Migraine-relevant sex-dependent activation of mouse meningeal afferents by TRPM3 agonists

Experiments were performed in 10- to 13-week-old male and female wildtype (WT) C57BL/6 J mice. Mice were bred in the Animal Facility of the University of Eastern Finland (UEF) and housed in special cages in rooms with controlled temperature 22 °C, humidity, and a 12-h light/dark cycle. Food and water were provided ad libitum. All experimental procedures were performed following the ethical guidelines of the European Community Council Directive of 22 September 2010 (2010/63/EEC). The study protocol was approved by the Animal Care and Committee of the University of Eastern Finland (licence EKS-008-2019, protocol from 25 November 2019). All measures were taken to minimize animal suffering in accordance with ARRIVE guidelines.

Isolated mouse hemiskull preparations for direct spike recordings from TG nerve endings were prepared as previously described [13, 23]. In brief, after CO₂ inhalation and checking for lack of a pedal withdrawal reflex, mice were sacrificed by cervical dislocation followed by decapitation. Subsequent cleaning procedures were carried out for 15–20 min in oxygenated artificial cerebrospinal fluid (aCSF), containing (in mM): 120 NaCl, 2.5 KCl, 2 CaCl₂, 1 MgCl₂, 11 glucose, 24 NaHPO₄ and 30 NaHCO₃, bubbled with 95% O₂/5% CO₂ at room temperature (RT), while pH was maintained at 7.25–7.35. Skin and cranial muscles were removed from the outer side of the skull, which was then divided into two parts along the sagittal line using a scissors. To provide access for the recording electrode to the meningeal nerves branching out from the TG ganglia, the brain was gently removed with a forceps without harming the TG ganglia. Special attention was paid to keep the dura mater with meningeal nerves and vessels on the bone tissue inside the hemiskull (virtually) intact. After that, the isolated hemiskull was placed in a recording chamber continuously perfused with aCSF (6–7 mL/min) and oxygenated with 95% O₂/5% CO₂ mixture.

Drugs were purchased from Tocris Bioscience, UK, i.e. TRPM3 channel agonists pregnenolone sulfate (PregS) and CIM0216, Piezo1 agonist Yoda1 and transient receptor potential vanilloid 1 (TRPV1) channel agonist capsaicin and dissolved in DMSO. Substances were diluted to a final concentration in aCSF immediately before usage and applied to the receptive field around the main meningeal branch of the TG nerve by fast perfusion (~ 6 mL/min).

To detect TRPM3 mechanosensitive channels in TG nerves of meningeal tissue, we employed direct electrophysiological spike recording of nociceptive firing from peripheral meningeal nerve terminals. In the preparation stage, after placing the hemiskull preparation into the recording chamber, the main meningeal branch of the TG nerve was cleaned from surrounding tissue and cut at a distance of ~ 0.5 mm from the TG ganglion. Next, a small incision was made in the dura mater, and the peripheral part of the cut meningeal branch was placed into a recording glass microelectrode filled with aCSF. The exposed tip of the microelectrode was adjusted to the nerve diameter to allow the nerve ending to entirely plug it. Next, a silver reference electrode was placed into the bath containing the hemiskull preparation. At the start of each experiment, to allow the nerve to be tightly sucked into the electrode and stabilize the baseline, 10 min of spontaneous spike activity was recorded as control. After baseline recording, 50 mM KCl with compensated osmolarity was applied to verify the neuronal activity of the preparation. Next, to demonstrate the presence of mechanosensitive fibres, mechanosensitive Piezo1 channels [12] were activated with Yoda 1 (5 μM) whereas TRPM3 mechanosensitive channels were activated with PregS (50 μM) or CIM0216 (5 μM). Finally, as a marker of non-mechanosensitive nociceptive neuronal activity mediated by TRPV1 receptors, capsaicin (1 μM) was applied. Note that all drug applications were made to the same peripheral terminal of the meningeal nerve and lasted for 10 min with subsequent perfusion for 20 min (washout). Because Yoda1, PregS, CIM0216 and capsaicin were all prepared in DMSO, a pre-application of vehicle solution containing the same concentration of DMSO was administered, which did not significantly affect spiking activity of the TG nerve.

Electrophysiological recordings of neuronal spiking activity generated in the peripheral part of the meningeal nerves were registered at RT using a low-noise digital amplifier (ISO-80; World Precision Instruments, Sarasota, FL, USA) with the following parameters: gain 10,000X and bandpass 300–3000 Hz. Obtained electrical signals were digitized at 8-μsec intervals using a NIPCI 6221 data acquisition board (National Instruments, Austin, TX, USA) and stored on a PC for offline analysis. Signals were visualized by WinEDR v.3.5.2 software (University of Strathclyde, Glasgow, UK) and analyzed with MATLAB-based software (MathWorks, Natick, MA, USA) [13].

Advanced cluster and spectral spike analysis was performed, as described previously [13, 24, 25]. In brief, before analysis, all original recordings were filtered at 100–9000 Hz using a Chebyshev type 2 filter for spike detection and subsequent cluster and spectral analysis. The baseline noise level was estimated at the beginning of each experiment in a spikeless interval lasting at least 20 s to determine the respective threshold for spike detection. A recording was considered to contain spikes when the potential amplitude exceeded 5 standard deviations (SD) of the level of baseline noise. Spike amplitudes were normalized by baseline noise and expressed as SD values (arbitrary units, a.u.). Two-phase signals with a duration in the range of 0.3–1.8 ms were considered to represent action potentials, for which the spike rate was presented as number of spikes per 10 s. Using MATLAB software, for each spike, we calculated parameters such as the rise and decay time, the amplitude of positive and negative phases, spike areas, and their total duration. The ‘KlustaKwik’ application [26] was used to automatically recognize the most compact groups of spikes (clusters). Amplitudes of positive and negative phases of spikes were used as input parameters for clusterization. This approach allowed us to separate the total flow of spikes into 7 to 30 individual clusters for each experiment. For spectral analysis, data were analyzed with respect to the number of interspike intervals (ISI) per second, both for the whole nerve and for spike activity of individual clusters.

To assess possible involvement of two types of mechanosensitive channels in the activity of meningeal TG nerve fibres in male and female mice, first 50 mM KCl was applied for 10 min to measure overall excitability of the preparation. Then, after washout of KCl, sequentially 5 μM Yoda1 (as Piezo1 agonist), 50 μM PregS (as TRPM3 agonist) and 1 μM capsaicin (as agonist of TRPV1 receptors, typically expressed in nociceptors) were tested in the same preparation. Examples of the firing activity of the nerve fibres before and after application of Yoda1 and PregS in male and female meningeal hemiskull preparations are shown in Fig. 1 A, B. Application of Yoda1 led to a moderate induction of persistent firing - compared to the spike activity during the 10 min control recording period directly prior to drug application (Fig. 1A and C) - with no difference between males (from 346.9 ± 73.9 prior to 1367 ± 297.3 during Yoda1) and females (from 370.2 ± 97.5 prior to 1507 ± 338.4 during Yoda1; p = 0.76, Mann–Whitney U test). In contrast, PregS application led to profoundly increased firing in females, namely from 515.1 ± 143.1 to 5084 ± 815.4 (n = 7, p = 0.01, Wilcoxon signed-rank test), while in males the observed increase in firing during PregS (from 343.2 ± 61.31 prior to 1503 ± 112.9) was not significant (n = 5, p = 0.06, Wilcoxon signed-rank test; Fig. 1 B-D). Thus, in females, PregS induced a much stronger increase in spiking activity than in males (p = 0.0025, Mann–Whitney U test; Fig. 1 C, D). The amplitude of action potentials in basal conditions did not show sex difference (7.6 ± 0.5 a.u. in females; n = 7 vs. 6.6 ± 0.4 a.u. in males; n = 5. p = 0.14; Mann–Whitney U test). However, there was a sex difference in amplitude of action potentials during PregS application which were larger in females (13.8 ± 1.2 a.u. in females; n = 7 vs. 8.9 ± 0.4 a.u. in males; n = 5, p = 0.01; Mann–Whitney U test).

The presence of two types of mechanosensitive channels in meningeal afferents, as detected during their sequential activation by Piezo1 and TRPM3 agonists, raises the issue whether or not the two channels interact, and if so, whether they promote or depress effects of the other. A functional calcium-dependent interaction has been presented for Piezo and TRPV1 channels [11]. As Piezo1 and TRPM3 mechanosensitive channels are both calcium-permeable [11, 27], one can envisage that they may also be able to provide sensitisation when activated after each other. To test for a possible interaction between Piezo1 and TRPM3 channels, the sequence of agonist application was swapped in female mouse meningeal preparations (Fig. 2 A, B). Spiking activity was calculated as the ratio between the number of spikes during the 10 min with the compound present, and the number of spikes during the 10-min period directly before compound application (i.e. control), the latter taken as 100%. When PregS was applied after Yoda1, firing induced by activation of TRPM3 channels by PregS increased to 1917 ± 1037% from the control level (n = 7; Fig. 2 C). Applying PregS before Yoda1 increased firing induced by activation of TRPM3 channels by PregS to 2339 ± 1221% from the control level (n = 5; Fig. 2 C), which was similar (p = 0.63, Mann–Whitney U test) to the alternate order of application. When Yoda1 was applied before PregS, the firing induced by activation of Piezo1 channels by Yoda1 increased to 415 ± 129.5% (n = 7; Fig. 2 D), while it increased to 466.8 ± 124.8% (n = 5; Fig. 2 D) when Yoda1 was applied after PregS. Thus, also for Yoda1 the spike frequencies were similar (p = 0.53, Mann–Whitney U test) regardless of the order of application of the agonists. This indicates a lack of interaction (including sensitization) between TRPM3 and Piezo1 channels when sequentially activated in female mouse meninges.

To confirm our finding that TRPM3 channels are expressed in meningeal afferents, we tested the potent, selective synthetic TRPM3 agonist, CIM0216 [27]. Following the KCl and Yoda1 application we now applied 5 μM CIM0216 instead of PregS. Figure 3 A, B show example traces of spiking activity in meningeal TG nerve fibres from male and female mice in the control condition, during application of 5 μM CIM0216, and after 1 μM capsaicin. Figure 3 C, D shows that 5 μM CIM0216 raised the number of spikes during a 2-min window from 115.8 ± 47.2 during the control condition to 638.3 ± 52.5 in males (n = 6, p = 0.03, Wilcoxon signed-rank test) and from 127.0 ± 50.6 to 1839.0 ± 259.4 in females (n = 6, p = 0.03, Wilcoxon signed-rank test). Hence, like PregS, CIM0216 induced profound firing activity in TG nerve terminals, especially in female mice (p = 0.0022, Mann–Whitney U test; Fig. 3 C, D). Note that following CIM0216 application, the response to capsaicin was strongly impaired (Fig. 3 A-C) in both males and females, as compared to that observed following PregS application (Fig. 1 C).

One key question is whether the observed sex difference in TG nociceptive firing is exclusively seen with TRPM3 channel activation or whether it is also observed for other nociceptive receptors and/or may relate to non-specific differences in neuronal excitability. To this end, we directly compared the general excitability features (as assessed with KCl application), effects of Piezo1 receptors (as assessed by Yoda1 application), and the capsaicin-induced firing of TG nerve terminals from the experiments from males vs. females. Figure 4 A, B shows the contribution of sex to the overall level of nociceptive neuronal firing by comparing the level of baseline firing to that after application of 50 mM KCl, for both sexes. Notably, the mean number of spikes during a 10-min window at baseline was similar (p = 0.78, Mann–Whitney U test) between males and females, i.e. 492.5 ± 158.9 (n = 11) and 472.5 ± 120.2 (n = 13), respectively. Following KCl application, the mean number of spikes within short burst of activity during the 30-s active phase was also similar (p = 0.29, Mann–Whitney U test) for both sexes, i.e. 241.4 ± 38.6 (n = 11) in males and 295.0 ± 37.5 (n = 13) in females. For the effect of Yoda1, the combined data (expressed as the mean number of spikes per 10 min) from all experiments during Yoda1 application revealed no sex difference for mechanosensitive Piezo1 channels (Fig. 4 C), with the mean number of spikes in males (1367.0 ± 297.3 (n = 11)) and females (1507.0 ± 338.4 (n = 13)) being similar (p = 0.94, Mann–Whitney U test). In addition, the mean number of spikes within the 2-min active phase of application of 1 μM capsaicin in males (166.8 ± 30 (n = 5)) and females (200.4 ± 32.9 (n = 7)) was also similar (p = 0.53, Mann–Whitney U test; Fig. 4 D). Notice that in the last comparison, data with 1 μM capsaicin were taken only from experiments with prior PregS application, as application of 5 μM CIM0216 had an impact on subsequent compound applications. Therefore, a lack of a sex difference for Piezo1 and TRPV1 channels indicates a specific role of sex in the regulation of mechanosensitive TRPM3 channels.

To assess whether the activation of mechanosensitive Piezo1 and TRPM3 channels varies for single nerve fibres or small groups of fibres (clusters), we used a clustering approach [13]. Figure 5 A shows examples of multiple-spike clusters of experiments from female and male mice under control conditions and during 50 μM PregS application, whereby each dot represents one spike and different colours indicate different spike clusters. The data reveal that, especially in females, the TRPM3 channel agonist PregS increased the number of spikes in each of the identified clusters and additionally ‘wakes up’ previously silent clusters that consisted of spikes with a relatively large positive and negative amplitude. One advantage of the clustering approach is that it is able to shed light on whether different nociceptors are co-expressed within the same fibre, i.e. one cluster. Figure 5 Ba-d and Ca-d show example recordings illustrating the activity of selected clusters, indicating that different combinations of the two mechanosensitive Piezo1 and TRPM3 channels may occur within one cluster (fibre), as well as revealing whether a fibre responds to activation of classical nociceptors by capsaicin or not. Our functional data indicate that the latter are typically expressed in fibres with small- and prolonged-amplitude of spikes (Fig. 5 Ba,c and Ca,c), a signature of C-fibres [28] that are typically responsive to capsaicin [13]. Based on the joint responsivity to Yoda1 and PregS, Piezo1 channels are co-expressed with TRPM3 in capsaicin-sensitive (Fig. 5 Ba and Ca) as well as capsaicin-insensitive fibres (Fig. 5 Bb and Cb). Notably, TRPM3 channels could be activated in most (98% in females vs. 80% in males) fibres (including fibres that were insensitive to Yoda1 and/or capsaicin, see Fig. 5 Bb-d or Cb-d). Notably, the application of PregS caused a massive and repetitive escalation of spike frequency in clusters that had been silent throughout the whole application in females (Fig. 5 Bd), while in males (Fig. 5 Cd) in the same type of clusters PregS induced short peaks in firing rate. Thus, the different shapes of recorded spikes correlated to different types of activated fibres, based on the distinct responses to Yoda1, PregS, and capsaicin. Figure 5 D and E summarizes the neurochemical response profile of different meningeal fibres (represented by the distinct spike clusters) in females and males. Interestingly, in females, up to 44% clusters (63 out of 143) were activated by PregS and Yoda1 (23%) or by PregS and Yoda1 as well as capsaicin (21%), whereas in males the number of such ‘supermechanosensitive’ fibres was only 24% (17 out of 70; including PregS and Yoda1 (7%); PregS, Yoda1 and capsaicin (17%)). In contrast, the profile of ‘only-PregS-sensitive’ (46% in females vs. 44% in males) or all capsaicin-sensitive (30% in females vs. 33% in males) fibres did not show sex-dependence. In summary, spike cluster analysis revealed a prevailing functional activity of mechanosensitive TRPM3 channels in individual meningeal nerve fibres in females and, typical for this sex, massive and repetitive escalation of spike frequency in all clusters following pharmacological activation of these channels.

Finally, we characterized the temporal patterns of spike activity in meningeal nerves induced by the TRPM3 channel agonists. Our spectral analysis showed that application of either CIM0216 or PregS remarkably increased nociceptive firing in the range of very brief interspike intervals (ISI), especially in females (examples of original traces in Fig. 6 A and B and averaged data in Fig. 6 C and D). Firing corresponding to the spiking activity in δ-range (1–4 Hz) was prevalent in the majority of fibres during control conditions with no difference between males (88 ± 5.9% of fibres, n = 6) and females (87.7 ± 5.6% of fibres, n = 6). In contrast, after application of 5 μM CIM0216, we observed additional spiking activity in the θ- and α-ranges (4–8 Hz and 8–13 Hz, respectively) in both sexes with a predominance in females, as well as β-range of spiking activity (13–20 Hz) only in females (Fig. 6 E, H). One can envisage that high-frequency firing induced by TRPM3 agonists is supposed to provide the temporal summation of nociceptive signalling at the level of second order nociceptive neurons in the brainstem [13, 25], thus facilitating nociceptive traffic to higher pain centers in the brain.