Non-canonical pore architecture underlies constitutive gating of human retinal TRPM1

Abstract

Transient receptor potential melastatin 1 (TRPM1), a Ca²⁺-permeable nonselective cation channel essential for retinal ON bipolar cell signaling and night vision, and implicated in congenital night blindness, has remained structurally and functionally poorly characterized. Here we report the cryo–electron microscopy structure of human TRPM1, revealing that although it assembles as a tetramer, it adopts an unexpected clockwise domain-swapped pore module with rotational geometry inverse to that observed in previously characterized 6-TM tetrameric channels. This inverted topology is accompanied by extensive remodeling of the S5–P–S6 module, dilation of the selectivity filter, expansion of the central cavity, and splaying of S6 to form a wide intracellular gate. Our single-channel recordings reveal constitutive activity consistent with the conductive state captured. Together, these findings uncover a 6-TM fold in the tetrameric channel and provide a framework for understanding TRPM1 gating, disease-associated dysfunction, and associated pharmacology.

TRPM1 is an ion channel essential for retinal signal transmission and pigmentation. Here, the authors determine the cryo-EM structure of human TRPM1, revealing a non-canonical clockwise pore rotation that underlies its constitutive activity and provides insights into retinal disease mechanisms.

Introduction

Members of the transient receptor potential (TRP) superfamily predominantly function as a tetrameric cation-conducting channels that play crucial roles in sensory physiology, including touch, pain, thermosensation, and vision^1–6. Transient receptor potential melastatin 1 (TRPM1) is a calcium-permeable, nonselective cation channel with essential function in visual signal transduction and pigmentation⁷. In the retina, TRPM1 enables synaptic communication of photoreceptors with ON bipolar cells^8,9. In the dark, binding of glutamate to its receptor mGluR6 in the dendritic tips of the ON bipolar cells tonically activates G protein, releasing Gαo and Gβγ subunits, which directly inhibit TRPM1 opening^7,10,11. Stimulation of photoreceptors by light leads to hyperpolarization and a transient suppression of glutamate release from their synaptic terminals. The resulting reduction of mGluR6 activity relieves TRPM1 inhibition by G proteins subunits, allowing channel opening and consequent depolarization of ON bipolar cells^7,9,12,13. Since ON bipolar neurons are the only synaptic partners of rod photoreceptors, TRPM1 is indispensable for rod-mediated scotopic vision under dim-light conditions^14,15. Mice lacking TRPM1 also have cognitive and behavioral abnormalities¹⁶. In humans, loss-of-function mutations in TRPM1 cause congenital stationary night blindness^12,17–22. Additionally, TRPM1 plays a crucial role in melanocyte differentiation, survival, and melanin synthesis, and its reduced expression is associated with melanoma progression and metastasis, highlighting its physiological significance beyond retinal function^23–26. However, despite being the founding member of the TRPM family, the structural basis of TRPM1 assembly and pharmacological modulation has remained largely unknown, greatly limiting mechanistic insight and therapeutic development.

Tetrameric ion channels evolved from simple two–transmembrane pore modules into the canonical six–transmembrane (6-TM) architecture, in which an S1–S4 voltage-sensor or voltage sensor–like domain is functionally coupled to an S5–P–S6 pore module. In most structurally characterized 6-TM channels, including voltage-gated sodium channels, calcium channels, and potassium channels, the pore domain is arranged in a counterclockwise orientation relative to the voltage-sensing domain when viewed from the extracellular side, typically resulting in a domain-swapped architecture^27–33. More recently, non–domain-swapped configurations have been identified, often associated with a shortened S4–S5 linker that maintains the pore within the same subunit^32,34–36. TRP channels adopt a 6-TM domain-swapped tetrameric structural framework but are distinguished by large cytoplasmic N- and C-terminal assemblies that integrate regulatory and signaling inputs^37,38. Structures of TRPM2, TRPM3, TRPM4, TRPM7, and TRPM8 have established shared mechanisms of ion permeation and gating within the subfamily^39–51. In contrast, TRPM1 exhibits a distinct mode of physiological regulation: in retinal ON bipolar cells, it is tonically active in light and is suppressed in darkness by mGluR6 signaling through the G protein Gα_o and βγ^10,11. This specialized mode of regulation suggests that TRPM1 harbors structural features that diverge from established TRPM paradigms. Although TRPM1 shares ~64% sequence identity with TRPM3, its pharmacological properties remain debated, as reported modulation by capsaicin, capsazepine, and voriconazole shows inconsistent effects between retinal bipolar cells and heterologous expression systems across different studies, suggesting unresolved uncertainties in its functional regulation^10,15,52,53. Notably, a previous structural study proposed a dimeric assembly for TRPM1, further highlighting uncertainties surrounding its quaternary organization⁵⁴.

Here, we report the cryo–electron microscopy structure of human TRPM1, which shows non-canonical clockwise domain-swapped pore modules that distinguish it from other characterized tetrameric ion channels and provides a structural basis for its constitutive activity. These findings establish a molecular framework for understanding TRPM1 gating regulation and pharmacology and lay the groundwork for understanding how disease-associated mutations disrupt channel function in visual and pigmentary disorders.

Results

Single-channel activity and constitutive gating of human wild-type TRPM1

The functional characterization of hTRPM1-WT has remained challenging when expressed in heterologous systems⁵⁵, limiting direct measurements of channel activity using whole cell assays. We observed that hTRPM1-WT is localized intracellularly as well as at the plasma membrane in HEK293F cells (Suppl. Fig. 1). Despite detectable surface expression, Fura-2 AM–based intracellular calcium assays failed to detect hTRPM1-WT activation in response to pregnenolone sulfate (PregS), voriconazole, and capsaicin (Fig. 1a, Suppl. Fig. 1b, c). To directly assess channel function, we purified heterologously expressed wild-type human TRPM1 from suspension-adapted HEK293F cells and reconstituted it into lipid bilayers for single-channel recording (Suppl. Fig. 2a). Using this approach, we showed that hTRPM1 exhibited spontaneous channel opening events with open probability (Po) of ~0.7 and a conductance of ~12 pS in the absence of agonist, indicating constitutive channel activity, as previously suggested but not directly measured (Suppl. Fig. 2e)⁷. Addition of PregS further increased the basal open probability to ~0.8 and conductance to ~15.5 pS, consistent with ligand-dependent potentiation rather than obligate ligand-dependent activation (Fig. 1d, e; Suppl. Fig. 2f). Unlike TRPM3, TRPM1 activation by PregS did not require PIP₂, although inclusion of PIP₂ increased P_o and stabilized channel openings (Fig. 1d, e), indicating a supportive role of membrane lipids⁴⁸.

Fig. 1. Functional characterization of human TRPM1 by Fura-2 calcium measurements and single-channel recordings.

a, b Intracellular Ca²⁺ measurements using Fura-2 AM in response to agonists for wild-type TRPM1 (a), the cryo-EM–optimized construct (TRPM1-EM) (b). Inset in (a) shows a western blot analysis of detergent solubilization (DDM, LMNG, Digitonin, GDN) confirming expression of wild-type TRPM1. S, soluble fraction; P, pellet fraction. The cryo-EM–optimized construct (TRPM1-EM) exhibits Ca²⁺ influx in response to pregnenolone sulfate and other modulators. Data are presented as mean ± s.e.m. from n = 3 independent biological experiments. Statistical significance was determined using an unpaired two-tailed Student’s t-test (ns, not significant; **P < 0.01; ***P < 0.001). c Dose–response curve for PregS activation of TRPM1-EM, shows an EC₅₀ of 23.4 ± 4.4 μM. Changes in the F340/F380 ratio were normalized to the maximum response at saturating agonist concentrations and fitted with a logistic equation. Data are presented as mean ± s.e.m. from n = 3 independent biological experiments. d–g Representative single-channel recordings of purified wild-type TRPM1 and TRPM1-EM reconstituted into planar lipid bilayers are shown in the presence of PregS and PIP₂. Wild-type TRPM1 displays spontaneous channel openings and sub-conductance states in the presence of PregS and PIP₂ at −140 mV and −80 mV, respectively (d, e). TRPM1-EM exhibits comparable gating behavior, with channel activity observed at −80 mV and −40 mV (f, g). Closed (C) and open (O) states are indicated. Corresponding amplitude histograms fitted with Gaussian distributions, revealing distinct closed, sub-conductance, and fully open states (right panels). Scale bars: 2 pA, 500 ms.

Strikingly, single-channel recordings revealed a low conductance state ( ~ 1 pA; ~6 pS at −140 mV) that either persisted or transitioned to a fully open state ( ~ 5 pA; ~33 pS at −140 mV), consistent with a multistep gating mechanism (Suppl. Fig. 3a). Channel activity exhibited voltage dependence and lead to increase in Po at more negative potentials (Fig. 1d, e; Suppl. Fig. 2). Together, these data identify TRPM1 as a constitutively active ion channel with discrete conductance states, and distinct pharmacological sensitivity.

Structure determination of TRPM1

To enable structural studies of TRPM1, we optimized channel expression and stability using fluorescence-detection size exclusion chromatography (FSEC)–based detergent screening for mouse and human TRPM1 orthologs^56,57. Among the two, human TRPM1 exhibited the most favorable expression and monodispersed profiles and was therefore selected for further analysis. To improve sample homogeneity for cryo-EM, we generated a truncated construct lacking exon 11 (residues 439–496), a flexible poly-lysine loop (residues 624–631), and truncating the distal C terminus (Δ1479–1642), hereafter termed as hTRPM1-EM. This construct displayed markedly improved monodispersity and stability relative to wild-type hTRPM1 (Suppl. Figs. 1c, 2b).

To determine whether these modifications altered channel function, hTRPM1-EM was reconstituted into lipid bilayers and examined using single-channel recordings. Similar to wild-type hTRPM1, hTRPM1-EM exhibited spontaneous opening events, including a ~1 pA low conductance preceding transition to a fully open state, and channel activity was potentiated by agonists (Fig. 1f, g; Suppl. Fig. 2g, h). Notably, hTRPM1-EM opened at less negative voltages in presence of PS and PIP₂ ( − 40 mV) compared to wild-type hTRPM1, suggesting that exon11 and/or the poly-lysine loop contribute to the regulation of hTRPM1 gating. In addition, hTRPM1-EM retained measurable activity at depolarized potentials ( + 80 and +140 mV), consistent with previous reports⁵⁸ (Suppl. Fig. 3e–h). The observed differences in voltage sensitivity likely arise from variations in experimental context, including heterologous expression, the absence of the native bipolar cell environment, and reconstitution into lipid bilayers. Next, we evaluated whether hTRPM1-EM could serve as a platform for ligand screening and functional characterization of channel activity. Fura-2 AM–based intracellular calcium measurements showed that hTRPM1-EM exhibits reliable agonist-evoked Ca²⁺ influx, validating its utility for probing TRPM1 pharmacology (Suppl. Fig. 4). In contrast to wild-type hTRPM1, hTRPM1-EM showed robust Ca²⁺ responses upon application of PregS, clotrimazole, and voriconazole, activating the channel with moderate potency (EC₅₀ = 23.4 ± 4.4 µM, 8.3 ± 0.6 µM, and 45.3 ± 4.4 µM, respectively), whereas no activation was observed with 2-APB or capsaicin (Fig. 1b, c; Suppl. Fig. 2c, d & 4d–f). Notably, PregS-evoked Ca²⁺ signals were effectively inhibited by 2-APB (IC₅₀ ≈ 44.3 µM), a compound that does not inhibit the closely related TRPM3 channel (Suppl. Figs. 3c, d, 4g, 5). Consistently, planar lipid bilayer recordings revealed a marked reduction in channel activity in the presence of 90 µM 2-APB, with the open probability decreasing to ~0.024 (Suppl. Fig. 3c, d). Together, these results establish hTRPM1-EM as a functional surrogate that enables pharmacological interrogation of TRPM1 and reveals ligand sensitivities distinct from related TRPM family members.

To define the overall architecture of TRPM1, we purified hTRPM1-EM and subjected it to single-particle cryo-EM. Initial 2D class averages revealed a fourfold symmetric tetramer with prominent intracellular secondary structure features corresponding to the melastatin homology regions, consistent with other members of the TRPM subfamily (Suppl. Fig. 6). The final three-dimensional reconstruction was resolved to an overall resolution of 4.36 Å (Suppl. Fig. 6b), with higher local resolution observed in the intracellular MHR domains and the C-terminal coiled-coil region (Suppl. Figs. 6, 7). Several regions, including 80 residues from the N terminus, 211 residues from the C terminus, and few intracellular loops (residues 605–635, 809–853, 1156–1180), were not resolved in the final model, likely due to conformational flexibility.

Overall structure of TRPM1

The overall cryo-EM density map of hTRPM1-EM reveals a broadly conserved architecture within the TRPM family, with intracellular domains closely resembling those of TRPM3. In contrast, the transmembrane region exhibits a divergent, noncanonical organization relative to other six–transmembrane (6-TM) tetrameric ion channels.

From the N-terminus, the protein contains the conserved melastatin homology regions MHR1/2 and MHR3/4 (Fig. 2). The MHR1/2 domains form a clamshell-shaped Rossmann-like fold composed of repeating β–α–β motifs, while the MHR3/4 domains are primarily helical, consisting of helix-turn-helix elements that mediate extensive intersubunit contacts. As in TRPM3, these intracellular domains assemble into a large cytosolic platform that supports the transmembrane region. This intracellular architecture is essential for tetrameric assembly and mediates protein–protein interactions that allosterically regulate channel gating, including interaction with G-proteins^10,11. Following the MHR domains, as in TRPM3 and other TRPM channels, each TRPM1 subunit contributes six transmembrane helices (S1–S6), with helices S1–S4 forming a voltage sensor–like domain (VSLD) and helices S5–P–S6 assembling into the central ion-conducting pore. The intracellular C-terminus begins with the conserved TRP helix, followed by the post-TRP elbow, rib helix, and a central coiled-coil domain that mediates tetramerization (Fig. 2).

Fig. 2. Overall structure and domain organization of human TRPM1.

a Linear domain topology of the cryo-EM–optimized TRPM1 construct (TRPM1-EM), showing major cytosolic domains (Pre-MHR, MHR1/2, MHR3, MHR4), the pre-S1 elbow, transmembrane segments (S1–S4 and S5–S6), TRP and rib helices, coiled-coil domain, and C-terminal truncation. Regions corresponding to exon 11 deletion and loop truncations are indicated (Schematic diagram was created in Adobe Illustrator®). b–d Cryo-EM density of human TRPM1 showing in side (b), top (c), and bottom view (d), with individual subunits colored differently. The detergent micelle is shown as a transparent surface. e–g Corresponding atomic models of TRPM1 in side (e), top (f), and bottom (g) views. Subunits are colored consistently with (b–d). h Structural model of a TRPM1 monomer showing key domains, including the voltage-sensor-like domain (S1–S4), pore domain (S5–S6), pre-S1 region, TRP helix, MHR domains, rib helix, and coiled-coil region and are colored consistently with (a).

Structural discovery of a non-canonical transmembrane architecture in tetrameric ion channels

Tetrameric ion channels are thought to have evolved from simple two–transmembrane pore modules into the canonical six–transmembrane (6-TM) architecture, in which an S1–S4 voltage-sensor or voltage sensor–like domain couples to an S5–P–S6 pore module. Across structurally characterized 6-TM channels, including voltage-gated sodium^29,59–61, calcium, and potassium channels, this arrangement is highly conserved, with the pore domain adopting a counterclockwise orientation relative to S1–S4, leading to a stereotyped domain-swapped or non-swapped topology^{28–33,43,61,62}. Surprisingly, the cryo-EM structure of hTRPM1-EM reveals deviations from this canonical topology observed within the tetrameric ion channels. While its intracellular assembly conforms to the TRPM family and closely resembles TRPM3, the transmembrane region adopts a fundamentally different organization (Suppl. Fig. 8). Each TRPM1 subunit retains the canonical 6-TM fold, yet their tetrameric arrangement is unconventional: the S5–P–S6 pore module is rotated clockwise relative to the S1–S4 domain, producing a topology opposite to that observed in other characterized 6-TM tetrameric channels (Fig. 3; Suppl. Fig. 8a–e).

Fig. 3. Clockwise pore swapping in hTRPM1-EM.

a, b Extracellular views of the tetrameric assemblies of hTRPM1-EM (a) and TRPM3 (b), with individual subunits colored differently and semi-transparent surfaces overlaid on cartoon representations. c, f S1–S4–aligned superposition of hTRPM1-EM (yellow) and TRPM3 (teal) (c), shown after a 90° rotation (f), reveals opposite pore-module rotations: clockwise in TRPM1 and anticlockwise in TRPM3 relative to the VSLD. d Isolated S1–S4 domain of hTRPM1-EM (yellow) shown alongside its associated pore domain (blue), showing domain organization within a subunit. e Comparison of the pore domain (S5–S6; blue) and S1–S4 domain (teal) of TRPM3, highlighting structural features.

Structural superposition of the S1–S4 domains of TRPM1 and TRPM3 reveals that this divergence originates at the S4–S5 junction. Beyond this point, the TRPM1 pore helices undergo substantial translational and rotational rearrangements relative to TRPM3: S5 is displaced by ~51 Å, S6 by ~56 Å, and the pore helix by ~21 Å (Fig. 3). These shifts correspond to an overall clockwise rotation of ~53° of the pore module, with S5 and S6 rotating ~50° relative to S1–S4 domain (Fig. 3c, f). Importantly, this reorganization of pore module is not a rigid-body motion. Instead, the pore domain exhibits internal remodeling, including a marked repositioning of S6 relative to S5 that alters helix packing within the conduction pathway. Together, these features establish TRPM1 as a structurally unique member of the tetrameric ion channel family, providing a framework for understanding its constitutive activity and distinctive gating behavior while expanding current views of how 6-TM channels can be assembled and regulated.

Ion conduction pathway in the TRPM1 pore

To define the functional state captured in the TRPM1 structure, we estimated the pore profile based on the traced Cα backbone and qualitatively compared the pore dimensions at the selectivity filter and intracellular gate with those of TRPM3^49,63. Given the limited side-chain resolution in the pore region, this analysis is intended to provide an overall assessment of the permeation pathway rather than a precise description of side-chain-defined constrictions. Consistent with an open-like conformation, the TRPM1 backbone architecture reveals a substantially wider permeation pathway at the selectivity filter than that observed in the closed-state structure of TRPM3, with an estimated minimum pore radius of ~8.35 Å compared with ~3.65 Å for TRPM3, and is accompanied by prominent lateral fenestrations (Fig. 4a–f & Fig. 5). A similarly widened selectivity filter was previously observed in PS-bound TRPM3, where ligand binding near the pore helix stabilizes an open selectivity filter conformation (Fig. 4d & Fig. 6)^44,63. Like the PregS and CIM0216 bound TRPM3, large lateral fenestrations are seen in the TRPM1 pore (Fig. 5). We observed additional density adjacent to the TRPM1 pore helix, suggesting lipid-mediated stabilization of the open selectivity filter (Fig. 6c).

Fig. 4. Ion permeation pathway of hTRPM1-EM.

a, b Side views of the pore domains of TRPM3 (a) and hTRPM1-EM (b), shown as cartoon representations with the pore surface rendered in gray. hTRPM1-EM displays an expanded pore with outward displacement of S5–S6 helices relative to TRPM3. c Pore-radius profiles calculated along the conduction axis using the HOLE program, comparing hTRPM1-EM (open; salmon) and TRPM3 (closed; teal). The selectivity filter, central cavity, and lower gate regions are indicated. d, e Top views of TRPM3 (d) and hTRPM1-EM (e) pore domains, showing differences in pore diameter. f Bottom view of hTRPM1-EM showing a dilated intracellular gate consistent with an open, conducting conformation.

Fig. 5. Electrostatic surface potential and lateral fenestrations of hTRPM1-EM in the open state.

Electrostatic surface potential of hTRPM1-EM shown in (a) side view, (b) top view, and (c) bottom view. Red and blue colors indicate negatively and positively charged regions, respectively. Major structural features, including the lateral fenestration, central pore, and coiled-coil domain, are labeled.

Fig. 6. Structural features of the pore and comparison of hTRPM1-EM with TRPM3 channels.

a Surface rendering of TRPM3–PS pore in gray, showing a wider opening at the selectivity filter. b Pore-radius profiles calculated using HOLE, comparing hTRPM1-EM with TRPM3 in closed state and in PS–bound state. Distinct regions corresponding to the selectivity filter, central cavity, and lower gate are indicated. c Unidentified lipid-like density observed near the pore helix of hTRPM1-EM, as viewed from extracellular side, d–f Structural superposition of hTRPM1-EM (salmon) with TRPM3α2 in the open state (blue; PDB ID: 9N1N). Side and top views highlight conformational differences in the transmembrane (TM) domain and the MHR regions. g–i Structural comparison of hTRPM1-EM (salmon) with rTRPM1-EM (yellow; PDB ID: 9OR5). Rotated views illustrate overall structural conservation as well as subtle rearrangements in the TM domain and S1–S4 domain.

Immediately below the selectivity filter, TRPM1 contains an expanded hydrophobic cavity lined by nonpolar residues, substantially larger than that of TRPM3, potentially accommodating hydrated cations during permeation (Fig. 4b, e). The intracellular activation gate is located at the S6 bundle crossing. In the TRPM1 structure, this gate is fully open, with a diameter of ~8.5 Å measured between opposing N1115 side chains (Fig. 4b, c, f). The expanded selectivity filter and fully open gate indicate that TRPM1 is captured in a constitutively conductive state.

Gating mechanism of TRPM1

The gating architecture of TRPM1 is closely coupled with a clockwise rotation of its pore domain relative to the voltage sensor–like domain, which leads to a distinct spatial arrangement of the S5–P–S6 helices. This rearrangement is associated with an expanded selectivity filter compared with TRPM3 and a widened intracellular gate, collectively biasing the channel toward a conductive conformation. The selectivity filter of TRPM1 adopts a comparatively open geometry that is consistent with permeation of partially or fully hydrated cations, in contrast to TRPM3, where the narrower filter favors ion dehydration during permeation (Fig. 7a, c, d). In TRPM3 resting closed state, the C-terminal end of the pore helix is positioned deeper within the membrane (approximately 20 Å from the extracellular side) and oriented toward the central hydrophobic cavity, where it contributes to stabilization of permeating ions (Fig. 7a). In contrast, in TRPM1 the C-terminal end of the pore helix lies closer to the extracellular leaflet (approximately 11 Å within the membrane) and adopts a more membrane-parallel orientation rather than projecting toward the cavity (Fig. 7c). This altered geometry is accompanied by an enlarged hydrophobic cavity, likely providing additional space to stabilize hydrated or partially hydrated ions. Notably, a similar rearrangement of the pore helix and selectivity filter expansion has been observed in the PregS–bound state of TRPM3 (Fig. 6, Fig. 7b), where channel activation correlates with increased selectivity filter dilation. At the intracellular bundle crossing, the S6 helices diverge substantially, forming a fully open gate lined by I1111, with a wide constriction consistent with an activated conformation (Fig. 4b, f & Fig. 7d). S6 exhibits pronounced outward kinking and splays away from the ion conduction axis, further contributing to gate opening (Fig. 7d). Notably, the connection between S6 and the TRP domain adopts a loop-like configuration rather than a continuous helix (Fig. 7h), a feature not observed in other TRPM channels during opening. The S6–TRP helix linker has been shown to adopt distinct conformations across different gating states of TRPM7 and TRPM8, suggesting that this region plays an important role in defining the gating behavior of TRPM family channels^46,51.

Fig. 7. Gating Mechanism of TRPM1.

a–c Side views of the pore region of closed state of TRPM3 (a), the PregS-bound state (b), and conducting pore of hTRPM1-EM (c) showing vertical displacement of the pore helix and selectivity filter. Lines indicate the relative membrane depth of the pore helix measured from the extracellular side. d The S6 segment of hTRPM1-EM forms a continuous α helix positioned ~8.5 Å away from the ion conduction axis (dark reference line drawn parallel to S6), in contrast to the kinked S6 observed in closed TRPM3. e, f Structural alignment of hTRPM1-EM (salmon) with closed TRPM3 (teal), referenced to the S1–S4 paddle domain, reveals coordinated repositioning of the MHR1/2, MHR3/4, rib, and coiled-coil domains, shown in side view (e) and bottom view (f). g Close-up view of the interface between the α21–α22 linker within the MHR3/4 domain that engages with the pre-S1 segment of a neighboring subunit in hTRPM1-EM. h Structural remodeling at the S6–TRP helix, S4-S5 elbow.

Structural alignment of TRPM1 with TRPM3 based on the S1–S4 scaffold (PDB: 8DDQ) reveals pronounced differences in the relative positioning of the cytoplasmic MHR1/2 and MHR3/4 domains, indicating coupling between intracellular regulatory elements and the transmembrane gate (Fig. 7e, f). More specifically, these comparisons show that, relative to the closed-state TRPM3 structure, the MHR1/2 domain in TRPM1 is moved away from central axis while the MHR3/4 domain twists toward the transmembrane region, the conformational rearrangements previously associated with channel activation in other TRPM family members^42,44,46. Consistent with an activated state, the D702 and D703 from α21–α22 motif of MHR3/4 engages the Y727 and T766 of pre-S1 elbow of a neighboring subunit, an interaction characteristic of open-state TRPM4, TRPM7, and TRPM8 structures (Fig. 7g)^46,47,51. These cytoplasmic rearrangements correlate with the expanded ion conduction pathway observed in TRPM1, indicating that the structure captures a conductive conformation. The structural superposition of the TRPM1-EM open state with the TRPM3 open state (PDB:9N1N) suggests that the MHR1 domain undergoes an outward displacement relative to the coiled-coil domain, while the pore helix adopts a more membrane-parallel orientation (Fig. 6d–f). Additionally, substantial remodeling occurs at the interface between the S4–S5 elbow, the TRP domain, and the pre-S1 region, where residues W776, Y994, and W1129 adopt altered conformations that may mark the pivot point for the clockwise swapping of the S4–S5 linker (Fig. 7h). Together, the clockwise rotation of the pore domain, dilation of the selectivity filter, and splayed S6 gate define a gating architecture in which TRPM1 is intrinsically biased toward conduction. This structural configuration provides a mechanistic basis for the constitutive activity of TRPM1 and distinguishes it from other TRPM channels that rely on anticlockwise pore rotation for activation.

Discussion

In this study, we establish that full-length human TRPM1 assembles into a bona fide ion channel capable of spontaneous and ligand-modulated activity when reconstituted into planar lipid bilayers (Fig. 1). Our single-channel recordings show that both the full-length protein and the cryo-EM–optimized construct (hTRPM1-EM) exhibit comparable basal gating and PregS-dependent potentiation, with only modest differences in voltage dependence, indicating that the engineered construct preserves the intrinsic functional properties of the channel (Fig. 1c, d). The agonist-evoked Ca²⁺ influx was detectable only with hTRPM1-EM in Fura-2–based cellular assays. The absence of a detectable Ca²⁺ influx in TRPM1-WT may likely reflect constraints of the heterologous expression system, as noted previously, rather than an intrinsic loss of channel function^26,58. Importantly, these findings align with prior work showing that removal of an exon-encoded segment improves functional detection of TRPM1 without disrupting channel activity⁵⁸. Notably, our finding showed that TRPM1-EM was inhibited by 2-APB, in contrast to the closely related TRPM3 channel, which is insensitive to this compound, highlighting a distinct mode of pharmacological modulation (Suppl. Figs. 3e, f & 4g). This differential sensitivity reveals gating features of TRPM1 that were previously difficult to resolve because of challenges in measuring its activity. Given that 2-APB is known to modulate multiple TRP channels through diverse mechanisms, its inhibitory effect on TRPM1 suggests the presence of unique allosteric sites or conformational states that are not conserved in TRPM3. Notably, 2-APB has been reported to inhibit the constitutively active TRPV6 channel while activating TRPV1–4, suggesting its context-dependent and channel-specific actions^64–67. In this framework, the sensitivity of TRPM1 to 2-APB further supports the notion of distinct gating and regulatory features within the TRPM subfamily. Together, these results validate that hTRPM1-EM is a faithful structural surrogate and support the conclusion that the cryo-EM structure captures physiologically relevant channel behavior⁵⁸.

Although functional measurements indicate that both hTRPM1-WT and hTRPM1-EM are potentiated by PregS, the cryo-EM structure determined here under apo conditions captures a pore in an open configuration with a noncanonical architecture. Despite strong sequence conservation and overall structural similarity between TRPM1 and TRPM3 (Suppl. Fig. 5), the clockwise pore-domain arrangement in TRPM1 repositions residues equivalent to those forming the PregS-binding interface in TRPM3, where ligand coordination occurs at the junction of the pore helix (residues W1040 and F1047) and neighboring S1 segment (residues N890 and Y891) in the canonical anticlockwise topology⁶³. In TRPM1, this geometry is disrupted by the reversed domain organization, implying that PregS recognition may occur through an alternative structural configuration (Fig. 7a, d). Determining how PregS engages TRPM1 within this unique architectural framework will be essential for understanding ligand-dependent modulation in channels with noncanonical pore organization and represents an important direction for future investigation.

Structural comparison of TRPM1 with other six–transmembrane (6-TM) ion channels reveals a non-canonical domain-swapping arrangement in which the pore module is rotated clockwise relative to the voltage sensor–like domains (Suppl. Fig. 8a–f). In contrast, all previously characterized 6-TM tetrameric channels, whether domain-swapped or non-swapped, display a canonical anticlockwise pore orientation^27,61,68 (Fig. 8a, b). Similar to classical tetrameric channels, the TRPM1 pore forms an inverted teepee-like assembly of S5 and S6 helices that encloses a hydrated central cavity⁶. Despite its reversed topology, the clockwise arrangement preserves a geometry capable of stabilizing permeant ions within this cavity (Fig. 4). The outward orientation of the pore helix and selectivity filter in the clockwise swapped state along with the splaying away of S6 from the ion conduction axis stabilizes a wide ion conduction pathway in TRPM1 (Fig. 7c), explaining the constitutive activity of the full-length channel observed in our single channel recordings and reported previously^7,11,26,69 (Fig. 1). Recently, the structure of rat TRPM1 (rTRPM1) has been reported, and its superimposition with human TRPM1 reveals a high degree of structural conservation, with an RMSD of ~1.8 Å, indicating near-identical overall architecture (Fig. 6g–i). Both structures display a similar global organization, including a clockwise rotation of the pore domain and a widely open conduction pathway, consistent with an activated state. However, in contrast to human TRPM1, which adopts a fully C4-symmetric arrangement, rTRPM1 displays symmetry mismatch, with a fourfold-symmetric intracellular domain (ICD) and a twofold-symmetric transmembrane domain (TMD) (Fig. 6g–i). Despite this structural similarity, notable functional differences are evident. rTRPM1 has been reported to lack sensitization to pregnenolone sulfate, whereas human TRPM1-EM exhibits channel activity, as demonstrated by both lipid bilayer recordings and Fura-2–based intracellular calcium measurements. This divergence suggests that subtle differences may underlie species-specific functional regulation despite the conserved structural framework⁶⁹.

Fig. 8. Proposed model for pore architecture and gating in TRPM1.

a Schematic representation of the canonical counter-clockwise pore swapping observed in typical six–transmembrane (6-TM) tetrameric ion channels, where each subunit contributes its pore domain to the adjacent subunit in a counter-clockwise arrangement when viewed from the extracellular side. b Non-canonical clockwise pore swapping observed in TRPM1. In contrast to the canonical arrangement, the pore domain of each subunit engages the neighboring subunit in a clockwise orientation, resulting in a distinct transmembrane topology and pore architecture. c Proposed gating mechanism of TRPM1 based on structural comparison with the closed-state structure of TRPM3. TRPM1 adopts a constitutively open conformation stabilized by clockwise pore-domain swapping. Intracellular MHR3/4 elements, particularly residues D702 and D703 within the α21–α22 motif, engage Y727 and T766 located in the pre-S1 elbow of a neighboring subunit. This inter-subunit interaction resembles open-state configurations observed in TRPM4, TRPM7, and TRPM8 (Fig. 7g) and likely contributes to stabilization of the conductive state. Additionally, an upward displacement of the pore helix toward the extracellular membrane side suggests a dynamic pore architecture that may facilitate gating transitions. (Diagrams were created in Adobe®).

Gating in TRPM channels is thought to arise from coordinated intracellular conformational changes that propagate to the lower S6 gate. Structural studies of TRPM4/5/7/8 have identified recurring activation markers, including coupling of the MHR3/4 domain to the voltage sensor–like domain (VSLD) of the cognate subunit and to the pre-S1 elbow of a neighboring subunit, as well as elongation of the lower S6 segment accompanied by shortening of the TRP helix^46,47,51,70. Although the stimuli that trigger these transitions differ across TRPM family members, the underlying mechanical coupling appears conserved. TRPM1 exhibits several of these activation-associated features despite being purified in an apo state, most notably rotation of the MHR1/2 domain and engagement of the MHR3/4 α21–α22 linker with the neighboring pre-S1 segment (Fig. 8c). These conformational signatures are consistent with stabilization of an open pore architecture and align with the constitutive activity observed electrophysiologically. Importantly, this intersubunit interface provides a structural context for regulation by Gβγ subunits in retinal bipolar cells. Gβγ binds to an exon-encoded segment within the MHR3/4 domain conserved between TRPM1 and TRPM3⁴⁰, and in TRPM3 such binding disrupts α21–α22 interactions with the pre-S1 region to favor a closed state^63,71(Fig. 8b). By analogy, Gβγ engagement in TRPM1 is likely to weaken this stabilizing contact, shifting the conformational equilibrium toward pore closure⁶³.

Finally, disease relevance reinforces the functional importance of this architecture. Mutations in TRPM1 cause congenital stationary night blindness by disrupting signal transmission from photoreceptors to ON-bipolar cells, with missense and nonsense variants impairing channel activity through truncation or destabilization of folding, tetramer assembly, and gating. Structural mapping localizes multiple disease-associated residues to subunit interfaces and S1–S4–pore contacts, revealing integrity hotspots not apparent from sequence analysis alone (Suppl. Fig. 9). TRPM1 function further depends on its assembly within a bipolar-cell macromolecular complex, where perturbations in associated components likewise compromise channel localization and signaling, suggesting the structural interdependence of this pathway^7,52. These observations connect a non-canonical pore architecture to constitutive gating and disease susceptibility, establishing a structural framework that links channel organization to retinal pathophysiology.

Methods

All plasmids generated in the manuscript are available upon request.

Construct Design

The cDNA encoding human TRPM1 (hTRPM1, splice variant with accession NP_001238949), originally cloned in pDONR221, was obtained from the DNASU Plasmid Repository (Clone ID HsCD00821818) and subcloned into a modified pEG-BacMam mammalian expression vector. The optimized construct (hTRPM1-EM) was engineered to improve biochemical behavior by deleting exon 11 (residues 439–496), removing a polylysine segment (residues 624–631), and truncating the distal C terminus (Δ1479–1642). The modified pEG-BacMam vector encodes, downstream of the hTRPM1 C terminus, a thrombin protease cleavage site (LVPRGS), followed by an enhanced green fluorescent protein (eGFP) reporter⁵⁶ and a Strep affinity tag (WSHPQFEK), enabling fluorescence-detection size-exclusion chromatography screening and affinity purification. For large-scale expression and structural studies, a non-GFP construct lacking the eGFP fusion was generated. All plasmids were generated using standard molecular cloning techniques and verified by DNA sequencing.

Protein expression and purification

Recombinant bacmids were made in DH10Bac cells. Baculovirus encoding hTRPM1-WT, and hTRPM1-EM were generated and amplified in Sf9 insect cells (Thermo Fisher Scientific, 11496015) using established protocols⁷². Briefly, to produce the baculovirus, Sf9 cells were transfected with approximately ~21 μg of bacmid DNA using Cellfectin II (Gibco, 10362100), and baculovirus was harvested after 5–6 days, filtered through a 0.22 μm membrane filter, and stored at 4 °C. For virus amplification, 600 ml of Sf9 cells cultured in SF900 III medium (Gibco, 12659017) were infected and incubated for 88–96 h, followed by concentration of P2 virus by ultracentrifugation at 63,900 g using a T-647.5 rotor (Sorvall WX100, Thermo Fisher Scientific) prior to transduction of HEK293F cells ((Thermo Fisher Scientific, Cat# R7900). hTRPM1-WT and hTRPM1-EM was expressed in suspension-adapted HEK293F cells (Gibco, R79007) cultured in FreeStyle 293 Expression Medium (Gibco,12338018) at 37 °C with 5% CO₂. Expression conditions were adapted with minor modifications from protocols established for TRPV6³¹ and GPR158⁷³. Cells were infected at a density of 2.6–3.0 × 10⁶ cells ml⁻¹ with P2 BacMam virus. Sodium butyrate (10 mM) was added 8–10 hours post-transduction, and cultures were incubated for an additional 66–76 hours at 30 °C before harvesting by centrifugation at 5422 g for 20 min. Cell pellets were washed once with 1x PBS (pH 8.0) and collected by centrifugation.

All protein purification steps were performed at 4 °C. Cell pellets from 2 L of culture were resuspended in lysis buffer containing 20 mM Tris-HCl (pH 8.0), 150 mM NaCl, 1 mM β-mercaptoethanol, and protease inhibitors (1.5 mM PMSF, 4.5 μM leupeptin, 1.5 μM pepstatin A, and 1.0 μM aprotinin) and lysed by sonication (12 × 10-s pulses, amplitude 20, with 15-s cooling intervals). Cell debris was removed by centrifugation at 9900 g for 18 min, and membranes were pelleted by ultracentrifugation at 186,000 g for 1 hour. Membranes were homogenized and solubilized for ~2.5 h in buffer containing 20 mM Tris-HCl (pH 8.0), 150 mM NaCl, 1% (w/v) digitonin, and 1 mM β-mercaptoethanol. Insoluble material was removed by ultracentrifugation at 186,000 g for 1 hour. The clarified supernatant was incubated with streptavidin-coupled resin for 11–14 h. After washing the resin with 20–30 column volumes of SEC buffer (20 mM Tris-HCl pH 8.0, 150 mM NaCl, 1 mM β-mercaptoethanol, 0.01% digitonin), bound protein was eluted with SEC buffer supplemented with 2.5 mM D-desthiobiotin. Eluted hTRPM1-WT and hTRPM1-EM was concentrated using a 100-kDa MWCO centrifugal filter and clarified by centrifugation at 40,000 g for 30 min. The sample was further purified by size-exclusion chromatography on a Superose 6 10/300 GL column equilibrated in SEC buffer. Peak fractions corresponding to tetrameric hTRPM1-WT and hTRPM1-EM were supplemented with 2.5 mM TCEP and concentrated to ~0.26 mg ml⁻¹ and ~2.2 mg ml⁻¹ respectively, for single channel recording and for cryo-EM analysis.

Cryo-EM sample preparation and data acquisition

Quantifoil R1.2/1.3 holey carbon film grids with a 300-mesh gold support were used for cryo-EM sample preparation as described previously⁷⁴. Immediately prior to vitrification, grids were rendered hydrophilic by glow discharge (2 mA for 60 s). Cryo-EM samples were prepared by applying 3 μl of purified hTRPM1-EM to the gold side of the grid using an FEI Vitrobot Mark IV maintained at 4 °C and 100% humidity. After a 20 s incubation, grids were blotted for 3 s with a blot force setting of 3 and plunge-frozen in liquid ethane. Cryo-EM data were collected on a Titan Krios G4 transmission electron microscope operated at 300 kV and equipped with a Gatan K3 Summit direct electron detector and a post-column GIF Quantum energy filter in counting mode. Movies were recorded at a nominal magnification of 81,000×, corresponding to a calibrated pixel size of 1.1 Å. A total of 6,752 dose-fractionated movies were acquired across a defocus range of −1.0 to −2.5 μm. Each movie comprised 50 frames with a cumulative exposure of 55 e⁻ Å⁻² collected over 2.0 s, corresponding to an estimated dose rate of ~33 e⁻ s⁻¹ per physical pixel.

Cryo-EM image processing and 3D reconstruction

The dataset was processed using RELION⁷⁵ and CryoSPARC⁷⁶, or both (refer to Figs. S5, S6, and Suppl. Table 1 for details). A total of 6,752 movies, each comprising 50 frames, were corrected for beam-induced motion using patch motion correction, followed by estimation of the contrast transfer function (CTF) with the patch CTF estimation routine in CryoSPARC v4.5.1⁷⁶. Following CTF estimation, the exposure curation tool available in CryoSPARC v4.5.1⁷⁶ was used to reject poor-quality micrographs, viz those with a CTF fit value lower than 6 Å, in addition to outliers astigmatism, resulting in 6543 micrographs for further processing. A small subset of micrographs was used initially to select representative particles/templates using blob and template picker respectively, in CryoSPARC v4.5.1, followed by reference-free 2D classification. Finally, a TOPAZ model was first trained against a good set of ~22517 particles representing multiple views of hTRPM1-EM, and then this trained model was subsequently used for particle picking on the entire dataset⁷⁷. The TOPAZ extraction led to a total of 3,867,524 particles. Following extraction, the particles were aligned using multiple rounds two-dimensional class averaging and subsequently removing the junk particles; this was followed by ab-initio reconstruction, heterogenous, homogenous, and non-uniform (NU) refinements, resulting in two major classes (111,508 and 94,538 particles). The final map was obtained from the 94,538-particle subset and refined with C4 symmetry, yielding a reconstruction of hTRPM1-EM at an overall resolution of 4.36 Å, suitable for model building. In contrast, the larger class, refined under C1 symmetry, remained at a lower resolution and did not improve with higher symmetry, particularly in the transmembrane and pore regions, indicating structural heterogeneity. The reported resolution of the final maps was estimated with the gold-standard Fourier shell correlation (GSFSC) at FSC = 0.143 cut-off⁷⁸ as implemented in CryoSPARC v4.5.1. The local resolution estimation was done using the local resolution estimation tool in CryoSPARC v4.5.1⁷⁶. EM density of the reconstructed maps was visualized using UCSF Chimera⁷⁹.

Model building and refinement

The cryo-EM reconstruction of the hTRPM1-EM achieved an overall resolution of 4.36 Å, with the intracellular domain resolved better compared to transmembrane domain reconstructions. We utilized a model predicted by AlphaFold as a reference to build the intracellular and S1-S4 domain of the hTRPM1-EM using real-space refinement in COOT⁸⁰. S5-P-S6 pore domain was built manually with the help of bulky residues as a guide to build the pore domain. The side chain density corresponding to bulky amino acids such as tryptophan, tyrosine, phenylalanine, and arginine helped in correct sequence assignment, with each residue manually adjusted and iteratively built into the cryo-EM density using COOT⁸⁰. All models underwent real-space refinement against their respective cryo-EM maps with restrained group ADP refinement and were subsequently validated using comprehensive cryo-EM validation in PHENIX⁸¹. The UCSF Chimera⁷⁹, ChimeraX⁸², and PyMOL⁸³ were used for structure visualization and figure preparation. The cryo-EM data collection parameters and refinement statistics are shown in Suppl. Table 1.

Planar lipid-bilayer recordings

Ion channel activity of purified human hTRPM1-WT and hTRPM1-EM were examined using planar lipid bilayer recordings as previously described in ref. ⁸⁴. A synthetic lipid mixture containing 1-palmitoyl-2-oleoyl-glycero-3-phosphocholine (POPC), 1-palmitoyl-2-oleoyl-glycero-3-phosphoethanolamine (POPE), and 1-palmitoyl-2-oleoyl-glycero-3-phosphoglycerol (POPG) at a 3:1:1 molar ratio was prepared at 30 mM in n-decane. Lipid bilayers were formed across a ~ 150-μm aperture of a Meca chip (Nanion) equipped with integrated Ag/AgCl electrodes. Bilayers were separated from symmetric bathing solutions containing 10 mM HEPES (pH 7.4), 30 mM CaCl₂, and 10 mM dextrose. All reagents were of ultrapure grade ( > 99%, Sigma-Aldrich). Bilayer formation was monitored by capacitance measurements, typically ranging from 5 to 9 pF. For channel reconstitution, 5 μl of hTRPM1-WT or hTRPM1-EM ( ~ 0.002 μg μl⁻¹) was mixed with 50 or 100 μM phosphatidylinositol 4,5-bisphosphate (PIP₂), with or without pregnenolone sulfate, and incubated with 5 μl lipid mixture at 30 °C for 30 min before painting onto the aperture. Unitary currents were recorded using an Orbit Mini amplifier (Nanion Technologies) controlled by Element Data Reader 3 software. Signals were acquired and digitized at 1.22 kHz. Single-channel events were analyzed using Clampfit together with the Element Data Analyzer (Nanion). Only recordings containing a single incorporated channel were included in the analysis; traces with multiple channel insertions were excluded. All experiments were performed at 20 °C, and data are reported as mean ± SEM.

Fura-2 Calcium Flux Assays

hTRPM1-WT and hTRPM1-EM constructs lacking GFP were expressed in HEK293F cells as described above and performed Fura-AM assay as described previously^34,67. Briefly, sixty hours after viral infection, cells were harvested by centrifugation at 300 g for 2 min and resuspended in divalent-free HEPES-buffered saline (118 mM NaCl, 4.8 mM KCl, 5 mM D-glucose, and 10 mM HEPES, pH 7.4). Cells were loaded with 5 μg ml⁻¹ Fura-2 AM (Thermo Fisher Scientific) and incubated for 45 min at 37 °C in the dark with shaking at 180 rpm. After a single wash, cells were incubated for an additional 30 min under identical conditions to allow dye de-esterification. Cells were washed twice and resuspended in HEPES-buffered saline supplemented with 2.5 mM CaCl₂ and 1 mM MgCl₂, then maintained on ice in the dark. Fluorescence measurements were performed at room temperature using a Cary Eclipse spectrofluorometer (Agilent Technologies) with continuous stirring. Excitation wavelengths were alternated between 340 and 380 nm at 1 s intervals, and emission was recorded at 510 nm. Intracellular Ca²⁺ changes were quantified as the 340/380 fluorescence ratio. Baseline signals were recorded for 1 min before addition of compounds, including pregnenolone sulfate, Diclofenac, voriconazole, 2-APB, and CIM0216⁶⁷. Dose–response relationships and EC₅₀ values were calculated using Origin 9.1.

Confocal microscopy imaging of hTRPM1 constructs

HEK293F cells were transiently transfected with GFP-tagged human TRPM1 constructs (hTRPM1-WT and hTRPM1-EM) using Lipofectamine™ 3000 Transfection Reagent (Invitrogen, L3000001) and plated onto poly-D-lysine–coated coverslips. Cells were harvested 36–48 h post-transfection, washed with 1× PBS (pH 7.4), and incubated with DiD dye for 20 min at 37 °C in a humidified incubator maintained at 5% CO₂. Excess dye was removed by washing the cells three times with 1× PBS. Cells were then fixed with 4% paraformaldehyde for 20 min at room temperature, followed by two washes with 1× PBS. Subsequently, cells were incubated with DAPI for 10 min at room temperature, washed three times with 1× PBS, and mounted using Fluoromount-G. For confocal microscopy, cells were imaged using a laser-scanning confocal microscope (LSM780 NLO, Carl Zeiss GmbH) equipped with a photomultiplier tube (PMT) and GaAsP detectors for signal acquisition. TRPM1 localization was assessed using a plasma membrane marker (DiD, red) and a nuclear dye (DAPI, blue), while GFP fluorescence (green). Images were acquired with identical settings across all samples and processed in ImageJ image analysis software^85,86.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

M.S., K.V., K.A.M., and A.K.S. conceptualized the study. Methodology was developed by M.S. and K.V. Investigation was carried out by M.S., K.V., and A.K.S. Validation was performed by M.S., K.V., V.K., A.C., S.N., N.V., R.C., and A.K.S. TRPM1-WT and TRPM1-EM protein purification and single-channel electrophysiology was done by M.S., and K.V. cloned the TRPM1-EM expression construct and performed the Fura-2 calcium imaging assays. M.S. processed the cryo-EM data and built the structural models under the supervision of A.K.S. Visualization was carried out by M.S., K.V., and A.K.S. Funding acquisition was provided by K.A.M. and A.K.S. Resources, project administration, and supervision were provided by A.K.S. The original draft was written by M.S., K.V., and A.K.S., and review and editing were performed by M.S., K.V., K.A.M., and A.K.S.

Peer review

Peer review information

Nature Communications thanks the anonymous reviewers for their contribution to the peer review of this work. A peer review file is available.

Funding

This work was supported by ICMR IIRPIG-2024-01-00662, EY034339 (to K.A.M. and A.K.S.), DBT-NIH RO1 IC-12048(12)/1/2022-ICD-DBT (to A.K.S), Department of Biotechnology BT/PR52447/MED/30/2538/2024, and ANRF/ARG/2025/001924/LS Kanpur (to A.K.S.). M.S., and V.K. acknowledges the CSIR, India; K.V. and A.C. acknowledge MHRD for the fellowships.

Data availability

The cryo-EM density maps and atomic coordinates have been deposited in the Electron Microscopy Data Bank (EMDB) and the Protein Data Bank (PDB), respectively, under accession codes EMD-81936 and PDB 43KC. Source data are provided with this paper. All other materials generated in this study are available from the corresponding author upon request. The source data underlying Figs. 1a, b, 6c, and Supplementary Fig. 4d–g are provided as a source data file. The atomic coordinates used for Supplementary Fig. 8 include: apo TRPM3 (PDB ID: 8ED7) [https://www.rcsb.org/structure/8ED7], KcsA (PDB ID: 1K4C) [https://www.rcsb.org/structure/1K4C], Kv1.2 (PDB ID: 2A79) [https://www.rcsb.org/structure/2A79], NavAb (PDB ID: 3RVY) [https://www.rcsb.org/structure/3RVY], TRPV1 (PDB ID: 5IRZ) [https://www.rcsb.org/structure/5IRZ], TRPM3 (PDB ID: 8DDQ) [https://www.rcsb.org/structure/8DDQ]. The structures used in Fig. 6 include: TRPM3 complex with cholesteryl hemisuccinate (PDB ID: 9B29) [https://www.rcsb.org/structure/9B29], TRPM3 in complex with CIM0216 (PDB ID: 9N1N) [https://www.rcsb.org/structure/9N1N], mTRPM3 in complex with PregS and CIM 0216 (PDB ID: 9B2A) [https://www.rcsb.org/structure/9B2A], rat TRPM1 (PDB ID: 9OR5) [https://www.rcsb.org/structure/9OR5]. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41467-026-76454-y.