The Small Hydrophobic Protein of the Human Respiratory Syncytial Virus Forms Pentameric Ion Channels*

Background: Few effective treatments exist for human respiratory syncytial virus infection. The absence of small hydrophobic (SH) protein in RSV leads to viral attenuation. Results: SH protein forms pentamers and shows pH-dependent ion channel activity. Conclusion: SH protein forms pentameric ion channels. Significance: The SH protein and its channel activity constitute a potential drug target. The small hydrophobic (SH) protein is encoded by the human respiratory syncytial virus. Its absence leads to viral attenuation in the context of whole organisms, and it prevents apoptosis in infected cells. Herein, we have examined the structure of SH protein in detergent micelles and in lipid bilayers, by solution NMR and attenuated total reflection-Fourier transform infrared spectroscopy, respectively. We found that SH protein has a single α-helical transmembrane domain and forms homopentamers in several detergents. In detergent micelles, the transmembrane domain is flanked N-terminally by an α-helix that forms a ring around the lumen of the pore and C-terminally by an extended β-turn. SH protein was found in the plasma membrane of transiently expressing HEK 293 cells, which showed pH-dependent (acid-activated) channel activity. Channel activity was abolished in mutants lacking both native His residues, His22 and His51, but not when either His was present. Herein, we propose that the pentameric model of SH protein presented is a physiologically relevant conformation, albeit probably not the only one, in which SH contributes to RSV infection and replication. Viroporins are short (∼100 amino acids) viral membrane proteins that form oligomers of a defined size, act as proton or ion channels, and in general enhance membrane permeability in the host. However, with some exceptions, their precise biological role of their channel activity is not understood. In general, viroporins resemble poorly specialized proteins but are nevertheless critical for viral fitness. In vivo, viruses lacking viroporins usually exhibit an attenuated or weakened phenotype, altered tropism, and diminished pathological effects. We have chosen to study the SH protein, 64 amino acids long, found in the human respiratory syncytial virus because of the effect of RSV on human health and the lack of adequate antivirals. We show that SH protein forms oligomers that behave as ion channels when activated at low pH. This study adds SH protein to a growing group of viroporins that have been structurally characterized. Although the precise biological role of this pentameric channel is still unknown, this report is nevertheless essential to fill some of the many gaps that exist in the understanding of SH protein function.

The small hydrophobic (SH) protein is encoded by the human respiratory syncytial virus. Its absence leads to viral attenuation in the context of whole organisms, and it prevents apoptosis in infected cells. Herein, we have examined the structure of SH protein in detergent micelles and in lipid bilayers, by solution NMR and attenuated total reflection-Fourier transform infrared spectroscopy, respectively. We found that SH protein has a single ␣-helical transmembrane domain and forms homopentamers in several detergents. In detergent micelles, the transmembrane domain is flanked N-terminally by an ␣-helix that forms a ring around the lumen of the pore and C-terminally by an extended ␤-turn. SH protein was found in the plasma membrane of transiently expressing HEK 293 cells, which showed pH-dependent (acid-activated) channel activity. Channel activity was abolished in mutants lacking both native His residues, His 22 and His 51 , but not when either His was present. Herein, we propose that the pentameric model of SH protein presented is a physiologically relevant conformation, albeit probably not the only one, in which SH contributes to RSV infection and replication. Viroporins are short (ϳ100 amino acids) viral membrane proteins that form oligomers of a defined size, act as proton or ion channels, and in general enhance membrane permeability in the host. However, with some exceptions, their precise biological role of their channel activity is not understood. In general, viroporins resemble poorly specialized proteins but are nevertheless critical for viral fitness. In vivo, viruses lacking viroporins usually exhibit an attenuated or weakened phenotype, altered tropism, and diminished pathological effects. We have chosen to study the SH protein, 64 amino acids long, found in the human respiratory syncytial virus because of the effect of RSV on human health and the lack of adequate antivirals. We show that SH protein forms oligomers that behave as ion channels when activated at low pH. This study adds SH protein to a growing group of viroporins that have been structurally characterized. Although the precise biological role of this pentameric channel is still unknown, this report is nevertheless essential to fill some of the many gaps that exist in the understanding of SH protein function.
The SH 4 protein is found in human respiratory syncytial virus (hRSV), an enveloped pneumovirus in the paramyxoviridae family that causes lower respiratory tract disease in infants, the elderly, and immunocompromised populations worldwide (1). Up to 64 million reported cases of hRSV infection and 160,000 deaths occur each year. Although the virus was identified almost half a century ago (2), there are still no vaccines or effective antiviral drugs available. hRSV can cause repeated reinfections throughout life, and little is known about its molecular mechanism of pathogenesis.
The hRSV genome composes a nonsegmented negativestranded RNA of ϳ15 kb that transcribes 11 proteins, including the three membrane proteins F, G, and SH. Proteins F and G are key factors during virus entry, attachment, and fusion (3,4). In contrast, the role of SH protein is less clear. RSV that lacks SH (RSV⌬SH) is still viable and still forms syncytia (5)(6)(7). However, RSV⌬SH was attenuated in in vivo mouse and chimpanzee models (6,9), which indicates that SH protein is important for RSV pathogenesis.
The SH protein is a 64-(RSV subgroup A) or 65 (RSV subgroup B)-amino acid-long type II integral membrane protein, with the C terminus oriented extracellularly/lumenally in the host and with a single predicted ␣-helical transmembrane (TM) domain (10). The sequence of SH protein is highly conserved, especially at the TM domain (11,12). In infected cells, most SH protein accumulates at the membranes of the Golgi complex, but it is also found in the endoplasmic reticulum and plasma membranes (13), which enables patch clamp studies. During infection, SH protein has been shown to exist in several forms, e.g. full-length truncated protein (4.5 kDa), and post-translationally modified by glycosylation and phosphorylation (14,15), although the full-length unmodified form is the major species (10).
SH protein permeabilizes membranes (16,17). Its synthetic TM domain (residues 18 -43) shows channel activity in synthetic bilayers (18) and forms homopentamers in perfluorooctanoic acid (PFO) gel electrophoresis. In that work, using the technique of linear infrared dichroism (19,20) and conformational searching methods (21,22), we provided a low resolution model of the TM homopentamer responsible for channel activity. More recently, an electron microscopy study of an 86-amino acid-long polypeptide containing the 64 amino acids of SH protein visualized in a membrane-mimetic environment (17) identified ring-like features interpreted as having either 5or 6-fold symmetry.
The use of full-length, tag-free SH protein is essential to resolve possible ambiguities in structure and function. In our hands, synthetic SH protein could not be purified satisfactorily, and native ligation of synthetic peptides produced very low yield. Similarly, SH protein alone or with a His tag could not be expressed in Escherichia coli, presumably due to toxicity. Thus, in this work, full-length SH protein was obtained using maltose-binding protein (MBP) as an N-terminal fusion protein, which in turn has an N-terminal His tag for easier purification. The construct His 6 -MBP-SH improved expression significantly. Between MBP and SH protein, a tobacco etch virus (TEV) cleavage site was introduced. Once SH protein was cleaved from the construct, it contained only three extra residues (SNA) at the N terminus, which do not interfere with the biophysical properties of the protein. SH protein was characterized by ATR-FTIR when reconstituted into model lipid bilayers and in parallel by solution NMR after isotopic labeling, using detergent micelles. The latter technique allowed reconstruction of the pentameric ␣-helical bundle that includes the extramembrane regions. We have also investigated the channel activity of the full-length SH protein and some of its mutants, especially at the protonatable His residues, using the whole-cell patch clamp technique.

MATERIALS AND METHODS
Cloning of SH Gene-The nucleotide sequence of SH protein was obtained from NCBI (NC_001803.1), strain S2 ts1C. The SH gene, accession number NP_044594.1 (23), was synthesized, and the purity was determined by agarose gel electrophoresis. The SH gene was cloned into pTBMalE vector (24) with MBP as fusion partner carrying a His tag at the N terminus to produce the construct His-MBP-SH. Mutants were constructed in the same way.
Expression of the His-MBP-SH Construct-The DNA plasmid containing the SH gene was transformed into an E. coli competent cell strain BL-21 (DE3) CodonPlus-RIL strain (Stratagene) for protein overexpression. Cells from a single colony were picked to inoculate 10 ml of LB media with 100 g/ml ampicillin and 34 g/ml chloramphenicol and grown overnight at 37°C with shaking. 8 ml of the overnight culture was then transferred to 800 ml of LB media and grown at 37°C with shaking to an A 600 of 0.6 -0.7. For unlabeled samples, cells were induced with 0.4 mM isopropyl ␤-thiogalactoside and grown at 23°C overnight with shaking.
For isotopically labeled samples, cells were harvested when A 600 0.6 -0.7 was reached by centrifugation and washed with M9 minimal media once. The cells were transferred to M9 minimal media containing [ 15 N]ammonium chloride or [ 13 C]glucose. Expression levels in minimal media were enhanced by increasing cell densities through a 4:1 concentrating method (25). The concentrated culture in M9 media was induced with 0.4 mM isopropyl ␤-thiogalactoside and grown overnight at 23°C with shaking.
After induction, cells were harvested and resuspended in Ni 2ϩ -NTA binding buffer containing 20 mM Tris-HCl, 500 mM NaCl, and 5 mM imidazole, pH 8.0, and then kept frozen at Ϫ20°C overnight. Thawed cells were incubated with 0.2 mg/ml lysozyme and 0.02 mg/ml benzonase for 10 min. Then, octyl-␤-D-glucopyranoside was added to the sample to a final concentration of 50 mM. The cells were lysed with a microfluidizer at 15,000 p.s.i. pressure, and the supernatant was collected after centrifugation at 20,000 ϫ g for 30 min and loaded to Econocolumn (Bio-Rad) packed with Ni 2ϩ -NTA-agarose resin (Qiagen) pre-equilibrated with binding buffer. The proteins were allowed to bind to the nickel resin with gentle shaking at 4°C overnight.
Purification of His-MBP-SH Construct by Ni 2ϩ -NTA Chromatography-The nickel resin with bound protein was washed with 20 column volumes of buffer containing 20 mM Tris-HCl, 500 mM NaCl, and 20 mM imidazole, pH 8.0. Bound proteins were eluted with elution buffer containing 20 mM Tris-HCl, 500 mM NaCl, 500 mM imidazole, pH 8.0, and 30 mM octyl-␤-Dglucopyranoside. All fractions collected were stored at 4°C, whereas fractions containing the His-MBP-SH fusion construct were analyzed by SDS-PAGE.
Expression and Purification of TEV Protease-Expression and purification of TEV protease were done in a similar way for SH protein except that no octyl-␤-D-glucopyranoside was added to the sample, as it is a soluble protein. TEV was stored at Ϫ20°C in buffer containing 50% glycerol, 10 mM Tris-HCl, 250 mM NaCl, 500 mM imidazole, 1 mM EDTA, and 5 mM DTT, pH 8.0.
Cleaving of His-MBP-SH Construct by TEV Protease-TEV protease was added to the His-MBP-SH construct at a mass ratio of 1:5 (TEV/fusion protein, mg/ml). The digestion was performed at room temperature with gentle shaking, and the progress of the reaction was monitored by SDS-PAGE. Usually, the reaction was completed in 4 -6 h. Digestion was stopped by addition of trichloroacetic acid (TCA) to a final concentration of 6% (v/v)/ and the precipitate was collected by centrifugation at 18,000 ϫ g for 30 min. The pellet was washed twice with water followed by lyophilization.
Purification of SH Protein by Organic Solvent Extraction and RP-HPLC-The cleaved SH protein was separated from the His-MBP fragment using a two-step purification strategy. First, SH protein was extracted by methanol by adding 10 ml of methanol per 1 liter of culture and mixing gently for 2 h at room temperature. The supernatant was collected by centrifugation at 18,000 ϫ g for 30 min and checked by MALDI spectroscopy and SDS-PAGE. SH protein was further purified by injecting the supernatant onto a Zorbax C3-300 Å column connected to an HPLC system. Purified SH protein was eluted with a linear gradient of solvent A (water/TFA, 99.9:0.1, v/v) and solvent D (isopropyl alcohol/acetonitrile/TFA, 80:19.9:0.1, v/v). Pooled fractions were lyophilized, and the purity of the samples was checked by MALDI.
Gel Electrophoresis-Reduced SDS-PAGE or PFO-PAGE using precast Tricine gradient gel (Bio-Rad), with or without heat treatment, was used to test the purity of the preparations. The gel used in Fig. 1C was an in-house 12.5% gel that was run using Tris-glycine buffer, ϳpH 9, a typical Laemmli system. The gel used in Fig. 2A was a commercially available pre-cast gel from Invitrogen, and the NuPAGE SDS-PAGE was prepared with BisTris buffer at near neutral pH and run using MES running buffer, ϳpH 7.0, to minimize protein modification at extreme pH. SDS or PFO sample buffer, for SDS or PFO-PAGE, respectively, was added to the lyophilized protein to a final protein concentration of 2 g/l. After mixing for 1 min followed by heating at 95°C for 5 min, the sample was loaded onto the gel. Electrophoreses were run at constant voltage of 80 V for 3 h at room temperature, and gels were stained with Coomassie Blue dye.
Analytical Ultracentrifugation-AUC ES experiments were performed in a Beckman XL-1 analytical ultracentrifuge (Beckman Coulter) using a six-channel carbon-epoxy composite centerpiece equipped with quartz windows for the absorbance experiments or sapphire windows for interference experiments. All experiments were performed at 25°C on samples at three different initial SH protein concentrations (0.3, 0.5, and 0.8 absorbance at 280 nm) solubilized in 5 mM C14 betaine (Sigma) and three different rotor speeds (16,000, 19,500, and 24,000 rpm). The software program Winmatch was used to monitor whether the equilibrium had been reached at each speed, every 3-4 h. Usually, the samples achieved equilibrium in about 20 h. For DPC or C8E5, the detergent concentration was 15 and 33 mM, respectively. SH protein in methanol was dried under a stream of nitrogen gas and lyophilized for at least 3 h to remove any residual organic solvent. 5 mM C14 betaine in buffer containing 50 mM Tris-HCl, 100 mM NaCl, 29.4% D 2 O, pH 7.3, was added to the dried protein samples and mixed by 5 min of vortexing. The sample was centrifuged at 15,000 rpm for 5 min before loading into the centerpiece. 110 l of sample and 120 l of reference buffer were loaded into the six-channel centerpiece. Data obtained by UV absorption at 280 nm were analyzed with Sedphat by nonlinear least squares curve-fitting of radial concentration profiles using the Levenberg-Marquardt algorithm. Different reversible association models were tried to fit the data to obtain the most suitable model. The monomeric mass, extinction coefficient at 280 nm, and partial specific volumes for each sample were calculated using the program SEDNTERP.
Infrared Spectroscopy-FTIR spectra were recorded on a Nicolet Nexus 560 spectrometer (Madison, WI) purged with N 2 and equipped with an MCT/A detector cooled with liquid nitrogen. Attenuated total reflection (ATR) spectra were measured with a 25-reflection ATR accessory from Graseby Specac (Kent, UK) and a wire grid polarizer (0.25 micrometers, Graseby Specac). Approximately 100 l of sample in water with 50:1 lipid/peptide molar ratio were applied onto a trapezoidal (50 ϫ 2 ϫ 20 mm) germanium internal reflection element. The lipids used here were DMPC and 1-palmitoyl-2-oleoyl-snglycero-3-phosphocholine (Avanti Polar Lipids). A dry or D 2 Osaturated N 2 stream flowing through the ATR compartment was used to remove bulk water or to achieve D 2 O exchange, respectively. After insertion of the plate in the ATR cell, spectra were collected. A total of 200 interferograms collected at a resolution of 4 cm Ϫ1 were averaged for every sample and processed with one-point zero filling and Happ-Genzel apodization. The area of amide I (CϭO stretching) was obtained by peak integration from 1600 to 1700 cm Ϫ1 , and the area of amide II (N-H bending, centered at ϳ1550 cm Ϫ1 ) was obtained by peak integration from 1510 to 1580 cm Ϫ1 . No difference in band area was observed employing other means of peak size estimation such as peak fitting and Fourier self-deconvolution. For the hydrogen-deuterium exchange experiment, the kinetics of exchange was calculated by measuring the relative area of amide II relative to the amide I band (26).
Electrophysiological Recordings and Data Analysis-HEK 293 cells were transfected with a plasmid containing the SH protein gene. Expression of wild type and mutants of the fulllength SH gene was confirmed using green fluorescent protein (GFP) (27). The full-length SH gene was cloned into the pIRES-AcGFP1 (Aequorea coerulescens green fluorescent protein) vector (Clontech), containing a green fluorescent protein gene, by using restriction enzymes BgIII and PstI. The internal ribosome entry site of the encephalomyocarditis virus permits the translation of two open reading frames from one messenger RNA. Thus, two proteins, in this case SH protein and AcGFP1, are expressed simultaneously from the same bicistronic mRNA transcript.
The cDNA was transiently transfected into HEK 293 cells using the standard calcium phosphate method (28). The vector alone, pIRES-AcGFP1, was also transiently transfected in separate experiments as a control. Whole-cell current was recorded at room temperature using the standard patch clamp technique 48 -72 h after transfection. The bath contained the following (mM): 124.0 NaCl, 3.5 KCl, 0.01 NaH 2 PO 4 , 26.2 NaHCO 3 , 1.3 MgSO 4 , 2.5 CaCl 2 , and 10.0 D(ϩ)-glucose; gassed with a mixture of 95% O 2 and 5% CO 2 , pH 7.4, and an osmolarity of 300 mosM/kg. The internal solution (pipette solution) contained the following (mM): 135.0 potassium gluconate, 10.0 KCl, 10.0 HEPES buffer, 0.5 EGTA, 2.0 Mg-ATP (pH adjusted to 7.3 with KOH, and an osmolarity of 275-285 mosM /kg). The voltages were uncorrected for a Ϫ9 mV junction potential, and actual voltage was obtained by subtracting 9 mV from the reported values. Whole-cell currents, obtained under voltage clamp with an Axonpatch Multiclamp 700B amplifier (Axon Instruments), were filtered at 1-5 kHz and sampled at 5-50 kHz. The access resistance R a (usually less than 20 MegaOhms) and the capacitive transients were not compensated.
Human embryonic kidney 293 (HEK 293) cells were cultured in Dulbecco's modified Eagle's medium (DMEM, Invitrogen), supplemented with 10% FBS and 1% penicillin and streptomycin, and maintained in 5% CO 2 incubator at 37°C. For transfection, cells were seeded on the 10-mm Petri dishes containing poly-D-lysine-coated coverslips and grown overnight. A total of 4 g of pIRES-AcGFP plasmids, with either SH wt or SH FF FLAG-tagged, were transfected into HEK 293 cells using Lipofectamine 2000 kit (Invitrogen) following the protocol of the manufacturer. Transfected HEK 293 cells were grown for 48 h, and then they were fixed in PBS containing 4% paraformaldehyde for 45 min at 4°C. The fixed cells were washed twice with PBS before blocking by 0.3% BSA/PBS with or without 0.1% Triton X-100 for 30 min. The HEK 293 cells were incubated with Alexa Fluor 647-conjugated anti-FLAG antibody (1:100 dilutions, Cell Signaling Technology) overnight at 4°C. After washing three times with PBS ϩ 0.3% BSA and one time with PBS, the immunolabeled cells were visualized using a laser scanning confocal microscope (Fluoview BX61; Olympus).
NMR Sample Preparation-15 N-and 15 N/ 13 C-labeled SH protein was tested in various detergents (SDS, DPC, and DHPC) for best dispersion in an HSQC spectrum. The best results were obtained in d 38 DPC. For paramagnetic probe measurements, 0.8 mg of dry 5-doxyl stearic acid (5-DSA) or 16-doxyl stearic acid (16-DSA) was first dissolved separately in 300 l of methanol. 30 l of methanol solubilized DSA was then aliquoted into an Eppendorf tube, corresponding to 1 mM in the final NMR samples. The DSA probes were then dried under a stream of dry N 2 gas and placed in a vacuum lyophilizer for at least 2 h to remove any residual methanol. The NMR sample containing SH/DPC was added to the NMR tube containing dried DSA and vortexed until the film dissolved completely. The sample was left to equilibrate for at least 2 h before NMR measurements was taken.
For Cu 2ϩ experiments, copper(II) chloride was made up as a 1 mM stock, of which 50 M Cu 2ϩ (final concentration) was added to the sample. Weakly aligned samples were prepared by soaking a solution of 0.5 mM 15 N-labeled SH protein in 100 mM DPC into polyacrylamide gels. Gels were prepared from stock containing 36% w/v acrylamide (Bio-Rad) and 0.94 w/v N,Nmethylenebisacrylamide (Bio-Rad), which yields an acrylamide/bisacrylamide molar ratio of 83:1. An 8% polyacrylamide gel was cast using a gel chamber of 5.4 mm inner diameter (New Era Enterprise). After polymerization was complete, gels were washed in large excess of H 2 O overnight to ensure removal of unreacted components. The gels were then dried to completeness at 37°C. The protein solution containing SH/DPC was soaked into the dried gels over 2 days to ensure complete re-hydration. The hydrated 8% gel was then radially compressed into a 4.2-mm inner diameter open-ended tube using the gel press assembly of Chou et al. (73) (New Era Enterprise, Inc.) and secured using the supplied support rod and end gel plug.
Approximately 1 mg of lyophilized SH protein was solubilized in 100 l of methanol and dried under a dry stream of N 2 gas, resulting in a thin protein film deposit. The tube was placed in a vacuum lyophilizer for at least 2 h to remove any residual methanol. The thin protein film was then solubilized with sample buffer containing 5 mg of d 38 DPC (Avanti Polar Lipids) and 100 mM acetate buffer to a give final detergent concentration of 150 mM and sample, pH 4.4, with the protein/detergent ratio of about 1:200. The sample pH was brought up to 4.4 by titrating with aqueous sodium hydroxide. The sample was subjected to repeated cycles of heating to 318 K, vortexing, cooling, and sonication until the solution was cleared indicating protein reconstitution into detergent micelles. The sample was spun at 13,000 rpm to clear nonreconstituted protein. For copper addition in NMR experiments, the SH protein was reconstituted in 100 mM acetate buffer at pH 4.5 and 160 mM DPC, to a final protein/detergent molar ratio of 1:200.
NMR Spectroscopy-NMR experiments were performed at 37°C (310 K) using Bruker Avance-II 700 and 600 NMR spectrometers equipped with cryogenic TXI probes (Bruker Bio-Spin). Complete sequence-specific assignment of backbone was achieved using standard triple resonance experiments (HNCA, HN(CO)CA, CBCACONH, HNCACB, and HNCO), whereas side-chain resonances were assigned using three-dimensional 15  To identify transmembrane residues, the NMR sample was lyophilized overnight and reconstituted in 99% D 2 O. Sodium 2,2-dimethyl-2-silapentane-5-sulfonate was used as the internal reference for 1 H nuclei. The chemical shifts of 13 C and 15 N nuclei were calculated from the 1 H chemical shifts (29). Analysis of the RDC data and rigid body modeling was done using the program MODULE (30). To determine protein backbone flexibility, 1 H-15 N heteronuclear steady-state NOEs were obtained by recording spectra with and without 3 s 1 H presaturation at 600 MHz.
Structure Reconstruction-The structure of the SH monomer in DPC micelles was calculated using the assignment of 1 H, 13 C, and 15 N resonances and unassigned NOEs as input for the program CYANA (31,32). Backbone dihedral angles were predicted using chemical shift data and PREDITOR (33). Structure calculations were started from 100 random conformers, using the standard simulated annealing protocol in CYANA. The statistics of meaningful NOE distance constraints in the final CYANA cycle showed a high density of structural constraints per amino acid. Seven cycles of NOE assignment and structure reconstruction resulted in a bundle of 20 conformers. All data sets used for CYANA were reformatted for CNS using PDBstat (34).
The program CNS 1.2 (35) was used to refine the SH structure by incorporating RDCs, dihedral angles, hydrogen bonds, and upper limit distance constraints (from CYANA). Structures were recalculated from an extended strand with random initial velocities using the default simulated annealing protocol of the CNS package. 100 conformers were calculated, and the structure with the lowest energy function after fitting the topology based on rigid body modeling was selected as the representative model for construction of the C 5 -symmetrical SH pentamers (supplemental Fig. S7).
The pentameric structure of SH was determined using the multibody interface of the HADDOCK webserver (36). The docking was performed from five copies of the representative model (from CNS with the lowest energy function) and inclusion of RDCs for all five monomers using the SANI module for RDC analysis (37). To prevent possible occlusion and stearic hindrance during the docking process, the flexible parts of the protein showing negative HNOE values were removed from the monomer. Center-of-mass restraints, C 5 symmetry restraints, and noncrystallographic restraints were imposed on all sequential monomer pairs (AB-BC-CD-DE-EA).
The pentameric model of SH-TM in lipid bilayers obtained previously using site-specific infrared dichroism was used to derive a set of plausible intermonomeric NOEs. These were verified against experimental side-chain NOEs from standalone peaks in 13 C HSQC and 13 C-resolved NOESY-HSQC spectra resulting in a list of NOEs incompatible with ␣-helical backbone geometry but exhibiting clearly detectable and unique NOEs. These inter-monomeric NOEs were implemented as the enforced restraints for the HADDOCK procedure to obtain the full atomic description of the pentameric structure.
The lumen dimensions for the pentameric model were calculated using HOLE (38) and visualized using Visual Molecular Dynamics (39). Chemical shift perturbations were monitored using 1 H-15 N TROSY spectra where the magnitude of the perturbation was calculated using the weighted-average chemical shifts for each amino acid residue as shown in Equation 1 (40).

Overexpression and Purification of SH Protein from E. coli-
We expressed SH protein with its N terminus fused to His 6 -MBP (41), reaching moderate levels of expression after isopropyl ␤-thiogalactoside induction, shown by a band migrating at 50 kDa (Fig. 1A, arrow). His 6 -MBP-SH fusion protein was mainly found in the membrane fraction (Fig. 1B, see arrowhead in lane labeled MEM), which suggests that SH protein in that instance was properly folded and likely in a native conforma-tion. Only a small percentage of the expressed fusion protein appeared in inclusion bodies (Fig. 1B, lane labeled INC). Similar observations regarding the membrane localization of MBPfused proteins have been reported in cytochrome b 6 , M2, CorA, and KdpF (41).
The His 6 -MBP-SH fusion protein construct was purified using Ni 2ϩ -NTA resin and subjected to tobacco etch virus (TEV) enzymatic cleavage to remove His 6 -MBP. Complete cleavage (Fig. 1C) was achieved at room temperature in 4 h, after mixing TEV and His 6 -MBP-SH at a 1:5 molar ratio under constant shaking. SH protein was extracted from the mixture with methanol, followed by reverse-phase high pressure liquid chromatography (RP-HPLC) purification. The HPLC chromatogram shows an intense peak (Fig. 1D, arrow). The MALDI mass spectrum of that fraction (Fig. 1E) shows a major peak corresponding to pure SH protein, with only three extra N-terminal amino acids (see sequence in Fig. 1F). This purified polypeptide showed channel activity when reconstituted in black lipid membranes (supplemental Fig. S1).
Gel Electrophoresis Analysis of SH Protein-It has been shown that SH protein can be cross-linked in infected cells to produce a ladder of oligomers, from dimers to pentamers, in an SDS gel (10,15). We have shown previously that the synthetic peptide corresponding to the predicted TM domain of SH protein (residues 18 -43) formed monomers in SDS but pentamers in the mild detergent PFO (18). The latter results showed that SDS is not a good detergent to monitor SH protein oligomerization and that milder detergents are required to retain the native interactions between monomers. This is confirmed when examining the full-length SH protein (ϳ7.5 kDa) in the presence of SDS ( Fig. 2A, WT), which migrates as a single band with an apparent molecular mass of ϳ17 kDa, consistent with dimers. However, in the presence of PFO (Fig. 2B, WT), mobility was consistent with a higher molecular weight (ϳ35-40 kDa), compatible with a pentameric form, and in agreement with the results we reported for SH-TM (18). We therefore conclude that the dimeric form observed for full-length SH protein in SDS is artifactual.
To confirm the pentameric oligomerization of SH protein, we performed a Blue Native (BN)-PAGE analysis with SH protein previously solubilized in zwitterionic 3-(N,N-dimethylmyristylammonio)propanesulfonate (C14 betaine) micelles (Fig. 2C, WT). Because water-soluble molecular weight markers may show abnormal mobility compared with membrane proteins in BN-PAGE (42), we included as an additional molecular weight marker a membrane protein, E. coli aquaporin Z (monomeric size 27 kDa), which forms a ladder with oligomers of increasing size in these conditions (Fig. 2C, lane AQP) (43). WT SH protein migrated as a single band with a mobility between monomeric and dimeric aquaporin Z (i.e. between 30 and 60 kDa), again consistent with a pentamer (ϳ40 kDa). These results show that WT SH protein appears to form a single oligomer, probably pentameric, in both PFO and in C14 betaine detergents.
Analytical Ultracentrifugation Equilibrium Sedimentation (AUC ES) of WT SH Protein in Detergent Micelles-An independent determination of WT SH protein oligomeric size was performed using AUC ES. Reversible association models corre-sponding to an equilibrium between monomers and n-mers, where n ϭ 2-7, were used to fit the experimental data.
The monomer/pentamer (1:5) model produced the best fit for the WT SH protein in C14 betaine (Fig. 3, A-C). The residuals are all closely distributed around zero, and 2 and root mean square deviations values were small, 1.09103 and 0.00704, respectively, suggesting that the 1:5 model is a good description of the experimental data. The calculated standard free energy (⌬G x 0 ) of pentamerization (44) of SH protein in C14 betaine micelles was Ϫ16.3, or Ϫ3.3 kcal/mol per monomer.
For the range of concentrations used in the AUC experiments (Fig. 3, A-C), the fraction that formed pentamers was 75-87% in C14 betaine, C14SB (Fig. 3D). The same panel shows that SH protein formed even more stable pentamers in DPC and C8E5 micelles, with K x ϭ 5.15 ϫ 10 13 and 1.83 ϫ 10 8 M Ϫ4 , which results in ⌬G x 0 per monomer of Ϫ3.74 and Ϫ4.9 kcal/mol, respectively.
Thus, results from PFO-PAGE, BN-PAGE, and equilibrium sedimentation indicate that SH protein forms homopentamers in detergent. As the polar headgroup of DPC is more relevant  biologically, this detergent was used for further NMR studies (see below).
Secondary Structure of SH Protein Reconstituted in Model Lipid Bilayers-Secondary structure predictions indicate that the ␣-helical region of SH protein spans residues 16 -46, i.e. ϳ40% of the molecule (45), whereas the TM domain is predicted to encompass residues ϳ20 -40 (46).
For full-length SH protein the relative area corresponding to ␣-helix and ␤-structure was 60% (ϳ40 amino acids) and ϳ30% (ϳ20 amino acids), respectively. The remaining of the structure in SH protein is likely to be represented by random coil or turn regions.
For SH-TM (18 -43), the ␣-helical content was calculated to be 74% (ϳ20 amino acids) and ϳ10% (2-3 amino acids) of ␤-structure, i.e. mostly ␣-helix as expected from a predicted ␣-helical TM domain. The full-length SH protein shows a higher content in ␣-helix relative to the TM, suggesting that in the full-length protein there is an extramembrane ␣-helical region of ϳ20 amino acids. Although it is difficult to extrapolate results obtained for a fragment, one can speculate that this extra ␣-helical region should be found in the N-terminal extramembrane domain SH(1-17) because the spectrum for and at three different speed as follows: 16,000 (blue), 19,500 (green), and 24,000 (red) rpm, in A-C, respectively. Data for WT was globally best-fitted to a monomer-pentamer equilibrium model. The graph shows both data points (black filled circles) and fitted function (in color). Lower panels represent fit residuals. D, WT oligomeric species distribution in C14 betaine, C8E5, and DPC detergents, where the thick bar on the curve represents the range of protein/detergent molar ratios used in the AUC experiment. The dotted line represents the protein/DPC molar ratio (1:200) used in the NMR experiments (see below) and its corresponding pentamer fraction (ϳ90%). C-14 betaine, DPC, and C8E5 detergent concentrations were 5, 15, and 33 mM, respectively. JULY 13, 2012 • VOLUME 287 • NUMBER 29

Hydrogen/Deuterium (H/D) Exchange of SH Protein Reconstituted in Model Lipid
Bilayers-The amide II band in the samples above was used to monitor protein H/D exchange, as the amide II band is due primarily to the peptide backbone N-H bending vibration. Upon H/D exchange, the peptidic N-H bond becomes N-D and the frequency of amide II downshifts from near 1545 to 1450 cm Ϫ1 (ϳ100 cm Ϫ1 ) (47). In a typical experiment, the percentage of H/D exchange can be measured from the decrease in the intensity of amide II band at 1550 cm Ϫ1 . The amide II band spectra (Fig. 4, D-F) were recorded either in the presence of H 2 O or after 1 h exposure to D 2 O. For full-length SH protein (Fig. 4D), 45 Ϯ 6%, i.e. 30 Ϯ 4 residues of the protein, was protected against H/D exchange, which therefore may include some extramembrane residues. For SH-TM(18 -43), 77 Ϯ 8% was resistant to H/D exchange (ϳ20 Ϯ 2 residues), consistent with the expected number of residues in the ␣-helical TM domain. Finally, the C-terminal peptide SH(45-64) exchanged almost completely, with Ͼ90% (Fig. 4F).
NMR Study of SH Protein in DPC Micelles-NMR spectra were recorded in DPC micelles, as it was found that SH forms stable pentamers in this detergent (Fig. 3D). Reconstitution of SH protein in DPC micelles resulted in well resolved 1 H-15 N-TROSY-HSQC spectra (Fig. 5A), where most of the resonances could be assigned to individual residues. However, for extramembrane residues 5-14, 57, and 59 -64, two or more closely spaced cross-peaks were observed, representing a slow equilibrium of two spectroscopically resolved conformations (see supplemental Fig. S2). This structural heterogeneity was also observed in other solubilization media such as SDS, DHPC micelles, and DMPC/DHPC isotropic bicelles with q ϭ 0.1, in the studied pH range from 4.5 to 6.5 (data not shown). For NMR

TABLE 1 Analysis of the amide I region in SH and SH-TM
Assignment, wave number (v), and relative areas of the component bands in the amide I region of SH protein and SH-TM were reconstituted in DMPC bilayers (Fig.  4). The standard deviation is a result of three independent experiments. For the SH-C20 peptide, it was estimated that the ␤-structure constituted at least Ͼ60% of the peptide.

SH protein(1-64) SH-TM(18 -43)
Assignment analysis, only the major component of each cross-peak was used. A comparison of the 1 H-15 N-TROSY-HSQC spectra in water and in 99% D 2 O (Fig. 5, A and B, respectively) shows that amide protons in 19 residues are significantly protected from exchange with water (Met 23 -Leu 41 ). This is consistent with the infrared H/D exchange data for the TM peptide (ϳ20 residues) obtained in lipid bilayers (Fig. 4) (18). However, this number is lower than the ϳ30 residues determined for the full-length SH protein in lipid bilayers by FTIR (Fig. 4B). The origin of this difference is discussed below. The dynamic properties of SH were studied using a 1 H, 15 N steady-state HNOE experiment (Fig. 5C), obtained for the backbone 1 H-15 N moieties. HNOEs provide information regarding backbone molecular motions in the picosecondnanosecond time scale and therefore help to identify structured regions (48). HNOE values of the majority of the residues of SH protein correspond to a fully folded protein (residues 5-57), flanked with highly mobile residues at the termini, although the central part of the protein is the most stable (residues ϳ12-49).
The structure of the SH protein monomer was calculated using the constraints derived from 1508 NOEs and 126 dihedral angle constraints using CYANA (Table 2) (31, 32). The upper limit, dihedral angle, and hydrogen bond constraints were input into CNS-1.2 (35) and recalculated starting from an extended conformation using the default-simulated annealing protocol. From the NMR model, the total ␣-helical content of SH protein can be estimated at ϳ53%, which is comparable with ϳ60% obtained using FTIR data ( Table 1). Analysis of intra-monomer NOEs shows that SH protein composes two distinct ␣-helical segments (Fig. 6, A and B). The first is close to the N terminus (residues 7-16) and is not protected from H/D exchange in DPC micelles (Fig. 5B), although FTIR data suggested otherwise. The second corresponds to the TM domain (residues 21-44), and is almost completely protected from H/D exchange in DPC micelles (Fig. 5B) and in lipid bilayers (Fig. 4).
For both N-terminal and TM-helices, backbone and sidechain heavy atom root mean square deviations of a set of 20 monomeric conformers were 0.50 and 0.80 Å, respectively, suggesting a well defined structure. These two ␣-helical domains are connected by a stretch of extended conformation (residues ϳ17-22).
At the C-terminal extramembrane domain, the structure consists of an extended strand (residues 46 -57), susceptible to H/D exchange in DPC micelles (Fig. 5B) and also in lipid bilayers (Fig. 4). The data suggest an overall appearance of an extended ␤-turn formed by 11 residues, stabilized by hydrogen bonds between amide and carbonyl groups at the tip of the turn structure, which is confirmed by NOEs (Fig. 6C). We note that from FTIR data, the number of residues forming ␤-structure was slightly higher (ϳ20 residues, see above). Despite these differences, the secondary structure composition of SH determined by NMR in DPC micelles and by FTIR in lipid bilayers is in reasonably good agreement.
Reconstruction of the Pentameric Model of SH Protein Using NMR Data-In the first stage, the relative orientations of N-terminal helix, TM helix, and C-terminal ␤-turn were defined using 51 RDCs in the program MODULE (30). In all models fitted, the N-terminal ␣-helix was found to be positioned orthogonal to the TM ␣-helix axis (see supplemental Fig.  S3). The C-terminal ␤-turn structure was also found to be orthogonal to the TM-helix, and it may be lying on the surface of the micelle (Fig. 7, A-C).
The multibody HADDOCK algorithm (36) was used to reconstruct the pentameric structure of SH protein. To identify inter-helical NOEs, an exhaustive search was undertaken of all available NOEs that cannot be satisfied within the structures of monomers. After verification against a previously obtained low resolution model (18), the inter-monomeric contacts were implemented as enforced restraints for the C5-symmetrical input in HADDOCK. In this way, seven unique NOEs stemming from regions between Ser 5 and Trp 15 were found between the "head" of the N-terminal helix of a monomer and the "tail" of an adjacent N-terminal helix, and another eight unique intermonomeric NOEs were found between side chains of the TM ␣-helical domain (supplemental Table S1). To improve convergence, residues with negative HNOE values (Met 1 -Thr 4 and Arg 59 -Thr 64 , Fig. 5C) were removed from the calculations. Residues comprising the "link" (Tyr 17 -Leu 20 ) between the N-ter-minal and TM ␣-helices were defined as fully flexible, whereas the two helices and the C-terminal ␤-turn (segments Ser 5 -Trp 15 and Ile 22 -Pro 61 ) were defined as semi-flexible. Thus, RDCs, center-of-mass restraints, C5-symmetry restraints, noncrystallographic restraints, and five copies of the truncated monomers were used to build the pentameric SH assembly (Fig.  7, A-C).
The convergent pentameric model of SH shows a funnel-like pore of about 45 Å in length (Fig. 7, D-F), spanning residues belonging to the N-terminal helix, TM helix, and C-terminal ␤-turn. At its narrowest, the pore diameter is about 3.5 Å, formed by side chains of Ile 36 . The N-terminal side (cytoplasmic) of the pentamer is very wide due to the ring of the N-terminal helices (up to 18 Å in diameter), whereas at the C-terminal side (extracellular/lumenal) the pore is more constricted, to a maximum of about 8 Å in diameter.
To validate the general topology of the pentameric structure obtained from HADDOCK, we obtained paramagnetic induced relaxation enhancement data of cross-peaks in TROSY-HSQC spectra. The following reagents were used: (i) 5-doxyl stearic acid (5-DSA) to probe the vicinity of the phosphate headgroup of the DPC detergent; (ii) 16-doxyl stearic acid (16-DSA) to probe the central hydrophobic core of the DPC micelles, and (iii) water-soluble Gd(DOTA) to probe 1 H N in residues that are water-exposed were measured.
In the presence of 1 mM 5-DSA (Fig. 8A), intensities of the N-terminal helix, C-and N-ends of the TM domain, and parts of the C-terminal ␤-turn decreased by more than 25%, whereas the TM region was not affected. This broadening effect indicates that the N-terminal helices and the C-terminal ␤-turn are either lying on or slightly embedded into the surface of the micelles. In the presence of 1 mM 16-DSA (Fig. 8B), residues of the N-terminal helix and C-terminal ␤-turn also showed some attenuation; however, the most affected residues were found in the TM domain, confirming their location at the center of the micelle hydrophobic core.
To delineate water-exposed residues, we employed a watersoluble probe Gd(DOTA). We found that at low concentrations (1-3 mM) of Gd(DOTA), the longitudinal proton relaxation of solvent-exposed residues was mostly enhanced, resulting in higher cross-peak volumes observed for the flexible regions at the protein termini (data not shown). However, in the presence of 100 mM Gd(DOTA), it was found that residues expected to be exposed to water are in fact broadened. The resonances still visible in the HSQC spectrum (Fig. 8D) closely correspond to residues protected from H/D exchange (Fig. 8C), thus confirming the position and topology of SH pentamer in the DPC micelles. The spectral regions in Fig. 8, C and D, are directly super-imposable, although slightly shifted due to the paramagnetic effect.
Patch Clamp of HEK 293 Cells Expressing SH Protein-To test the channel activity of SH protein, full-length protein was transiently expressed in HEK 293 cells. At neutral pH, no significant difference was observed between currents mediated by vector alone (Fig. 9A, left panel) and SH protein (Fig. 9A, middle panel) (unpaired t test, p ϭ 0.1928, for currents evoked at 70 mV). However, a more acidic pH led to a significant increase in a Intermolecular NOEs were not included in the total distance constraints for SH monomer calculation. b Backbone hydrogen bond and dihedral restraints of ␣-helices and ␤-turns are applied for regions confirmed to be ␣-helices or ␤-turns according to local NOE pattern and chemical shifts (assessed with the PREDITOR program for predicting polypeptide secondary structure). c Statistics are calculated and averaged over an ensemble of the 20 structures with lowest target function from CYANA. d Ramachandran analysis indicates that 1.6% of residues are disallowed. This residue (Thr 4 ) is found within the flexible region of the protein and can be safely ignored.
current density by almost 5-fold (unpaired t test, p ϭ 0.0013, for currents evoked at 70 mV) (Fig. 9A, right panel). A rectifier behavior, i.e. more outward conductance (positive ions going out) than inward (positive ions going in), is observed at this lower pH (Fig. 9B). The bath solution used in this experiment, artificial cerebrospinal fluid, contained a high concentration of NaCl (124 mM), whereas the internal solution contained a high concentration of potassium ions (145 mM), to mimic physiological conditions. Thus, the estimated equilibrium potentials for sodium and potassium, E Na and E K , were ϩ65 and Ϫ87 mV, respectively. In case of a strong selectivity for sodium or potassium, the reversal potential (potential at which no net current is observed) should be present near the respective equilibrium potential of each of these ions. The observed value of reversal potential of HEK 293 cells expressing SH protein was ϳ0 mV (Fig. 9B), indicating no selectivity of SH channel for sodium or potassium.
We reasoned that the pH dependence observed in the pH range 5.5-7.4 may be related to the protonatable His residues His 22 and His 51 (pK a ϳ6.5) (49). His 22 is located at the interface between the TM domain and the cytoplasmic side (see sequence in Fig. 6B), and in our previous low resolution model for the SH-TM pentamer (18) it had a lumenal orientation. This is reminiscent of His 37 in the TM domain of influenza A M2, M2A, which is lumenally oriented and involved in pH-mediated gating (50), and it also has a stabilizing role (51). The other His residue in SH protein, His 51 , is located at the extracellular C-terminal side and should be readily protonated when the pH changes in the patch clamp experiment. Thus, the experiment was repeated with single mutants H51A, H22A, and double mutants H22A/H51A (AA) and H22F/H51F (FF) (Fig. 9, C-G).
Although none of the mutants showed channel activity at pH 7.4, mutant H51A showed activity and pH activation comparable with WT (Fig. 9C), which supports a role for His 22 protonation in activation. Although His 22 is near the cytoplasmic side of the TM domain, it still may be exposed to changes in pH at the extracellular side, at least when His 51 is missing. Surprisingly, mutant H22A showed an even larger increase in current-density relative to either WT or H51A under acidic conditions (7-fold; p ϭ 0.0049, un-paired t test) (Fig. 9D). The mutation at His 22 may lead to unstable TM-TM interactions and a very leaky membrane when the pH is lowered, may be through His 51 protonation. Finally, none of the double mutants, AA or FF, lacking His showed activation upon lowering the pH (Fig. 9, E and F). All the traces obtained at pH 5.5 were compared in Fig. 9G).
Overall, these results suggest that the pH-sensitive behavior observed is due to His protonation. It is difficult to delineate the relative importance of these two histidines because absence of one of them may affect the overall conformation of the protein, particularly His 22 , located within the TM domain. However, given the pH-sensitivity of both single His mutants, it is likely that both residues have some gating role.
SH Protein Localization and Topology in Cells Expressing SH Protein-Both WT (active) and FF mutant (inactive) were found to be expressed abundantly in HEK 293 cells, and the C-terminal FLAG tag was localized at the plasma membrane when cells were not permeabilized with Triton X-100, as shown by using an antibody against a C-terminally located FLAG tag (Fig. 10). These conclusions are in agreement with previous studies performed in infected cells (10,13). However, similar experiments performed to localize an N-terminally located FLAG tag were not successful, with or without Triton X-100 permeabilization, which suggests that the N-terminal end of SH may be membrane-embedded or -truncated. In any case, these results show that the lack of channel activity in the double mutant FF cannot be attributed to deficient expression or anomalous localization.
BN-PAGE of SH Mutants-To support that the lack of activity observed in mutant FF is not due to protein unfolding or defective oligomerization, we compared the mobility of mutant FF and WT SH protein in BN-PAGE (Fig. 11). Mutant FF produced a single band after solubilization in a variety of detergents, with a mobility similar to that of WT SH protein. In the case of C8E5 or C14 betaine, SH WT has been shown to form homopentamers (see Fig. 3); therefore, the similar mobility observed for WT and mutant FF in BN-PAGE suggests mutant FF in these detergents is also pentameric.
In contrast, species containing the mutation H22A, i.e. mutants H22A and double mutant AA, behaved most differently from species that have His 22 , i.e. WT or H51A. For example in SDS, H22A produced a distinct pattern with larger molecular weights, compared with WT and H51A, whereas double mutant AA had a tendency to aggregate before entering the gel (supplemental Fig. S4A).
Similar conclusions can be drawn from BN-PAGE experiments, where the sample was presolubilized in C14 betaine, DPC, or C8E5. For example, in C14 betaine (supplemental Fig.  S4B), both H22A and AA ran faster than WT and H51A, and AA also showed a tendency to aggregate. In DPC (supplemental Fig. S4C), both H22A and AA produced smeared large aggre- gates, compared with WT and H51A. In C8E5 (supplemental Fig. S4D), both H22A and AA mutants aggregated before entering the gel, whereas WT and H51A produced almost identical patterns.
These results suggest that although His 22 is critical for preserving TM interactions, it can be substituted for Phe but not for Ala. The latter is consistent with the patterns of conserva-tion observed in multiple alignments (supplemental Fig. S6). Finally, although mutant H22A showed channel activity at low pH in patch clamp experiments, the AA mutant was inactive, suggesting that His 51 also has an important functional role in activation.
pH Activation and Role of the Two Histidines-We have shown that the I/V plots obtained in the samples WT, H51A,  and H22A are not equivalent (Fig. 9G); in order of decreasing conductance, they can be arranged as H22A Ͼ WT Ͼ H51A. His 51 is located extracellularly; therefore it should be readily protonated by changing the pH in the whole-cell patch clamp experiment.
Thus the activation of mutant H22A can be explained by protonation at His 51 , whereas the larger conductance than WT may be due to the disrupted native TM interactions mediated by H22A.
Activation in mutant H51A may be reduced relative to WT due to the more hidden location of His 22 , more difficult to protonate. Alternatively, the absence of His 51 may make a conformational change mediated by His 22 less efficient. Finally, mutants AA and FF cannot be activated because they both lack His side chains.
In summary, lack of channel activity correlates with predicted functional features of SH and not with loss of SH structural integrity or anomalous folding. The latter would be expected if the observed conductance was due to direct or indirect interaction with an endogenous channel.
In an attempt to localize the structural variations associated with pH changes, the pH of the SH protein sample in DPC micelles was increased from 4.4 to 6.5. Chemical shift changes were observed not only at His 22 and His 51 but also at other mostly polar residues as follows: Thr 7 , Phe 10 -Ser 11 , Lys 13 -Trp 18 , Thr 19 -His 22 , Thr 25 -Thr 26 , and residues located at the

Structure and Function of RSV SH Protein
C-terminal end of the TM, Asn 42 -Leu 44 , and Arg 56 and Leu 57 (Fig. 12A). Fitting a 3.6 periodic sine wave to the N-terminal helix, maximal perturbation was observed in residues that are in phase: Thr 7 , Ser 11 , Phe 14 , which are oriented toward the lumen of the channel (Fig. 12B).
Cu 2ϩ Interaction with SH His Residues-For influenza A M2 proton channel, it has been shown that Cu 2ϩ can inhibit channel activity by chelation to the histidine imidazole groups (52). To test if a similar effect could be observed in the SH protein, the bath solution was perfused with external buffer containing increasing concentrations of copper chloride, from 1 nM to 1 mM. A concentration of 1 M Cu 2ϩ successfully inhibited the current evoked through SH protein by a factor of 2 (0.53 Ϯ 0.06) (data not shown). However, this inhibitory effect was also observed when using mutant H22A, which indicates that His 51 is also involved in this regulation. At the same time, this suggests that His 51 from different monomers may be in contact in biological membranes, i.e. the fragment from the TM to His 51 may be in a different conformation in DPC micelles relative to biological membranes. This was also suggested by comparing H/D exchange in NMR and FTIR experiments (see above).
We then used TROSY-HSQC spectra to measure the chemical shift variation and intensity decrease of the 1 H-15 N backbone amide moieties when SH protein in DPC micelles was exposed to 1-50 M of Cu 2ϩ . Although more localized to the interfacial regions of the pore, chemical shift changes were observed throughout the protein (supplemental Fig. S5), and neither His residue was particularly affected. The interaction of copper and SH protein in detergent micelles therefore is not specific to His residues.
SH Protein Is Not a Proton Channel-Inspired by the possible similarities between SH protein and the proton channel M2A, we tested if SH acts as a proton channel. However, we could not detect any proton transport when SH protein was reconstituted in E. coli liposomes or a mixture involving 1-palmitoyl-2oleoyl-sn-glycero-3-phosphocholine, 1-palmitoyl-2-oleoyl-sn-glycero-3-[phospho-L-serine], and cholesterol, an assay used to test proton channel activity in influenza A M2 (53) (results not shown) in any of the protein/lipid ratios tested.

RSV SH Protein Forms Pentameric
Oligomers-In studies toward oligomeric size determination, the source of protein, i.e. synthetic or recombinant, presence of tags, mutations, and reconstitution environment may strongly affect the final outcome. In our hands, the electrophoretic mobility pattern of synthetic SH protein was always slightly different from recombinant SH protein, 5 possibly due to some side chain modifications during synthesis. Therefore, in this study, we have used recombinant SH protein, where the extra amino acids are only three, N-terminally located to minimize any possible disruptive effect. In addition, the fusion construct used was expressed in the membranes and not in inclusion bodies. This not only suggests that MBP has a chaperone-like effect but indicates that in our work the state of SH protein when expressed should be close to the native conformation.
The results presented show that in a variety of the following environments, SH protein forms a single oligomer that is pentameric: (i) in PFO detergent; (ii) in a BN-PAGE system, and (iii) in sedimentation equilibrium experiments in C14 betaine, DPC, and C8E5, which is consistent with results obtained previously for peptide SH-TM (18). This confirms the importance of the TM domain in the pentamerization of SH protein.
In a recent electron microscopy study on full-length SH protein in liposomes or DHPC detergent micelles (17), very large (ϳ9 nm) structures were interpreted as pentameric or hexameric forms, with a central pore diameter of 19 and 26 Å, respectively. The authors proposed a hexameric model, and not a pentameric one, although a rationale for that choice was not made explicit (17). In that study, however, SH protein was in 5 S. W. Gan, unpublished results. An arbitrary cutoff of ⌬␦ ϭ 0.05 is shown. P represents proline residues, W⑀ represents the side chain of Trp 15 , and * represents a nonobserved peak. B, SH pentameric assembly showing perturbations above the cutoff (in red) in A. One monomer has been removed for clarity.
fact an 86-residue construct, i.e. SH was prolonged by 19 additional N-terminal amino acids, plus 3 C-terminal ones. This means that the construct was almost one-third longer than SH protein itself, and this fusion protein appeared in inclusion bodies when expressed. We note that a central pore of 19 -26 Å is not compatible with the cation-selective ion channel activity observed by patch clamp. In addition, it is questionable how such a large pore could be formed with only 5-6 TM helices (one per SH monomer). The model we report herein, may offer an alternative explanation to a pentameric arrangement of TM domains; indeed, a ring formed by the N-terminal helices connected via a flexible linker to the TM domain would be compatible with the dimensions reported by electron microscopy.
However, we also note that although our AUC results indicate that WT SH monomer-monomer interactions are sufficiently strong, the resulting oligomerization state may still be sensitive to small changes in the sequence. This underlines the problems reported in the literature regarding oligomeric state determination of viroporins, especially when mutants or artificial constructs are used, as these may disrupt native interactions (54 -57).
Secondary Structure and H/D Exchange of RSV SH Protein-Overall, the NMR results, combined with FTIR results, indicate that SH protein has one TM ␣-helix, an N-terminal extramembrane helix, and a C-terminal hairpin-like region. The N-terminal region is linked to the TM by an extended linker peptide formed by residues 17-21. An intriguing structural feature of SH protein is the observation of conformational heterogeneity manifested as doubling of cross-peaks, starting from the N terminus and propagating to residue 19, and also in the C terminus (supplemental Fig. S2). The N-terminal region corresponds to the ring of tightly packed short N-terminal helices, a feature reminiscent of the C-terminal extramembrane helix found in influenza A M2 (fragment 18 -60) (58). Because of the significantly different RDCs observed for the 1 H-15 N indole group of Trp 15 in minor and major conformations, with Ϫ17 and 0.4 Hz, respectively, one may conclude that the heterocycle ring of Trp 15 can be aligned along or orthogonal to the membrane surface. This variable orientations of the indole ring of Trp 15 , and the pH activation likely mediated by histidine side chains, are also reminiscent of similar features encountered in influenza A M2, which has a "Trp gating" mechanism (58). Further studies are required to assess the physiological relevance of the minor alternative conformation.
We note that, although NMR and FTIR results are remarkably consistent despite the different environments used, there are several differences in terms of secondary structure content and in H/D exchange. We determined a higher ␣-helical content for SH protein using FTIR; therefore, it may be possible that the ␣-helical structure extends from the TM domain on both sides in lipid bilayers, rather than forming an extended structure.
These differences may be caused by the following: (i) the well known curvature effects of micelles used in NMR relative to the planar lipid bilayers used in FTIR; (ii) the different hydration between FTIR and NMR samples (higher hydration in NMR); and (iii) the molar ratio protein/detergent, which is significantly lower in NMR (1:200) relative to FTIR (1:50), in addition to inaccuracies of determination in the two techniques, especially in the case of FTIR.
SH Protein Forms Cation-selective Channels in Transfected Mammalian Cells-The fact that SH protein is found in plasma membrane (13) allowed the study of SH channel activity using whole-cell patch clamp. SH protein transiently expressed in mammalian cells produced nonselective cation channel activity, with passage of Na ϩ or K ϩ . Poor selectivity has also been observed in other viroporins (60,61), and it may be a hallmark of incomplete specialization of these channels. Indeed, in vitro studies show that SH protein is not essential, although it is important for effective viral infection in vivo, an importance also suggested by its strong sequence conservation.
In SH protein, only His 22 is close to the TM domain and is facing the lumen of the channel (Fig. 12B) (18). However, inspection of sequence alignments shows that SH proteins in human RSV, PIV5, or mumps virus always have at least one His residue near the equivalent position of His 51 in RSV SH protein, i.e. the extracellular/lumenal side (highlighted in yellow, supplemental Fig. S6), but the equivalent residue to His 22 is substituted in some cases, for example in bovine RSV to Phe. This suggests His to Phe at this position is a conservative mutation for this class of proteins, consistent with our results, because mutant FF forms pentamers like the WT. It would therefore appear that the His 51 residue in RSV is even more important than His 22 . Overall, these considerations suggest that a mutant H22F would form oligomers like WT and FF and would be activated by low pH.
It has been suggested that ion leakage may lead to dissipation of membrane potential and disruption of cell homeostasis, a common sign of apoptosis (67), leading to cell damage as infection progresses. In fact, the viroporins of Sindbis virus 6K, murine hepatitis virus E protein, influenza A M2 protein, HCV p7 protein, and poliovirus 2b and 3A protein have all been reported to affect apoptosis in infected cells (68 -70). In contrast, the SH protein encoded by members of Paramyxoviridae family simian virus 5 (SV5) (71), parainfluenza virus 5 (PIV 5) (72), mump virus (8), and hRSV (72) all inhibit apoptosis in several mammalian cell lines. Although promotion of apoptosis may help with virus release, inhibition of apoptosis during infection may give an advantage to the virus to replicate.
For both RSV SH and another member of the Paramyxoviridae family, parainfluenza virus 5 (PIV5), it has been observed (71) that SH protein is necessary for the inhibition of tumor necrosis factor-␣ (TNF-␣)-induced apoptosis (59,72). The fact that the absence of SH can cause cell death in RSV⌬SH-infected cells, compared with wild type RSV, and the fact that this is the case even in A549 cells, which are insensitive to TNF-␣-induced cell death (72), suggest that this toxicity is not uniquely mediated by a TNF-␣ pathway. Also, in MDBK and L929 cells, SH from PIV5 itself or from RSV A or B subgroups has a protective role against the cytopathic effect produced by PIV5 (8,59,71). Similarly, SH from PIV5 could be substituted by SH from mumps virus (8), even though these two SH proteins have no sequence homology. These data argue against a mechanism mediated by a specific protein-protein interaction with an unknown protein. We postulate that the protective effect of SH may be mediated by the protein itself, for example via membrane permeabilization using the channel structure reported herein. Further studies on hRSV-infected cells should gain insight into the significance of SH viroporin activity in the hRSV life cycle.