Mechanisms of sensory adaptation and inhibition of the cold and menthol receptor TRPM8

Abstract

Our sensory adaptation to cold and chemically induced coolness is mediated by the intrinsic property of TRPM8 channels to desensitize. TRPM8 is also implicated in cold-evoked pain disorders and migraine, highlighting its inhibitors as an avenue for pain relief. Despite the importance, the mechanisms of TRPM8 desensitization and inhibition remained unclear. We found, using cryo-electron microscopy, electrophysiology, and molecular dynamics simulations, that TRPM8 inhibitors bind selectively to the desensitized state of the channel. These inhibitors were used to reveal the overlapping mechanisms of desensitization and inhibition and that cold and cooling agonists share a common desensitization pathway. Furthermore, we identified the structural determinants crucial for the conformational change in TRPM8 desensitization. Our study illustrates how receptor-level conformational changes alter cold sensation, providing insights into therapeutic development.

Fig. 1.. Cryo-EM structure determination of TRPM8_MM in complex with antagonists.

(A) Cartoon diagram of the PIP₂ and cooling agonist-dependent gating pathway of TRPM8 channels. Published TRPM8_MM structures in the C₀, C₁, and O states (PDB 8E4P, 8E4N, and 8E4L, respectively) are used for illustration. (B) Chemical structures of TC-I 2014, AMG2850, and AMTB. (C) 3D reconstructions of the conformation A (left; brown) and the C₀ state (right; silver-gray). Neighboring protomers are colored in gray. (D) TMD comparison between the conformation A (left; brown) and the C₀ state (right; silver-gray). S2-S3 linkers and S3 were omitted for clarity. (E and F) EM density at the VSLD cavity (E) and the S4-S5 linker (F) from the conformation A reconstruction in (C). Densities corresponding to antagonists are colored in lime for TC-I, teal for AMG, and blue for AMTB. Red dashed circles indicate the lack of antagonist densities. Thresholding 0.3 in (E) and 0.6 in (F). (G and I) Binding site and EM densities for TC-I [(G) left, lime sticks], AMG [(G) right, teal sticks], and AMTB [(I) blue sticks]. Densities in magenta mesh are contoured at thresholding 0.3 for TC-I, 0.34 for AMG, and 0.2 for AMTB. (H and J) Summary of current inhibition by 10 μM TC-I or 10 μM AMG (H) or by 25 μM AMTB (J) measured by TEVC recording on the WT and mutant TRPM8_MM channels activated by 10 to 30 μM C3 at −60 mV. The antagonist inhibition level is quantified by the percentage of current inhibited by antagonists over full inhibition by 50 μM RR (see Materials and Methods). Values for individual oocytes are shown as open circles with means ± SEM (n = 3 to 8 oocytes). ns > 0.05, **P < 0.01, ***P < 0.001, using one-way ANOVA followed by Dunnett’s post hoc test.

Fig. 2.. Molecular details of state-dependent inhibition.

(A) Mean normalized concentration-response relations for TC-I (left), AMG (middle), and AMTB (right) against TRPM8_MM activation by 1 mM menthol in the presence of increasing concentrations of extracellular Ca²⁺ in the TEVC recording buffer. Data are shown as means ± SEM. n = 4, 5, and 4 oocytes for measurement with 0, 0.5, and 2 mM Ca²⁺ in the left panel, respectively; n = 4, 5, and 5 oocytes for measurement with 0, 1, and 3 mM Ca²⁺ in the middle panel, respectively; n = 5, 5, and 6 oocytes for measurement with 0, 0.5, and 2 mM Ca²⁺ in the right panel, respectively. The continuous curves were fit to the Hill equation with IC₅₀ values indicated in the figure. (B) Thermodynamic model of TRPM8 desensitization and inhibitor binding for the equilibria (K_n) among C, O, D, OI, and DI states and the associated thermodynamic coupling (c). Red arrows indicate transition among the thermodynamically favored states based on fit values. (C) AMG dose-response data fit with the thermodynamic model for 0, 1, and 3 mM extracellular Ca²⁺. Data are from (A), taking into account the initial level of desensitization. Data points for individual replicate (n ≥ 4) shown. (D) Model fit quality as SSQ plotted as a function of fixed coupling parameter (c) value. (E) Calcium dependence of desensitization (K₃) for AMG and TC-I. Error bars represent 95% confidence intervals. (F) Shown in surface (upper) and cylinder (lower) representations, comparison of the antagonist binding site above the S4-S5 linker in TRPM8 channels adopting the C₀ (PDB 8E4P), C₁ (PDB 8E4N), O (PDB 8E4L), and D (current study) states. Red dashed lines highlight changes in the size of the binding pocket. TC-I and AMG shown as spheres, and residues involved in antagonist binding shown as sticks.

Fig. 3.. TRPM8 activation by distinct stimuli converges on desensitization.

(A) Representative whole-cell recordings on HEK293T cells expressing the WT TRPM8_MM channels at −60 mV. Current trace elicited by 100 μM C3 was inhibited by increasing concentrations of AMG in the presence of 0 μM (left) and 100 μM (right) extracellular Ca²⁺. (B) Mean normalized concentration-response relations for AMG against the WT TRPM8_MM activation by 100 μM C3 in the presence of 0 μM (black trace; n = 5) or 100 μM (red trace; n = 4) extracellular Ca²⁺. Data are means ± SEM. The continuous curves were fit to the Hill equation with IC₅₀ values indicated in the figure. (C and D) Representative whole-cell recordings on HEK293T cells expressing the WT TRPM8_MM channels at −60 mV. Current trace elicited by cold (C) or 200 μM menthol (D) was inhibited by increasing concentrations of AMG in the presence of 0 μM (left) or 40 μM (right) extracellular Ca²⁺. Currents were first stabilized after three to six repeated cold or menthol activations, then increasing concentrations of AMG were coapplied with subsequent cold or menthol pulses, as indicated. (E) Summary of IC₅₀ values for WT TRPM8_MM activation by cold (blue circles) and menthol (green circles) in the presence of 0 and 40 μM extracellular Ca²⁺ (n = 3 to 4). Data are means ± SEM. The continuous curves fit to the Hill equation with IC₅₀ values are in fig. S9. (F) Simplified schematic diagram showing the TRPM8 gating pathway from the C to the O state followed by D and DI with antagonist bound. Dashed lines in (A), (C), and (D) indicate the zero-current level.

Fig. 4.. Mechanism of TRPM8 desensitization.

(A) Aligned at the VSLD, comparison of the O (blue) the D (brown) states at the pore, viewed from the extracellular side. Red circle indicates the S6 gate residue Val⁹⁷⁶. (B and C) Ion permeation pathway (B) and pore radii (C) in the O state and the D state. (D to F) Comparison of the O and the D states at the interfacial cavity for PIP₂ binding [(D) and (E)] and at S4b and the S4-S5 linker (F). Val⁹⁷⁶ shown as yellow spheres. Asterisks denote the π helix position. Red dashed lines and arrows in (E) indicate the structural rearrangements from the O to the D state. ⊗ symbol denotes the ion conduction pathway. (G) Comparison of the O and the D states at the interaction network among S4, the S4-S5 linker, and the neighboring pore domain (S5′ and S6′) with Phe⁸⁶⁹′ highlighted in red and Ile⁸⁵⁷, Leu⁸⁶⁰, and Leu⁹⁷⁰′ in teal. (H) Mean normalized concentration-response relations for AMG against the WT (n = 3) and F869A (n = 3) in the presence of 0 or 100 μM extracellular Ca²⁺. Representative current traces shown in fig. S14E. (I) Mean normalized concentration-response relations for AMTB against the WT (black trace; n = 3) and F869A (red trace; n = 3). Representative current traces shown in fig. S14F. (J) Representative current traces of the WT and mutant TRPM8_MM channels activated by 200 μM menthol (upper) and cold (lower) at −60 mV in HEK293T cells. Dashed lines indicate the zero-current level. (K and L) Summary of currents remaining after desensitization for the WT and mutant TRPM8_MM channels activated by menthol (K) and cold (L) (n = 3 to 5). *P < 0.05, **P < 0.01, ***P < 0.001, using one-way ANOVA followed by Dunnett’s post hoc test. Data are means ± SEM in (H), (I), (K), and (L).

Fig. 5.. Mechanism of TRPM8 inhibition.

(A to E) Comparison of the pore domain (S5′ and S6′) between the O (A) and the D states [(B) and (D)]. Trp⁸⁷⁷ on S5 is rotated and exposed to the membrane-facing side, and the π helix position on S6 is shifted in the D state [(C) and (E)]. TC-I or AMG binding stabilizes the alternate π helix and the rotated Trp⁸⁷⁷ positions in the D state [(B) and (D)]. Dashed lines indicate interactions that stabilize the specific helical configurations. (F) AMTB interaction stabilizes the VSLD in the D state. Residue side chains and antagonists shown as sticks.

Fig. 6.. Roles of PIP₂ and Ca²⁺ in TRPM8 desensitization.

(A) Table of channel conformation and ligands present for five TRPM8_MM structures used for analyses in (B) to (D). (B to D) TMD superimposition among structures as indicated. Channel structures shown as ribbons and colored as in (A). Dashed boxes highlighting conformations at S4b, S4-S5, S6, and the TRP domain (residues 839 to 877 and 966 to 999). Cα RMSD values at this region against the reference structure (Ref) are indicated in parentheses. PDB IDs or structures from the current study and the corresponding channel conformation are specified. (E) Comparison of interactions between the S4-S5 linker and the TRP domain in the PIP₂-bound C1 state (left, orange; PDB 8E4N), the O state (middle, blue; PDB 8E4L), and the D state (right, brown; TRPM8_MM_TC-I). PIP₂ shown as red sticks and key residues shown as sticks and labeled in red. Dashed lines indicate residue interactions. Secondary structural change at S4b and movements at the S4-S5 linker and S6-TRP are indicated by arrows. (F and G) Summary of normalized currents of the TRPM8_MM channel upon repeated activation by menthol (F) and cold (G) (n = 4 to 5) at −60 mV. Open circles indicate the individual data points for each experiment. ***P < 0.001, using two-way ANOVA followed by Sidak post hoc test. Representative time courses shown in fig. S16 [(B) to (E)]. (H and I) Summary of currents remained after desensitization of the TRPM8_MM channel subject to prolonged activation by menthol (H) and cold (I) (n = 3 to 4). Open circles indicate the individual data points for each experiment. ns > 0.05, **P < 0.01, ***P < 0.001, using two-way ANOVA followed by Sidak post hoc test. Representative time courses shown in fig. S16 [(F) and (G)]. Data are means ± SEM in (F) to (I).

Fig. 7.. Summary of ligand binding and ligand-dependent gating of TRPM8 channels.

(A) The binding sites for PIP₂ (1), type I agonists (2), type II agonists (3), Ca²⁺ ions (4), competitive antagonists (5), and noncompetitive agonists (6) are located surrounding the S4b in TRPM8. Representative ligands, PIP₂ (red), C3 (yellow), AITC (teal), TC-I (lime), and Ca²⁺ (green), are shown in surface or sphere and highlighted by dashed lines. (B) Conformational changes of S4b, S5, and the S4-S5 linker in the closed C₀ (silver-gray), closed C₁ (yellow), O (blue), and D (brown) states. Secondary structures and S4-S5 linker motions indicated by lines and arrows. PIP₂, C3, AITC, and TC-I shown as sticks and Ca²⁺ ions as spheres. Neighboring protomers colored in white. (C) Structural diagram illustrating the conformational change from the O state to the D state and the D state–dependent inhibition by antagonists. Ligands and S6 gate residue Val⁹⁷⁶ shown as sticks and Ca²⁺ as green spheres. Insets show antagonist binding highlighted in yellow shades and red arrows.