Elucidation of the Rotavirus NSP4-Caveolin-1 and -Cholesterol Interactions Using Synthetic Peptides

Rotavirus (RV) NSP4, the first described viral enterotoxin, is a multifunctional glycoprotein that contributes to viral pathogenesis, morphogenesis, and replication. NSP4 binds both termini of caveolin-1 and is isolated from caveolae fractions that are rich in anionic phospholipids and cholesterol. These interactions indicate that cholesterol/caveolin-1 plays a role in NSP4 transport to the cell surface, which is essential to its enterotoxic activity. Synthetic peptides were utilized to identify target(s) of intervention by exploring the NSP4-caveolin-1 and -cholesterol interactions. NSP4112–140 that overlaps the caveolin-1 binding domain and a cholesterol recognition amino acid consensus (CRAC) motif and both termini of caveolin-1 (N-caveolin-12–20,  19–40 and C-caveolin-1161–180) were synthesized. Direct fluorescence-binding assays were employed to determine binding affinities of the NSP4-caveolin-1 peptides and cholesterol. Intracellular cholesterol alteration revealed a redistribution of NSP4 and disintegration of viroplasms. These data further imply interruption of NSP4112–140-N-caveolin-119–40 and cholesterol interactions may block NSP4 intracellular transport, hence enterotoxicity.


Introduction
As the leading cause of gastroenteritis in young children under the age of five, rotavirus (RV) infections annually are responsible for approximately 600,000 deaths worldwide [1,2]. During infection, the RV nonstructural protein 4 (NSP4) functions as a viral enterotoxin by binding an extracellular receptor and activating a signal transduction pathway which increases intracellular calcium ([Ca 2+ ] i ) levels through the release of ER calcium stores [3,4]. This increase in [Ca 2+ ] i induces secretory chloride currents which result in diarrhea, but only when initiated from the exofacial leaflet of the plasma membrane (PM) [3]. The initiated increases in intracellular, calcium levels fail to induce the chloride secretory response [4][5][6].
Traditionally defined as an ER glycoprotein, NSP4 contains a single transmembrane domain that serves to anchor the protein into the ER membrane, such that a short Nterminal domain (amino acids [aa] 1-24) remains within the lumen of the ER while the longer C-terminus (aa  extends into the cytoplasm [7]. Interactions with numerous viral and cellular proteins occur within the extended Cterminal tail of NSP4 [8][9][10][11][12]. Also contained within the Cterminal cytoplasmic tail is an amphipathic α-helix (AAH), coiled-coil domain (aa 95-137) [13,14]. Cross-linking and crystallographic experiments reveal that this region of NSP4 primarily oligomerizes into dimers and tetramers and contains a cation-binding site [13][14][15]. Residing within the amphipathic α-helix, coiled-coil domain is the enterotoxic peptide region (aa 114-135) as well as the caveolin-1 (cav-1) binding domain [16] and a putative cholesterol recognition amino acid consensus (CRAC) sequence [17].
Many cholesterol binding proteins contain a CRAC sequence characterized by-L/V-(X) ( Journal of Amino Acids where (X) (1−5) represents between one and five unspecified residues [17]. Numerous proteins involved in cholesterol transport contain this sequence, including cav-1 [18]. However, the presence of such a sequence does not assure an interaction with cholesterol, as this sequence is highly variable and likely is not the only requirement for cholesterol binding [19]. Yet a number of viral proteins bind cholesterol, including the HIV-1 gp41 transmembrane glycoprotein [17,20], the influenza M2 protein [21], and the F1 subunit of the fusion protein of Sendai virus [22].
There is compelling evidence that the enterotoxic peptide, NSP4 114−135 , and the full-length NSP4 protein interact with cav-1. In vivo laser scanning confocal microscopy (LSCM) colocalizations, fluorescent energy transfer (FRET) analyses, and co-immunoprecipitation data from our laboratory verify that NSP4 binds cav-1, the main constituent protein of caveolae, at multiple sites within the cell, including the PM [23]. Yeast two-hybrid and in vitro binding assays confirm an interaction between NSP4 and cav-1 and map the cav-1 binding domain to NSP4 114−135 [23]. Using deletion and site-directed mutagenesis, the binding site delineated to three hydrophobic residues in the enterotoxic region indicating that NSP4 and cav-1 associate via a hydrophobic interaction (unpublished). Additional studies reveal that the cav-1 binding site for NSP4 maps, to both the N-and C-termini of cav-1, residues 2-31 and 161-178, respectively [24].
Cav-1 is an intracellular, 21 kD protein bound by the inner leaflet of the PM in a hairpin-like structure such that both the N-and C-termini are oriented towards the cytoplasm [23,25]. Numerous signaling molecules, such as those involved in calcium signaling, are localized to caveolae and their function appears to be dependent on interaction with cav-1 [16,26,27]. Cav-1 also binds cholesterol, which has been shown to be vital to caveolae biogenesis and the transport of cholesterol to caveolae at the PM [28][29][30][31].
Purified NSP4 and NSP4 114−135 peptides interact with caveolae-like model membranes, that is, those that have a high radius of curvature and are rich in anionic phospholipids and cholesterol [32,33]. Upon interaction with model membranes, both the protein and the peptide undergo a change in secondary structure characterized by an increase in α-helix formation. Additional secondary structure studies of other NSP4 peptides that overlap the enterotoxic peptide and cav-1 binding domain confirm that this region is important for membrane interactions. When caveolae are isolated from enriched PM fractions of RV-infected MDCK cells, the full-length, fully-glycosylated NSP4 is present in the isolated caveolae fractions verifying transport of NSP4 to the PM and caveolae [34]. It is possible that the transport of NSP4 to caveolae occurs through an interaction with cav-1 and cholesterol, although additional studies are needed to for verification. Our studies reveal the exposure of the NSP4 C-terminus and the enterotoxic region, but not the Nterminus, on the outer PM leaflet and subsequent release into culture media at a time when the membrane is intact as verified by a lack of exposure of cav-1 and other intracellular markers [34][35][36].
In this paper, we have extended these studies using synthetic peptides to address the following questions. (i) Does NSP4 interact with both termini of cav-1 with equal affinity? (ii) Do the two termini of cav-1 interact with one another and thus facilitate binding of NSP4 to both termini? (iii) Is there a structural change when NSP4 and cav-1 peptides interact? (iv) Do the overlapping NSP4-cav-1 binding, CRAC motif, and enterotoxin regions interact with cholesterol? (v) Does alteration(s) of NSP4-cholesterol binding have a biological effect? To answer these questions, we employed peptide-peptide and -cholesterol direct fluorescence-binding assays to ascertain the interaction and determine the binding affinities (K d ). We also employed circular dichroism (CD) to investigate secondary structural changes upon interaction of the peptides in the absence or presence of model membranes.
Antibodies specific to the NSP4 peptide aa 150-175 (NSP4 150−175 ; deduced from the simian rotavirus SA11 NSP4 sequence) were generated in rabbits by immunizing with peptide cross-linked to keyhole limpet hemocyanin as previously described [23]. Anti-NSP4 150−175 was affinity purified against the inoculating peptide linked to preactivated cyanogen bromide Sepharose 4B beads according to the manufacturer (Amersham Pharmacia Biotech, Piscataway, NJ) [37]. Guinea pig anti-NSP5, a viroplasm (virus factory in the cell) marker, was a gift from Dr. Oscar Burrone (International Center for Genetic Engineering and Biotechnology, Trieste, Italy). F(ab ) 2 fragments of goat anti-guinea pig IgG-Cy3 and goat anti-rabbit IgG-Cy2 were purchased from Jackson Immuno Research (West Grove, PA).

Synthesis and Purification of NSP4 and Cav-1 Peptides.
All peptides (Table 1) were synthesized by fluorenylmethoxycarbonyl (Fmoc) solid-phase chemistry with either To separate the peptides from the solid support, the mixture was filtered through a sintered glass filter into a 50 mL conical tube containing cold diethyl ether in a dry ice/ethanol bath. Following 2-3 rinses of the filter with TFA, the ether solution containing the precipitated peptide was centrifuged at 300 g for 4 minutes. The supernatant was removed and additional cold diethyl ether was added to the peptide pellet. The mixture was allowed to cool in the dry ice/ethanol bath for several minutes before repelleting. This step was repeated an additional two times; the crude peptide was dried under N 2 , resolubilized in 10% acetic acid and lyophilized.
The lyophilized crude peptide was resuspended in 5-10% acetic acid and purified from organic contaminants and incomplete peptide fragments by gravimetric gel filtration chromatography (Sephadex G25 medium). The eluted peptide was lyophilized and further purified by reverse-phase HPLC using either a reverse phase C4 Delta Pak column (Waters Chromatography Division, Milford, MA) or a reverse phase C18 column (Beckman-Coulter, Fullerton, CA) depending on the hydrophobicity of the peptide. Eluted peaks were lyophilized and characterized by matrix-assisted laser desorption/ionization (MALDI) mass spectrometry (Laboratory for Biological Mass Spectrometry, Department of Chemistry, Texas A&M University, College Station, TX).

Preparation of Small Unilamellar Vesicles (SUV).
Small unilamellar vesicles (SUVs) composed of the neutral lipid POPC, cholesterol, and the anionic phospholipid DOPS in the molar ratio 55 : 35 : 30 were prepared following a previously published protocol [33]. This membrane composition was chosen as a result of previous data showing that this composition and curvature promote an interaction with NSP4-specific peptides as observed by an increase in α-helix formation [32,33]. Stock solutions of the lipids dissolved in chloroform were mixed together in the proper ratio in an amber glass vial. The solvents were removed under N 2 with constant rotation so that the dried lipids formed a thin film on the wall of the glass vial. The vial containing the dried lipids was further dried under vacuum for a minimum of 4 h. The dried lipids were resuspended in 10 mM MOPS buffer, pH 7.4 (filtered through a 0.45 μm filter; Millipore, Bedford, MA), vortexed and bath sonicated. The resulting multilamellar membrane suspension was sonicated with a microprobe under N 2 at 4 • C at 2 min intervals, followed by 1 min pauses to prevent overheating of the lipid solution. The sonicated lipid solution then was centrifuged at 110,000 g for 4 h to remove any multilamellar vesicles and titanium debris from the sonicator probe. The lipid concentration of the SUV solution was determined by a standard phosphate assay as described [2]. Briefly, the SUV solutions were mixed with a 10% magnesium nitrate (Mg(NO 3 ) 2 ) solution and heated for 30 min at 120 • C. Each sample was ashed in a flame, 600 μL of 0.5 N HCl was added, and the samples were heated in boiling water for 15 min. Each sample was mixed with 1400 μL of a 10% ascorbic acid solution that contained 6 parts of a 0.42% ammonium molybdate solution in 1 N sulfuric acid (H 2 SO 4 ) and incubated for 20 min at 45 • C before reading the absorbance at 660 nm (A 660 ). Lipid concentrations of the SUVs were determined by comparison to a standard curve of 0-200 nmol phosphate.  [20,39]. All of the cav-1 peptides directly were labeled with a Cy3 fluorescent dye, using the Cy3 Monoreactive dye pack from Amersham Biosciences according to the manufacturer's protocol. Briefly, each peptide was dissolved in 50 nM phosphate buffer, pH 7.0, at a concentration of 1 mg/mL, and added to a vial containing lyophilized Cy3 dye. The solution gently was mixed and incubated at 25 • C for at least 2 h with intermittent mixing. Labeled peptides were separated from unlabeled peptides by Sephadex G25 gel filtration chromatography.
To demonstrate a direct peptide-peptide interaction between the NSP4 112−140 peptide and each of the cav-1 peptides, increasing concentrations of NSP4 112−140 (10-500 nM) were added to 25-100 nM of the Cy3-labeled cav-1 peptides in 2 mL of phosphate-buffered saline (PBS, pH 7.4) in either the absence or presence of small unilamellar vesicles (SUVs, 1 μM). The Cy3 fluorophore was excited at 550 nm and emission spectra were scanned from 560 to 700 nm. Fluorescence emission spectra were obtained at 25 • C with a PC1 photon-counting spectrofluorometer with excitation and emission slit widths of 1.0 (ISS, Inc.). Spectra were corrected for background and maximal fluorescence intensities were recorded. Calculation of the dissociation constants (K d ) was performed from the titration curves plotted as quenching in Cy3-peptide fluorescence intensity (F 0 − F, where F 0 and F represented the Cy3-cav-1 peptide fluorescence intensities in the absence and presence of NSP4 112−140 or cav-1 peptide, resp., at each titration point) as a function of peptide concentration. The K d also was obtained by a reciprocal plot of 1/(1 − F/F max ) and C L /F/F max according to the following equation: y = bx + y 0 , where F is the fluorescence intensity at a given concentration of ligand, F max is the maximal fluorescence obtained, and C L is the ligand concentration. The slope of the line (b) is equal to 1/K d .

Cholesterol-Peptide Interaction.
Similarly, the NSP4 112−140 -cholesterol interaction and binding affinity were determined by a direct fluorescent binding assay in which the peptide was labeled with a Cy5 fluorophore using the Cy5 Monoreactive dye pack from Amersham Biosciences (Piscataway, NJ). A concentrated cholesterol stock solution (20 uM) was prepared by dissolving the cholesterol in ethanol. To analyze a binding interaction between NSP4 112−140 and cholesterol, an increasing quantity of cholesterol (5-35 nM) was added to 10 nM of the Cy5-labeled NSP4 112−140 in 2 mL of PBS (pH 7.4). The Cy5 fluorophore was excited at a wavelength of 649 nm, and the emission spectra were scanned from 655 to 720 nm using a PCI photoncounting spectrofluorometer (ISS, Inc., Champaign, IL). Fluorescence emission spectra were obtained at 25 • C with excitation and emission slit widths of 1.0. As with the peptidepeptide interactions, spectra were corrected for background and maximal fluorescence intensities were measured. Calculation of the K d was determined from the plotted titration curves whereby the Cy5-peptide fluorescence intensity was quenched and measured as a function of cholesterol concentration (F 0 − F, where F 0 and F represented the Cy5-NSP4 112−140 fluorescence intensities in the absence and presence of cholesterol, resp., at each titration point). The sigmoidal curves also were fitted to a Hill plot according to the following equation: , where y and x correspond to (F 0 − F) and the ligand concentration at each point, while a, b, and c represent the maximum binding (B max ), the number of binding sites (n), and the K d value, respectively. K d values were calculated with the sigmoidal function of Sigma Plot (SPSS, Chicago, IL) utilizing the Hill plot feature.

Circular Dichroism (CD) and Secondary Structure Estimations.
The secondary structures of the NSP4 and cav-1 specific peptides were determined by circular dichroism following a previously published protocol [32,33]. Briefly, each peptide (15-35 μM) was suspended in 10 mM potassium phosphate buffer, pH 7.4, or 50% trifluoroethanol (TFE) to promote a hydrophobic environment and folding of the peptide, in the presence or absence of lipid vesicles (1 mM). Samples containing buffer or SUVs without peptide were used for background correction. Peptide concentrations were determined by amino acid analysis (Protein Chemistry Laboratory, Department of Biochemistry, Texas A&M University, College Station, TX). CD spectra were obtained in a 1 mm circular quartz cell using a Model J-710 JASCO spectropolarimeter (JASCO, Easton, MD) or in a 1 mm rectangular quartz cell in a Model 202 Aviv spectrometer (Aviv Biomedical, Lakewood, NJ). The spectra for each of the peptides were recorded from 185 nm to 260 nm, with a step resolution of 1 nm, speed of 50 nm/min, response of 1 sec, bandwidth of 2.0 nm, and sensitivity of 10 mdeg. Data were averaged from 5 scans, background subtracted, smoothed, and converted into mean residue molar ellipticity [θ] (deg cm 2 /dmol).
To determine the percent α-helix for each of the NSP4and cav-1 specific peptides, the following equation was used: [12,21,22]. In this equation, θ 222 is the mean residue molar ellipticity at 222 nm, f h is the fraction in α-helical form,ί is the number of helices, κ is a wavelength-specific constant with a value of 2.6 at 222 nm, N is the number of residues in the peptide, and θ h 222∞ is the molar ellipticity for a helix of infinite length at 222 nm, that is, −39,500 deg cm 2 /dmol.

Treatment of RV-Infected MDCK and -HT29.f8 Cells with
Cholesterol-Altering Drugs. MDCK cells were infected with RV at a multiplicity of infection (MOI) of 2 and incubated for 12 hpi. The cells were washed and treated in triplicate with the cholesterol altering drugs, fillipin [40], and nystatin [41] exactly as described. Chlorpromazine [40,42] and nocodazole [39] were utilized as negative and positive controls, respectively. Cells were fixed with methanol : acetone (1 : 1) at −20 • C for 10 min, blocked with 5% dry milk for 20 min, incubated with anti-NSP4 150−175 and anti-rabbit-IgG-FITC, and examined under a UV microscope (Zeiss). The cell surface carefully was observed and the presence or absence of fluorescence carefully noted. Comparisons also were made to untreated cells. Primary and secondary antibody controls were negative. To ensure the lack of drug toxicity, a separate set of cells were treated and then stained with trypan blue.

Direct Fluorescence-Binding Assays
with Greater Affinity Than the C-Terminus. We previously demonstrated that NSP4 114−135 interacts with both the Nand C-termini of cav-1 using yeast-2-hybrid and pull-down assays, which is unusual [24]. To further investigate this finding and determine if there was a preferential binding of NSP4 to the N-or C-terminus of cav-1, the NSP4 112−140 peptide that encompasses aa 114-135 and various cav-1 peptides (N-Cav 2−20 , N-Cav 19−40 , and C-Cav 161−178 ) were utilized in a direct fluorescence-binding assay.
Upon titration with increasing concentrations of NSP4 112−140 , each of the Cy-3-labled N-or C-terminal cav-1 peptides showed an increase in fluorescence intensity at 565 nm, indicative of an interaction. Plots of NSP4 112−140 concentrations versus maximum fluorescence intensities revealed saturated binding curves for each of the cav-1 peptides ( Figure 1). Linear reciprocal plots were used to calculate the K d for each peptide-peptide pair (     [23]. It is unusual for a protein to interact with both termini of cav-1. We now show that there is a preferential interaction of NSP4 112−140 with the N-terminus of cav-1, which, to our knowledge, is the first report of a preferential binding to one termini of cav-1. Previous CD analyses of NSP4 peptides corresponding to the cav-1 binding domain show that aa 114-135 interacts with model membranes (SUVs) [32,33]. Therefore, NSP4cav-1 peptide-peptide interactions also were investigated in the presence of the same SUV membranes. The SUVs were mixed with each of the Cy3-labeled cav-1 peptides in PBS and then titrated with increasing concentrations of NSP4 112−140 (Figure 2).
The presence of lipid vesicles had no effect on the interaction of NSP4 112−140 with either N-Cav 19−40 or C-Cav 161−178 (Figures 2(b) and 2(c)). Binding affinities in the presence of membranes were similar to those calculated in the absence of membranes (Table 2). However, the presence of the SUVs resulted in an increased affinity between NSP4 112−140 and N-Cav 2−20 . The calculated K d value was determined to be 26±4 nM, an approximate 3-fold increase over that observed in the absence of membranes (85 ± 6 nM) and about a1.5fold increase over the K d of N-Cav 19−40 in the absence of membranes (40 ± 10 nM). These results strongly suggest that the presence of lipid vesicles enhances binding between NSP4 112−140 and the extreme N-terminus of cav-1 (aa 2-20), but not the adjacent N-terminal cav-1 peptide, N-Cav 19−40 . (aa 161-178). To evaluate the mechanism of the interaction of NSP4 with both cav-1 termini, we examined whether the two cav-1 termini bound one another to facilitate the interaction with NSP4 by employing the fluorescence-binding assay. C-Cav 161−178 was labeled with a Cy3-fluorophore and titrated with either the N-Cav 2−20 or N-Cav 19−40 peptide. Titration of C-Cav 161−178 with increasing concentrations of either N-terminal peptide resulted in no change in fluorescence intensity at 565 nm, indicative of a lack of association ( Figure 3). Hence, we propose that the N-and C-termini of cav-1 do not bind one another and the interaction between NSP4 and the cav-1 termini does not result from an initial binding between the cav-1 N-and C-termini. To our knowledge, this is the first report of the cav-1 termini not interacting. Additional studies are needed to dissect the mechanism and implications of both cav-1 termini binding NSP4.  (Figure 4, dark circles). The CD spectrum showed double minima at 208 and 222 nm and a single maximum peak at 190 nm, indicative of some α-helical secondary structure. Using the molar ellipticity value at 222 nm [43], the α-helical content of the NSP4 112−140 peptide was calculated to be 30.6 ± 2.4%.

Circular Dichroism
To enhance innate peptide structure, the secondary structure of NSP4 112−140 was determined in 50% TFE, a hydrophobic solvent that promotes intramolecular hydrogen bonding ( Figure 4, open circles). The CD spectrum showed a dramatic increase in α-helix formation of the peptide, as evidenced by an increase in both the negative molar ellipticity at 208 and 222 nm and the positive molar ellipticity at 190 nm. The α-helical content was calculated using the molar ellipticity value at 222 nm and determined to be 79.6± 4.6% (Table 3).
Previous CD experiments with the enterotoxic peptide, NSP4 114−135 demonstrate that the α-helical content increases in the presence of SUVs containing POPC, cholesterol, and DOPS at a 55 : 35 : 10 molar ratio [32,33]. To determine the secondary structural changes that occur in the presence of membranes, NSP4 112−140 was mixed with SUV membranes and changes in α-helical content were noted by CD.
The CD spectrum of the NSP4 112−140 peptide in the presence of 1 mM SUV showed nearly a 2-fold increase in α-helix formation over that in aqueous buffer, as evidenced by the increase in both the negative molar ellipticity as 208 and 222 nm and the positive molar ellipticity at 190 nm ( Figure 5). The α-helical content was calculated using the molar ellipticity value at 222 nm and determined to be 57.9 ± 1.6% (Table 3).   (Figure 6, open circles). Only N-Cav 19−40 and C-Cav 161−178 showed an increase in α-helix formation in the presence of 50% TFE. When mixed with the SUVs, none of the cav-1 peptides demonstrated a change in α-helix formation over that seen in aqueous buffer ( Figure 6, dark triangles). This result was anticipated as none of these peptides corresponded to a cav-1 membrane interacting or transmembrane domain. of each peptide (theoretical). The α-helical content was calculated and significant differences (Student's t-test, P < 0.05) between the observed and theoretical values indicated a change in secondary structure and binding between the two peptides. However, a lack of secondary structural alterations upon mixing the peptides did not negate binding between the NSP4 112−140 and the cav-1 peptide being tested. While direct fluorescent binding assays revealed the affinity of each of the peptide-peptide interactions, the CD experiments disclosed the extent to which the interaction(s) caused a change in the secondary structure. When each of the cav-1 peptides was mixed with NSP4 112−140 in an aqueous buffer, only the NSP4 112−140 -C-Cav 161−178 mixture showed a significant difference between the observed and theoretical values (25.0 ± 1.7% versus 35.3 ± 1.6% α-helix) (Figure 7(d); Table 4 Table 4). Similarly, when each peptide-peptide pair was placed in 50% TFE, only the NSP4 112−140 -C-Cav 161−178 mixture showed a significant change in secondary structure (44.2 ± 10.6% versus 67.8 ± 3.2%) ( Table 4). These data confirmed the weak binding between NSP4 112−140 and C-Cav 161−178 and verified that secondary structure alterations do not always occur upon an interaction.

Effect of SUV Model Membranes on the NSP4 112−140 -Caveolin-1 Peptide-Peptide Interactions.
Cav-1 is critical to both the structure and function of caveolae, is the major constituent of caveolae, and is controller of caveolae formation [44]. NSP4 is present in isolated caveolae from RVinfected cells [34], and together with NSP4 114−135 , preferentially interacts with caveolae-like model membranes [32]. Therefore we evaluated the extent the SUVs influenced secondary structure alterations of the NSP4-cav-1 peptidepeptide interactions. Each NSP4 112−140 -cav-1 peptide pair was monitored by CD (Figure 8).  Even though the NSP4 112−140 -C-Cav 161−178 peptide pair showed the most change in α-helical content in aqueous buffer and 50% TFE, in the presence of SUVs there was not a significant conformational change indicating that this interaction was not enhanced by the presence of model membranes (Table 4). Similarly, the NSP4 112−140 -N-Cav 19−40 showed the strongest affinity by direct fluorescent binding assays (Table 2), yet this interaction failed to result in a significant change in secondary structure when mixed with membranes. The observed α-helical content did not vary from the expected theoretical value, regardless of whether the peptide mixture was in aqueous buffer, 50% TFE, or mixed with SUVs. These results suggest that the strong binding between NSP4 112−140 and N-Cav 19−40 required neither a change in structure nor the presence of specific lipids.
Only the NSP4 112−140 -Cav 68−80 (negative control) mix ture showed a significant change in the α-helical content in the presence of the SUVs (31.3 ± 0.2% versus 37.5 ± 1.9%) (Figure 8(c); Table 4). It is currently unclear why the control peptide mixture resulted in an alteration in helical structure and requires further study.

NSP4 112−140
Peptide-Cholesterol Interaction. The NSP4 112−140 peptide, which contains the CRAC motif, was labeled with a Cy5-fluorophore and titrated with cholesterol in a direct fluorescence-binding assay. The maximum solubility of cholesterol in water is 4.7 μM and the critical micellar concentration (CMC) is 25-40 nM [18], so cholesterol concentrations were kept below these limits in the binding assays.
In the absence of cholesterol, the Cy5-NSP4 112−140 pep tide showed maximum fluorescence at 665 nm ( Figure 9). Upon titration with increasing concentrations of cholesterol

Cholesterol Altering Drugs Disrupt NSP4 Transport.
Specific inhibitors aid in the dissection of biological processes. We evaluated the effect(s) of drugs known to disrupt cholesterol (fillipin, nystatin, lovastatin) [40,41,45], hence caveolin/caveolae transport properties [46], an inhibitor of clathrin-coated pit processes (chloropromazine) [40,42], and a disruptor of the cytoskeleton (nocodazole) [39]. Cells were treated at the recommended effective dose, evaluated with trypan blue to ensure a lack of toxicity, and NSP4 transport to the cell periphery was noted. Whenever cholesterol trafficking was disrupted, NSP4 likewise failed to transport to the cell surface as evidenced by a lack of peripheral staining ( Table 5).

Discussion
Several reports highlight the use of peptides as a means of investigating the interactions of full-length proteins [3,[47][48][49][50]. This study similarly demonstrated the utility of synthetic peptides by showing differential binding of NSP4 112−140 to the cav-1 termini and disclosing secondary structural changes that were not necessary for the interaction between the NSP4 and cav-1 binding domains. We also demonstrated a direct interaction between NSP4 and cholesterol. The K d was calculated as 8.0 ± 1 nM and a Hill plot revealed a single binding site for cholesterol when reacted with NSP4 112−140 . An interaction between NSP4 114−135 and the cellular protein cav-1 [23][24][25] as well as caveolae is well established [34]. Additionally, we report the interaction of NSP4 with both the N-and C-termini of cav-1 (aa 2-31 and 161-178, resp.) based on yeast two-hybrid analyses and  in vitro peptide binding assays [24]. Herein, we confirmed the interaction between NSP4 and the N-and C-termini of cav-1 and calculated the binding affinities of those interactions using peptides and direct fluorescence-binding assays. NSP4 112−140 , which encompasses the cav-1 binding While we did not specifically analyze the oligomerization of the NSP4 112−140 peptide in this study, our previous data on a similar and overlapping peptide (NSP4 114−135 ) showed oligomerization/aggregation at a peptide concentration of ∼50 uM [32]. The peptide concentration utilized in the binding assays stayed within the nM range, well below the 50 uM concentration at which the oligomerization occurred with the NSP4 114−135 peptide.
Following confirmation of an interaction between NSP4 and both termini of cav-1, we analyzed potential secondary structural changes upon peptide-peptide association by CD. Previous secondary structure analysis of NSP4 114−135 yielded an estimate of 37% α-helix for the peptide [32,33]. Analyses of the secondary structure of the full-length NSP4 protein showed ∼26% α-helix when analyzed by CD [32]. To our knowledge, all other structural studies of NSP4 were completed on specific fragments of NSP4 and not the full-length  protein. The C-terminal region of NSP4 clearly is helical [13]. CD analysis of NSP4 112−140 showed 30% α-helix, which is within the range calculated for the enterotoxic peptide and the full-length protein. When placed in the presence of SUVs, which mimics the more biologically relevant caveolae environment, NSP4 112−140 demonstrated an increase in αhelical content. This increase in α-helix formation indicated that this peptide behaved similarly to the NSP4 114−135 peptide and the presence of membranes influenced the structure.
To date the crystallographic structure of cav-1 has not been resolved; however, limited secondary structural studies and bioinformatics analyses have provided insight into its folded conformation. CD analysis of aa 1-101 reveals that the N-terminus contains 20% α-helix, with aa 79-96 constituting the α-helix portion, while aa 1-78 likely lacks significant secondary structure [51]. An additional study, which used a bioinformatics approach, notes that aa 95-101 contains α-helical structure, while aa 84-94 forms two βstrands [52]. Further, cav-1 is known to form high molecular weight oligomers that requires the cav-1 scaffolding domain (CSD, aa 82-101) [52][53][54]. Additionally, the C-terminus is palmitoylated on three cysteines, which helps anchor the protein to the membrane and may thus explain the preferential binding of NSP4 to the N-terminus of cav-1 [54][55][56].
The data presented herein expand on the currently reported structural information of cav-1. Peptides corresponding to the N-terminus of cav-1 displayed ∼20% αhelix, whereas the cav-1 C-terminal peptide unexpectedly showed 43% α-helix, twice that of the N-terminus when peptides were analyzed by CD. This was surprising as the bioinformatics study predicted that the C-terminus would contain primarily random secondary structure. Hence, our data show that the cav-1 C-terminus contains helical structure, suggesting that it may not exclusively be random, and supports the consensus structural predictions of cav-1 aa 134-167 [52].
CD analysis of NSP4-cav-1 peptide-peptide interactions revealed that there was no significant change in α-helix formation upon binding of NSP4 112−140 and the N-Cav 19−40 peptide. However, a structural change was noted upon interaction of NSP4 112−140 with N-Cav 2−20 and C-Cav 161−178 . Taken together with the binding assays, these results reveal that while the interaction between NSP4 112−140 and N-Cav 19−40 occurs with a strong affinity, binding does not require a conformational change. However, the secondary structure change observed when the NSP4 112−140 and C-Cav 161−178 peptides were mixed likely occurred to allow the two peptides to interact, albeit more weakly.
A statistically significant conformational change was observed when the NSP4 112−140 -Cav 68−80 mixture was combined with model membranes. Since cav-1 68−80 is located near the N-terminal membrane attachment domain (N-MAD; aa 82-101), as well as the transmembrane domain (aa 102-134) of cav-1, it was postulated that the observed conformational change was due to the interaction of Cav 68−80 with the lipid vesicles. However, when Cav 68−80 was mixed with SUVs in the absence of NSP4 112−140 , no change in secondary structure was observed. Extensive studies are needed to dissect how the SUVs alter the structure of the NSP4 112−140 and Cav 68−80 mixture.
The interaction of NSP4 with both termini of cav-1 is unique, as most cav-1 binding proteins interact with the CSD [57,58]. Since NSP4 (aa 112-140) interacted with the N-terminus of cav-1 more strongly than the C-terminus, it is possible that the interaction with the C-terminus is transitory. NSP4 activates signaling pathways that mobilize [Ca 2+ ] i , requiring the protein to interact with membraneassociated signaling molecules [59,60]. The weak binding between NSP4 and the C-terminus of cav-1 may serve to present and/or orient the viral protein for interactions with signaling molecules or lipids, such as the phosphoinositides. Once NSP4 is properly presented to signaling molecules within the membrane, the C-terminus of cav-1 may then quickly disassociate thus allowing NSP4 to carry out its biological function(s). Since the C-terminus of the cav-1 protein is closely associated with membranes and does not extend into the cytoplasm to the same extent as the N-terminus, it likely has less flexibility and/or accessibility for interactions with other proteins. This lack of freedom is another potential explanation for the weak binding observed between NSP4 and the C-terminus of cav-1.
It has been shown that NSP4 is released from RV-infected cells in different forms dependent on the cell type, virus strain, and MOI [36,61,62]. The differential binding of NSP4 to the N-and C-termini of cav-1 could also contribute to the presentation and transport of NSP4 across the PM and provide a key target to block NSP4 from exiting the cell. Many models can be envisioned to explain how NSP4 interacts with both cav-1 termini. Results of the fluorescencebinding assay showed that the N-and C-termini of cav-1 do not interact, so this possibility can be ruled out. It is likely that different molecules of NSP4 associate with the individual cav-1 termini.
Lastly, we demonstrated that an RV NSP4 peptide directly interacted with cholesterol and this interaction overlapped the amphipathic α-helix, the enterotoxic peptide, and cav-1 binding domains (aa 112-140), as well as a putative CRAC motif. The interaction between cholesterol and the NSP4 CRAC sequence has implications for the trafficking of NSP4 in RV-infected cells. Given that caveolae are involved in de novo cholesterol transport from the ER to PM caveolae [30,39] and cav-1 assists in cholesterol transport [29,30,63,64], contains a CRAC motif, and binds cholesterol in a 1 : 1 ratio, it is reasonable to propose that cav-1 and cholesterol function in NSP4 intracellular localization and transport [65,66]. Indeed the interaction between NSP4 and cholesterol supports the hypothesis that NSP4 traffics via a cholesterol transport pathway in RV-infected cells. Likewise, NSP4 interacts with cav-1, has a proposed CRAC motif, traffics to the cell surface via a Golgi-bypassing, unconventional pathway [7,34,36,67], and, as we show herein, interacts with cholesterol with a K d of 8 ± 1 nM. We also showed that the NSP4-cholesterol interaction was specific to aa 112-140, as there was no binding to a C-terminal NSP4 150−175ΔAla peptide.
Based on these results using synthetic peptides, we hypothesize that NSP4 traffics with cholesterol and cav-1 from the ER to the PM, and to viroplasms via association to cholesterol. Additional studies are needed to verify this hypothesis, dissect NSP4 transport to the PM, and resolve viroplasm localization, all of which may constitute novel targets of reducing NSP4 effects.