Rapamycin activates TRPM8 in HEK293 cells and sensory neurons.

(A) Time course of the intracellular calcium concentration in HEK293 cells expressing TRPM8, showing robust responses to rapamycin (10 µM) and menthol (50 µM), and inhibition of the responses by AMTB (2 µM). Shown are mean ± SEM, N=63 cells

(B) Concentration dependence of rapamycin-evoked calcium responses. The dashed line respresents the best fit using a Hill equation (EC50=3.1; nH=3.8). Shown are mean ± SEM, N=52-95 cells/concentration. The red symbol indicates the lack of response to rapamycin (10 µM) in non-transfected HEK293 cells (N=23).

(C) Left, time course of whole-cell currents in HEK293 cells expressing TRPM8 evoked by repetitive voltage steps to +120 and -80 mV, showing the activation of outwardly rectifying currents by rapamycin (10 µM) and inhibition by AMTB (2 µM). Right, voltage steps recorded at the indicated time points.

(D) Concentration dependence of rapamycin-evoked whole-cell currents at +120 mV. The dashed line respresents the best fit using a Hill equation (EC50=3.8; nH=1.0). Shown are mean ± SEM, N=5 cells per concentration

(E) Examples of calcium signals in individual DRG neurons from Trpm8+/+ and Trpm8-/- mice in response to rapamycin (Rapa; 10 µM), menthol (Menth; 50 µM), pregnenolone sulphate (PS; 40 µM), cinnamaldehyde (CIN; 10 µM), capsaicin (Caps; 100 nM) or high K+ (50 mM).

(F) Fraction of sensory neurons from Trpm8+/+ and Trpm8-/- mice that responded to rapamycin and menthol.

Direct interaction between rapamycin and TRPM8.

(A) Left, time course of currents in a cell-free inside-out patch pulled from a HEK293 cell expressing TRPM8 evoked by repetitive voltage steps to +80 and -80 mV, showing the activation of outwardly rectifying currents by rapamycin (10 µM) and menthol (50 µM) applied from the cytosolic side. Right, voltage steps recorded at the indicated time points. This example is representative for 5 similar experiments.

(B) Cartoon representing the steps to obtain the direct interaction of rapamycin with TRPM8. Individual STD spectra were recorded on different sample compositions, then non-specific interactions were filtered out with multiple subtractions resulting in the double difference spectra (STDD-1,2) and the final triple difference spectrum (STTD) shown on the right side.

(C) The four STD spectra (STD-1,2,3,4) are overlayed for comparison. Double difference spectra were computed from the respective STD pairs showing the specific and non-specific binding of rapamycin (STDD-1, green) and the non-specific binding of rapamycin alone (STDD-2, purple). All non-specific interactions were filtered out in the final STTD spectrum (red) computed by subtracting STDD-2 from STDD-1. Here, only one experimental set is shown (dataset B), spectra on the replicates can be found in the supplementary (Figures S5 and S6).

(D) Rapamycin resonances involved in the direct interaction with TRPM8 were assigned by the comparison of the reference 1H NMR spectra of rapamycin (black) with the computed STDD effects (red). The reference spectrum was recorded on a 0.2 mM rapamycin sample in a D2O buffer solution at 25 °C. The number of scans was 256 and a watergate sequence was used to suppress the residual water signal.

(E) Hydrogen atoms involved in the interaction with TRPM8 are mapped to the structure of rapamycin (yellow circles).

TRPM8 residues involved in the interaction with rapamycin and menthol.

(A) Representative time courses of the intracellular calcium concentration in HEK293 cells expressing wild type TRPM8 or the indicated mutants, when stimulated with rapamycin (10 µM), menthol (50 µM) and the calcium ionophore ionomycin (2 µM).

(B) Quantification of the relative calcium response to rapamycin and menthol for wild type and the indicated TRPM8 mutants. Values indicate the ratio between the calcium response amplitude to rapamycin, divided by the sum of the responses to rapamycin and menthol. The dotted line represents the mean value for wild type TRPM8. Values above this line (in yellow) indicate a relative reduction in the response to menthol, whereas values below the line (cyan) indicate a relative reduction in the response to rapamycin.

(C) Amplitude of the calcium response to the agonist (menthol or rapamycin) that gave the largest response for wild type and the indicated TRPM8 mutants. Data in B and C represent mean ± SEM, N=34-156/group.

(D) Whole-cell current-voltage relations for the currents in control, and in the presence of rapamycin (10 µM) or menthol (50 µM) in HEK293 cells expressing wild type TRPM8 or the indicated mutants. *, **, *** indicate p<0.05, p<0.01 and p<0.001 compared to WT.

(E) Quantification of the relative current response to rapamycin and menthol for wild type and the indicated TRPM8 mutants. Values indicate the ratio between the current amplitude increase at +80 mV to rapamycin, divided by the sum of the responses to rapamycin and menthol. The dotted line represents the mean value for wild type TRPM8. Values above this line (in yellow) indicate a relative reduction in the response to menthol, whereas values below the line (cyan) indicate a relative reduction in the response to rapamycin. Data in represent mean ± SEM; N=5-8 per mutant.

Structural model showing the distinct interaction sites for menthol and rapamycin.

(A) Side view (left) and top view (right) of TRPM8, showing the known interaction site for menthol (green) and the proposed rapamycin interaction site (red) based on our present molecular docking and mutagenesis studies.

(B) Closer view of rapamycin docked onto the TRPM8 structure. Amino acid residues that, when mutated, influence rapamycin responses are indicated in green.

(C) 2D projection of interactions between rapamycin and TRPM8 created using Ligplot+.

Activation of TRPM8 by rapamycin is independent of intracellular calcium.

(A) Whole-cell currents during 800-ms voltage steps from -80 to +80 mV under control conditions and in the presence of rapamycin (10 µM) or icilin (10 µM). At the time points indicated by the arrows, a 1-ms UV falsh was applied, leading to a rapid increase in intracellular calcium. Magenta lines and scale bar indicate Fura-FF fluoresence ratios at the indicated time points.

(B) Whole-cell current-voltage relations measured during voltage ramps 2 s before and 1 second after the UV flashes shown in panel A.

(C) Current amplitudes at +80 and -80 mV before and after UV uncaging of calcium, under control conditions and in the presence of rapamycin or icilin.

(D) Quantification of the relative potentiation of inward and outward currents following UV uncaging of calcium. Data in C and D represent the mean ± SEM from 5 experiments. *,**, ***: p<0.05 in a paired t-test comparing currents before and after UV flash.

Effect of rapamycin and macrolid analogs on TRPM8.

(A) Overview of the macrolids tested in this study.

(B) Relative calcium response to rapamycin and the indicated analogs tested at 10 µM. N=6 in each group.

(C) Fura2-based calcium response to rapamycin (10 µM) in the presence of everolimus (10 µM) or vehicle. N=6 in each group.

(D) Fura2-based calcium response to menthol (50 µM) in the presence of everolimus (10 µM) or vehicle. N=6 in each group.

(E) Summary of the responses to menthol (50 µM) and rapamycin (10 µM) in the absence or presence of everolimus (10 µM). Responses were normalized to the response to a saturating concentration of menthol (300 µM). N=6 in each group. ***: p<0.001.

Rapamycin activates hTRPM8 in a concentration-dependent manner.

(A-B) Concentration dependence of rapamycin-evoked calcium responses measured by a microplate reader at room temperature (23 °C) (A) and 37°C (B). Shown are mean ± SEM; N=5 wells for each concentration. The lines represent the best fit using a Hill equation, yielding EC50 values of 6.0 ± 0.3 µM and 10.1 ± 0.2 µM, and Hill coefficients of 1.9 and 4.8, at 23°C and 37°C respectively.

Rapamycin does evoke calcium signals in naïve HEK293 cells.

(A) Time course of the intracellular calcium concentration in non-transfected HEK293 cells, when stimulated with rapamycin (10 µM), menthol (50 µM) and the positive control stimulus ionomycin (2 µM). Shown are mean ± SEM, N=23 cells.

(B) Whole-cell current-voltage relations obtained in non-transfected HEK293 cells, using the same approach as used for wild type and mutant TRPM8 in Figure 3D. In N=5 similar experiments, changes in current amplitude upon addition of menthol or rapamycin never exceeded 50 pA at -100 and +100 mV.

Rapamycin does not activate TRPA1, TRPM3 or TRPV1.

(A-C) Representative whole-cell current-voltage relations in CHO-cells expressing mouse TRPA1 (A), HEK293 cells expressing mouse TRPM3 (B) or HEK293 cells expressing human TRPV1 (C) upon stimulation with rapamycin (30 µM) or the respective agonists Allyl isothiocyanate (AITC), pregnenolone sulphate (PS; 40 µM) and capsaicin (Caps; 100 nM). Current responses to rapamycin where <5% of the response to the respective channel agonists in n=5 cells.

Rapamycin allows distinghuising between TRPM8-mediated and TRPM8-independent menthol responding sensory neurons.

(A) Representative examples of Fura2-based calcium signals in two sensory neurons that were stimulated three times at the indicated times with rapamycin (10 µM) or menthol (50 µM). During the second application, the TRPM8 antagonist AMTB was present at a concentration of 2µM.

(B) Average ratio between the response to menthol in the presence and absence of AMTB (second response/first response) in neurons that did or did not respond to rapamycin.

The variability of peak heights in 1H NMR experiments and the normalization procedure.

Reference 1H experiments are shown on three individual sets of samples from which later three parallel STTD spectra were computed by the linear combinations of STD experiments. Conditions and sample numbering are shown in the right corner of the plots. The size of the spectra varied significantly despite the strict production protocol due to the complexity of the samples. Therefore, the scaling of the follow-up STD experiments was performed prior to the linear combinations based on the peak integral sums from the 0-4 ppm region of 1H experiments. The corresponding STD spectra were scaled to an external spectrum recorded on a cell sample with an overexpressed ion channel and rapamycin. The following factors were used for multiplication: Dataset A: (1): 0.29, (2): 0.24, (3): 0.48, (4): 0.40. Dataset B: (5): 0.70, (6): 0.82, (7): 1.05, (8): 4.92. Dataset C: (9): 0.34, (10): 0.28, (11): 0.60, (12): 0.46.

Replicates of STTD measurements.

(A) An STTD spectrum obtained by the linear combinations of scaled STD spectra [(4-3) - (2-1)]. A direct interaction between rapamycin and the ion channel is confirmed and interaction sites of rapamycin were revealed upon peak assignments (presented in Figure 2).

(B) An STTD spectrum obtained by the linear combinations of scaled STD spectra [(8-7) - (6-5)]. A direct interaction between rapamycin and the ion channel is confirmed and the same interaction sites of rapamycin were revealed upon peak assignments.

(C) An STTD spectrum obtained by the linear combinations of scaled STD spectra [(12-11) - (10-9)]. Here, there is no reliable STD effect confirming the direct interaction between TRPM8 and rapamycin. Increased cell sedimentation was observed in sample 12 after measurements, an unconotrolled process that might have reduced the detected STD effects. Consequently the extent of non-specific binding outweighed specific binding resulting in zero or „negative” effects in the STTD spectrum.

Rapamycin acts as a type I agonist.

(A) Representative examples of whole-cell currents in HEK293 cells expressing TRPM8, in response to a 200-ms voltage step from -80 to +120 mV and back to -80 mV, in control conditions and in the presence of the indicated concentrations of rapamycin or menthol.

(B) Zoomed-in time course of current relaxation at +120 mV. Dashed lines indicate monoexponential fits.

(C) Zoomed-in time course of current relaxation at -80 mV. Dashed lines indicate monoexponential fits.

(D) Monoexponential time constants for current activation at +120 mV in control conditions and in the presesence of the indicated concentrations of menthol or rapamycin.

(E) Monoexponential time constants for current deactivation at -80 mV in control conditions and in the presesence of the indicated concentrations of menthol or rapamycin. **, ***: P<0.01, P<0.001 versus control. ##, ###: P<0.01, P<0.001 versus menthol. Mean ± SD, dots represent individual cases.

Additive effects of Rapamycin and menthol and TRPM8 deactivation.

(A) Representative examples of whole-cell currents in HEK293 cells expressing TRPM8, in response to a 200-ms voltage step from -120 to +120 mV and back to -120 mV, in control conditions, in the presence of rapamycin (10 μM) and in the combined presence of rapamycin (10 μM) and menthol (50 μM).

(B) Mean current deactivation, calculated as the fraction of the peak current upon the final step to -120 mV that deactivated after 100ms. Data from N=6 cells.

Potential binding sites and poses of rapamycin on TRPM8 obtained from blind pilot dockings.

(A-B) site 1, (C-D) site 2, (E-F) site 3. Rapamycin was docked onto the structure of full-length TRPM8 from the collared flycatcher (Ficedula albicollis; pdb code: 6NR2). For clarity, the amino acid numbering refers to the corresponding residues in the human TRPM8 ortholog.

Rapamycin, but not menthol, moderately activates HEK cells expressing TRPM8F847A

(A) Time course of the intracellular calcium concentration in HEK293 cells expressing TRPM8F847A, showing moderate responses to rapamycin (10 µM) but no response to menthol (50 µM). Ionomycin was used as positive control. Shown are mean ± SEM, N=114 cells

(B) Statistical analysis on the amplitude of Ca2+ transients evoked by menthol (50 µM), rapamycin (10 µM) and ionomycin. Shown are mean ± SEM; N=114 cells.

Structural model indicating side chains for all mutations, colored by menthol (green)/rapamycin (red) selectivity.

Visualizing and comparing the STTD NMR results with the predicted rapamycin binding pocket.

(A) 2D projection of interactions between rapamycin and TRPM8 created using Ligplot+. The yellow circles mark the carbon atoms of the rapamycin which the interacting hydrogen atoms identified by STTD NMR belong to.

(B) Structure of rapamycin indicating hydrogen atoms involved in the interaction with TRPM8. The interacting hydrogen atoms proposed by the STTD NMR experiments are marked by the yellow circles.

Molecular docking of rapamycin and everolimus to the groove between voltage sensor-like domain and the pore domain of TRPM8.

(A-B) In silico molecular docking indicates that everolimus binds to TRPM8 in a similar pose as rapamycin, albeit with lower binding energy (-8.5 kcal/mol for everolimus versus -11.6 kcal/mol for rapamycin).

(C-D) A zoom-in on the binding site shows the hydrogen bond between the hydroxyl group on the cyclohexane ring in rapamycin (C) and the side-chain amide of residue Gln861, whereas no such a hydrogen bond can be formed between the longer hydroxyethyl moiety on everolimus (D).