The Nociceptor Ion Channel TRPA1 Is Potentiated and Inactivated by Permeating Calcium Ions*

The transient receptor potential A1 (TRPA1) channel is the molecular target for environmental irritants and pungent chemicals, such as cinnamaldehyde and mustard oil. Extracellular Ca2+ is a key regulator of TRPA1 activity, both potentiating and subsequently inactivating it. In this report, we provide evidence that the effect of extracellular Ca2+ on these processes is indirect and can be entirely attributed to entry through TRPA1 and subsequent elevation of intracellular calcium. Specifically, we found that in a pore mutant of TRPA1, D918A, in which Ca2+ permeability was greatly reduced, extracellular Ca2+ produced neither potentiation nor inactivation. Both processes were restored by reducing intracellular Ca2+ buffering, which allowed intracellular Ca2+ levels to become elevated upon entry through D918A channels. Application of Ca2+ to the cytosolic face of excised patches was sufficient to produce both potentiation and inactivation of TRPA1 channels. Moreover, in whole cell recordings, elevation of intracellular Ca2+ by UV uncaging of 1-(4,5-dimethoxy-2-nitrophenyl)-EDTA-potentiated TRPA1 currents. In addition, our data show that potentiation and inactivation are independent processes. TRPA1 currents could be inactivated by Mg2+, Ba2+, and Ca2+ but potentiated only by Ba2+ and Ca2+. Saturating activation by cinnamaldehyde or mustard oil occluded potentiation but did not interfere with inactivation. Last, neither process was affected by mutation of a putative intracellular Ca2+-binding EF-hand motif. In conclusion, we have further clarified the mechanisms of potentiation and inactivation of TRPA1 using the D918A pore mutant, an important tool for investigating the contribution of Ca2+ influx through TRPA1 to nociceptive signaling.

The transient receptor potential A1 (TRPA1) channel is the molecular target for environmental irritants and pungent chemicals, such as cinnamaldehyde and mustard oil. Extracellular Ca 2؉ is a key regulator of TRPA1 activity, both potentiating and subsequently inactivating it. In this report, we provide evidence that the effect of extracellular Ca 2؉ on these processes is indirect and can be entirely attributed to entry through TRPA1 and subsequent elevation of intracellular calcium. Specifically, we found that in a pore mutant of TRPA1, D918A, in which Ca 2؉ permeability was greatly reduced, extracellular Ca 2؉ produced neither potentiation nor inactivation. Both processes were restored by reducing intracellular Ca 2؉ buffering, which allowed intracellular Ca 2؉ levels to become elevated upon entry through D918A channels. Application of Ca 2؉ to the cytosolic face of excised patches was sufficient to produce both potentiation and inactivation of TRPA1 channels. Moreover, in whole cell recordings, elevation of intracellular Ca 2؉ by UV uncaging of 1-(4,5-dimethoxy-2-nitrophenyl)-EDTA-potentiated TRPA1 currents. In addition, our data show that potentiation and inactivation are independent processes. TRPA1 currents could be inactivated by Mg 2؉ , Ba 2؉ , and Ca 2؉ but potentiated only by Ba 2؉ and Ca 2؉ . Saturating activation by cinnamaldehyde or mustard oil occluded potentiation but did not interfere with inactivation. Last, neither process was affected by mutation of a putative intracellular Ca 2؉ -binding EF-hand motif. In conclusion, we have further clarified the mechanisms of potentiation and inactivation of TRPA1 using the D918A pore mutant, an important tool for investigating the contribution of Ca 2؉ influx through TRPA1 to nociceptive signaling.
Members of the transient receptor potential (TRP) 2 family of ion channels that are expressed by sensory neurons in dorsal root and trigeminal ganglia serve as sensors for temperature and noxious stimuli (1,2). Of these, TRPA1 is one of the key sensors for pungent chemicals and environmental irritants and is essential for behavioral responses of mice to conditions that evoke inflammatory pain (3)(4)(5)(6)(7). Inflammatory mediators, such as bradykinin, bind to G protein-coupled receptors on nociceptors, initiating a second messenger signaling cascade that leads to Ca 2ϩ influx mediated in part by the opening of Ca 2ϩ -permeable TRPA1 channels (5,8,9). TRPA1 is also activated directly by a wide range of chemicals that cause painful sensations, including food additives, such as mustard oil (MO), cinnamaldehyde (Cin), onion, raw garlic, and thyme; environmental irritants, such as formaldehyde and acrolein (a component of automobile exhaust); and products of oxidative stress (4, 8, 10 -16). Many of these chemicals activate TRPA1 by binding covalently to reactive cysteine residues in the amino terminus of the protein (17,18), producing a modification of the channel that can last for more than 1 h and which leads to persistent activation of TRPA1 currents (18,19).
Ca 2ϩ plays at least two roles in regulating the activity of TRPA1 channels. TRPA1 currents that are activated by pungent chemicals are rapidly potentiated in the presence of extracellular Ca 2ϩ , an effect that may be mediated by Ca 2 entry through TRPA1 (4,20,21). This mechanism is attractive, since it could also account for the observed activation of TRPA1 by some inflammatory mediators (5,8). For example, bradykinin, acting through its cognate receptor, stimulates a phospholipase C-based signaling pathway, leading to elevation of intracellular Ca 2ϩ , and it has been proposed that Ca 2ϩ is the proximate stimulus that gates TRPA1 (4, 5) (but see also Refs. 8 and 22). Activation of TRPA1 by intracellular Ca 2ϩ has been observed in experiments where Ca 2ϩ is dialyzed into cells or applied to excised patches (20,21,23), although it is not known whether this is the mechanism for Ca 2ϩ potentiation or receptor activation of TRPA1. The site for Ca 2ϩ activation of TRPA1 has been proposed to be an EF-hand present in the N terminus of the protein (20,21).
A second and equally important effect of extracellular Ca 2ϩ is to inactivate TRPA1 (e.g. see Refs. 4,19,20,and 24, but see also Ref. 19). TRPA1 currents activated by pungent chemicals in the absence of extracellular Ca 2ϩ , decay within seconds when extracellular Ca 2ϩ is introduced (24). Inactivation may be mediated by binding of Ca 2ϩ to the outside of the channel or could be mediated by elevation of intracellular Ca 2ϩ , which is known to cause desensitization of other TRP channels (25)(26)(27)(28)(29)(30)(31). Alternatively, inactivation of TRPA1 might simply reflect the stochastic and obligatory entry of the channel into the inacti-vated state, following Ca 2ϩ -dependent activation, similar to inactivation of voltage-gated ion channels (32).
By studying mutant TRPA1 channels in which Ca 2ϩ permeability is reduced, we show that both the potentiating and inactivating effects of extracellular Ca 2ϩ can be attributed to entry of Ca 2ϩ through TRPA1 and the subsequent elevation of intracellular Ca 2ϩ . We also provide evidence that potentiation and inactivation are mediated by independent mechanisms, each with a distinct sensitivity and specificity for divalent cations.

EXPERIMENTAL PROCEDURES
cDNAs and Expression in HEK-293 Cells-For most experiments, we used an N-terminal YFP fusion of rat TRPA1 generated by inserting a PCR fragment of YFP into an AgeI site at the 5Ј-end of a rat TRPA cDNA. To confirm that the YFP fusion did not alter function, in some experiments we used an unaltered clone of rTRPA1, which was co-transfected with green fluorescent protein (20:1). To assess whether some of our results could be attributed to species differences between rat and human variants of TRPA1, in some experiments, we used hTRPA1, co-expressed at a ratio of 20:1 with green fluorescent protein. Point mutations were generated by QuikChange mutagenesis (Stratagene, La Jolla, CA) and were verified by sequencing (Retrogen (San Diego, CA) or Macrogen (Rockville, MD)). All constructs were transiently transfected into HEK-293 cells using TransIT-LT1 (Mirus Bio Corp., Madison, WI), as suggested by the manufacturer, and cells were cultured at 37°C. Prior to recording, cells were treated with 0.05% trypsin and 4 mM EDTA in phosphate-buffered saline for 5 min at 37°C and replated in the recording chamber with Tyrode's solution. Transfected cells were identified under epifluorescence. Recordings were performed ϳ24 -48 h after transfection at room temperature.
Patch Clamp Recording-Patch clamp electrophysiology was performed as previously described (27,33). In brief, recordings were made with an Axopatch 200B amplifier, sampled at 5 kHz, and filtered at 1 kHz. Data were digitized with a Digidata 1322a digitizer, acquired with pClamp 8.2, and analyzed with Clampfit 8.2 (Axon Instruments, Union City, CA). Representative data shown in figures were exported into Origin (Microcal, Northampton, MA) and CorelDRAW (Corel Corp., Eden Prairie, MN). Pipettes were fabricated from borosilicate glass, fire-polished to a resistance of 2-3 megaohms. For whole-cell recordings, the membrane potential was ramped from Ϫ80 mV to ϩ80 mV (1 V/s). Excised patch recording was performed as previously described (27). Briefly, following formation of a gigaohm seal, the patch was excised into Ca 2ϩ -free solution and rapidly moved in front of a linear array of microperfusion pipes (Warner Instruments, Hamden, CT).
Flash Photolysis of Caged Compound-All experiments were performed on an Olympus IX71 microscope. To allow uniform dialysis of the cell with the caged compounds, we waited at least 100 s after breaking into whole cell recording mode before uncaging. UV light from a mercury arc lamp that was then passed through a ϫ350/50 band pass filter (Chroma Technology Corp., Rockingham, VT) was controlled by a Uniblitz shutter (Vincent Associates, Rochester, NY). The shutter was opened for 100 ms to uncage Ca 2ϩ . For experiments in which Ca 2ϩ levels were measured, UV light from a xenon arc lamp was passed through a band pass filter (380xv1; Chroma Technology) that was part of a Ludl Electronic Products (Hawthorne, NY) filter wheel/shutter operated under computer control (Simple PCI; Compix Corp., Sewickley, PA). The shutter was opened for 1 s to uncage Ca 2ϩ .
Ca 2ϩ Imaging-Ca 2ϩ imaging was performed as previously described (33). Images were acquired on an Olympus IX71 microscope equipped with a Ludl Electronic Products filter wheel and ORCA ER camera (Hamamatsu Photonics) and were analyzed with simple PCI (Compix Corp.). The low affinity Ca 2ϩ indicator Fluo-5F (10 M) or Fura-4F (20 M) was loaded through the patch pipette. The fluorescence emission from Fluo-5F was detected using a YFP filter set (Chroma Technology). The fluorescence emission from Fura-4F was detected with a Fura 2 filter set (Chroma Technology) using excitation alternately at 340 and 380 nM. The ratio of the fluorescence emission (r ϭ F 340 /F 380 ) was used to calculate the intracellular Ca 2ϩ concentration as follows, where K d is the Ca 2ϩ dissociation constant of the indicator Fura-4F (770 nM), Q is emission ratio at 380 nM recorded at 0 Ca 2ϩ and at a saturated Ca 2ϩ concentration, and R max and R min are the ratios under saturating Ca 2ϩ or 0 Ca 2ϩ conditions, respectively. R min was measured from a separate set of cells in which intracellular solution was heavily buffered by 5 mM EGTA, and the extracellular solution was Ca 2ϩ -free (buffered with 2 mM HEDTA). To obtain R max and Q, at the end of each experiment, ionomycin was added to the bath, and the emission with 2 mM Ca 2ϩ and 0 Ca 2ϩ in the bath was measured. Ion Selectivity Measurements-TRPA1 currents were activated by brief exposure to Cin, and the current in the presence of different external cations was measured by rapidly exchanging the extracellular solution using a linear array of microperfusion pipes. The internal solution contained Cs ϩ (HB-Cs ϩ ), and the ion permeability relative to Cs ϩ was calculated from the reversal potential under bi-ionic conditions according to Equations 2 and 3, where P Ca /P Cs and P X /P Cs are the Ca 2ϩ and monovalent cation permeability relative to Cs ϩ , respectively, E rev is the reversal potential, F is Faraday's constant, R is the universal gas constant, and T is the absolute temperature (34). E rev was corrected for the liquid junction potentials measured under the conditions of our experiments. Chemicals-Cinnamaldehyde, menthol, mustard oil, and polyP 3 were purchased from Sigma. Fluo-5F, Fura-4F, and ionomycin were from Invitrogen Molecular probes (Eugene, OR). HC-030031 was from ChemBridge (San Diego, CA). For stock solutions, Cin, HC-030031, mustard oil, and ionomycin were dissolved in dimethyl sulfoxide, menthol was in ethanol, and other chemicals were in water. Prior to use, the stock solution was diluted in the appropriate bath solution. The final dimethyl sulfoxide or ethanol concentration was Ͻ0.2%.

RESULTS
A Dual Effect of Extracellular Ca 2ϩ ; Potentiation and Inactivation of TRPA1-To study Ca 2ϩ regulation of TRPA1 channels, we expressed rTRPA1 fused to YFP in HEK 293 cells and recorded responses to TRPA1 agonists with patch clamp electrophysiology. Bath perfusion of Cin (100 M) for ϳ40 s in the absence of extracellular Ca 2ϩ strongly activated a nonselective cation current of 3.9 Ϯ 0.4 nA at ϩ80 mV (n ϭ 10) in TRPA1expressing cells (Fig. 1A) but not in untransfected cells. The TRPA1 current persisted following washout of the agonist (Fig.   1A), consistent with the observation that Cin binds covalently to cysteine residues in the N terminus of the channel, and therefore its binding is not readily reversible (17,18). The addition of 2 mM Ca 2ϩ to the bath caused a strong potentiation, which was followed by a rapid decay of the current, as previously noted (4, 24) (Fig. 1A). We quantitated the degree of potentiation by the -fold increase in the current after introduction of Ca 2ϩ , where 1 indicates no increase and 2 indicates a doubling of the current. Note that this analysis underestimates the full degree of potentiation, which is partly masked by the concomitant process of inactivation. In these experiments, the addition of 2 mM Ca 2ϩ potentiated the currents by 1.85 Ϯ 0.06-fold at ϩ80 mV and 3.98 Ϯ 0.50-fold at Ϫ80 mV (n ϭ 10). The greater potentiation of the currents at negative voltages reflects a decrease in the outward rectification of the currents (from 6.56 Ϯ 0.79 to 3.58 Ϯ 0.76, measured as the current at ϩ80 mV divided by the current at Ϫ80 mV), as is commonly observed during strong activation of many TRP channels (35). The currents subsequently decayed to 50% of their peak value (T50) in 4.8 Ϯ 0.5 s (measured at ϩ80 mV; n ϭ 10). Similar results were obtained with MO as the agonist, indicating that as previously noted, Ca 2ϩ potentiation and inactivation are independent of the specific agonist used to activate TRPA1 (21). Ca 2ϩ potentiation and inactivation appeared to be unaffected by fusion of YFP to the N terminus of TRPA1, since there was no difference in the degree of potentiation or rate of inactivation when an unfused rTRPA1 cDNA was used (2.52 Ϯ 0.45-fold and 4.85 Ϯ 1.29-fold potentiation at ϩ80 and Ϫ80 mV, respectively; T50 ϭ 4.7 Ϯ 0.9 s, n ϭ 3; see supplemental Fig. 1).
To confirm that the decay of TRPA1 currents upon the addition of external Ca 2ϩ represents inactivation rather than a return to the resting state, we determined whether TRPA1 currents could be reactivated following exposure to external Ca 2ϩ . For these experiments, we used a low concentration of menthol, which does not bind covalently to TRPA1 and can be used to repeatedly activate TRPA1 currents in the absence of external Ca 2ϩ (36). Following activation of TRPA1 by menthol, the addition of external Ca 2ϩ caused a rapid potentiation and subsequent decay of the current similar to that observed when the channels had been activated by Cin (Fig. 1B). Following decay of the current, no additional responses to menthol could be evoked for a period up to 20 min ( Fig. 1B; n ϭ 5). Thus, external Ca 2ϩ promotes the entry of TRPA1 channels into a long lasting inactivated state.
Complete Activation of TRPA1 by Cin Occludes Ca 2ϩ Potentiation-For both practical and theoretical reasons, it was important to understand whether extracellular Ca 2ϩ and Cin act independently or in concert to regulate gating of TRPA1. Toward this end, we determined whether TRPA1 currents could be potentiated by Ca 2ϩ when the currents had been maximally activated by Cin. Because Cin binds covalently to TRPA1, the fraction of channels activated by a fixed concentration of the agonist is expected to be sensitive to the duration of the exposure. Consistent with this prediction, we found that a prolonged exposure (156 Ϯ 17 s) to 100 M Cin in the absence of external Ca 2ϩ activated an average current of 7.7 Ϯ 1.1 nA (at ϩ80 mV, n ϭ 4), which was twice as large as the average current measured in response to an exposure of ϳ40 s (Fig. 1, C and D; see above). After exposure to Cin for ϳ120 s, the currents reached a plateau, presumably reflecting the full activation of TRPA1 channels (Fig. 1C). Interestingly, when fully activated, the currents were significantly less rectifying than when they were partially activated ( Fig. 1C; rectification ratio was 1.32 Ϯ 0.20; n ϭ 4). At this point, the addition of extracellular Ca 2ϩ only slightly potentiated the currents (Fig. 1, C and D; -fold increase in current was 1.07 Ϯ 0.04 and 1.12 Ϯ 0.05 at ϩ80 and Ϫ80 mV, respectively, n ϭ 4), whereas inactivation was slightly slowed (T50 ϭ 10.5 Ϯ 3.0 s; p Ͻ 0.05 compared with brief exposure). Therefore, Ca 2ϩ potentiation and Cin activation are not independent and may share a final common mechanism (23). Similar results were obtained for MO activation of hTRPA1 (supplemental Fig. 6). In the following experiments, we took care to only partially activate the channels with Cin or MO, so that Ca 2ϩ potentiation would not be occluded.
Potentiation and Inactivation of TRPA1 by Micromolar Concentrations of Extracellular Ca 2ϩ -The dual effects of Ca 2ϩ on TRPA1 currents could be explained by several different mech-anisms. First, inactivation might reflect an obligatory and stochastic conformational change that follows Ca 2ϩ -dependent activation, like someformsofinactivationofvoltagedependent ion channels (32). This "coupled" model predicts that potentiation and inactivation will show a similar sensitivity and specificity for Ca 2ϩ and that conditions that promote potentiation will similarly promote inactivation. Alternately, potentiation and inactivation may be mediated by "independent" Ca 2ϩ -dependent processes. This model predicts that inactivation could be separable from activation. For each of these two mechanisms, external Ca 2ϩ could act by binding to a site on the outside of the channel, or it could enter the cell, through TRPA1 channels, to act internally.
To determine if the processes of Ca 2ϩ -dependent potentiation and Ca 2ϩ -dependent inactivation are independent, we tested if the two could be separated based on their sensitivity to extracellular Ca 2ϩ or other divalent cations. Following Cin activation of TRPA1, we varied the concentration of external Ca 2ϩ that was introduced into the bath. Surprisingly, potentiation of TRPA1 currents could be induced by a Ca 2ϩ concentration as low as 12 M, and at 50 M Ca 2ϩ , the magnitude of the response was saturated, although the response kinetics were slowed as compared with responses to 2 mM Ca 2ϩ (Fig. 2, A and C, 1.26 Ϯ 0.05-fold increase at 12 M Ca 2ϩ , n ϭ 4; 2.22 Ϯ 0.16-fold increase at 50 M Ca 2ϩ , n ϭ 4). In contrast, the rate of inactivation varied with Ca 2ϩ concentration over the entire range that we tested (12 M to 10 mM) (Fig. 2, A and D). Importantly, at low Ca 2ϩ concentrations that still induced maximal potentiation, (e.g. 50 M), inactivation was substantially slower as compared with control conditions (2 mM Ca 2ϩ ), suggesting that inactivation does not derive its Ca 2ϩ dependence from activation.
Potentiation and Inactivation of TRPA1 by Other Divalent Cations-In a similar series of experiments, we tested whether divalent cations other than Ca 2ϩ could potentiate or inactivate TRPA1 currents. Introduction of Ba 2ϩ (2 mM) produced both robust potentiation and rapid inactivation of TRPA1 currents (-fold increase of 1.83 Ϯ 0.05; T50 of 12.0 Ϯ 1.9 s; n ϭ 4). In contrast, introduction of 2 mM Mg 2ϩ caused little or no potentiation but significant inactivation (-fold increase of 1.13 Ϯ 0.09; T50 of 47.5 Ϯ 8.1 s; n ϭ 6) (Fig. 2, B-D), and a higher FIGURE 1. TRPA1 channels are potentiated and inactivated by extracellular Ca 2؉ . Currents were recorded from HEK-293 cells heterologously expressing an N-terminal YFP fusion of rat TRPA1. A and B, in the absence of extracellular Ca 2ϩ , brief exposure to TRPA1 agonists, as indicated, elicited a current, which was first potentiated and then decayed to base line in response to the addition of extracellular Ca 2ϩ (2 mM). Inset, the I-V relationship in response to a ramp depolarization (1 V/s) at the times indicated. Agonists were menthol (30 M) or Cin (100 M). Note that following the introduction of Ca 2ϩ , no additional responses to menthol could be measured, indicating that the channels had entered a long lasting inactivated state. C, TRPA1 currents in response to prolonged exposure to Cin (100 M) in the absence of extracellular Ca 2ϩ reached a plateau after ϳ100 s. The addition of 2 mM Ca 2ϩ did not cause further activation but still promoted rapid inactivation of the currents. Inset, the I-V relationship in response to a ramp depolarization (1 V/s) at the times indicated. Note the linearization of the I-V curve with strong activation. D, magnitude of the TRPA1 currents at ϩ80 mV before and after the addition of 2 mM Ca 2ϩ from experiments as in A and C. The Cin-evoked currents measured with the brief exposure protocol were ϳ50% of the maximal Cin-evoked currents, and only these subsaturating responses could be potentiated by Ca 2ϩ .
concentration of Mg 2ϩ (20 mM) promoted more rapid inactivation (T50 of 11.6 Ϯ 2.2; -fold increase of 1.03 Ϯ 0.01; n ϭ 6; Fig. 2, C and D). Similar results were obtained using a clone of rTRPA1 that was not fused to YFP (see supplemental Fig. 1). We conclude that potentiation and inactivation are mediated by distinct processes that can be separated on the basis of their specificity for divalent cations.
Inactivation of TRPA1 by Mg 2ϩ was unexpected, given that under physiological conditions, this ion is present at millimolar concentrations inside and outside cells. If inactivation is medi-ated by the entry of divalent cations, then it might be predicted that physiologically relevant concentrations of intracellular Mg 2ϩ would inactivate the channel. To determine whether this is the case, we measured TRPA1 currents in response to prolonged exposure to Cin in cells perfused with varying concentrations of free Mg 2ϩ . Inhibition by the TRPA1-specific blocker HC-030031 (10 M) (15) confirmed that the currents were entirely mediated by TRPA1 channels. Perfusion with 8.8 mM free Mg 2ϩ strongly reduced the magnitude of the currents, whereas perfusion with 1.8 mM free Mg 2ϩ , a physiologically relevant concentration, had no effect when compared with perfusion with a low concentration of free Mg 2ϩ (0.35 mM; Fig. 2, E and F). Therefore, physiologically relevant resting levels of free Mg 2ϩ are not expected to inactivate TRPA1. Moreover, the steep dependence of the TRPA1 currents on the intracellular Mg 2ϩ concentration suggests that the process by which divalent cations speed inactivation of TRPA1 is highly cooperative.
Conventional Methods Cannot Be Used to Assign a Location for Ca 2ϩ Regulation of TRPA1-Having established that potentiation and inactivation of TRPA1 are mediated by distinct processes, we next examined whether the Ca 2ϩ regulation of each process was due to Ca 2ϩ binding to the outside of the channel or to entry of Ca 2ϩ through the channel and binding of Ca 2ϩ to an intracellular target. If Ca 2ϩ entry mediates either process, it might be possible to disrupt the effect of extracellular Ca 2ϩ with an intracellular Ca 2ϩ chelator. The preceding experiments were all performed with 5 mM EGTA, indicating that at this concentration, EGTA does not block potentiation or inactivation. Moreover, reducing the buffer to 20 M EGTA only slightly increased the potentiation of the currents and had no effect on the rate of inactivation (-fold increase of 2.50 Ϯ 0.29, T50 of 5.0 Ϯ 0.5 s, n ϭ 5; Fig. 2, C and D).
Because Ca 2ϩ might act close to the site of entry, where EGTA would be ineffective, we also tested whether potentiation or inactivation of TRPA1 was affected by including the fast Ca 2ϩ chelator BAPTA in the pipette. For these experiments, TRPA1 currents were activated by MO, which has a similar mechanism of action as Cin (37). The Ca 2ϩ -dependent potentiation and inactivation of TRPA1 currents were not different between cells that were dialyzed with 5 mM BAPTA and those dialyzed with 20 M EGTA (supplemental Fig. 2). In similar experiments, other groups have reported only a partial block of TRPA1 activation by BAPTA (21,24). Therefore, either Ca 2ϩ acts on the extracellular side of the channel, or the Ca 2ϩ buffers at the concentrations we used were not effective in interfering with the binding of Ca 2ϩ to its intracellular target. This might be the case if Ca 2ϩ acts at a site very close to the permeation pathway or if the buffer is saturated by the large Ca 2ϩ influx through the channels.
Mutation of a Proposed EF-hand Does Not Affect Potentiation or Inactivation-It has recently been proposed that an EF-hand located within the N terminus of TRPA1 mediates Ca 2ϩ activation (20,21). To confirm this result and determine if this site also mediates inactivation of TRPA1, we mutated each of three residues proposed to participate in activation of TRPA1 by Ca 2ϩ (S471R/D472A, L475A, D480A; see supplemental Fig. 3). None of the mutations had any effect on Ca 2ϩ -dependent potentiation or Ca 2ϩ -dependent inactivation (supplemental . Note that each trace represents a recording from a separate cell, and the traces have been aligned by the time at which Ca 2ϩ was introduced. The currents were scaled by the magnitude of the response to Cin, and therefore the y axis is represented by an arbitrary unit. B, currents evoked in response to brief exposure to Cin (100 M) and the subsequent addition of Ba 2ϩ or Mg 2ϩ (2 mM). C, average data from experiments as in A and B. Potentiation was measured as the -fold increase in the current (at ϩ80 mV) following the addition of the divalent cation, where 1 represents no change in the current. Note the "threshold" for potentiation of the current by extracellular Ca 2ϩ was ϳ12 M. The bar corresponding to the response to "0 Ca 2ϩ " was duplicated to allow comparison with responses to Ba 2ϩ and Mg 2ϩ . D, inactivation was measured by the T50 or time, relative to the peak, at which the currents had decayed to 50% of their maximum value (at ϩ80 mV). There was a continuous slowing of inactivation as the Ca 2ϩ concentration was lowered. Data represent the mean Ϯ S.E. (n ϭ 4 -10). Intracellular solution was HB-Cs ϩ (which contained 5 mM EGTA) for A and B. E, TRPA1 currents elicited in response to a saturating exposure to Cin (100 M) with varying concentrations of free Mg 2ϩ in the pipette (as indicated). HC-030031 (10 M) was introduced at the end of each experiment to confirm that the current was entirely due to the gating of TRPA1 channels. Intracellular solution was HB-Cs ϩ with 0, 2, or 10 mM MgCl 2 added. F, magnitude of the current (ϩ80 mV) from experiments as in E. Data represent the mean Ϯ S.E. *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001. Fig. 3). Similar results were obtained with a clone of rTRPA1 that was not fused to YFP (L475A and D480A were tested; supplemental Fig. 4). To determine whether the differences between our results and those of Doerner et al. (21) were due to the use of different species variants of TRPA1 or to different activating chemicals, we introduced mutations into two sites on hTRPA1 (L474A and D479A) and measured potentiation following both Cin and MO activation. Mutation of Leu 474 in hTRPA1 had no effect on Ca 2ϩ potentiation following Cin or MO activation, whereas mutation of Asp 479 abrogated channel function, as previously noted (21) (supplemental Figs. 5 and 6). Therefore, as assessed under a variety of conditions, residues in a putative EF-hand do not play a critical role in potentiation or inactivation of TRPA1, and EF-hand mutant channels cannot be used to assign a location for Ca 2ϩ regulation. We noted that mutation of Asp 480 in rTRPA1 destabilized opening of the channel by Cin, and this region, which is in close proximity to N-terminal cysteine residues that are the targets for pungent chemicals (17,18), may therefore be more generally involved in channel gating.
A Pore Region Mutation Disrupts Ca 2ϩ -dependent Potentiation and Inactivation of TRPA1-A more direct way to assess the role of Ca 2ϩ entry in potentiation and inactivation of TRPA1 is to reduce the Ca 2ϩ permeability of the channel and determine whether it can still be potentiated or inactivated by extracellular Ca 2ϩ . In other structurally related ion channels, this has been accomplished by mutation of acidic residues in the pore region (34,38). An alignment of TRPA1 with other cation channels places Asp 918 in the selectivity filter at the same position as the tyrosine residue in the GYG motif of the potassium channel (39) (Fig. 3C). We therefore mutated Asp 918 and characterized the ion selectivity of the Cin-activated currents by rapidly exchanging the external solution with those containing either Na ϩ , Cs ϩ , K ϩ , or Ca 2ϩ as the only permeable cation. The reversal potential of the current under these bionic conditions can be used to calculate permeability relative to Cs ϩ , which was the only cation present in the pipette solution (Fig. 3A) (34). Consistent with previous reports (3), we found that wild-type TRPA1 channels are highly Ca 2ϩ -permeable (P Ca /P Cs ϭ 3.28 Ϯ 0.58, n ϭ 6). Mutation of Asp 918 to alanine (D918A) led to a reduction in calcium permeability of nearly 2 orders of magnitude (P Ca /P Cs ϭ 0.08 Ϯ 0.01, p Ͻ 0.05; n ϭ 3; Fig. 3, A and B). A chargeneutralizing substitution that did not change the size of the amino acid side chain, D918N, produced channels with an intermediate Ca 2ϩ permeability (P Ca /P Cs ϭ 0.26 Ϯ 0.01, p Ͻ 0.01; n ϭ 6; Fig. 3, A and B), and the charge conserving substitution D918E increased Ca 2ϩ permeability (P Ca /P Cs ϭ 13.56 Ϯ 2.63, p Ͻ 0.05; n ϭ 8; Fig. 3, A and B). Note that the D to E substitution is found in TRPA1 orthologs from several species, including zebrafish and Drosophila (Fig. 3D).
If Ca 2ϩ entry mediates either potentiation or inactivation, we expected that one or both of these processes would be impaired in the D918A mutant, which is poorly permeable to Ca 2ϩ . Indeed, the D918A mutant showed virtually no potentiation or inactivation in response to the addition of 2 mM Ca 2ϩ following activation with Cin (-fold increase of 1.06 Ϯ 0.02, p Ͻ 0.001; T50 of 105.8 Ϯ 19.3 s, p Ͻ 0.001; n ϭ 5; Fig. 4, B, E, and F). According to this logic, the D918N mutant, which has an intermediate permeability to Ca 2ϩ , should show a milder deficit in potentiation and inactivation. As seen in Fig. 4, D918N currents showed robust potentiation in response to extracellular Ca 2ϩ (-fold increase of 1.52 Ϯ 0.15; n ϭ 5), but potentiation was slowed and resembled the response of wild-type channels to a much lower external Ca 2ϩ concentration (Fig. 4C; compare with Fig. 2A). Inactivation of D918N channels was also signifi- . Plots show the current in response to a ramp depolarization (1 V/s) in the presence of the indicated extracellular cation (Na ϩ , NMDG ϩ , or Ca 2ϩ ). The extracellular solution was rapidly exchanged to minimize changes in the magnitude of the currents during the recording. The arrow indicates the reversal of the current with 128 mM Ca 2ϩ in the bath. Intracellular solution was HB-Cs ϩ . B, ion selectivity was calculated based on the reversal potential of the current as measured from experiments as in A. D918A (n ϭ 3) and D918N (n ϭ 6) showed dramatically reduced Ca 2ϩ permeability, whereas D918E (n ϭ 8) had increased Ca 2ϩ permeability as compared with wild type (n ϭ 6). C, the S5-S6 region of TRPA1 aligned with that of related cation-permeable channels. Conserved acidic residues important for ion permeability are shown in red. Gray shading highlights the conserved GXG sequence in the selectivity filter. Note that Asp 918 is at the same position as the Tyr in the GYG motif of the K ϩ channel. Pore region alignment from multiple channels was based partly on Ref. 49. D, alignment of the putative selectivity filters of TRPA1 from multiple species. Data represent the mean Ϯ S.E. Significance was determined by two-tailed Student's t test and was Bonferroni-corrected. *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001. Ca2ϩI and NaϩI, the first transmembrane domains of the Ca 2ϩ and Na ϩ channel, respectively. Accession numbers are as follows: POA334 (KcsA), ZP_02523257 (NaKbc), ZP_02523257 (CNGA1), NP_997491 (TRPA1), NP_076460 (TRPV4), NP_446239 (TRPV5), Q13936 (Ca 2ϩ I), NP_062139 (Na ϩ I), NP_015628 (human TRPA1), NP_997491 (rat TRPA1), NP_808449 (mouse TRPA1), XP_544123 (dog TRPA1), XP_581588 (cow TRPA1), NP_001007066 (fish TRPA1), NP_001097554 (fly TRPA1).

, D-F).
To determine if the decrease in potentiation of D918A currents by Ca 2ϩ could be attributed to saturation of the activation of the channels by Cin, we measured the magnitude of D918A currents following prolonged Cin exposure, which is expected to activate all available channels. Prolonged Cin exposure elicited currents of 4.1 Ϯ 0.9 and Ϫ1.9 Ϯ 0.7 nA at ϩ80 and Ϫ80 mV, respectively, (n ϭ 6), which was ϳ2 times the size of the currents used for assessing Ca 2ϩ potentiation (2.1 Ϯ 0.1 and Ϫ0.54 Ϯ 0.07 nA at ϩ80 and Ϫ80 mV, respectively, n ϭ 5). Note that current magnitudes of all pore mutants were somewhat reduced as compared with the wild-type current, but compared with each other, they were not significantly different (2.3 Ϯ 0.1 and Ϫ0.27 Ϯ 0.04 nA at ϩ80 and Ϫ80 mV, respectively, n ϭ 5 for D918N; 2.4 Ϯ 0.1 and Ϫ1.4 Ϯ 0.9 nA at ϩ80 and Ϫ80 mV, respectively, n ϭ 4 for D918E).
Potentiation and Inactivation of TRPA1 Channels Can Be Attributed to Permeation of Ca 2ϩ through the Channel-It is well recognized that permeant ions can alter the gating of ion channels, presumably because their occupancy in the permeation pathway impedes channel closure (40). It has also been shown that mutations of the pore of ion channels can affect gating (e.g. see Refs. 41 and 42). Indeed, the activation of the TRPA1 pore mutants was somewhat more sluggish than that of wild-type currents (Fig. 4, A-D); this was confirmed by measurement of the 50% activation time, which showed a 2-fold slowing for D918A currents as compared with wild-type currents (79.3 Ϯ 6.9 s for D918A and 48.9 Ϯ 5.4 s for wild type, p Ͻ 0.01, measured from experiments where Cin was applied until the currents reached saturation, as in Fig. 1C). It was therefore important to determine whether mutation of the TRPA1 pore disrupted potentiation and inactivation through reduced Ca 2ϩ influx, as we hypothesized, or via a direct effect on the gating of the channel. If the mutation principally affected Ca 2ϩ influx, we reasoned that potentiation and inactivation might be rescued by reducing the intracellular Ca 2ϩ buffering so as to allow accumulation of intracellular Ca 2ϩ . As shown in Fig. 5A, Cin-activated D918A currents were not potentiated or inactivated by 2 or 10 mM extracellular Ca 2ϩ when intracellular Ca 2ϩ was buffered with 5 mM EGTA (Fig. 5, A and B), and lowering the concentration of EGTA to 20 M partially rescued both potentiation and inactivation (Fig. 5, A and B).
We confirmed that the small Ca 2ϩ flux through D918A channels produced a significant elevation of intracellular Ca 2ϩ under conditions of low Ca 2ϩ buffering by repeating the experiment with the ratiometric Ca 2ϩ indicator fura-4F in the pipette. The addition of 2 or 10 mM Ca 2ϩ to the bath produced a Ca 2ϩ elevation of 1-3 M in D918A-expressing cells activated with Cin (Fig. 5, C and D). Moreover, this elevation of intracellular Ca 2ϩ was blocked by including 5 mM EGTA in the pipette (Fig. 5, C and D). Interestingly, in the cells expressing wild-type channels, EGTA did not block potentiation or inactivation, and Ca 2ϩ imaging revealed a substantial elevation of intracellular Ca 2ϩ , despite the presence of the buffer (Fig. 5, C and D). These data confirm that potentiation and inactivation are mediated by elevation of intracellular Ca 2ϩ and that mutation of Asp 918 disrupts inactivation by reducing Ca 2ϩ permeability.
TRPA1 Is Potentiated but Is Not Inactivated by Uncaging Ca 2ϩ -The previous experiments indicate that the ability of extracellular Ca 2ϩ to potentiate and inactivate TRPA1 requires Ca 2ϩ entry. To determine if elevation of intracellular Ca 2ϩ is sufficient to potentiate and inactivate TRPA1 channels, we  ). B, extracellular Ca 2ϩ caused no significant potentiation and only slow inactivation of the mutant D918A, which is poorly permeable to Ca 2ϩ . C, currents carried by the D918N mutant, which has an intermediate Ca 2ϩ permeability, were strongly potentiation by Ca 2ϩ , but potentiation was slowed as compared with wild type. D, currents carried by the D918E mutant, which has an enhanced Ca 2ϩ permeability, were strongly activated and rapidly inactivated by extracellular Ca 2ϩ . E and F, average data from experiments as in A-D, quantified as in Fig. 1. Wild-type data are reproduced from Fig. 1. Significance was determined by two-tailed Student's t test with Bonferroni correction for multiple samples. ***, p Ͻ 0.001. Internal solution was HB-Cs ϩ , which contained 5 mM EGTA.
loaded cells with Ca 2ϩ bound to the chelator DMNP-EDTA, which changes affinity for Ca 2ϩ from 5 nM to 3 mM upon exposure to UV light (43). Fig. 6A shows that UV uncaging of DMNP-EDTA in the absence of extracellular Ca 2ϩ caused a sustained potentiation of the Cin-activated TRPA1 current (2.70 Ϯ 0.20-fold increase at ϩ80 mV, n ϭ 5) (Fig. 6C). The effect of elevating intracellular Ca 2ϩ partially occluded further potentiation by extracellular Ca 2ϩ , consistent with our previous observations. To confirm that the observed responses were due to elevation of intracellular Ca 2ϩ and not to direct effects of the two photolytic products (44), we loaded cells with DMNP-EDTA without Ca 2ϩ . Under these conditions, UV exposure caused only a slow, small increase in TRPA1 currents (Fig. 6, B and C; 1.14 Ϯ 0.04-fold increase at ϩ80 mV, n ϭ 5), and this increase was significantly less than what we observed when DMNP-EDTA was loaded with Ca 2ϩ (p Ͻ 0.001).
A surprising result of these experiments was that uncaging Ca 2ϩ did not inactivate TRPA1 currents. This could be explained if the Ca 2ϩ elevation produced by the uncaging protocol was insufficient to promote inactivation. To test this hypothesis, we compared the Ca 2ϩ elevation produced by UV uncaging of DMNP-EDTA with that observed upon subsequent entry of Ca 2ϩ through TRPA1 channels. Ca 2ϩ levels were measured with the indicator Fluo-5F, which can be visualized with wavelengths of light that do not uncage DMNP-EDTA. Under these conditions, we found that Ca 2ϩ entry produced a significantly greater global elevation of intracellular Ca 2ϩ than did Ca 2ϩ uncaging (p Ͻ 0.001; Fig. 6, D and E). It should be noted that Ca 2ϩ levels at the plasma membrane upon entry through TRPA1 channels could be considerably higher than what we have measured. Together, these data indicate that elevation of intracellular Ca 2ϩ by uncaging DMNP-EDTA can potentiate TRPA1 currents but not inactivate them. Inactivation may require a greater elevation of intracellular Ca 2ϩ than can be achieved by Ca 2ϩ uncaging.
Effect of Intracellular Ca 2ϩ on TRPA1 Single Channels-To study the effects of Ca 2ϩ on TRPA1 channels in more detail, we recorded the gating of single TRPA1 channels in cell-attached patches. If intracellular Ca 2ϩ is sufficient to both potentiate and inactivate TRPA1 channels, the addition of extracellular Ca 2ϩ to a region outside the patch pipette should potentiate and inactivate TRPA1 channels under the patch pipette. Fig. 7A shows a cellattached patch clamp recording with 100 nM Ca 2ϩ in the patch pipette from a TRPA1-expressing HEK 293 cell. Constitutive channel activity was observed, and this activity was increased by the addition of Cin to the bath, indicating that it represented opening of TRPA1 channels. Moreover, this activity was never observed in cells not transfected with TRPA1 (not shown). Measurement of single channel amplitudes in recordings of voltage families gave a slope conductance of 139 Ϯ 2 pS (n ϭ 3; Fig. 7B). Following Cin activation, the addition of Ca 2ϩ to the intact cell produced dramatic potentiation, followed by a rapid decay in channel activity. Measurement of single channel currents and fitting of all points histograms indicated that the single channel conductance was unchanged in response to Cin activation or Ca 2ϩ potentiation (141 Ϯ 12 pS (n ϭ 3) before Cin, 148 Ϯ 3 pS (n ϭ 5) after Cin, and 143 Ϯ 4 pS (n ϭ 5) after Ca 2ϩ ; Fig. 7, C-E). Following inactivation, sparse channel openings were observed, and these openings were of variable amplitude, possibly reflecting the sparse opening of TRPA1 channels as well as the opening of smaller endogenous Ca 2ϩ -activated channels. 3 We conclude that potentiation and inactivation of TRPA1 can be 3 Y. Y. Wang and E. R. Liman, unpublished observations. induced by intracellular Ca 2ϩ in a manner that is not membrane-delimited and that potentiation involves a change in channel gating and not in conductance.
Finally, we were interested in knowing whether Ca 2ϩ potentiation and Ca 2ϩ inactivation of TRPA1 was retained in excised patches. A technical difficulty with these experiments is that TRPA1 activity induced by MO or Cin is not preserved upon patch excision (45). Recently, it was reported that polyphosphates can restore TRPA1 activity after patch excision (45), making it now possible to study TRPA1 channels in excised patches. As shown in Fig. 8A, following activation by Cin and in the presence of polyP3, TRPA1 channel activity is a rapidly and reversibly potentiated by intracellular Ca 2ϩ . The dose dependence of activation can be fit with the Hill equation with K1 ⁄ 2 of 225 nM and slope of 1.8 (n ϭ 6; Fig. 8B). We also tested whether the channels undergo inactivation under these conditions. TRPA1 currents showed only modest decay in response to 12 M intracellular Ca 2ϩ (Fig. 8C). At higher Ca 2ϩ concentrations, more substantial decay could be observed (Fig. 8, D and E). Thus, in cell-free excised patches preactivated with Cin, TRPA1 current can be potentiated and inactivated by cytoplasmic Ca 2ϩ , and potentiation occurs at lower Ca 2ϩ concentrations.

DISCUSSION
Inactivation of TRPA1 can rapidly terminate signaling and therefore may be one of the key mechanisms that limit sensory activation by painful or irritating substances. Previous work has demonstrated that extracellular Ca 2ϩ both potentiates and inactivates TRPA1 (4,24). In this study, we show that inactivation of TRPA1 is not coupled to potentiation and that both processes are mediated by elevation of intracellular Ca 2ϩ . In addition to these findings, we have discovered a key residue in TRPA1 that controls Ca 2ϩ permeability of the channel, providing experimental evidence that the S5-S6 linker of this channel forms part of the pore region. Together, these results give insights into the mechanisms of regulation of TRPA1 and provide tools for future research.
Identification of the Ion Selectivity Filter in TRPA1-An important outcome of our work is the identification of an amino acid residue that determines Ca 2ϩ selectivity of TRPA1. Ca 2ϩ influx through TRPA1 undoubtedly plays an important role in the physiology and pathophysiology of sensory neurons, possibly mediating local release of inflammatory mediators as well as generating cross-desensitization of co-expressed TRPV1 channels (46).
The architectural similarity between TRP channels and voltage-gated ion channels suggests that the pore (P) region in TRP channels will be formed from the S5-S6 linker as it is in K ϩ channels (39,47). Within the P region, a "selectivity filter" is formed in the K ϩ channel by a short sequence of highly conserved amino acids, with the motif TVGYG, the backbone carbonyls of which form a series of binding sites for ions within the permeation pathway (47). Recent studies suggest that the pore of Ca 2ϩ -and Na ϩ -selective ion channels is structurally similar to that of K ϩ channels (48,49). The reported crystal structure of NaKbc, a nonselective cation channel from bacteria that shares homology with cyclic nucleotide-gated ion channels, shows that the carboxyl groups of conserved acidic residues do not line the permeation pathway, as previously assumed, and are instead tangential to it, acting through electrostatic interactions to stabilize permeating divalent cations (48,50).
An alignment of TRPA1 with NaKbc and KcsA shows that Asp 918 , which we have identified as a key determinant of the Ca 2ϩ permeability of TRPA1, aligns with the acidic residue in NaKbc that stabilizes permeating Ca 2ϩ and with the tyrosine in the K ϩ channel that is essential for K ϩ selectivity (Fig. 2C). It also aligns with a methionine in TRPV4 and an aspartate in TRPV5, which are key determinants of ion selectivity in these channels (51,52), and it is near an aspartate in Drosophila TRP that has been shown to determine ion selectivity in vivo (38). Mutation of Asp 918 to Ala nearly eliminated Ca 2ϩ permeability of the channel, whereas substitution with the size-conserving uncharged Asn residue caused a more moderate decrease in Ca 2ϩ permeability, and substitution with the charge-conserving larger Glu residue led to an increase in Ca 2ϩ permeability. These results are inconsistent with a direct coordination of Ca 2ϩ by Asp 918 within the constrained environment of the ion permeation pathway and instead indicate that Asp 918 stabilizes the permeating Ca 2ϩ ions at a distance by electrostatics, like the stabilization of Ca 2ϩ by acidic residues in the pore of the NaKbc channel (48,50).
Potentiation of TRPA1 by Intracellular Ca 2ϩ -Our data support and extend previous observations that TRPA1 channels can be activated by elevation of intracellular Ca 2ϩ (20,21,23). Gating of TRPA1 by intracellular Ca 2ϩ may be the mechanism by which TRPA1 is activated downstream of bradykinin and other inflammatory mediators that bind to G protein-coupled receptors, although a role for other signaling molecules has also been proposed (8,9,53). Activation of TRPA1 has been observed in response to intracellular dialysis of Ca 2ϩ , with halfactivation concentrations of 900 nM and 6 M measured by two different groups (20,21). Activation of TRPA1 by intracellular Ca 2ϩ has also been observed in excised patches (20,21), but recent data indicate that a co-factor lost upon patch excision is required to maintain these responses (23,45).
In the presence of polyP3, which restores TRPA1 activity (45), we found that TRPA1 channels could be activated at concentrations of Ca 2ϩ just slightly higher than the resting level (K1 ⁄ 2 ϳ225 nM); this is in contrast to two other Ca 2ϩ -activated TRP channels, TRPM4 and TRPM5, which in excised patches require micromolar concentrations of Ca 2ϩ to be activated (27, 30, 54 -57). By studying the D918A TRPA1 channel, in which Ca 2ϩ permeability is reduced, we provide evidence that potentiation of TRPA1 by extracellular Ca 2ϩ is mediated by Ca 2ϩ entry. Importantly, Ca 2ϩ potentiation of D918A currents was blocked by including 5 mM EGTA in the pipette and could be restored by lowering the concentration of EGTA. In addition to demonstrat-  ing that intracellular Ca 2ϩ potentiates TRPA1, these data suggest that, under some conditions, TRPA1 channels can respond to a global change in the intracellular Ca 2ϩ concentration (58). This was further confirmed by our observation that TRPA1 channels in cell-attached patches could be potentiated by the addition of Ca 2ϩ to a region outside of the patch. Similar results were recently reported for activation of TRPA1 in the absence of a pungent chemical (23). We noted that potentiation was accompanied by an increase in channel activity, without a change in channel conductance. A previous report showing a decrease in channel conductance during potentiation may be attributed to a concomitant block of the channel by extracellular Ca 2ϩ (24).
Our report is also the first to show that TRPA1 can be potentiated by elevation of intracellular Ca 2ϩ through UV uncaging of the Ca 2ϩ cage DMNP-EDTA. This method has been extensively used in assessing the role of Ca 2ϩ signaling in synaptic transmission and vesicle release as well as for studying Ca 2ϩ regulation of ion channels, including TRP channels (33, 43-45, 57, 59). In studying TRPA1, which can be modified by a number of reactive compounds, it was important to establish that the effects of uncaging could not be attributed to effects of the two iminodiacetic acids generated by the photolysis of DMNP-EDTA (43,44). Our observation that UV uncaging of DMNP-EDTA in the absence of intracellular Ca 2ϩ caused only a small, slow increase of the TRPA1 currents indicates that the potentiation of TRPA1 currents upon UV uncaging of DMNP-EDTA could be largely attributed to Ca 2ϩ elevation. Therefore, with the proper controls, flash photolysis of DMNP-EDTA is a useful method for studying Ca 2ϩ activation of TRPA1. Interestingly, with this method, we observed a long lasting potentiation of TRPA1, which was not accompanied by inactivation. In similar experiments with other Ca 2ϩ -activated TRP channels, a transient activation of the currents was observed, consistent with the observed transient elevation of Ca 2ϩ (33,57). The absence of inactivation could be explained if UV uncaging of DMNP-EDTA did not produce sufficiently high intracellular Ca 2ϩ levels. This interpretation is supported by our observation that the Ca 2ϩ levels attained by UV uncaging of DMNP-EDTA were significantly lower than those attained through entry of Ca 2ϩ through TRPA1 channels.
Intracellular Ca 2ϩ Elevation Mediates Inactivation of TRPA1-Inactivation of heterologously expressed TRPA1 by extracellular Ca 2ϩ has been well documented (3,4,8,19,20,24) and is also observed in neurons (e.g. see Ref. 12). Our results show that, like potentiation, inactivation by extracellular Ca 2ϩ requires Ca 2ϩ entry. Thus D918A channels, which are poorly permeable to Ca 2ϩ , inactivate only slowly in response to extracellular Ca 2ϩ , and inactivation can be largely restored by lowering intracellular Ca 2ϩ buffering. Moreover, TRPA1 channels in excised and cell-attached patches could be inactivated through elevation of intracellular Ca 2ϩ , in the absence of extracellular Ca 2ϩ . It is perhaps not surprising that elevation of intracellular Ca 2ϩ mediates inactivation of TRPA1, since it induces inactivation/desensitization of many ion channels and receptors, including both Ca 2ϩ -permeable and Ca 2ϩ -impermeable TRP channels (25)(26)(27)(28)(29)(30)(31)60).
Given that intracellular Ca 2ϩ potentiates and inactivates TRPA1, is there a mechanism that ensures that potentiation precedes inactivation? In voltage-gated Na ϩ and K ϩ channels, inactivation is coupled to voltage-dependent activation, such that the channels first open and then inactivate (61). A similar coupling between Ca 2ϩ -dependent activation/potentiation and inactivation of TRPA1 appears unlikely for several reasons. First, in response to moderate elevation of intracellular Ca 2ϩ by UV uncaging of DMNP-EDTA, potentiation can be observed in the absence of inactivation, indicating that the two processes are not strictly coupled. Moreover, inactivation can be induced by extracellular Mg 2ϩ in the absence of potentiation, suggesting that the two processes are mediated by intracellular signaling events with different ionic specificities. Excised patch recording suggests that inactivation follows potentiation because inactivation is relatively slow at the low Ca 2ϩ concentrations that maximally activate the channel. Moreover, these data predict that a modest Ca 2ϩ elevation in response to signaling by G protein-coupled receptors or receptor tyrosine kinases might activate without inactivating TRPA1. The subsequent entry of Ca 2ϩ through TRPA1 channels may then terminate the signal.
Where is the Site for Ca 2ϩ Potentiation and Inactivation of TRPA1?-An attractive hypothesis is that Ca 2ϩ activates TRPA1 by binding directly to a putative EF-hand on the N terminus of TRPA1, as proposed recently by two groups (20,21). Within the stretch of 12 amino acids that comprise the Ca 2ϩ binding site on a canonical EF-hand, 6 residues coordinate Ca 2ϩ (62). Doerner et al. (21) found that of these sites, mutations of only one (Leu 474 ) significantly reduced Ca 2ϩ activation of the channels. This is surprising in that Leu 474 occupies a position that contributes a backbone carbonyl to the Ca 2ϩ binding site of a canonical EF-hand, and Leu is not commonly observed at this position (62). Zurborg et al. (20) found, in contrast, that mutant channels with charge-neutralizing substitutions of the acidic residues that flank the EF-hand were no longer sensitive to Ca 2ϩ , as assayed by intracellular Ca 2ϩ dialysis and other methods. Our experiments show that neither Leu 474 nor Asp 479 is necessary for Ca 2ϩ potentiation of TRPA1. We observed robust potentiation of hTRPA1 L474A and of rTRPA1 L475A (analogous to L474A in hTRPA1), using both a fusion protein with YFP and the unfused channel. The absence of potentiation of hTRPA1 L474A currents in the experiments by Doerner et al. (21) could be due to a saturation of the response to the agonist (MO), which can occlude activation by Ca 2ϩ (supplemental Fig. 6). In agreement with Doerner et al. (21) and contrary to the report by Zurborg et al. (20), we found that the D479A mutation of hTRPA1 abrogated channel activity. The equivalent mutation in rTRPA1, D480A, did not interfere with Ca 2ϩ potentiation but appeared to destabilize channel activation. Together, these results make it unlikely that an N-terminal region proposed to form an EF-hand contributes to Ca 2ϩ potentiation of TRPA1, although it may otherwise regulate gating of TRPA1.
Other mechanisms by which Ca 2ϩ could potentiate or inactivate TRPA1 include direct binding to other acidic residues on the intracellular side of TRPA1, of which there are more than 100, or binding to other proteins that regulate TRPA1 either directly or indirectly. A prime suspect for the calcium sensor is calmodulin, which mediates the Ca 2ϩ -dependent facilitation and inactivation of voltage-gated Ca 2ϩ channels (63)(64)(65)(66)(67)(68) and desensitization of Ca 2ϩ -permeable cyclic nucleotide-gated ion channels (69,70) and which binds to several TRP channels (71)(72)(73)(74)(75)(76). However, expression of TRPA1 with a dominant negative form of calmodulin did not abrogate Ca 2ϩ potentiation or inactivation (20), suggesting that calmodulin is unlikely to be the sensor for Ca 2ϩ regulation of TRPA1. 3 Alternatively, elevation of intracellular Ca 2ϩ could activate enzymatic processes that modulate channel activity. Other TRP channels are regulated by PI(4,5)P 2 , which can be depleted by elevation of intracellular Ca 2ϩ (77). Depletion of PI(4,5)P 2 underlies desensitization of TRPV1, TRPM4, TRPM5, TRPM7, and TRPM8 (27-31, 78 -80). PI(4,5)P 2 depletion has also been proposed to underlie sensitization of TRPV1 in response to the proalgesic agents bradykinin and nerve growth factor that activate receptors coupled to phospholipase C (81,82). A role for PI(4,5)P 2 depletion in the desensitization of TRPA1 is supported by recent experiments showing that cross-desensitization of MO-evoked currents in sensory neurons by capsaicin, acting on TRPV1, can be blocked by infusion of PI(4,5)P 2 through the pipette (19) and by evidence that PI(4,5)P 2 can restore channel activation following partial, but not complete, inactivation in excised patches (83). But other data suggest that PI(4,5)P 2 depletion contributes to activation or sensitization of TRPA1 (9,53). Using newly developed methods for manipulating PI(4,5)P 2 levels (79, 84), we found that PI(4,5)P 2 depletion neither activates nor inactivates TRPA1 (see supplemental Fig. 7), although we cannot rule out the possibility that changes in PI(4,5)P 2 levels might have more subtle effects on channel function. Yet another mechanism by which Ca 2ϩ regulates TRP channels is through Ca 2ϩ -activated phosphatase calcineurin (85), although a recent report does not support a role for calcineurin in desensitization of TRPA1 (19).
Ca 2ϩ Regulation of TRPA1 in Sensory Neurons-In sensory neurons, intracellular Ca 2ϩ can be elevated through several distinct mechanisms, each of which could affect the gating of TRPA1. For example, inflammatory mediators, such as bradykinin, can act on cognate G protein-coupled receptors to release intracellular Ca 2ϩ ; thermal and chemical stimuli can cause entry of Ca 2ϩ through TRPV1 or TRPA1 channels; and depolarization can activate voltage-gated Ca 2ϩ channels (86,87). How elevation of Ca 2ϩ contributes to activation and inactivation of TRPA1 in sensory neurons is not well understood and may require a better understanding of the complex spatial and temporal dynamics of Ca 2ϩ signaling in these cells. An example of this complexity is the surprising observation that targeted deletion of either TRPA1 or TRPV1 eliminates bradykinin-evoked Ca 2ϩ entry in sensory neurons (5). This suggests that the Ca 2ϩ release that is evoked by bradykinin is insufficient to activate TRPA1 and that concurrent entry of Ca 2ϩ through TRPV1 may be necessary. Adding to this complexity is the possibility that changes in other signaling molecules, such as depletion of PI(4,5)P 2 , also contribute to activation of TRPA1 in sensory neurons (9,22). A similar complexity applies to inactivation (or desensitization) of TRPA1 in sensory neurons. In behavioral studies, responses to MO show both homologous desensitization and cross-desensitization by TRPV1 activation (88,89). Recent data indicate that the cross-desensitization of MO responses is Ca 2ϩ -dependent (46) and therefore may be mediated by the mechanism we and others observe in heterologous cells. However, homologous desensitization of MO responses in sensory neurons is reported to be Ca 2ϩ -independent (46) and may be mediated instead by TRPV1-dependent internalization (19). The absence of Ca 2ϩ -sensitive homologous desensitization may reflect the lower current density of TRPA1 in sensory neurons (e.g. see Ref. 4) or differences between native and expressed channels in subunit composition or post-translational modification. Clearly, additional studies are needed to understand the complex regulation of TRPA1 in native cells.