Calmodulin is responsible for Ca2+-dependent regulation of TRPA1 Channels

TRPA1 is a Ca2+-permeable ion channel involved in many sensory disorders such as pain, itch and neuropathy. Notably, the function of TRPA1 depends on Ca2+, with low Ca2+ potentiating and high Ca2+ inactivating TRPA1. However, it remains unknown how Ca2+ exerts such contrasting effects. Here, we show that Ca2+ regulates TRPA1 through calmodulin, which binds to TRPA1 in a Ca2+-dependent manner. Calmodulin binding enhanced TRPA1 sensitivity and Ca2+-evoked potentiation of TRPA1 at low Ca2+, but inhibited TRPA1 sensitivity and promoted TRPA1 desensitization at high Ca2+. Ca2+-dependent potentiation and inactivation of TRPA1 were selectively prevented by disrupting the interaction of the carboxy-lobe of calmodulin with a calmodulin-binding domain in the C-terminus of TRPA1. Calmodulin is thus a critical Ca2+ sensor enabling TRPA1 to respond to diverse Ca2+ signals distinctly.

(F) Summary of TRPA1 potentiation induced by different Ca 2+ in experiments similar to those in (E). The number of experiments is given above each bar. Significance was compared to bar 1. All data are mean ± SEM. NS, not significant; **P < 0.01; ***P < 0.001. activation, inhibition and desensitization. For clarity, both TRPA1 inhibition and desensitization are termed inactivation, though it is unknown whether the two processes employ the same mechanism.
To define the [Ca 2+ ] e range that triggers potentiation and inactivation, respectively, we plotted TRPA1 currents elicited by carvacrol as a function of different concentrations of [Ca 2+ ] e , yielding a biphasic response curve (Fig. 1D). Based on this curve, [Ca 2+ ] e evoking the maximal TRPA1 response was estimated to be around 1 mM (Fig. 1D). Ca 2+ below 1 mM is thus defined as low Ca 2+ . Within this range, Ca 2+ mainly promoted TRPA1 activation with little inhibition or desensitization (Fig. 1D), a process manifesting Ca 2+ -dependent potentiation (CDP). By contrast, the inhibition and/or desensitization of TRPA1 became more and more prominent with further increases in Ca 2+ at the high Ca 2+ range (i.e. > 1 mM), resulting in gradually reduced TRPA1 responses, a process reflecting Ca 2+ -dependent inactivation (CDI), in addition to CDP (Fig. 1D, also see below).
We next examined TRPA1 potentiation induced by low and high Ca 2+ , respectively, in real-time. Figure 1E shows that carvacrol elicited stable inward TRPA1 currents in nominal 0 Ca 2+ . The currents were then rapidly potentiated shortly after Ca 2+ was perfused. Of note, Ca 2+ -induced TRPA1 potentiation progressively enhanced with increases in perfused Ca 2+ , with a peak potentiation at around 1 mM Ca 2+ (Fig. 1E and F), consistent with predicated peak TRPA1 response at 1 mM Ca 2+ in Fig. 1D. Notably, peak TRPA1 currents were not desensitized in the presence of carvacrol at low Ca 2+ (Fig. 1E), showing that low Ca 2+ induces a pure CDP process. By contrast, high Ca 2+ -induced peak TRPA1 currents rapidly reduced following initial potentiation, a process known as acute desensitization (Fig. 1E), leading to a gradual reduction in the overall CDP. CDP at high Ca 2+ is thus a consequence of counteracting actions of concurrent CDP and CDI. Collectively, Ca 2+ is a key player in multiple TRPA1 functions including the basal sensitivity, CDP and CDI. TRPA1 binds to CaM depending on Ca 2+ . We next investigated how TRPA1 senses and responds to different Ca 2+ levels. Previous studies did not consistently support the idea that direct binding of Ca 2+ to TRPA1 mediates the Ca 2+ effects (see introduction), we wondered whether Ca 2+ act through an intermediate protein, which is not only sensitive to Ca 2+ , but also binds to TRPA1, forming a Ca 2+ -sensitive channel complex. CaM is such a candidate with remarkable Ca 2+ -sensing capability. Indeed, CaM mediates Ca 2+ gating of many ion channels, such as TRP, SK, Na + and Ca 2+ channels 34-39 . To test this hypothesis, we first examined whether CaM binds to TRPA1 using CaM pull-down assay. TRPV1 is known to be regulated by CaM 36 , and was thus used as a positive control. We found that CaM bound to much more TRPA1 than to TRPV1 in the presence of Ca 2+ . The binding was, however, absent without either Ca 2+ or TRPA1/TRPV1 co-expression ( Fig. 2A). Notably, the most potent binding to CaM was observed with Scientific RepoRts | 7:45098 | DOI: 10.1038/srep45098 TRPA1 among thermo-TRP channels (Fig. 2B), suggesting a high binding affinity of CaM for TRPA1. Indeed, TRPA1-CaM binding can even be triggered by nanomolar ranges of Ca 2+ (Fig. 2C), implying that TRPA1 binds to CaM in the resting [Ca 2+ ] i (∼ 100 nM). Furthermore, despite a much lower affinity of CaM for Ba 2+ over Ca 2+ ions 40 , Ba 2+ elicited substantial TRPA1-CaM binding, albeit reduced when compared to Ca 2+ (Fig. 2D), further supporting a high binding affinity between TRPA1 and CaM. Consistent with Ba 2+ -induced TRPA1-CaM binding, Ba 2+ was capable of regulating TRPA1 (see below). However, Mg 2+ did not induce significant TRPA1-CaM binding (Fig. 2D), and accordingly did not modulate TRPA1 (see below).
In support of CaM pull-down assay, TRPA1 was found to bind to CaM in a co-immunoprecipitation experiment (Fig. 2E). CaM contains two different lobes (N-and C-lobe) and each lobe contains two EF hand domains responsible for Ca 2+ binding. CaM C-lobe exhibits a higher binding affinity for Ca 2+ (Kd, 10 −7 M) than N-lobe (Kd, 10 −6 M) 41 . The binding of Ca 2+ to CaM can thus be selectively disrupted by mutating two different lobes 42 . There was about a 40% reduction in TRPA1-CaM binding after selectively preventing Ca 2+ binding to N-lobe by mutating the first two Ca 2+ binding sites (EF 12 ) on N-lobe (CaM 12 mutant) ( Fig. 2E and F). Mutating the third (EF 3 ) and the fourth Ca 2+ binding sites (EF 4 ) on C-lobe (CaM 3 , CaM 4 mutants) caused a 60% and an 87% reduction in TRPA1-CaM binding, respectively ( Fig. 2E and F). TRPA1-CaM binding was, however, completely abolished by mutating both Ca 2+ binding sites on C-lobe (CaM 34 mutant), or by Ca 2+ -insensitive CaM 1234 in which all four Ca 2+ -binding sites are mutated 34 (Fig. 2E and F). Therefore, binding of CaM to TRPA1 critically depends on Ca 2+ loading of CaM, and different Ca 2+ -binding sites on CaM play distinct roles in TRPA1 binding: CaM C-lobe is more important in triggering TRPA1 binding, while N-lobe may only play an auxiliary role in the binding process. Even within CaM C-lobe, the role of EF 3 and EF 4 is not identical, with EF 4 having a higher capability of triggering TRPA1 binding than EF 3 .

CaM binds to a non-canonical CaM-binding domain (CaMBD) in TRPA1.
CaM binds to a variety of molecular targets through several classes of CaMBD such as the "IQ" motif in Ca 2+ channels [43][44][45] . However, TRPA1 does not contain any known canonical CaMBDs. To delineate a CaMBD on TRPA1, the cytoplasmic Nand C-terminal fragments of TRPA1 coupled to the GST tag were first purified and then used for CaM pull-down assay. CaM bound prominently to the C-terminus of TRPA1, but not to GST, though there was a negligible binding to the N-terminus (Fig. 3A). Similar results were also obtained with Flag pull-down assay in which purified Flag-coupled TRPA1 cytoplasmic tails were used to pull down pure CaM (Fig. 3B). To further narrow down the binding, we truncated the C-terminus of TRPA1 progressively. Deleting the distal 114 amino acids (AA) (R1012-F1125) enhanced CaM binding, but a further deletion of 17 residues (L995-N1011) completely eliminated CaM binding (Fig. 3C), suggesting that the 17AA is a key CaMBD.
To validate whether the 17AA is a bona fide CaMBD, we synthesized a peptide identical in sequence to the 17AA. As anticipated, the binding of CaM to the C-terminus of TRPA1 was abrogated by incorporating the peptide in CaM pull-down assay (Fig. 3D). A scrambled peptide was, however, ineffective (Fig. 3D). As a positive control, the binding was also abolished by the CaM antagonistic peptide CALP2 (Fig. 3D) 46 . The specificity of the CaMBD peptide was further tested on full-length TRPA1. Consistently, TRPA1-CaM binding was almost abolished by the CaMBD peptide, but not by the scrambled peptide (Fig. 3E).
To verify whether CaM binds directly to the CaMBD peptide forming a CaM-peptide complex, we incubated the peptide with pure CaM in different molar ratios and then resolved protein complexes on non-denaturing PAGE gels followed by silver staining for visualization. CaM did not exhibit a mobility shift after incubation with the peptide (Fig. 3F, top panel), presumably the peptide is not large enough (17aa) to alter CaM mobility. But intriguingly, the intensity of CaM gradually enhanced with increasing ratios of peptide to CaM in the presence of Ca 2+ . The effect was abolished by removing Ca 2+ ( Fig. 3F and G), suggesting the formation of a Ca 2+ -dependent CaM-peptide complex, which contains more amino acid side chains for binding to silver ions than pure CaM, resulting in enhanced silver stain. These experiments conclusively demonstrate that the 17 AA in the C-terminus of TRPA1 is a bona fide CaMBD and that CaM binds directly to TRPA1.
We next deleted the CaMBD from TRPA1. As expected, CaM binding was dramatically reduced in CaMBD-lacking TRPA1 (Δ TRPA1), though incompletely ( Fig. 3H), suggesting the presence of other minor CaM binding regions elsewhere in TRPA1, likely in the N terminus.
Of note, the CaMBD is rich in hydrophobic residues, consistent with the binding preference of CaM 44 . Interestingly, based on the recently resolved TRPA1 structure 47 , the CaMBD corresponds precisely to the β -strand domain with unknown function flanked by the TRP-like domain and coiled-coil domain in the C-terminus of TRPA1 (Fig. 3I), two important structural domains critical for modulating TRPA1 gating 47 . Notably, the CaMBD is freely exposed in the periphery of TRPA1 structure ( Fig. 3I) 47 , making this domain physically accessible to CaM.
CaM is essential for TRPA1 CDP. To test a possible role for CaM in regulating TRPA1, we first investigated whether CaM potentiates TRPA1, as does low Ca 2+ . Indeed, CaM over-expression potently increased TRPA1 responses to all doses of carvacrol in nominal 0 [Ca 2+ ] e (Fig. 4A), suggesting that TRPA1 is not fully occupied by endogenous CaM under the resting condition. The effect of CaM was, however, absent in Ca 2+ -free CaM 1234 , which was also deficient for TRPA1 binding (Figs 2E and 4A). These data demonstrate that CaM binding is sufficient to potentiate the basal sensitivity of TRPA1 even without changes in [Ca 2+ ] e . It is likely that the basal [Ca 2+ ] i (∼ 100 nM) is sufficient to trigger enhanced TRPA1-CaM binding in the presence of excess CaM.
We then examined whether CaM mediates CDP. To isolate CDP from CDI and to avoid the mutual interference of the two processes, we used the Ca 2+ range between 10 μ M and submaximal 0.5 mM that elicits no appreciable desensitization. TRPA1 current evoked by carvacrol was rapidly potentiated by 10 μ M Ca 2+ (Fig. 4B). A similar potentiation was also observed with Ba 2+ , though to a less degree ( Fig. 4C and D). Ca 2+ -induced TRPA1 potentiation was significantly enhanced when cells overexpressed CaM or CaM 12 or CaM 3 ( Fig. 4B and D), consistent with their ability to bind to TRPA1 (Fig. 2E). Such enhancement was absent in cells expressing CaM 4 , or CaM 34 or CaM 1234 mutants (Fig. 4D), all of which exhibited little (CaM 4 ) or no binding (CaM 34 , CaM 1234 ) to TRPA1 (Fig. 2E), suggesting that CaM binding is critical for TRPA1 CDP and that EF 4 on CaM C-lobe is indispensable for this process. Enhanced CDP may be caused by a larger Ca 2+ influx due to increased initial TRPA1 responses evoked by carvacrol in the presence of overexpressed CaM. However, there were no positive correlations between the initial peak current amplitudes and CDP (r = 0.044, Fig. 4E), suggesting that enhanced CDP is not due to different [Ca 2+ ] e entry. Taken together, increases in either Ca 2+ or CaM can elicit CDP. Ca 2+ and CaM thus act cooperatively to potentiate TRPA1 under low Ca 2+ .
To determine whether CDP could be prevented by disrupting TRPA1-CaM binding, we took advantage of the potent binding between CaM and TRPA1 CaMBD, and substituted the CaMBD for the cytoplasmic tail in Tac antigen, an α -subunit of interleukin-2 receptor with a single membrane-spanning domain 48 , to produce a Tac-A1-CaMBD chimera (Fig. 4F). We then co-expressed Tac-A1-CaMBD to sequester endogenous CaM. Co-expressed Tac-A1-CaMBD markedly reduced CDP induced by 0.5 mM Ca 2+ , and completely abolished CDP evoked by 10 μ M Ca 2+ (Fig. 4F and G). In contrast, CDP was not significantly affected by Tac co-expression ( Fig. 4F and G). Similarly, TRPA1 CDP was also blocked by the CaM antagonist W-7 ( Fig. 4F and G), validating an essential role for CaM in eliciting TRPA1 CDP. TRPA1 CDP was also observed in DRG neurons (Fig. 4H). Importantly, CDP was significantly prevented by including the CaMBD peptide in the pipette, while a scrambled peptide was without effect (Fig. 4H and I). Collectively, these experiments demonstrate a critical role for CaM in governing TRPA1 sensitivity and CDP. CaM is critical for TRPA1 CDI. To investigate a role of CaM in regulating TRPA1 CDI, we first examined whether CaM inhibits TRPA1 in high Ca 2+ (> 1 mM). Indeed in 2 mM Ca 2+ , TRPA1 currents evoked by all doses of carvacrol were robustly inhibited by over-expressed CaM (Fig. 5A and B), an effect analogous to that induced by higher 10 mM Ca 2+ (Fig. 1B). Thus, increases in either Ca 2+ or CaM inhibit TRPA1 in the high Ca 2+ range. The inhibitory effect was also observed with the N-lobe mutant CaM 12 , but not with the C-lobe mutant CaM 34 or Ca 2+ -insensitive CaM 1234 (Fig. 5B), suggesting a critical role for CaM C-lobe in this process. Similar effects were also observed with another TRPA1 agonist AITC (Fig. 5C). In contrast to a previous report 21 , the sensitivity of TRPA1 was also inhibited by blocking endogenous CaM with the CaM antagonist W-7 (Fig. 5D), further supporting the idea that CaM is essential for the basal sensitivity of TRPA1.
We then investigated whether CaM is involved in TRPA1 desensitization. To this end, TRPA1 was activated by consecutive pulses of carvacrol. Peak TRPA1 current was typically reduced in the second activation resulting in tachyphylaxis (Fig. 5E). CaM overexpression promoted TRPA1 tachyphylaxis (Fig. 5E and F). To measure acute desensitization, TRPA1 was maximally activated by AITC to allow desensitization to fully evolve following initial channel activation (Fig. 5G). Remarkably, CaM overexpression accelerated the desensitization rate of TRPA1, leading to a marked reduction in the time constant of desensitization ( Fig. 5G and H), despite overall smaller peak currents with CaM ( Fig. 5G and I). The accelerated desensitization of TRPA1 was not affected by CaM 12 , CaM 3 and CaM 4 , all of which retained TRPA1 binding, but was absent in the C-lobe mutant CaM 34 and CaM 1234 , both of which were deficient for TRPA1 binding (Figs 2E and 5H), suggesting that CaM binding is also crucial for inducing TRPA1 desensitization and that EF 34 in CaM C-lobe is critical in this process. Taken together, CaM inhibits TRPA1 sensitivity in the basal state and promotes TRPA1 desensitization in the activation state of the channel in high Ca 2+ , and both processes require CaM C-lobe.
To investigate whether TRPA1 desensitization could be prevented by disrupting TRPA1-CaM binding, we used the CaM chelator Tac-A1-CaMBD and the CaMBD peptide. In HEK293 cells expressing TRPA1, co-expression of Tac-A1-CaMBD significantly reduced the desensitization rate of TRPA1 induced by 2 mM Ca 2+ (Fig. 5J and L). An analogous effect was also observed with the CaM antagonist W-7 ( Fig. 5J and L). In native DRG neurons, Currents induced by the TRPA1 agonist AITC underwent similar rapid desensitization ( Fig. 5K and L). The acute desensitization was significantly prevented by including in the pipette the CaMBD peptide, but not by a scrambled peptide (Fig. 5L), further demonstrating that CaM is critical for TRPA1 desensitization. As an alternative approach, we used Ba 2+ , which elicited a reduced TRPA1-CaM binding compared to Ca 2+ (Fig. 2D). Accordingly, Ba 2+ prevented desensitization induced by consecutive pulses of carvacrol ( Fig. 5M and N). These experiments further demonstrate that CaM binding is required for TRPA1 desensitization.

CaM binding sites responsible for TRPA1 CDP and CDI. To identify CaM effector sites on TRPA1
responsible for CDP and CDI, we mutated 16 residues on CaMBD individually to the negatively charged glutamic acid, in an attempt to disrupt TRPA1-CaM interaction. To investigate the effect of these mutants on CDP, the mutated channels were activated using lower dose of carvacrol in order to avoid saturating TRPA1 mutants based on their dose-response curves (data not shown). TRPA1 CDP was abolished by mutating W996, R999, and P1010 ( Fig. 6A and B). A pronounced deficit in CDP was also observed in Y997E, V1008E and Y1009E TRPA1 (Fig. 6A and B). Interestingly, three of these mutants (W996E, V1008E and P1010E) also exhibited significant impairment in acute desensitization induced by AITC (Fig. 6A and B). The three sites are thus critical for transducing both CDP and CDI. In contrast, three other mutants (i.e. Y997E, R999E, and Y1009E) exhibited selective deficit in CDP without an impairment in CDI (Fig. 6A and B). Notably, impaired desensitization in W996E, V1008E and P1010E was rescued by over-expressing CaM ( Fig. 6A and C), showing that impaired desensitization is due to diminished CaM binding and is not intrinsic to these mutants. On the other hand, V1001E exhibited an enhanced potentiation and accelerated desensitization (Fig. 6B). A significant enhancement in CDP was also observed in Q1003E and S1005E (Fig. 6B). Presumably, mutating these sites caused over-compensation of CaM binding to other functional sites such as W996. Neither potentiation nor desensitization was affected in other mutants (Fig. 6C).
We also tested whether CDP and CDI are affected in Δ TRPA1 in which CaMBD has been deleted, but found that Δ TRPA1 was non-functional (data not shown). The lack of function in Δ TRPA1 could be due either to impaired CaM binding or to structural disturbance. Of note, tryptophan is a conserved residue targeted by CaM among many other CaM effectors, such as CaMKII 49 , underscoring a pivotal role for W996 in governing Ca 2+ gating of TRPA1. Taken together, these results demonstrate that TRPA1 CDP and CDI are mediated by overlapping but distinct sets of effector sites on the CaMBD of TRPA1.
We finally demonstrated whether CaM binding is disrupted in the TRPA1 mutants with defective CDP/CDI. In silver staining of CaM-peptide complex, enhanced CaM staining due to bound CaMBD peptide was significantly diminished by mutating either W996, or V1008 or P1010 (Fig. 3F and G), suggesting that CaM-peptide binding is disrupted in these mutants. A further experiment from co-immunoprecipitation also validated that CaM binding was significantly impaired in the W996E, V1008E and P1010E mutants (Fig. 6D and E).

Discussion
Ca 2+ exerts multiple and opposing effects on TRPA1. However, it has been a mystery how TRPA1 senses different Ca 2+ levels (Ca 2+ sensor) and translates Ca 2+ signals into the gating of the channel (Ca 2+ effector), leading to opposing effects. In this study we have found that Ca 2+ gating of TRPA1 depends on CaM, which binds to TRPA1, forming a Ca 2+ -sensing channel complex. In the basal state, CaM either enhances (low Ca 2+ ) or inhibits (high Ca 2+ ) TRPA1 sensitivity. In the activated state, CaM promotes either TRPA1 activation (low Ca 2+ ) or desensitization (high Ca 2+ ). These effects were prevented by either mutating the Ca 2+ -binding sites on CaM or by mutating the CaM binding sites on TRPA1. CaM thus acts as a Ca 2+ sensor and an effector responsible for regulating TRPA1 sensitivity and activation in both the basal state and activation state.
Unexpectedly, the multiple Ca 2+ effects were mediated by CaM C-lobe, without a significant role for N-lobe in these processes. The two lobes of CaM are thus not the source that drives opposing Ca 2+ effects on TRPA1. This is in contrast to CaV 2.1 and TMEM16 chloride channels that employ the C-lobe and N-lobe of CaM to transduce channel facilitation and inactivation, respectively 45,50 . Interestingly, EF 3 and EF 4 in the C-lobe triggered different degrees of TRPA1 binding (Fig. 2E), suggesting that EF 3 and EF 4 have different Ca 2+ -binding capabilities and maybe the origin responsible for detecting different Ca 2+ levels for TRPA1. Consistently, EF 3 and EF 4 participate in different aspects of TRPA1 regulation. Specifically, EF 3 is only involved in TRPA1 CDI, but EF 4 participates in both TRPA1 CDP and CDI. In keeping with distinct roles of EF 3 and EF 4 in TRPA1 regulation, some CaM effector sites on TRPA1 mediated only CDP, while others transduced both CDP and CDI (Fig. 6B). It is thus conceivable that TRPA1 CDP and CDI may be mediated by two different forms of interactions between CaM C-lobe and CaM effector sites on TRPA1, with one form of interaction promoting CDP and another favouring CDI. This possibility could arise from different binding stoichiometries between TRPA1 and CaM.
It is noteworthy that over-expressed CaM can still exert an additional effect on TRPA1 CDP/CDI, supporting that TRPA1 is far from saturated by endogenous CaM. TRPA1-CaM binding can even be triggered by nanomolar ranges of Ca 2+ (Fig. 2C), suggesting that basal [Ca 2+ ] i is sufficient to cause a preassociation of CaM with TRPA1, which may explain why TRPA1 still exhibited Ca 2+ sensitivity in inside-out excised patches 21-23 , a patch configuration in which all associated intracellular molecules (e.g. CaM) are assumed to have been lost. It is very likely that residual CaM remains associated with excised TRPA1 channels due to, for example, local Ca 2+ microdomains 51 , which could prevent the complete loss of associated CaM. However, Ca 2+ -free CaM 1234 (apoCaM) did not pre-associate with TRPA1, in contrast to other ion channels, such as L-type Ca 2+ channels 43,52 . The lack of binding of apoCaM with TRPA1 explains why CaM 1234 had no dominant-negative action on TRPA1, which is the main evidence leading to the exclusion of a possible role for CaM in TRPA1 regulation in a previous study 21 .
Our results support that the binding of CaM C-lobe is critical for Ca 2+ -dependent regulation of TRPA1. Indeed, all the TRPA1-CaM binding can be eliminated by the C-lobe mutant CaM 34 (Fig. 2E). In contrast, no associated deficits in TRPA1 modulation were observed in the N-lobe mutant CaM 12 , even though CaM 12 caused a 40% reduction in TRPA1 binding, suggesting that N-lobe may only play an accessory role in the binding process. It is possible that once loaded with Ca 2+ , CaM C-lobe acts as both a tether anchoring CaM to CaMBD in TRPA1 and an effector transmitting Ca 2+ -dependent channel gating, whereas Ca 2+ -loaded N-lobe may trigger the binding of CaM to other regions in TRPA1, such as the N terminus, acting as a second tether but without influencing channel gating. This possibility is supported by the evidence that there was a weak binding between CaM and the N terminus of TRPA1 ( Fig. 3A and B) and that deleting CaMBD from C-terminal TRPA1 did not completely abolish CaM binding (Fig. 3H). However, it remains to be tested whether there is a second CaMBD in the N terminus of TRPA1 contributing to Ca 2+ -dependent modulation of TRPA1.
In this case, a proximal N-terminal ankyrin repeat (AR) domain in TRPA1 has been implicated as a critical region for mediating TRPA1 desensitization 29 . However, we did not find the binding of CaM to the AR domain (data not shown). The AR domain is thus unlikely to be a second CaMBD. A most likely second CaMBD seems to lie in the region nearby the linker domain and/or pre-S1 helix in the N terminus of TRPA1. These regions are structurally in close proximity to the C-terminal CaMBD (Fig. 3I) 47 , and could cooperate with each other to modulate TRPA1 CDP and CDI.
Scientific RepoRts | 7:45098 | DOI: 10.1038/srep45098 The C-terminal CaMBD of TRPA1 is adjacent to the ion permeation pathway allowing CaM to detect rapid Ca 2+ oscillations, agreeing with previous studies demonstrating that [Ca 2+ ] e regulates TRPA1 through binding to a channel site that needs to be very close to the channel pore 22,30 . Notably, the identified six CaM effector sites on TRPA1 are very close either to the TRP domain or to the coiled-coil domain (Fig. 3I), two important structural domains implicated in TRPA1 gating. It is conceivable that a slight conformational change in these effector sites due to CaM binding could have significant impact on either the TRP domain or the coiled-coil domain or both, leading to distinct functional outcomes. The unique position of CaMBD thus not only enables TRPA1 to sense Ca 2+ , but also allows the channel to translate Ca 2+ signals into channel gating. The revealed accessible CaMBD in TRPA1 could thus be a potential drug targeting region for tuning the channel for therapeutics.
Molecular biology. Thermo-TRP ion channels including TRPV1-4, TRPM8 (rat) and TRPA1 (mouse) cDNAs were subcloned into pcDNA3-V5-His-TOPO vector (Life Technology) as described previously 56 . Calmodulin (rat) cDNA was a kind gift of Dr. Ruth Lagnado (University of Sussex). CaM was added a HA tag to the N-terminus and subcloned into the pcDNA3.1 vector (Invitrogen) using standard PCR via KpnI and XbaI. The interleukin-2 receptor (Tac) cDNA was obtained from Dr. Juan Bonifacino (NIH, USA). To produce chimeric Tac-A1-CaMBD, the cytoplasmic tail in the C-terminus of Tac (T260-I272) was replaced by the corresponding C-terminus of mTRPA1 (G965-N1011) containing CaMBD via HinIII and XbaI using standard PCR. GST-coupled N-and C-terminus of TRPA1 were constructed by amplifying the corresponding cytoplasmic fragments (N-terminus: 1 M-R719; C-terminus: G965-F1125) followed by in-frame subcloning into a GST-pcDNA3 vector via BamHI and EcoRI. The GST-pcDNA3 vector was prepared by fusing the GST tag amplified from the pGEX-2T vector (GE healthcare) to the pcDNA3 vector (Invitrogen) via HindIII and BamHI. The Flag tag (DYKDDDDK) was also added to the C-terminus of cytoplasmic fragments of TRPA1 by PCR followed by subcloning into the pcDNA3.1 vector via KpnI and XbaI. To generate truncated GST-coupled C-terminus of TRPA1, a stop codon was introduced at relevant sites using the Quick-Change mutagenesis kit (Agilent Technologies). Quick-Change mutagenesis was also used to prepare all other mutations. We also employed Quick-Change mutagenesis to generate CaMBD-deleted TRPA1 using partially overlapping primers lacking the CaMBD region. All the constructs and mutations were validated by DNA sequencing.

Pull down assay and co-immunoprecipitation. CaM pull down assay was performed by incubating
CaM-agarose (Sigma) with HEK293 cell lysate expressing TRPA1-V5-His (× 6) or other TRP channels in a lysis buffer (20 mM Tris-HCl, pH, 7.4, 150 mM NaCl, 1% NP-40 plus protease inhibitor cocktails (Roche)) at 4 °C under different Ca 2+ or Ba 2+ concentrations as indicated in related Figures. For CaM pull down assay in Fig. 3H and co-immunoprecipitation in Fig. 6E, TRPA1 and its mutants containing the 6× His tag expressed from HEK293 cells were first purified using Ni-NTA beads (Qiagen) as described previously 55 . After thorough washing, TRPA1 and mutant proteins were then eluted with 250 mM imidazole (Sigma). Equal amount of eluted TRPA1 proteins was then used for CaM pull down assay or co-immunoprecipitation. To obtain Ca 2+ -free, 0.5 mM EGTA was included to deplete background Ca 2+ in the solution. Similarly in experiments for studying the effect of Ba 2+ , corresponding Ba 2+ ions were added to the above Ca 2+ -free solution to exclude the effect of background Ca 2+ . For CaM pull down in Fig. 2C, free Ca 2+ of 100 nM and 500 nM were obtained by adding Ca 2+ (in mM) of 0.28 and 0.422, respectively, to the above Ca 2+ -free binding solution buffered by 0.5 mM EGTA. Free Ca 2+ concentrations were calculated using the programme MaxChelator (University of Stanford). After 4 times of thorough washing (20 mins each) with corresponding free Ca 2+ , bound proteins were eluted by boiling in sample buffer followed by separation on a 7.5% SDS-PAGE gel and detection with anti-V5 (Life Technologies). A peptide (NH 2 -LWYLRKVDQRSTIVYPN-COOH) with identical sequence to CaMBD in TRPA1 and corresponding scrambled peptide (NH 2 -YNQIRVYKVTPRLSLDW-COOH) (Biomatik) were also included in CaM pull down assay to investigate the effect of peptide on TRPA1-CaM binding.
To examine the binding of CaM to cytoplasmic tails of TRPA1, GST-coupled TRPA1 cytoplasmic tails expressed from HEK293 cells were purified using GST-agarose (Sigma) followed by elution with Glutathione (10 mM), similarly as previously described 55 . Purified N-and C-terminus of TRPA1 coupled to GST were then incubated with CaM-agarose for CaM pull down assay. Bound GST-coupled TRPA1 fragments were then eluted with 100 mM EGTA followed by separation on a 10% SDS-PAGE gel and blot detection with anti-GST (GE Healthcare). Similarly, Flag-coupled N-and C-termini of TRPA1 expressed from HEK293 cells were first isolated using Flag-agarose (Sigma) and purified through thorough washing. Flag-pull down assay was then performed by incubating pure CaM (Sigma) with the isolated Flag-tagged TRPA1 fusion proteins bound to Flag agarose, followed by extensive washing and subsequent elution with the Flag peptide (Sigma). Eluted CaM was next analysed on 10% SDS-PAGE and detected by anti-CaM (Millipore). Flag-tagged fusion proteins were probed by anti-Flag (Sigma). Co-immunoprecipitation between CaM and TRPA1 was performed similarly as described previously 55,56 . Briefly, HA-CaM or related mutants was co-expressed with TRPA1-V5 in HEK293 cells followed by solubilisation. HA-CaM was then precipitated by monoclonal anti-HA.11 (Covance) and Protein A/G PLUS-Agarose (Santa Cruz Biotechnology). Co-precipitated TRPA1 was then resolved on a 10% SDS-PAGE gel followed by immunodetection with anti-V5. The band intensity of blots was quantified using Image J.
Silver staining of CaM-peptide complex was performed as described by others with slight modifications 52 . Briefly, 600 nM CaM was incubated with different molar ratios of peptides in a buffer containing 10 mM Scientific RepoRts | 7:45098 | DOI: 10.1038/srep45098 Na-HEPES (pH 7.2) and 2 mM Ca 2+ or without Ca 2+ (5 mM EGTA) at room temperature for 1 h. The formed protein complexes were then resolved on 12% non-denaturing polyacrylamide gels in the presence of 2 mM Ca 2+ or without Ca 2+ (2 mM EGTA) followed by staining with a silver kit (Sigma) in accordance with the manufacturer's instructions.
Electrophysiology. Whole-cell recordings were conducted using Axopatch 200B patch clamp amplifier (Axon) controlled by pClampx 10.2 softwares (Molecular Device) as before 55,56 . The basic extracellular solution (nominal 0 Ca 2+ ) consisted of (in mM): 140 NaCl, 4 KCl, 1.0 MgCl 2 , 10 HEPES, 5 Glucose, pH 7.4 with NaOH. Ca 2+ -free solution was prepared by adding 5 mM EGTA to the basic extracellular solution. Solutions with millimolar free Ca 2+ or Ba 2+ were obtained by adding Ca 2+ or Ba 2+ directly to the basic extracellular solution. Internal pipette solution contained (in mM): 140 KCl, 2.0 MgCl 2 , 5.0 EGTA, 10 HEPES, pH 7.4 with KOH. To obtain 10 μ M free Ca 2+ , 1.99 mM Ca 2+ was added to a Ca 2+ -free solution buffered by 2 mM EGTA. To study the effect of the CaM antagonist W-7, cells were perfused with 100 μ M W-7 (Tocris) throughout the recording. To study the effect of peptides on TRPA1 currents in DRG neuros, peptides were included in the pipette to dialyse the cells as described previously 56 . All the recordings were held at − 60 mV and signals were analogue filtered using a 1 kHz low-pass Bessel filter. Time constants (τ ) of desensitization induced by AITC were determined by fitting TRPA1 peak currents with the equation y = A * exp (− x/τ ) + C using the Clampfit 10.2 software (Molecular Devices).
Statistics. All data were presented as mean ± SEM. Differences between two groups were determined using Student's t test and was considered significant if P < 0.05. Comparisons among multiple groups were assessed by one-way analysis of variance with Bonferroni's post hoc test.