A duo of redox-sensitive pore-loop cysteines controls the activity of the neural ion channel TRPM3

To investigate the modulatory effects of redox reagents on the activity profile of TRPM3, we performed whole-cell patch clamp recordings on HEK293T cells stably expressing murine TRPM3α2 (HEK-TRPM3). The TRPM3 activity during application of the agonist pregnenolone sulfate (PS) 17 and the TRPM3 modulator clotrimazole (Clt) 8 was monitored in standard, non-treated (NT) conditions and after pretreatment with the reducing agent 1,4-Dithiothreitol (DTT-treated, 10 mM for 40 to 60 minutes). Application of PS (40 µM) evoked robust activation of outwardly rectifying currents in NT HEK-TRPM3 cells, but was without effect in DTT-pretreated cells (Fig.  1A, B and D ). Oppositely, in line with earlier studies 8 , sole application of Clt did not evoke any detectable current activation in NT HEK-TRPM3 cells, but activated currents with a linear current-voltage relation in DTT-pretreated cells (Fig.  1A, B and D ). In NT cells, application of PS in the continued presence of Clt evoked a strong potentiation of the PS-induced currents, including a large inwardly rectifying current component, in line with earlier work 8 . In contrast, PS caused a significant inhibition of the Clt-evoked currents in DTT-treated cells (by 65.3 ± 6.0%) (Fig.  1B and D , and Supplementary Table  1 ). Importantly, non-transfected HEK cells pretreated with DTT did not exhibit any current activation upon stimulation by PS and Clt (Supplementary Fig.  1 ), similar to what was reported earlier 8 . Since DTT is membrane-permeable, it could affect both extracellular and intracellular redox-sensitive processes. To assess the sidedness of the observed redox effect, the membrane-impermeable reducing agent tris(2-carboxyethyl)phosphine (TCEP) was tested in further experiments. Extracellular exposure of HEK-TRPM3 cells to TCEP resulted in comparable modifications of the TRPM3 responsivity towards PS and Clt stimulation as in DTT-treated conditions (Fig.  1C and D , and Supplementary Table  1 ). These results indicate that the redox modulation of TRPM3 is occurring at the extracellular side of the plasma membrane. Fig. 1 Sensitivity of heterologous TRPM3 to chemical activating stimuli is reversibly modulated by redox reagents. A Time course of whole-cell currents at ±80 mV recorded in HEK-TRPM3 cells upon application of PS (40 µM), Clt (10 µM) and co-application of PS+Clt. Inset: corresponding Current (I)–Voltage (V)-plot. Colors represent similar colored time points in the time course. B Same as in ( A ) but for HEK-TRPM3 cells pretreated ≥40 min with the reductive agent DTT (10 mM). C Similar as in ( B ) but cells were pretreated (≥40 min) with TCEP (10 mM). D Normalized current changes for HEK-TRPM3 cells during application of the different stimuli (X) as used in ( A )–( C ) for non-treated (NT) cells and cells pretreated with the reductive agents DTT or TCEP. Normalization was accomplished by dividing the current increases by the sum of the current increases upon PS stimulation and upon Clt stimulation. Statistical comparisons were performed with Kruskal-Wallis test and subsequent Dunn’s multiple comparison posthoc test or One-way ANOVA with subsequent Tukey’s multiple comparison posthoc test ( n  = 5 for NT, n  = 12 for DTT, n  = 6 for TCEP; for PS and for Clt: Kruskal-Wallis: p  = 6.6*10 − 04 , H = 14.65., DF = 2, Dunn’s post-hoc: NT vs. DTT, p  = 9.1*10 − 03 , NT vs. TCEP, p = 1.36*10 −04 , TCEP vs. DTT, p  = 6.53*10 −02 ; for PS+Clt: Kruskal-Wallis: p  = 3.45*10 −03 , H = 11.34., DF = 2, Dunn’s post-hoc: NT vs. DTT, p  = 1.98*10 −03 , NT vs. TCEP, p  = 3.05*10 −03 , TCEP vs. DTT, p  = 0.77. E Mean time course of fluorescence ratios (F340/F380) of HEK-TRPM3 cells recorded with calcium fluorimetric measurements upon application of indicated TRPM3 agonists/modulators PS (40 µM), and Clt (10 µM) in non-treated conditions and DTT-pretreatment (10 mM for ≤ 40 min). F Normalized fluorescence ratio increases during application of PS, Clt, and PS+Clt in non-treated or DTT-pretreated conditions. Statistical analyses between DTT-treated and non-treated groups were performed with a Student’s unpaired t-test ( N  = 3 dishes for all conditions; PS: p  = 4.20*10 −08 , t -value = 109.34, DF = 4; Clt: p  = 4.20*10 −08 , t -value = -109.34, DF = 4; PS+Clt: p  = 0.14*10 −02 , t -value = 24.96, DF = 2.06). A Time course of whole-cell currents at ±80 mV recorded in HEK-TRPM3 cells upon application of PS (40 µM), Clt (10 µM) and co-application of PS+Clt. Inset: corresponding Current (I)–Voltage (V)-plot. Colors represent similar colored time points in the time course. B Same as in ( A ) but for HEK-TRPM3 cells pretreated ≥40 min with the reductive agent DTT (10 mM). C Similar as in ( B ) but cells were pretreated (≥40 min) with TCEP (10 mM). D Normalized current changes for HEK-TRPM3 cells during application of the different stimuli (X) as used in ( A )–( C ) for non-treated (NT) cells and cells pretreated with the reductive agents DTT or TCEP. Normalization was accomplished by dividing the current increases by the sum of the current increases upon PS stimulation and upon Clt stimulation. Statistical comparisons were performed with Kruskal-Wallis test and subsequent Dunn’s multiple comparison posthoc test or One-way ANOVA with subsequent Tukey’s multiple comparison posthoc test ( n  = 5 for NT, n  = 12 for DTT, n  = 6 for TCEP; for PS and for Clt: Kruskal-Wallis: p  = 6.6*10 − 04 , H = 14.65., DF = 2, Dunn’s post-hoc: NT vs. DTT, p  = 9.1*10 − 03 , NT vs. TCEP, p = 1.36*10 −04 , TCEP vs. DTT, p  = 6.53*10 −02 ; for PS+Clt: Kruskal-Wallis: p  = 3.45*10 −03 , H = 11.34., DF = 2, Dunn’s post-hoc: NT vs. DTT, p  = 1.98*10 −03 , NT vs. TCEP, p  = 3.05*10 −03 , TCEP vs. DTT, p  = 0.77. E Mean time course of fluorescence ratios (F340/F380) of HEK-TRPM3 cells recorded with calcium fluorimetric measurements upon application of indicated TRPM3 agonists/modulators PS (40 µM), and Clt (10 µM) in non-treated conditions and DTT-pretreatment (10 mM for ≤ 40 min). F Normalized fluorescence ratio increases during application of PS, Clt, and PS+Clt in non-treated or DTT-pretreated conditions. Statistical analyses between DTT-treated and non-treated groups were performed with a Student’s unpaired t-test ( N  = 3 dishes for all conditions; PS: p  = 4.20*10 −08 , t -value = 109.34, DF = 4; Clt: p  = 4.20*10 −08 , t -value = -109.34, DF = 4; PS+Clt: p  = 0.14*10 −02 , t -value = 24.96, DF = 2.06). Subsequently, similar experimental protocols were performed using Fura2-based calcium microfluorimetry to further confirm the redox-sensitivity of TRPM3. These microfluorimetric experiments revealed strong increases in intracellular calcium, [Ca 2+ ] I , in NT cells upon application of PS, whereas these responses were almost completely abolished in DTT-treated HEK-TRPM3 cells (Fig.  1E, F ). Oppositely, the sole application of Clt did not induce a calcium response in NT cells, while evoking robust responses in DTT-pretreated cells, which were partially attenuated by subsequent co-application of PS (25.1 ± 1.8%) (Fig.  1E, F ). In addition, non-transfected HEK cells pretreated with DTT did not exhibit any increase in [Ca 2+ ] I upon treatment with PS and Clt (Supplementary Fig.  1 ). Concentration-response curves of PS and Clt in NT and DTT-treated HEK-TRPM3 cells showed an EC 50 value for PS stimulation of 8.6 ± 1.2 µM in NT-treated cells, while no PS-induced response was observed even at higher doses up to 150 µM PS in DTT-treated cells. In contrast, HEK-TRPM3 cells showed a dose-dependent response to Clt after DTT-treatment (EC 50 value of 16.6 ± 1.9 µM), while the responses up to 10 µM Clt were completely absent in NT cells (Supplementary Fig.  2 ). To further investigate the PS-induced block of the Clt-evoked [Ca 2+ ] I influx in DTT-treated HEK-TRPM3 cells, a dose-dependent effect of PS was examined on the Clt-induced calcium influxes. Interestingly, the Clt-induced [Ca 2+ ] I increase could be blocked in a dose-dependent manner by PS (IC 50 value of 21.6 ± 1.5 µM) (Supplementary Fig.  3 ). Taken together, these results reveal that the reducing agents DTT and TCEP have a robust effect on the gating properties of TRPM3, leading to the complete elimination of responses to PS while enabling the activation of the channel by Clt. Next, we investigated whether the redox modulation of TRPM3 could be reversed by application of the oxidizing agent hydrogen peroxide (H 2 O 2 ) 50 . In whole-cell patch clamp experiments, DTT-pretreated cells were subjected twice to the application with PS, Clt, and the co-application of PS and Clt (PS+Clt), where the second set of stimuli was applied following a five min treatment with either 10 mM H 2 O 2 or vehicle. Notably, H 2 O 2 -treatment, but not vehicle treatment, fully reversed the effects of the DTT-pretreatment: cells recovered their responsivity towards PS stimulation, lost the direct response to Clt, and showed a potentiation of the PS-induced response in the presence of Clt (Fig.  2A–D ). Accordingly, Fura2-based calcium fluorimetric experiments showed that application of H 2 O 2 did not affect the responsivity profile of TRPM3 towards PS and Clt stimulation in NT cells (Fig.  2E , Supplementary Fig.  1 ), but restored responsiveness to PS in TCEP-pretreated cells (Fig.  2E, F ). Importantly, the fluorescence ratio increases recorded during PS, Clt and PS+Clt application in TCEP-pretreated HEK-TRPM3 cells after H 2 O 2 treatment were similar to the increases recorded in NT-cells after H 2 O 2 application (Fig.  1E ). Fig. 2 Sensitivity of heterologous TRPM3 to chemical activating stimuli is reversibly modulated by redox reagents. A Time course of whole-cell patch clamp recordings at ±80 mV performed in DTT-pretreated (10 mM) HEK-TRPM3 cells during application of TRPM3 stimuli (PS (40 µM), Clt (10 µM), PS+Clt) before and after H 2 O 2 (10 mM) treatment. B Corresponding IV -curves to the indicated colored dots in the time course before (left panel) and after H 2 O 2 treatment (right panel). C Same as in A) but with vehicle (Vhc) application. D PS-induced current density changes normalized to the sum of the current density changes upon sole application of PS or Clt. Statistical comparisons within the same group were performed with a Student Paired-sample t -test (H 2 O 2 : n  = 7 cells, p  = 9.70 × 10 −09 , t -value = -43.63, DF = 6; Vhc: n  = 4 cells, p  = 5.7*10 −03 , t -value = -7.12, DF = 3). For the comparison between groups a Student’s unpaired t -test was performed (before H 2 O 2 and Vhc: t -value = 3.37, DF = 6, after H 2 O 2 and Vhc: t -value = -1.88, DF = 10) E Mean time course of the calcium fluorescence ratio (F340/F380) of control or TCEP: Tris-(2-carboxyethyl)-phosphine pretreated (10 mM for 30 min) HEK-TRPM3 cells upon application of PS (40 µM) and Clt (10 µM) before and after treatment (5 min) with 10 mM H 2 O 2 (green color). F PS-induced Fura2-ratio amplitude increases normalized to the sum of the amplitude increases upon sole application of PS or Clt. Statistical comparisons within the same treatment group were performed with a Paired-sample Wilcoxon-signed-rank test (NT: p  = 0.55, W = 27163.5, Z = 0.6; TCEP: p = 1.09 × 10 −69 , W = 1718, Z = −17.65). A Time course of whole-cell patch clamp recordings at ±80 mV performed in DTT-pretreated (10 mM) HEK-TRPM3 cells during application of TRPM3 stimuli (PS (40 µM), Clt (10 µM), PS+Clt) before and after H 2 O 2 (10 mM) treatment. B Corresponding IV -curves to the indicated colored dots in the time course before (left panel) and after H 2 O 2 treatment (right panel). C Same as in A) but with vehicle (Vhc) application. D PS-induced current density changes normalized to the sum of the current density changes upon sole application of PS or Clt. Statistical comparisons within the same group were performed with a Student Paired-sample t -test (H 2 O 2 : n  = 7 cells, p  = 9.70 × 10 −09 , t -value = -43.63, DF = 6; Vhc: n  = 4 cells, p  = 5.7*10 −03 , t -value = -7.12, DF = 3). For the comparison between groups a Student’s unpaired t -test was performed (before H 2 O 2 and Vhc: t -value = 3.37, DF = 6, after H 2 O 2 and Vhc: t -value = -1.88, DF = 10) E Mean time course of the calcium fluorescence ratio (F340/F380) of control or TCEP: Tris-(2-carboxyethyl)-phosphine pretreated (10 mM for 30 min) HEK-TRPM3 cells upon application of PS (40 µM) and Clt (10 µM) before and after treatment (5 min) with 10 mM H 2 O 2 (green color). F PS-induced Fura2-ratio amplitude increases normalized to the sum of the amplitude increases upon sole application of PS or Clt. Statistical comparisons within the same treatment group were performed with a Paired-sample Wilcoxon-signed-rank test (NT: p  = 0.55, W = 27163.5, Z = 0.6; TCEP: p = 1.09 × 10 −69 , W = 1718, Z = −17.65). Taken together, these results indicate a bidirectional modulation of TRPM3 by reducing and oxidizing agents, which affects the responsivity to PS and Clt as well as the shape of the current-voltage relationship. Notably, similar redox modulation of TRPM3 was also observed in the human TRPM3 ortholog (Supplementary Fig.  4 and Supplementary Table  1 ). Next, the redox sensitivity of TRPM3 was investigated in primary cells endogenously expressing TRPM3. Our previous work has established functional TRPM3 expression in a subset of sensory neurons of the mouse dorsal root ganglia (DRG) and trigeminal ganglia (TG), where it plays a role in the detection of noxious heat and in the development of pathological pain 1 . To address redox modulation of endogenous TRPM3 in sensory neurons, primary DRG neurons derived from Trpa1 -/- _ Trpv1 -/- double knockout (DKO) mice were isolated, preventing potential interference by TRPV1- and TRPA1-mediated responses to Clt 51 or redox reagents 52 – 57 . Using Fura2-based calcium microfluorimetric experiments, NT- and DTT-pretreated neurons were investigated for their sensitivity towards PS and Clt stimulation (Fig.  3 and Supplementary Fig.  5 ). In line with earlier work, a large subset of NT DKO neurons responded to PS stimulation (67% of DRG neurons, n = 243 out of 363), whereas responses to application of Clt were very sparse (2%, n  = 6 out of 363; Fig.  3A, E ). In contrast, PS responses were almost completely abolished in DTT-pretreated neurons (5% of DRG neurons, n  = 12 out of 229), whereas a large subset of these neurons showed a robust calcium response to Clt (75% of DRG neurons, n  = 172 out of 229) (Fig.  3B, E ). To ascertain the TRPM3 specificity of these responses, similar experiments were performed using primary neurons isolated from Trpa1 -/- , Trpv1 -/- and Trpm3 -/- triple knockout (TKO) mice 29 . Indeed, TKO neurons did not respond to PS or Clt stimulation, neither in NT- nor in DTT-treated conditions (Fig.  3C–E ). Fig. 3 The activity of endogenous TRPM3 is reversibly modulated by redox reagents. A Representative traces of the calcium fluorescence ratios (F340/F380) of dorsal root ganglia (DRG) neurons isolated from Trpa1_Trpv1 double-knockout (DKO) mice upon application of the TRPM3 stimuli PS (40 µM), Clt (10 µM) and co-application of PS+Clt or high concentrations of potassium (High K + (50 mM)). B Similar as in ( A ) but for DRG neurons pretreated (≥40 min) with the reductive agent DTT (10 mM). C Similar as in ( A ) but for non-treated (NT) DRG neurons derived from Trpa1_Trpv1_Trpm3 triple-knockout (TKO) animals. D Similar as in ( C ) but for DRG neurons pretreated with DTT (10 mM). E Percentage of DRG neurons that responded to the indicated stimuli ( X-axis ) in DKO and TKO animals in non-treated conditions or pretreated with the reductive agent DTT (as shown in ( A )–( D )). Statistical comparisons were done with a Chi-square test followed by Fisher’s exact tests to compare individual groups (DKO NT: n  = 363, DKO DTT: n = 229, TKO NT: n  = 512, TKO DTT: n  = 422, Chi-Square test for all groups (PS, Clt and PS+Clt): p  < 0.1*10 −05 , DF = 3). F Normalized fluorescence ratio increases for DRG neurons to indicated stimuli ( X-axis ) in control conditions or pretreated with the reductive agent DTT. Statistical testing was performed with a Student’s unpaired t -test ( N  = 3 dishes per condition; PS: p  = 0.009, t -value = 10.05, DF = 2.03; Clt: p  = 0.009, t -value = -10.05, DF = 2.03; PS+Clt: p = 0.036, t-value = 5.10, DF = 2). G Mean time courses of the calcium fluorescence ratios (F340/F380) of DRG (yellow) neurons isolated from DKO animals that were treated with different TRPM3 stimuli (PS (40 µM) and Clt (10 µM)) or high K + (50 mM). Black: cells were untreated before the recordings. Yellow: cells were pretreated with 10 mM DTT. In both conditions cells were subsequently treated with the oxidative agent H 2 O 2 before application of TRPM3 agonists/modulators. H Left panel: Percentage of DRG neurons that responded to the indicated stimuli ( X-axis ) in non-treated and DTT-pretreated DKO animals after H 2 O 2 treatment. Right panel: Fluorescence ratio increases of DRG neurons to the indicated stimuli ( X-axis ) in non-treated conditions or pretreated with the reductive agent DTT. Statistical testing was performed using a Student’s unpaired t-test for each stimulus ( N  = 3 dishes per condition; PS: p  = 0.28, t -value = −1.26, DF = 4; Clt: p  = 0.28, t-value = 1.26, DF = 4; PS+Clt: p  = 0.84, t -value = 0.22, DF = 4). A Representative traces of the calcium fluorescence ratios (F340/F380) of dorsal root ganglia (DRG) neurons isolated from Trpa1_Trpv1 double-knockout (DKO) mice upon application of the TRPM3 stimuli PS (40 µM), Clt (10 µM) and co-application of PS+Clt or high concentrations of potassium (High K + (50 mM)). B Similar as in ( A ) but for DRG neurons pretreated (≥40 min) with the reductive agent DTT (10 mM). C Similar as in ( A ) but for non-treated (NT) DRG neurons derived from Trpa1_Trpv1_Trpm3 triple-knockout (TKO) animals. D Similar as in ( C ) but for DRG neurons pretreated with DTT (10 mM). E Percentage of DRG neurons that responded to the indicated stimuli ( X-axis ) in DKO and TKO animals in non-treated conditions or pretreated with the reductive agent DTT (as shown in ( A )–( D )). Statistical comparisons were done with a Chi-square test followed by Fisher’s exact tests to compare individual groups (DKO NT: n  = 363, DKO DTT: n = 229, TKO NT: n  = 512, TKO DTT: n  = 422, Chi-Square test for all groups (PS, Clt and PS+Clt): p  < 0.1*10 −05 , DF = 3). F Normalized fluorescence ratio increases for DRG neurons to indicated stimuli ( X-axis ) in control conditions or pretreated with the reductive agent DTT. Statistical testing was performed with a Student’s unpaired t -test ( N  = 3 dishes per condition; PS: p  = 0.009, t -value = 10.05, DF = 2.03; Clt: p  = 0.009, t -value = -10.05, DF = 2.03; PS+Clt: p = 0.036, t-value = 5.10, DF = 2). G Mean time courses of the calcium fluorescence ratios (F340/F380) of DRG (yellow) neurons isolated from DKO animals that were treated with different TRPM3 stimuli (PS (40 µM) and Clt (10 µM)) or high K + (50 mM). Black: cells were untreated before the recordings. Yellow: cells were pretreated with 10 mM DTT. In both conditions cells were subsequently treated with the oxidative agent H 2 O 2 before application of TRPM3 agonists/modulators. H Left panel: Percentage of DRG neurons that responded to the indicated stimuli ( X-axis ) in non-treated and DTT-pretreated DKO animals after H 2 O 2 treatment. Right panel: Fluorescence ratio increases of DRG neurons to the indicated stimuli ( X-axis ) in non-treated conditions or pretreated with the reductive agent DTT. Statistical testing was performed using a Student’s unpaired t-test for each stimulus ( N  = 3 dishes per condition; PS: p  = 0.28, t -value = −1.26, DF = 4; Clt: p  = 0.28, t-value = 1.26, DF = 4; PS+Clt: p  = 0.84, t -value = 0.22, DF = 4). To investigate the bidirectional redox modulation of TRPM3 in sensory neurons, primary DRG and TG neurons were first pretreated with DTT, followed by H 2 O 2 treatment. The responsivity towards PS stimulation was recovered by treatment with H 2 O 2 (Fig.  3G , Supplementary Fig.  5 ). Moreover, the percentage of DRG responsive neurons to PS and Clt was in a similar range for DTT-treated and NT neurons. Furthermore, similar amplitudes to these stimuli were detected after H 2 O 2 treatment in DTT-treated and NT cells (Fig.  3H , Supplementary Fig.  5 ). Altogether, these data confirm the bidirectional redox sensitivity of TRPM3 in primary peripheral neurons. The sensitivity of endogenously expressed TRPM3 to DTT modulation was further investigated in mouse pancreatic islet cells. Pancreatic islet cells derived from DKO animals were subjected to whole-cell patch clamp recordings to compare the responsivity of NT and DTT-treated cells towards PS, Clt and PS+Clt (Supplementary Fig.  6 ). We found that PS activates an outwardly rectifying current in NT pancreatic islet cells from DKO mice, while Clt does not cause any direct current activation but potentiates the response to PS. Following DTT treatment, PS-evoked current responses were suppressed, whereas Clt causes direct activation of currents with a linear current-voltage relation. Importantly, NT or DTT-treated pancreatic islet cells derived from TKO mice did not show any responses to PS, Clt and PS+Clt, further confirming the TRPM3 specificity of the responses (Supplementary Fig.  6 ). To further investigate the effects of redox modulation, we used spheroids formed from the mouse MIN6 β-cell line. We recently demonstrated that glucose induced spheroid-wide calcium oscillations leading to insulin release, and that PS causes an increase in the frequency of glucose-induced oscillations in a TRPM3-dependent manner 47 . Importantly, when MIN6 spheroids were pretreated with TCEP, the effect of PS on calcium oscillations induced by 8 mM glucose, quantified as the frequency of oscillations and the area under the curve, was largely eliminated (Fig.  4A, B ). Overall, these data demonstrate redox modulation of TRPM3 in pancreatic beta cells, modulating TRPM3-dependent effects on glucose responses. Note that the TCEP treatment prevented the accurate detection of released insulin, as reduction of disulfide bridges in insulin directly affect its detection using ELISA. Fig. 4 TCEP is able to reduce the Ca 2+ -oscillatory behavior of MIN6 β-cell spheroids. A Example traces of normalized Ca²⁺ responses (F/F0) of individual MIN6 β-cell spheroids exposed to extracellular glucose (8 mM) with and without 50 µM pregnenolone sulfate (PS) and with or without pre-treatment with 5 mM TCEP. A 5 min baseline period was measured in an extracellular solution containing 3 mM glucose. B Analysis of the normalized area under the peaks for the different conditions as presented in ( A ). Each dot represents one spheroid. Statistical analysis was performed with a Kruskal-Wallis rank sum test with a Dunn’s posthoc test. Relevant p values are shown on the graph with stars (# control vs TCEP: p  = 0.0027, z-value = 2.99, DF = 3; * control vs PS: p  = 1.26*10–24, z -value = −10.35, DF = 3; PS vs PS + TCEP: p  = 3.4 *10−8, z-value = 5.59, DF = 3). C Analysis of the amount of peaks per hour (frequency) for the different conditions as presented in ( A ). Each dot represents one spheroid. Statistical analysis was performed with a Kruskal-Wallis rank sum test with a Dunn’s posthoc test. Relevant p values are shown on the graph with stars (control vs TCEP: # p  = 2.05*10−4, DF = 3, z -value = 3.76; control vs PS: * p  = 1.10*10-20, z-value = −9.44, DF = 3; PS vs PS + TCEP: p  = 5.47 *10−11, z-value = 6.66, DF = 3). A Example traces of normalized Ca²⁺ responses (F/F0) of individual MIN6 β-cell spheroids exposed to extracellular glucose (8 mM) with and without 50 µM pregnenolone sulfate (PS) and with or without pre-treatment with 5 mM TCEP. A 5 min baseline period was measured in an extracellular solution containing 3 mM glucose. B Analysis of the normalized area under the peaks for the different conditions as presented in ( A ). Each dot represents one spheroid. Statistical analysis was performed with a Kruskal-Wallis rank sum test with a Dunn’s posthoc test. Relevant p values are shown on the graph with stars (# control vs TCEP: p  = 0.0027, z-value = 2.99, DF = 3; * control vs PS: p  = 1.26*10–24, z -value = −10.35, DF = 3; PS vs PS + TCEP: p  = 3.4 *10−8, z-value = 5.59, DF = 3). C Analysis of the amount of peaks per hour (frequency) for the different conditions as presented in ( A ). Each dot represents one spheroid. Statistical analysis was performed with a Kruskal-Wallis rank sum test with a Dunn’s posthoc test. Relevant p values are shown on the graph with stars (control vs TCEP: # p  = 2.05*10−4, DF = 3, z -value = 3.76; control vs PS: * p  = 1.10*10-20, z-value = −9.44, DF = 3; PS vs PS + TCEP: p  = 5.47 *10−11, z-value = 6.66, DF = 3). Next, we tested the effects of redox treatment on the TRPM3 responses to other activating stimuli, such as temperature and the chemical ligands CIM0216 and nifedipine. To determine the effect of redox-treatment on the temperature-dependent Ca 2+ -responses, non-treated and DTT-treated HEK-TRPM3 cells were tested in a 96-well plate-based assay in the presence and absence of PS. In line with previous findings, we observed that PS at 10 µM shifted the thermal response profile of HEK-TRPM3 to lower temperatures in NT cells 1 . Interestingly, the Ca 2+ -responses to increasing temperature were completely absent in DTT-treated HEK-TRPM3 cells, even in the presence of PS, indicating that the heat sensitivity of TRPM3 depends on the redox status (Fig.  5A ). In addition, DTT-treated HEK-TRPM3 cells showed a significant shift in EC 50 values for CIM0216 (EC 50  ~ 22.4 µM) towards higher doses compared to NT cells (EC 50  ~ 2.0 µM) (Fig.  5B ). Additional patch clamp experiments showed that in NT HEK-TRPM3 cells, application of CIM0216 (1 µM) induced a robust increase in current amplitudes with a typical current ( I) – voltage (V) relationship showing double rectification, as described earlier 2 . Following DTT-treatment, however, the amplitude of CIM0216-evoked currents was significantly reduced compared to NT conditions (Fig.  5C–E ). Moreover, the CIM0216-evoked I-V relationship showed reduced inward rectification, which could be quantified as a significant increase in the ratio of the amplitude of outward over inward currents (ΔI +80mV / ΔI -80mV ) (Fig.  5F ). In addition, DTT-treatment also significantly reduced TRPM3-mediated responses evoked by nifedipine, when compared to NT-conditions (Supplementary Fig.  8 ). Fig. 5 Heat and CIM0216 sensitivity of TRPM3 is strongly affected by redox reagents. A Fluorescent ratio increases of non-transfected HEK293T cells (Empty HEK; green) or HEK293T cells stably expressing TRPM3α2 cells upon increasing temperatures. Cells were either non-treated (NT; black) with (open square) or without (filled square) the addition of 10 µM PS, or DTT-pretreated (for ≥ 40 min with 10 mM; yellow) with (open diamond) or without (filled diamond) the addition of 10 µM PS. Fluo4 signals were normalized to the baseline values at 16°C (ΔF/F 16°C ). B CIM0216 dose-dependency of TRPM3 in non-treated (black) and DTT-pretreated (yellow) condition. Data is represented as mean ± SEM. HEK-TRPM3 cells were tested in a 96-well plate-based assay in the presence and absence of DTT and stimulated with different doses of CIM0216. C Time course of whole-cell currents at ±80 mV recorded in non-treated (NT) HEK293T cells stably expressing TRPM3α2 upon application of the TRPM3 activator CIM0216 (1 µM). D Same as in A) but for HEK293T cells stably expressing TRPM3α2 that were pre-treated ≥40 min with the reductive agent DTT (10 mM). E Current density changes for non-treated and DTT-pre-treated HEK293T TRPM3α2 cells upon application of CIM0216 as shown in ( A ) and ( B ). Statistical comparisons performed with Student’s two-sample t -test (at +80 mV: p  = 0.012, t -value = 4.15, DF = 4.40; at −80 mV: p  = 0.005, t -value = −5.34, DF = 4.35). F Ratio of currents at +80 mV versus currents at −80 mV for non-treated and DTT-pre-treated HEK293T TRPM3α2 cells upon application of CIM0216. Statistical comparisons performed with a Mann-Whitney test ( p  = 0.014, U = 1, Z = −2.46). A Fluorescent ratio increases of non-transfected HEK293T cells (Empty HEK; green) or HEK293T cells stably expressing TRPM3α2 cells upon increasing temperatures. Cells were either non-treated (NT; black) with (open square) or without (filled square) the addition of 10 µM PS, or DTT-pretreated (for ≥ 40 min with 10 mM; yellow) with (open diamond) or without (filled diamond) the addition of 10 µM PS. Fluo4 signals were normalized to the baseline values at 16°C (ΔF/F 16°C ). B CIM0216 dose-dependency of TRPM3 in non-treated (black) and DTT-pretreated (yellow) condition. Data is represented as mean ± SEM. HEK-TRPM3 cells were tested in a 96-well plate-based assay in the presence and absence of DTT and stimulated with different doses of CIM0216. C Time course of whole-cell currents at ±80 mV recorded in non-treated (NT) HEK293T cells stably expressing TRPM3α2 upon application of the TRPM3 activator CIM0216 (1 µM). D Same as in A) but for HEK293T cells stably expressing TRPM3α2 that were pre-treated ≥40 min with the reductive agent DTT (10 mM). E Current density changes for non-treated and DTT-pre-treated HEK293T TRPM3α2 cells upon application of CIM0216 as shown in ( A ) and ( B ). Statistical comparisons performed with Student’s two-sample t -test (at +80 mV: p  = 0.012, t -value = 4.15, DF = 4.40; at −80 mV: p  = 0.005, t -value = −5.34, DF = 4.35). F Ratio of currents at +80 mV versus currents at −80 mV for non-treated and DTT-pre-treated HEK293T TRPM3α2 cells upon application of CIM0216. Statistical comparisons performed with a Mann-Whitney test ( p  = 0.014, U = 1, Z = −2.46). Next, we investigated the effect of the different redox conditions on the potency of the TRPM3 inhibitors isosakuranetin and primidone, as well as on the TRPM3 inhibition by activation of GPCRs 19 – 23 . In particular, the inhibitory effect of TRPM3 antagonists was studied under conditions of maximal current activation, namely upon combined stimulation by PS+Clt for the NT condition and by Clt for the DTT-treated condition (Fig.  6A–C ) 19 . Isosakuranetin (5 µM) fully blocked PS+Clt-induced currents in NT HEK-TRPM3 cells (Fig.  6A, C ), while only a partial block (±40%) of the Clt-induced currents was observed in DTT-treated cells (Fig.  6B, C ). Primidone 20 (25 µM) induced a full block of PS+Clt-induced currents in NT conditions, but did not show any inhibitory effect on Clt-induced currents in DTT-treated cells (Fig.  6D–F ). Finally, the inhibitory effect of the µ-opioid receptor agonist [D-Ala 2 , N-MePhe 4 , Gly-ol]-enkephalin (DAMGO) was studied in NT and DTT-treated cells co-expressing the µ-opioid receptor. Application of DAMGO (1 µM) induced 80% block of the PS+Clt-induced current in NT cells but was without detectable effect on the Clt-induced currents in DTT-treated cells (Fig.  6G–I ). In conclusion, these results indicate that DTT treatment markedly affects the efficacy of different types of TRPM3 modulations. Fig. 6 Sensitivity of TRPM3 to antagonists is altered by redox reagents. A Time course of whole-cell currents at ±80 mV recorded in HEK-TRPM3 cells upon co-application of Clt (10 µM) and PS (40 µM) or single application of Clt (10 µM) in presence and absence of the TRPM3 inhibitor isosakuranetin (Iso; 5 µM). B Same as in A) but for HEK-TRPM3 cells pretreated (≥40 min) with DTT (10 mM). C Percentage of inhibition induced by isosakuranetin in NT ( n  = 7) and DTT-pretreated ( n  = 6) conditions recorded in ( A ) and ( B ) (at +80 mV: p  = 9.17*10 -05 , t -value = 11.31, DF = 5.02; at -80 mV: p  = 1.01*10 -04 , t -value = 11.08, DF = 5.02). D Similar as in ( A ) but for the TRPM3 antagonist primidone (Prim; 25 µM). E Same as in ( D ) but for cells pretreated with DTT (10 mM). F Percentage of inhibition induced by primidone in NT ( n  = 5) and DTT-pretreated ( n  = 6) conditions recorded in ( D ) and ( E ) (at +80 mV: p  = 1.72*10 − 11 , t -value = 40.43, DF = 9; at -80 mV: p  = 9.61*10 − 10 , t -value = 25.77, DF = 9). G Time course of whole-cell currents at ±80 mV recorded in HEK293T cells transiently co-expressing TRPM3α2 and µ-opioid receptor upon co-application of Clt (10 µM) and PS (40 µM) or single Clt (10 µM) stimulation in presence and absence of the µ-opiod receptor agonist DAMGO: [D-Ala 2 , N -MePhe 4 , Gly-ol]-enkephalin (1 µM). H Same as in ( G ) but for cells pretreated with the reductive agent DTT (10 mM). I Percentage of inhibition induced by DAMGO in NT ( n  = 7) and DTT-pretreated ( n  = 6) conditions recorded in ( G ) and ( H ) (at +80 mV: p  = 3.71*10 − 06 , t -value = 12.61, DF = 7.18; at −80 mV: p  = 2.11*10 -07 , t-value = 17.05, DF = 7.18). Statistical comparisons in ( C ), ( F ) and ( I ) were performed with Student’s unpaired t-tests. A Time course of whole-cell currents at ±80 mV recorded in HEK-TRPM3 cells upon co-application of Clt (10 µM) and PS (40 µM) or single application of Clt (10 µM) in presence and absence of the TRPM3 inhibitor isosakuranetin (Iso; 5 µM). B Same as in A) but for HEK-TRPM3 cells pretreated (≥40 min) with DTT (10 mM). C Percentage of inhibition induced by isosakuranetin in NT ( n  = 7) and DTT-pretreated ( n  = 6) conditions recorded in ( A ) and ( B ) (at +80 mV: p  = 9.17*10 -05 , t -value = 11.31, DF = 5.02; at -80 mV: p  = 1.01*10 -04 , t -value = 11.08, DF = 5.02). D Similar as in ( A ) but for the TRPM3 antagonist primidone (Prim; 25 µM). E Same as in ( D ) but for cells pretreated with DTT (10 mM). F Percentage of inhibition induced by primidone in NT ( n  = 5) and DTT-pretreated ( n  = 6) conditions recorded in ( D ) and ( E ) (at +80 mV: p  = 1.72*10 − 11 , t -value = 40.43, DF = 9; at -80 mV: p  = 9.61*10 − 10 , t -value = 25.77, DF = 9). G Time course of whole-cell currents at ±80 mV recorded in HEK293T cells transiently co-expressing TRPM3α2 and µ-opioid receptor upon co-application of Clt (10 µM) and PS (40 µM) or single Clt (10 µM) stimulation in presence and absence of the µ-opiod receptor agonist DAMGO: [D-Ala 2 , N -MePhe 4 , Gly-ol]-enkephalin (1 µM). H Same as in ( G ) but for cells pretreated with the reductive agent DTT (10 mM). I Percentage of inhibition induced by DAMGO in NT ( n  = 7) and DTT-pretreated ( n  = 6) conditions recorded in ( G ) and ( H ) (at +80 mV: p  = 3.71*10 − 06 , t -value = 12.61, DF = 7.18; at −80 mV: p  = 2.11*10 -07 , t-value = 17.05, DF = 7.18). Statistical comparisons in ( C ), ( F ) and ( I ) were performed with Student’s unpaired t-tests. TRPM3 is a cation permeable channel that shows a substantial permeability for Zinc ions (Zn 2+ ) 58 . To investigate whether ROS modulation affects the Zn 2+ permeability of TRPM3, NT and DTT-treated HEK-TRPM3 cells were loaded with the fluorescent Fluo-Zin dye 4 . NT HEK-TRPM3 cells showed a robust increase in Fluo-Zin fluorescence after application of a bath solution containing 1 mM Zn 2+ (Supplementary Fig.  10 ), indicating significant Zn 2+ permeability under basal conditions, as described earlier 58 . The increase in intracellular Zn 2+ concentration was completely prevented in the presence of the TRPM3 inhibitor isosakuranetin (Supplementary Fig.  9 ). Importantly, DTT-treated TRPM3 cells showed only a minor increase in Fluo-Zin fluorescence, comparable to the fluorescence increase in non-transfected HEK cells. Taken together, these data show that redox modulation has a significant effect on the basal Zn 2+ permeability of TRPM3 (Supplementary Figs.  1 and 9 ). In earlier publications, we showed the existence of a non-canonical pore within the voltage-sensing like domain (VSLD) of TRPM3 8 , 59 . The opening of the alternative pore was induced by the combined co-application of PS and Clt and was characterized by the appearance of a large inwardly rectifying monovalent cation current. To investigate whether the large, linear currents activated by Clt under reducing conditions involve ion flux via the non-canonical pore, we investigate the behavior of a previously described TRPM3 mutant (D988R). In this mutant, substitution of a negatively charged aspartate in the VSLD by a positively charged arginine eliminates ion flux through the non-canonical pore, while preserving currents through the central pore 59 . In NT conditions, stimulation with PS elicited strongly outwardly rectifying currents, whereas sole Clt stimulation was without effect; co-application of PS+Clt induced outwardly rectifying currents, lacking the inward component corresponding to ion flux through the non-canonical pore seen in WT-TRPM3 (Supplementary Fig.  9 ), in line with our earlier work 59 . Importantly, following DTT pretreatment of D988R-expressing HEK293 cells, PS-evoked currents were largely eliminated, but Clt-stimulation resulted in robust linear currents, which were blocked by PS, similar to the behavior of DTT-treated WT TRPM3 (Supplementary Fig.  9 and Supplementary Table  1 ). Overall, these data indicate that the Clt-evoked current following DTT treatment occurs via the central pore of TRPM3 rather than via the alternative pathway in the VSLD. Important differences in the pharmacological profile have been previously reported between the long TRPM3 pore variant (TRPM3α1) and short pore variants (represented by TRPM3α2) 3 , 13 , 25 , 26 . As TRPM3α2 represents the standard isoform used for the here presented work, it is referred to as TRPM3 throughout the manuscript. Notably, the biophysical and pharmacological properties of DTT- or TCEP-treated TRPM3 channels described here are highly reminiscent of the profile reported for the long pore-loop isoform TRPM3α1 3 . We therefore used Fura2-based microfluorimetric and whole-cell patch clamp experiments to test the effects of DTT treatment on HEK293T cells overexpressing the long-pore loop variant TRPM3α1. Interestingly, no significant difference in the response profile was detected between NT and DTT-pretreated HEK cells overexpressing TRPM3α1 after stimulation by PS or Clt (Fig.  7A–E and Supplementary Table  1 ). Indeed, both NT and DTT-pretreated cells expressing TRPM3α1 failed to respond to PS but showed robust calcium and current responses to Clt application (Fig.  7A–E and Supplementary Table  1 ). Moreover, no modification of the response profile was observed after H 2 O 2 treatment of TRPM3α1 expressing cells (Fig.  7F ). Taken together, these results indicate that the long pore TRPM3 isoform (TRPM3α1), contrary to the short pore isoform (TRPM3α2), is insensitive to redox-modulation. Interestingly, the redox-modulatory properties seem to be preserved in all short pore-loop TRPM3 isoforms, as Fura2-based calcium imaging showed that other short pore-loop isoforms (TRPM3α3-α6) were also modulated by DTT-pretreatment, resulting in reduced PS responses and enhanced responses to Clt stimulation (Supplementary Fig.  10 ). Fig. 7 TRPM3 pore-loop length is important for its redox sensitivity. A Mean time courses of the calcium fluorescence ratios (F340/F380) of HEK293T cells transiently expressing TRPM3α1 in non-treated (NT) or DTT-pretreated conditions upon application of PS (40 µM), Clt (10 µM), and co-application of PS+Clt. B Normalized fluorescence ratio increases during application of the different TRPM3 stimuli as shown in ( A ). Statistical analyses between DTT-treated and NT groups performed with a Student’s unpaired t -test or a Mann-Whitney test ( N  = 3 dishes; PS: p  = 1, U = 4, Z = 0; Clt: p  = 1, U = 5, Z = 0; PS+Clt: p  = 0.38, t -value = -0.98, DF = 4). C Time course of whole-cell currents at ±80 mV recorded in NT HEK293T cells transiently expressing TRPM3α1 upon single and co-application of PS (40 µM) and Clt (10 µM). D Same as in ( C ) but for TRPM3α1 expressing HEK293T cells that were pretreated with DTT (10 mM). E Normalized current density changes in cells during application of the indicated stimuli ( X-axis ) as shown in ( C ) and (D ) in NT or DTT-pretreated conditions. Statistical comparisons were performed with Student’s unpaired t-test and Mann-Whitney test ( n  = 5 cells; PS: p  = 0.53, U = 9, Z = -0.63; Clt: p  = 0.53, U = 16, Z = −0.63; PS+Clt: p  = 0.014, t -value = −3.13, DF = 8). F Time course of whole-cell currents at ±80 mV recorded in DTT-pretreated (≥40 min) cells upon single and co-application of PS (40 µM) and Clt (10 µM) before and after H 2 O 2 application (green bar). A Mean time courses of the calcium fluorescence ratios (F340/F380) of HEK293T cells transiently expressing TRPM3α1 in non-treated (NT) or DTT-pretreated conditions upon application of PS (40 µM), Clt (10 µM), and co-application of PS+Clt. B Normalized fluorescence ratio increases during application of the different TRPM3 stimuli as shown in ( A ). Statistical analyses between DTT-treated and NT groups performed with a Student’s unpaired t -test or a Mann-Whitney test ( N  = 3 dishes; PS: p  = 1, U = 4, Z = 0; Clt: p  = 1, U = 5, Z = 0; PS+Clt: p  = 0.38, t -value = -0.98, DF = 4). C Time course of whole-cell currents at ±80 mV recorded in NT HEK293T cells transiently expressing TRPM3α1 upon single and co-application of PS (40 µM) and Clt (10 µM). D Same as in ( C ) but for TRPM3α1 expressing HEK293T cells that were pretreated with DTT (10 mM). E Normalized current density changes in cells during application of the indicated stimuli ( X-axis ) as shown in ( C ) and (D ) in NT or DTT-pretreated conditions. Statistical comparisons were performed with Student’s unpaired t-test and Mann-Whitney test ( n  = 5 cells; PS: p  = 0.53, U = 9, Z = -0.63; Clt: p  = 0.53, U = 16, Z = −0.63; PS+Clt: p  = 0.014, t -value = −3.13, DF = 8). F Time course of whole-cell currents at ±80 mV recorded in DTT-pretreated (≥40 min) cells upon single and co-application of PS (40 µM) and Clt (10 µM) before and after H 2 O 2 application (green bar). The only difference between the redox-insensitive long pore loop TRPM3α1 and the redox-sensitive short pore loop isoform (TRPM3α2) resides in a stretch of 13 amino acid residues in the part of the extracellular loop between TM5 and TM6 that follows the selectivity filter (Fig.  8A ) 3 , 25 . The reducing agents DTT and TCEP are able to break cysteine-bridges in proteins, which may induce conformational changes in the channel structure that could affect the channel gating. As TCEP is a membrane-impermeable reducing agent, which was applied to the extracellular side of the cells, we focused on potential candidate cysteine (Cys) residues in the extracellular loops of TRPM3. Interestingly, we identified two cysteine residues at positions 1079 and 1096 in the region between the selectivity filter and TM6, immediately C terminal to the pore loop variation between TRPM3α1 and TRPM3α2 (Fig.  8A ). Notably, a recent cryoEM structure of TRPM3 of a short-pore TRPM3 isoform shows that these cysteines form a cystein bridge, which is absent in an earlier structure obtained under reducing conditions 5 , 6 (Fig.  8B ). Notably, a homology model of the long-pore loop TRPM3α1 isoform does not show this disulfide bridge, as the additional pore loop residues cause the two cysteines to be positioned too far away from each other to engage (Fig.  8B ). Mutagenesis in TRPM3α2 of these cysteines to alanines, individually or combined (mutants C1079 A, C1096A and C1079A_C1096A), resulted in mutant channels that showed no response upon stimulation with PS in both NT and DTT-treated conditions, whereas Clt consistently evoked robust calcium influxes (Fig.  9A–D ) and current responses (Fig.  9E–H and Supplementary Table  1 ). Moreover, addition of PS in the presence of Clt attenuated the response to Clt in both redox conditions. In conclusion, these data demonstrate that substitution of the cysteine residues to alanines in the pore region of TRPM3 results in mutant channels that are no longer sensitive to the redox status of the cell. Fig. 8 Pore loop length determines TRPM3 gating properties based on the sterical (in)ability of two pore loop cysteines to form intra-subunit cysteine bridges. A Amino acid sequence alignment of the pore-loop regions of the TRPM3 isoforms α1 and α2. 12-amino acid insertion in the long pore-loop isoform TRPM3α1 is indicated in orange. The amino acid difference at position 13, following that insertion, is indicated in blue. Two cysteine residues situated in the pore loop of TRPM3α1 and TRPM3α2 are highlighted in magenta. The numbers above indicate the amino acid positions within TRPM3. B TRPM3 cryo-EM structure and structural models illustrating the location of the two cysteines and their proximity within the short but not the long pore loop TRPM3 isoform. Upper left panel: Top view of the TRPM3 cryo-EM structure. Upper right panel: Side view of the TRPM3 cryo-EM structure within the membrane. The cysteine residues are indicated in magenta. The four-channel subunits are marked in different colors for clarification (blue, green, yellow, and cyan). PDB structural number: 9B29. Lower left panel: An AlphaFold structural model from TRPM3α2 is represented, indicating the proximity of the two pore loop cysteines that engage in a cysteine bridge. Lower middle panel: An AlphaFold model of TRPM3α1 with insertion of 12 additional amino-acid residues specific to TRPM3α1 indicates the sterical inability of the two pore loop cysteines to form a cysteine bridge in the long pore loop TRPM3α1 variant. The pore loop insertion (in orange) is increasing the distance between the two cysteines to a level that disallows cysteine bridge formation. Lower right panel: Mutation of one of the 12 amino acids (K1066) to cysteine, which restores the potential to form a cysteine bridge. Fig. 9 Redox sensitivity of TRPM3 determined by two cysteine residues located in the extracellular pore-loop. A Mean time courses of the calcium fluorescence ratios (F340/F380) of HEK293T cells transiently expressing TRPM3α2 C1079A mutant in NT or DTT-pretreated (10 mM) conditions upon single and co-application of PS (40 µM) and Clt (10 µM). B Same as in ( A ) but for HEK293T cells transiently transfected with TRPM3α2 C1096A mutant. C Same as in ( A ) and ( B ) but for HEK293T cells transiently transfected with TRPM3α2 C1079A + C1096A double-mutant. D Fluorescence ratio increases for the different TRPM3 mutant channels during application of PS (40 µM) and Clt (10 µM) in non-treated (NT) and DTT-pretreated conditions. Statistical analyses between DTT-treated and NT groups performed with a Student unpaired t -test or a Mann-Whitney test ( N  = 3 dishes for each condition, C1079A: PS: p  = 0.66, U  = 6, Z  = 0.44; Clt: p  = 0.66, U = 3, Z  = -0.44; C1096A: PS: p  = 0.20, t -value = -1.87, DF = 2.01, Clt: p  = 0.20, t -value = 1.86, DF = 2.01; C1079A + C1096A: PS: p  = 0.22, t -value = -1.44, DF = 4; Clt: p  = 0.22, t -value = 1.46, DF = 4). E Time course of patch clamp whole-cell currents at ±80 mV recorded in HEK293T cells transiently expressing TRPM3α2 C1079A mutant upon single and co-application of PS (40 µM) and Clt (10 µM). F Same as in ( E ) but for HEK293T cells transiently transfected with TRPM3α2 C1096A mutant. G Same as in ( E ) and ( F ) but for HEK293T cells transiently transfected with TRPM3α2 C1079A + C1096A double-mutant. H Normalized current changes in NT and DTT-pretreated HEK293T cells transiently expressing TRPM3α2 (TRPM3), C1079A, C1096A, or the C1079A + C1096A double-mutant during application of Clt (10 µM). Statistical comparisons of each group with the control group were performed with Student’s unpaired t -test and Mann-Whitney test (TRPM3 NT: n  = 5; TRPM3 DTT: n  = 12, PS: p  = 1.87*10 -03 , U  = 60, Z  = 3.11, Clt: p  = 1.87*10 − 03 , U  = 0, Z  = −3.11; C1079A NT as well as C1096A NT: n  = 6, PS: p  = 8.11*10 −03 , U = 30, Z  = 2.65, Clt: p  = 8.11*10 −03 , U = 0, Z = −2.65; C1079A + C1096A NT: n  = 5, PS: p  = 1.22*10 −02 , U  = 25, Z = 2.51, Clt: p  = 1.22*10− 02 , U  = 0, Z  = −2.51). A Amino acid sequence alignment of the pore-loop regions of the TRPM3 isoforms α1 and α2. 12-amino acid insertion in the long pore-loop isoform TRPM3α1 is indicated in orange. The amino acid difference at position 13, following that insertion, is indicated in blue. Two cysteine residues situated in the pore loop of TRPM3α1 and TRPM3α2 are highlighted in magenta. The numbers above indicate the amino acid positions within TRPM3. B TRPM3 cryo-EM structure and structural models illustrating the location of the two cysteines and their proximity within the short but not the long pore loop TRPM3 isoform. Upper left panel: Top view of the TRPM3 cryo-EM structure. Upper right panel: Side view of the TRPM3 cryo-EM structure within the membrane. The cysteine residues are indicated in magenta. The four-channel subunits are marked in different colors for clarification (blue, green, yellow, and cyan). PDB structural number: 9B29. Lower left panel: An AlphaFold structural model from TRPM3α2 is represented, indicating the proximity of the two pore loop cysteines that engage in a cysteine bridge. Lower middle panel: An AlphaFold model of TRPM3α1 with insertion of 12 additional amino-acid residues specific to TRPM3α1 indicates the sterical inability of the two pore loop cysteines to form a cysteine bridge in the long pore loop TRPM3α1 variant. The pore loop insertion (in orange) is increasing the distance between the two cysteines to a level that disallows cysteine bridge formation. Lower right panel: Mutation of one of the 12 amino acids (K1066) to cysteine, which restores the potential to form a cysteine bridge. A Mean time courses of the calcium fluorescence ratios (F340/F380) of HEK293T cells transiently expressing TRPM3α2 C1079A mutant in NT or DTT-pretreated (10 mM) conditions upon single and co-application of PS (40 µM) and Clt (10 µM). B Same as in ( A ) but for HEK293T cells transiently transfected with TRPM3α2 C1096A mutant. C Same as in ( A ) and ( B ) but for HEK293T cells transiently transfected with TRPM3α2 C1079A + C1096A double-mutant. D Fluorescence ratio increases for the different TRPM3 mutant channels during application of PS (40 µM) and Clt (10 µM) in non-treated (NT) and DTT-pretreated conditions. Statistical analyses between DTT-treated and NT groups performed with a Student unpaired t -test or a Mann-Whitney test ( N  = 3 dishes for each condition, C1079A: PS: p  = 0.66, U  = 6, Z  = 0.44; Clt: p  = 0.66, U = 3, Z  = -0.44; C1096A: PS: p  = 0.20, t -value = -1.87, DF = 2.01, Clt: p  = 0.20, t -value = 1.86, DF = 2.01; C1079A + C1096A: PS: p  = 0.22, t -value = -1.44, DF = 4; Clt: p  = 0.22, t -value = 1.46, DF = 4). E Time course of patch clamp whole-cell currents at ±80 mV recorded in HEK293T cells transiently expressing TRPM3α2 C1079A mutant upon single and co-application of PS (40 µM) and Clt (10 µM). F Same as in ( E ) but for HEK293T cells transiently transfected with TRPM3α2 C1096A mutant. G Same as in ( E ) and ( F ) but for HEK293T cells transiently transfected with TRPM3α2 C1079A + C1096A double-mutant. H Normalized current changes in NT and DTT-pretreated HEK293T cells transiently expressing TRPM3α2 (TRPM3), C1079A, C1096A, or the C1079A + C1096A double-mutant during application of Clt (10 µM). Statistical comparisons of each group with the control group were performed with Student’s unpaired t -test and Mann-Whitney test (TRPM3 NT: n  = 5; TRPM3 DTT: n  = 12, PS: p  = 1.87*10 -03 , U  = 60, Z  = 3.11, Clt: p  = 1.87*10 − 03 , U  = 0, Z  = −3.11; C1079A NT as well as C1096A NT: n  = 6, PS: p  = 8.11*10 −03 , U = 30, Z  = 2.65, Clt: p  = 8.11*10 −03 , U = 0, Z = −2.65; C1079A + C1096A NT: n  = 5, PS: p  = 1.22*10 −02 , U  = 25, Z = 2.51, Clt: p  = 1.22*10− 02 , U  = 0, Z  = −2.51). In addition, further characterization of the C1079A mutant showed a strongly reduced block of the Clt-induced current amplitude by isosakuranetin, primidone, and G βγ subunits (Supplementary Fig. 11) in comparison to wild type TRPM3 in NT-conditions (Fig.  6 ). These results are comparable to wild type TRPM3 in the ‘reduced’ state after DTT-treatment (Fig.  6 ). Notably, a similar cysteine modification in the longer pore-loop mTRPM3α1 isoform (C1079A), did not affect the responsivity towards PS and Clt stimulation (Supplementary Fig.  12 ). Next, the impact of the length of the pore loop on the redox modulation was investigated by stepwise insertion of alanine residues in the short pore isoform TRPM3α2. In a previous study, five channel mutants were designed in the pore loop, ranging from TRPM3α2 + 1 (one alanine was inserted between D1076 and P1077) up to TRPM3α2 + 5. In NT conditions, the TRPM3α2 + 1 mutant showed robust responses to both PS and Clt stimulation, thus exhibiting a mixed phenotype between TRPM3α1 and TRPM3α2. Interestingly, the TRPM3α2 + 1 mutant showed strongly reduced increases in fluorescence to PS stimulation after DTT treatment, while Clt-responses were unaffected (Supplementary Fig.  13 ). In contrast, DTT-treatment was without effect on the TRPM3α2 + 2 mutant, showing direct activation by clotrimazole and no response to PS in both NT- and DTT-treated conditions. Previously, it was shown that all other insertion mutants (TRPM3α2 + 2 to +5) recapitulated the properties of TRPM3α1 3 . These data could suggest that insertion of two or more alanine residues in the pore loop of TRPM3α2 will disturb the formation of the cysteine bridge in the pore loop and therefore cause the channel to become insensitive towards redox modulation. Collectively, these results indicate that the reversible formation of a disulfide bridge between Cys1079 and Cys1096 in the extracellular pore loop of TRPM3α2 underlies the channel’s redox-sensitive properties, and that PS responses are only observed when the disulfide bridge is present. Finally, to evaluate whether the formation of a disulfide bridge in the outer pore loop of the long-pore TRPM3α1 can introduce sensitivity to PS, we introduced cysteines at positions in the longer extracellular loop predicted to be in close-enough proximity with C1109 (the equivalent of C1096 in TRPM3α2) to form a disulfide bridge. Interestingly, in one mutant (K1066C), we indeed detected small but significant PS responses in NT conditions, which were absent in the DTT-pretreated condition (Supplementary Fig.  14 ). These findings confirm the importance of the disulfide bridge in the outer pore loop for PS sensitivity in TRPM3 and indicate that artificial introduction of a disulfide bridge in TRPM3α1 can confer redox-sensitive PS responses (Fig.  8 ).

In this work, we reveal reversible redox modulation of TRPM3, with profound effects on channel gating and pharmacology. Exposure to reducing agents such as DTT and the membrane-impermeable TCEP significantly diminished TRPM3 activation by heat and the ligands pregnenolone sulfate (PS), CIM0216, and nifedipine. This effect was reversed by oxidative treatment with H₂O₂, confirming the redox-dependent nature of TRPM3 modulation. Interestingly, clotrimazole, which acts as a modulator of oxidized TRPM3, exhibited potent agonist activity under reducing conditions, revealing a redox-dependent switch in ligand efficacy. Furthermore, we also found that reducing conditions attenuated TRPM3 inhibition by antagonists or µ-opioid receptor activation and the channel’s basal permeability to Zn 2+ ions. Importantly, these effects were not limited to heterologously expressed TRPM3, as we observed similar redox-dependent TRPM3 modulation in mouse primary sensory neurons and pancreatic beta cells. Finally, based on site-directed mutagenesis and a direct comparison of TRPM3 splice isoforms, we reveal that formation/disruption of a disulfide bridge in the outer pore loop underlies this particular form of redox modulation. Our findings indicate that TRPM3 is an interesting example of a channel that can operate in two different gating “modes” depending on the redox state, which dictates the presence or absence of an intra-subunit disulfide bridge formed between two extracellular pore-loop cysteines (C1079 and C1096) (Fig.  10 ). When the cysteine-bridge is formed, TRPM3 exhibits the “canonical” properties of TRPM3, including activation by PS and heat, basal Zn 2+ permeation, potentiation (but no activation) by Clt, and potent inhibition by compounds such as isosakuranetin and primidone as well as by GPCR activation (Mode 1). When the cysteine bridge is not formed, either due to reducing conditions or due to mutation of critical cysteines, a drastically different gating mode occurs (Mode 2) (Fig.  10 ). Here, the channel is no longer activated by PS or heat and lacks basal Zn 2+ permeation, but is directly activated by Clt, resulting in currents with a linear current-voltage relation. Currents activated under these conditions show a severely reduced sensitivity to inhibition by TRPM3 antagonists or GPCR activation, while PS acts as an antagonist. This model also perfectly aligns with our previously published data, which showed how the length of the pore loop of TRPM3 determines its gating properties 3 . Indeed, we found that introducing two or more alanines in the pore loop of TRPM3α2 immediately N-terminal to C1079, which would likely pull the cysteines too far apart to engage in a disulfide bond, changes the channel properties to Mode 2, irrespective of the redox conditions. In particular, this is the case for the naturally occurring TRPM3α1 splice variant, where an additional 13 amino acids are present immediately N-terminal to C1079 (Fig.  8 ). Interestingly, we found that introducing an extra cysteine residue in the pore loop of TRPM3α1 (K1066C), at a position where it can form a disulfide bridge with the second cysteine in the outer pore loop, rendered this isoform sensitive to PS-induced activation, suggesting that the presence of this disulfide bridge determines, at least partly, PS-induced channel activation. Fig. 10 Model of TRPM3 redox gating. In an oxidized state, short pore loop isoforms possess a cysteine bridge within the pore-loop, which translates to a specific gating mode (Mode 1). Disruption of the cysteine bridge (Mode 2) either by chemical reduction or by insertion of amino acids (AA) between the two pore-loop cysteines (as present in long pore loop isoforms) will induce a switch to another gating mode. In an oxidized state, short pore loop isoforms possess a cysteine bridge within the pore-loop, which translates to a specific gating mode (Mode 1). Disruption of the cysteine bridge (Mode 2) either by chemical reduction or by insertion of amino acids (AA) between the two pore-loop cysteines (as present in long pore loop isoforms) will induce a switch to another gating mode. The recently published cryoEM structure by Yin et al. 2025 5 describes the presence of an extracellular cap-like structure, formed by the pore loop, which can be divided into pore loop (PL)a and PLb. Interestingly, the authors also highlight the unique sequence pattern of the pore loop with two important cysteine residues forming a disulfide bond at the base of PLb, which stabilizes the cap domain. Our data support a model in which the length of PLb strongly affects the position of the two Cys residues and as such the formation of the disulfide bridge. When this bridge is not formed, the channel is no longer activated by PS. The same authors also reported two potential binding sites of PS in TRPM3 5 . The first binding site of PS was identified where it is surrounded by residues from S1 and the pore helix, the binding site 2 is surrounded by the pore helix and S6 and the pore helix of the adjacent subunit 17 . Binding of PS resulted in significant conformational changes across the transmembrane domains and the melastatin homology regions. In particular, PS binding resulted in significant changes in the outer pore, including the splitting of the pore helix into two shorter helices. These changes in the pore helix and the pore loop lead to major rearrangements of S5 and S6, reshaping the pore. Notably, rearrangement of the pore helix also induces rearrangements into PS binding site 2. The substantial conformational changes in the outer pore upon PS binding have made the pore loop highly flexible. Probably, breaking up the disulfide bridge between the Cys residues in the pore loop may affect the PS-induced conformational changes that lead to channel opening. Notably, we found that Clt-activated currents in the reduced channel are inhibited by PS, suggesting that breaking of the disulfide bridge does not impair binding of PS to the channel, but rather affects the conformational changes that occur downstream of PS binding. In earlier work, other TRP channels were shown to be modulated by redox reagents, including the sensory TRP channels TRPA1, TRPV1, TRPM8 57 , 60 , 61 and TRPM7 62 , 63 . However, the modulation of TRPM3 by redox reagents markedly differs from the redox modulation of TRPA1 and TRPV1. In particular, TRPA1 was shown to be directly activated by H 2 O 2 and blocked by reductive agents such as DTT 52 , 53 , 64 , while a sensitization of the heat activation occurred in the presence of H 2 O 2 29 . In contrast to this, no direct actions of redox reagents were detected on the TRPM3 activity. For TRPV1, it was described that the sensitivity towards known activation stimuli like capsaicin and protons is increased in the presence of H 2 O 2 56 . In contrast, TRPM3 undergoes a redox-dependent shift in gating-mode, which is associated with a profound change in the responsivity to different TRPM3 modulators. In addition, the redox-modulation of TRPM3 appears to be solely dependent on two cysteines located in the extracellular pore-loop, which is not aligning with the higher number of redox-sensitive cysteine residues that were reported for TRPV1 and TRPA1. In fact, some of the redox-sensitive residues of TRPV1 and TRPA1 are described to be located intracellularly at the N- and C-termini 53 , 54 , 56 . Nevertheless, cysteine bridge formation in the extracellular pore-loop was suggested for TRPV1 and later also for TRPC5, TRPM7 and TRPM8 channels 55 , 61 , 62 , 65 . Interestingly, mutation of the conserved Cys-residues in TRPM7 altered the rectification of the I - V curve, very similar to TRPM3 channels in reducing conditions, while the presence of these two cysteines in the pore of TRPM8 are essential for responses to the TRPM8 agonists, icilin and menthol 61 . Given the functional evidence for cysteine bridge-controlled gating in TRPM7 62 , and the conserved nature of these two cysteine residues among the TRPM family 62 , it is intriguing to have a closer look whether similar cysteine gating and redox sensitivity exists in other members of the TRPM subfamily. Whereas our cellular experiments suggest that it is possible to induce a reversible switch between two TRPM3 gating modes under experimental conditions, it remains unclear whether such a mechanism occurs in vivo. It is conceivable that, under physiological conditions, a dynamic equilibrium might exist, influenced by the tissue’s redox status and the relative expression of short versus long pore loop TRPM3 isoforms across different cell types. If present, such dynamic redox modulation of TRPM3 could potentially contribute to (patho)physiological processes in which the channel is implicated. For example, oxidative stress has been associated with insulin release 17 and with various forms of pathological pain, including inflammatory, chemotherapy-induced, and neuropathic pain 38 – 40 , 63 , 66 , 67 , and previous studies have suggested an important role for TRPM3 in these pain states 1 , 31 , 63 . Moreover, oxidative stress is a common factor during the development and progression of a multitude of neurological diseases of the CNS, including epilepsy 68 . Considering the recent discoveries implicating TRPM3 in brain development and encephalopathies 28 , 32 – 36 , 69 – 72 , redox modulation of TRPM3 activity in the context of brain disease could potentially represents an important and exciting new direction for further investigation. Importantly, our study also contains an important cautionary message for drug development efforts, aimed at targeting TRPM3 to combat neurological disorders. Indeed, our findings demonstrate that the redox state can have a profound impact on the effect of small-molecule agonists, antagonists and modulators, which needs to be considered when designing safe and effective TRPM3 blockers for specific diseases. A limitation of this study is that TRPM3 modulation by ROS-related reagents has only been demonstrated in vitro. Attempts to confirm this in vivo were unsuccessful, largely due to the non-specific effects of reducing agents such as DTT and TCEP, as well as the limited specificity of clotrimazole. Moreover, there is currently no evidence for the existence of a strongly reducing extracellular environment in vivo that could enable the proposed channel switch. Importantly, the reduced state of TRPM3 observed in our experiments requires artificial intervention with potent reducing agents, and such a state has not been reported under physiological or pathophysiological conditions.

Transient Receptor Potential Melastatin 3 (TRPM3) is a non-selective, Ca 2+ -permeable cation channel that can be activated by a plethora of stimuli, including chemical ligands, cell swelling, and heat 1 – 4 . Recently, the cryo-EM structure of mouse TRPM3 was presented as a tetramer of subunits with 6 transmembrane segments (TM) and cytosolic N and C termini, with a central pore formed by TM5 and TM6 5 , 6 . A distinctive feature of TRPM3 is the extracellular cap-like structure formed by the pore loop 5 . Most abundant expression of TRPM3 is found in the kidney, eye, and pancreas, as well as within several structures of the nervous system (e.g. brain, spinal cord and sensory neurons) 7 – 12 . Over the years, several endogenous ligands were described to induce or block the strongly outwardly-rectifying currents in TRPM3 expressing cells 2 , 13 – 16 . At this point, the neurosteroid pregnenolone sulfate (PS) presents the most potent endogenous activator of the channel. Several other ligands have been described to modulate TRPM3 channel activity, including the anti-fungal drug clotrimazole (Clt), nifedipine (Nif), and the small-molecule synthetic compound CIM0216 2 , 17 . Interestingly, both Clt and CIM0216 were shown to open an alternative ion permeation pathway in TRPM3, distinct from the central canonical pore 2 , 8 . TRPM3 activity can be blocked by synthetic compounds such as the anti-epileptic drug primidone 18 – 20 , the flavonoid isosakuranetin and by activation of G-protein-coupled receptors such as µ-opioid, GABA-B and NPY receptors, due to direct binding of the G βγ subunit to the N-terminus of the channel 21 – 24 . Notably, a multitude of isoforms have been described for TRPM3, arising from alternative splicing 13 , 25 – 28 , . Interestingly, it was recently shown that significant differences in the activation and blocking pattern exist between the long-pore α1 and short-pore α2-α6 isoforms of TRPM3 3 , 25 . In general, most research has been done using TRPM3α2, representing a channel isoform with properties similar to endogenous TRPM3 channel activity in cell types such as dorsal root ganglion (DRG), trigeminal ganglion (TG) neurons and pancreatic islet cells 1 , 17 . TRPM3 acts as a sensor of noxious stimuli in the peripheral nervous system (PNS), contributing to the detection of acute heat 29 , 30 and to the development of heat hyperalgesia during inflammation 1 and after neuronal damage 31 . TRPM3 is also highly expressed in the central nervous system (CNS), but its (patho)physiological roles in the brain remain largely unknown. Recently, de novo gain-of-function variants in TRPM3 were identified to cause neurodevelopmental and epileptic diseases in humans, indicating an important function of TRPM3 within the CNS 28 , 32 – 37 . Redox signaling and reactive oxygen species (ROS) play important roles in many neurological diseases, including persistent pain 38 – 40 and brain disorders such as epilepsy and stroke 41 , 42 . ROS are generated during the mitochondrial oxidative metabolism or during cellular responses to xenobiotics, cytokines, and bacterial invasion. ROS-induced oxidation can result in the formation of disulfide bridges between cysteine residues, thereby influencing protein structure and function 43 . Here, we demonstrate a specific, bidirectional modulation of the gating properties of TRPM3 by redox reagents and identify a molecular mechanism in which a cysteine bridge, located in the extracellular domain adjacent to the pore loop, is responsible for the modulation of the gating properties. These results clearly indicate that the redox status of TRPM3 strongly affects the pharmacology and heat sensitivity of the channel.

All experiments were performed according to the European legislation and were approved by the local ethics committee of the KU Leuven (project 184/2021; Scientific Research involving laboratory animals — Research involving laboratory animals). The design of TRPA1_TRPV1 double knock-out and TRPA1_TRPV1_TRPM3 triple knock-out mice lines is described in earlier work 29 . HEK293T cells stably expressing murine TRPM3α2 (hereafter named HEK-TRPM3) were designed and cultured as described previously 1 . For the overexpression of other TRPM3 plasmids (human TRPM3 and TRPM3 mutants), HEK293T cells were transiently transfected with 2 µg of DNA using TransIT transfection reagent (Mirus) 12–36 h before the start of the experiments. To test the blocking efficacy of DAMGO on TRPM3 channel activity, mTRPM3α2 and µ-opioid receptor DNA were co-transfected in a 1:1 ratio of 1 µg each. Dorsal root ganglion (DRG) and trigeminal (TG) neurons were isolated as described previously 44 along the entire vertebral column of both male and female mice. Isolation of mouse pancreatic islets and dissociation into single cells were performed as described previously 45 . The murine pancreatic β-cell lines MIN6 and INS-1 were incubated at 5% CO 2 and 37°C and cultured as previously described 45 , 46 . To obtain homogenously shaped pancreatic β-cell spheroids, MIN6 cells were seeded in glass-bottom 96-well plates (P96-1.5H-N, Cellvis Inc.) containing 19 micro-pillar structures per well, as described previously 47 . All mutants were obtained by the standard PCR overlap extension method using mTRPM3α2 ( AJ544535 ) and mTRPM3α1 ( AJ544532 ) from the pCAGGSM2/IRES/GFP vector 48 . The accuracy of all mutant sequences was verified by sequencing of cDNA constructs. Since several different splice isoforms of TRPM3 have been described, a new numbering system for the amino acid positions within TRPM3 was introduced 28 to simplify and unify reported TRPM3 variants. Here, we adhere to this counting system throughout the entire manuscript when describing the newly constructed TRPM3 mutant channels. The imaging system for standard calcium measurements has been described before 49 . Briefly, cells were incubated with 2 µM Fura-2 acetoxymethyl ester for 20–40 min at 37 °C. Fluorescent signals were evoked during alternating illumination at 340 and 380 nm using a Lambda XL illuminator (Sutter Instruments, Novato, USA), and recorded using an Orca Flash 4.0 camera (Hamamatsu Photonics Belgium, Mont-Saint-Guibert, Belgium) on a Nikon Eclipse Ti fluorescence microscope (Nikon Benelux, Brussels, Belgium). The imaging data were recorded and analyzed using NIS-elements software (Nikon, Japan). The standard imaging solution contained (in mM) 150 NaCl, 2 CaCl 2 , 1 MgCl 2 , and 10 HEPES (pH 7.4 with NaOH). In the case of primary sensory neurons, the solution was supplemented with 10 mM glucose. Increases in fluorescent ratios were calculated per individual cell as the difference between the maximum value during the stimulus period and the mean value at the end of the preceding stimulus period. To determine an average temperature-response relation for TRPM3 in non-treated cells and cells that underwent DTT-treatment, a fluo-4-based assay was performed using 96-well plates and the 7500 Real-Time PCR system (Applied Biosystems) as described previously 1 . HEK-TRPM3 or non-transfected HEK293T cells were loaded with Fluo-4-AM for 30 min, trypsinized, centrifuged, resuspended in a solution containing (in mM) 150 NaCl, 6 KCl, 2 MgCl 2 , 2 CaCl 2 , and 10 HEPES (pH 7.4 with NaOH), and transferred to a 96-well plate (at 15,000–25,000 cells/well; 50 ml). When indicated, PS (10 mM) was added to the wells. Fluo-4 fluorescence was measured while the well temperature was raised from 16 °C to 43 °C in 3-degree steps. Background-subtracted fluorescence signals were used to calculate temperature-induced changes in fluorescence as ΔF/F 16°C , where F 16°C is the background-corrected fluorescence at 16 °C and ΔF = F- F 16°C . To measure the dose-response correlation of PS and Clt in non-treated and DTT-treated conditions and to measure PS and Clt responses under different redox potentials, HEK293T cells stably expressing murine Trpm3 (HEK-mTRPM3) and non-transfected HEK293T cells were cultured as described previously 1 and used until passage number 25. Cells were seeded on poly-L-lysine (PLL)-coated 96-well black walls and clear bottom assay plates (Azenta 4ti-0263) and incubated 18 to 24 hours at 37 °C and 10% CO2. Intracellular Ca 2+ levels were measured using a fluorescence microplate reader. Forty to sixty minutes prior to the experiment, the growth medium was removed from the 96-well plate, and the cells were treated with 100 µL/well loading buffer (1 µM Fura-2 acetoxymethyl ester (Thermo Fisher Scientific), 250 µM sulfinpyrazone (Invitrogen), and 3 µM pluronic acid (Invitrogen) in 11 ml culture medium). The loading buffer was supplemented with 10 mM 1,4-Dithiothreitol (DTT) reduction agent under DTT-treated conditions. Afterwards, the loading buffer was removed from the 96-well plate, and the cells were washed with 100 µL/well bath solution (in mM) (150 NaCl, 2 CaCl 2 , 1 MgCl 2 , 10 HEPES (pH 7.4, adjusted with NaOH)). Intracellular changes in Ca 2+ levels were measured as the ratio of the Fura-2 fluorescence signals at both wavelengths (F340/F380) using a FDSS/μCELL kinetic 96-well plate imager (Hamamatsu Photonics K.K.) at room temperature. The temperature was fixed to 24 °C to maintain the calibrated redox conditions. Various concentrations of PS and Clt were used to generate the corresponding dose-response curves. Four to five replicates per condition were included for each 96-well plate. As a positive control, ionomycin (2 µM) was used at the end of each experiment. In the case of calibrated redox conditions, ionomycin (2 µM) was used in a buffer solution containing (in mM): 150 NaCl, 20 CaCl 2 , 1 MgCl 2 , 10 HEPES (pH 7.4, adjusted with NaOH), to ensure complete saturation of the fluorescent signals. Non-transfected cells were used as a negative control. To perform a background correction, an empty 96-well plate was measured, and the corresponding signal was subtracted from the experimental signal of the same day. FluoZin-3, AM (Invitrogen) was dissolved in DMSO to a concentration of 5 mM. Cells seeded on glass coverslips were loaded with 5 µM FluoZin-3 AM for 30–50 min at 37 °C. Fluorescent signals were evoked during illumination at 490 nm at 0.5 Hz using a Lambda XL illuminator (Sutter Instruments, Novato, USA), and recorded using an Orca Flash 4.0 camera (Hamamatsu Photonics Belgium) on a Nikon Eclipse Ti fluorescence microscope (Nikon Benelux, Belgium). The imaging data were recorded using NIS-elements software (Nikon, Japan). First analysis was performed with ImageJ, where individual regions of interest (ROI) were defined after background subtraction, and traces from each ROI were obtained. OriginPro 8.6 (OriginLab, MA, USA) was further used for analysis and data display. All traces were normalized to the baseline value, defined as the mean of the first 20 values. The standard imaging solution contained (in mM) 150 NaCl, 2 CaCl 2 , 1 MgCl 2 , and 10 HEPES (pH 7.4 with NaOH). A stock solution of 100 mM ZnCl 2 was prepared in MilliQ water and further diluted to the indicated dose. Standard patch clamp recordings were measured with an EPC-10 amplifier and the PatchMasterPro Software (HEKA Elektronik, Germany). Current measurements were performed at a sampling rate of 20 kHz, and currents were digitally filtered at 2.9 kHz. For whole-cell patch clamp recordings on HEK293T cells, the standard internal solution contained (in mM): 100 CsAsp, 45 CsCl, 10 EGTA, 1 MgCl 2 , and 10 HEPES (pH 7.2 with CsOH). The standard extracellular solution contained (in mM): 150 NaCl, 1 MgCl 2 , 10 HEPES (pH 7.4 with NaOH). For whole-cell patch clamp recordings on primary pancreatic islet cells, the extracellular solution was supplemented with 10 mM glucose. The standard patch pipette resistance for all experiments was between 2 MΩ and 4 MΩ when filled with pipette solution. Between 50% to 70% of the series resistance was compensated using the auto-compensation function of PatchMaster, reducing series resistance errors to below 1-5 mV. No compensations for liquid junction potentials were performed. Differences in current density were calculated as the difference between the mean value during the last 6 s of the stimulus period and the mean value of the basal current densities before and after stimulation. To determine the abilities of the different chemicals to induce TRPM3-current increases, treatment-specific current increases (PS, Clt, co-application of PS+Clt) were normalized to the sum of the current increases upon PS and Clt treatment, respectively. Electrophysiological and calcium microfluorimetric data were analyzed using NIS-Elements software (Nikon, Japan), Excel (Microsoft, USA), IgorPro 6.2 (WaveMetrics, USA), GraphPad Prism 9.0.1 (GraphPad Software, USA), and OriginPro 8.6 (OriginLab, USA). OriginPro 8.6 was further used for data display. All data sets were tested for normalit,y and a student’s two-tailed unpaired t -test or a Mann-Whitney U test was used for statistical comparison between two different data sets. Where applicable, a Bonferroni correction was used to correct for multiple testing. For pairwise comparisons, data was tested with a Paired-sample Wilcoxon-signed-rank test or a Student Paired-sample test. For multiple groupwise comparisons (groups > 2) a One-way ANOVA or a non-parametric Kruskal-Wallis test was applied, followed by a Tukey’s post hoc or a Dunn’s post hoc test. For normally distributed data, the variance homogeneity was tested. P values below 0.05 were considered significant. Data points represent means ± s.e.m of the given number ( n ) of individual experiments. Cryo-EM structures of TRPM3α2 were retrieved from the RCSB Protein Data Bank (PDB ID: 9B29) and used as a template for homology modeling. The pore loop region of TRPM3α1 was formed using the homology modeling feature in YASARA, with the PDB structure 8DDQ serving as the template. Additionally, an AlphaFold3.0 predicted model of TRPM3α1 was consulted to guide and validate the homology model. All structural models were visualized and analyzed using PyMOL.

Transparent Peer Review file Supplementary Information Transparent Peer Review file Supplementary Information