Cooling-Sensitive TRPM8 Is Thermostat of Skin Temperature against Cooling

We have shown that cutaneous cooling-sensitive receptors can work as thermostats of skin temperature against cooling. However, molecule of the thermostat is not known. Here, we studied whether cooling-sensitive TRPM8 channels act as thermostats. TRPM8 in HEK293 cells generated output (y) when temperature (T) was below threshold of 28.4°C. Output (y) is given by two equations: At T >28.4°C, y = 0; At T <28.4°C, y  =  -k(T – 28.4°C). These equations show that TRPM8 is directional comparator to elicits output (y) depending on negative value of thermal difference (ΔT  =  T – 28.4°C). If negative ΔT-dependent output of TRPM8 in the skin induces responses to warm the skin for minimizing ΔT recursively, TRPM8 acts as thermostats against cooling. With TRPM8-deficient mice, we explored whether TRPM8 induces responses to warm the skin against cooling. In behavioral regulation, when room temperature was 10°C, TRPM8 induced behavior to move to heated floor (35°C) for warming the sole skin. In autonomic regulation, TRPM8 induced activities of thermogenic brown adipose tissue (BAT) against cooling. When menthol was applied to the whole trunk skin at neutral room temperature (27°C), TRPM8 induced a rise in core temperature, which warmed the trunk skin slightly. In contrast, when room was cooled from 27 to 10°C, TRPM8 induced a small rise in core temperature, but skin temperature was severely reduced in both TRPM8-deficient and wild-type mice by a large heat leak to the surroundings. This shows that TRPM8-driven endothermic system is less effective for maintenance of skin temperature against cooling. In conclusion, we found that TRPM8 is molecule of thermostat of skin temperature against cooling.


Introduction
In mammals, thermostat is a key device to maintain body temperature against changes in ambient temperatures. Many investigators have studied thermostats [1,2,3,4,5,6]. Since Adrian [7], it has been assumed that thermo receptors are transducers (or sensors) that change temperature into the firing-rate code of impulse trains, travelling to the brain. Along this line, a target neuron receiving impulses from thermal receptors has been assumed to be ''thermostat'' that compares skin temperature with a set-point temperature at the level of the firing-rate code, and generates impulses as drive signal inducing thermoregulatory responses [1,3,4,8].
If thermo receptors are transducers of temperature, their outputs should be related to temperature strictly. However, output of thermo receptors shows adaptation [7,9,10,11,12], even when temperature is constant. In addition, receptors show threshold response to temperature changes [10,11]. Moreover, TRPM8, a member of thermo TRP channel family [13], responds to menthol and icilin as well as temperature [10,14]. These results deny the assumption that receptors are transducers. Therefore, it is impossible for cerebral target neurons to detect skin temperature from afferent impulses sent from cooling-sensitive receptors. In fact, the ''thermostat'' neuron hypothesized in the brain [3,4,8,15] has not been found. By lack of the ''thermostat'', regulation of skin temperature is not explained.
In contrast, we have shown that cutaneous cooling-sensitive receptor itself can act as thermostat, comparing skin temperature with its threshold temperature [11,16]. Single-channel recording shows that channel activities appear when temperature is lower than threshold temperature [11]. These results indicate that phase transition is a biological direct method for comparison, different from artificial indirect one using a code. If cooling-induced activities of receptors elicit responses to warm the skin recursively, a negative feedback loop is formed for regulation of skin temperature against cooling [16]. However, molecule of the thermostat has not been clarified.
Recently, cooling-and menthol-sensitive TRPM8 [10,14] has been identified. TRPM8 is expressed in C and Ad sensory fibers [17,18] of cutaneous sensory nerves [19,20]. Experiments of TRPM8-deficient mice show that TRPM8 is required for cold sensation [17,21,22]. Menthol application to the whole trunk skin [23] and intravenous menthol injection [24] induce behavioral responses to warm the skin. In contrast, TRPM8-induced autonomic thermoregulatory responses are not fully analyzed [23,25,26]. Here, with TRPM8-deficient mice, we studied whether TRPM8 in the skin induces behavioral and autonomic responses to warm skin against room cooling.

TRPM8 is comparator, but not transducer
We examined thermal responses of TRPM8 channels expressed in HEK293 cells by calcium imaging [12]. When temperature (T) was decreased slowly from 38 to 10uC (Fig. 1B), activities (y) of TRPM8 appeared at threshold temperature (mean, 28.4uC), and increased with a fall in temperature, consistent with prior studies [10,14,17]. To explore a role of TRPM8, we schematically showed relationship between output (y) and temperature (T) in Fig. 1C. Due to the threshold response, output (y) is expressed by two equations: Equation 1: y~0 at Tw28:4 0 C Equation 2: y~{k T{28:4 0 C ð Þat Tv28:4 0 C Threshold temperature (28.4uC) is constant specific to TRPM8, and T is variable applied to TRPM8. In Equation 2, -k (,0) is a slope constant. Two equations show that TRPM8 has silent (Equation 1) and active (Equation 2) phases, and that phase transition occurs at 28.4uC [11]. In terms of control theory, TRPM8 functions as directional comparator judging whether T is lower than 28.4uC, and generates output (y) proportionately with a negative thermal difference (DT = T -28.4uC). For instance, when T is 22.4uC, TRPM8 enters active phase (Equation 2), and generates output (y) according to 26uC.
To aid comprehension, we symbolically show TRPM8 by a cooling-sensitive switch (Fig. 1D). At DT .0uC, this switch is turned off. At DT ,0uC, this switch is turned on and output (y) increases with a fall in T from 28.4uC (Fig. 1C). In sensory neurons, output of sensory receptors generates receptor potential [11], eliciting impulse trains on sensory nerves [17]. If negative DT-dependent impulse activities induce thermoregulatory responses to warm the skin for minimizing DT recursively, TRPM8 switch acts as thermostats of skin temperature having 28.4uC as set-point temperature against cooling [16]. Then, intensity of the thermoregulatory activities (e.g., cold sensation and BAT activities) increases proportionately with expansion of DT, which corresponds to the proportional control in technology [16].
TRPM8 is also activated by menthol, when menthol concentration is higher than threshold level of 20 mM [10]. This shows that TRPM8 acts as directional comparator of menthol concentration, generating output (y) depending on the positive difference from 20 mM to menthol concentration. In addition, TRPM8 is activated by icilin, when its concentration is higher than 0.1 mM [10]. These properties indicate that TRPM8 are polymodal comparator of several kinds of stimuli. In later experiments, menthol is applied to the skin to explore TRPM8-driven thermogenic activities at neutral temperature [23]. It should be notified, however, that even when menthol-induced activities of TRPM8 elicits responses to warm the skin, negative feedback loops are not formed, because responses to warm the skin do not reduce menthol concentration in the skin.
In contrast, if TRPM8 is transducer of temperature, output (y) of TRPM8 should be expressed by one continuous function of T [8], In fact, in the transducer hypothesis [7], threshold responses of sensory receptors were not described at all. However, threshold response ( Fig. 1) is essential feature of TRPM8 [10,14,17], and output (y) is not expressed by one function of T. Moreover, output of TRPM8 is changed by menthol, icilin and other chemicals, even when T is constant [10,14,27]. When menthol and T changes are applied in turn, output of TRPM8 shows complicated trace, different from T [14]. Thus, target neurons can hardly detect skin temperature from impulses sent from TRPM8. These results reject the hypothesis that the ''thermostat'' of skin temperature is present in the brain [1,4,8].

TRPM8-driven behavior for warming the sole skin against room cooling
We examined whether output of TRPM8 induces behavior for warming the sole skin for regulation of skin temperature against cooling (Fig. 2). A free-moving mouse was placed in a cage with two floor temperatures for thermal preference test. When room temperature was neutral (27uC), stay time on 35uC floor was similar between wild-type and TRPM8-deficient mice, showing that TRPM8 is not involved in this behavior at 27uC. When room temperature was 10uC, stay time on 35uC floor was increased in both mice. However, stay time on 35uC floor in wild-type mice was longer than that in TRPM8-deficient mice. This indicates that TRPM8 comparators induce heat-seeking behavior for warming the sole skin against room cooling. This heat-seeking behavior corresponds to TRPM8-induced cold sensation [17,21,22], showing that TRPM8-induced cold sensation triggers the behavior for warming the sole skin. Cooling-induced TRPM8-independent behavior seems to be responses induced by non-TRPM8 receptors [28]. These results show that TRPM8 comparators work as thermostats of skin temperature, inducing behaviors for warming the sole skin against cooling.
TRPM8-driven autonomic responses for warming the skin against cooling TRPM8-induced activities of thermogenic BAT. Pathways from cooling-sensitive receptors to thermogenic brown adipose tissues (BAT) are well characterized [29,30]. We examined whether TRPM8 channels induce BAT activities Ca 2+ level (ratio) in five HEK293 cells expressing TRPM8. When temperature (T) was decreased slowly from 38 to 10uC (B), activities (output, y) of TRPM8 appeared at threshold temperature (28.460.5uC, mean 6 SEM; n = 5) and increased with a fall in temperature. (C) Relationship between output (y) and temperature (T). Threshold response in (A) is shown approximately by two lines (green), which are expressed by two linear equations. When T is above 28.4uC, output (y) is 0. When T is below 28.4uC, output (y) = -k(T -28.4uC). For explanation of the automatic control theory, thermal difference, DT ( = T -28.4uC), is defined. When T is higher than 28.4uC, DT is positive. When T is lower than 28.4uC, DT is negative. Thus, y increases with expansion of negative DT. (D) Representation of TRPM8 by a coolingsensitive switch. When DT is positive, this switch is turned off. When DT is negative, this switch is turned on and y increases with expansion of negative DT as shown in C. doi:10.1371/journal.pone.0017504.g001 against room cooling (Fig. 3). To study cooling-induced BAT activation, mice were put at 10 or 27uC for 1h. When room temperature was 27uC, fluorescence of UCP1 was low in both TRPM8-deficient and wild-type mice (Fig. 3A, C). When room temperature was 10uC, UCP1 signals were increased in both mice (Fig. 3B, D), but the signal strength in wild-type mice was higher than that of TRPM8-deficient mice (Fig. 3B, D). To confirm this observation, we examined cooling-induced activities in phosphorylated NFkB (pNFkB) in nuclei, as a marker of BAT cell activities [23,31]. At 27uC (Fig. 3E, G), fluorescence of pNFkB was detected in neither TRPM8-deficient nor wild-type mice. At 10uC, the number of pNFkB-positive nuclei was increased in both mice (Fig. 3F, H), but the ratio of the number of pNFkB-positive nuclei to all nuclei in wild-type mice was higher than that in TRPM8-deficient mice (Fig. 3I). These results show that TRPM8 induces activities of thermogenic BAT used for warming the skin in response to cooling.
Menthol-and cooling-induced TRPM8-driven thermogenic responses. We examined whether TRPM8 in the skin induces thermogenic responses for warming the skin. When room temperature was 27uC, core temperature was 36.8uC and skin temperature was 34.2uC in both mice (Fig. 4). This shows that TRPM8 is not involved in maintaining normal body temperatures at 27uC, consistent with the earlier study [17]. When menthol was applied to the whole trunk skin of mice at 27uC (Fig. 4A,B, arrows), TRPM8-dependent rise occurred in core (1.3uC) and skin temperatures (0.4uC).
In contrast, when room was cooled from 27 to 10uC (Fig. 4C,  D), core temperature of TRPM8-deficient mice was decreased (Fig. 4C), but that of wild-type mice was increased slightly. The TRPM8-dependent difference (1.0uC) reflects TRPM8-induced thermogenesis in the core. By the same room cooling, however, skin temperature was decreased severely from 34.2uC to 24.1uC in both mice (Fig. 4D), as if regulation is not functioning. Room cooling-induced decrease in skin temperature in wild-type mice is similar to the earlier study of rabbits [32]. These results indicate that TRPM8-induced thermogenesis in the core is not powerful enough to warm the skin against cooling.

TRPM8 is not transducer of temperature
It has generally been assumed that a sensory receptor is transducer that changes physical quantity into the firing-rate code of nerve impulses, and that sensation appears by decoding the code [7]. Along this line, a target neuron receiving impulses sent from thermo receptors has been assumed to be the ''thermostat''  that compares skin temperature with a set-point temperature at the code level [1,3,4,8,15].
However, the present study of TRPM8 shows that the above assumptions are not valid as a whole. TRPM8 is a polymodal receptor, responding to temperature, menthol, icilin and other chemicals [10,14,17,27]. This is strong evidence to reject the view that TRPM8 is transducer of temperature, because TRPM8 does not respond to skin temperature solely. From the very beginning, the view that a mouse measures skin temperature with receptor is inadequate, for measurement is done exclusively by a person who knows the meaning of temperature (e.g., 28.4uC) and the principle of measurement. Thus, the hypothesis that a cerebral target neuron is the ''thermostat'' of skin temperature [1,3,4,8] is not valid.
Temperature (e.g., 22.4uC) or menthol concentration (e.g., 100 mM) is physical quantity, and completely different from subjective feeling that the skin is cold. Therefore, the view that temperature or menthol concentration is transformed into cold sensation by coding-decoding system is not reasonable.

TRPM8 is thermostat of skin temperature
In contrast, the present study shows that TRPM8 is polymodal comparator of temperature, menthol concentration and icilin concentration. The present experiments show that output of TRPM8 induces responses to warm the skin, which indicates that TRPM8 acts as thermostat of the skin temperature against cooling (Fig. 5).
TRPM8-induced afferent impulses are drive signals, which do not convey a code for skin temperature to target neurons. Then, how is low skin temperature related to cold sensation? TRPM8induced impulses should excite a cerebral target neuron involved in cold sensation, which triggers behavior to warm sole skin. Thus, we assume that information to induce cold sensation is innately stored in a target neuron. If not stored, cold sensation would not be elicited even when a target neuron is excited by arriving impulses. Electric stimulations of local areas of the somatosensory cortex in conscious persons induce feeling that various spots of the whole-body skin are touched [33]. This famous experiment supports the present assumption that information to induce cold sensation is stored in a target neuron. Cell theory states that a cell is the basic unit for structure and function of all plants and animals [34], which gives the basis for the modern cell biology [35]. This cell theory also supports the present assumption that information to induce cold sensation is stored in a target neuron, because cold sensation is a critical function to survive in low ambient temperatures. These explanations are also applicable to menthol- [36,37] and icilin-induced cold sensation [38].
In autonomic regulation, when room is cooled deeply, the skin temperature is severely decreased in both mice (Figs. 4), as if regulation of the skin temperature is not functioning. This is an unexpected finding that TRPM8-induced autonomic responses are less powerful to maintain the skin temperature against cooling. To maintain skin temperature against cooling (Fig. 4), TRPM8induced thermogenesis should warm the core drastically. However, an excessive rise in core temperature is harmful for mice. Thus, TRPM8-induced autonomic regulation of skin temperature has limited power in nature.
Many investigators have been interested in homeostasis of core temperature in mammals [1,3,4,8]. When room temperature is decreased from 27 to 10uC (Fig. 4D), skin temperature is decreased from 34.2uC to 24.1uC but core temperature is maintained near 36.5uC in both mice. The difference between core temperature and room temperatures is as high as 26uC. Thus, there should be thermostat working for homeostasis of core temperature. Intraperitoneal injection of 59-AMP induces torpor-like response, in which core temperature is reduced from normal level to ambient temperature [39]. This suggests that autonomic thermoregulatory system of core temperature is a device inhibited by 59-AMP. However, the thermostat of core temperature remains unresolved.

Ratiometric calcium imaging
Doxycycline-inducible TRPM8-expressing HEK293 cells were established by Flp-In T-Rex system (Clontech, La jolla, CA), as described previously [40]. Briefly, TRPM8-expressing cells on a collagen-I-coated coverslip were loaded with 2 mM Fura-2 AM. After washing with Krebs' solution, these cells on a coverslip was placed on a recording chamber mounted on the stage of fluorescence microscope (BS50WI, Olympus). With a digital image analysis system (AQUACOSMOS; Hamamatsu Photonics, Hamamatsu), intracellular Ca 2+ ion concentration was measured by Fura-2 fluorescence ratio (F 340 /F 380 , excitation 340/380 nm and emission 510 nm), and displayed as TRPM8 activity.

Heat-seeking behavior
A plastic cage (16 cm 627 cm) was placed in the incubator. Floor of a cage was divided into two zones. One half (

Temperature measurement and stimulation
Mice were implanted with E-mitter probe (minimitter, Bent, OR) in peritoneal cavity under pentobarbital anesthesia (70 mg/ kg). 7 to 14 days after implantation, abdominal temperature was measured as core temperature. Mean temperature of the whole back skin was detected with thermography (Thermotracer TH3100MR, NEC San-ei Instrum, Kyoto, Japan) as skin temperature. By using anesthetized mice (n = 5), skin temperature of a thermography (15-33uC for room cooling experiment and Figure 5. Schematic representation of the present findings. Cooling-sensitive TRPM8 is directional comparator of skin temperatures, inducing output (y) when DT ( = T -28.4uC) is negative (Fig. 1). If negative DT-dependent output (y) induces thermoregulatory responses to warm the skin recursively, negative feedback loops having 28.4uC as set-point temperature are formed and TRPM8 functions as thermostat for minimizing DT against cooling. In behavioral regulation, when room is cooled, TRPM8 induces behavior to move to heated floor to warm the sole skin, which effectively maintains temperature of the sole skin against cooling. In autonomic regulation, when room is cooled, TRPM8 induces a small rise in core temperature, but skin temperature is severely decreased, as if regulation of skin temperature is not functioning. doi:10.1371/journal.pone.0017504.g005 30-35uC for menthol application experiment) was calibrated with that of a thermocouple fixed on the back lumber with a tape.
Free-moving mice were put in an incubator (CRB-41, HITACHI; 62 cm 650 cm 6132 cm) with a built-in regulator, and temperature in the incubator was used as room temperature. For thermal stimulation, room temperature was decreased from 27uC (neutral temperature) to 10uC. Room temperature was sampled every 10 min.
For menthol application, mice were anesthetized with 5% isoflurane for 30 sec. Menthol (l-menthol 10% w/v (640 mM), Wako, Japan) in ethanol was applied once to their whole trunk skin with paper (Kimwipes 665 cm) soaked with 600 ml solution of menthol.

Statistical analysis
Data were represented as mean 6 SEM. Comparisons of mean values between two groups were performed by one-tailed t-test. Statistical significance level, a, was set as 0.01 (*: p,0.01).