Molecular determinants of permeation through the cation channel TRPV4.

We have studied the molecular determinants of ion permeation through the TRPV4 channel (VRL-2, TRP12, VR-OAC, and OTRPC4). TRPV4 is characterized by both inward and outward rectification, voltage-dependent block by Ruthenium Red, a moderate selectivity for divalent versus monovalent cations, and an Eisenman IV permeability sequence. We identify two aspartate residues, Asp(672) and Asp(682), as important determinants of the Ca(2+) sensitivity of the TRPV4 pore. Neutralization of either aspartate to alanine caused a moderate reduction of the relative permeability for divalent cations and of the degree of outward rectification. Neutralizing both aspartates simultaneously caused a much stronger reduction of Ca(2+) permeability and channel rectification and additionally altered the permeability order for monovalent cations toward Eisenman sequence II or I. Moreover, neutralizing Asp(682) but not Asp(672) strongly reduces the affinity of the channel for Ruthenium Red. Mutations to Met(680), which is located at the center of a putative selectivity filter, strongly reduced whole cell current amplitude and impaired Ca(2+) permeation. In contrast, neutralizing the only positively charged residue in the putative pore region, Lys(675), had no obvious effects on the properties of the TRPV4 channel pore. Our findings delineate the pore region of TRPV4 and give a first insight into the possible architecture of its permeation pathway.

The TRPV subfamily of transient receptor potential (TRP) 1 proteins homologous to the vanilloid receptor consists of at least five mammalian Ca 2ϩ -permeable cation channels, which are activated by a variety of signals including noxious chemical and thermal stimuli (TRPV1 (VR1) and TRPV2 (VRL-1)), increased cell volume (TRPV4 (VR-OAC, OTRPC4, and TRP12)), or decreased intracellular Ca 2ϩ (TRPV5 (ECaC1) and TRPV6 (ECaC2)) (Ref. 1; for a unifying nomenclature see Ref. 2). TRPV channels typically contain three to six ankyrin repeats in the N terminus, six transmembrane segments with a putative pore region between transmembrane segments 5 and 6 but lack the so-called TRP motif (1,2). Recent findings indicate that TRPV4, which was originally identified as an osmotically activated channel (3)(4)(5), is not only activated by mechanical stimuli but also by ligands such as phorbol derivatives (6). Therefore, the functional importance of this channel may be connected to its role as a promiscuous Ca 2ϩ influx channel integrating multiple physical and chemical stimuli.
Based on their pore properties, the TRPV subfamily can be subdivided into two groups. TRPV5 and TRPV6, on the one hand, are highly Ca 2ϩ selective channels displaying P Ca /P Na values of more than 100 and a monovalent cation permeability sequence corresponding to a strong field binding site (Eisenman X or XI) (7)(8)(9)(10). It has been demonstrated that their high Ca 2ϩ selectivity crucially depends on a single negatively charged aspartate residue in the pore region (11). On the other hand, TRPV1, TRPV2, and TRPV4 are only weakly Ca 2ϩ -selective with P Ca /P Na values below 10 (4 -6, [12][13][14], and their monovalent cation permeability sequences are more indicative of a weak field strength binding site (6,(12)(13)(14). The precise molecular basis for these divergent pore properties is not yet fully understood. In the present study, we were led by a sequence comparison of the pore region of the TRPVs to search for possible molecular determinants for the pore features of TRPV4 and identify several residues that contribute to the permeability profile and blocker sensitivity of the channel.

MATERIALS AND METHODS
Cell Culture and Molecular Biology-We used the recombinant bicistronic expression plasmid pdiTRP12, which carries the entire protein-coding region for murine TRPV4 (mouse mTRP12; accession number CAC20703) and for green fluorescent protein coupled by an IRES sequence. Human embryonic kidney cells, HEK293, were grown in Dulbecco's modified Eagle's medium containing 10% (v/v) human serum, 2 mM L-glutamine, 2 units/ml penicillin, and 2 mg/ml streptomycin at 37°C in a humidity controlled incubator with 10% CO 2 . HEK293 cells were transiently transfected with the above-described vector using methods described previously (9). Positively transfected cells were visually identified based on their green fluorescence. Nontransfected cells from the same batch were used as controls. Mutations to the TRPV4 pore were made using the QuikChange TM site-directed mutagenesis kit (Stratagene). The nucleotide sequences of the mutants have been verified by sequencing of the corresponding cDNAs.
Solutions-The standard extracellular solution contained 150 mM NaCl, 1 mM MgCl 2 , 5 mM CaCl 2 , 10 mM glucose, 10 mM HEPES, buffered at pH 7.4 with NaOH. The osmolality of this solution, as measured with a vapor pressure osmometer (Wescor 5500, Schlag, Gladbach, Germany), was 320 Ϯ 5 mosmol. When indicated in the figure legends, the Ca 2ϩ concentration of this solution was varied between 0 and 30 mM. To study the relative permeability of mono-and divalent cations, we used extracellular solutions containing 1 mM MgCl 2 , 10 mM glucose, 10 mM HEPES, and either 150 mM XCl (where X ϭ sodium, lithium, cesium, potassium, or rubidium) or 30 mM XCl 2 (X ϭ calcium or magnesium) and 120 N-methyl-D-glucamine chloride. These solutions were titrated to pH 7.4 with the appropriate base. Two different intracellular solutions were used yielding virtually identical results after correcting for the liquid junction potential (see below). The cesium-based solution contained 20 mM CsCl, 100 mM cesium aspartate, 1 mM MgCl 2 , 4 mM Na 2 ATP, 0.022 mM CaCl 2 , 10 mM BAPTA, 10 mM HEPES, pH adjusted to 7.2 with CsOH. The sodium-based solution contained 150 mM NaCl, 1 mM MgCl 2 , 4 mM Na 2 ATP, 0.037 mM CaCl 2 , 5 mM EGTA, 10 mM HEPES, pH adjusted to 7.2 with NaOH. Free intracellular Ca 2ϩ was calculated to be ϳ1 nM for both solutions. The specific TRPV4 activator 4␣-phorbol 12,13-didecanoate (4␣-PDD; Sigma) (6), a non-protein kinase C-activating phorbol ester, was applied to the extracellular solution at a concentration of 1 M from 30 mM stock solutions in ethanol.
Electrophysiological Recordings-Whole cell membrane currents were monitored with an EPC-9 (HEKA Elektronik, Lambrecht, Germany; 8-Pole Bessel filter, 2.9 kHz) using ruptured patches. Patch electrodes had a DC resistance between 2 and 4 M⍀ when filled with intracellular solution. An Ag-AgCl wire was used as a reference electrode. Capacitance and access resistance were monitored continuously. Between 50 and 70% of the series resistance was electronically compensated to minimize voltage errors. Unless mentioned otherwise, we have applied a ramp protocol consisting of a 20-ms voltage step to Ϫ100 mV followed by a 380-ms linear ramp to ϩ100 mV. This protocol was repeated every 5 s. The currents were sampled at 1.25 kHz. The time course of the whole cell current and current densities were obtained by averaging the current in a narrow window around Ϫ80 mV during the voltage ramp protocol. All of the experiments were performed at room temperature (20 -23°C).
Calculation of the Relative Permeability of Mono-and Divalent Cations-The relative permeability of monovalent cations was calculated from the shift in reversal potential after complete substitution of extracellular Na ϩ by the specific cation, according to the following equation.
where ⌬V rev is the measured shift in reversal potential. Permeability of the divalent cations Ca 2ϩ and Mg 2ϩ relative to Na ϩ was calculated from the absolute reversal potential measured with 30 mM of the respective cation in the extracellular solution, according to the following equation.
where P X represents the permeability of the divalent cation, [X] e represents its extracellular concentration, ␣ is P Cs /P Na obtained from Equation 1, [Na ϩ ] e , [Na ϩ ] i , [Cs ϩ ] e , and [Cs ϩ ] i are the extra-and intracellular concentrations for Na ϩ and Cs ϩ , respectively, and V rev represents the reversal potential (9 -11). Before calculation of the relative permeabilities, all reversal potentials were corrected for liquid junction potentials (15).
where V LJ , the liquid junction potential, was calculated according to Barry (16). Note that the current-voltage relations in Figs. 2-5 are not corrected for liquid junction potential. Statistical Analysis-The data are expressed as the means Ϯ S.E. Overall statistical significance was determined by analysis of variance. In case of significance (p Ͻ 0.01), individual groups were compared using Student's t test. Fig. 1 shows an alignment of the putative pore region of murine TRPV4 (residues 663-686) with the corresponding regions of the other members of the TRPV subfamily and of the bacterial potassium channel KcsA. In this region, TRPV4 has the highest homology with TRPV1 and TRPV2, with which it shares a central stretch of 11 highly conserved amino acids. There is also significant homology with TRPV5 and TRPV6 and, to a lesser extent, with the pore region of KcsA. When comparing TRPV1/2/4 with the highly Ca 2ϩ -selective TRPV5/TRPV6 pores two differences are immediately obvious. Firstly, the TRPV1/ 2/4 pores contain a positively charged lysine (Lys 675 in TRPV4), which is not present in TRPV5/6. Secondly, TRPV1/2/4 contain a methionine (Met 680 in TRPV4) at the position corresponding to Asp 542 in TRPV5 (Asp 541 in TRPV6), which is crucial for the high Ca 2ϩ selectivity of the latter channel. Moreover, TRPV4 contains two negatively charged residues within the conserved stretch of 11 amino acids: Asp 672 , which is conserved among all TRPVs, and Asp 682 , which is not present in TRPV5/6 and corresponds to Asp 646 in TRPV1, a residue involved in Ruthenium Red block and Mg 2ϩ permeability. To assess the contribution of these TRPV4 pore residues to the pore properties of the channel, we constructed seven cDNAs coding for TRPV4 channels in which point mutations were introduced into the pore. Charged residues were neutralized (mutants D672A, D682A, and K675A and double mutant D572A/D682A), the pore methionine residue was substituted by an alanine (mutant M680A), and, to mimic the TRPV5/6 pore, a negative charge was introduced, either alone or in combination with a deletion of the positively charged Lys 675 (mutants M680D and M680D/⌬K675). As shown in Fig. 2, all seven mutants expressed as functional ion channels, as indicated by the activation of a cation current upon stimulation of transfected HEK293 cells with the specific TRPV4 activator 4␣-PDD. When compared with wild type TRPV4, the current densities were significantly larger for the D572A, D582A, and D572A/D582A mutants, whereas mutations to Met 680 led to a reduction in current density. These differences in current density could arise from alterations in plasma membrane expression, singlechannel conductance, or sensitivity toward activation by 4␣-PDD. This latter possibility would not be unprecedented, because gating of TRPV1 by extracellular protons critically depends on a negatively charged residue (Glu 646 ) within the TRPV1 pore region (17), but more work is needed to elucidate a possible involvement of pore residues in activation of TRPV4. In this manuscript, we focus on the contribution of these residues to typical pore properties: permeability, rectification, and voltage-dependent block.

Pore Mutants of TRPV4 Yield Functional Channels-
Effects of Pore Mutations on Permeability to Mono-and Divalent Cations-It has been previously shown that TRPV4 is moderately selective for Ca 2ϩ over Na ϩ ions and displays a monovalent cation permeability sequence corresponding to Eisenman IV. We investigated whether these properties are affected by mutations to the putative pore region by determining the relative cation permeabilities for all mutant channels. After full activation of the current with 4␣-PDD, we switched between extracellular solutions containing a single permeant cation species and calculated relative permeabilities from the shifts in the reversal potential after correcting for liquid junction potentials (see "Materials and Methods"). For wild type TRPV4, switching from a Na ϩ -containing extracellular solution to solutions containing 30 mM Ca 2ϩ or Mg 2ϩ as the sole permeant cation caused a decrease of the inward current at negative potentials and a rightward shift of the reversal potential (Fig. 3A). Relative permeabilities calculated according to Equation 3 were 6.85 Ϯ 0.55 (n ϭ 25) and 2.52 Ϯ 0.29 (n ϭ 7) for Ca 2ϩ and Mg 2ϩ , respectively. Substituting extracellular Na ϩ by other monovalent cations caused only small shifts in the reversal potential (Fig. 4A), confirming that TRPV4 poorly discriminates between monovalent cations. The deduced monovalent cation permeability sequence was K ϩ Ͼ Cs ϩ Ն Rb ϩ Ͼ Na ϩ Ͼ Li ϩ , which corresponds to Eisenman sequence IV for a weak field strength site.
A possible explanation for the relatively low Ca 2ϩ selectivity and the weak field strength of the TRPV4, TRPV1, and TRPV2 pores could be the presence of a positively charged lysine (Lys 675 in TRPV4) in the central part of the putative pore region, which is not present in the strong field strength site TRPV5 and TRPV6 pores. However, neutralizing this lysine into alanine (mutant K675A) had no significant effect on the divalent selectivity of the channel (Fig. 3B). Likewise, the K675A pore still poorly discriminates between monovalent cations (Fig. 4B) and retains the Eisenman IV permeation profile. Thus, the low field strength binding site is conserved, indicating that Lys 675 does not significantly influence the electrical field within the TRPV4 channel pore.
The two negatively charged residues in the pore region,  Table I. G and H, leftward shift of the reversal potential for the M680D and M680D/⌬K675 mutants.
Asp 672 and Asp 682 , could potentially function as (part of) Ca 2ϩbinding sites within the TRPV4 pore, similar to the role played by aspartate and glutamate residues in the Ca 2ϩ selective pores of TRPV5, TRPV6, and voltage-gated Ca 2ϩ channels. Consistent with such a role, we found that the D672A and D682A mutants both displayed a reduced Ca 2ϩ selectivity, as can be judged from the less positive reversal potentials upon switching to a Ca 2ϩ -containing extracellular solution (Fig. 3, C  and D). The relative Ca 2ϩ permeability (P Ca /P Na ) was approximately halved in both mutants, to 3.20 Ϯ 0.70 (n ϭ 4; p Ͻ 0.05) and 4.09 Ϯ 0.22 (n ϭ 5; p Ͻ 0.05) for D672A and D682A, respectively (Table I). Likewise, inward current densities with 30 mM Ca 2ϩ as the sole charge carrier were significantly smaller for both mutants than for wild type TRPV4 despite the significantly larger current densities in Na ϩ -containing solutions (Fig. 2), indicating that Ca 2ϩ conduction is hindered when these pore aspartates are neutralized. Like wild type TRPV4, the D672A and D682A pores discriminate poorly between monovalent cations and display the Eisenman IV permeation profile ( Fig. 4C and data not shown). However, we found that both channels display a significantly higher relative permeability for K ϩ ions, with P K /P Na values of 1.63 Ϯ 0.10 (D672A, n ϭ 5; p Ͻ 0.05) and 1.58 Ϯ 0.07 (D682A, n ϭ 6; p Ͻ 0.05) compared with 1.35 Ϯ 0.03 (n ϭ 10) for wild type TRPV4 (Table I).
Simultaneous neutralization of both aspartates (double mutant D672A/D682A) induced some more drastic changes in permeation phenotype. The relative Ca 2ϩ permeability was even more reduced than in the single mutants ( Fig. 3E; P Ca / P Na ϭ 2.45 Ϯ 0.19, n ϭ 12; p Ͻ 0.01), and there was no longer a measurable permeability for Mg 2ϩ . Moreover, the permeability sequence for monovalent cations was different from that of wild type, with both Cs ϩ and Rb ϩ being more permeant than K ϩ in 7 of 8 cells (Fig. 4D). The observed Cs ϩ Ϸ Rb ϩ Ͼ K ϩ Ͼ Na ϩ Ͼ Li ϩ permeability sequence is consistent with a reduction of the field strength of the cation-binding site to Eisenman sequence II or I (Table I). We also observed important differences in the rectification properties between wild type TRPV4 and the D672A/D682A mutant, which, as will be discussed below, arise from differences in Ca 2ϩ block.
From the sequence comparison in Fig. 1, two lines of evidence indicated that Met 680 is positioned at a potentially crucial position within the putative pore region of TRPV4. Firstly, at the corresponding position in TRPV5 we found an aspartate (residue 542), which is crucial for the high Ca 2ϩ selectivity of this channel. Secondly, at the corresponding position in voltage-gated K ϩ channels there is a tyrosine, part of the so-called K ϩ channel "signature sequence" (T(V/I)GYG), whose backbone atoms contribute to the lining of the K ϩ selectivity filter. Mutants in which Met 680 was substituted by alanine or aspartate gave rise to sizable but strongly reduced current densities upon stimulation with 4␣-PDD (Fig. 2), which may have some quantitative bearings on the determination of relative permeabilities of the M680A, M680D, and M680D/⌬K675 mutants. Still, some drastic changes in divalent permeability of these mutants were obvious. In the M680A mutant, switching from a Na ϩcontaining extracellular solution a solution containing 30 mM Ca 2ϩ as the sole permeant cation caused a rightward shift of the reversal potential (Fig. 3F), corresponding to a P Ca /P Na value of 3.26 Ϯ 0.09 (n ϭ 6; p Ͻ 0.05). Under the same conditions, we observed a leftward shift of the reversal potential for the M680D and M680D/⌬K675 mutants (Fig. 3, G and H), indicating that these pore mutations abolish the Ca 2ϩ selectivity of the TRPV4 pore (P Ca /P Na values Ͻ Ͻ1; p Ͻ 0.01). Compared with wild type TRPV4, no changes in the monovalent permeability sequence (Eisenman IV) were observed (data not shown), but all three mutants had a significantly higher relative permeability for K ϩ ( Table I).
Effects of Pore Mutations on the TRPV4 Rectification Properties-The current-voltage relation of wild type TRPV4 in extracellular solutions containing "physiological" concentrations of mono-and divalent cations displays both inward and outward rectification. As illustrated in Fig. 5A, this behavior strongly depends on the extracellular Ca 2ϩ concentration. In the absence of divalent cations, the current-voltage relation is linear with identical current amplitudes at ϩ100 and Ϫ100 mV. Increasing concentrations of Ca 2ϩ cause an inhibition of the current that is more pronounced at negative potentials, characteristic of a voltage-dependent block. However, Ca 2ϩ concentrations up to 30 mM do not cause a complete block of the inward current, because Ca 2ϩ itself permeates the channel, as evidenced by the rightward shift of the reversal potential. To quantify the effect of extracellular Ca 2ϩ we normalized the inward current at Ϫ100 mV to the outward current at ϩ100 mV, assuming that the Ca 2ϩ block disappears at highly positive potentials (Fig. 5D). For wild type TRPV4, the I Ϫ100 mV / I ϩ100 mV ratio decreased from ϳ1 in nominally Ca 2ϩ -free solution to ϳ0.25 in the presence of 30 mM Ca 2ϩ . Virtually identical results were obtained for the K675A mutant, again indicating that this residue is not an important determinant of TRPV4 pore properties (Fig. 5D and data not shown). In contrast, neutralization of Asp 672 and Asp 682 strongly reduced the sensitivity of the channel to extracellular Ca 2ϩ . In the D672A/ D682A double mutant, the current-voltage relation was still completely linear with 1 mM Ca 2ϩ in the extracellular solution ( Fig. 5B), and outward rectification remained reduced for Ca 2ϩ concentrations up to 30 mM (Fig. 5, B and D). The Ca 2ϩ -dependent rectification properties of the single mutants D672A and D682A were intermediate between wild type and the D672A/D682A double mutant (Fig. 5D), which can also be appreciated from the normalized current-voltage relations measured with 5 mM extracellular Ca 2ϩ (Fig. 5C). These results indicate that both aspartate residues contribute to the Ca 2ϩ sensitivity of the TRPV4 pore.
Effects of Pore Mutations on Ruthenium Red Sensitivity-Ruthenium Red (RR) is a polycationic molecule that has been shown to inhibit members of the TRPV family with nanomolar to low micromolar affinity. As illustrated in Fig. 6 (A and B), extracellular RR is a potent, voltage-dependent blocker of wild type TRPV4. At a concentration of 1 M, extracellular RR completely inhibited inward currents, whereas significant outward currents were measured at positive potentials more positive than ϩ20 mV (Fig. 6, A and B). The apparent fraction of unblocked channels, measured from tail currents, increased from Ͻ0.05 at Ϫ100 mV to ϳ0.8 at ϩ80 mV, with 50% block at ϩ60.8 Ϯ 0.8 mV (n ϭ 4; Fig. 6E). The rate of RR block was relatively fast, with a time constant for block of 0.9 Ϯ 0.1 ms at Ϫ100 mV (n ϭ 4). We also performed experiments in which a high concentration (50 M) of RR was included in the pipette solution. Under this condition, we did not observe inhibition of the TRPV4 current, whereas a subsequent addition of 1 M RR to the extracellular medium resulted in the expected voltagedependent block. These data establish that RR blocks TRPV4 from the extracellular side by binding in a voltage-dependent manner to a site in the channel pore, within the transmembrane electrical field. Virtually identical results were obtained for the D572A and K680A mutants ( Fig. 6E and data not shown). In contrast, the D682A and D672A/D862A mutants were much less sensitive to block by extracellular RR (Fig. 6, C and D, and data not shown), and the voltage dependence of the block was shifted to more negative potentials (Fig. 6E). For both mutants, the apparent fraction of unblocked channels increased from ϳ0.15 at Ϫ100 mV to ϳ1 at potentials Ͼϩ40 mV (Fig. 6E), with 50% block at Ϫ21.0 Ϯ 1.4 mV (n ϭ 4) and Ϫ23.5 Ϯ 2.5 mV (n ϭ 3) for D672A/D682A and D682A, respectively. The rate of RR block was also much slower for these mutants, with time constants of 25.4 Ϯ 1.3 ms (n ϭ 4, D672A/ D682A) and 24.7 Ϯ 2.5 ms (n ϭ 3, D682A). Thus, the negative side chain of Asp 682 appears to be crucial for the high sensitivity of TRPV4 for RR, similar to Asp 646 in the pore of TRPV1 (18). DISCUSSION In recent years, the persistent discovery of new members within the TRP superfamily of cation channels has greatly advanced our understanding of Ca 2ϩ influx mechanisms in many cell types. However, it is not always clear whether the ionic currents that are observed after expression of members of the TRP family are directly mediated by the TRP proteins themselves or whether they merely act as modulators of endogenous channels. The situation is further complicated by the findings that TRP proteins can form heteromultimers (19) and/or reside in intracellular membranes (20). The analysis of channel permeation properties and modulation of these properties by mutations to pore residues form the ultimate proof of the channel nature of these proteins. At present, this goal has only been achieved for TRPV1 and TRPV5 (11,18,21). The present study adds another member of the TRPV family to this list, the osmosensitive and ligand-gated TRPV4. We have demonstrated that two negatively charged aspartate residues and TABLE I Relative permeabilities of wild type and mutant TRPV4 channels The relative permeabilities were calculated as described under "Materials and Methods." The data are given as the means Ϯ S.E. from at least three cells.
Channel P Ca /P Na P Mg /P Na P K /P Na P Rb /P Na P Cs /P Na P Li /P Na Eisenman a  the non-polar amino acid methionine are critical determinants of the pore properties of TRPV4.
In the presence of both monovalent and divalent cations in the extracellular solution, the current-voltage relation of TRPV4 displays both inward and outward rectification, a feature it shares with TRPV1, TRPV2, and several members of the TRPM and TRPC families (22). Our present results show that, at least for TRPV4, this behavior depends on extracellular Ca 2ϩ ions, which both block the channel in a voltage-dependent manner and permeate the channel. We found that two negatively charged residues, Asp 672 and Asp 682 , contribute to the channel's Ca 2ϩ sensitivity, as evidenced by the reduced Ca 2ϩdependent rectification and ϳ2-fold lower relative Ca 2ϩ permeability in both the D672A and D682A mutants. The D672A/ D682A double mutant displays an even stronger reduction in Ca 2ϩ -dependent rectification and Ca 2ϩ permeability, indicating that the effect of neutralizing these aspartates on the Ca 2ϩ sensitivity of TRPV4 is additive. Moreover, this double mutant has a very low selectivity for monovalent cations and a Cs ϩ Ϸ Rb ϩ Ͼ K ϩ Ͼ Na ϩ Ͼ Li ϩ permeability sequence, indicative of a very weak field binding site (Eisenman I or II). These results seem to indicate that these aspartates are fully equivalent and possibly contribute to a single negatively charged binding site for cations within the channel pore. However, both aspartates contribute differently to the voltage-dependent block of TRPV4 by the large polycation RR; the D672A mutant is blocked with the same affinity and voltage dependence as wild type by TRPV4, whereas block of the D682A or D672A/D682A mutants is strongly impaired. The potential for half-maximal block of these latter mutants by 1 M RR is shifted to the left by ϳ80 mV, and the rate of block at Ϫ100 mV is ϳ25 times slower. Taken together, these results indicate that both aspartates contribute to the lining of the TRPV4 pore but differ in their accessibility to extracellular cations. Asp 682 is positioned close to the outer mouth of the pore, where it can interact with both Ca 2ϩ and RR. In contrast, Asp 672 is likely to be located more toward the cytoplasmic site of the channel, because it can interact with the permeant cation Ca 2ϩ but not with the bulky, impermeant cation RR.
Our results also indicate that Met 680 is crucial for the proper functioning of the TRPV4 pore. Substituting this residue with an alanine results in a decreased Ca 2ϩ permeability and overall current amplitude. Introducing a negative charge at this position (mutant M680D) causes a further reduction of the current density and a complete loss of Ca 2ϩ permeation. One possible explanation is that this additional negative charge enhances the affinity of the Ca 2ϩ -binding site to such an extent that it can no longer permeate the channel.
We had hypothesized that neutralization of the positively charged lysine residue Lys 675 could be involved in determining the field strength of the putative cation-binding site within the pore. This hypothesis originated from the fact that this residue is not present in the highly Ca 2ϩ -selective channels TRPV5 and 6. However, permeation properties for monovalent cations are not altered, and the relative Ca 2ϩ permeability remains unchanged in the K675A mutant. Likewise, the two other mutants that were constructed to transfer the high Ca 2ϩ selectivity and strong field binding site of TRPV5/6 onto TRPV4 (M680D and M680D/⌬K675) did not give the expected result and yielded channels with strongly reduced Ca 2ϩ selectivity. It thus appears that the differences in pore properties between TRPV1/2/4 and TRPV5/6 cannot be reduced to single amino acid substitutions within the pore region.
As shown in Fig. 1, the pore region of TRPV4 shows significant homology with the pore region of KcsA, the bacterial K ϩ channel whose crystal structure has been determined at 2.0 Å resolution (23,24). This raises the possibility that both channels share a similar pore architecture. The results from our mutagenesis study are at least consistent with this hypothesis. Firstly, as discussed above, we concluded that Asp 682 is located at the extracellular part of the TRPV4 pore where it can interact with RR, whereas Asp 672 seems to be located more toward the cytoplasm. Indeed, in the KcsA structure, Asp 80 (corresponding to Asp 682 in TRPV4) is located at the outer mouth of the selectivity filter, whereas Glu 71 (corresponding to Asp 672 ) is directed more toward the intracellular side of the channel. Secondly, we found that Met 680 is crucial for proper channel function, because mutating it to an alanine or aspartate strongly affects the permeability for monovalent and divalent cations. In the KcsA structure, the corresponding residue, Tyr 78 , is at the center of the selectivity filter, where its backbone oxygens interact with dehydrated K ϩ ions (23,24). Finally, we concluded that the positively charged Lys 675 does not significantly influence the pore properties of TRPV4. Interestingly, KcsA, like TRPV5/6, lacks a positive charge at the corresponding residue. However, if the structure homology between KcsA and TRPV4 holds true, Lys 675 is expected to be part of the so-called pore helix (24), the residues of which do not directly contribute to the lining of the channel pore.
In conclusion, we have unambiguously shown that TRPV4 is a channel-forming protein that responds to mutations in its putative pore region with changes in ion permeability and block. Two negatively charged aspartates and a non-polar methionine residue were identified as critical determinants of the pore properties of the channel. Moreover, our results are compatible with the hypothesis that TRPV channels have the same FIG. 6. Neutralization of Asp 682 reduces TRPV4 block by Ruthenium Red. A-D, currents through wild type (WT) TRPV4 (A and B) and the D672A/ D682A mutant (C and D) in the absence (A and C) and presence (B and D) of 1 M RR in the extracellular solution in response to the voltage protocol shown below. The extracellular solution contained 150 mM Na ϩ and 5 mM Ca 2ϩ . E, voltage dependence of the block by 1 M RR for wild type TRPV4 and three mutants. The unblocked fraction in the presence of RR was obtained by measuring peak tail currents during the second step to Ϫ100 mV and normalizing to the current in the absence of the blocker. basic pore architecture as K ϩ channels, although further work, including a determination of the multimeric structure of the channel, may be needed to substantiate this notion. The present work may also form a starting point for future studies aimed at the unraveling of the exact gating mechanism or the discovery of more selective TRPV4 antagonists.