Outer pore topology of the ECaC-TRPV5 channel by cysteine scan mutagenesis.

The substituted cysteine accessibility method (SCAM) was used to map the external vestibule and the pore region of the ECaC-TRPV5 calcium-selective channel. Cysteine residues were introduced at 44 positions from the end of S5 (Glu515) to the beginning of S6 (Ala560). Covalent modification by positively charged MTSET applied from the external medium significantly inhibited whole cell currents at 15/44 positions. Strongest inhibition was observed in the S5-linker to pore region (L520C, G521C, and E522C) with either MTSET or MTSES suggesting that these residues were accessible from the external medium. In contrast, the pattern of covalent modification by MTSET for residues between Pro527 and Ile541 was compatible with the presence of a alpha-helix. The absence of modification by the negatively charged MTSES in that region suggests that the pore region has been optimized to favor the entrance of positively charged ions. Cysteine mutants at positions -1, 0, +1, +2 around Asp542 (high Ca2+ affinity site) were non-functional. Whole cell currents of cysteine mutants at +4 and +5 positions were however covalently inhibited by external MTSET and MTSES. Altogether, the pattern of covalent modification by MTS reagents globally supports a KcsA homology-based three-dimensional model whereby the external vestibule in ECaC-TRPV5 encompasses three structural domains consisting of a coiled structure (Glu515 to Tyr526) connected to a small helical segment of 15 amino acids (527PTALFSTFELFLT539) followed by two distinct coiled structures Ile540-Pro544 (selectivity filter) and Ala545-Ile557 before the beginning of S6.

The substituted cysteine accessibility method (SCAM) was used to map the external vestibule and the pore region of the ECaC-TRPV5 calcium-selective channel. Cysteine residues were introduced at 44 positions from the end of S5 (Glu 515 ) to the beginning of S6 (Ala 560 ). Covalent modification by positively charged MTSET applied from the external medium significantly inhibited whole cell currents at 15/44 positions. Strongest inhibition was observed in the S5-linker to pore region (L520C, G521C, and E522C) with either MTSET or MTSES suggesting that these residues were accessible from the external medium. In contrast, the pattern of covalent modification by MTSET for residues between Pro 527 and Ile 541 was compatible with the presence of a ␣-helix. The absence of modification by the negatively charged MTSES in that region suggests that the pore region has been optimized to favor the entrance of positively charged ions. Cysteine mutants at positions ؊1, 0, ؉1, ؉2 around Asp 542 (high Ca 2؉ affinity site) were non-functional. Whole cell currents of cysteine mutants at ؉4 and ؉5 positions were however covalently inhibited by external MTSET and MTSES. Altogether, the pattern of covalent modification by MTS reagents globally supports a KcsA homology-based three-dimensional model whereby the external vestibule in ECaC-TRPV5 encompasses three structural domains consisting of a coiled structure (Glu 515 to Tyr 526 ) connected to a small helical segment of 15 amino acids ( 527 PTALFSTFELFLT 539 ) followed by two distinct coiled structures Ile 540 -Pro 544 (selectivity filter) and Ala 545 -Ile 557 before the beginning of S6.
The TRP ion channels form a large class of cationic channels that are related to the product of the Drosophila TRP gene. TRP channels share a similar predicted topology of six transmembrane segments in which the amino acids that link the fifth and sixth transmembrane domains line the pore region (1). According to the recent IUPHAR classification of ion channels (2), the 21 members of the TRP family can be divided by sequence homology into three subfamilies (3,4) as short (TRPCx), long or melastatin (TRPMx), and osm-9-like or vanilloid-like (TRPVx) channels. The molecular domains that are mostly conserved among TRP channels include part of the S6 segment, ankyrin repeats in the N terminus, and a "TRP domain" in the C terminus (EWKFAR) (5), the latter being absent from TRPV channels. The TRPC and TRPV proteins have 2-4 N-terminal ankyrin domains suggesting that these proteins are coupled to the spectrin-based membrane cytoskeleton.
TRP channels vary significantly in their biophysical properties and gating mechanisms. In contrast to other members of the TRP family, TRPV5 and TRPV6 channels show strong inward rectification, exhibit anomalous mole-fraction effect, are activated by low [Ca 2ϩ ] i and inactivated by higher [Ca 2ϩ ] i (6 -8). TRPV5 and TRPV6 are also highly Ca 2ϩ -selective channels with PCa/PNa Ͼ 100. In particular, ECaC-TRPV5 displays a high Ca 2ϩ affinity with a K d of Ϸ2 M (7) that is comparable to the K d of Ϸ1 M for voltage-dependent Ca V channels (9). A single residue in the S5-S6 linker (Asp 542 ) was found to account for the high Ca 2ϩ affinity of ECaC-TRPV5 (7). The absence of the aspartate residue at the equivalent position in the pore region of TRPV1-4 channels might explain, together with the presence of a lysine residue, the Ϸ20-fold lower Ca 2ϩ selectivity of TRPV1-4 channels (10). TRPV5 and TRPV6 channels can also form homo-and heterotetramers suggesting that they are structurally and functionally related (11).
There is currently very little structural data available on the pore architecture of Ca 2ϩ -selective TRP channels. It is possible that the four aspartate residues form an extracellular ion binding site as it has been shown for the Glu 71 /Asp 80 residues in the KscA crystal structure (12). It has also been proposed that the four aspartate residues project in the pore lumen as it has been suggested with the four EEEE-residues locus that accounts for the channel high Ca 2ϩ -affinity (13) of Ca 2ϩ -selective Ca V 1.2 channels. In two landmark studies, cysteine mutation of each of the four EEEE residues locus of the Ca V 1.2 channel rendered the channel susceptible to irreversible inhibition by external sulfhydryl modifiers, indicating that the side chain was covalently modified by the methanethiosulfonate (MTS) 1 reagent (14,15). Cysteine substitutions at positions immediately adjacent to the EEEE locus (Ϯ1 positions) were also generally susceptible to sulfhydryl modification. Sulfhydryl modifiers had lesser effects on channels substituted one position further from the EEEE locus (Ϯ2 positions). These results suggested that the carboxylate-bearing side chains of the high affinity EEEE locus and their immediate neighbors were accessible from the water-filled extracellular medium in Ca V 1.2 channels.
To examine the topology of the pore region and the external vestibule in TRPV channels, we undertook a systematic analysis of pore residues accessibility using the substituted cysteine accessibility method (SCAM) (16). Mutant channels were expressed in Xenopus oocytes and their covalent modifications by externally applied membrane-impermeant methanethiosulfonate compounds of different charge and cross-section were measured. Based on the reactivity/accessibility to external sulfhydryl reagents, we report that some of the structural features of the bacterial KcsA channel, namely the coiled region in the S5-pore linker (turret) and the pore ␣-helix of 15 residues that follows the turret and precedes the selectivity filter, are conserved in TRPV5 channels. Furthermore, our results show that Thr 528 , Ser 532 , Glu 535 , and Thr 539 in the pore helix region are selectively accessible to positive MTS reagents from the external medium suggesting that they constitute part of the external vestibule in ECaC-TRPV5.

MATERIALS AND METHODS
Site-directed Mutagenesis of the Rabbit ECaC-TRPV5-The cDNAs coding for the wild-type ECaC-TRPV5 (GenBank TM AJ133128) (17) and the wild-type CaT2-TRPV5 (GenBank TM AF209196) (18) were obtained after reverse transcription of rabbit distal tubule mRNA as reported before (7). ECaC-TRPV5 and CaT2-TRPV5 were subcloned into the pT7TS vector (generously provided by Dr. Paul A. Krieg, University of Texas) using exonuclease III (19) for optimal expression in Xenopus laevis oocytes. Point mutations in ECaC-TRPV5 were performed with 39-mer synthetic oligos using the QuickChange TM XL-mutagenesis kit (Stratagene, La Jolla, CA). The C556S channel was used as a template for all cysteine mutations (see "Results"), and oligos were carefully designed to preserve that mutation. The nucleotide sequence of the S5-S6 linker including the background C556S mutation (over 600 bp) was bi-directionally analyzed using automatic sequencing by BioST (Lachine, Qué). DNA constructs were linearized at the 3Ј-end by BamHI digestion. Run-off transcripts were prepared using methylated cap analog m 7 G(5Ј)ppp(5Ј)G and T7 RNA polymerase with the mMessage mMachine ® transcription kit (Ambion, Austin, TX).
Expression of CaT2-TRPV5 Wild Type, ECaC-TRPV5 Wild Type, and Mutants in Xenopus Oocytes-Female Xenopus laevis clawed frog (Nasco, Fort Atkinson, WI) were anesthetized by immersion in 0.1% tricaine or MS-222 (3-aminobenzoic acid ethyl ester, Sigma) for 15 min before surgery as detailed before (7,20). cRNA was injected at a concentration of 0.46 -4.6 ng per oocyte depending upon the channel (wild type or mutant) being expressed. With only 0.46 ng of RNA, large whole cell inwardly rectifying currents (ϷϪ50 A) were routinely recorded with Li ϩ as the charge carrier (see below) for the wild-type channel less than 24 h after injection. ECaC-injected oocytes were incubated at 18°C in a calcium-free and serum-free Barth's solution for 24 -48 h before experiments.
Whole Cell Recordings-Whole cell currents were measured at room temperature with a two-electrode voltage-clamp amplifier (OC-725, Warner Instruments) as described before (7). Voltage and current electrodes (0.1-0.2 M⍀ tip resistance) were filled with 3 M KCl; 1 mM EGTA; 10 mM HEPES (pH 7.4). Instantaneous current-voltage relationships were measured using voltage ramps from ϩ80 to Ϫ150 mV at a rate of 0.575 mV/ms from a holding potential of Ϫ50 mV. Whole cell current-voltage curves (I-V) were measured under control conditions in the presence of the nominally calcium-free Li ϩ solution (in mM): 120 LiOH; 5 EGTA; 2 KOH; 20 HEPES titrated to pH 7.35 with methane sulfonic acid. Ca 2ϩ -free solutions were used since ECaC-TRPV5 undergoes a near-irreversible inactivation in response to a steady Ca 2ϩ influx over a 5-10-min period (7,8). Oocytes were perfused by gravity flow at a rate of 10 ml/min or 167 l/s. Taking into account the volume of the experimental bath and the dead volume in the perfusion line, the bath solution was completely exchanged within the first 30 s of perfusion. PClamp software Clampex 8 (Axon Instruments, Foster City, CA) was used for on-line data acquisition and analysis. Unless stated otherwise, data were sampled at 10 kHz and low pass filtered at 5 kHz using the amplifier built-in filter.
The 44 cysteine mutants were systematically tested for inward rectification and high Ca 2ϩ affinity to ensure that the introduction of cysteine residues did not distort the channel structure. Whole cell peak currents measured at Ϫ150 mV varied from Ϫ10 to Ϫ30 A for most cysteine mutants (see Table I for data on the pore helix region). Out of the 44 mutants tested, only 14 cysteine mutations failed to express whole cell currents significantly larger than non-injected oocytes. Alto-gether, the non-functional mutants, E515C, N518C, N519C, F523C, F531C, F534C, L536C, I541C, D542C, G543C, P544C, L551C, P552C, and Y555C were tested in a minimum of 3 different series of oocyte injections over a 1-year period using RNA concentrations 10-times higher than for the wild-type channel for culture periods of 2-4 days. In addition, non-functional mutants were incubated for periods up to 3 h with 5 mM dithiothreitol prior to experiments to reduce eventual intradisulfide bonds.
ECaC1 wild-type and mutant channel affinity for Ca 2ϩ ions was assessed from the calcium block of whole cell Li ϩ currents as described previously (7,9,21). Ca(OH) 2 was added to the solution to obtain the desired level of free calcium. The stability constants used to calculate the free calcium concentration were taken from Fabiato and Fabiato (22). Ca 2ϩ block was reversible at Ͼ 90% in all experiments reported here. Data were analyzed using Origin 6.1 (OriginLab Corporation, Northampton, MA) software. Results are presented as mean Ϯ S.E. Unpaired Students's t test was used for statistical comparison.
Substituted Cysteine Accessibility Method-MTS reagents MTSEA ((2-aminoethyl)methanethiosulfonate bromide), MTSET (2-(trimethylammonium)ethyl methanethiosulfonate bromide), and MTSES (sodium (2-sulfonatoethyl) methanethiosulfonate) were purchased from Toronto Research Chemicals (Toronto, Canada). Positively charged MTSEA and MTSET are both 1-nm long but differ substantially in surface area with diameters of 3.6 Å (10 Å 2 ) and 5.8 Å (26 Å 2 ), respectively (23). MTSET and MTSES are predicted to be of similar size although they differ in charge. Because the MTS reagents are rapidly hydrolyzed (5-10 min), they were always prepared fresh before use as described before (16). MTSEA is positively charged but also potentially membrane permeant under certain conditions (24) whereas MTSET (positively charged) and MTSES (negatively charged) are considered to be membrane impermeant (24). Whole cell current traces were routinely recorded using voltage ramps in the presence of Li ϩ as the charge carrier (120 mM LiMeS ϩ 5 mM EGTA) in the following sequence: 1) under control conditions, 2) after the bath addition of 5 mM MTS reagent for 30 s, 3) after 5-min perfusion with 5 mM MTS, and 4) after washing out the unreacted MTS reagent for 10 min with the control solution. In addition of providing the MTS response over a wide range of potentials, voltage ramps allowed us to check for the presence of rectifying currents throughout the course of an experiment. For most mutants, the gradual increase in nonspecific currents or leaky currents could be immediately assessed from the gradual disappearance of the trademark inward rectification of ECaC-TRPV5 currents. Experiments where nonspecific leaks developed during the experiment were simply discarded. The reversibility of covalent bonds was tested with 5 mM BMS (bis(2-mercaptoethylsulfone)) (Calbiochem, San Diego, CA) a water-soluble reagent considered to be a superior reducing agent than dithiothreitol (25,26). The percentage of currents remaining after MTS modification was computed at V m ϭ Ϫ150 mV using [(I wash )/(I ctrl )] ϫ 100 where I wash is the whole cell peak current remaining after MTS application and washout and I ctrl is the whole cell peak current measured before MTS application. For the figures, current traces were averaged from n Յ 4 separate experiments and are shown as the mean Ϯ S.E. Hence, the thickness of the traces actually reflects the experimental variability of the MTS response. The standard errors tended to be smaller in the absence of functional modification. Inhibition was considered "significant" at p Ͻ 0.01.
Computer-predicted Structure and Homology Modeling of the Pore Region in ECaC-TRP5-Sequence alignments were performed with the INSIGHTII HOMOLOGY module, which integrated the threading technique Profiles-3D developed in the laboratory of David Eisenberg and co-workers (27). The analysis of the three-dimensional scoring table led to the choice of the KcsA channel as a possible template to be used for ECaC-TRPV5 homology modeling. The score was considerably lower for MthK, and the alignment would require the introduction of several gaps in the structure. Computer-based homology modeling was performed with Modeler V6.2 (28) using the crystal coordinates of KcsA (PDB 1BL8) as a template and involved the generation of 50 monomer models of the ECaC-TRPV5 channel pore. Energy minimization was carried out with CHARMM. The dimer was obtained by superposing two ECaC monomers onto the KcsA tetramer. The surface three-dimensional representation of the ECaC-TRPV5 was generated with the INSIGHTII software (Accelrys, San Diego) as described elsewhere (16).

RESULTS
Modification of Wild-type TRPV5 Channels by MTS Reagents-To examine the topology of the external vestibule and pore region in TRPV channels, we undertook a systematic analysis of pore residues accessibility using the substituted cysteine accessibility method (SCAM) with hydrophilic MTS reagents (MTSEA, MTSET, MTSES). MTS reactivity of the wild-type rabbit ECaC-TRPV5 (GenBank TM AJ133238) was investigated in the presence of Li ϩ . As explained earlier, whole cell currents were measured in the absence of Ca 2ϩ since ECaC undergoes Ca 2ϩ -dependent inactivation over a 5-10-min period in the presence of Ca 2ϩ (8) whereas current levels of ECaC-TRPV5 and CaT1-TRPV6 are stable in Ca 2ϩ -free solutions with either Na ϩ (8) or Li ϩ (7, 29) as the charge carrier. As seen, whole cell currents measured under these conditions are strongly rectifying at positive voltages ( Fig. 1, A and B and Table I). Perfusion with 5 mM MTSES for up to 5 min did not modify the whole cell currents through ECaC-TRPV5 (not shown). However, external application of 5 mM MTSEA (  (Fig. 1B) suggesting that the former inhibition was nonspecific whereas the latter inhibition was truly covalent. This conclusion was further supported by the application of the reducing agent BMS that was found to fully reverse the MTSET-inhibition confirming that the MTS-induced inhibition involved the formation of a disulfide bridge between the MTS reagent and the channel protein (Fig. 1B).
In contrast to ECaC-TRPV5, the highly homologous CaT2-TRPV5 (GenBank TM AF209196) channel was not irreversibly modified by either MTS reagent (Fig. 1, C and D and Table I).
As seen, whole cell Li ϩ currents through CaT2-TRPV5 rectified strongly at positive voltages. CaT2-TRPV5 also displayed an anomalous mole fraction effect between Ca 2ϩ and Li ϩ and external Ca 2ϩ was found to inhibit whole cell Li ϩ currents with a K d ϭ 2 Ϯ 1 M (4), which is comparable to the value reported for ECaC-TRPV5 (7). CaT2-TRPV5 was completely insensitive to MTSET (Fig. 1D) and MTSES (not shown) whereas MTSEA inhibition observed after a 5-min exposure, was completely abolished upon the reagent washout (Fig. 1C). A comparison of their primary structure indicated that 11 out of the 15 cysteine residues of ECaC-TRPV5 were strictly conserved in CaT2-TRPV5 (Fig. 2). Six conserved cysteine residues are located in putative transmembrane segments S1, S4, and S5 whereas that the remaining five residues are located either in the N terminus (Cys 70 , Cys 112 , Cys 172 , and Cys 213 ) or in the C terminus (Cys 619 ). The three non-conserved residues namely Cys 4 , Cys 270 , and Cys 653 are located either in the intracellular N or C termini where they should be inaccessible to modification by external perfusion of membrane-impermeant reagents. The cysteine residue at position 556 in the pore region of ECaC-TRPV5 corresponds to His 549 of CaT2 where it could potentially be accessible from the external medium.
The conservative mutation of the Cys 556 residue to a serine resulted in an ECaC-TRPV5 channel completely insensitive to MTSET (Fig. 3B) and MTSES (not shown) as demonstrated by the current traces superimposed under all experimental conditions. The smaller MTSEA, documented to be somewhat membrane-permeant (24), caused current inhibition of whole cell currents after a 5-min perfusion period that did not persist upon extensive washout of the reagent (Fig. 3A). The global time course of MTS modification of C556S channels following a  Table I for complete details. 10-min period was reported for MTSEA (Fig. 3C) and MTSET (Fig. 3D) at two voltages ϩ20 mV and Ϫ100 mV. MTSEA steadily decreased currents measured at Ϫ100 mV over that period but currents could be recovered at 92 Ϯ 3% (4) after washout of the reagent (Fig. 3C). Whole cell currents at ϩ20 mV were not affected indicating that the MTSEA-induced modification did not affect the channel rectification. Longer incubation periods up to 20 min with MTSET or MTSES did not significantly affect whole cell currents (not shown). Hence C556S channels did not undergo covalent modification by either MTSEA, MTSET, or MTSES. The key biophysical features of the wild-type channel namely the steep inward rectification and the high affinity for Ca 2ϩ ions (Table I) were preserved in the C556S channel. The C556S channel was thus used as the template channel for the cysteine mutations in order to extend the characterization of the pore properties of the ECaC-TRPV5 channel undertaken in a previous work (7). Cysteine mutations were introduced one by one into the C556S channel in the S5-S6 linker region from Asp 515 to Tyr 555 . In all cases, whole cell currents were measured after 5-min exposure to 5 mM MTS (MTSEA, MTSET, or MTSES). The MTS-induced modification was reported after extensive washout of the reagent solution to TABLE I Covalent modification of ECaC-TRPV5 wild-type and mutants in the pore helix region Biophysical properties and covalent modification of wild-type TRPV5 channels and mutants in the putative pore helix region. Most mutant channels displayed the strong inward rectification of wild-type TRPV5 channels. Peak currents were measured in the presence of the nominally Ca 2ϩ -free Li ϩ solution. The percentage of control currents remaining after external MTS modification with 5 mM reagent was computed after washout of the unreacted reagent (see "Materials and Methods''). A value of 100 means there was no effect. Mutant channels preserved the high-Ca 2ϩ affinity as shown by the percentage of whole cell currents measured after exposure to 1 M free-Ca 2ϩ . Data were computed at V m ϭ Ϫ150 mV and are shown as mean Ϯ S.E. with the number n of experiments in parentheses. N/E, non-expressor.
Ϫ30 Ϯ 3 (28) 108 Ϯ 6 (9) 109 Ϯ 7 (6) 100 Ϯ 7 (9) 80 Ϯ 1 (4) FIG. 2. Putative secondary structure of the epithelial ECaC-TRPV5 Ca 2؉ channel. The deduced primary sequences of ECaC1 (Gen-Bank TM AJ133128) and CaT-2 (GenBank TM AF209196) are highly homologous with 84% overall identity. The 15 endogenous cysteine residues of ECaC-TRPV5 are shown as circles with the empty circles showing residues conserved between ECaC and CaT2 whereas filled circles highlight cysteine residues absent in CaT2. The inset underscores the pore region in the S5-S6 linker region with 46 out of 52 amino acids being strictly conserved between the 2 channels in that region. A single cysteine residue (shaded box) differs in that region, the Cys 556 in ECaC1 that is aligned with His 549 in CaT2. The residue Asp 542 previously identified as the key molecular determinant of high Ca 2ϩ affinity in ECaC-TRPV5 is shown with an asterisk. ensure that channel modification did not result from nonspecific effects.
External MTS Reactivity of the S5-to-Pore Linker Region Asp 515 -Tyr 526 -The MTS reactivity was first studied in the region referred to as the S5-to-pore linker region spanning 11 amino acids from Asp 515 to Tyr 526 . With the exception of L520C, all mutant channels in that region featured the typical inward rectification and the high Ca 2ϩ affinity of TRPV5 channels. Three consecutive mutant channels L520C, G521C, and E522C were strongly inhibited by positively and negatively charged MTS. Inhibition was nearly completed within 30 s and persisted through extensive washout periods with the control solution. Average I-V data are shown in Fig. 4 for E522C. As seen, perfusion with 5 mM BMS, a potent reducing agent almost completely restored whole cell currents in MTSEA-and MTSET-modified E522C channels confirming that the channel had formed a covalent disulfide bridge with the MTS reagents. BMS nonetheless failed to restore under the same conditions the whole cell currents of MTSES-modified channels. Altogether, these data suggest that L520C, G521C, and E522C were readily accessible from the aqueous external medium.
Such robust MTS reactivity was nonetheless limited to these three residues in that region. The SCAM data for the S5-pore linker are summarized in Fig. 5. Nothing can be inferred from positions 515, 518, 519, and 523 since E515C, N518C, N519C, and F523C mutants failed to express significant whole cell currents. Of the remaining mutants in that region, only D525C was covalently modified by the three MTS reagents and this, to a moderate extent with residual currents ranging from 60 Ϯ 7% (5) for MTSEA (p Ͻ 0.01), and 61 Ϯ 8% (4) for MTSET (p Ͻ 0.01), to 77 Ϯ 8% (4) for MTSES (p Ͻ 0.1) as measured at Ϫ150 mV. Mutants P517C and S524C were only partially inhibited by MTSEA with residual currents of 64 Ϯ 8% (4) and 45 Ϯ 10% (3) respectively, suggesting a limited access to larger reagents at these positions.
SCAM Analysis of the Pro 527 -Ile 540 Region-Cysteine substitution in the P527-I540 region resulted in 11 of 14 channels with typical inward rectification properties and high Ca 2ϩ affinity (Table I). Only two mutants, F531C and L536C failed to express whole cell currents larger than Ϫ1 A at Ϫ150 mV. The overall pattern of MTS modification in that region is summarized in Fig. 6A. Covalent modification by positively charged MTSEA and MTSET reagents resulted in a significant inhibition (p Ͻ 0.01) of whole cell currents at positions T528C, A529C, S532C, and E535C that persisted upon extensive washout (Table I). Average I-V data for T528C (Fig. 7, A-C) and E535C (Fig. 7, B-D) are shown in the presence of MTSEA (Fig.  7, A and B) and MTSET (Fig. 7, C and D). As seen, MTSEA and MTSET inhibition of T528C currents occurred within the time frame of bath perfusion (Ͻ 30 s). The E535C mutant channel was also significantly inhibited by MTSEA (Fig. 7B) but inhibition by MTSET (Fig. 7D) was reduced considerably and required a longer perfusion period. It follows that MTSET reactivity was strong within the first half of the region (528C, 529C,

FIG. 3. The endogenous cysteine at position 556 accounts for the MTS reactivity of the wild-type ECaC-TRPV5. A and B, C556S
currents display the trademark inward rectification of TRPV5 wild-type channels and the high Ca 2ϩ affinity (see Table I for details). Whole cell currents remaining after 5-mM exposure with MTSEA (A) or MTSET (B) were reported as mean currents Ϯ S.E. MTSEA-induced inhibition was fully reversed by washout with the saline solution indicating the non-covalent nature of the MTSEA inhibition. As seen, the C556S mutation eliminated the covalent inhibition by MTSET. Experimental conditions were described in Fig. 2. C and D, time course of MTS modification in C556S channels was measured at ϩ20 mV (empty circles) and at Ϫ100 mV (filled circles) every 30 s up to 10 min perfusion with 5 mM MTSEA (C), 5 mM MTSET (D), and 5 mM MTSES (not shown) followed by a 15-min washout with the saline 120 mM Li ϩ solution. Perfusion with MTSEA induced a steady decline in currents that was however washed out at Ն 90% by the saline solution to a level that is comparable to the level reached under the same conditions with MTSET or MTSES perfusion. 532C, and 535C) but absent after position 535 whereas the smaller MTSEA reagent induced robust inhibition down to position 539 (Table I). Indeed, F537C and T539C were functionally modified by MTSEA with residual currents of 59 Ϯ 4% (5) and 36 Ϯ 8% (4) respectively, after a 5-min perfusion period. Mutants L538C and I540C were completely insensitive to modification by MTSEA and MTSET.
Although residues Thr 539 and Ile 540 are located a few residues away from the Asp 542 residue responsible for the high affinity Ca 2ϩ binding site (7), the MTS response of either mutant was not altered when experiments were conducted in the presence of 1 M free Ca 2ϩ , the concentration required to inhibit 50% of the whole cell Li ϩ currents (results not shown). This suggests that Ca 2ϩ binding/transit through the channel did not significantly alter the pore structure to such an extent that it would modify the side chain accessibility to MTSEA and MTSET.
Finally, MTSES reactivity was absent in the entire Pro 527 -Ile 540 region. A slight increase in whole cell currents was however observed at some positions upon washout of the unreacted MTSES reagent although in both cases there was no discernible change in the whole cell currents in the presence of MTSES. The absence of MTSES-induced modification suggests the presence of an intrinsic electrostatic field, which would prevent negatively charged ions to penetrate into the pore.
Cysteine Mutations of the Ile 541 -Pro 544 Region Yielded Nonfunctional Channels-Cysteine substitutions of residues Ile 541 -Pro 544 surrounding the high affinity Ca 2ϩ binding site located at Asp 542 , namely I541C, D542C, G543C, and P544C all yielded non-functional channels. Although this absence of functional expression could result from a dysfunction of the perme-ation pathway, mutations of Asp 542 to Ala (A), Gly (G), Glu (E), and Asn (N) in the wild-type ECaC-TRPV5 channels were shown to produce functional channels with whole cell currents exhibiting the typical inward rectification properties (7).
The external accessibility of S6 residues was tested intermittently at positions Ala 560 , Ala 561 , Ala 563 , and Ala 566 . The A563C channel was non-functional whereas the A560C, A561C, and A566C channels remained completely insensitive to modification by MTSET and MTSES (results not shown) but A560C was moderately inhibited by MTSEA with residual currents of 67 Ϯ 8% (7) as measured at Ϫ150 mV. DISCUSSION A systematic analysis of some of the structural features of the external vestibule of ECaC-TRPV5 was performed using SCAM. Positively charged MTS reagents were generally the most reactive as expected for cation selective channels with MTSEA Ն MTSET Ͼ Ͼ MTSES. MTS-induced inhibition of ECaC-TRPV5 was reported after a 5-min exposure period, which is 10 times larger than the time constants reported for the rate of MTS modification in other ion channels (30,31). Under these conditions, inhibition was not significantly voltagedependent between Ϫ50 and Ϫ150 mV. Given that it is the rate rather than the level of inhibition that is usually influenced by the membrane electrical field (32), this observation suggests that channel inhibition had reached a steady state (33). The majority of cysteine mutants preserved the robust I-V rectification and the high Ca 2ϩ affinity that are typical features of ECaC-TRPV5. Nonetheless four (4) mutant channels (L520C, A545C, N546C, and Y547C) lacked both properties suggesting that the two molecular mechanisms could be linked in TRPV5 as it has been inferred in TRPV6 (34).
Homology Modeling of the S5-pore-S6 Region of ECaC-TRPV5-To establish a structural correspondence with crystallized K ϩ channels, the S5-pore-S6 region of ECaC-TRPV5 was analyzed using the structure-based threading PRO-FILES-3D method (INSIGHT II) (Fig. 8). Although there is basically no homology between ECaC-TRPV5 and KcsA at the primary sequence level, the scoring table generated by PRO-FILES-3D revealed that the pore region of ECaC-TRPV5 is structurally compatible with the K ϩ channel KcsA, which can thus be used as a template for homology modeling. The resulting sequence alignment as well as three-dimensional represen-tations of the pore region are shown in Figs. 8 and 9. The structure prediction globally supports a model whereby two helical regions comprising the transmembrane S5 and S6 segments encompass a "pore domain" consisting of a coiled structure (Glu 515 to Tyr 526 ) connected to a small helical segment of 15 amino acids ( 527 PTALFSTFELFLT 539 ), which in turn is attached to two coiled domains, the first one (Ile 540 -Pro 544 ) containing the high Ca 2ϩ affinity Asp 542 residue (7) and the second one extending from Ala 545 to Ile 557 . The PROFILES-3Dbased alignment establishes a structural correspondence between Asp 542 in ECaC-TRPV5 and Tyr 78 in the GYGD signature sequence known to contribute to the S 0 -S 1 K ϩ external binding site (35) in the KscA crystal structure obtained at 2.1Å resolution (12) thus positioning the high affinity Ca 2ϩ binding site at the entrance of the external vestibule. As seen in Fig.  9A, the selectivity filter between Ile 540 and Pro 544 could behave as an extended ␤-strand or a coil region, both structures being compatible with the x-ray structure of the selectivity filter or GYGD region in KcsA and MthK channels. This region is followed by a coiled structure between Ala 545 and Ile 557 predicted to be longer than in KcsA. The three-dimensional-based structure alignment finally suggests that the S6 ␣-helix should start at Thr 558 whereas the current secondary structure model (17) based on hydrophobicity plots predicts S6 should begin at Cys 556 .
The surface three-dimensional representation (Fig. 9B) of the model ECaC-TRPV5 channel displays a comprehensive color-coded picture of the residues modified by MTSET and their relative accessibility from the aqueous medium. Residues that displayed the strongest MTSET reactivity are colored in red whereas partial inhibitions are shown in pink. Yellowcolored residues formed non-functional cysteine channels. Residues that failed to react with MTSET are presented in blue. As seen, residues that displayed the strongest reactivity appear to be located at the external interface and surround the pore area. Residues that caused a partial inhibition (pink) or failed to inhibit ion fluxes appear to be positioned further away from the pore. Finally cysteine mutations that yielded non-functional channels are mostly found in the selectivity filter or close to the pore helix region. It is however important to bear in mind that the implication of nonexistent or partial functional modification cannot be immediately translated into structural information. The absence of modification (residues shown in blue in Fig. 9B) could result either from the inaccessibility of the residue or else from a "silent" covalent modification that does not affect ion fluxes. Partial inhibitions could be produced by the limited accessibility of the residue or from a decrease in the channel single channel conductance coupled with an increase in the channel mean open time (16) whereby the MTS reagent behaves like an open channel blocker. Despite these limitations, the surface three-dimensional representation globally supports our SCAM data to the extent that the strong reactivity of some residues was correlated with their accessibility from the external medium whereas non-reactive residues appear to be buried within the protein.
The Asp 515 -Tyr 526 Region Could Form a Coiled Structure in ECaC-TRPV5-The three-dimensional model obtained by homology (Fig. 9A) predicts that the region spanning 11-amino acids from Asp 515 to Tyr 526 in ECaC-TRPV5 should form a coiled structure. This region projects in the aqueous external medium and is called the "turret" in the crystal structure of KcsA (36), MthK (37), and KirBac1.1 (38) channels. The residues located in such a structure are expected to be readily accessible without any apparent periodicity in MTS-induced modification. The robust reactivity of three consecutive residues in that region (L520C, G521C, and E522C) argues for the presence of a coiled region that is easily accessible from the aqueous external medium. The SCAM data for the remaining S5-pore linker residues showed however milder reactivity for since only D525C was moderately modified by the three MTS reagents. Overall, the strong reactivity of three consecutive residues irrespectively of the charge of the MTS reagent agree with the surface representation of the S5-pore-S6 region as these residues correspond to the red-colored spots surrounding the channel pore (Fig. 9B).
The Pore Helix in ECaC-TRPV5-The structural region in ECaC-TRPV5 spanning from Pro 527 to Thr 539 could correspond to the "pore helix" brought to light by the x-ray structures of the bacterial KcsA (12,36), MthK (37), and KirBaC1.1 (38) channels. It is assumed that the pore helix provides a structural support to the selectivity filter. Its net negative dipole should contribute to the stabilization of K ϩ in the water-filled central cavity of the closed KcsA channel (39) and could account for charge selectivity in Ca 2ϩ -permeable glutamate receptors FIG. 6. A, histogram of the SCAM data for the pore helix region in the ECaC-TRPV5 channel. Effects of MTS reagents on the Cys mutants spanning the Pro 527 to Ile 541 region in the ECaC-TRPV5 channel. Mutants were perfused with 5 mM of the MTS solution for 5 min using the protocol described previously. The histogram reports the percent of whole cell currents remaining after washout of the MTS reagents with the saline solution such that a ratio of 100 indicates that whole cell currents were not modified by MTS reagents as compared with control current traces. Whole cell currents through F531C, F534C, and L536C were not significantly different than currents measured in non-injected oocytes. In some cases (P527C, L538C, and I540C), whole cell currents increased upon washout of the MTS reagent but this increase was not statistically significant (p Ͼ 0.01). MTSES did not significantly modify any mutant in this region. Inhibition by MTSET was significant at p Ͻ 0.001 for T528C and A529C and at p Ͻ 0.01 for S532C and E535C. Inhibition by MTSEA was significant at p Ͻ 0.001 for A529C and T539C while significant at p Ͻ 0.01 for T528C, S532C, E535C, and F537C (Table I). B, histogram of the SCAM data for the pore-to-S6 linker region. Effects of MTS reagents on the Cys mutants spanning the Ala 545 to Cys 556 region. The Cys 556 channel is the wild-type channel (see Fig. 1). Whole cell currents for L551C, P552C, and Y555C were not significantly different than currents measured in non-injected oocytes. Strong reactivity for MTSEA and MTSET was observed at the beginning of this region with A545C, N546C, Y547C, and V549C being significantly inhibited by MTSEA and MTSET at p Ͻ 0.001. In addition, MTSES significantly inhibited A545C, N546C, and V549C suggesting that charge was not critical in that region. D550C was inhibited by MTSEA at p Ͻ 0.001 and MTSET at p Ͻ 0.01. F553C was inhibited by MTSEA at p Ͻ 0.01. C556S was inhibited by MTSET at p Ͻ 0.001. Mutants S548C and M554C were not significantly modified by either MTS reagent. *, p Ͻ 0.01; **, p Ͻ 0.001.  (4, dark gray). MTSEA-induced covalent inhibition of T528C and E535C was achieved within 30 s because the averaged current traces recorded at 30 s were superimposed to the current traces recorded after washout. MTSET-induced inhibition of E535C required longer perfusion times suggesting that the rate of accessibility was decreased in that case. No inhibition was observed after 5 min of exposure to 5 mM MTSES for either mutant (not shown). See Table I for complete values. AMPAR (40). In the three-dimensional model obtained by homology (Fig. 9A), the side chains of hydrophilic residues Thr 528 , Ser 532 , Glu 535 , and Thr 539 are projecting toward the selectivity filter. This orientation is compatible with the observation that these hydrophilic positions were covalently modified by MTSEA and MTSET. In addition, the location of the hydrophilic Thr 533 residue could also explain its relative insensitivity to MTSET and MTSEA. A529C is the only residue that was covalently modified by MTSEA and MTSET despite being apparently inaccessible in the three-dimensional model. It could be speculated that the relative accessibility of Ala 529 derives from its relative proximity to the external medium being located at the same height as Asp 542 . The three-dimensional model patterned after the dimensions of KcsA should also be refined to take into account the observation that MTSET (26 Å 2 ) was able to reach position E535C whereas MTSEA (10 Å 2 ) could go further down to T539C at the junction where the pore helix finishes and the selectivity filter starts suggesting that the pore region is larger in ECaC-TRPV5 than KcsA. Finally, the three-dimensional model suggests that the access to the pore has been designed to select positively charged ions. Given that the side chains of Glu 535 and Asp 542 appear to project in the same direction, we can speculate that these negatively charged residues provide together a potent negative field that could prevent MTSES to reach its targeted cysteine residue in the pore helix.
The Structural Features of the Pore-to-S6 Linker Region in ECaC-TRPV5-The selectivity filter is predicted to encompass Ile 540 to Pro 544 , a region that was devoid of functional cysteine mutant channels and shown in yellow in Fig. 9B. Furthermore, the Pro 544 to Ile 557 region that links the selectivity filter to S6 is predicted to form a coiled structure (Fig. 9A) considerably longer than the equivalent region in the KcsA. The robust MTSEA and MTSET reactivity of the first three consecutive residues (A545C, N546C, Y547C) is indeed compatible with the presence of a coiled structure. Furthermore, A545C, N546C, and V549C channels were significantly inhibited by the negatively charged MTSES indicating that access to these residues was not limited as we observed in the pore helix region. The SCAM data were not however helpful in determining the beginning of the ␣-helix in S6 since most mutants after Asp 550 were either non-functional (L551C, P552C, and Y555C) or else non-reactive (M554C). The Cys 556 residue found to confer high MTSET-reactivity to the wild-type channel is found at the end of this coiled structure (Fig. 9A) where it could be accessible from the extracellular medium. We can only speculate that MTSEA was too small to block efficiently ion fluxes through the selectivity filter since the dimensions of the three-dimensional representation are very approximate. Finally, the A560C, A561C, and A566C residues present in the S6 segment were found to be completely insensitive to modification by external MTSET and MTSES reagents whereas A563C was non-functional. As predicted by the three-dimensional model (Fig. 9B), these hydrophobic residues should be buried within the channel.
Comparison with the SCAM Analysis of Models K ϩ Channels-There exists little data on the cysteine reactivity/accessibility of the external vestibule of crystallized K ϩ channels (KcsA, MthK, KirBac1.1) besides the observation that Y82C located upstream to the GYGD (77-80) signature sequence, can be functionally modified by external MTSET in the KcsA channel (41). SCAM was used to investigate the structural features of the extra long extracellular S5-P linker that is absent from ECaC-TRPV5 channels and was also used to study the external vestibule of voltage-gated K V 2.1 (42) channels. In this last case, cysteine-substituted residues in the Ile 369 -Gly 377 region that includes the lower part of the pore helix and the FIG. 8. Computer-predicted secondary structures generated by PROFILES-3D (INSIGHT II) using KcsA (PDB 1BL8) as the template. Structure homology was performed for the 499 -568 region of ECaC-TRPV5 with only part of the actual S5 and S6 transmembrane regions. The domains were identified as they appear in KcsA. The coiled structure following the selectivity filter is predicted to be longer in ECaC-TRPV5 than in KcsA such as the ␣-helix TM2 (shown as a dotted box) starts at Trp 87 in KcsA whereas it starts at Thr 558 in ECaC-TRPV5.
FIG. 9. A, ribbon three-dimensional representation of two ECaC-TRPV5 monomers obtained by homology modeling using Modeler 6.2 as viewed from a perspective parallel to the membrane. The side chains of MTSET-modified positions in the pore helix region (Thr 528 , Ser 532 , Glu 535 , Thr 539 ) and in the wild-type channel at Cys 556 are shown. Asp 542 is projecting toward the selectivity filter. B, surface three-dimensional representation of ECaC-TRPV5 obtained by homology modeling using Modeler 6.2 as viewed from a perspective parallel to the membrane. Residues are color-coded according to their reactivity toward MTSET. Residues that were very significantly inhibited by MTSET at p Ͻ 0.001 are shown in red whereas pink-colored residues were inhibited by MTSET at p Ͻ 0.01. Blue residues remained insensitive to MTSET and non-functional residues are shown in yellow. White residues were not tested. Asp 542 is shown in green to indicate the region of the selectivity filter. The arrow indicates the direction of the ion fluxes. selectivity filter were either non-functional or insensitive to external MTS modification (42) just as we have shown in ECaC-TRPV5. The alignment produced by PROFILES-3D would correlate A545C in ECaC-TRPV5 with Y380C in Kv2.1 (42). These residues were shown to be functionally modified by MTSEA, MTSET, and MTSES in both channels. Thr 539 in ECaC-TRPV5 would also be aligned with Thr 373 in Kv2.1. Since T373C in Kv2.1 was unaffected by MTSEA (42) whereas Thr 539 in ECaC-TRPV5 was significantly inhibited by MTSEA, it is suggested that the pore region is wider in ECaC-TRPV5 than in K ϩ channels. Definite answers to these questions will await the x-ray crystal structure of TRPV channels.