assay, a decrease in apparent absorbance was correlated with a decreased light scatter that was due to the remodeling of large vesicles into smaller structures, such as small vesicles or tubules. Both peptides transformed the large POPG vesicles into smaller structures, although 2C-wt displayed a higher clearance than 2C-L3A, as observed by a decrease in apparent absorbance due to scatter (Figure 2A). Since 100% POPG vesicles are highly negatively charged, we used vesicles that were more physiologically relevant. We used two lipid compositions that were representative of the endoplasmic reticulum membrane [30]: one with a 2:1 POPC/POPE molar ratio and another with a 1:1 POPC/POPE molar ratio, since viral replication is known to affect the lipid metabolism, resulting in an increased production of phosphatidylcholine lipids in certain cases [31]. The 2C-wt peptide still displayed a significant transformation of large lipid vesicles with both POPC/POPE 2:1 and POPC/POPE 1:1 lipid vesicles ( Figure 2B,C). A slightly higher vesicle clearance was observed with POPC/POPE 2:1 vesicles when compared to POPC/POPE 1:1 lipid vesicles ( Figure 2B,C). The vesicle clearing ability of 2C-L3A seemed to be highly compromised in the presence of both POPC/POPE 2:1 and POPC/POPE 1:1 lipid vesicles ( Figure 2B,C).

An Increase in Helical Conformation Is Associated with Membrane Binding and Remodeling
We used circular dichroism to observe any conformational changes associated with the peptide during vesicle clearance. Minima at 208 nm and 222 nm indicated the presence of α-helical conformation in a protein. Both peptides, 2C-wt and 2C-L3A, displayed spectra characteristic of αhelical structure in the presence of 100% negatively charged POPG vesicles ( Figure 3A). However, 2C-L3A had a highly reduced helicity when compared to the 2C-wt peptide. This could be either due to a reduced number of bound peptides to the lipid vesicles in the case of 2C-L3A, or peptides bound to the lipid vesicle with a partial helical structure. This difference is magnified in the presence of POPC/POPE 2:1 and POPC/POPE 1:1 vesicles ( Figure 3B); 2C-L3A shows a complete loss of conformational change with both POPC/POPE 2:1 and POPC/POPE 1:1 vesicles. The three mutations introduced to affect the hydrophobicity of the 2C helical domain seemed to significantly affect the membrane binding property of this domain. Similar to the vesicle clearance data, we observed a slightly higher helicity with POPC/POPE 2:1 vesicles compared to POPC/POPE 1:1 vesicles ( Figure  2B, compare orange and grey lines).

An Increase in Helical Conformation Is Associated with Membrane Binding and Remodeling
We used circular dichroism to observe any conformational changes associated with the peptide during vesicle clearance. Minima at 208 nm and 222 nm indicated the presence of α-helical conformation in a protein. Both peptides, 2C-wt and 2C-L3A, displayed spectra characteristic of α-helical structure in the presence of 100% negatively charged POPG vesicles ( Figure 3A). However, 2C-L3A had a highly reduced helicity when compared to the 2C-wt peptide. This could be either due to a reduced number of bound peptides to the lipid vesicles in the case of 2C-L3A, or peptides bound to the lipid vesicle with a partial helical structure. This difference is magnified in the presence of POPC/POPE 2:1 and POPC/POPE 1:1 vesicles ( Figure 3B); 2C-L3A shows a complete loss of conformational change with both POPC/POPE 2:1 and POPC/POPE 1:1 vesicles. The three mutations introduced to affect the hydrophobicity of the 2C helical domain seemed to significantly affect the membrane binding property of this domain. Similar to the vesicle clearance data, we observed a slightly higher helicity with POPC/POPE 2:1 vesicles compared to POPC/POPE 1:1 vesicles ( Figure 2B, compare orange and grey lines).

Large Unilamellar Vesicles Are Remodeled into Tubular Structures by the Helix Domain from 2C Protein
In order to examine the morphological changes associated with lipid vesicles, we visualized them using transmission electron microscopy. First, we used POPG vesicles with which maximum vesicle clearance was observed ( Figure 3). Both 2C-wt and 2C-L3A displayed the ability to remodel larger vesicles ( Figure 4A) into tubular membranes ( Figure 4B,C). However, similar to vesicle clearance data and structural data, a drastic reduction in membrane remodeling was observed with 2C-L3A ( Figure 4C). The tubes formed by the 2C-wt peptide were stable even up to six days, suggesting a stable association of the peptide to the lipid ( Figure 5). A similar behavior has been previously observed with other membrane remodeling amphipathic α-helical proteins, such as α-synuclein [25].
Membrane remodeling was also observed with POPC/POPE 2:1 and POPC/POPE 1:1 vesicles ( Figure 6). However, here we observed the remodeling of vesicles into two types of smaller tubules ( Figure 6C-F), as opposed to the very long membrane tubes that were observed with POPG vesicles ( Figure 4B). The first group was smaller tubes, as shown in Figure 6C for POPC/POPE 2:1 and Figure 6D for POPC/POPE 1:1. The second group was smaller vesicular structures, as demonstrated in Figure 6E for POPC/POPE 2:1 and Figure 6F for POPC/POPE 1:1. No membrane remodeling was observed with POPC/POPE 2:1 or POPC/POPE 1:1 vesicles in the presence of the 2C-L3A peptide ( Figure 6G,H).

Tubulation Is Accompanied by Vesicle Leakage
We inferred that a robust membrane remodeling that results in a significant change of membrane architecture would be associated with membrane disruption. To monitor this, we employed a vesicle leakage assay. The intact vesicles contained both ANTS, a fluorophore, and DPX, a quencher of fluorescence. Any disruption of liposomes that resulted in the leakage of ANTS or DPX would separate the fluorophore from the quencher and cause an increase in the fluorescence, therefore, an increase in fluorescence was correlated with an increase in leakage. Our data showed a significant leakage by both 2C-wt and 2C-L3A peptides with highly negatively charged POPG vesicles when 50 or 100 µM peptides were used ( Figure 7A). Although 2C-L3A displayed a reduced leakage compared to 2C-wt, the difference was not statistically significant. However, a significant difference (p < 0.05) was observed between 2C-wt and 2C-L3A peptides when they were incubated with POPC/POPE 2:1 ( Figure 7B) or POPC/POPE 1:1 ( Figure 7C) vesicles in that 2C-L3A displayed a significantly reduced leakage compared to 2C-wt. It is worth noting that 2C-L3A peptides did not show any noticeable leakage with POPC/POPE 2:1 or POPC/POPE 1:1 vesicles ( Figure 7B,C). A higher percentage of leakage was observed with POPC/POPE 2:1 vesicles compared to POPC/POPE 1:1 vesicles, which was in accordance with vesicle clearance and structural data using circular dichroism.

Mutations Reducing Helicity of the Amphipathic α-Helix of 2C Protein Block Poliovirus Infection
Given our results that the L3A peptide was defective to tubulate liposomes, we expected that the mutations, when incorporated into the viral genome, would affect the viral replication. To test this, we introduced the specific mutations L3A to the full-length cDNA of PV-wt to construct the PV-L3A mutant. Transcripts of PV-wt and the PV-L3A mutant were subsequently produced and used in the plaque assay to examine their infectivity. HeLa cell monolayers were transfected with a series of diluted PV transcripts. As shown in Figure 8, the PV-wt infectivity was readily detected at two days post RNA transfection, as plaques can be observed at all dilutions. However, the PV-L3A infectivity was non-detectable, as no visible plaque was formed even at the non-diluted condition. These results suggested that the three mutated amino acids are critical for PV infection, and the affected remodeling of liposomes by the amphipathic α-helix might account for the defective infection.
It has been demonstrated that synthesized peptides of BMV 1a helix A and the helix in SFV nsP1 only form helical structure upon binding to membranes, and are necessary and/or sufficient to target proteins to membranes [32,36,37]. Similarly, a recent report showed that the peptide of CHIKV nsP1 amphipathic α-helix also forms a helical structure when it binds to negatively charged lipid vesicles [35]. Although the purified CHIKV nsP1 protein was able to induce membrane remodeling, the peptide representing the amphipathic α-helix failed to reorganize liposomes [35]. In poliovirus 2C protein, the N-terminal fragment with 77 amino acids or 39 amino acids has been shown to be sufficient for the association with membranes [19] or LDs [21], respectively. In addition, the amphipathic α-helix in poliovirus 2C is necessary for the association with LDs [21]. These results indicate the critical role of the amphipathic α-helix for membrane association of poliovirus 2C protein. Our results for the first time confirmed that this predicted amphipathic α-helix in poliovirus 2C protein behaves as a helix: the 2C helix peptide increased helical conformation when binding to the lipid membrane, and the mutations in the hydrophobic face of the amphipathic region largely abolished this enhancement ( Figure 5). In addition, sole 2C helix peptide is sufficient to induce the remodeling of the membrane structure (Figures 2, 4 and 5), in contrast to CHIKV nsP1, of which the cytoplasmic domain of nsP1 is also required for membrane reorganization [35]. Many cellular proteins contain such amphipathic α-helices that are necessary for their membrane-associated functions [38]. Previous studies with endophilin A1 has shown that there could be multiple helices in a single protein that can mediate or regulate membrane interaction in a protein [39,40]. Altering the charge of the central helix of endophilin A1 changed the insertion depth of this helix that altered membrane remodeling [39]. Another example is of epsin 1, which is also known to bind membranes through an amphipathic α-helix [41]. Any mutation that affected the hydrophobic face of helix 0 modulated the membrane interaction of the protein [41].
Overall, our data show that the 2C peptide from the N-terminal amphipathic domain of poliovirus 2C protein has enough structural features to induce membrane remodeling, and the three amino acid residues that are part of the hydrophobic face of this amphipathic region could have a critical structural role for membrane interaction of this domain. Both electrostatic and hydrophobic interactions contribute to the membrane binding of 2C peptides, however, the contribution from electrostatic interactions is small (Figure 3). This is evident from only a marginal decrease in the helical content of 2C-wt in the presence of POPG vesicles compared to POPC/POPE lipid compositions. Furthermore, 2C-L3A displayed only a small increase in helical content in the presence of fully negatively charged phospholipid vesicles. The hydrophobic interaction, on the other hand, should have a major influence on the membrane interaction of the 2C peptide, as supported by the fact that mutations affecting the hydrophobic face of the 2C peptide strongly disrupt membrane interaction ( Figure 3).
The data from the plaque forming assay further underscore the role of the three hydrophobic amino acids in viral replication. Although the three substitutions may affect the processing of the 2BC intermediate to release 2C protein or the 2C protein conformation to interact with other viral or host proteins, it is highly likely that the inability of the 2C-L3A protein to infect HeLa cells is due to the compromising of the membrane remodeling ability. In the absence of detailed structural data, it is difficult to speculate how the mutations might affect the positioning of the amphipathic α-helix on the membrane surface. Further investigations to unravel the membrane-bound structure of the 2C amphipathic α-helix could provide answers to such questions.
Our observation provides the opportunity to design or search for ligands that can bind to the 2C N-terminal amphipathic α-helix, thereby disrupting the assembly of viral replication complexes and viral replication. Further studies could reveal whether targeting amphipathic α-helices in viral proteins, such as polioviral 2C, BMV 1a, nsP1 of SFV, or CHIKV, could be an antiviral approach in both plants and animals.

Conflicts of Interest:
The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.