Complex modulation of the cold receptor TRPM8 by volatile anaesthetics and its role in complications of general anaesthesia

Cellular membrane ion channels participate in all physiological processes ranging from electrogenesis and the activation of numerous functional response(s) to gene expression and the determination of cell fate. They are also important determinants of numerous pathological states and are the targets of therapeutic, as well as adverse actions, of clinically used pharmacological agents. Volatile general anaesthetics (VAs) induce safe and reversible loss of consciousness, enabling surgical procedures to be carried out (Stachnik, 2006; Bovill, 2008; Franks, 2008). Several membrane ion channels and receptors have been identified as targets of various VAs (Rudolph and Antkowiak, 2004; Hemmings et al., 2005). Enhancement of the activity of inhibitory ionotropic receptors (Lobo and Harris, 2005; Zeller et al., 2008), depression of the excitatory ionotropic receptors (Yamakura et al., 2001), potentiation of the 2P-domain K⁺ channels (Patel and Honore, 2001), inhibition of voltage-gated Na⁺ channels (Hemmings, 2009) and of the transient receptor potential (TRP) family member TRPC5 (Bahnasi et al., 2008) probably all contribute to the central clinically relevant anaesthetic effects, whereas the action of the same VAs on other ion channels might, at least in part, be responsible for their numerous adverse effects (Stachnik, 2006; Bovill, 2008). Indeed, activation of TRP member TRPA1 by the ‘pungent’ VAs (those that are known to excite peripheral nociceptive neurons), isoflurane and desflurane (Matta et al., 2008; Cornett et al., 2008; Eilers et al., 2010), and sensitization of TRPV1 to its agonists, capsaicin and protons (Cornett et al., 2008; Eilers et al., 2010), might contribute to postoperative pain and inflammation, as well as producing airway irritation. Hypersensitivity to cold temperatures and a decreased shivering threshold are also common complications observed after administering VAs (Kurz et al., 1997; Kurz, 2008; Sessler, 2008), suggesting that at least some of their representatives might sensitize peripheral cold receptors. Other peripheral effects of these agents are still mechanistically poorly studied. Among these effects are excitation by VAs of mammalian nociceptor afferents (MacIver and Tanelian, 1990), hypersensitivity of laryngeal C-fibers (Mutoh T and Tsubone, 2003) and excessive respiratory reactions, such as coughing, inhibition of breathing (apnea), laryngospasm and secretion (Drummond, 1993). In fact, respiratory complications are the most frequent complications and the leading cause of death during anaesthesia. Respiratory depression, laryngospasm, apnea and other causes induce hypoxia. These complications are resolved spontaneously by correct treatment, but are potentially lethal if they are not detected in time, or in case of inability to intubate or ventilate the patient. Interestingly, several studies have reported that a cold air stimulus consistently causes an inhibition of breathing (Sekizawa et al., 1996) and that this inhibition of respiratory drive can also be caused by menthol (Eccles, 1994), suggesting that at least some VAs might sensitize peripheral cold receptors. Anaesthetics are also implicated in perioperative hypothermia (Sessler, 2008) by, at least, sensitizing peripheral thermoreceptors. Hypothermia is a multifactorial, common and serious complication of anaesthesia and surgery manifested by the drop of core body temperature by as much as 6°C. The combination of anaesthetic-induced impairment of thermoregulatory control and exposure to a cool operating room environment causes most surgical patients to become hypothermic. Mild intraoperative hypothermia triples the occurrence of postoperative wound infections and myocardial events, as well as increasing perioperative blood loss (Kurz, 2008). Furthermore, it delays postoperative recovery and prolongs the effect of almost all anaesthetic drugs (Kurz, 2008). This side effect could be reduced by perioperative warming, and recent evidence suggests that perioperative warming could be given routinely to all patients of various surgical disciplines in order to counteract the consequences of hypothermia (Sajid et al., 2009). Overall, these data show that the perioperative warming of surgical patients reduces postoperative wound pain, wound infection and shivering but does not entirely solve this problem (Sajid et al., 2009). As a consequence, a better understanding of this multifactorial side effect could be very relevant in medical practice.

Members of the TRP channel superfamily are the principal source of thermoreceptors of various modalities (Schepers and Ringkamp, 2009). Of these, transient receptor potential melastatin 8 (TRPM8), which is expressed in a subset of dorsal root ganglion (DRG) and trigeminal sensory neurons, is activated by innocuous cold (<25°C) and chemical imitators of cooling sensation such as menthol, icilin and eucalyptol (McKemy et al., 2002). The activation of the TRPM8 channel by cold or chemical agents depolarizes sensory nerve fibres causing their excitation and transmission of the signal to the integrative centres of the CNS. However, despite being proven to be a principal detector of environmental cold (Bautista et al., 2007; Dhaka et al., 2007), the role of TRPM8 in how VAs induce complications to general anaesthesia has yet to be assessed.

Application of halothane (0.5 mM) to the HEK-293 cells stably transfected with human TRPM8 (HEK_M8 cells) activated a membrane current with biophysical properties, such as a prominent outward rectification and a reversal potential of close to 0 mV, which is similar to the current activated by the established TRPM8 stimulus (i.e. cold temperatures) (Fig. 1A). TRPM8 activation by halothane did not increase monotonically with concentration, but exhibited a bell-shaped concentration dependence (Fig. 1B); at room temperature (23°C) anaesthetic potently activated TRPM8 in the narrow concentration range of 0.20–1 mM, with 0.5 mM being the most effective. Comparison of the steady-state activation curves of the background membrane current carried through TRPM8 channels (I_TRPM8) in HEK_M8 cells at room temperature and I_TRPM8 over-activated by further exposure to halothane showed that anaesthetic produces a depolarizing shift in the voltage-dependence of the TRPM8 channel activation by ∼100 mV (Fig. 1C,D), which might underlie its agonistic action mechanism.

As far as other VAs used in clinical practice are concerned, clinically relevant concentrations of isoflurane (0.2 mM, n = 5) (Fig. 2A), desflurane (0.2 mM, n = 3) (Fig. 2B) and sevoflurane (0.2 mM, n = 3) (Fig. 2C) also rapidly activated a membrane current in HEK_M8 cells with similar biophysical properties to the one activated by halothane. Fig. 2D summarizes the activation of I</emph>_TRPM8 upon VA application compared with basal current at room temperature. Because cold stimulus, menthol or any of the VAs tested were unable to induce such a current in the control HEK-293 cells (HEK_M8 cells without tetracycline induction, HEK_Ctrl cells), we concluded that in HEK_M8 cells, this current is associated with TRPM8 activation (I_TRPM8).

TRPM8 is a non-selective Ca²⁺-permeable cationic channel, therefore its activity in response to VAs can be evaluated by measuring the cytosolic Ca²⁺ concentration ([Ca²⁺]_c). Ca²⁺ imaging experiments using the Ca²⁺ probe fura-2AM, clearly demonstrated that isoflurane (0.5 mM, n = 9) induces a pronounced elevation of [Ca²⁺]_c (up to 1 µM in HEK_M8 cells) but is ineffective in control HEK_Ctrl cells. This would suggest that this elevation probably results from isoflurane-stimulated TRPM8-mediated Ca²⁺ entry (Fig. 2E). Interestingly, withdrawal of isoflurane led not only to the recovery of [Ca²⁺]_c in HEK_M8 cells, but also to the substantial decrease in their basal room temperature [Ca²⁺]_c level from the pre-isoflurane value of 430±75 nM, to the post-isoflurane value of 205±7 nM (n = 9) (Fig. 2E). We also performed additional Ca²⁺ imaging experiments (at 37°C) with fura-2 in which the effect of TRPM8 inhibitor (BCTC, 10 µM) on isoflurane-induced (0.5 mM) sustained Ca²⁺ entry was assessed. This Ca²⁺ entry was effectively antagonized by the TRPM8 inhibitor BCTC (Fig. 2F).

To more thoroughly investigate how general anaesthetics affect the TRPM8 activity, we performed measurements of the single-channel activity of the channel. A high expression level of TRPM8 protein in HEK_M8 cells led to the presence of multiple channels in all patches tested and limited our analysis to characterization of overall probability of TRPM8 opening (NP_open). Fig. 2G,H compares basal and isoflurane-stimulated (0.2 mM, n = 5) TRPM8 activity by analyzing fragments before and 20 seconds after isoflurane application, and shows a significant increase in TRPM8 NP_open upon stimulation by isoflurane.

The stimulatory action of VAs on TRPM8 is followed by its sustained inhibition; as shown in Fig. 3, initial activation of I_TRPM8 in HEK_M8 cells in response to clinically relevant concentrations of halothane (Fig. 3A), isoflurane (Fig. 3B), desflurane (Fig. 3C) and sevoflurane (Fig. 3D) was followed by a profound current decrease despite a continuing presence of the drug. This residual current was 75–85% of the maximally stimulated value (Fig. 3E) and commonly brought the current down to the steady-state level, or even below the baseline I_TRPM8 at room temperature. Chloroform (CHCl₃) also produced a delayed and sustained I_TRPM8 inhibition, but its extent was notably smaller than the other VAs studied (i.e. a residual current of 40%, Fig. 3E). Moreover, incubation of HEK_M8 cells in 0.25 mM CHCl₃ for 15 minutes resulted in ∼40% decrease in the amplitude of basal I_TRPM8 at room temperature (Fig. 3F). This is consistent with the notion that prolonged action of VAs induces TRPM8 inhibition.

Prolonged exposure to VAs was able to inhibit not only background non-stimulated TRPM8 activity at room temperature but also stimulated activity. Fig. 3H shows that exposure of HEK_M8 cells to isoflurane at 37°C, besides activating a TRPM8-mediated [Ca²⁺]_c rise, at the same time diminished the response to subsequent cold stimuli, whereas without isoflurane, a series of cold stimuli produced ever-increasing [Ca²⁺]_c responses (Fig. 3G). Taken together, these data clearly demonstrate a dual action of VAs on TRPM8: a rapid and transient activation followed by delayed sustained inhibition.

Our results clearly show that the rapid TRPM8-activating mode of the VAs is associated with a depolarizing shift in the voltage-dependence of the channel activation (Fig. 1C,D). Thus, we next asked what kind of molecular interactions might underlie this phenomenon. First, there might be a direct interaction of the VAs with a channel protein, as has been suggested for the mammalian K_2P channels, TREK-1 and TASK (also known as KCNK2 and KCNK3, respectively) (Patel et al., 1999). In the case of the TREK-1 and TASK channels, their C-terminal regions were found to be crucial for anaesthetic-induced activation (Patel et al., 1999). Second, a TRPM8 channel might change its gating owing to mechanical tension coming from an alteration in membrane microcurvature as a result of the partitioning of highly hydrophobic VAs into the lipid bilayer. Such a mechanism has been suggested for a number of lipid modifiers on the mechano-gated TREK and TRAAK (also known as KCNK4) K_2P channels (Patel and Honoré, 2001). Third, VAs might influence the sensitivity of a TRPM8 channel to its endogenous modulators, first of all to phosphatidylinositol 4,5-bisphosphate (PIP₂), which regulates the activation of TRPM8 through interaction with the TRP domain of the channel (Rohács et al., 2005).

To test the possibility that the site of the interaction between VA and the TRPM8 channel might be the same as the one for its classical agonist menthol, we used two channel mutants, Y745H and L1009R, that have been shown to have strongly reduced menthol and icilin, but not cold, voltage or PIP₂ sensitivity (Bandell et al., 2006). As shown in Fig. 4A, both halothane and menthol are capable of activating I_TRPM8 in HEK-293 cells transfected with the wild-type TRPM8. However, in HEK-293 cells transfected with the Y745H (n = 3) and L1009R (n = 3) mutants, exposure to menthol failed to activate any measurable current, whereas current activation by halothane was not affected (Fig. 4B,C). Thus, the molecular determinants, within a TRPM8 channel, required for channel gating by VAs are distinct from those used by menthol.

To verify the ‘bilayer-coupled’ hypothesis, we used a lipid bilayer modifier, namely arachidonic acid (AA), which is a polyunsaturated fatty acid with proven efficacy on mechanosensitive K_2P channels. By partitioning into the outer leaflet of the lipid bilayer, AA is known to induce a convex membrane curvature (i.e. membrane crenation), leading to the activation of TREK and TRAAK channels (Patel and Honoré, 2001). In our hands, the effect of TRPM8 opening either by menthol or by VAs, was not mimicked by AA (10 µM; n = 5) (Fig. 5A). We also failed to detect any TRPM8 activation by anionic amphipath trinitrophenol (TNP), which has proven efficacy on mechanosensitive K_2P and TRPA1 channels (Patel and Honoré, 2001; Hill and Schaefer, 2007). Moreover, consistent with previous studies (Andersson et al., 2007; Bavencoffe et al., 2011), AA inhibited the I_TRPM8. Hence, it is not likely that the anaesthetic action of TRPM8 is mediated by an effect on the lipid bilayer.

In our previous study, we showed that cationic amphipaths, such as chlorpromazine (CPZ), which preferentially inserts into the inner leaflet of the bilayer, thereby facilitating the formation of the membrane cup-shapes (cup-formers), are able to antagonize the activating effects of menthol and lysophospholipids on TRPM8 (Vanden Abeele et al., 2006). When CPZ (1 µM) was applied together with CHCl₃, it not only prevented CHCl₃-mediated TRPM8 activation but also reversibly inhibited baseline I_TRPM8 at room temperature (Fig. 5B). Similar antagonistic actions of cup-formers have been reported for the mechano-gated K_2P channels (Patel and Honoré, 2001). Thus, we cannot entirely rule out possible TRPM8 modulation by a bilayer alteration mechanism, particularly with respect to the inhibitory influence of membrane invaginations on TRPM8 channel activation. However, it is unlikely that this mechanism underlies VA action. Given that modulation of TRPM8 by VAs correlates with their ability to partition into the membrane (halothane and sevoflurane are better than desflurane) one cannot also exclude VAs signalling through membrane fluidity.

Next, we examined a possible dependence of the VA-activated TRPM8-mediated current on the PIP₂ regulatory system, which is important for supporting TRPM8 channel function (Rohács et al., 2005). After establishment of whole-cell configuration, TRPM8-mediated current usually runs down, due to the depletion of PIP₂ by lipid phosphatase. This rundown could be accelerated by intracellular poly-lysine dialysis, which acts as a PIP₂ scavenger (Rohács et al., 2005). In our experiments, inclusion of 3 µg/ml poly-lysine in the pipette solution within 15–25 minutes of cell dialysis, inhibited more than 80% of the cold-activated I_TRPM8. However, application of CHCl₃ (Fig. 5C) or halothane (Fig. 5D) was still able to reactivate the current (n = 3), thus suggesting that the agonistic action of VAs on TRPM8 occurs through a mechanism distinct from that described for menthol, temperature or icilin, because it seems to be independent of PIP₂ level.

Next, we studied the effects of halothane on endogenous TRPM8 in isolectin B4 (IB4)-negative small-diameter DRG neurons (see the Materials and Methods). These are known to preferentially express the TRPM8 cold receptor, but they do not express the TRPV1 heat receptor and only express negligible levels of the pungent-compound receptor TRPA1 (Kobayashi et al., 2005). Given that VAs can potentially modulate or directly activate different types of endogenous channels, we first evaluated the ability of VAs to specifically affect menthol-activated current in DRG neurons. Fig. 6A shows that the application of halothane (0.5 mM) on top of menthol (100 µM) potentiates menthol-activated I_TRPM8 by ∼30% at 100 mV and by ∼50% at −100 mV (n = 5). The current in the presence of halothane had the same reversal potential and apparent rectification as the menthol-activated one (Fig. 6B), suggesting that the stimulation of current occurs because of the action of the anaesthetic on TRPM8 channels and not on some other channel type(s). When neurons were clamped at a holding potential (V_h) close to the resting potential of the neurons (i.e. −60 mV), application of menthol (100 µM) activated an inward current (Fig. 6C; n = 3); this current was sufficient to produce depolarization and action potentials (APs) that were readily detectable in the current-clamp mode (Fig. 6D; n = 4). Application of halothane on top of menthol augmented the inward current (Fig. 6C; n = 3), further depolarized the neuron and substantially increased the frequency of AP firing (Fig. 6D; n = 4), which is consistent with the ability of halothane to enhance TRPM8-mediated signalling.

In a second series of experiments, we examined whether VAs not only potentiated the menthol-activated current, but also directly activated TRPM8 in IB4-negative small-diameter DRG neurons. Fig. 7A shows representative recordings of the background currents at different step potentials in DRG neuron and the currents following exposure to halothane (0.5 mM). One can see that application of halothane alone was sufficient to activate membrane current above the baseline (Fig. 7A,B) and this current had biophysical properties similar to that activated by menthol (compare the current–voltage relationship shown in Fig. 6B with that shown in Fig. 7C). In our hands, isoflurane (0.5 mM) also activated membrane current with biophysical properties similar to I_TRPM8. Moreover, this current was effectively antagonized by the TRPM8 inhibitor BCTC (10 µM) (supplementary material Fig. S1).

The current activated by halothane in the IB4-negative menthol-responding DRGs was suppressed by the known TRPM8 inhibitor, capsazepine (20 µM, Fig. 7D). Given that of the two other VA-sensitive TRP members, TRPV1 is not expressed in these neurons and TRPA1 is not sensitive to capsazepine (Matta et al., 2008; Cornett et al., 2008; Eilers et al., 2010), this provides additional confirmation of the involvement of TRPM8 in the observed effects (Simon and Liedtke, 2008). Moreover, as in HEK_M8 cells, prolonged exposure of menthol-responding DRG neurons to halothane after the initial current activation, caused pronounced current inhibition (Fig. 7E), indicating that VAs exert dual action not only on recombinant TRPM8 but also on the native neuronal one. Ca²⁺ imaging experiments with fura-2AM (at 37°C) demonstrated that halothane (0.5 mM, n = 5) also induces a transient [Ca²⁺]_c increase in menthol-sensitive DRG neurons (Fig. 7F). The fact that in the current-clamp mode, a brief application of halothane (0.5 mM, n = 3) alone was able to cause depolarization and AP firing that was inhibited by the TRPM8 antagonist BCTC (10 µM) (Fig. 7G), demonstrates the physiological significance of VA action on the TRPM8 cold receptor in sensory neurons. Furthermore, to exclude the possibility of possible involvement of TRPA1, we assessed the effect of halothane (0.5 mM) on membrane currents in DRG neurons in the presence of the TRPA1 inhibitor HC-030031. In these conditions, halothane also activated membrane current with biophysical properties similar to those of I_TRPM8, suggesting that they are independent of TRPA1 (Fig. 7H). Additionally, the halothane-activated current was inhibited by BCTC (10 µM), again confirming the involvement of TRPM8 (Fig. 7H).

The involvement of TRPM8 in the mechanism(s) of VA-induced complications of general anaesthesia has yet to be evaluated. On the one hand, several studies have established that a cold air or menthol stimulus consistently causes an inhibition of respiratory drive (Eccles, 1994; Sekizawa et al., 1996). On the other hand, a pharmacological blockade of TRPM8 produces hypothermia (Knowlton et al., 2011; Almeida et al., 2012). We hypothesized that transient TRPM8 activation by VAs might be at least partly responsible for the inhibition of respiratory drive, whereas delayed sustained inhibition of TRPM8 following VA administration, might contribute to the induction of hypothermia. To test this hypothesis, we conducted comparative functional tests on wild-type and Trpm8-knockout mice.

Fig. 8A,B shows a comparison of the body temperatures and respiratory frequency of the wild-type and Trpm8^−/− mice during isoflurane anaesthesia. Body temperature and respiratory frequency of the wild-type and Trpm8^−/− mice were not statistically different in the resting state: 38.9°C±0.2 in wild-type animals versus 39.3°C±0.2 in Trpm8^−/− mice (Student's t-test; P>0.05 for both). Following exposure to isoflurane, both functional parameters underwent an expected reduction in both animal groups, although the reduction was more pronounced in the wild-type animals. The body temperature of Trpm8^−/− mice remained significantly higher than that of wild-type mice (Student's t-test; ***P<0.01; n = 10) during the whole 20-minute monitoring period, consistent with the notion that the TRPM8 cold receptor contributes to the lowering of body temperature in response to isoflurane. On the other hand, the respiratory frequency in Trpm8^−/− mice was significantly higher than in wild-type mice (Student's t-test; ***P<0.01; n = 10) only during the first half of the 20-minute-long monitoring period. During the second half, no statistically significant differences were detected, suggesting a role for the TRPM8 cold receptor in this complication only at the beginning of anaesthesia. These results agree perfectly with previous reports showing that a cold air or menthol stimulus causes an inhibition of respiratory drive (Eccles, 1994; Sekizawa et al., 1996) and with our present findings that VAs produce a transient activation of the TRPM8 cold receptor. Indeed, we demonstrate that VAs, as well as activating TRPM8, also induce its strong desensitization to the physiological agonist, cold (Fig. 3G,H). Such desensitization could be viewed as an equivalent to pharmacological TRPM8 blockade, which was recently reported to evoke a decrease in body temperature in rats and mice (Knowlton et al., 2011; Almeida et al., 2012) and, thus, can explain the hypothermia observed in our experiments.

In the present study, we show for the first time that VAs, including the archetypal anaesthetic chloroform as well as halothane, isoflurane and modern anaesthetics such as desflurane and sevoflurane, activate both the heterologously expressed and the endogenous TRPM8 cold receptor. Activation of endogenous TRPM8 in rat DRG neurons by halothane results in depolarization and increased firing, thereby indicating induction of peripheral cold sensitivity by VAs. Our results suggest that: (1) TRPM8 modulation by a bilayer alteration mechanism cannot account for the agonistic actions of the VAs; (2) the molecular determinants required for channel gating by VAs are distinct from those used by menthol; and (3) the agonistic action of VAs on TRPM8 occurs through a distinct mechanism from the one described for menthol, temperature or icilin, because it seems to be independent of PIP₂ level.

Recent ultra-structural studies on general-anaesthetic-sensitive bacterial homologues of mammalian cysteine-loop-type ionotropic receptors, have, at least for this type of receptors, revealed a common general-anaesthetic-binding site. This site consists of mainly hydrophobic amino acids located within the cavity on the extracellular side of the receptor-forming subunits (Nury et al., 2011). It remains to be determined, the extent to which structural considerations derived from cysteine-loop-type ionotropic receptors can be applied to TRP channels.

In addition to inducing unconsciousness, amnesia and analgesia, VAs also cause a number of side-effects, which include an excitation of mammalian nociceptor afferents (MacIver and Tanelian, 1990), hypersensitivity of laryngeal C-fibres (Mutoh and Tsubone, 2003), excessive respiratory reactions, such as coughing, inhibition of breathing (apnea), laryngospasm and secretion (Drummond, 1993). Hypersensitivity to cold temperatures and a lower shivering threshold are also common complications observed after administering VAs (Kurz et al., 1997; Kurz, 2008; Sessler, 2008; MacIver and Tanelian, 1990), suggesting that at least some of their representatives might sensitize peripheral cold receptors. Our results establish a link between cold sensation and VA administration, since we show that VAs directly activate the TRPM8 cold receptor.

Recently, it has been shown that ‘pungent’ inhalation general anaesthetics (which are known to excite peripheral nociceptors), namely isoflurane and desflurane, activate TRP-member TRPA1 and sensitize the TRPV1 heat receptor to its agonists, capsaicin and protons (Cornett et al., 2008; Matta et al., 2008; Eilers et al., 2010), but do not affect the TRPM8 cold receptor. These authors concluded that the pro-nociceptive effects of these anaesthetics combined with surgical tissue damage may specifically contribute to postoperative pain and inflammation, as well as producing airway irritation. In our hands, TRPM8 was activated with quite similar potency by the non-pungent VAs, halothane, chloroform and sevoflurane, as well as the pungent ones, isoflurane and desflurane. TRPM8 activation by these compounds was validated in HEK_M8 cells as well as DRG neurons using both patch-clamp and Ca²⁺ imaging techniques, which together strongly confirm our observations. We currently have no explanation why our results on TRPM8 sensitivity to the VAs are different from those reported previously (Matta et al., 2008). However, we maintain that, aside from being able to excite TRPA1/TRPV1-expressing nociceptors, VAs are also capable of exciting a separate population of TRPM8-expressing cold-sensitive afferents, and this might contribute to the set of complications associated with impaired peripheral sensitivity to cold temperatures.

Concerning perioperative hypothermia, it has been demonstrated that TRPM8 agonists, such as menthol, icilin and 1.8-cineole enhance thermogenesis (Masamoto et al., 2009), whereas TRPM8 antagonists induce a decrease in body temperature (Knowlton et al., 2011; Almeida et al., 2012). However, in view of possible non-specific side effects of these compounds, attribution of their thermogenicity solely to TRPM8 should be taken with caution. Our functional tests on wild-type and Trpm8-knockout mice clearly show that a lack of TRPM8 reduces hypothermia associated with isoflurane anaesthesia and that these effects are probably linked to the strong desensitization of TRPM8 by VAs to the physiological agonist, cold. Such desensitization might be viewed as equivalent to pharmacological TRPM8 blockade, and thus, can explain the hypothermia observed in our experiments. Although more studies are necessary to explore the role of other thermo-receptors (i.e. TRPA1, TRPV1 and TREK) in VA-induced hypothermia, these data unequivocally demonstrate that VA-mediated activation of TRPM8 is, at least partially, involved in this process.

The relationship between respiratory depression and TRPM8 modulation by VAs appears to be more complex. Similarly, respiratory depression by anaesthetics is also multifactorial. Respiratory complications are the most frequent complications and the leading cause of death during anaesthesia. This depressant effect on respiration by VAs and isoflurane, in particular, is enhanced by premedication and concomitant administration of other preparations, which also have a depressing effect on respiration. In medical practice, it is therefore essential to closely monitor breathing and to support it, where necessary, by mechanical ventilation. Patients with myasthenia gravis are particularly susceptible to anaesthetic preparations that cause respiratory depression. These complications are resolved by correct treatment, but can be potentially lethal if they are not detected in time or if there is an inability to intubate or ventilate the patient. Interestingly, several studies have reported that, on the one hand, a cold air stimulus consistently causes an inhibition of breathing (Sekizawa et al., 1996) and, on the other hand, that inhibition of respiratory drive can also be caused by menthol (Eccles, 1994). It has also been demonstrated that a subpopulation of airway vagal afferent nerves express TRPM8 receptors and that activation of TRPM8 receptors by cold excites these airway autonomic nerves increasing airway resistance (Xing et al., 2008). These data suggest the involvement of TRPM8 activation in the inhibition of respiratory drive. Our findings, that in wild-type mice, the progression of isoflurane anaesthesia is accompanied by a more pronounced decrease in respiratory frequency than in the Trpm8^−/− mice, are fully consistent with this notion. Our data clearly show that the difference between wild-type animals and Trpm8^−/− mice was statistically significant during the first half of the monitoring period. During the second half, no statistical differences were found, suggesting a role for TRPM8 cold receptor activation at the beginning of anaesthesia. These observations correlate with the dual effect of VAs on TRPM8. At the same time, our results demonstrate that TRPM8 is not the only target in this process, given that the depressant effect of isoflurane on respiratory frequency is partially preserved in the Trpm8^−/− mice. We have to bear in mind that respiratory depression by anaesthetics is also multifactorial, and that the TRPA1 channel could be also involved, given the evidence that desflurane induces airway contraction through an activation of TRPA1 in sensory C-fibres (Satoh and Yamakage, 2009).

In conclusion, our results provide a better understanding of the side effects of volatile anaesthetics, thereby contributing to the public health domain in general and in particular to the elaboration of new anaesthetic strategies.

Studies of recombinant TRPM8 were performed on HEK-293 cells stably transfected with human TRPM8 under the control of a tetracycline-inducible promoter (HEK_M8 cells) (Thebault et al., 2005). HEK_M8 cells without tetracycline induction were used as controls (HEK_Ctrl cells). Neurons were isolated from the lumbar DRG of adult Wistar rats (250–300 g) using an enzymatic digestion procedure. All animal experiments were performed according to local animal ethical committee guidelines. Cell suspensions were plated onto Petri dishes filled with 10% fetal-calf-serum- and 8 µg/ml gentamicine-supplemented DMEM (Gibco, UK) culture medium and incubated for 18–24 hours at 37°C in a 95% air, 5% CO₂ atmosphere, prior to use in electrophysiological experiments. Immediately before electrophysiological experiments, neurons were stained with 10 µg/ml IB4–Alexa-Fluor-488 (Molecular Probes) for 20 minutes and then rinsed for 10 minutes in extracellular solution. Only IB4-negative small-diameter neurons were used for subsequent experiments.

Membrane currents were recorded in the whole-cell configuration of the patch-clamp techniques using the Axopatch 200B amplifier (Molecular Devices, Union City, CA). The resistance of the patch pipettes, fabricated from borosilicate glass capillaries (World Precision Instruments, Sarasota, FL), when filled with the intracellular solution, was 2–3 megaohms for the whole-cell recordings. In the whole-cell experiments, series resistance was compensated for by ∼70%. Currents were filtered at 1 or 2 kHz and sampled at 10 kHz.

Whole-cell currents were measured under nearly identical ionic conditions [i.e. standard extracellular solution containing (in mM): 150 NaCl, 1 MgCl₂, 5 glucose, 10 HEPES pH 7.3 (adjusted with NaOH), whereas the pipette was filled with the intracellular solution containing (in mM): 150 NaCl, 3 MgCl₂, 5 EGTA, 10 HEPES, pH 7.3 (adjusted with NaOH)]. For temperature control and solution exchange, the TC1-SL25 system (Bioscience Tools, San Diego, CA) was used with the temperature probe placed near the patch pipette tip. The system provided a temperature stability of at least 0.2°C. All experiments were carried out at 23°C. During whole-cell patch-clamp recordings, DRG neurons were bathed in the standard extracellular solution (in mM): 140 NaCl, 5 KCl, 5 glucose, 10 HEPES, 2 CaCl₂, 1 MgCl₂, pH 7.4, while being dialyzed with Cs⁺-based intracellular patch-pipette solution containing (in mM): 140 CsCl, 5 EGTA, 10 HEPES, 2 Mg-ATP, 0.5 Li-GTP, pH 7.3. For current-clamp recordings, the Cs⁺ was replaced with K⁺ in the pipette solution.

Single-channel recordings were performed in cell-attached patches to HEK_M8 cells stably expressing TRPM8 protein. Cells were immersed in high-KCl solution, in order to bring cell potential close to zero (in mM): 150 KCl, 2 MgCl₂, 1 CaCl₂, 5 glucose, 10 HEPES PH 7.3 (adjusted with KOH) and the pipettes were filled with the standard extracellular solution described above. Recording of the basal activity of TRPM8 was followed by application of 0.2 mM isofluorane and results before and 20 seconds after application of isofluorane were analyzed with the help of Clampfit-10 program (Molecular Devices, Union City, CA). The obtained NP_open values were normalized to the basal NP_open levels individually for each trace.

These experiments were performed using the membrane-permeable Ca²⁺-sensitive dye fura-2AM, as detailed previously (Vanden Abeele et al., 2006).

Halothane, chloroform and diethyl-ether were purchased from Sigma-Aldrich (Saint-Quentin Fallavier, France); isoflurane from Laboratoires Belamont (Neuilly Sur Seine, France); sevoflurane from Abbott (Rungis, France) and desflurane from Baxter (Maurepas, France). Other reagents were purchased from Sigma-Aldrich. Saturated stock solutions of VAs were prepared in gas-tight bottles by dissolving excess anaesthetic agents in bath solutions and stirring vigorously overnight. From these stock solutions, fresh dilutions were made up immediately prior to experiments.

Membrane current recordings obtained from cells were analyzed and plotted using the pCLAMP 9 (Axon Instruments, Inc.) and Origin 5 software (Microcal Software Inc.). Results are expressed as the means±s.e.m. Statistical analysis was performed using Student's t-test (differences were considered significant when P<0.05).

We are grateful to D. Julius for providing us with the Trpm8-knockout mice and to A. Patapoutian for providing us with the mutants Y745H and L1009R.