Research ArticleBiochemistry

Intracellular cavity of sensor domain controls allosteric gating of TRPA1 channel

See allHide authors and affiliations

Science Signaling  23 Jan 2018:
Vol. 11, Issue 514, eaan8621
DOI: 10.1126/scisignal.aan8621

Regulatory cavity in TRPA1

The ion- and voltage-gated channel TRPA1 contributes to acute and chronic pain and itch, as well as vascular function. The channel is sensitive to temperature, pressure, various environmental irritants, and other molecules generated by injury or inflammation in an allosteric manner, meaning that the channel’s response to one stimulus alters its response to another. Zimova et al. investigated the structure of TRPA1 and found an intracellular region within the sensor domain that coordinates this complex allostery. For example, mutations in this cavity prevented stabilization of the channel by interacting intracellular phospholipids, resulting in dysregulated sensitivity to calcium ions. These findings suggest that alterations to the cavity may underlie chronic pain or itch in patients.

Abstract

Transient receptor potential ankyrin 1 (TRPA1) is a temperature-sensitive ion channel activated by various pungent and irritant compounds that can produce pain in humans. Its activation involves an allosteric mechanism whereby electrophilic agonists evoke interactions within cytosolic domains and open the channel pore through an integrated nexus formed by intracellular membrane proximal regions that are densely packed beneath the lower segment of the S1–S4 sensor domain. Studies indicate that this part of the channel may contain residues that form a water-accessible cavity that undergoes changes in solvation during channel gating. We identified conserved polar residues facing the putative lower crevice of the sensor domain that were crucial determinants of the electrophilic, voltage, and calcium sensitivity of the TRPA1 channel. This part of the sensor may also comprise a domain capable of binding to membrane phosphoinositides through which gating of the channel is regulated in a state-dependent manner.

INTRODUCTION

Sensory neurons receive and react to a variety of extracellular stimuli, including those that cause or threaten tissue damage. To detect pungent and proalgesic agents, these neurons use a sophisticated system of transduction molecules that respond to noxious stimuli by opening intrinsic ion channel gates. One such molecule, the transient receptor potential ankyrin 1 (TRPA1) channel (1), gates in response to a wide range of thiol-reactive electrophiles and oxidants and to a number of chemically unrelated irritants such as menthol, carvacrol, camphor, certain cannabinoids, and many others [reviewed in (24)]. The opening of this cation channel leads to an increase in intracellular calcium concentration, depletion of membrane lipids, and depolarization of the membrane, and all these signals, in turn, modulate the channel’s activity. Such a complexity of TRPA1 regulation at the molecular level requires the presence of a significant number of interacting domains, and despite the large number of identified sites to date, we are still far from a complete identification of them and understanding of their functions (5, 6).

All transient receptor potential (TRP) channels form functional tetramers, with each subunit consisting of six transmembrane segments (S1 to S6) flanked by N- and C-terminal cytosolic domains (Fig. 1A). The helices S1–S4 form isolated sensor domains (S1–S4) arranged radially around the periphery of the central ion-conducting pore, which is lined with four S5-reentrant pore loop-S6 domains. The recent high-resolution structures of TRPA1 and its relatives TRPV1, TRPV2, TRPV6, TRPP1, and TRPN1, captured in different conformations, have yielded valuable new insights into the general principles of TRP channel functioning (512). In particular, the structures of three of these channels (TRPV1, TRPP1, and TRPN1) were determined in a native bilayer environment, adding especially important information about the putative sites of the TRP channel’s interactions with annular and regulatory lipids (11, 1315). At the same time, the recent structures and sequence comparisons among TRPs point to apparent differences, suggesting different regulatory mechanisms (Fig. 1, B to D).

Fig. 1 Schematic of TRPA1 channel sensor domain.

(A) Inner cavity (light blue funnel) formed by the lower part of the S1–S4 sensor domain and adjacent structures. S5-P-S6 refers to the central ion-conducting pore. The cavity regulates the gating of the channel (gray arrows) right in the center of the integrated nexus formed by the web of interactions between the transient receptor potential (TRP)–like domain (brown), pre-S1 helix (helix preceding S1), and S4-S5 linker (violet). Red asterisks indicate TRPA1-activating stimuli. (B) The positions of the cavity-facing polar residues. Amino acids are annotated by their single-letter abbreviation and residue number. (C) The HOLLOW script (53) was used for a “casting” of the inner cavity of TRPA1 (model constructed in this study; based on the TRPA1 3J9P structure and TRPP1 structure 5K47) and TRPV1 (5IRZ) by filling the voids with dummy atoms defined on a grid. The inner cavity of the sensor in TRP channels is predicted to be hydrated, and the extracellular part is tightly packed with bulky hydrophobic residues (16). Note the differences between the predicted hydration in TRPV1 and TRPA1 sensors. (D) Amino acid sequence conservation within the S2-S3 region of TRPA1, TRPV1, and TRPV2 proteins (222, 183, and 293 sequences) represented as sequence logo (54). Residues participating in phosphatidylcholine binding to TRPV1 (yellow triangles). (E) Sequence alignment of the human TRPP1 and TRPA1 channels used for the homology modeling. The identical, strongly conserved, and weakly conserved residues are denoted with asterisk, double dots, and single dot marks, respectively.

Using approaches from information theory and probabilistic modeling, Palovcak et al. (16) recently performed a comparative sequence analysis of almost 3000 different TRP proteins and a large comprehensive ensemble (>3700) of distantly related voltage-gated potassium (Kv) channel sequences. Remarkable differences in the evolutionarily conserved features of these two families were found between their S1–S4 sensor domains. In Kv channels, the sensors can be hydrated from both the intracellular and extracellular sides of the membrane bilayer (17). Penetrating water interacts with polar residues and regulates the Kv channel’s voltage gating by shaping the transmembrane electric field (18). In contrast, solvation of the TRP’s sensor is predicted to be limited to the intracellular side, whereas its upper part is packed with hydrophobic and aromatic residues (16, 19). In the structurally and functionally best-characterized TRPV1 and TRPV2, the lower segment of the S1–S4 sensor domain contains highly conserved hydrophilic residues that may facilitate the solvation of this region (7, 8, 16). This portion of the channels directly interacts with endogenous lipids and C-terminal amino acids following the TRP box that probably help to stabilize the sensor during channel gating (9, 13, 20). The TRPA1’s sensor is also occupied by polar residues at the base of the S1–S4 helices and the S2-S3 loop (Fig. 1, B and C), but there is practically no clear homology across this domain between TRPV1, TRPV2, and TRPA1 at the primary sequence level (Fig. 1D).

The role that the lower vestibule of the S1–S4 sensor domain plays in TRPA1 regulation is of particular interest, because this region is an important component of an allosteric nexus at which activation signals are integrated and transmitted through the TRP-like domain to the intracellular channel gate (Fig. 1A). The specific structural features of TRPA1 apparently distinct from those of TRPV channels (5) raise intriguing questions and prompted us to use electrophysiology combined with molecular dynamics (MD) simulations and systematic mutagenesis to explore the hypothesis that conserved polar residues form a uniquely charged and functionally important cavity in the lower part of the TRPA1 sensor domain. We show that this region does more than just act to regulate the electrophilic, voltage, and calcium sensitivity of the TRPA1 channel. The sensor’s cavity is capable of directly binding membrane phosphoinositides, such as phosphatidylinositol 4,5-bisphosphate [PI(4,5)P2; also known as PIP2], through which the channel’s gating can be regulated in a state-dependent manner.

RESULTS

The intracellular side of the S1–S4 sensor domain forms a hydrated cavity

We constructed a model of human TRPA1 with the S1–S4 sensor domain using the 3J9P structure determined by cryo-electron microscopy (5). The intracellular loop connecting helices S2 and S3 was modeled with the use of sequence homology with the polycystin-2 TRP channel TRPP1 (Fig. 1E) on the basis of its structures with Protein Data Bank (PDB) IDs: 5K47 (12), 5T4D (14), and 5MKE and 5MKF (15) (see Materials and Methods for a detailed description of the procedure used for homology modeling and subsequent MD and ligand docking studies). The model showed that the lower cavity inside the S1–S4 sensor domain was highly hydrophilic. The following residues were oriented inward: His719 and Asn722 (S1); Lys787, Gln791, and Gln794 (S2); Lys796 (S2 and S3); Asp802, Ser804, Asn805, and Glu808 (S3); Tyr849 and Arg852 (S4); and Lys989 (TRP-like box). The S1–S4 sensor domain was surrounded by phospholipids; however, its bottom part was in contact with water molecules. MD simulations confirmed that water molecules can permeate and reside in the lower cavity of the S1–S4 sensor domain (Fig. 2A). Further, we calculated the electrostatic potential around the TRPA1 channel (3J9P completed with the S2-S3 linker as described above) and compared it with the electrostatic potential obtained for TRPV1 [PDB ID: 5IRZ (13)] (Fig. 2B). In TRPA1, there were a negative electrostatic potential in the selectivity filter and inside the pore and a positive electrostatic potential around the intracellular ankyrin moieties. Compared to TRPV1, a notable feature of TRPA1 was that the positive electrostatic potential permeated into the intracellular part of the S1–S4 sensor domain, where there is a cluster of basic amino acids composed of Lys787, Lys796, Arg852, Lys989, and, possibly, protonated His719.

Fig. 2 Modeling the sensor domain of human TRPA1.

(A) Model of the intracellular side of the S1–S4 sensor domain, which forms a hydrated cavity. The intracellular loop connecting helices S2 and S3 was modeled using sequence homology with TRPP1 (5K47, 5T4D, 5MKE, and 5MKF) as described in Materials and Methods. Residues His719, Asn722, Lys787, Lys796, Asp802, Asn805, Glu808, Arg852, and Lys989 were mutated in this study. (B) Schematic of electrostatic potential surrounding TRPA1 [Protein Data Bank (PDB) ID: 3J9P] and TRPV1 (PDB ID: 3J5P) structures was determined by means of visual molecular dynamics (41). Red mesh indicates a negative electrostatic potential in the selectivity filter and inside the pore. Compared to TRPV1, a notable feature of TRPA1 is that the positive electrostatic potential (blue mesh) permeates into the intracellular part of the S1–S4 sensor domain. (C) Model of the putative phosphatidylinositol 4,5-bisphosphate (PIP2) binding site on the S1–S4 sensor domain of TRPA1. Left: Homology model of the S1–S4 sensor domains of TRPA1. The negatively charged inositol trisphosphate head group of PIP2 contacts residues His719, Asn722, Lys787, Lys796, Arg852, and Lys989. Right: Four PIP2 molecules located in the four identical binding pockets (left) shown in the context of the template 3J9P structure completed with the model of the S1-S2 and S3-S4 linkers.

The S1–S4 sensor domain contains a putative PIP2 binding site

Recently, the determined structures of the TRPV1 and TRPV2 channels [PDB IDs: 5IRZ (13), 5AN8 (9), and 5HI9 (20)] revealed membrane lipids bound to the crevice formed by the S1–S4 helical bundle of the sensor above the TRP domain. Here, we found by ligand docking that TRPA1 was also capable of binding phospholipids including PIP2 at an analogous position. Subsequent MD simulations performed with TRPA1 in solution showed that the negatively charged inositol trisphosphate head group of PIP2 may adopt several different conformations within the S1–S4 sensor domain of TRPA1 to contact residues His719, Asn722, Lys787, Lys796, Arg852, and Lys989 (Fig. 2C).

Polar residues in the inner cavity of the sensor regulate the voltage-dependent gating of TRPA1

To assess the functional roles of polar amino acids predicted to face the cavity of the sensor domain of TRPA1, we individually mutated the polar residues located in this region (Figs. 1B and 2A) and measured the voltage-dependent activation properties of the mutants transiently expressed in human embryonic kidney (HEK) 293T cells using whole-cell electrophysiology. The conductance-to-voltage (G-V) relationships were assessed in control bath solution using a voltage step protocol from −80 to +200 mV, in 20-mV increments (Fig. 3A). Intracellular Ca2+ was routinely buffered to low levels with 5 mM EGTA in the patch pipette to prevent the synergistic potentiation of voltage-induced currents by permeating calcium ions (21). In other experiments, low-buffer intracellular solution (LB-ICS) containing 100 μM instead of 100 nM free Ca2+ was additionally used to assess the activation capacity of the less responsive mutants (Fig. 3B).

Fig. 3 Voltage-dependent gating of TRPA1 mutants.

(A) Representative family of whole-cell currents recorded from human embryonic kidney (HEK) 293T cells transfected with either wild-type human TRPA1 channel (WT) or the indicated mutants elicited with a voltage step protocol consisting of 100-ms depolarizing pulses from −80 up to +200 mV in steps of 20 mV and a holding potential of −70 mV. The voltage protocol is shown in the top right panel. Bath solution contained 160 mM NaCl, 2.5 mM KCl, 1 mM CaCl2, 2 mM MgCl2, 10 mM Hepes, and 10 mM glucose (adjusted to pH 7.3 and 320 mosmol). The currents were recorded ~1 min after whole-cell formation. Steady-state currents were measured at the end of the pulses as indicated by colored symbols atop each record. (B to D) Average conductances obtained from recordings as in (A). Data are means ± SEM (n = 132 cells for wild-type and n = 6 to 30 cells for mutants from at least two independent transfections). The lines represent the best fit to a Boltzmann function for wild-type and mutant TRPA1 (gray and colored lines) using high-buffer (solid lines) or low-buffer intracellular solution [(LB-ICS); dashed lines]. (E) Deactivation kinetics of TRPA1 mutants. Averaged tail currents normalized to the maximum amplitude at +200 mV obtained as indicated by dashed box in (A, top left) for the wild-type channel. The average currents are shown with color bars indicating means ± SEM [number of cells indicated in (B) to (D)]. The gray lines with gray bars (SEM) represent the averaged tail currents obtained from data for wild-type TRPA1. Dashed lines indicate zero current. (F) The average time constants obtained from single exponential fits of tail currents. The asterisks indicate a significant difference from wild-type channels [**P < 0.001; n as in (B) to (D)]. (G) Summary of half-activation voltage (V50) and apparent number of gating charges (z) from experiments in (B) to (D). *P < 0.05; analysis of variance (ANOVA) on ranks followed by Dunn’s test versus WT.

In cells expressing the H719A, N722A, N722I, K787A, K796A, R852A, and E808K mutant channels, the G-V curves were significantly shifted rightward compared to wild-type channels (Fig. 3, B to D). Notably, currents through the H719A and N722I channels were significantly suppressed by using LB-ICS in the pipette (Fig. 3B, middle and right), whereas currents through the less responsive K787A mutant channels were not significantly affected (Fig. 3C, left), indicating the involvement of these residues in Ca2+ sensitivity. The E788A, E788K, N805A, E808A, and K989A mutants exhibited significantly increased currents at positive membrane potentials. The E788A and E808A constructs had much steeper G-V relationships, reflecting a changed apparent number of gating charges from z = 0.69 eo for wild-type channels to 0.80 and 0.94 eo for E788A and E808A, respectively. The most notable effect of the mutations at His719, Asn722, Lys787, Lys796, Glu788, and Glu808 was that these channels closed extremely quickly upon repolarization from +200 to −70 mV, resulting in faster tail currents (Fig. 3, A, E, and F). Because inward currents were significantly attenuated at hyperpolarized voltages in these mutants, the steady-state half-activation voltage (V50) values and z derived from G-V relationships could only be reliably estimated over the positive voltage range (Fig. 3, B to D and G). D802A channels opened slowly and exhibited slowly decaying tail currents that remained transiently open upon repolarization to −70 mV (Fig. 3E). The R797A mutant displayed significantly increased basal conductance at negative potentials (Fig. 3C, right), suggesting that this mutation disturbed the closed-open equilibrium in favor of the open state.

Together, these data indicated that the neutralizations of negative charges quite deep in the cavity facilitated conformational transitions upon voltage stimulation, and the other polar residues also critically contributed to the voltage-dependent gating of TRPA1. Of these, H719 and K787 appeared to play a central role, because their neutralization mostly disrupted the ability of the channel to gate in response to depolarization. Likewise, substitution of the crevice facing asparagine N722 with a large hydrophobic residue disrupted voltage-dependent gating.

Mutations deep in the cavity strengthen Ca2+-induced inactivation

To test the overall chemical sensitivity of the mutants, we used a previously established protocol (21, 22) in which currents were first induced by the electrophilic agonist in the absence of external Ca2+ (Fig. 4, A to L, and fig. S1). The agonist was then washed out for 10 s, and 2 mM Ca2+ was added to the extracellular solution to assess the allosteric effects of permeating calcium ions. The membrane potential was linearly ramped up each second from −80 to +80 mV (1 V/s). In the standard procedure, cinnamaldehyde (100 μM) or allyl isothiocyanate (AITC; 100 μM) was used as a partial and a full agonist, whereas intracellular Ca2+ was buffered to low levels with 5 mM EGTA in the patch pipette. The less responsive mutants were also tested using the LB-ICS. The average responses through wild-type channels (Fig. 4, A and G) were in full accordance with the ones from previous papers (21, 22). In H719A, N722I, and K787A, the AITC-induced currents were slowly developing and outwardly rectifying (Fig. 4, B, D, and E). These changes were even more evident from the rectification ratio (−current at −80 mV/current at +80 mV) plotted as a function of time (fig. S2). The currents through N722A channels activated slowly but, ultimately, within 40 s, reached their maximum amplitude at +80 mV similar to the wild-type channels (Fig. 4C). K787A was partially rescued by increasing intracellular Ca2+ in the patch pipette (Fig. 4E), and the currents resembled the AITC responses obtained from R852A channels [Fig. 4F; (23)]. In the above mutants, the addition of Ca2+ to the bath solution induced an immediate inactivation that was almost complete at negative membrane potentials. The average responses to cinnamaldehyde (100 μM) through the K796A and R797A channels were consistently much larger than those through wild-type channels (Fig. 4, H and I), and their rectification ratio was apparently increased in the presence of external Ca2+ (fig. S2, H and I). Responses to cinnamaldehyde through D802A were initially smaller, but the addition of Ca2+ to the bath solution potentiated the currents so that their magnitudes reached the maximum responses of the wild-type TRPA1 (Fig. 4J). The AITC currents through D802A were slower, their average amplitudes exceeded the wild-type responses at negative potentials, and the addition of external Ca2+ inactivated the currents more intensively than in the wild type (Fig. 4K). The K989A channels were constitutively active in the absence of external Ca2+ and characterized by saturating cinnamaldehyde-induced currents that were not further potentiated by external calcium (Fig. 4L).

Fig. 4 Mutations in the inner cavity of sensor module affect chemical-dependent gating of TRPA1.

(A) Time course of average whole-cell currents induced by 100 μM allyl isothiocyanate (AITC) measured at +80 and −80 mV in HEK293T cells transfected with wild-type TRPA1 (open circles). Inset shows voltage-ramp protocol used for measuring currents. The cells were first exposed to the electrophilic agonist (AITC) in the absence of external Ca2+ using the bath solution containing 2 mM HEDTA. The agonist was then washed out for 10 s, and 2 mM Ca2+ was added to the extracellular solution as indicated above the current traces. Data are mean + SEM (open circles; n = 34 cells). In some cases, the error bars are smaller than the symbol. The dashed line represents the average currents obtained for WT using low-buffer intracellular solution (LB-ICS) (n = 11 cells). Zero current is indicated by the horizontal line. (B to F) Time course of average AITC-induced currents recorded from the indicated mutant channels using either high-buffer intracellular solution (circles) or LB-ICS (squares). Data are means + SEM (n = 6 to 13 cells). The average current for WT is overlaid as a gray line with gray bars indicating mean + SEM (high-buffer pipette solution) or dashed gray line representing the average current for WT obtained with LB-ICS. (G) Average whole-cell currents induced by 100 μM cinnamaldehyde (CA) in Ca2+-free solution and then exposed to 2 mM Ca2+ measured at +80 and −80 mV in WT. The application of CA and subsequent addition of 2 mM Ca2+ are indicated above. Data are mean + SEM (n = 45 cells). (H to L) Average currents recorded from mutant channels. The average current for the WT is shown as a gray line with bars indicating + SEM. Data are mean + SEM (n = 7 to 9 cells). In some cases, the error bars are smaller than the symbol.

Together, these results suggested that the mutations at residues His719, Asn722, Lys787, Asp802, Arg852, and Lys989 induced functional defects manifested as significant changes in rectification that depend on the activation state of the channel, indicating that these residues are important for the allosteric coupling between voltage and agonist sensing and gate opening. Moreover, the prominent and immediate inactivation by external Ca2+ observed in some of these mutants suggested that neutralizations in the inner cavity of the sensor may also strengthen the allosteric coupling between the putative domain(s) responsible for Ca2+-dependent inactivation and the channel’s gate. In D802A, the time constant of Ca2+ inactivation was independent of the agonist used and the extent of channel activation (23 ± 1 s and 25 ± 1 s for cinnamaldehyde and AITC, respectively; Fig. 4, J and K), suggesting that the external Ca2+ stimulus avoids the electrophile-dependent activation machinery and works through a separate pathway.

Glu808 interacts with key Lys787, whereas Glu788 helps to transduce the signals to the gate

The E788A and E808A mutants displayed robust responses to cinnamaldehyde, reaching more than twice the maximum currents obtained from the WT-TRPA1 at +80 mV (Fig. 5, A and B). The cinnamaldehyde-induced currents mediated by E808A exhibited a slow onset at negative membrane potentials, as was visible from the time course of the rectification ratio (Fig. 5B, bottom). Although, in wild-type channels, the rectification ratio typically increased slightly upon stimulation with cinnamaldehyde from about 0.5 to 0.7 in 40 s; the currents through E808A reached a maximum rectification of only about 0.4 and displayed a shift in gating equilibrium toward positive membrane potentials and a hindered opening at negative potentials. The mutation E788K produced channels that readily opened in response to cinnamaldehyde, exhibiting a rectification ratio of about 0.9, suggesting a close-to-saturation state (Fig. 5C, bottom). The charge-reversal mutant E808K closely resembled the double-mutant E788A/E808A in that the currents recorded during the 40-s application of cinnamaldehyde were fully superimposable onto those obtained with WT-TRPA1 (Fig. 5, D and E). Notably, in all the constructs above, the subsequent exposure to extracellular Ca2+ caused a rapid inactivation of cinnamaldehyde-induced currents, instead of potentiation seen with wild-type channels. Except for E788K, the Ca2+-induced block was more pronounced at negative membrane potentials. AITC-induced responses were saturated in E788A, E788K, and E808A, whereas in E808K and E788A/E808A, they did not reach the wild-type level (fig. S1), indicating the activation-dependent roles of these residues.

Fig. 5 Mutations in internal sensor domain increase the sensitivity of TRPA1 and strengthen Ca2+-induced inactivation.

(A to E) Time course of average cinnamaldehyde (100 μM)–induced currents recorded from HEK293T cells transfected with either wild-type TRPA1 channel (WT) or the indicated mutants. The overlaid gray line with gray bars represents the average currents + SEM obtained for wild-type TRPA1 as in Fig. 4G. Below each panel, average rectification of currents shown above (−current at −80 mV/current at +80 mV) plotted as a function of time. Colored symbols and lines with gray bars indicate means + SEM (n = 45 cells for WT and n = 7 to 9 cells for mutants). (F) Bottom view of the inner cavity of TRPA1 sensor domain. (G) Representative current traces from E788I in response to voltage step protocol shown in Fig. 3A recorded in control extracellular solution containing 1 mM Ca2+ (left; indicated above as 1 mM [Ca2+]o) or in Ca2+-free bath solution (right; indicated above as 0 mM [Ca2+]o). Bottom: Averaged tail currents recorded from cells expressing E788I (lines with yellow and pink bars indicating means ± SEM; n = 20 and 16 cells, respectively) and WT (superimposed gray lines with gray bars indicating means ± SEM; n = 132 and 10 cells, respectively). (H to I) Average conductances obtained from HEK293T cells transfected with the indicated mutants compared with WT human TRPA1 channel. The currents elicited with a voltage step protocol were measured at the end of the pulses as indicated by colored circles atop the traces in (G). The average conductance obtained from the WT in control extracellular solution (n = 132 cells) is shown as a gray line. Data are means ± SEM for WT measured in Ca2+-free bath solution (open circles; n = 10 cells) and for N805A and Y799A recorded in control extracellular solution containing 1 mM Ca2+ (colored circles; n = 15 and 13 cells, respectively). Solid lines are best fits to a Boltzmann function. (J to L) Time course of average CA-induced currents through indicated mutants measured at +80 and −80 mV. The average current for the WT is shown as a gray line with bars indicating means ± SEM (n = 45 cells for WT and n = 8 cells for each of the mutants). Bottom: Mean rectification ratio for the cells shown above (colored lines with gray bars indicating means + SEM) plotted against time.

These distinctive effects of mutations argued against a mechanism of simple electrostatic interactions among the charged residues in the internal sensor cavity. Rather, they indicated that some other factor is required as an intermediary for a series of allosteric steps that are necessary for a proper regulation of TRPA1. One possibility emerging from our molecular modeling was that conformational changes underlying the transition of the channel between the closed and open states are influenced by changes in the water accessibility of the inner crevice of the sensor domain. The other possibility was indicated by our docking studies implying that Asn722, and Lys787, together with His719, Lys796, Arg852, and Lys989 may constitute a binding site for phosphoinositides (Fig. 2C). These models are not mutually exclusive, because changing the occupancy of water within the crevice and/or a disruption of the binding pocket may, in a state-dependent manner, affect the strength of allosteric coupling between the putative voltage- and Ca2+-sensing domains and the gate.

Inner cavity of the sensor is a Ca2+-sensing domain

In our structural model, the key lysine residue, K787, forms salt bridge interactions with one glutamate residue, Glu808, whereas the neighboring Glu788 is oriented away from the cavity (Fig. 5F) and located about 2 to 3 Å from Tyr799, which, in turn, is about 3 Å from Asn805. Because the size and not the charge of the residue at position Glu788 was important for effective gating, we further substituted Glu788 with a bulky hydrophobic isoleucine to explore whether Glu788 might contribute to the processes by which the various signals from the cavity could be propagated throughout the sensor domain (Fig. 5, G and H). The voltage-induced currents through E788I mutant channels closely resembled those mediated by E788K mutant channels (Fig. 3D). The tail currents upon repolarization from +200 to −70 mV normalized to the maximum amplitude at +200 mV, measured in the standard bath solution, which contained 1 mM Ca2+, were much faster in E788I mutant channels, but upon the removal of Ca2+, the traces were superimposable onto those obtained with wild-type TRPA1 (Fig. 5G). Upon the removal of Ca2+ from the extracellular medium, the G-V relationship of E788I channels was significantly shifted leftward by −35 mV (z from 0.7 to 0.6 eo; Fig. 5H). This result supported the involvement of Glu788 in the Ca2+-dependent regulation of TRPA1 and argued against the relevance of the polarity of this residue for channel deactivation under Ca2+-free conditions. Also consistent with our hypothesis, the voltage- and agonist-induced currents through N805A and Y799A channels almost completely overlapped, resembled those of the E788I channels, and were characterized by large and saturating cinnamaldehyde-induced currents that were not further potentiated by Ca2+, indicating that these residues use the same transduction pathway and participate in the putative mechanism through which TRPA1 is modulated by permeating Ca2+ (Fig. 5, J to L; note the striking resemblance of the mutant phenotypes).

On the basis of this information, we further explored a working model of channel activation that can account for the observed voltage- and calcium-dependent TRPA1 current characteristics (Fig. 6). In this model, we hypothesized that the lower cavity needs to be occupied by a phospholipid (as in the wild-type channels) to enable proper gating upon voltage stimulation and upon agonist stimulation (Fig. 6A). When the mutations (H719A, K787A, and N722A/I) prevent the stabilization of the sensor domain by phospholipids, the channel tends to be closed (Fig. 6B), whereas mutations (E788A and E808A) with putative propitious effects on PIP2 binding cause gain-of-function phenotypes (Fig. 6C), perhaps by keeping the channel in a state that is primed for activation. Moreover, any alternation of the polarity balance deep in the cavity causes a Ca2+-dependent block of currents at negative membrane potentials (Fig. 6D).

Fig. 6 Proposed mechanism for regulation of human TRPA1 by the inner cavity of sensor domain.

The diagram of the transmembrane part of the TRPA1 channel only shows two of the four subunits for clarity. The sensor domains (S1–S4; green) are connected through the S4-S5 linkers (lilac) to the pore domains (S5 and S6; gray). The TRP-like domain (brown) interacts with the S4-S5 linker through hydrophobic interactions. The inner cavities of the sensor domains are shown as a light blue cone for the wild-type channel. Sodium and calcium ions (blue and red circles) are indicated. (A) In the absence of external Ca2+, the cavity can be occupied by a phospholipid (PIP2; lipids with yellow and red phosphate head groups) to enable proper gating at negative membrane potentials. (B and C) When the mutations (K787A, H719A, N722A/I, and R852A) prevent the stabilization of the sensor by phospholipids (indicated with a white cone), the channel tends to be closed (B), whereas mutations (E788A and E808A) with putative beneficial effects on phosphoinositide binding (indicated with a dark blue cone) cause an increase in currents in the absence of external Ca2+ (C). (D) Any alteration to the polarity balance deep in the cavity causes a Ca2+-dependent block of currents elicited by electrophilic agonists.

Structural comparisons of the lipid binding site densities among the TRPV1 and TRPV2 at different conformations suggested that interaction of the crevice formed by the S1–S4 helical bundle above the TRP domain with lipids is dynamic and possibly depends on the channel activation state (20). Our results supported a similar role of lipids for TRPA1. In our model, the residues Asn722, Lys787, and Glu808 directly contact Tyr726 (fig. S3, A and B) and help to keep its side chain in an orientation analogous to Tyr400 in TRPV2. This tyrosine is in a direct contact with the membrane lipid in TRPV2 (20). It is noteworthy that in TRPA1, the neighboring residue of the cognate Tyr726 is the reactive cysteine Cys727 and its reactivity depends on the extent of TRPA1 activation (24). This may indicate that this part of the channel undergoes structural changes during electrophilic activation. To test this hypothesis, we measured cinnamaldehyde- and AITC-induced whole-cell currents from the C727A and C727S mutants (fig. S3, C and D). In the absence of external Ca2+, these mutants did not exhibit changes in their responsiveness to electrophilic agonists. However, the addition of Ca2+ to the bath solution induced a significantly stronger inactivation of cinnamaldehyde-induced responses than in wild-type channels but not of the currents induced by the full agonist AITC. This result supported the hypothesis that the sensor domain was involved in Ca2+-induced inactivation and that this process was activity-dependent.

Membrane PIP2 may stabilize TRPA1 in an open conformation

Reports are inconsistent regarding the effects of PIP2 on TRPA1 [reviewed in (25)]. Although there are indications that PIP2 has a positive modulatory effect on TRPA1 (26, 27), some reports suggest that PIP2 either does not affect (21, 28) or down-regulates TRPA1 (29, 30). In our study, mutations of the residues putatively involved in PIP2 binding (His719, Asn722, Lys787, Lys796, Arg852, and Lys989) resulted in a partial loss of functional response to voltage and the low-affinity electrophilic agonist cinnamaldehyde (Figs. 3 and 4). Thus, if phosphoinositides (particularly PIP2) bind the TRPA1 channel through the inner cavity of the sensor domain, then we would expect that the depletion of membrane PIP2 could down-regulate the wild-type channels (Fig. 6). To clarify this issue, we used three different approaches to manipulate the membrane PIP2 levels and measured responses from wild-type TRPA1 (Fig. 7).

Fig. 7 Sequestering or reducing membrane PIP2 inhibits voltage- and cinnamaldehyde-induced TRPA1 currents.

(A to D) Top: Average conductances obtained from HEK293T cells transfected with the indicated protein combinations measured in control bath solution containing 160 mM NaCl, 2.5 mM KCl, 1 mM CaCl2, 2 mM MgCl2, 10 mM Hepes, and 10 mM glucose (adjusted to pH 7.3 and 320 mosmol). The solid lines of the WT (gray) or E808A mutant (brown) are the best fits to a Boltzmann function, as described in Materials and Methods and shown in Fig. 3 (B and D). Data are means ± SEM (n = 7 to 13 cells from at least two independent transfections). Bottom: The time course of average currents induced by cinnamaldehyde (100 μM) and by external Ca2+ (2 mM) measured at +80 and −80 mV, as described in Fig. 4G. The average current through TRPA1 expressed alone is overlaid as a gray line, with bars indicating SEM (n = 45 cells) for comparison. A schematic model of the mechanism is indicated for each “pipmodulin” above. PIP2 molecules are indicated as lipids with yellow and red phosphate head groups. Data are means ± SEM (n = 7 to 11 cells from at least two independent transfections). Bottom: Average rectification of currents shown above expressed as absolute values of the amplitudes of inward currents at −80 mV divided by outward currents at +80 mV and plotted as a function of time. The average rectification of TRPA1 expressed alone is overlaid as a gray line with gray bars indicating SEM (n = 45 cells) for comparison. MARCKS, myristoylated alanine-rich C-kinase substrate; GAP43, growth-associated protein 43. (E) The same experiments described in (A) to (D) performed with TRPA1 coexpressed with voltage-sensitive lipid 5-phosphatase from Danio rerio (Dr-VSP). The voltage step protocol was preceded by a 2-s depolarization to +80 mV. Colored symbols and lines with gray bars indicate average ± SEM (n = 7 to 9 cells).

First, we coexpressed TRPA1 with myristoylated alanine-rich C-kinase substrate (MARCKS). This protein has basic domains capable of laterally sequestering membrane PIP2, depending on a local increase in intracellular Ca2+ concentration (31, 32). The expression of MARCKS in HEK293T cells should cause membrane PIP2 to be sequestered by MARCKS and, therefore, less available for the supposed interaction with coexpressed TRPA1. In line with our hypothesis, we observed that the sequestration of PIP2 by MARCKS had detrimental effects on wild-type TRPA1 functioning (Fig. 7A). Upon voltage stimulation, the responses were completely inactivated at negative potentials, resembling the currents mediated by H719A, K787A, N722I, and K796A. The G-V curves were significantly shifted rightward, and the cinnamaldehyde-evoked currents were significantly smaller than the wild-type responses. Moreover, the onset of agonist-induced responses was delayed at negative membrane potentials as in the mutants N722A, K787A, and R852A, which we predicted to be disruptive of the TRPA1-phospholipid interaction. The addition of external Ca2+ to cells preactivated for 40 s with cinnamaldehyde caused an immediate influx of Ca2+, which should result in a release of PIP2 sequestered by MARCKS. Next, we assumed that the neutralization of Glu808 might positively influence a putative channel-phosphoinositide interaction. We hypothesized that if the interacting lipid is PIP2, then its sequestration by MARCKS should attenuate the current responses of E808A. The actual changes observed in HEK293T cells coexpressing MARCKS with E808A channels matched this assumption reasonably well (Fig. 7B). The voltage-gated currents were suppressed to the wild-type level at positive holding potentials, and the channels were completely blocked at negative potentials. The cinnamaldehyde-induced currents also decreased to the wild-type level, and a subsequent exposure to Ca2+ caused an immediate block, as was previously observed for E808K and E788A/E808A.

In the second approach, we coexpressed TRPA1 with growth-associated protein 43 (GAP43; also known as neuromodulin). Whereas MARCKS sequesters PIP2 under low-Ca2+ conditions, GAP43 sequesters PIP2 when the cytoplasmic Ca2+ concentration is increased (31, 32). Furthermore, only one point mutation in GAP43, R43A, produces a phenotype that sequesters PIP2 regardless of the Ca2+ concentration (33), thus providing an excellent control for our experiments. The same series of experiments as with MARCKS was performed with either the wild-type GAP43 (WT-GAP43) or R43A-GAP43 mutant coexpressed with wild-type TRPA1 (Fig. 7, C and D). Coexpression of WT-GAP43 did not change the TRPA1 profiles of current responses induced by cinnamaldehyde in the absence of external calcium, as expected for these control measurements. Switching to the extracellular solution containing 2 mM calcium sensitized the responses to a greater extent than that observed for the wild-type TRPA1 alone. Coexpression of the mutant R43A-GAP43, which was expected to sequester PIP2 independently of calcium concentration, resulted in reduced current amplitudes, practically identical to those obtained with the expression of MARCKS (Fig. 7A), which supported the idea that the observed changes in activation kinetics of TRPA1 may be caused by a lack of available PIP2 in the plasma membrane.

In the third approach, we used a voltage-sensitive lipid 5-phosphatase from Danio rerio (Dr-VSP). The activity of this enzyme can be induced by depolarization greater than +50 mV and results in the hydrolysis of PIP2 [PI(4,5)P2 to PI(4)P] (34, 35). Whereas the above coexpression approaches allow for the equilibration of free PIP2 with closed channels, the acute stimulation of Dr-VSP only dynamically affects the channels upon depolarization to +80 mV. At these potentials, the channels opened with a probability of about 25% in the absence of any agonist (Fig. 3B). We coexpressed Dr-VSP with TRPA1 and stimulated the cells with a 2-s depolarizing prepulse to +80 mV before each application of the standard voltage step protocol (Fig. 7E). In cotransfected cells, the rightward shift in the G-V curves was similar to the effects seen with MARCKS and R43A-GAP43, in support of a PIP2-promoting role in voltage-dependent gating. In the next series of experiments, we measured responses to cinnamaldehyde using a standard protocol in which the membrane potential was linearly ramped up each second from −80 to +80 mV (1 V/s). The cinnamaldehyde-induced currents measured from the cells coexpressing TRPA1 with Dr-VSP were selectively suppressed at negative membrane potentials (Fig. 7E). The subsequent addition of extracellular Ca2+ potentiated the currents at both positive and negative membrane potentials, more than we observed in cells expressing TRPA1 alone. Collectively, the results indicated that a reduction in PIP2 leads to a rightward shift in G-V characteristics and, unambiguously at least at negative membrane potentials, a reduction in cinnamaldehyde-induced currents.

Apparently, the effects of PIP2 modulation by the “pipmodulins” MARCKS, GAP43, R43A-GAP43, and Dr-VSP were obscured upon the addition of external Ca2+ (Fig. 7, A to E). This indicated that PIP2 may modulate TRPA1 in a state-dependent manner, with PIP2 preferentially occupying closed states until the potentiating effect of Ca2+ prevails. To further clarify this observation, we measured the cinnamaldehyde-induced responses from cells coexpressing TRPA1 with MARCKS, GAP43, or R43A-GAP43 while using the LB-ICS containing 100 μM free Ca2+. The channels appeared to more readily respond to cinnamaldehyde and exhibited a greater degree of inactivation upon the addition of Ca2+ to the extracellular medium (fig. S4, A to C). Notably, the presence of GAP43, but not of R43A-GAP43, significantly increased the currents, suggesting a specific and Ca2+-dependent action of GAP43 on TRPA1 (fig. S4, B and C). This observation indicated that PIP2 may compete with Ca2+ to confer the potentiation of TRPA1.

Glu808 regulates activation of TRPA1 under physiological temperatures

Because TRPA1 is considered as a thermosensitive channel, we further explored whether the identified residue Glu808 plays a general role in channel activation under physiological temperatures. We measured currents at 25° and 35°C using 200-ms voltage ramps from −100 to +100 mV (fig. S5, A to H). Currents through wild-type and mutant channels measured in control extracellular solution at 25°C exhibited outward rectification that was significantly more pronounced by increasing the temperature to 35°C (fig. S5, A to E). The outward currents at +80 mV were potentiated 1.9-fold in wild-type channels, which is in agreement with a previously reported finding (36). E808A mutation produced currents that were potentiated only 1.5-fold at 35°C. In the presence of an agonist (50 μM carvacrol), increasing the temperature from 25° to 35°C reduced the rectification ratio about 2-fold in wild-type channels but only ~1.5-fold in E808A (fig. S5, F to H). This mutation rendered the channel insensitive to the agonist at negative membrane potentials and significantly decreased its temperature dependence. This result demonstrated that Glu808 may participate in temperature regulation of TRPA1. The significant effects of the alanine mutation causing the channel to remain closed at physiological membrane potentials and to be less affected by warm temperatures indicated that this residue is, in essence, irreplaceable for the proper functioning of the TRPA1 channel under close to physiological conditions.

DISCUSSION

TRPA1 is the only known ion channel that is capable of responding to both cold and heat (36). Such a bidirectional temperature dependency has been recently explained by a theoretical possibility that the conformational changes that occur upon opening the channel may cause a transfer of specific residues between hydrophobic and aqueous environments, which leads to changes in the heat capacity of the channel protein complex (3739). When premodified with redox-active compounds or noncovalent ligands, TRPA1-mediated currents were sensitized by warm and cold temperatures (36), indicating that conformational changes induced by chemical agonists may be associated with changes in the solvent accessibility of residues within a putative allosteric nexus converging on the region that encompasses the TRP-like domain, pre-S1 helix, and S4-S5 linker (5). The inner cavity of the sensor is an integral part of this key region, and thus, it may represent an important locus that undergoes changes in solvation when the channel opens. Our results have identified possible candidate residues in the lower sensor domain that may be the main determinants of water occupancy. Among these, Glu788, Glu808, and Lys787 seem to play important roles. In the presence of external Ca2+, mutations at His719, Asn722, Lys796, Glu788, and Glu808 biased the channel toward the closed state at negative membrane potentials, indicating that PIP2 binding or changes in hydration of these residues may provide the energy required to open the channel. On the other hand, the mutations E788I, N805A, Y799A, and K989A biased the channel toward the open state in the absence of Ca2+, suggesting that this may be an inherently less stable conformation and that calcium provides the energy required to adopt the closed conformation. Although this hypothesis awaits confirmation by future experiments, our results demonstrate that substitutions of polar residues predicted to face the crevice lead to large changes in TRPA1 sensitivity to voltage and chemical stimuli. In addition, our molecular model indicated that the inner cavity is also capable of binding phosphoinositides, such as PIP2, which might further increase the number of allosteric states that depend on the degree of channel activation and phospholipid environment.

Our current understanding of how the TRPA1 channel gates is based on the electron cryo-microscopic structure obtained with an agonist and two antagonists (5) and on the recent structural analyses of related TRPV1, TRPV2, TRPV6, and TRPP1 channels (710, 14, 15). The remarkable sequence variability among the TRP proteins observed in the lower part of the sensor and the S2-S3 intracellular linker (16) indicates that this domain might serve a specific function in TRPA1 channel gating. The lower sensor domain in TRPA1 may contain a putative site for interactions with annular or regulatory lipids as seen in its relative TRPV1 (13). Using several approaches to manipulate PIP2 levels in cells, we found that the depletion of membrane PIP2 down-regulates the wild-type TRPA1 channels in the absence of external Ca2+. An even more selective effect of PIP2 depletion observed with Dr-VSP indicated that TRPA1 needs PIP2 to be properly activated at negative membrane potentials. When the channel is weakly activated by depolarization or cinnamaldehyde, the intracellular cavity of the sensor domain can bind PIP2 and then release it in response to a local increase in Ca2+. Our observation that the increased cinnamaldehyde-induced activity of the gain-of-function mutant E808A can be suppressed to wild-type levels by PIP2 sequestration is a strong indication of the role of the internal sensor cavity in TRPA1 regulation.

During the preparation of this work, a study reported that calmodulin binds to TRPA1 in a Ca2+-dependent manner and that this binding is essential for the basal sensitivity and Ca2+ potentiation and inactivation of the channel (40). MARCKS and GAP43, but not R43A-GAP43, interact with the Ca2+/calmodulin complex and the free calmodulin, respectively, and this interaction tightly depends on intracellular calcium concentration (33). We have found here that TRPA1 coexpressed with the pipmodulins MARCKS, GAP43, or R43A-GAP43 exhibited a higher maximum rectification ratio (~0.9) than TRPA1 alone and that this effect was seen when the intracellular calcium concentration was raised. This means that the channels come close to saturation, which could indicate that PIP2 modulates TRPA1 in a Ca2+-dependent manner or vice versa. This could explain the divergent results on the effect of PIP2 in reports from groups using different activation states of the channel. Whether PIP2 and calmodulin agonistically regulate TRPA1 through direct or indirect competition should be further analyzed, and such analysis could reveal important details about the activation mechanisms of TRPA1.

MATERIALS AND METHODS

Homology modeling, ligand docking, MD simulations

To obtain a model of human TRPA1 with the S1–S4 sensor domain, we used the structure with PDB ID: 3J9P determined by cryo-electron microscopy (5). The intracellular loop connecting helices S2 and S3 was modeled with the use of sequence homology with the polycystin-2 TRP channel TRPP1 [structures with PDB IDs: 5K47 (12), 5T4D (14), 5MKE and 5MKF (15)]. The homology model of the S2-S3 linker of TRPA1 was created using the Swiss-Model web server (https://swissmodel.expasy.org/). UCSF Chimera (www.cgl.ucsf.edu/chimera/) and AutoDock Vina (http://vina.scripps.edu/) were used for the docking of phospholipids into the S1–S4 sensor domain of TRPA1. The electrostatic potential surrounding TRPA1 and TRPV1 channels was determined by means of Visual Molecular Dynamics (VMD) (41). The TRPA1 tetrameric structure was inserted into the patch of the 1-palmitoyl-2-oleoylphosphatidylcholine (POPC) bilayer and solvated in transferable intermolecular potential 3-point (TIP3P) (42) water molecules to ensure at least 10 Å of solvent on both sides of the membrane and neutralized in 0.5 M NaCl. This gives a periodic box with a size of ~133, ~133, and ~164 Å for a simulated system consisting of ~239,000 atoms. All-atom structure and topology files were generated using VMD (41). Forces were computed using a CHARMM27 force field for proteins, lipids, and ions (4345). All MD simulations were produced with the aid of the software package NAMD2.9 (46) running on a local workstation equipped with an NVIDIA graphics processing unit. The particle mesh Ewald method with a grid size of 128 × 128 × 192 was used for long-range electrostatic forces (47). The nonbonded cutoff was set to 12 Å. The SETTLE algorithm (tolerance, 1 × 10−8) was applied to constrain bonds in water molecules (48). Langevin dynamics was used for temperature control with the target temperature set to 310 K, and the Langevin piston method was applied to reach an efficient pressure control with a target pressure of 1 atm (46). The integration time step was set to 2 fs. Simulated systems were energy-minimized and heated to 310 K, and production MD runs reached lengths of 200 ps. Data were recorded every 1 ps and analyzed using the CPPTRAJ module from the AmberTools suite (49). MD trajectories were visualized with the aid of the VMD 1.9 software package (41). Figures were produced with the software packages UCSF Chimera (50), ICM (Molsoft LLC), and CorelDraw X7 (Corel Corporation).

Cell culture, constructs, and transfection

HEK293T cells were cultured in Opti-MEM I medium (Invitrogen) supplemented with 5% fetal bovine serum as described previously (51). The magnet-assisted transfection (IBA GmbH) technique was used to transiently cotransfect the cells in a 15.6-mm well on a 24-well plate with 200 ng of plasmid encoding green fluorescent protein (TaKaRa), 300 ng of plasmid encoding wild-type or mutant human TRPA1 (pCMV6-XL4 vector, OriGene), and, for particular experiments, 200 ng of plasmid of wild-type or mutant GAP43 (pCMV6-XL5 vector, OriGene) or 100 ng of plasmid encoding MARCKS (pCMV6-XL5 vector, OriGene). The cells were used 24 to 48 hours after transfection. At least two independent transfections were used for each experimental group. The wild-type channel was regularly tested in the same batch as the mutants. The mutants were generated by polymerase chain reaction using the QuikChange II XL Site-Directed Mutagenesis Kit (Agilent Technologies) and confirmed by DNA sequencing (GATC Biotech).

Electrophysiology

Whole-cell membrane currents were recorded by using an Axopatch 200B amplifier and pCLAMP 10 software (Molecular Devices). Patch electrodes were pulled from borosilicate glass and heat-polished to a final resistance between 3 and 5 megohms. Series resistance was compensated by at least 70% in all recordings. The experiments were performed at room temperature (23° to 25°C). Only one recording was performed on any one coverslip of the cells to ensure that recordings were made from cells not previously exposed to chemical stimuli. A system for rapid superfusion and heating of the cultured cells was used for drug application (52). The extracellular bath solutions contained 150 mM NaCl and 10 mM Hepes, with an added 2 mM HEDTA [N-(2-hydroxyethyl)ethylenediamine-N,N′,N′-triacetic acid] for the Ca2+-free solution and 2 mM CaCl2 for the Ca2+-containing solutions (adjusted to pH 7.3 with NaOH and to 300 mosmol). The current-to-voltage (I-V) relationships were measured in control bath solution containing 160 mM NaCl, 2.5 mM KCl, 1 mM CaCl2, 2 mM MgCl2, 10 mM Hepes, and 10 mM glucose (adjusted to pH 7.3 and 320 mosmol). The I-V relationships were recorded using 100-ms voltage steps from −80 to +200 mV (+20-mV increments) and a holding potential of −70 mV recorded in control extracellular solution ~1 min after whole-cell formation. The high-buffer internal pipette solution containing 145 mM CsCl, 5 mM EGTA, 3 mM CaCl2, 10 mM Hepes, and 2 mM MgATP (adjusted to pH 7.3 with CsOH and to 290 mosmol) was used unless the usage of low-buffer internal solution (LB-ICS) is noted, which contained 145 mM CsCl, 10 mM EGTA, 10.24 mM CaCl2 (corresponding to 100 μM free Ca2+), 10 mM Hepes, and 2 mM MgATP (adjusted to pH 7.3 with CsOH and to 290 mosmol). For experiments shown in fig. S5, extracellular control solution contained 140 mM NaCl, 5 mM KCl, 2 mM MgCl2, 5 mM EGTA, 10 mM Hepes, and 10 mM glucose (pH 7.4 was adjusted by tetramethylammonium hydroxide). Intracellular solution contained 140 mM KCl, 5 mM EGTA, 2 mM MgCl2, and 10 mM Hepes (adjusted to pH 7.4 with KOH). Cinnamaldehyde and AITC solution was prepared before use from a 0.1 M stock solution in Me2SO. All of the chemicals were purchased from Sigma-Aldrich. The agonist sensitivity was tested with a standard protocol where the membrane potential was ramped every second from −80 to +80 mV (1 V/s) or from −80 to +160 mV (1 V/s) for the inactivating mutations H719A and K787A. In experiments with Dr-VSP, the voltage step protocol was preceded by a 2-s depolarization to +80 mV. The current responses for further processing were always measured at −80 and +80 mV.

Statistical analysis

The electrophysiological data were analyzed using pCLAMP 10 (Molecular Devices), and the curve fitting and statistical analyses were done in SigmaPlot 10 (Systat Software Inc.). G-V relationships were obtained from steady-state whole-cell currents measured at the end of voltage steps from −80 to +200 mV in increments of +20 mV. Voltage-dependent gating parameters were estimated by fitting the conductance G = I/(VVrev) as a function of the test potential V to the Boltzmann equation: G = [(GmaxGmin)/(1 + exp (−zF(VV50)/RT))] + Gmin, where z is the apparent number of gating charges; V50 is the half-activation voltage; Gmin and Gmax are the minimum and maximum whole-cell conductance, respectively; Vrev is the reversal potential; and F, R, and T have their usual thermodynamic meanings. Statistical significance was determined by Student’s t test or the analysis of variance (ANOVA), as appropriate; differences were considered significant at P < 0.05 unless stated otherwise. Data are means ± SEM.

SUPPLEMENTARY MATERIALS

www.sciencesignaling.org/cgi/content/full/11/514/eaan8621/DC1

Fig. S1. Mutations in the inner cavity of the sensor module affect chemical-dependent gating of TRPA1.

Fig. S2. Average rectification of whole-cell currents through the mutant channels.

Fig. S3. Structural comparison with TRPV2 and a central role for Tyr726.

Fig. S4. LB-ICS containing Ca2+ abolishes the inhibitory effects of MARCKS and mutant GAP43.

Fig. S5. Polar residues in the sensor cavity regulate the activity of TRPA1 under physiological temperatures.

REFERENCES AND NOTES

Funding: This work was supported by the Czech Science Foundation (15-15839S to V.V., L.Z., L.V., V.S., and A.K.) and the Grant Agency of Charles University (365215 to V.S.). Author contributions: V.V. and L.Z. conceived the project, designed the experiments, and wrote the manuscript. L.Z., I.B., A.K., V.S., and V.Z. performed the experiments and analyzed the data. L.V. helped read the manuscript and was involved in data discussions. Competing interests: The authors declare that they have no competing interests. Data and materials availability: The structural model for human TRPA1 is available in the Model Archive (www.modelarchive.org) under the accession code ma-auqu1.
View Abstract

Navigate This Article