Mutational analysis of the rotavirus NSP4 enterotoxic domain that binds to caveolin-1

Background Rotavirus (RV) nonstructural protein 4 (NSP4) is the first described viral enterotoxin, which induces early secretory diarrhea in neonatal rodents. Our previous data show a direct interaction between RV NSP4 and the structural protein of caveolae, caveolin-1 (cav-1), in yeast and mammalian cells. The binding site of cav-1 mapped to the NSP4 amphipathic helix, and led us to examine which helical face was responsible for the interaction. Methods A panel of NSP4 mutants were prepared and tested for binding to cav-1 by yeast two hybrid and direct binding assays. The charged residues of the NSP4 amphipathic helix were changed to alanine (NSP446-175-ala6); and three residues in the hydrophobic face were altered to charged amino acids (NSP446-175-HydroMut). In total, twelve mutants of NSP4 were generated to define the cav-1 binding site. Synthetic peptides corresponding to the hydrophobic and charged faces of NSP4 were examined for structural changes by circular dichroism (CD) and diarrhea induction by a neonatal mouse study. Results Mutations of the hydrophilic face (NSP446-175-Ala6) bound cav-1 akin to wild type NSP4. In contrast, disruption of the hydrophobic face (NSP446-175-HydroMut) failed to bind cav-1. These data suggest NSP4 and cav-1 associate via a hydrophobic interaction. Analyses of mutant synthetic peptides in which the hydrophobic residues in the enterotoxic domain of NSP4 were altered suggested a critical hydrophobic residue. Both NSP4HydroMut112-140, that contains three charged amino acids (aa113, 124, 131) changed from the original hydrophobic residues and NSP4AlaAcidic112-140 that contained three alanine residues substituted for negatively charged (aa114, 125, 132) amino acids failed to induce diarrhea. Whereas peptides NSP4wild type 112−140 and NSP4AlaBasic112-140 that contained three alanine substituted for positively charged (aa115, 119, 133) amino acids, induced diarrhea. Conclusions These data show that the cav-1 binding domain is within the hydrophobic face of the NSP4 amphipathic helix. The integrity of the helical structure is important for both cav-1 binding and diarrhea induction implying a connection between NSP4 functional and binding activities.


Background
Rotaviruses (RV) induce a secretory and malabsorptive diarrhea, both of which are multi-factorial [1,2]. RV NSP4 is the first described viral enterotoxin and induces early secretory diarrhea in rodents [1,[3][4][5][6]. Exogenously added NSP4 mobilizes intracellular calcium ([Ca 2+ ]i) levels through an integrin receptor-mediated, phospholipase C (PLC)-dependent pathway [3,6], whereas endogenously expressed NSP4 mobilizes [Ca 2+ ]i by a PLC-independent mechanism, i.e. by NSP4 functioning as a viroporin in the endoplasmic reticulum (ER) [7,8]. These data indicate unique activities of NSP4 in different environments. At the exofacial surface of the plasma membrane (PM), increasing evidence has established that NSP4 activates a calcium signal transduction pathway with the release of chloride and water into the lumen of the gut [9,10]. It has been hypothesized that de novo NSP4 is released from RV-infected cells whereupon it binds to neighboring or the same cell to initiate secretory diarrhea [5,11,12].
In addition to its enterotoxic activity, NSP4 performs multiple intracellular functions that contribute to RV morphogenesis and replication. Early reports show NSP4 is an ER transmembrane glycoprotein that serves as an intracellular receptor that binds double layered particles [13][14][15][16], and facilitates entry into the ER and acquisition of the outer coat to form triple layered particles with a transient ER membrane [10,[15][16][17]. Subsequent silencing studies (siRNA) of RV-infected cells reveal that in the absence of NSP4 there are: (1) abnormal distributions of viral proteins in the viroplasm [18]; (2) little to no infectious viral particles present in the cell; (3) accumulations of empty viral particles [18,19], and (4) an up-regulation of viral transcription [19]. Taken together, NSP4 appears to function in viral pathogenesis, replication, and morphogenesis, as well as serving as an enterotoxin that induces calcium signaling events and fluid loss [20].
An NSP4 C-terminal cleavage fragment (residues 112-175) isolated from culture media of RV SA11-infected (MOI = 20; Sf9 cells) or NSP4 112-175-transfected cells show that at least a portion of NSP4 leaves the ER and the cell [21]. Bugarcic and Taylor report the secretion of a larger, alternately glycosylated 32 kD protein (MOI = 10, HT29 cells) [22], but signaling function of this larger NSP4 form was not reported. Other studies note the presence of NSP4 at multiple locations in RV SA11-infected cells [11,[23][24][25][26]. For example, Boshuizen et al report the presence of NSP4 at the basolateral surface of polarized cells in association with the extracellular matrix proteins, lamininbeta 3 and fibronectin [11], while also showing evidence of apical release [22]. In addition, NSP4 colocalizes with LC3, an autophagic vesicle marker, tubulin, and VP5, a RV viroplasm marker [12,27,28]. The recent identification of integrins alpha 1 beta 1 and alpha 2 beta 1 as the NSP4 cell receptors support the previous observations that NSP4 binds to the outside of the cell and stimulates a PLC signaling pathway [12]. Our recent data show full-length (FL) NSP4 is released from intact cells and binds neighboring cells [29]. Taken together, these data indicate that a pool of NSP4 interacts with several host-cell molecules that may influence its movement from the ER to the outer leaflet of the PM and subsequent release from the cell.
Collectively, our previous data show (a) preferential binding of FLNSP4 and NSP4 114-135 peptide to caveolaelike model membranes [30,31]; (b) transport of FLNSP4 to the PM and caveolae by a Golgi-bypassing, unconventional transport pathway [26]; (c) a direct interaction cbetween the structural protein of caveolae, caveolin-1 (cav-1), and NSP4 in yeast and mammalian cells [25]; (d) the presence of full length, endoglycosidase H (Endo-H)sensitive FLNSP4 in caveolae microdomains isolated from PM fractions of infected cells [26]; (e) FLNSP4 traffics to the PM, exposing the C-terminus to the extracellular space and subsequently is released from the cell; and (f ) K d analyses of FLNSP4 and cav-1 peptides showed preferential binding to the N-terminus of cav-1, with an increased affinity in the presence of cholesterol-rich model membranes, as well as a direct interaction between FLNSP4 and cholesterol [32].
Cav-1 is localized in the intestine, interacts with cholesterol, functions to transport de novo synthesized cholesterol to and from the ER and caveolae, and organizes signaling molecules, including those regulating calcium homeostasis [33][34][35]. The NSP4 binding site for cav-1 maps to aa 114-135, the enterotoxic domain, using in vivo yeast two-hybrid (Y2H) assays, in vitro peptide binding assays, co-immunoprecipitation (co-IP) reactions, and laser scanning confocal microscopy (LSCM) co-localization and fluorescent resonance energy transfer (FRET) analyses [25]. The enterotoxic domain of NSP4 interacts with both the N-and C-termini of cav-1 [36]. This result was surprising as cav-1 forms a hairpin loop in the cytofacial leaflet of the PM such that the two termini appear physically separated in the cytosol of the cell [37][38][39][40][41].
The goal of this study was to resolve the details of the NSP4-cav-1 interaction by (i) elucidating the type of interaction between NSP4 and cav-1, (ii) mapping the precise binding residues, and (iii) determining the extent to which interaction with cav-1 influences enterotoxic activity.
The interactions of these FLNSP4 mutants with cav-1 were evaluated by yeast-two-hybrid (Y2H) assays by reacting the Gal4 activation domain fusion protein with the NSP4:Gal4 binding domain fusion proteins. The plasmids were co-transformed into the yeast strain MaV203 as previously described [25]. The growth patterns of the co-transformed yeast confirmed the activation of the three reporter genes of the Y2H assay and established FLNSP4-Ala6 as positive for interacting with cav-1. In contrast, the FLNSP4-HydroMut failed to activate the three reporter genes indicating a lack of binding (data not shown).
Previous Y2H assays show that the reactions are clearer in the absence of the first two N-terminal hydrophobic domains of FLNSP4 [25]. Therefore, new mutants, NSP4, NSP4 Ala6 , and-HydroMut , that lack the first two hydrophobic domains, were cloned into the Gateway® entry clone, pENTR11, and inserted into the pDest32 plasmids to produces the Gal4 binding domain fusion proteins. The Gateway® entry clone, pECav-1 was used to insert the cav-1 sequence into the pDest22 plasmid that produces a fusion protein with the Gal4 activating domain for Y2H assays [25].
The growth patterns of the co-transformed yeast were monitored for activation of transcription from three chromosomally integrated reporter genes Ura3, His3, and lacZ ( Table 2). To assess the success of the co-transformation and transcription activation, four growth phenotypes were determined using media lacking the following amino acids: 1) Leucine and tryptophan; 2) Leucine, tryptophan, and histidine with 50 mM 3-Amino-1,2,4-Triazol (3-AT); 3) Leucine, tryptophan, and uracil; 4) Leucine, tryptophan, with 5-fluoroorotic acid (5FOA). The growth patterns were qualitatively scored and compared to the phenotypes that are defined in the Y2H manual [43] as positive and negative interactions (data not shown).
Induction of the lacZ reporter gene was confirmed using the quantitative chlorophenol red-beta-D-galactopyranoside (CPRG) assay for beta-galactosidase activity and reported as beta-galactosidase units (BGU). The negative yeast control had 0.134 BGU while truncated NSP4 46-175 had 1.39 BGU. NSP4 HydroMut and NSP4 Ala6 had 0.105 and 1.822 BGU, respectively, indicating a negative and positive interaction, respectively ( Table 2).
The growth patterns of the co-transformed yeast confirmed the activation of the three reporter genes of the Y2H assay. The N-terminally deleted NSP4 Ala6 showed a positive reaction for interacting with cav-1. In contrast, the N-terminally deleted NSP4 HydroMut failed to activate the three reporter genes. These data suggested alteration of the hydrophobic face of the amphipathic alpha helix of the NSP4 enterotoxic domain (I113, V124, and Y131), failed to bind cav-1.
All three NSP4  HydroMut revertants associate with cav-1 by Y2H analyses and peptide pull-down assays To further localize the cav-1 binding domain of the Nterminal deletion mutant (NSP4  , each of the mutant residues in NSP4 HydroMut individually were reverted back to the original hydrophobic amino acid [Rev1(113I), Rev2(124 V) and Rev3(131Y), Table 1], and transformed into yeast with cav-1 for evaluation by Y2H assays. All three revertants showed growth patterns of the cotransformed yeast confirming the activation of the three reporter genes of the Y2H assay (data not shown). Induction of the lacZ reporter gene was confirmed using the quantitative CPRG assay for beta-galactosidase activity reported as BGU. The revertants (Rev1, Rev2, Rev3) yielded 3.941, 3.042, and 3.195 BGU, respectively, which was 30 times higher than the negative yeast control (0.134 BGU), and over 2 times higher than NSP4  (1.39 BGU) ( Table 2). These data indicated all 3 revertants interacted with cav-1, (See figure on previous page.) Figure 1 A and B Organization of the Polar and Hydrophobic faces of the NSP4 crystal structure of the enterotoxic AAH (A and B respectively). A shows the acidic amino acids, D114, E122, E125 and D132 depicted in a maroon color. The basic amino acids K115, R119, R129 and K133 are shown in blue. B shows the hydrophobic amino acids of the AAH of NSP4 I113, L116, V124, L127, I130, Y131 and L134 in purple. C PyMol representations of the hydrophobic face of NSP4-(residues 46-175) and mutants. Amino acids in purple (I113, V124 and Y131) are indicated by arrows represent the wild type NSP4. Eight mutant clones are viewed below the wild type NSP4. The mutations, I113R, V124D and Y131K are orange, and amino acids that are mutated to alanines are shown as blue. To the right of each represented clone, the results of the yeast two hybrid and the peptide binding assays are given. The names and the sequences of the oligonucleotides that were used to clone the mutant clones.
so a single hydrophobic residue between NSP4 124-131 was sufficient for binding.

NSP4-and cav-1-fusion proteins are present in the co-transformed yeast
To verify the presence of the NSP4 and cav-1 GAL-4 fusion proteins in the yeast that showed no binding to cav-1 in the Y2H assay, the co-transformed lysates were probed with either anti-NSP4 150-175 ( Figure 3A and C) or anti-cav-1 2-31 ( Figure 3B and D) peptide-specific antibodies. Controls included untransformed MaV203 yeast lysates. The controls failed to show an antibodyspecific band at the correct molecular weight of the fusion proteins for both the activating domain (AD)-NSP4 and binding domain (BD)-cav-1. However, a non-specific band at~28 kD was observed when blotted with both antisera (Figure 3, lane 1). All NSP4 and cav-1 fusion proteins tested expressed at the correct molecular weight (34.4 kD and 34.9 kD, respectively) and were reactive with NSP4- (Figure 3, panels A&C) and cav-1 peptide-specific antisera (panel B&D). These data established that the encoded sequences from both plasmids were translated as fusion proteins in the co-transformed yeast, even when no reactivity was detected.
With cav 161-178 , a NSP4 specific band at~21 Kd (lane 8), also was observed. Two rev2 mutants likewise bound both cav-1 peptides (lanes 4, 5 and 11, 12), but only bound the NSP4dimeric form. A larger dimer also was present in lanes 5 and 12, further indicating the larger molecular weight band is a multimeric form of NSP4. Together, these data revealed that L116 is important for a proteinprotein interaction with cav-1.

Mutations within the amphipathic helix of NSP4 peptides abolish diarrheagenic activity in mouse pups
Our previous study delineates the cav-1 binding site between NSP4 residues 114-135 (24). Hence, three peptides encompassing amino acids (aa) 112-140 of NSP4 were synthesized, which contained mutated residues in the hydrophobic (NSP4 Hydro112-140 ), acidic (NSP4 AlaAcidic112-140 ) or basic (NSP4 AlaBasic112-140 ) face of the amphipathic alpha-helix (AAH). The sequences of each of the NSP4 mutant peptides are given in Additional file 1: Table S1. NSP4 Hydro112-140 contains three charged amino acids (aa113, 124, 131) that were altered from the original hydrophobic residues. NSP4 AlaAcidic112-140 and NSP4 AlaBasic112-140 each contained three alanine residues substituted for negatively charged (aa114, 125, 132) and positively charged (aa115, 119, 133) residues, respectively. Balb/C mouse pups (6-10 days) were administered wild type (wt) or mutant peptides by the intraperitoneal route and monitored for diarrhea as previously described [3]. Prior to injection, all purified peptide samples were tested for endotoxin by the Limulus Amebocyte Lysate (LAL) test (Associates of Cape Cod, Inc.) to ensure that samples were endotoxin free. A value of 0.5 EU/mL or less was considered acceptable for use in the diarrhea studies. Mouse pups injected with endotoxin free PBS served as a negative control. Mouse pups were monitored for diarrhea every 2 hours for 12 hours and then at 24 hours post injection. The severity of the diarrhea was scored on a scale of 1 to 4 (data not shown) (5). Following injection, both the wt-NSP4 112-140 and NSP4 AlaAcidic112-140 peptides caused diarrhea in 50% and 67% of the mouse pups tested, respectively (Additional file 1: Table S1). NSP4 Hydro112-140 and NSP4 AlaBasic112-140 peptides, however, did not induce diarrhea in any of the mouse pups tested. These preliminary results suggest that residues within both the hydrophobic face and the basic face of the amphipathic alpha-helix of NSP4 are also important for enterotoxic function. The mutant NSP4 peptides had an altered secondary structure when compared to NSP4 112-140wt To determine any alteration in structure of the mutant peptides, which may influence function, circular dichroism (CD) was employed. CD spectra of the mutant NSP4 (mtNSP4) peptides in aqueous buffer differed from that of the wild-type peptide ( Figure 5B-D, dark circles). The alpha-helical content was calculated based on the molar ellipticity values at 222 nm. The NSP4 HydroMut112-140 peptide seemed most affected by the mutations made in the AAH, as the alpha-helical content was determined to be 19.4 ± 2.2% (Additional file 1: Table S1). The mutations made in each of the charged faces of the AAH had less of an effect on alpha-helix formation within the  corresponding peptides. The alpha-helical content for the NSP4 AlaAcidic112-140 and NSP4 AlaBasic112-140 peptides was determined to be 23.7 ± 0.6% and 26.2 ± 2.9%, respectively (Additional file 1: Table S1).
In the presence of 50% trifluoroethanol (TFE, known to promote a hydrophobic environment and to enhance folding of synthetic peptides) each of the mutant NSP4 peptides increased in their alpha-helical secondary structure (2.5-3 times that in aqueous buffer) ( Figure 5B-D, open circles). As in aqueous buffer, the NSP4 HydroMut112-140 peptide exhibited less alpha-helical formation (57.8 ± 0.6%) than the NSP4 AlaAcidic112-140 and NSP4 AlaBasic112-140 peptides (73.1 ± 7.3% and 75.3 ± 2.8%, respectively) Additional file 1: Table S1. Therefore, the mutations made within the hydrophobic face of the AAH/cav-1 binding domain of NSP4 most negatively affected the alpha-helix folding of the peptide. Mutation of residues within the charged faces, however, only had a slight affect (NSP4 AlaAcidic112-140 ) or no affect (NSP4 AlaBasic112-140 ) on the alpha-helix structure of the peptides.

Conservation of key NSP4 residues between different RV strains
NSP4 amino acids 113-135 obtained from GenBank of twelve RV strains that represent groups A through E. (accession no. ABZ04174, BAA24144, ABZ04170, BAB83830, AAA64924, AAL11029, AAB58698, AAD 50676, ABV66094, BAA13728, P08434, AND P04512) were aligned with CLUSTAL ( Figure 6). The following observations were made: (i) all aligned sequences included I113 and L127 except Human RV C I113M and L127K; (ii) all twelve strains contained L116; (iii) position 124 was V with one exception, Human RV C V124D; (iv) L134 was conserved with the exception of EDIM L134M and Human RV C L134I, another hydrophobic amino acid; and the residue 131 varied from Y131, H131 (NCDV, OSU, Wa, RRV), S131 (Human RV C). The high similarity of I113, L116, V124, L127, and L134 infer the importance of these residues. Given the high sequence divergence recently reported in NSP4 sequences, the conservation of the hydrophobic residues is all the more remarkable [44].

Methods
Antibodies NSP4 and cav-1 peptide-specific antibodies were generated in New Zealand white rabbits to NSP4 residues 150-175 and cav-1 residues 2-31, respectively, following established protocols (5). Antibodies were purified against the peptide and used as primary antibodies in Western blots as previously described [25]. All primary antibodies were detected by horseradish peroxidase (HRPO)-labeled goat anti-rabbit IgG (Pierce, Rockford, Il).

Mammalian cell lines and yeast strains
MDCK cells were obtained from ATCC (Rockville, MD) and maintained in Dulbecco modified Eagle medium (DMEM; Gibco, Grand Island, NY) supplemented with 10% fetal bovine serum (FBS), glutamine (2 mM), penicillin-streptomycin (100 ug/ml) and non-essential aa (1×) (Sigma, St. Louis, MO). To compare the NSP4 expressed in yeast to that expressed in RV-infected mammalian cells, MDCK cells were infected with RV SA114F, lysates were collected, and subjected to Endo H digestion as previously described [45,46]. The cell lysates also were utilized as a positive control with the pulldown assays.
The transformed yeast colonies were grown at 30°C for 3 days on CSM Leu -Trpto identify colonies containing both plasmids. The activation of transcription of three independent reporter genes (URA, His3,lacZ) were monitored by the observation of yeast growth patterns on specific media. URA3 was monitored by cell growth patterns on CSM Leu -Trp -Uraand CSM Leu -Trp -+ 0.2% 5-fluoroorotic acid (FOA); 5FOA inhibits the growth of yeast. When HIS3 was transcribed, the growth of the co-transformed yeast was inhibited in a dosedependent manner by adding 3-amino-1, 2, 4-triazole (3AT) to CSM Leu -Trp -His - [50]. MaV203 expresses a basal level of HIS3 that was suppressed with increasing concentrations of 3AT to allow discrimination of HIS3 activation.
Expression of FLNSP4, NSP4 46-175 and a panel of NSP4 mutants The entry vectors encoding FLNSP4, FLNSP4-Ala6, FLNSP4-HydroMut, and NSP4 46-175 were used to shuttle the NSP4 sequences into the inducible yeast expression plasmid, pYES-DEST52 (Invitrogen) and transformed into the inducible yeast strain, InVSc-1 [25]. Since our previous data demonstrate stronger binding of NSP4 to cav-1 in Y2H assays when the first two NSP4 N-terminal hydrophobic domains are deleted, we continued our experiments with the truncated NSP4 [25] and constructed NSP4 Ala6 and NSP4 HydroMut for use in the Y2H experiments. All additional mutants (Rev1, Rev2, Rev3, Rev2 116,124,127A, Rev2 116A, Rev1,2 127A, and Rev1 124,127A) were constructed using site directed mutagenesis of NSP4  , introduced into the Gateway® entry vector, and cloned into the inducible yeast expression plasmid, pYES-DEST52, as above ( Table 1). The transformed yeast colonies were grown on CSM Uraand the encoded NSP4 proteins were induced with YPAG medium yeast extract, peptone (Difco), 0.01% adenine sulfate and 2% galactose (Sigma). Cells were grown in YPAD at 30°C for 24 h, washed and re-suspended to an OD 600 of 0.5 in YPAG and incubated at 30°C for 24 h. Yeast lysates were prepared using the Zymo Yeast Protein Extraction kit (Zymo Research, Orange, CA) as previously described and were utilized in the binding assays [25]. Approximately 1 × 10 6 cells were pelleted, Y-Lysis buffer and zymolase were added to the samples and incubated at 37°C for 1 h. The yeast were centrifuged at 400×g for 5 min, resuspended in PBS (pH 7.2), containing protease inhibitors 100 μM AEBSF, 80 μM aprotinin, 5 μM bestatin, 1.5 μM E-64, 2 μM leupeptin, 1 μM pepstatin A and 100 μM PMSF (Calbiochem-Novabiochem Corp., San Diego, CA)] and quantified by BCA (Pierce). These lysates were used in the in vitro peptide binding and Western blot assays as described below.  [32]. Following synthesis, the peptides were cleaved from the solid resin support and all side-chain protecting groups removed by addition of Reagent R (90% trifluoroacetic acid (TFA), 5% thioanisole, 3% ethanedithiol and 2% anisole). Peptides were separated from the solid support by filtration into cold diethyl ether in a dry ice/ethanol bath. Following three organic precipitations, peptides were purified from organic contaminants and incomplete peptide fragments by gravimetric gel filtration chromatography (Sephadex G25 medium) and reverse-phase HPLC using a reverse phase C4 Delta Pak (Waters Chromatography Division, Milford, MA) or C18 (Beckman-Coulter, Fullerton, CA) column. Peptide characterization was by matrix-assisted laser desorption/ionization (MALDI) mass spectrometry (Laboratory for Biological Mass Spectrometry, Department of Chemistry, Texas A&M University, College Station, TX). Mass chromatograms revealed if the full-length peptide was present, the mass of the peptide and the extent of contaminants in the eluted fraction.

Peptide synthesis, purification and characterization
All purified peptide samples were tested for endotoxin by the Limulus Amebocyte Lysate (LAL) test (Associates of Cape Cod, Inc.) according to the manufacturer's protocol, to ensure that the samples were endotoxin free. A value of 0.5 EU/mL or less was considered acceptable for use in the diarrhea studies.

Diarrhea induction in mouse pups
To examine the biological relevance (enterotoxic activity) of the mutations made in the hydrophobic face of the amphipathic alpha-helix/enterotoxic peptide/cav-1 binding domain of NSP4, the NSP4 HydroMut112-140, NSP4 AlaAcidic112-140 , NSP4 AlaBasic112-140 and the wt-NSP4 112-140 peptides were tested for diarrhea induction as previously described (5, 54) (Additional file 1: Table S1). Briefly, each of the NSP4-specific peptides (100 nmol in 50 μl total volume) was administered by intraperitoneal (IP) delivery to 6-10 day old mouse pups (5-6 pups per peptide). Mouse pups injected with endotoxin-free PBS served as a negative control. Mouse pups were monitored for diarrhea every 2 hours for 12 hours and then at 24 hours post injection. The severity of the diarrhea was scored on a scale of 1 to 4 as previously described (5).

Western blot assays
The co-transformed yeast colonies were grown in liquid CSM Leu -Trp -, and yeast protein extracts were prepared using the Zymo Yeast Protein Extraction kit (Zymo Research, CA). The cell pellets were resuspended in PBS (pH 7.2) containing protease inhibitors and quantified by BCA (Pierce). All lysates were separated by 12% SDS-PAGE, electroblotted onto nitrocellulose membranes, probed with NSP4 or cav-1 peptide-specific antibodies, and reactive bands visualized by the addition of HRP-conjugated IgG and Super Signal® West Pico chemiluminescent substrate (Pierce) followed by exposure to Kodak X-OMAT film [51,52].

Discussion
This study demonstrated that the hydrophobic face of the AAH of the enterotoxic domain of NSP4 is critical for the binding to cav-1 and structure may play a role. An NSP4 construct that disrupted the charged face of the amphipathic helix (NSP4-Ala6) continued to bind cav-1, but alteration of three residues (I113R, V124K, Y131D) in the hydrophobic face disrupted cav-1 binding (NSP4-HydroMut). When any one of the three mutated hydrophobic amino acids was restored, NSP4 binding to cav-1 was restored as observed with the Y2H and peptide binding assays. We had not anticipated that all three hydrophobic amino acids selected for mutation in NSP4-HydroMut would individually restore the binding to cav-1, but anticipated reversion of more than one of the three mutated residues would be required to reconstitute the binding site. All three revertants included L116 and L127 suggesting these hydrophobic residues were critical to cav-1 binding. Further analysis of L116 and L127, using the revertant mutants that substituted L116A and/or L127A identified L116 as a key residue for NSP4 binding to cav-1. The ability of L116 to bind cav-1 changes the reactivity from binding in native NSP4 to not binding in NSP4 HydroMut , to binding in the Rev 2 and finally to not binding in Rev2 with mutated L116A ( Figure 1C).
Yet, NSP4 HydroMut contained L116 and failed to bind cav-1, suggesting more than one amino acid is required to accomplish binding. Protein-protein interactions can rely on the recognition of linear motifs yet, binding sites frequently depend on (a) exposure of the binding motif at the surface of the protein; (b) the environment of the binding residues; and (c) may require conformation-dependent epitopes for binding [54]. Given that the binding site is within the hydrophobic face, typical protein folding patterns would not place these residues on the protein surface, but would protect them from the aqueous environment of the cell. Perhaps the NSP4-cav-1 binding site was dependent on the secondary structure of a region that included L116 that was disrupted in the NSP4 HydroMut . The CD analysis confirmed that the NSP4 HydroMut peptide has a different secondary structure than the NSP4 wild type peptide.
Notably the identified NSP4 residue that binds cav-1 overlaps the enterotoxic peptide. To begin to understand the functional connection between NSP4 enterotoxic activity and cav-1 binding, diarrhea induction in neonatal mice was evaluated with peptides corresponding to the wt-NSP4 112-140 , NSP4 AlaAcidic112-140 , NSP4 HydroMut112-140 and NSP4 AlaBasic112-140 . The NSP4 AlaBasic112-140 and NSP4 HydroMut112-140 peptides failed to induce diarrhea, whereas when three acidic residues were altered to alanines, NSP4 AlaAcidic112-140, diarrhea was induced, similar to the wild type peptide. As the CD analysis of NSP4 AlaAcidic112-140 and NSP4 AlaBasic112-140 indicated disruption of the AAH, the helix appeared to tolerate changes in the charged face. Computer modeling of the mutants using PyMol showed constraints for structure especially with I113R (data not shown). These data support the model that the integrity of the secondary structure of the hydrophobic face of the AAH of NSP4 may be important for the induction of diarrhea.
Despite our diarrhea data, it is unclear if cav-1 binding in itself plays a role in diarrhea induction or if cav-1 directs NSP4 to caveolae for signal induction. A previous study highlighted the importance of the tyrosine residue in the hydrophobic face of the AAH. A synthetic peptide in which the tyrosine residue was mutated to a lysine lacked diarrhea-inducing function [55]. In this study, when tyrosine was changed to aspartic acid (D), NSP4 failed to induce diarrhea. Although Y131 is important for the induction of diarrhea in some but not all strains of RV [55], Y131 appears to be dispensable for cav-1 binding. Both Rev1 and Rev2 bind cav-1 in the absence of Y131, and the lack of Y131 does not influence cav-1 binding of Rev1,2 127A or Rev1 124,127A. Our data herein supports and further suggests that the structure of the AAH of NSP4 may play a significant role in promoting diarrhea.
A recent study revealed that the α1β1 integrins are receptors for NSP4 [12]. Binding to these receptors initiates cellular signaling processes (in particular, activation of phospholipase C) that mobilize calcium and result in diarrhea induction. While NSP4 binds the integrin receptor at residues 114-135, signaling occurs via an interaction between NSP4 131-140 and a separate integrin domain [12]. NSP4 HydroMut 112-140 contains a mutation at aa131, which is within the integrin signaling domain, and does not cause diarrhea. Lack of enterotoxic function with this mutant peptide may be a result of its inability to initiate integrin mediated signaling. These results agree with previous data demonstrating that the exogenous addition of NSP4 mobilizes [Ca 2++ ] I from ER stores through a phospholipase C-inositol 1,4,5-triphosphate (PLC-IP 3 ) pathway [6].
The data presented suggested that the integrity of the hydrophobic face of NSP4 was critical to both cav-1 binding and modulating diarrhea induction. Localization of the binding site to the conserved hydrophobic residues between RV strains and the lack of diarrhea induced by the synthesized peptide with the corresponding mutations in the hydrophobic face of NSP4 indicated the importance of the cav-1 binding site of NSP4. However, the lack of diarrhea observed with the peptide NSP4 AlaBasic112-140 indicated that the NSP4 enterotoxin might extend into the basic face. Additional studies are needed to dissect the complexity of NSP4's multiple activities and binding domains, and how these domains are coordinated during a viral infection.

Ethics
Research was conducted in compliance with the Animal Welfare Act and other federal statutes. The authors adhered to the principles stated in the Guide for the Care and Use of Laboratory Animals, National Research Council. Animals were housed in a fully accredited facility.