Androgenic TRPM8 activity drives sexual dimorphism in a murine model of chronic migraine

The mechanisms contributing to the high prevalence of chronic migraine in females are yet elusive. Here, we used a mouse model of nitroglycerin-induced chronic migraine that displays a sexual dimorphic phenotype and unveiled a role of TRPM8 as a testosterone receptor that provides antinociceptive resilience exclusively in males. Nitroglycerin induced similar mechanosensitivity to both sexes trough activation of TRPA1 channels, but triggered persistent hypersensitivity solely in females, as males readily recovered from the migraine crisis. Notably, we found that testosterone exerted an antinociceptive activity through its interaction with the TRPM8 channel. Downregulation of this protective mechanism in males led to persistent mechanical hypersensitivity, whereas administration of testosterone to females favoured their recovery. Thus, our ndings reveal a novel protective function of TRPM8 through pre-clinical models of acute and chronic pain and highlights the interest of molecular solutions mimicking the pain-relieving activity of testosterone on TRPM8.


Introduction
Chronic migraine is a highly prevalent and recurrent headache a iction particularly severe in women 1 .
The mechanisms underlying such sex dimorphism of chronic migraine remain largely unknown. A migraine model with high predictive validity is the sensory sensitization induced by nitroglycerin (NTG) 2 .
Acute NTG treatments lead to a delayed hypersensitivity to mechanical stimulation that lasts hours in humans and rodents 2,3 . Furthermore, repeated NTG exposure causes a chronic hypersensitivity that lasts several weeks in murine models 4 . This chronic hypersensitivity is characterized by generalized cutaneous sensitization, which has also been described as a reliable predictor of migraine chroni cation in humans 5 .
Multiple migraine triggers including NTG have demonstrated a crucial involvement of transient receptor potential ankyrin 1 (TRPA1) in rodent models of acute migraine 3,6 . Indeed, TRPA1 is expressed in primary afferent neurons innervating the meninges where its activation favours the release of α-calcitonin generelated peptide (αCGRP) 6−8 a neuropeptide that plays a pivotal role in migraine development 2,9,10 . An additional TRP tightly associated with the expression of chronic migraine in humans is the transient receptor potential melastatin 8 (TRPM8). Several single-nucleotide polymorphisms affecting TRPM8 have been linked to migraine 11,12 and TRPM8 agonists such as menthol have been used medicinally for the alleviation of migraine-related pain 13,14 .
TRPM8 is a cation channel expressed in primary afferent neurons, known for being the menthol receptor and the principal detector of environmental cold [15][16][17][18] . As such, TRPM8 activity shows modulatory effects on thermal and mechanical hypersensitivity in preclinical models of pain 19 . However, its presence in internal structures kept at euthermic temperature 20 and its recent description in central brain areas 21 suggests additional functions of this protein that may go beyond cold perception. In this line, TRPM8 was previously described as a testosterone receptor in cellular models 22 . While high testosterone levels have been associated with decreased pain sensitivity in mice 23,24 and humans 25 , it is unknown whether testosterone-TRPM8 interactions could have functional relevance in pain perception.
Here we implemented a murine model of chronic migraine that displays a sexual dimorphism characterized by enhanced pain sensitivity of females as described in humans. Mechanical sensitivity was assessed in mice of both sexes chronically exposed to NTG and the participation of TRPA1 and TRPM8 was evaluated through genetic and pharmacological approaches. To dissect the functional and molecular consequences of TRPA1 and TRPM8 activities, murine cultures of trigeminal neurons and transfected cell lines expressing murine and human receptors were evaluated. After nding a malespeci c function of TRPM8, the role of testosterone-TRPM8 interactions and the effects of exogenous TRPM8 stimulation were elucidated in in vitro and in vivo models of acute and chronic pain. Molecular docking in murine and human ligand-receptor models provided further insight on TRPM8 function as a testosterone receptor. Collectively, our data suggest that testosterone, through its interaction with TRPM8, drives sexual dimorphism in chronic migraine and likely in other pain-related behaviours. Methods 2.1. Animals. Adult male and female mice with a C57BL/6J background (Envigo, Horst, The Netherlands), wild-type or defective in Trpa1 26 or Trpm8 17 were bred in the animal facility at Universidad Miguel Hernández (UMH, Elche, Alicante, Spain). TRPM8 knockout mice were a gift from Dr. F. Viana (Instituto de Neurociencias de Alicante, Alicante, Spain). Care was taken to minimize the number of animals used and the pain and stress they experienced. All experimental procedures were approved by the Animal Care and Use Committees of Universidad Miguel Hernández and the regional government and were conducted according to the ethical principles of the International Association for the Study of Pain (IASP) for the evaluation of pain in conscious animals 27 , the European Parliament and the Council Directive (2010/63/EU) and the Spanish law (RD 53/2013). Housing conditions were maintained at 21 ± 1°C and 55 ± 15% relative humidity in a controlled light/dark cycle (light on between 8:00 a.m. and 8:00 p.m.).
Animals had free access to food and water except during manipulations and behavioural assessment.
Experiments were performed blinded for NTG, genotype or pharmacological treatment depending on the studied condition.
2.2. Drugs for behavioural studies. Two NTG formulations were used: 5 mg/1.5 ml ampoules and 50 mg/50 ml vials (Bioindustria LIM, Novi Liguri, Italy). The ampoules contained NTG dissolved on a vehicle made of 1 ml propylene glycol and 0.5 ml ethanol (Bioindustria LIM). This initial solution was dissolved in saline to obtain 1 mg/ml NTG, reaching nal concentrations of 10% ethanol and 20% propylene glycol. The 50 mg/50 ml NTG vials contained a vehicle made of 5% dextrose and 0.105% propylene glycol in pure water (Bioindustria LIM). This NTG or its vehicle was administered without further dilution. The TRPM8 selective blocker AMTB hydrochloride (AMTB, N-(3-aminopropyl)-2-{[(3-methylphenyl) methyl] oxy}-N-(2-thienylmethyl) benzamide hydrochloride, Tocris, Bristol, UK) was dissolved in dimethyl sulfoxide (DMSO, Merck, Darmstadt, Germany) and was further diluted in saline to reach 2.5% DMSO. The potent and selective TRPM8 agonist WS12 ((1R,2S,5R)-2-Isopropyl-N-(4-methoxyphenyl)-5-methylcyclohexanecarboxamide, Tocris) was dissolved in DMSO and diluted in corn oil to reach 2.5% DMSO. In a previous preparation, WS12 was dissolved in ethanol and diluted in 45% 2-Hydroxypropyl-βcyclodextrin in water to reach 5% ethanol, although precipitation was found at these concentrations. All these compounds and vehicles were injected intraperitoneally at a volume of 10 ml/kg. In the formalin test, WS-12 was dissolved in DMSO and diluted in saline up to 0.6% DMSO to achieve an amount of 6 nmol in 20 µl as previously described 28 . We used WS12 and not menthol or icilin as a TRPM8 agonist to avoid unspeci c signaling over TRPA1 29 . Testosterone (T1500, Merck) was dissolved in 45% 2-Hydroxypropyl-β-cyclodextrin in water to obtain a solution of 22 mg/ml. 2.3. Model of chronic migraine. Animals were exposed to a schedule of repeated NTG injections previously used to precipitate long lasting mechanical hypersensitivity 4 . Brie y, mice were injected with 10 mg/kg NTG or its vehicle every other day for 8 days (5 i.p. injections total). Mechanical sensitivity was rst assessed before the repeated NTG treatment (days -1 and 0). Afterwards, mechanical thresholds were measured again every day of the treatment before and 2 h after each injection (days 0, 2, 4, 6 and/or 8). Then, measurements continued up to 20 days after the beginning of the procedure (days 10, 12, 14, 16, 18 and/or 20).
2.4. Assessment of mechanical sensitivity. Mechanical thresholds were quanti ed by measuring the hind paw withdrawal response to von Frey lament stimulation. Brie y, animals were placed in Plexiglas® chambers (10x10x14 cm) with a wire grid bottom through which the von Frey laments (bending force range from 0.008 to 2 g) (PanLab, Cornellá, Barcelona, Spain) were applied, by using the up-down paradigm as previously described 30 . The lament of 0.4 g was rst applied. Then, the strength of the next lament was decreased when the animal responded or increased when the animal did not respond. The upper limit value (2 g) was recorded as a positive response even if there was no withdrawal response, and the lower limit was recorded as negative even if there was withdrawal response (0.008). This up-down procedure was stopped 4 measures after the rst change in animal responding (i.e. from response to no response or from no response to response). The sequence of the last 6 responses was used to calculate the mechanical threshold. To decrease stress, prior to baseline measurements mice were habituated for 4 hours to the testing environment during 2 days. On the evaluation days, animals were also allowed to habituate for 1-2 hours before testing in order to obtain appropriate behavioural immobility. Both ipsilateral and contralateral hind paws were alternatively tested whenever possible, and stimuli were applied at a minimum of 2 min intervals to avoid hypervigilance or sensitization between successive lament applications. Filaments were completely bent before considering responses and hold up to 4-5 s to consider a negative response. Clear paw withdrawal, shaking or licking were considered as nociceptivelike responses. The responses of both hind paws were averaged to obtain the mechanical threshold of each individual.
Afterwards, NTG 100 µM or its vehicle (5% dextrose and 0.105% propylene glycol) were co-applied with DD04107 10 µM or its vehicle for 30 additional min. Incubation solutions were made in culture medium and kept at 37ºC and 5% CO 2 .
2.7. Immunocytochemistry. 30 min after NTG exposure (vehicle 5% dextrose and 0.105% propylene glycol), the media was removed from the cells and the culture was washed with PBS 1X (D8662, Merck) 3 times. Afterwards, paraformaldehyde 4% (158127, Merck) was applied for 20 min at room temperature.  10% FBS and penicillin/streptomycin 1% at 37°C in a 5% CO 2 atmosphere. IMR90 experiments were performed when con uence reached 50-60%. were seeded in 12 mm coverslips at 50.000 cells/well and were maintained in Minimum Essential Medium (MEM) enriched with 10% FBS and penicillin/streptomycin 1% at 37°C in a 5% CO 2 atmosphere. IMR90 experiments were performed when con uence reached 50-60%. Cells were kept overnight in a hormone free medium, by eliminating FBS from the composition of the culture medium, substituting MEM or DMEM by no phenol red opti-MEM (Gibco 11058021). 2.11. Chemically-induced nocifensive behaviour (Formalin test). Mice were individually placed into transparent chambers and were habituated for 1 h before testing. Afterwards, 20 µl of a 45% 2-Hydroxypropyl-β-cyclodextrin solution containing 5% formalin (F8775, Merck) and 0.6% DMSO with or without 6 nmol of WS-12 were injected subcutaneously into the plantar aspect of the right hind paw by using a Hamilton syringe (Hamilton Syringe Gastight™ serie 1700, TLL end, Merck) coupled to a 30-gauge needle. Nocifensive behavior (licking or biting of injected paw) was quanti ed in 5 min intervals during 60 min as previously described 32 . For the AMTB experiment, formalin was dissolved in saline.
2.12. Orchidectomy. Mice were anesthetized with a mixture of i.p. ketamine (75 mg/kg; Imalgene, 100 mg/ml, Boehringer Ingelheim, Ingelheim/Rhein, Germany) and xylazine (15 mg/kg, Merck) and a midline scrotal incision was made. The testes were exposed, and the vas deferens and testicular blood vessels were ligated with 2 tight knots of 6-0 black silk (8065195601, Alcon Cusi S.A., Barcelona, Spain). An incision was made between the 2 knots to remove testes and epididymis and the incision was closed with three additional square knots after ensuring haemostasis. Sham surgeries were performed similarly but the testicles were exposed and not ligated or removed. Subsequent nociceptive evaluations were conducted 3 weeks after surgeries.
2.14. Computational studies. Homology models of murine and human TRPM8 channel were designed considering the structure of the TRPM8 from Ficedula albicollis, determined by cryo-electron microscopy at 4.1 Å (Protein Data Bank code 6BPQ) (https://www.rcsb.org/). The sequence of murine TRPM8 (Uniprot Q8R455) or human TRPM8 (Uniprot Q7Z2W7) was completely modelled against the reference structure, following the standard protocol implemented by Yasara (version 20.12.24, http://www.yasara.org). Sequence alignments between murine or human TRPM8 and F. albicollis were performed with ClustalO from the European Bioinformatic Institute (EBI, https://www.ebi.ac.uk).
Blind docking experiments were carried out with AutoDock implemented in Yasara. WS12 (PubChem CID: 11266244), testosterone (PubChem CID: 6013), progesterone (PubChem CID: 5994) and estradiol (PubChem CID: 5757) structures were obtained from the National Center for Biotechnology Information (NCBI) PubChem database (https://pubchem.ncbi.nlm.nih.gov/). 800 docking runs with exible ligands were xed and results clustered around binding hot spots. By using the Assisted Model Building with Energy Re nement (AMBER03) force eld, a simulated annealing optimization of the complexes was performed, which moved the structure to a stable energy minimum. The best binding energy in each cluster was saved and solutions grouped according to putative TRPM8 binding sites.
Local docking experiments focused in the menthol binding pocket were also performed with the murine and human TRPM8 model. A total of 50 runs were set with the side chain of critical residues in the menthol binding pocket kept exible.

Statistical
Analyses. Time courses of nociceptive behavioural data conducted in male and female mice were analysed using 2-way repeated measures ANOVA with time as within-subjects factor and NTG treatment or genotype as between-subject factors. The time courses involving orchidectomized animals were analysed with 3-way repeated measures ANOVA, with time as within-subjects factor and either NTG and orchidectomy or genotype and testosterone as between-subject factors. Levene's test of equality of error variances and Mauchly's sphericity tests were used to assess normality of the data and Bonferroni post-hoc pairwise comparisons were subsequently conducted when appropriate. Three-way ANOVA was also used to analyse the data of WS12 experiments (time point, WS12, NTG) whereas a within-design was chosen to analyse the effects of the AMTB doses in wild-type males recovered from sensitization (Friedman's test followed by Benjamini adjustment). A 3-way ANOVA was also used to analyse AMTB effects on NTG-exposed orchiectomized animals (Time point, Testosterone, Genotype). The time-course data of the chemically-induced nocifensive behaviour was analysed with repeated unadjusted t-tests to avoid assumptions of similar variances for the rst and the second phases of the formalin test and posterior measurements. For the cellular studies, data normality was rst assessed with the D'Agostino-Pearson test. Comparisons of 2 groups were analysed accordingly with T-tests or Mann-Whitney-U tests. Comparisons of more than 2 groups were analysed with either One-way ANOVA followed by Bonferroni or  Figure 1) and, therefore, it was discarded to prevent misinterpretation. Alternatively, a vehicle composed of 5% dextrose and 0.105% propylene glycol in water did not alter the animal nociception and was selected. We injected 10 mg/kg NTG or vehicle i.p. every other day during 8 days to male and female mice, and mechanical sensitivity was assessed before and after each treatment. Measurements were extended up to 12 additional days after the last NTG injection ( Figure 1A). Two hours after each NTG treatment, acute hypersensitivity was observed in male and female mice, whereas the vehicle did not induce nociceptive sensitization ( Figures 1B and 1C, left panels, P<0.001 treatment effect). Mechanical sensitivity assessed before NTG injections and after the end of the repeated treatment (Up to 20 days after beginning of the treatments, left panels of Figures 1B and 1C) showed a long-lasting hypersensitivity in male mice (Days 2-16, P<0.05 vs baseline and vehicle, Figure  1B) that returned to baseline values 8 days after the last NTG injection (Day 18, Figure 1B, nonsigni cant vs. baseline or vs. vehicle). In marked contrast, female mice presented a persistent cutaneous hypernociception that was signi cant until the end of the experimental procedure (P<0.05 vs. baseline, vs. vehicle, Figure 1C). Thus, these data indicate a sexual dimorphism as males can fully recover from NTG sensitization, while females improve partially.
2.2. Repeated NTG treatment induces long-lasting TRPA1-dependent hypersensitivity in female and male mice.
Migraine-related pain produced by acute NTG treatment has been associated to TRPA1 activity in trigeminal ganglia of male mice 3 . We investigated the possible participation of TRPA1 on the model of chronic migraine. Wild-type and TRPA1 knockout mice received NTG injections and mechanical sensitivity was assessed (Figure 2A-B). As expected, wild-type mice showed acute hypersensitivity 2 hours after each NTG injection. On the contrary, this sensitization was absent in TRPA1 knockout mice of both sexes (Figure 2A-B, P<0.001 genotype effect, left panels), in agreement with a previous study describing lack of acute NTG sensitization in TRPA1 knockout males 3 . TRPA1 deletion also prevented the development of chronic hypersensitivity in males and females (Figure 2A-B, right panels).
To understand the role of TRPA1 in mediating NTG-induced sensitization in mice, we assessed TRPA1 mRNA expression and calcium imaging activity in trigeminal ganglia of male and female mice chronically exposed to NTG ( Figures 2C and D). Samples from NTG-treated wild-type animals showed an increased TRPA1 mRNA expression regardless of the sex ( Figure 2C, P<0.001 vs. vehicle). In parallel, trigeminal neuronal cultures of male and female mice revealed calcium transients in response to the TRPA1 agonist allyl isothiocyanate (AITC) at 70 µM ( Figure 2D). These responses were of similar magnitude in samples of both sexes and had higher intensity and duration in neurons of mice chronically exposed to NTG (P<0.01 vs. vehicle, Figure 2E). The neuronal population responding to AITC was composed of small to medium size neurons (100 to 700 µm 2 , Suppl. Fig. 2A). This increased TRPA1 responsiveness was also characterized by a higher percentage of cells showing signi cant activity (percentage of KCl-sensitive cells, P<0.001 vs. vehicle, Suppl. Fig.2B), and the size of these responses was proportional to the percentage of sensitive cells (Suppl. Fig. 2C). Thus, the trigeminal cultures revealed small to medium size neurons with increased TRPA1 activity after chronic NTG treatment, regardless of the sex.
To further investigate TRPA1 involvement on NTG sensitization, trigeminal neurons of naïve wild-type and TRPA1 knockout mice were cultured to assess calcium responses. These neurons were challenged with 100 µM NTG followed by 70 µM AITC ( Figure 2F). A 15.7±3.2% of the cells responded to both stimuli in wild-type mice, while this percentage decreased drastically in TRPA1 knockouts (0.6±0.3%, P<0.001, Figure 2G), revealing that NTG sensitized through TRPA1 activation in mice. To assess the translatability of this NTG activity to human TRPA1, NTG was applied to human TRPA1-transfected HEK293 cells and IMR90 broblast cells natively expressing this receptor ( Figure 2H). Both cell types showed a dosedependent relationship for NTG-evoked calcium in ux, whereas control HEK293 cells did not respond. Thus, NTG activated both human and murine TRPA1 channels.
TRPA1-mediated trigeminal sensitization may involve the release of the vasodilator peptide aCGRP 6,8 .
This neuropeptide is an essential neurotransmitter for migraine neuroin ammation and pain 34 . Hence, we next examined NTG-induced aCGRP release in cultured trigeminal neurons through aCGRP immuno uorescence. A treatment with DD04107, an exocytosis inhibitor that interacts with the exocytosis-related protein SNAP-25 35 was used to investigate vesicular release. Control neurons exposed to NTG vehicle showed stronger aCGRP immunoreactivity than NTG-treated neurons ( Figure 2I-J, P<0.001) suggesting vesicular aCGRP release after NTG. In contrast, neurons pre-treated with DD04107 showed signi cantly higher aCGRP immunoreactivity ( Figure 2J, P<0.01 vs. NTG+vehicle), indicating inhibition of the neuropeptide release after NTG. Collectively, these data indicate that TRPA1 activity was essential for acute and chronic NTG hypersensitivity, although it could not explain the observed sexual dimorphism.
2.3. TRPM8 activity determines the recovery of normal sensitivity in male mice exposed to the model of chronic migraine.
TRPM8 is a thermoTRP channel that has been signalled in the pathophysiology of chronic migraine. Thus, we next investigated the involvement of TRPM8 in NTG chronic sensitization. For this purpose, male and female wild-type and TRPM8 knockout mice were exposed to the chronic NTG treatment ( Figure  3A-B). Wild-type males presented the expected sensitization ( Figure 3A, P<0.01 vs. Baseline on days 8 and 15) that was resolved by the end of the experimental procedure (day 20, Figure 3A). Noteworthy, TRPM8 knockout males maintained a persistent sensitization that was signi cant until the last day of measurements, akin to the persistent sensitization observed in females (P<0.01 vs. Baseline and wildtype, Right panel of Figures 3A and 3B). This nding revealed a protective function of TRPM8 in males subjected to NTG-mediated chronic sensitization and signalled a potential role of this channel in sex dimorphism in the model of chronic migraine.
To investigate the involvement of TRPM8 in sex dimorphism, we next analysed its expression in trigeminal ganglia of male and female wild-type mice chronically exposed to NTG. TRPM8 mRNA values were similar regardless of treatment and sex ( Figure 3C). Functionally, perfusion with the selective and potent TRPM8 agonist WS12 (500 nM) elicited calcium transients of similar morphology and size in both sexes, after chronic vehicle or NTG (Figures 3D, 3E). WS12 activated 8 ± 2.1% of the cultured neurons (Supplementary Figure 3A), which were small size cells (50-400 µm 2 , Supplementary Figures 3B-C). This percentage was similar in samples of vehicle and NTG-treated mice. Hence, trigeminal cultures showed that TRPM8 activity and expression were similar between sexes and after the NTG treatment.
Given this similar functionality, we conducted additional experiments to test the antinociceptive e cacy of WS12 in wild-type female mice sensitized after chronic NTG. On day 20, females received WS12 or its vehicle. The observed responses were highly variable and nonsigni cant results were found after 10 mg/kg WS12 (i.p.), a dose with reported antinociceptive e cacy in male mice 36 ( Figure 3F). In a separate experiment, we tested the effect of lower (5 mg/kg, Supplementary Figure 4A) and higher doses of WS12 (20 mg/kg, Supplementary Figure 4B) in wild-type and TRPM8 knockout females. An antinociceptive trend was observed with the highest dose (Supplementary Figure 4B, P=0.053 vs. vehicle), whereas the lowest dose was completely ineffective. WS-12 also showed e cacy when administered in a vehicle containing 5% ethanol and 45% cyclodextrin, however possible antinociceptive effects of this vehicle were detected (Supplementary Figure 4C). To further characterize the in vivo activity of WS12 in females, we decided to examine the pain-relieving effect of TRPM8 activity on the formalin pain model, another TRPA1-dependent pain model 37 . As illustrated in Figure 3G, WS12 induced signi cant antinociception in the acute phase, whereas the late phase was unaltered, revealing a short-lasting effect of this compound.
To further investigate the function of TRPM8 in wild-type males, mice previously exposed to NTG that had recovered their baseline sensitivity were exposed to increasing doses of the TRPM8 blocker AMTB ( Figure 3G). This compound precipitated a signi cant re-sensitization when administered at 10 or 15 mg/kg (P<0.05 vs. vehicle, Figure 3H) in this model. Interestingly, similar results were obtained in male mice previously subjected to the formalin test. Thus, male mice treated with AMTB after the extinction of formalin-induced nocifensive activity showed a signi cant reinstatement of licking behaviour evident 30 min after administration of the compound (P<0.05 vs. vehicle, Figure 3I). Altogether, these results suggest the presence of endogenous TRPM8 activity with pain-relieving function in male mice.
2.4. Testosterone activates TRPM8 to resolve mechanical sensitization in male mice exposed to the model of chronic migraine.
Testosterone has been suggested as an endogenous TRPM8 agonist 22 and shows antinociceptive functions in mice 23 . We hypothesized that this endogenous androgen could have a protective function in the mouse model of chronic migraine. To address this question, mice were rst subjected to a sham surgery or to an orchidectomy to deplete gonadal testosterone. Once their nociceptive sensitivity was restored the animals received the NTG treatment ( Figure 4A). Acute NTG produced similar hypernociception in sham and orchidectomized animals ( Figure 4A, left panel, P<0.001 vs. vehicle). In contrast, the repeated NTG evidenced a persistent chronic sensitization selectively expressed in orchidectomized animals ( Figure 4A, right panel, P<0.01 vs. sham), indicating possible protective role of testosterone.
To investigate whether testosterone could have TRPM8-mediated antinociceptive effect in males, wildtype and TRPM8 knockout males were orchidectomized and received subcutaneous osmotic pumps lled with testosterone or its vehicle (cyclodextrin 45% in water). Afterwards, all mice were chronically treated with NTG, and their mechanical sensitivity was assessed. Testosterone induced complete recovery of mechanical thresholds in wild-type mice ( Figure 4B, right panel, day 20, P<0.001 vs. Vehicle wild-type, nonsigni cant vs. baseline), while vehicle-treated animals remained sensitized by the end of the experiment ( Figure 4B, right pane, P<0.001 vs. baseline on day 20). Remarkably, TRPM8 knockout mice lacked this restorative effect of testosterone ( Figure 4B, right panel, day 20, P<0.001 vs. wild-type, P<0.001 vs. baseline), although an antinociceptive effect independent of TRPM8 activity was also evidenced in knockouts ( Figure 4B, right panel, day 20, P<0.05 vs. vehicle TRPM8 knockout). To further clarify the involvement TRPM8 in the protective effect of testosterone, AMTB was administered to all mice after the nociceptive measurement on day 20 ( Figure 4C). A signi cant drop in the mechanical thresholds was selectively observed in wild-type animals treated with testosterone ( Figure 4C, P<0.01 vs. values before AMTB), whereas mice receiving vehicle and knockouts showed unaltered mechanosensitivity. These results revealed a prominent testosterone-TRPM8 antinociception in males.
Next, to determine whether testosterone could have rapid antinociceptive effects we treated NTG-exposed wild-type and TRPM8 knockout females with a single subcutaneous administration of 1 mg/kg testosterone or its vehicle (2.5% DMSO in corn oil, Figure 4D). A signi cant testosterone-induced antinociception was observed selectively in wild-type mice (P<0.05 vs. pre-treatment values, P<0.05 vs. testosterone-treated TRPM8 knockouts), whereas TRPM8 knockout females showed a trend for a hypersensitivity. Overall, the present behavioural results suggest nongenomic antinociceptive effects of testosterone through TRPM8.
We next investigated if testosterone activated TRPM8 channels in trigeminal cultures as a mechanism to account for the protective role of the androgen. Trigeminal cultures of wild type mice showed calcium transients in response to 10 pM testosterone ( Figure 4E). Cells responding to testosterone also presented calcium transients in response to 500 nM WS12 (7.9± 0.8 % of KCl-sensitive cells; Figure 4E). These responses were abolished in neural cultures from TRPM8 knockout animals tested side-by-side ( Figures   4E, P<0.001 vs. wild-type), revealing a testosterone activity through TRPM8 in trigeminal neurons.
Previous studies described structural and functional differences between murine and human TRPM8 38 .
To evaluate this possibility, we assessed testosterone activity on HEK293 cells constitutively expressing murine or human TRPM8 ( Figure 4F). Similar calcium transients were elicited after testosterone 10 pM in cells expressing human or murine TRPM8, whereas control HEK293 cells lacked this response ( Figure   4F). Altogether, the present data reveal a testosterone activity on murine and human TRPM8 that exerts antinociceptive restorative effects on the mouse model of chronic migraine.

Discussion
The most salient contribution of our study is the discovery of a novel role of TRPM8 as an androgen receptor that provides antinociceptive resilience and favours recovery in a mouse model of NTG-induced chronic migraine that produces similar acute mechanosensitivity in males and females but persistent hypersensitivity exclusively in females. Testosterone by activating TRPM8 channels drives a sexual dimorphism characterized by recovery of normal sensitivity in males. The lack of this protective mechanism in females leads to a persistent mechanical hypersensitivity. Noteworthy, our model of chronic migraine mimics the sexual dimorphism observed in humans, characterized by stronger transitions to chronic sensitization in women 24,39 and possible higher resilience in men. Previous studies also described higher female sensitivity in models of formalin-in ammatory pain 40 and models of persistent pain such as stress-induced visceral hypersensitivity or muscle pain after repeated saline injections and forced activity 23,40 Overall, these ndings suggest that the exposure to repeated noxious insults such as ongoing in ammation and stress-related stimuli favour the perpetuation of painful responses in females, whereas males show higher resilience and are able of reinstating their normal sensitivity.
In our model of chronic migraine, NTG provokes mechanical hypersensitivity by signalling through the TRPA1 channel, although this thermoTRP is not involved in the sex dimorphism observed. Indeed, repeated NTG exposure induced similar TRPA1 mRNA overexpression in trigeminal ganglia of male and female mice. Increased expression translated into exacerbated TRPA1 activity and trigeminal cultures of males and females chronically exposed to NTG presented stronger and longer-lasting TRPA1 activity in response to the speci c agonist AITC. In line with our results, previous research described TRPA1-speci c neuronal responses after acute NTG 3 , and male rodents exposed to NTG were sensitive to TRPA1 antagonism 41  The NTG-TRPA1 mechanism involved in promoting the mechanical hypersensitivity relies on the release for aCGRP from trigeminal neurons 6,8 , in agreement with the critical role of this neuropeptide in the aetiology of chronic migraine 9 . Our data further substantiate this tenet as sensitivity to the exocytosis inhibitor DD04107 revealed a mechanism involving large dense core vesicles and vesicular fusion protein SNAP-25 35 . Notably, the data obtained with DD04107 suggest a potential e cacy of this peptide for the treatment of chronic migraine through inhibition of aCGRP release, similar to botulinum neurotoxin 10 . Accordingly, we propose antagonistic roles of TRPA1 and TRPM8 activities determined by the continuous exposure to exogenous and endogenous agonists. Namely, nitric oxide derived from the treatment with NTG elicits a hyperalgesic state via TRPA1 stimulation both in males and females, whereas testosterone exerts an antinociceptive role through high-a nity interactions with TRPM8.
Notably, our results in TRPM8 knockout mice subjected to chronic NTG sensitization reveal a protective function of TRPM8. In line with our data, TRPM8 stimulation showed e cacy alleviating thermal hyperalgesia and nocifensive behaviours in models of headache-related pain 45,46 . A protective function was also described in models of noxious heat or chemically-induced pain such as the injection of capsaicin or the TRPA1 agonist acrolein 18,28 . TRPM8 stimulation also provided alleviation of mechanical and cold sensitivity in models of chronic neuropathic pain 47,48 . In accordance with these preclinical ndings, TRPM8 agonists have been used medicinally for alleviating a variety of pain conditions including migraine 13,14,49,50 , although using these compounds is not a rst-line treatment for this clinical condition 10 . Contradictory results were also published showing pro-algesic effects of TRPM8 agonists on migraine, although the use of TRPM8 agonists with partial effect on TRPA1 such as icilin could have yielded misleading results 51 . Similarly, we only observed an antinociceptive trend after administration of high WS12 doses in female mice treated with NTG. The absence of a prominent antinociception after WS12 could be associated to the potent and short-lasting effect of this compound, as suggested by the results obtained in the formalin test where signi cant effects lasted only 5 min in the acute phase. In agreement, calcium-dependent desensitization of TRPM8 is described after application of the canonical TRPM8 agonists menthol, icilin or WS12 15,52 . In our von Frey experiments, reliable assessment of mechanosensitivity requires spaced and sequential application of von Frey laments during periods of 15-30 min. Hence, mechanosensitivity could not be precisely assessed at that speci c temporal resolution and a transient effect of WS12 could have been overlooked. Short-lasting pain-relieving responses to menthol and its derivatives can be found clinically and were described elsewhere 49 .
The protective role of TRPM8 was rati ed by the ATMB-induced resensitization of males already recovered from the chronic NTG treatment. This protective TRPM8 activity was corroborated in males previously subjected to the classical formalin test and is compatible with a latent pain sensitization masked by tonic TRPM8 activity. The development of latent pain sensitizations or hyperalgesic priming has been related to the establishment of chronic pain conditions [53][54][55] and an endogenous opioid tone has been associated with this tonic pain inhibition 54 . In line with studies on acute pain models demonstrating involvement of opioidergic activity in TRPM8-induced antinociception 36,56 , our results suggest the interest of investigating opioid-TRPM8 interactions for the promotion of pain resilience.
Noteworthy, testosterone application provided rapid TRPM8-mediated antinociception in females. In line with this nding, orchidectomized males showed persistent sensitization after chronic NTG similar to females. Previous studies also described decreased ability of orchidectomized males in restoring normal mechanosensitivity after in ammatory or stress-related insults 23,40,57 . These studies elucidated testosterone antinociceptive mechanisms including transformation to di-hydrotestosterone and binding to androgen receptor 40 , down-regulation of anti-opioid neurotransmitter Brain-derived Neurotrophic Factor (BDNF) 57 or modulation of serotonin transporters 23 . In our study, orchidectomized wild-type males exposed to testosterone recovered their normal sensitivity after cessation of the NTG treatment, and this effect was largely dependent on TRPM8. These data reveal a novel function of TRPM8 providing endogenous pain relief in males though testosterone stimulation. Male-speci c alterations have been previously observed in TRPM8 knockouts, including delayed cold acclimation and lower bone mineral density, similar to females of either genotype that showed this same phenotype 58 . Interestingly, these features are also tightly linked to the activity of sexual steroids 59 . We observed rapid antinociceptive effects of testosterone in NTG-exposed females, in a time frame in which acute testosterone anxiolytic effects are also found 60 . In agreement, preclinical works show pain-relieving testosterone e cacy in male and female rodent models of acute and chronic pain 23,40,57 22 . These results and the behavioural data strongly correlate with computational docking studies revealing higher a nity for testosterone-TRPM8 interactions when compared to WS12-TRPM8 binding ( Figure 5AB and Supplementary Figure 5AB). Interestingly, the results obtained both in murine and human TRPM8 models indicate that testosterone most likely binds to the active pocket described for WS12 and menthol 52,66 . Interestingly, female hormones such as estradiol or progesterone show lower TRPM8 a nity than testosterone (Figures 5CD and Supplementary Figure 5CD).
In this line TRPM3, another TRP channel of the melastatin family, has been established as a thermoreceptor 67 that also responds to sexual hormones such as pregnenolone and progesterone 68 , suggesting a dual role of these protein types as detectors of both exogenous and endogenous stimuli.
Our data suggest a direct testosterone-TRPM8 interaction that could be helpful for the design of novel compounds mimicking this channel agonist activity without the unwanted effects of hormonal treatments. Additional less popular docking solutions include regions without described impact on TRPM8 functionality, and mutagenesis studies will be needed to clarify the speci c role of these interactions.
In conclusion, a high sensitization level of females has been found in a model that reproduces the sexual dimorphism and the enhanced cutaneous sensitivity associated with chronic migraine in humans 5 . This type of chronic sensitization could be aggravated after recurrent exposure to environmental factors involving TRPA1 stimulation such as pungent substances, oxidative stress or cyclic proin ammatory events 69 . The present data reveals that the difference between males and females is mainly due to an increased nociceptive resilience of males. This effect relies partly on TRPM8 activity in response to endogenous testosterone, an androgenic hormone that is present at much higher levels in male individuals. Hence, novel molecules mimicking testosterone activity on TRPM8 could lack the unwanted effects of hormonal treatments and may provide effective pain relief for individuals with low testosterone levels of any gender. The sexual dimorphism described here is based on a restorative natural process selectively expressed in males and encourages further investigation on pathophysiological processes promoting resilience. In addition, our data endorse the interest of analgesic drugs inhibiting TRPA1 activity for the treatment or prevention of migraine and pain syndromes characterized by enhanced mechanosensitivity in males and females 43,70,71 , although careful modulation may be desirable since TRPA1 shows neuroprotective effects as a hypoxia sensor that drives vasodilation and reduces ischemic  TRPA1-related activity is essential for the development of generalized mechanical hypersensitivity after nitroglycerin treatment and TRPA1 expression and function are increased in trigeminal ganglia of males and females chronically exposed to nitroglycerin. A, B Deletion of TRPA1 prevents the development of acute and chronic mechanical hypersensitivity after nitroglycerin in male (A) and female (B) mice. C Trigeminal ganglia of male and female mice chronically exposed to nitroglycerin (red) show increased TRPA1 mRNA expression when compared to vehicle-treated mice (black perfusion. I Trigeminal culture exposed to vehicle, nitroglycerin or the exocytosis inhibitor DD04107. Neurons labelled with pan-neuronal marker antiMAP2 in green, antiCGRP in red and nuclei marked blue with DAPI. J CGRP immunoreactivity decreases in cells exposed to nitroglycerin for 30 min and this  TRPM8 allows recovery of normal sensitivity in male mice exposed to nitroglycerin or formalin. A Wildtype and TRPM8 knockout males display similar acute mechanical hypersensitivity after nitroglycerin treatment (left panel), but the lack of this receptor in prevents the recovery of baseline mechanical sensitivity after the chronic nitroglycerin treatment (right panel) B Acute and long-lasting mechanical hypersensitivity after nitroglycerin treatment is similar in wild-type and TRPM8 knockout female mice. C TRPM8 expression in trigeminal ganglia of mice chronically exposed to nitroglycerin or vehicle is similar regardless of the sex or the treatment. D Calcium responses of trigeminal cultures from males and females chronically exposed to nitroglycerin or vehicle when challenged with the selective TRPM8 agonist WS12 (500 nM). E The size of calcium transient currents in response to WS12 is similar in neurons from males and females chronically treated with nitroglycerin or vehicle. F Female mice chronically exposed to nitroglycerin show increased mechanical sensitivity when compared to vehicletreated mice. This sensitivity could not be signi cantly alleviated after treatment with WS12 10 mg/kg i.p.
G Nocifensive behavior in the acute phase of the formalin test (5 min) is signi cantly alleviated when females receive formalin co-injected with 6 nmol WS12. H Administration of speci c TRPM8 antagonist AMTB (i.p.) to mice chronically exposed to nitroglycerin that already recovered their basal sensitivity acute hypersensitivity after the rst dose of nitroglycerin, however orchidectomized TRPM8 knockout mice develop stronger sensitization after the last nitroglycerin dose, regardless of the hormonal treatment. Right panel, mice receiving testosterone supplementation show inhibition of the long-lasting hypersensitivity induced after chronic nitroglycerin treatment. This attenuation leads to recovery of normal sensitivity in wild-type mice, but not in TRPM8 knockout mice. C Testosterone-supplemented wildtype mice that recovered their normal sensitivity reinstate their pain sensitization after administration of TRPM8 blocker AMTB on day 21. This response is absent in control and TRPM8 knockout mice. D