Transient Receptor Potential Channel A1 Is Directly Gated by Calcium Ions*

Members of the superfamily of transient receptor potential (TRP) channels are proposed to play important roles in sensory physiology. As an excitatory ion channel TRPA1 is robustly activated by pungent irritants in mustard and garlic and is suggested to mediate the inflammatory actions of environmental irritants and proalgesic agents. Here, we demonstrate that, in addition to pungent natural compounds, Ca2+ directly gates heterologously expressed TRPA1 in whole-cell and excised-patch recordings with an apparent EC50 of 905 nm. Pharmacological experiments and site-directed mutagenesis indicate that the N-terminal EF-hand calcium-binding domain of the channel is involved in Ca2+-dependent activation. Furthermore, we determine Ca2+ as prerequisite for icilin activity on TRPA1.

The transient receptor potential channel A1 (TRPA1) 2 is a member of the superfamily of TRP channels. In mammals, A1 forms its own subfamily and is distinguished from other TRP channels by the presence of ϳ14 ankyrin repeats in its N terminus (1). TRPA1 was initially described as a cold sensitive nonselective cation channel (2), but it also functions as a ligand-gated channel in heterologous expression systems and sensory neurons (3). The pungent ingredients in mustard (allyl isothiocyanate, AITC) and garlic (allicin) robustly activate TRPA1 currents (4 -6). In addition, TRPA1 appears to be regulated by phospholipase C (PLC)-coupled receptors, suggesting that channel opening can be mediated by second messengers (4,7,8).
TRPA1 expression was first described in a subset of sensory neurons of dorsal root and trigeminal ganglia that contribute to nociception and co-express calcitonin gene related peptide, substance P and TRPV1 (2,4). Recent studies on TRPA1 deficient mice support a role of the channel in inflammatory pain and sensation of noxious cold (8,9). A model suggests TRPA1 activation by bradykinin, a potent algogenic substance released due to tissue injury and inflammation, in two possible ways: through PLC-mediated increases in intracellular Ca 2ϩ or other metabolites (e.g. diacylglycerol) and via Ca 2ϩ influx through TRPV1 (9). Whether an increase in intracellular Ca 2ϩ is sufficient to activate TRPA1 is still debated (4,7), but several findings indicate a role of Ca 2ϩ on TRPA1 function. It was shown that extracellular Ca 2ϩ enhances the current rate and magnitude of AITC-induced currents (4,10). Furthermore, Ca 2ϩ is thought to be responsible for fast channel closure (10). In addition, single channel recordings of heterologously expressed TRPA1 revealed an AITC-induced conductance, which is reduced in the presence of Ca 2ϩ (10). Together these reports emphasize the importance of Ca 2ϩ for TRPA1 function.
Very recently the existence of a putative EF-hand calciumbinding domain (EF-hand CBD) at the N terminus of TRPA1 was reported (11). The EF-hand CBD is the most common motif among Ca 2ϩ -binding sites of a large number of Ca 2ϩinteracting proteins (12). The classical EF-hand is a helix-loophelix motif that coordinates the Ca 2ϩ ion in a pentagonal pyramidal configuration. The domain consists usually of 12 residues, whereas six residues in positions 1, 3, 5, 7, 9, and 12 are postulated to be involved in Ca 2ϩ -binding (12).
In this study we set out to clarify the role of Ca 2ϩ on TRPA1 channel activation by studying human TRPA1 transiently expressed in HEK293 cells. Consistent with previous studies on AITC, our results show that Ca 2ϩ potentiates the cinnamaldehyde-and carvacrol-induced responses. In addition, we determine Ca 2ϩ as essential co-agonist for icilin activity on TRPA1. Interestingly, Ca 2ϩ also directly activates TRPA1 (EC 50 of 905 nM) in whole-cell and excised inside-out patch recordings. The mechanism underlying Ca 2ϩ -dependent activation is analyzed using pharmacological approaches and mutagenesis studies, thereby identifying the EF-hand CBD as potential Ca 2ϩ -binding site.
Transient Expression of Human TRPA1 and Mutagenesis-For transient expression of human TRPA1 (hTRPA1) we used a recombinant expression plasmid (pcDNA5-FRT) carrying the entire protein coding region for hTRPA1. The plasmid was kindly provided by H.-J. Behrendt. Semiconfluent HEK293 cells were transiently transfected (2 g of hTRPA1 cDNA per dish) in 35-mm dishes (Flacon, BD Bioscience, Heidelberg, Germany) using the CaP-precipitation method as described previously (13). Co-transfected pIRES-EGFP (0.2 g per dish) served as transfection marker. All recordings were performed at room temperature ϳ24 -48 h after transfection. Untransfected cells were used for control recordings.
For whole-cell recordings of Ca 2ϩ -activated currents, CaCl 2 was added either to the standard extracellular solution resulting in a concentration of 10 mM CaCl 2 or to the standard intracellular solution resulting in a concentration of 5 mM CaCl 2 . The intracellular solution for the Ca 2ϩ dose-response curve contained 140 KCl, 10 HEPES, 1 MgCl 2 , 10 EGTA. Ca 2ϩ was added as follows: 4.918 mM CaCl 2 for 100 nM free Ca 2ϩ , 7.421 mM CaCl 2 for 300 nM free Ca 2ϩ , 9.05 mM CaCl 2 for 1 M free Ca 2ϩ , 9.663 mM CaCl 2 for 3 M free Ca 2ϩ , 9.906 mM CaCl 2 for 10 M free Ca 2ϩ , 9.995 mM CaCl 2 for 30 M free Ca 2ϩ . The free Ca 2ϩ concentration was calculated by the software WCabuf (G. Droogmans, Leuven, Belgium).
The influence of Ca 2ϩ on icilin-induced currents was studied by adding 5 mM BAPTA to the standard intracellular solution or increasing the extracellular Ca 2ϩ concentration to 5 mM CaCl 2 .
For excised-patch recordings pipette and bath solutions were symmetric (in mM): 140 NaCl, 10 HEPES, 2 EGTA. Solutions with nano-or micromolar concentrations of free Ca 2ϩ were obtained by adding CaCl 2 to this EGTA-based buffer as follows: 1.012 mM CaCl 2 for 100 nM free Ca 2ϩ , 1.509 mM CaCl 2 for 300 nM free Ca 2ϩ , 1.823 mM CaCl 2 for 1 M free Ca 2ϩ and 1.940 mM CaCl 2 for 3 M free Ca 2ϩ .
Chemicals were prepared as concentrated stock solutions in either distilled water or Me 2 SO and diluted to the final concentration using standard extracellular solution or standard pipette solution as indicated in the text. AITC, carvacrol, icilin, and BAPTA were obtained from Sigma; ruthenium red (RR), U73122, and calmidazolium (CMZ) were purchased from Calbiochem; CALP2 was obtained from Tocris bioscience; and cinnamaldehyde was from Henkel.
Electrophysiological Recordings-All recordings were performed with an EPC7 amplifier (List-Medical Electronic, Darmstadt, Germany). Data were acquired using Pulse software (HEKA, Lambrecht, Germany). Excised patches were sampled at 2 kHz and filtered at 1 kHz. Patch pipettes were pulled from borosilicate glass (GC150TF-10, Harvard Apparatus Ltd.) and fire polished to 3-5 M⍀ tip resistance using a horizontal pipette puller (Zeitz Instruments, Munich, Germany). Solution exchange was achieved by placing cells in front of a theta-capillary and moving manually from one side of the outlet to the other. Excised patches were placed in front of a microperfusion pipette and solution exchange was achieved by switching from one solution to another under computer control.
Data Analysis-Electrophysiological data were analyzed using the software Pulse (HEKA), IgorPro (Wavemetrics), Sig-maPlot (SPSS Science), OriginPro (Origin Lab Corp.), TAC (Bruxton), and Microsoft Excel. Significance was tested using Student's independent t test (p Ͻ 0.05 is marked by an asterisk). The dose-response curve was fitted with a Hill equation of the form y ϭ base ϩ (max-base)/[1 ϩ (x half /x) n ]. Data are presented as mean Ϯ S.E.
Western Blotting-HEK293 cells were transfected with equal amounts of cDNA for wild-type TRPA1 and EF-hand mutants. Whole cell lysates were prepared 24 h after transfection, mixed with Laemmli buffer (30% glycerol, 3% SDS, 125 mM Tris-HCl, pH 6.8) and heated at 95°C for 5 min. Equal amounts of protein were loaded and resolved by 8% SDS-PAGE and transferred to nitrocellulose membrane (Protran; Schleicher & Schuell). The nitrocellulose membranes were stained with Ponceau S (Sigma) and blocked with TBST (150 mM NaCl, 50 mM Tris-HCl, Tween 20, pH 7.4) containing 2% ECL Advance Blocking Agent (Amersham Biosciences). TRPA1 was detected using two reported antibodies directed either against the C terminus of mouse TRPA1 (1:500) or the N terminus of mouse TRPA1 (1:1000) (10,15). The primary antibodies were diluted in 2% ECL Advance Blocking Agent in TBST. After washing and incubation with horseradish peroxidase-coupled secondary antibody, detection was performed with ECL Advance (Amersham Biosciences) on Hyperfilm ECL (Amersham Biosciences).

RESULTS
Icilin Requires Ca 2ϩ for Its Agonist Efficacy-Ca 2ϩ is known to modulate agonist-induced responses of various TRP channels (16 -18). Voltage-clamp recordings from TRPA1-expressing Xenopus oocytes or HEK293 cells have shown that AITCinduced responses are potentiated by external Ca 2ϩ ions (4, 10). We were interested to examine the effect of external Ca 2ϩ ions TRPA1 Is Directly Gated by Ca 2؉ MAY 4, 2007 • VOLUME 282 • NUMBER 18 on agonist efficacy of various TRPA1 agonists. Therefore, we performed whole-cell patch clamp recordings from HEK293 cells transiently transfected with cDNA of hTRPA1 and applied AITC, cinnamaldehyde, carvacrol, and icilin in the presence and absence of external Ca 2ϩ ions at a holding potential of V h ϭ Ϫ60 mV (Fig. 1). As expected, AITC (25 M) induced a slowly developing current in the absence of external Ca 2ϩ ions. Addition of 2 mM CaCl 2 to the bath solution evoked a strong potentiation of the inward current (Fig. 1A). The same was true for cinnamaldehyde (500 M) and carvacrol (500 M). The currents developed slowly in the absence of external Ca 2ϩ ions and were boosted when Ca 2ϩ was replenished in the extracellular recording solution ( Fig. 1, B and C).
In contrast, icilin (100 M) displayed very little or no agonist activity in the absence of external Ca 2ϩ ions (40.3 Ϯ 15.5 pA, Fig. 1, D and F). This low icilin efficacy under Ca 2ϩ -free conditions does not seem to reflect a decrease of agonist potency, since increasing icilin concentration from 100 to 500 M did not produce larger inward currents under Ca 2ϩ -free conditions (Fig.  1E). A recovery of icilin agonist efficacy was observed when Ca 2ϩ was added to the extracellular recording solution (Fig. 1

, D and E).
These data let us suggest that Ca 2ϩ serves as co-agonist with icilin by interacting directly with TRPA1, in a manner resembling the effect of Ca 2ϩ on icilin efficacy at TRPM8 (16). Raising the extracellular Ca 2ϩ concentration from 2 to 5 mM did not significantly increase icilin-induced currents (1815 Ϯ 302 pA, Fig.  1F). To test for the hypothesis that intracellular Ca 2ϩ is required for icilin efficacy, we added BAPTA (5 mM) to the pipette solution while recording in standard extracellular solution containing 2 mM CaCl 2 . Interestingly, icilin-evoked currents were significantly reduced (863 Ϯ 235 pA, p ϭ 0.048) compared to control conditions with standard intracellular solution (1586 Ϯ 222 pA), indicating a role of intracellular Ca 2ϩ for icilin agonist efficacy (Fig. 1F).
Taken together, our results argue for a distinct mechanism of activation by icilin as compared with other TRPA1 agonists.
Ca 2ϩ Influx Is Sufficient to Activate TRPA1-In light of the clear dependence of agonist activity on Ca 2ϩ , we asked whether Ca 2ϩ itself is sufficient to activate TRPA1. Whole-cell voltage-clamp recordings in standard extracellular solution for extended periods of time revealed the activation of an inward current that developed after on average 331 Ϯ 58 s in the absence of any agonist in 12 out of 15 recorded cells ( Fig. 2A). The current first showed a slow activation kinetic, which was boosted when the initial current reached on average 26.3 Ϯ 3.3% of the total current. Both the initial current and the sudden sharp increase in current were blocked by 1 M

TRPA1 Is Directly Gated by Ca 2؉
RR, indicating that both components of the biphasic response are mediated through TRPA1 activation (Fig. 2B). Ca 2ϩ may enter through spontaneously active TRPA1 channels, thereby triggering further channel opening until the intracellular Ca 2ϩ concentration reaches a certain threshold level responsible for the sudden sharp increase in current.
To clarify whether the leak influx of extracellular Ca 2ϩ activates TRPA1, we recorded under the same conditions at a positive holding potential of V h ϭ ϩ60 mV. The driving force for Ca 2ϩ is reduced at depolarized membrane potentials resulting in less Ca 2ϩ entry. Therefore, we expected the biphasic response to be abolished or delayed at V h ϭ ϩ60 mV. Indeed, no sudden sharp increase in current developed within the given time frame of 8 min in which currents developed at Ϫ60 mV (Fig. 2C). Application of AITC (10 s, 25 M) elicited currents, confirming the expression of TRPA1 in the recorded cells. The hypothesis that Ca 2ϩ leak influx activates TRPA1 is further supported by the finding that the biphasic current activation is induced faster when extracellular Ca 2ϩ is increased to 10 mM (59 Ϯ 6 s, Fig. 2D). Again, this current was completely blocked by 1 M RR (data not shown) and was absent in untransfected HEK293 cells (Fig. 2E). Taken together, these results indicate that influx of extracellular Ca 2ϩ into the cell is sufficient to activate TRPA1.
Increase in Intracellular Ca 2ϩ Elicits TRPA1 Currents-To investigate the impact of intracellular Ca 2ϩ on activation of TRPA1, we directly increased the intracellular Ca 2ϩ concentration to 5 mM resulting in a free Ca 2ϩ concentration of 23 M. A prominent inward current developed almost instantaneously after establishing the whole-cell configuration in cells recorded either in standard extracellular solution (219 Ϯ 38 pA) or Ca 2ϩfree solution (214 Ϯ 68 pA) ( Fig. 3A and B, V h ϭ Ϫ60 mV). This current was blocked by 1-5 M RR (Fig. 3A) and was absent or only exiguous in untransfected HEK293 cells (35 Ϯ 10 pA, Fig.  3C), supporting the finding that intracellular Ca 2ϩ triggers TRPA1 activation.
Next, we quantified the sensitivity of TRPA1 channels to intracellular Ca 2ϩ concentrations ranging from 100 nM to 30 M. We chose a holding potential of V h ϭ Ϫ80 mV to increase the driving force for Na ϩ ions, which should elicit larger inward currents. Ca 2ϩ was added to an EGTA-buffered pipette solution to define various Ca 2ϩ concentrations. The dose-response relationship was fitted with a Hill equation determining an EC 50 of 905 Ϯ 249 nM and a Hill-Coefficient of 0.9 Ϯ 0.2 (Fig. 3D).
Ca 2ϩ Activates TRPA1 in Excised Inside-out Patches-In whole-cell recordings many receptors or ion channel proteins might be modulated by intracellular processes or factors like enzymatic activity or cytosolic signaling molecules. To reduce a possible influence of cytosolic factors that might control the behavior of channels, we examined the effect of Ca 2ϩ on excised patches from transfected HEK293 cells in the insideout configuration. Interestingly, application of Ca 2ϩ in nanomolar concentrations to the intracellular side of the membrane was sufficient to elicit TRPA1 single-channel currents with amplitudes of on average Ϫ9.5 pA at a membrane potential (V m ) of Ϫ80 mV resulting in a single-channel conductance of 119 Ϯ 6.3 pS (Fig. 4, A and B). Application of higher Ca 2ϩ concentrations (1-3 M) led to activation of more channels or even induced macroscopic currents in some patches (Fig. 4A). Furthermore, we observed desensitization of Ca 2ϩ -induced currents (see Fig. 4A).
Consistent with the Ca 2ϩ dose-response curve for TRPA1 in whole-cell recordings (see Fig. 3D), we also found dose dependence in inside-out patches (Fig. 4C).
In untransfected control cells we observed an endogenous Ca 2ϩ -sensitive ion channel with a single-channel conductance of 40 pS at V m ϭ Ϫ80 mV, which interferes with the overexpressed TRPA1 in transfected cells (Fig. 4D). Further analysis of single-channel currents in transfected cells during voltage ramp-protocols showed that currents evoked by Ca 2ϩ (3 M) or AITC (25 M) have identical reversal potentials and rectification properties arguing that Ca 2ϩ -activated currents are due to TRPA1 activation (Fig. 4E).
Ca 2ϩ Activates TRPA1 in a PLC-independent Fashion-In light of the clear activation of TRPA1 by intracellular Ca 2ϩ we asked whether Ca 2ϩ directly gates TRPA1 or via a PLC-dependent signaling pathway. Since endogenous PLC activity remains preserved in inside-out patches (19) we stimulated inside-out patches with 3 M Ca 2ϩ after 30 s preincubation with the PLCinhibitor U73122 (10 M). Ca 2ϩ was still able to induce TRPA1mediated currents (Fig. 4F), indicating that Ca 2ϩ may directly gate TRPA1 probably by binding to a high affinity Ca 2ϩ -binding site at the channels cytosolic side.
TRPA1 Exhibits a Putative EF-hand CBD-While screening for Ca 2ϩ -binding domains, we identified a putative EF-hand motif at the N terminus of TRPA1 (Asp-468-Leu-480). To test for the hypothesis that Ca 2ϩ activates TRPA1 by binding through the EF-hand CBD, we used a 12-mer Ca 2ϩ like peptide, CALP2, known to function as antagonist at the EFhands of calmodulin and troponin C (20, 21). We examined the effect of CALP2 on Ca 2ϩ -induced TRPA1 single-channel currents in excised inside-out patches at V m ϭ Ϫ80 mV. Interestingly, Ca 2ϩ -induced (1 M) activity was strongly reduced in the presence of CALP2 (50 M). The open probability decreased by ϳ90%, whereas the single-channel amplitude was unaffected (8.3 Ϯ 0.5 pA before CALP2; 7.7 Ϯ 1.7 pA during CALP2) (Fig. 5A). The inhibitory effect of CALP2 was reversed by removal of the peptide and channel activity reverted back to the prior level (Fig. 5A). To exclude that CALP2 blocks the channel pore, we evaluated the effect of CALP2 on AITC-induced currents (Fig. 5B). Co-application of CALP2 (50 M) had no influence on AITC-induced (25 M) open probability nor on single-channel currents (13.1 Ϯ 4.3 pA before CALP2; 14.5 Ϯ 2.8 pA during CALP2). It should be noted that AITC-induced current amplitudes varied extremely (Fig. 5B) and reached on average larger amplitudes as compared with Ca 2ϩ -induced currents (Fig. 5A).
Up to now a role of calmodulin in TRPA1 function has not been reported. However, CALP2 might interact with calmodulin bound to TRPA1. To exclude a role of calmodulin in Ca 2ϩdependent activation, we preincubated the excised patches with CMZ, a known calmodulin inhibitor. We used a concentration known to be active in other calmodulin-dependent processes (10 M) (22,23). After 20 s preincubation we exposed the patches to 1 M Ca 2ϩ in the presence of CMZ. Ca 2ϩ still elicited TRPA1 single-channel currents (8.9 Ϯ 3 pA), arguing that calmodulin is not involved in Ca 2ϩ -dependent activation of TRPA1 (Fig. 5C).
A Single Mutation within the EF-hand CBD Impairs Ca 2ϩ Sensitivity-Based on the effect of CALP2 on Ca 2ϩ -induced single-channel currents, we were interested to see whether sensitivity of TRPA1 to Ca 2ϩ could be limited to a single residue within the EF-hand CBD. The EF-hand motif of TRPA1 appears to be highly conserved within different spe- cies (Fig. 6A). To identify the site(s) responsible for Ca 2ϩ binding, we performed alanine scanning mutagenesis for the residues proposed to be involved in Ca 2ϩ binding (Fig. 6B) (12). The resulting mutants (D468A, S470A, T472A, L474A, N476A, and D479A) were expressed in HEK293 cells and expression was first confirmed by Western blot analysis (Fig.  6C). Mouse TRPA1 antibodies recognized an expected band of ϳ128 kDa for wild-type TRPA1 and EF-hand mutants, indicating effective expression of all mutants. It should be noted that although the antibodies only recognized one band for the mutant T472A, this was of lower molecular weight than predicted. No band was detected in the whole cell extract from untransfected cells (Fig. 6C).
To analyze for functionality of the mutant receptors, we verified the sensitivity to AITC (25 M) before examining changes in Ca 2ϩ sensitivity of the mutants. Mutations at positions Asp-468, Ser-470, and Leu-474 were found to retain AITC sensitivity (Fig. 6D). Although slight differences were observed in mean current amplitudes elicited after 10 s application of AITC (Fig. 6E), currents showed identical reversal potentials and rectification properties for mutants as compared with the wild-type channel, arguing for plenary functionality of the mutants D468A, S470A, and L474A (Fig. 6, D  and E). No currents could be observed for cells expressing the T472A, N476A, and D479A mutants (Fig. 6D). Even prolonged exposure to AITC (25 M, 60 s) did not induce any inward or outward currents (Fig. 6D). To exclude the possibility that these mutants lost sensitivity to AITC, we applied cinnamaldehyde (500 M, 60 s). Again, we failed to elicit any reliable currents (data not shown). Therefore, we concentrated in further studies on the mutants D468A, S470A, and L474A which showed unaltered AITC responses.
To examine the Ca 2ϩ sensitivity of the EF-hand mutants D468A, S470A, and L474A, we compared mean current amplitudes elicited with a saturating Ca 2ϩ concentration (10 M, see Fig. 3D) in the pipette solution while recording in Ca 2ϩ -free extracellular solution at V h ϭ Ϫ80 mV.
A prominent inward current with on average 999 Ϯ 140 pA developed almost instantaneously after establishing the whole-cell configuration in cells expressing wild-type TRPA1 (Fig. 6, F and J). Mutations at positions Asp-468 and Ser-470 were found to retain Ca 2ϩ sensitivity analogue to wild-type TRPA1 (Fig. 6, G and H). Currents reached on average 1057 Ϯ 130 pA (D468A) and 1259 Ϯ 186 pA (S470A) (Fig. 6J). Interestingly, substitution of leucine in position 474 with alanine resulted in an almost Ca 2ϩ -insensitive channel (69 Ϯ 23 pA), indicating that the leucine in position 474 is likely to be involved in Ca 2ϩ -dependent activation (Fig. 6, I and J). The expression of the L474A mutant was confirmed by application of AITC (25 M, 40 s) and recording of strong inward currents (Fig. 6I).
Taken together, these results highlight the importance of a single residue within the EF-hand CBD in mediating sensitivity of TRPA1 to Ca 2ϩ .
Agonist Activity Is Diminished in Cells Expressing the L474A Mutant-In the first section we showed that Ca 2ϩ potentiates agonist induced responses in TRPA1 expressing cells. In light of the finding that a single mutation of the residue Leu-474 results in a Ca 2ϩ -insensitive channel, we next verified the sensitivity of this mutant to AITC in the presence and absence of extracellular Ca 2ϩ . In Ca 2ϩ -free solution currents elicited by AITC (25 M) were about the same for wild-type TRPA1 and the L474A mutant (Fig. 7, A-C, wild-type 1856 Ϯ 345 pA, L474A 1928 Ϯ 479 pA). As expected, addition of 2 mM CaCl 2 to the bath solution resulted in a strong potentiation of the wild-type response (Fig. 7, A and C, 3066 Ϯ 430 pA). Interestingly, the L474A mutant showed no potentiation of AITC-induced responses by Ca 2ϩ (Fig. 7B). Addition of 2 mM CaCl 2 to the bath solution resulted in a reduction of current amplitudes (Fig. 7, B  and C, 1423 Ϯ 281 pA), presumably reflecting the reported transition of the channel to a state with lower conductance (10).
Whereas Ca 2ϩ was found to potentiate AITC-induced responses, it was shown to be an essential co-agonist for icilin efficacy on TRPA1 (see Fig. 1). Thus, we tested icilin to elicit currents in cells expressing the Ca 2ϩ -insensitive L474A mutant. Performing whole-cell recordings in standard extracellular solution containing 2 mM CaCl 2 showed no substantial currents (9 Ϯ 5 pA), consistent with a loss of Ca 2ϩ activation of the L474A mutant (Fig. 7, D and F). In contrast mutations at positions Asp-468 and Ser-470 were found to retain sensitivity for icilin analogue to wild-type TRPA1 (Fig. 4, E and F, wild-type 1217 Ϯ 256 pA, D468A 1159 Ϯ 199 pA, S470A 1310 Ϯ 284 pA).
Taken together, the results indicate that the residue in position Leu-474 participates in Ca 2ϩ -mediated potentiation of agonist (AITC)-induced responses and also contributes to icilin agonist efficacy on TRPA1.

DISCUSSION
In the present study, we investigated the effects of Ca 2ϩ on the gating behavior of the Ca 2ϩ -permeable cation channel TRPA1. We showed that Ca 2ϩ potentiates responses to various TRPA1 agonists and serves as an essential co-agonist for icilin activity. We further demonstrated that Ca 2ϩ activates TRPA1 and shed light on the mechanism in which Ca 2ϩ may directly gate the channel.
Several TRP channels are reported to be regulated or modulated by Ca 2ϩ . For example, gating of TRPV4 depends on both extra-and intracellular Ca 2ϩ (18), and TRPM2 activation requires Ca 2ϩ (24). Here, we show that agonist efficacy of some TRPA1 agonists is modulated by Ca 2ϩ . As reported for AITC (4, 10), we found that cinnamaldehyde and carvacrol responses are potentiated by Ca 2ϩ , suggesting a common mechanism for these agonists. In contrast, we observed icilin to require Ca 2ϩ for its agonist efficacy, arguing for a distinct mechanism of activation. Whereas concomitant exposure of AITC and Ca 2ϩ displays synergy, the necessity of simultaneous exposure of icilin and Ca 2ϩ points toward a mechanism of co-dependence. Our data obtained by mutagenesis support the hypothesis for distinct mechanisms of activation by AITC and icilin. We describe a single residue within the N-terminal EF-hand CBD to participate in Ca 2ϩ -mediated modulation of agonist-induced responses. Ca 2ϩ binding to leucine in position 474 appears to be involved in potentiation of AITC-induced responses and seems to play a major role for icilin agonist efficacy on TRPA1. The ineffective-ness of icilin on the Ca 2ϩ -insensitive L474A mutant argues for a more complex mechanism in icilindependent activation of TRPA1 different from that previously reported for TRPM8 (16). Possibly Ca 2ϩ primarily induces activation of TRPA1 and icilin acts modulatory on the Ca 2ϩ -induced responses. Since icilin activates TRPA1 currents with variable delay of onset in the presence of Ca 2ϩ (data not shown) resembling the activation kinetics of Ca 2ϩ -influx induced currents (see Fig. 2), an appropriate scenario is conceivable. However, it remains to be determined whether icilin modulates the response thereby facilitating Ca 2ϩ -mediated activation of the channel rather than directly provoking TRPA1 channel opening.
Activation of TRP channels by Ca 2ϩ ions has been reported previously for the monovalent-selective channels TRPM4 and TRPM5 (25,26). Although a prior study suggested that TRPA1 can be activated by store depletion (4), the finding is still disputed as a second study failed to reproduce activation of TRPA1 by store depletion (7). Other studies showed that TRPA1 currents can be activated downstream of G protein-coupled receptors (4, 7), arguing for a role of second messengers in TRPA1 activation. Importantly, neither previous study tested whether Ca 2ϩ directly activates TRPA1 channel.
Evidence for Ca 2ϩ -dependent activation of TRPA1 is obtained from our excised-patch and whole-cell recordings. The data show that Ca 2ϩ influx through spontaneously active TRPA1 channels is sufficient to initiate further channel opening. Intracellular Ca 2ϩ activates the channel with an apparent EC 50 of 905 nM in whole-cell recordings. The exposure of excised inside-out patches to Ca 2ϩ elicits TRPA1 single-channel currents and the number of Ca 2ϩ -activated channels increased with higher Ca 2ϩ concentrations.
In general, Ca 2ϩ can bind and activate proteins either by itself or by binding through adaptor proteins. A direct interaction of TRPA1 with Ca 2ϩ was analyzed using pharmacological approaches and site-directed mutagenesis. Our data suggest that Ca 2ϩ activates TRPA1 in a PLC-and calmodulin-independent fashion. CALP2 effectively reduces Ca 2ϩ -induced channel activity and a mutation of a single residue within the N-terminal EF-hand CBD induces loss of Ca 2ϩ -dependent activation, favoring the N-terminal EF-hand domain as a putative Ca 2ϩbinding site.
Within the EF-hand motif six residues are described to be involved in binding of the Ca 2ϩ ion (12). Accordingly, we introduced point mutations at these sites and screened the channel  7). E, same as in D but from a representative cell overexpressing the S470A mutant. Analogue to the wild-type, the S470A mutant retained sensitivity to icilin (n ϭ 5). F, average inward currents evoked by 100 M icilin for wild-type TRPA1 in standard extracellular solution containing 2 mM CaCl 2 (n ϭ 7) or Ca 2ϩ -free solution (n ϭ 7), for mutant L474A (n ϭ 7) in standard extracellular solution, for mutant D468A (n ϭ 5) in standard extracellular solution, and for mutant S470A (n ϭ 5) in standard extracellular solution. Icilin-evoked responses are significantly reduced in cells overexpressing the L474A mutant. mutants by examining their functionality and Ca 2ϩ sensitivity. Of the six potentially involved residues, we observed that a single mutation at position Leu-474 altered Ca 2ϩ sensitivity of TRPA1. We find that Ca 2ϩ -dependent activation is strongly impaired for the L474A mutant, assuming that the leucine contributes to Ca 2ϩ sensitivity of TRPA1. The D468A and S470A mutants remain functional and exhibit wild-type responses to a saturating Ca 2ϩ concentration, suggesting that mutations at these sites may be accommodated without affecting Ca 2ϩ activation. Reports describe a considerable variability in length and amino acid sequence of the N-terminal part of the Ca 2ϩ -binding loop in EF-hand proteins (27). In natural EF-hand loops the planar position 3 is one of the most variable positions. Depending on the type of mutation the effect can vary over a wide range from nearly no effect up to virtually preventing Ca 2ϩ binding (28). The residue in the last coordinating position is reported to be required for the pentagonal bipyramidal coordination geometry and provides commonly two oxygen atoms for liganding Ca 2ϩ (27). Point mutations at this site are postulated to decrease the Ca 2ϩ binding affinity of EF-hand proteins (29). Whether the same is true for TRPA1 is not clear, as the D479A mutant displayed no activity upon application of AITC or cinnamaldehyde. Similar findings are observed for the T472A and N476A mutants. Whether this site(s) is essential to the basic opening and closing of the channel pore or whether the level of expressed protein at the membrane is to low to elicit any reliable currents remains an open question.
Although our experiments identify the residue Leu-474 within the EF-hand motif as involved in Ca 2ϩ -dependent activation of TRPA1 and although CALP2 reduces Ca 2ϩ -induced channel activity, further analysis comprising Ca 2ϩ binding studies are needed to determine the N-terminal EF-hand CBD as Ca 2ϩ -binding site in TRPA1.
In any event, the observation that Ca 2ϩ modulates agonistinduced responses and directly activates TRPA1 channels with high sensitivity has important functional consequences for the physiology and pathophysiology of TRPA1 expressing cells. TRPA1 has recently been shown to mediate the inflammatory actions of proalgesic agents (9) that commonly act via stimulation of G protein-coupled receptors linked to a PLC. The resulting increase in intracellular Ca 2ϩ would be sufficient to activate TRPA1 currents. In general, all processes that lead to an increase in intracellular Ca 2ϩ might directly stimulate TRPA1 or alternatively sensitize the channel and amplify agonist provoked responses. Environmental irritants that directly stimulate either TRPA1 or other Ca 2ϩ -permeable ion channels will increase intracellular Ca 2ϩ levels, which in turn would enhance TRPA1 activity. Since TRPA1 co-expresses with the Ca 2ϩ -permeable channel TRPV1 in native tissues (2, 4), a modulation of TRPA1 activity downstream of TRPV1 activation is likely. Stimulation of TRPV1 along with the accompanying Ca 2ϩ -influx could affect TRPA1 activity thereby modulating the initial excitatory response. Since TRPA1 conducts Ca 2ϩ ions with P Ca /P Na ϭ 0.84 (2), activation and modulation by Ca 2ϩ represents a regenerating and self-amplifying positive feedback loop that augments the initial Ca 2ϩ and voltage signal. This positive feedback mechanism requires tight control to avoid danger of cellular Ca 2ϩ overload. As for other receptors, the activity of TRPA1 is likely to be controlled by a Ca 2ϩ -dependent negative feedback mechanism (17). The excessive Ca 2ϩ influx is supposably shut down by a Ca 2ϩ -dependent desensitization process (10) that minimizes the Ca 2ϩ conductivity of TRPA1 in a selflimiting process.