Cold receptor TRPM8 as a target for migraine-associated pain and affective comorbidities

Abstract

Background

Genetic variations in the Trpm8 gene that encodes the cold receptor TRPM8 have been linked to protection against polygenic migraine, a disabling condition primarily affecting women. Noteworthy, TRPM8 has been recently found in brain areas related to emotional processing, suggesting an unrecognized role in migraine comorbidities. Here, we use mouse behavioural models to investigate the role of Trpm8 in migraine-related phenotypes. Subsequently, we test the efficacy of rapamycin, a clinically relevant TRPM8 agonist, in these behavioural traits and in human induced pluripotent stem cell (iPSC)-derived sensory neurons.

Findings

We report that Trpm8 null mice exhibited impulsive and depressive-like behaviours, while also showing frequent pain-like facial expressions detected by an artificial intelligence algorithm. In a nitroglycerin-induced migraine model, Trpm8 knockout mice of both sexes developed anxiety and mechanical hypersensitivity, whereas wild-type females also displayed depressive-like phenotype and hypernociception. Notably, rapamycin alleviated pain-related behaviour through both TRPM8-dependent and independent mechanisms but lacked antidepressant activity, consistent with a peripheral action. The macrolide ionotropically activated TRPM8 signalling in human sensory neurons, emerging as a new candidate for intervention.

Significance

Together, our findings underscore the potential of TRPM8 for migraine relief and its involvement in affective comorbidities, emphasizing the importance of addressing emotional symptoms to improve clinical outcomes for migraine sufferers, especially in females.

Supplementary Information

The online version contains supplementary material available at 10.1186/s10194-025-02082-4.

Introduction

Chronic migraine is a painful and debilitating condition that disproportionately affects women. Beyond intense head pain, it is associated with substantial emotional impairments, including anxiety and depression, which are also more prevalent in women and may have a bidirectional relationship with migraine [1]. Additionally, chronic migraine often coexists with confounding conditions like medication overuse headache, which is linked to an altered reward system and heightened impulsivity [2].

Many migraine triggers have been identified including stress, light or sound stimuli, or menstruation, and while knowledge about pathophysiological mechanisms of migraine has improved in recent years, its pathogenesis remains poorly understood [1]. A few inheritable mendelian forms exist; however, from twin studies, migraine is estimated to have a heritability of 0.4 [3], constituting often a polygenic disorder. One of the genes most consistently associated with migraine is TRPM8. Thus, single-nucleotide polymorphisms (SNPs) upstream of the TRPM8 gene are associated with differences in migraine prevalence, particularly in women [4]. This gene encodes Transient Receptor Potential Subfamily M member 8 (TRPM8), a non-selective Ca²⁺ channel receptor initially described in peripheral sensory neurons as sensor of cold, responsive to menthol and its derivatives [5]. Interestingly, TRPM8 has been detected in brain regions kept at euthermic temperature [6] where its physiological role should be different. These include areas such as amygdala and prefrontal cortex which are involved in emotional processing, underscoring potential contributions of TRPM8 to affective dimensions of migraine. This has promoted the present study to explore a role of TRPM8 in common affective traits associated with migraine, using as proxies mouse models of anxiety and depression. In addition, recent identification of TRPM8 activity of classical macrolide compounds like rapamycin and other rapalogs [7, 8], has motivated the experiments presented here to explore new avenues for migraine therapy. These TRPM8 ligands with longer in vivo half-lives may offer benefits versus canonical ligands with short-lived effects such as menthol or icilin [9, 10].

Here, we identify emotional behavioural phenotypes in Trpm8 knockout mice, in naïve conditions and when exposed to a chronic migraine model precipitated by nitroglycerin. We assess rapamycin potential as a modulator of migraine-related behavioural impairments through TRPM8 and validate TRPM8 druggability in human sensory neurons derived from induced pluripotent stem cells (iPSC).

Results

An affective behavioural signature of Trpm8 knockout mice

Humans carrying TRPM8 SNPs related to migraine susceptibility [4] have such genetic feature during their entire life, hence we speculated that TRPM8-defective mice could serve to identify potential phenotypes altered by distinct channel expression. Thus, we subjected Trpm8 knockout mice to a battery of tasks to characterise nociceptive and affective behaviour. In agreement with previous literature [5, 8], naïve Trpm8 null mice showed reduced cold sensitivity (Supplementary Figure 1A). We also observed greater female sensitivity to cold, regardless of genotype (Supplementary Figure 1A). In contrast, wild-type and Trpm8 knockouts showed similar withdrawal responses to mechanical stimulation with von Frey filaments, indicating normal motor responses and mechanical sensitivity. After verification of the reduced cold sensitivity and integrity of motor withdrawal responses, we aimed to explore affective behavioural traits. First, defensive anxiety was assessed with the Marble Burying task, in which mice are exposed to a number of marbles in a cage with abundant bedding, which mice bury according to their anxious-like phenotype. Wild-type and Trpm8 knockout mice displayed similar phenotypes (Supplementary Figure 1B). Afterwards, we used the Novelty-Suppressed Feeding test to assess conflict anxiety (Fig. 1A). In this test, a hungry mouse is exposed to food in the centre of a novel environment. Mice with higher anxiety levels explore more the surroundings before deciding to bite the pellet in the centre of the arena. Interestingly, male and female Trpm8 knockouts spent consistently less time exploring the environment and bite the pellet earlier than their wild-type counterparts. This difference could not be attributed to enhanced hunger of Trpm8 knockouts (Supplementary Figure 1C); hence it could be interpreted either as lower anxiety-like behaviour or as higher impulsivity as previously described [11]. To clarify this, mice were exposed to another classical paradigm of conflict anxiety, the Elevated Plus Maze. In this test, mice choose between exploring the open arms of a plus-shaped maze or staying within the closed arms which offer more safety. Anxious-like animals spend longer time in the closed arms, whereas animals with anxiolytic phenotype are more adventurous and explore more frequently the open arms. In our experimental conditions, we observed similar behaviour among genotypes and sexes (Fig. 1A), hence we assumed that differences in Novelty-Suppressed Feeding were due to enhanced impulsivity.

Fig. 1.

Affective Behavioral Signature of Trpm8 knockout mice. A Affective behaviour of naïve mice. From left to right: First panel, Novelty Suppressed Feeding. Trpm8 knockout mice of either sex show shorter latency to bite the food pellet in the novel environment than wild-type mice (P < 0.05, genotype effect), a phenotype compatible with reduced anxiety-like behaviour or with stronger impulsivity. Second panel, Elevated Plus Maze. Wild-type and Trpm8 knockout mice of either sex show similar percentage of entries to the open arms of the elevated plus maze, indicating similar anxiety-like behaviour in both genotypes. Third panel, Porsolt Swim Test. Male and female Trpm8 knockouts show longer-lasting immobility than their wild-type counterparts in the Porsolt swim test (P < 0.05, genotype effect), indicative of depressive-like phenotype. Fourth panel, Facial Expression of Pain. Trpm8 knockout mice display more frequent facial expressions of pain than wild-type mice (P < 0.001). B Affective behaviour of mice subjected to the nitroglycerin model of chronic migraine. Left panel, Novelty Suppressed Feeding. Mice chronically exposed to nitroglycerin show a significant delay in the latency to bite the food pellet when compared to those exposed to vehicle (P < 0.05 treatment effect). This delay is more evident in Trpm8 knockouts of both sexes and in females, whereas wild-type males appear unaltered. Middle panel, Porsolt Swim Test. Male and female knockout mice show a robust increase in immobility in the Porsolt swim test, regardless of the pharmacological treatment (P < 0.001, genotype effect). In addition, wild-type females exposed to nitroglycerin show a significant increase in immobility (P < 0.05 vs. wild-type female vehicle) unlike wild-type males which remain stable, suggesting greater female susceptibility to this depressive-like phenotype induced by nitroglycerin. Right panel, Facial Expression of Pain. An overall higher facial expression of pain is kept in Trpm8 knockout mice (P < 0.01), but no significant effect of nitroglycerin is observed. *P < 0.05, **P < 0.01, ***P < 0.001 Genotype effect, ###P < 0.001, #P < 0.05 Treatment effect, A) 2-way ANOVA; B), 3-way ANOVA; all followed by Tukey. Dots indicate individual values, Error bars are SEM; SEM, Standard Error of the Mean; Trpm8 KO, Trpm8 knockout; Veh, vehicle; NTG, nitroglycerin

After assessment of anxiety-like behaviour, animals were subjected to the Porsolt Swim Test to evaluate depressive-like behaviour. In this test, mice are exposed to forced swimming in an unescapable environment and despair-like behaviour (lack of movement trying to escape) is quantified. Remarkably, this test revealed significant increase of immobility in male and female Trpm8 knockouts (Fig. 1A), suggesting protective effect of Trpm8 in this depressive trait. Finally, we also evaluated facial expressions of pain using a computerized neural network [12]. Surprisingly, naïve Trpm8 knockout mice of both sexes exhibited more facial expressions of pain at rest when compared to wild-type (Fig. 1A, last graph to the right). In summary, naïve Trpm8 knockouts displayed reduced cold sensitivity while keeping normal withdrawal responses to mechanical stimulation, but these mice also showed enhanced impulsivity, depressive-like behaviour and enhanced facial expressions of pain at rest, unaffected by sex.

Enhanced affective vulnerability of Trpm8-deficient mice exposed to the model of chronic migraine

We previously described enhanced mechanical sensitivity of female mice when compared to males after chronic treatment with nitroglycerin, a trigger of migraine-like pain in sensitive individuals [13]. In the same study, we also observed that males became as sensitive as females when they lacked TRPM8 [13], revealing a protective function in mechanical hypersensitivity. Here, we replicated these phenotypes (Supplementary Figure 2A) and further evaluated the affective phenotypes altered in naïve Trpm8 knockouts. Interestingly, in the Novelty-Suppressed Feeding test, mice exposed to nitroglycerin spent longer time before deciding to bite the pellets than mice exposed to vehicle (Fig. 1B P < 0.05 Treatment effect), suggesting enhancement of conflict anxiety. This was evident in Trpm8 knockouts of both sexes and in wild-type females, whereas wild-type males appeared largely unaffected (Fig. 1B). Increased latency was unrelated to estimated hunger status (Supplementary Figure 2B). In addition, we quantified depressive-like behaviour with the Porsolt Swim test (Fig. 1B). Notably, wild-type females exposed to nitroglycerin exhibited significant depressive-like phenotype (Fig. 1B), whereas wild-type males remained unaltered. In contrast, both male and female Trpm8 knockouts presented prominent despair-like behaviour regardless of vehicle or nitroglycerin exposures. The immobility of vehicle-treated Trpm8 knockouts was remarkable and enhanced when compared to previous data of naïve Trpm8 knockouts (Supplementary Figure 2C), suggesting an effect of the experimental paradigm on their depressive-like phenotype.

Regarding the facial expression of pain, the enhanced time displaying facial expression of pain observed in naïve Trpm8 knockouts persisted after nitroglycerin (Fig. 1B, graph on the right), although nitroglycerin itself did not significantly alter pain expressions (Fig. 1B). Hence, overall data indicate protective function of TRPM8, diminishing negative affect and preventing nitroglycerin-induced anxiety and depressive-like behaviour, traits commonly associated with migraine.

TRPM8-dependent and independent effects of Rapamycin on the murine model of chronic migraine

Following previous reports of rapamycin-induced antinociception in the chronic migraine model [14], wild-type and Trpm8 knockout female mice received daily low rapamycin doses (1 mg/kg) or vehicle during the chronic nitroglycerin treatment (Fig. 2A). We used female mice in these experiments because wild-type females -unlike males- developed both long-lasting mechanical hypersensitivity (Supplementary Figure 2A) and depressive-like behaviour associated to this migraine model (Fig. 1B), mimicking the heightened migraine sensitivity observed clinically in women. Rapamycin significantly alleviated nitroglycerin-induced hypersensitivity to mechanical stimulation in wild-type mice, remarkably 2 h after the last exposure (Fig. 2A, left and AUC). In contrast, this antinociception was absent in Trpm8 knockouts (Fig. 2A, left). Additionally, the delayed long-lasting sensitization after nitroglycerin treatment was reduced in wild-type mice treated with rapamycin when compared to rapamycin-treated Trpm8 knockouts (Fig. 2A, right). This evidenced rapamycin TRPM8-dependent antinociception in the migraine model.

Fig. 2.

Rapamycin effects on the murine model of chronic migraine. A Left, Nitroglycerin-exposed wild-type mice treated with low rapamycin doses (1 mg/kg, filled circles) show significant alleviation of mechanical sensitivity 2 h after last nitroglycerin exposure (Left time-course graph, day 8, P < 0.05 vs. wild-type receiving rapamycin vehicle, represented by empty circles; left bar graph showing AUCs, P < 0.05 vs. vehicle of rapamycin). This rapamycin effect is absent in Trpm8 knockout mice (vs. green AUCs P < 0.05). Right, Before each nitroglycerine exposure and after the chronic nitroglycerin treatment, wild-type mice treated with rapamycin show a significant alleviation of mechanical hypersensitivity when compared to Trpm8 knockout mice receiving the same treatment (right bar graph, rapa wild-type vs. rapa Trpm8 KO, P < 0.01 in AUCs; No significant diference in time course). B After nociceptive measurements in A), mice were subjected to the Porsolt swim test after one last additional dose of 1 mg/kg rapamycin, but no effect of this treatment was observed. C Nitroglycerin-exposed mice showed also unchanged depressive-like behavior after repeated treatment with high rapamycin doses (10 mg/kg per day during 4 days). Trpm8 knockouts preserved their depressive phenotype (P < 0.05 vs. wild-type). D Mice treated with high rapamycin doses showed a reduction in the facial expression of pain, regardless of their genotype (P < 0.05 Rapamycin treatment effect). NTG, nitroglycerin. KO, knockout. Rapa 1, 1 mg/kg rapamycin. Rapa 10, 10 mg/kg rapamycin. Veh, Vehicle. SEM, Standard Error of Mean. Error bars are SEM, dots are individual values, bars or dots with error bars are average values. *P < 0.05; **P < 0.01. Mixed-effects model followed by Tukey post-hoc for time-course. Kruskal–Wallis followed by Dunn’s for 3-group bar graphs, 2-way ANOVA followed by Tukey for 4-group bar graphs. Three-way ANOVA followed by Tukey for rapamycin effect on facial expression

Once the long-term evaluation of mechanical sensitivity ended, we assessed depressive-like behaviour after one additional injection of 1 mg/kg rapamycin (Fig. 2B). While depressive-like phenotypes persisted in Trpm8 knockout mice, rapamycin lacked effect at this low dose (Fig. 2B). To explore potential effects of higher doses, we followed a prior study describing rapamycin antidepressant effects, giving 10 mg/kg rapamycin once a day 4 consecutive days [15] to nitroglycerin-treated mice. However, depressive-like behaviour of wild-type mice persisted, while Trpm8 knockout mice kept their enhanced depressive trait (Fig. 2C). Hence, this rapamycin treatment was ineffective in reducing depressive-like behaviour associated with chronic migraine, most likely because poor brain distribution.

Finally, when we evaluated facial expressions of pain in nitrgolycerin-treated mice after exposure to the high-dose rapamycin (10 mg/kg), we observed that rapamcyin significantly reduced the percentage of time with facial expressions of pain regardless of their genotype (Fig. 2D). Hence, rapamycin induced an inhibition of facial expressions of pain independent of TRPM8. Overall, rapamycin exerted both TRPM8-dependent and independent pain-relieving effects, alleviating nociception and facial expressions indicative of ongoing pain, respectively.

Activity of canonical and novel TRPM8 ligands in human sensory neurons

Given the protective role of TRPM8 in the murine models, we aimed verification of rapamycin activity in human neurons expressing functional TRPM8. Thus, to strengthen the translational relevance of potential findings, we specifically assessed TRPM8 expression and agonist-evoked calcium responses in human iPSC-derived sensory neurons. This is in contrast with prior studies relying on rodent models or in HEK293 cells, which lack endogenous TRPM8 expression and may not fully recapitulate human neuronal physiology. These human neurons showed immunoreactivity to pan-neuronal marker TUJ1 (Tubulin β-III) and BRN3 A protein, which peripheral neurons express (Fig. 3A). An average of 28 ± 3% of TUJ1-positive neurons exhibited also TRPM8 immunolabelling, mainly distributed within the cell body (Fig. 3B, Supplementary Figure 3A-C).

Fig. 3.

Validation of TRPM8 agonists in human iPSC-derived peripheral sensory neurons. A Human iPSC-derived sensory neurons show immunoreactivity for TUJ1, a β-III tubulin characteristic from neurons (white), and for BRN3 A (green), a well-established marker of peripheral neurons. B Part of the human iPSC-derived peripheral neurons also express TRPM8 (green puncta). Nuclei are stained with DAPI (blue). C Fluo-4 calcium imaging conducted in human iPSC-derived sensory neurons show calcium transients in response to 1 µM WS12, a selective TRPM8 agonist. These calcium transients are sensitive to the specific TRPM8 antagonist AMTB hydrochloride (P < 0.01). D 10 pM testosterone, and E) 10 µM rapamycin also elicit AMTB-sensitive calcium transients in human peripheral sensory neurons (P < 0.05). Dots indicate individual values and bars are averages normalized to vehicle. Error bars are SEM and scale bars are of 20 µM. *P < 0.05, **P < 0.01, ***P < 0.001. One-way ANOVA followed by Tukey, or Mann–Whitney U

Functional TRPM8 activity was assessed by intracellular calcium imaging in response to the selective agonist WS12 (1 µM), with calcium transients elicited in 27% of KCl-responding cells and inhibited with the specific antagonist AMTB (10 µM) (Fig. 3C, Supplementary Figure 4A-C). Thus, potassium chloride (25 mM) elicited pronounced responses, confirming neuronal identity of the cells (Fig. 3C, Supplementary Figure 4A,C-E). We also tested testosterone (10 pM), which evoked modest AMTB-sensitive calcium transients also in 27% of the neurons (Fig. 3D, Supplementary Figure 4B,D), consistent with prior findings in murine neurons and HEK cells expressing TRPM8 [13]. Finally, rapamycin (10 µM [7, 8]) was also evaluated, revealing prominent AMTB-sensitive calcium transients in up to 33% of the neurons (Fig. 3E, Supplementary Figure 4B,E), which substantiate ionotropic activation of human TRPM8 by this macrolide.

Discussion

The salient contribution of this study is the pronounced influence of the Trpm8 gene on affective behavioural states relevant to migraine pathophysiology. Namely, TRPM8-defective mice display remarkable depressive-like behaviour, increased impulsivity and frequent facial expressions of pain at rest. When subjected to a chronic migraine model, Trpm8 knockouts are vulnerable to mechanical nociception [13], and exacerbate anxiety and depressive-like phenotypes. Accordingly, the immunosuppressant and senostatic drug rapamycin, also described as TRPM8 agonist [7, 8], induces TRPM8-dependent relief of mechanical hypersensitivity in mice subjected to the migraine paradigm. Interestingly, rapamycin also inhibits facial expressions of pain through TRPM8-independent mechanisms. Notably, akin to WS12 and testosterone, rapamycin also activates neuronal TRPM8 channels expressed in human sensory neurons, substantiating human TRPM8 engagement in the macrolide activating activity.

The cold hyposensitivity of Trpm8-deficient mice [5] is accompanied by a phenotype associated to enhanced impulsivity in the Novelty Suppressed Feeding test [11, 16]. The reduced latency to bite the food pellet in the novel scenario was interpreted as impulsive behaviour as described elsewhere [16] To distinguish this from reduced anxiety-like behaviour, we used the elevated plus maze, a canonical test to identify anxiety-like behaviour that revealed normal performance in Trpm8 knockout mice. Interestingly, impulsivity was subject of research repeatedly in the context of headache conditions [17] and it involves maladaptations of the reward system linked with medication overuse headache [2]. Such association between TRPM8 and impulsivity could offer mechanistic insight into this poorly understood headache condition.

Naïve Trpm8 knockouts show depressive-like phenotype in the Porsolt swim test. In agreement, menthol elicited dose–response inhibition in this despair-like behaviour [18] and it can modulate the brain reward system [19]. In line with a disrupted affective phenotype, naïve Trpm8 knockouts also consistently present more facial expressions of pain at rest than their wild-type counterparts. While full explanation to this unexpected phenotype requires additional experimentation, it is reasonable to expect discomfort in animals with reduced ability to control temperature oscillations [20]. Such a persistent discomfort may be reflected in facial expressions of pain at rest. Overall, TRPM8 appears essential to maintain normal affective phenotypes in naïve mice.

The chronic migraine model promotes behavioural changes in Trpm8 knockout mice. Knockout mice of both sexes and wild-type females exposed to nitroglycerin show prolonged latency to bite the pellet in the novelty suppressed feeding test, contrasting with the reduced latency observed in naïve knockout mice. While this opposite result might be interpreted as a normalization of the altered impulsive phenotype in the knockouts, the behavioural status of nitroglycerin-exposed mice, including facial expression of pain at rest, mechanical hypersensitivity, together with depressive-like phenotype, suggests that the extended latency likely reflects an impaired affective status manifested also as neophobia and/or conflict anxiety, which is more perceptible in knockouts of both sexes and wild-type females. Such alteration unveils protective roles of TRPM8 in this emotional trait. In line with this, TRPM8 modulates anxiety-like behaviour at several environmental temperatures [21] and in migraine patients, a TRPM8 SNP was linked to anxiety prevalence [22]. Interestingly, knockouts exposed to vehicle dramatically enhance despair-like behaviour when compared to naïve knockouts of previous experiments, suggesting a depressive phenotype exacerbated by the experimental paradigm itself. Nitroglycerin-treated wild-type females also develop enhanced depressive-like behaviour, parallel to their mechanical hypersensitivity. Such behavioural repertoire closely mimics the greater vulnerability for depression observed in women with migraine [1]. Indeed, bidirectional influence and significant genetic overlap were observed between migraine and depression [1]. Overall, the data suggests involvement of the TRPM8 gene in emotional resilience and the interest of exploring TRPM8 modulation for understanding these migraine comorbidities.

Repeated exposure to low rapamycin doses elicits TRPM8-dependent antinociception. The absence of TRPM8-dependent effects after the first doses suggests the needs of a minimum concentration to obtain effective antinociception. Since rapamycin half-life is of around 18 h in mice [10], it is likely that repeated dosing led to accumulation and significant antinociception by the last day of treatment. Alternatively, this delayed response could be related to TRPM8 signalling downstream of receptor activation [23]. Nevertheless, our results do not preclude TRPM8-independent antinociceptive mechanisms involving canonical targets of rapamycin, particularly mTOR pathway inhibition, autophagy or induction of cell cycle arrest in immune cells [24–26]. Although we did not evaluate the possible antinociceptive effect in Trpm8 knockout mice due to the lack of a Trpm8 knockout group treated with vehicle, the reduced antinociceptive effect of rapamycin observed in Trpm8 knockouts when compared to the wild-type group reveals an impact of the lack of TRPM8 activity on rapamycin-induced antinociception. Despite this antinociception, low or high rapamycin doses did not modify depressive-like phenotypes in our experimental conditions. Possible reasons include TRPM8-independent mechanisms, relatively short exposure to rapamycin, or poor blood–brain barrier permeability [10]. Nevertheless, rapamycin antidepressant effects were described in basic studies [15] and clinically in combination with ketamine [27] While the rapamycin dosing schedule may be further optimized in this model, the present study focused on elucidating potential effects of rapamycin elicited through TRPM8 activity. We propose future studies using rapalogs with higher solubility, brain penetrance and reduced off-target effects for such optimization efforts [7, 8]. The persistence of depressive phenotypes despite alleviation of mechanical hypernociception also suggests independence of the degree of mechanical sensitisation. Hence, sustained hypernociception may be consequence rather than cause of the affective behavioural status. In agreement, other chronic pain models in females showed emotional-like affectation independent of the relief of hypersensitivity [28, 29], and continuous stressful stimuli facilitated mechanical sensitization in females [30]. Overall, our results are compatible with the view that depressive status could be a key factor upstream to the pain sensitization in the behavioural representation of chronic migraine [31]. Rapamycin also inhibits facial expressions of pain in mice exposed to nitroglycerin, independently of TRPM8 signalling. Different mechanisms were proposed for rapamycin-induced antinociception, including mTOR inhibition [25], autophagy modulation [14] or normalization of TNF-α, IL1β and IL-6 levels [26]. In addition, previous reports described improvements in grimace scale face expressions after rapamycin [32], although opposite effects were also observed at high rapamycin doses (20 mg/kg [25]). Our data reveal improvement in facial expressions of pain after rapamycin 10 mg/kg independently of TRPM8, in agreement with the multiplicity of targets described for this compound [33].

TRPM8 immunoreactivity and functional expression are revealed in iPSC-derived human sensory neurons. Menthol responses were previously described in these cells [34], however menthol also evokes TRPM8-unrelated responses [8]. Our data confirms TRPM8-elicited calcium transients after the selective menthol derivative WS12 and sensitivity of those to the specific antagonist AMTB. In agreement with previous works in different cell types [13, 35] we also identify sensitivity of human neurons to testosterone through TRPM8, implying different cold sensitivity between humans of different sex. In addition, human sensory neurons clearly respond to rapamycin and this response is blocked by AMTB. Overall, these results validate previous studies in HEK cells [7, 8] and evidence neuronal responses of human TRPM8 to this naturally-occurring compound.

In summary, we report TRPM8 participation in emotional traits related to impulsivity, anxiety and depression, all of them common migraine comorbidities. In addition, the immunosuppressant and senostatic drug rapamycin is revealed as a primary compound with TRPM8-dependent and TRPM8-independent pain-relieving effects that can modulate the sensitivity of human primary afferent neurons and, therefore, increase the pharmacological armamentarium for chronic migraine intervention.

Methods

Animals

Adult male and female mice with a C57BL/6J background (Envigo, Horst, The Netherlands), wild-type or defective in Trpm8 [5] were bred in the animal facility at Universidad Miguel Hernández (UMH, Elche, Alicante, Spain) and placed in an isolated room in the same institution at least one week before starting the experimental procedures. 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 stress they experienced. Housing conditions were maintained at 21 ± 1°C and 55 ± 15% relative humidity in a controlled light/dark cycle (light on between 8:00a.m. and 8:00p.m.). Animals had free access to food and water except during manipulations and behavioural assessment. Behavioural tests were conducted in a progressive order, starting with the least stressful paradigm and moving to the most stressful, to minimize potential influences between tests when animals were exposed to more than one. The testing sequence began with nociception assessment, followed by the evaluation of anxiety-like behaviour, and concluded with the assessment of depressive-like phenotypes. All procedures were conducted with approval from the UMH Ethical Committee and the regional government (code: 2022 VSCPEA0078-2), adhering to European Community guidelines (2010/63/EU).

Model of chronic migraine

Mice received 10 mg/kg nitroglycerin (50 mg/50 mL Nitroglycerin, Bioindustria LIM, Novi Liguri, Italy) or its vehicle (5% dextrose and 0.105% propylene glycol in pure water) every other day for 9 days (five injections total), administered at 10 mL/kg intraperitoneally (i.p.), as previously described [13].

Drugs

(1R,2S,5R)−2-Isopropyl-N-(4-methoxyphenyl)−5-methylcyclohexanecarboxamide (WS12; 3040/50, Tocris, Bristol, UK), N-(3-aminopropyl)− 2-{[(3-methylphenyl) methyl] oxy}-N-(2-thienylmethyl) benzamide hydrochloride (AMTB; Tocris), testosterone (#T1500, Merck, Darmstadt, Germany) and rapamycin (#J62473.MF, ThermoFisher, Waltham, Massachusetts, United States) were used in cellular studies, all dissolved in 0.01% dimethyl sulfoxide (DMSO, Merck). Used concentrations were based on previous works showing cellular responses in calcium imaging and electrophysiology [7, 8, 13]. In behavioural experiments, rapamycin was administered dissolved in DMSO at 2 mL/kg as previously described [36], at a dose of 1 mg/kg i.p. for the inhibition of nitroglycerin-induced mechanical sensitisation [14], and at a dose of 10 mg/kg to explore inhibition of depressive-like behaviour [15].

Behavioural assessment

Nociception

Cold plate

Cold response latencies were assessed with a Cold/Hot plate test with a 16.5 x 16.5 cm arena set at 0°C (Bioseb 760,112, PanLab, Harvard Bioscience, Cornellà, Barcelona, Spain). Latencies to forepaw withdrawal and licking were recorded as raw latencies in seconds, and cut-off for animal responding was established at 90s.

Mechanical sensitivity

Punctate mechanical sensitivity to von Frey filament stimulation was quantified through the up–down paradigm, as previously reported [13]. Filaments equivalent to 0.04, 0.07, 0.16, 0.4, 0.6, 1 and 2 g were used, applying first the 0.4 g filament and increasing or decreasing the strength according to the response. Filaments were bent and held for 4–5 s against the plantar surface of the hind paws and clear paw withdrawal, shaking or licking were considered nociceptive-like responses. Four additional filaments were applied since the first change of response (from negative to positive or from positive to negative), and the sequence of the last six responses was used to calculate the withdrawal threshold.

Anxiety-like behaviour

Marble burying task

Mice were placed in clean translucid cages (42.5 × 27.5 x 30 cm) with 5 cm sawdust bedding overlaid by twenty-eight glass marbles distributed in a 4 × 7 arrangement. Mice were allowed to explore the cage for 30 min and the number of marbles buried (> 2/3 of the marble covered by the bedding) was counted.

Novelty-suppressed feeding

Briefly, animals were food-restricted for 24 h and placed in a 51 × 51 cm arena filled with 5 cm sawdust, with three food pellets placed in a 12 × 12 cm filter paper situated in the centre. The test ended either when the animal began chewing or when 10 min transpired. Immediately afterwards, animals were placed in their home cage and the amount of food consumed in 5 min was measured as a relative measure of hunger (mg of pellet consumed).

Elevated plus maze

Anxiety-like behaviour was evaluated with an elevated plus maze made consisting of four arms (27 x 6 cm), two open and two closed, set in cross from a neutral central square (5 × 5 cm) elevated 40 cm above the floor. Light intensity in the open and closed arms was 45 and 5 lx, respectively. Mice were placed in the central square facing one of the open arms and tested for 5 min. Percentage of entries into the open and closed arms was determined.

Depressive-like behaviour

The Porsolt Swim test was used to evaluate depressive-like behaviour [37]. Mice were placed for 6 min into transparent Plexiglass beakers (ENDOglassware, 2000 mL CBB020, Akralab SL., Alicante, Spain) filled with 1800 mL of water at 22 ± 0.2°C to a depth of 22 cm. Time of immobility was assessed afterwards for the last 4 min. Immobility was considered when the animal made no movements in order to escape (swimming, climbing walls). Water was changed between subjects and beakers cleaned.

Facial expressions of pain

An artificial intelligence tool, specifically a convolutional neural network trained to analyse video recordings of mouse faces was used to score facial expressions of pain in mice. The procedure has been previously described elsewhere [12]. Briefly, mice were placed individually in custom-made test compartments (50 x 120 x 60 mm) with black walls and a mesh-bottomed platform (0.5 cm² grid) elevated 1.1 m above the ground. Each compartment was arranged in arrays of four and positioned at the edge of the platform, with one wall open facing a high-resolution infrared video camera (1440 x 1024 pixels, Kuman RPi camera, USA). Cameras were equipped with two infrared light-emitting diodes and positioned 25 cm from the test compartments to encourage the mice to face the visual cliff and, consequently, the camera. Each camera could simultaneously record two mice and was controlled by a Raspberry Pi Zero single-board computer (Kubii, France). Recordings were stored on USB drives for later transfer and analysis. No experimenters were present during testing, except during the first 1–2 min when mice were being placed into the compartments. Video recordings lasted a minimum of 15 min, but only the 5 to 10 min window was analysed to exclude potential artifacts caused by the researcher’s presence at the beginning and to avoid sleep-related features beyond the 10 min (e.g., partially closed eyes). The neural network, based on Google’s InceptionV3 model, was trained with facial images of mice highlighting features such as the ears, eyes, cheeks, and nose. We included 245 to clearly exemplify"pain"from animals treated intraperitoneally with cyclophosphamide (300 mg/kg), and 300 images labelled as"no pain” from control mice [38]. After over 30,000 training iterations, the model was able to evaluate each frame from new video recordings, assigning a probability value between −1 (no pain) and 1 (pain). We considered that a facial expression denoted pain when the probability value was greater than 0.1. Scripts for DeepLabCut and InceptionV3 were written in Python (v3.5). Network training and video scoring were performed remotely on an Ubuntu Linux computer equipped with an NVIDIA 2080Ti GPU.

Generation of iPSC sensory neurons

We followed the protocol described previously [34] to obtain sensory neurons from human Pluripotent Stem Cells (hPSCs). Briefly, female human pluripotent stem cells (Healthy Control Human iPSC Line, Female, SCTi003-A, #200–0511, STEMCELL Technologies, Cambridge, UK, CB25 9 TL) were cultured under feeder-free conditions using Essential 8 (E8) medium (Thermo Fisher Scientific) on vitronectin-coated plates (Thermo Fisher Scientific). Cells were passaged using 0.5 mM EDTA in PBS without calcium or magnesium (Thermo Fisher Scientific) for 5 to 6 min to dissociate the hPSC colonies. Cells were maintained in a humidified 5% CO2 atmosphere at 37°C. For neuronal differentiation, hPSCs were plated at 1.5 × 10⁵cells/cm² on vitronectin-coated 6 well plates in E8 medium containing CEPT cocktail (50 nM Chroman 1, 5 μM Emricasan, Polyamine supplement (1:1000 dilution) and 0.7 μM Trans-ISRIB (#7991, BioTechne)) to improve viability. 24 h later cells were switched to Essential 6 (E6) medium (Thermo Fisher Scientific) containing 2 µM A83-01 (Transforming Growth Factor-β inhibitor) and 0.2 µM CHIR98014 (GSK-3β inhibitor and WNT signalling pathway activator). On day 3, cells were passaged to a single cell suspension with Accutase (STEMCELL Technologies) and seeded at 5.5 × 10⁶ cells/well in 6 well AggreWell 800 plates (STEMCELL Technologies). Cells were maintained in E6 medium containing 0.5 µM CHIR98014, 2 µM A83-01, 1 µM DBZ (γ-secretase inhibitor), and 25 nM PD173074 (FGFR inhibitor). CEPT cocktail was included for the initial 24 h during nocisphere formation. On day 14, resulting nocispheres were dissociated using a MACS EB dissociation kit following manufacturers guidelines (#130–096–348, Miltenyi Biotec) and plated on poly-L-lysine/laminin-coated dishes in DMEM/F12 medium, supplemented with N2 supplement, B27 supplement (w/o Vitamin A), 1 µM PD0332991 (CDK4/6 inhibitor) and neurotrophic factors (BDNF/GDNF/NGF/NT-3; 25 ng/mL each) as described before [34]. By day 28, BRN3 A⁺(POU4 F1) and Tuj-1⁺ (Tubulin βIII) nociceptor-like neurons were obtained.

Immunocytochemistry

Cultures were washed with 1X PBS (D8662, Merck) three times. Afterwards, cells were fixed with 4% paraformaldehyde (#28909, Thermo Fisher Scientific) for 20 min at room temperature. Permeabilization was achieved with 0.1% v/v Triton 100X (P8787, Merck) for 5 min and blocking with 5% bovine serum albumin (#A7906, Merck) for 30 min, both in 1X PBS. Cells were labelled with primary antibodies mouse anti-BRN3 A 1:100 (#MAB1585, Millipore Sigma, Merck KGaA, Darmstadt, Germany), rabbit anti-TUJ1 1:400 (#5568, tubulin beta-3, Cell Signalling Technology, London, UK) and/or as previously suggested [39] mouse anti-TRPM8 Clone OTI7 A11 1:100 (#TA811228S, Origene Technologies GMBH, Herford, Germany, generous gift from E. de la Peña, Instituto de Neurociencias, San Juan, Alicante, Spain) and incubated for 1 h at room temperature. Secondary antibodies Donkey anti-mouse-488 1:200 (#A-21202, Thermo Fisher Scientific) and Donkey anti-Rabbit-555 1:200 (#A-31572, Thermo Fisher Scientific) were incubated for 1 h at room temperature protected from light. Slides where mounted with ProLong Gold antifade reagent with 4′,6-diamidino-2-phenylindole (DAPI, #P36931, Thermo Fisher Scientific) and images acquired with a confocal microscope (LSM 880, ZEISS, Jena, Germany).

Calcium imaging

Fluo4-AM (F14201, Molecular Probes) was dissolved in DMSO at a concentration of 10 mM and used at 2 µM to load the cells for calcium imaging experiments. D28 iPSC sensory neurons were incubated with Fluo4-AM for 30 min at 37°C in standard extracellular solution (in mM: 140 NaCl, 3 KCl, 2.4 CaCl2, 1.3 MgCl2, 10 HEPES, and 5 glucose, adjusted to pH 7.4 with 1 M NaOH). Cells were washed three times with extracellular solution and equilibrated for 30 min prior to imaging. Fluorescence measurements were obtained on an LSM 880 confocal fluorescent microscope using a 20 × objective. Basal fluorescence was captured every 3 s over a 60 s period prior to application of stimulants with images captured for a further 3 min. Mean fluorescent intensity values were recorded per cell and normalised to pre-simulation. Data expressed as ΔF/F0 using the equation: ΔF/F0 = (F_max-F₀)/F₀.

Statistical analyses

Behavioural and Calcium Imaging data were analysed using GraphPad Prism9 (GraphPad Software Inc., USA). Sample Sizes were based in previous works evaluating similar behavioural paradigm or calcium transients [13, 16, 29, 36]. For the behavioural experiments in naïve mice, 2-way Analysis of Variance (ANOVA) was used, with factors “genotype”, “sex” and their interaction, followed by post-hoc Tukey recommended tests to compare between experimental groups. When mice were exposed to nitroglycerin, data were assessed with 3-way ANOVA, adding the factor “treatment” and the interactions. Time-courses were analysed with 3-way (Time, Treatment, Genotype) or 2-way (Group, Time) Repeated measures ANOVA followed by Tukey. Calcium Imaging data were analysed with One-way ANOVA (WS12 vs WS12 + AMTB vs. KCl) followed by Tukey or with Kruskal Wallis tests (Testosterone vs. Testosterone + AMTB; Rapamycin vs. Rapamycin + AMTB) followed by Dunn’s Multiple Comparison Tests. Differences were considered statistically significant when P < 0.05. Outliers (± 2SD from the mean) were excluded. Artwork was designed using GraphPad Prism, Excel and Power Point. Experimenters were blinded to the key factors being assessed in each behavioural paradigm. Specifically, these factors included genotype in experiments in naïve mice, nitroglycerin or vehicle treatment in the migraine model experiments, and rapamycin vs. vehicle in experiments assessing the effects of rapamycin. In experiments with three experimental groups, only the treatment condition for wild-type mice was blinded. Blinded treatments were allocated randomly through the Random tool of Excel. Raw data and statistical analyses are provided in Supplementary material 2.

Abbreviations

AMTB

N-(3-aminopropyl) − 2-{[(3-methylphenyl) methyl] oxy}-N-(2-thienylmethyl) benzamide hydrochloride

ANOVA

Analysis of Variance

CEPT cocktail

Chroman, Emricasan, Polyamine and Trans-ISRIB cocktail.

DAPI

4′,6-Diamidino-2-phenylindole

DMSO

Dimethyl sulfoxide

E6 medium

Essential 6 medium

E8 medium

Essential 8 medium

HEK cell

Human Embrionic Kidney cell

hPSCs

Human Pluripotent Stem Cells

iPSC

Induced Pluripotent Stem Cells

SNPs

Single-nucleotide polymorphisms

TRPM8

Transient Receptor Potential Melastatin 8

TRPM8

Transient Receptor Potential Melastatin 8 human gene

Trpm8

Transient Receptor Potential Melastatin 8 mouse gene

TUJ1

Tubulin β-III

WS12

(1R,2S,5R) − 2-Isopropyl-N-(4-methoxyphenyl) − 5-methylcyclohexanecarboxamide

Authors’ contributions

D.C. conducted behavioural assays, analysed the data, conceptualized and designed the study and experiments, coordinated in vivo and in vitro experiments and wrote and edited the first and subsequent drafts of the manuscript. E.P.C. performed cell culture and differentiation assays, executed and analysed calcium imaging and immunocytochemical assays, edited and revised manuscript. R.G.C. developed the neural network and recording device, trained neural network, assisted and trained D.C. in face data management, edited and revised manuscript. E.J.C. developed the neural network and recording device, trained neural network, assisted and trained D.C. in face data management, edited and revised manuscript. A.F.C. supervised and designed experiments, conceptualized the project, revised the manuscript and provided funding. A.F.M. supervised and designed experiments, conceptualized the project, wrote, revised and edited the manuscript and provided funding.

Funding

Projects “Sex dimorphism in migraine: thermoTRPs as hormonal and drug targets (GIOCONDA)” Grant number: PID2021-126423OB-C21, and “A pre-clinical human nociceptive in vitro model for investigating sexual dimorphism in chronic migraine and screening drug candidates (HEADaCHE)” Grant: RTI2018-097189_B-C21, Ministerio de Ciencia e Innovación– Agencia Estatal de Investigación co-funded with FEDER funds from EU “Una manera de hacer Europa”. EPC is supported by a Royal Society research grant (RG\R1\251011).

Data availability

Data is provided within the manuscript or supplementary information files.

Declarations

Ethics approval and consent to participate

All procedures were conducted with approval from the UMH Ethical Committee and the regional government (code: 2022 VSCPEA0078-2), adhering to European Community guidelines (2010/63/EU).

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.