For the behavioral experiments, we used adult TRPM8^−/− (Dhaka et al., 2007) which were backcrossed for six generation on C57BL/6J background and obtained from pairs of homozygous TRPM8^−/−, and TRPA1^−/− (Kwan et al., 2006), back-crossed on 13 generations on C57BL/6J mice. From both we crossed and bred DKO and we used one common age- and sex-matched control group obtained from the DKO crossing and named C57BL/6J, because backcrossing were of the C57BL/6J strain purchased from Charles River (Sulzfeld, Germany). All mice were between 60 and 85 days; transgenic mice were genotyped according to our previously published procedures (Vetter et al., 2012, 2013). For the skin-nerve recordings of DKO shown in Figures 5, 6 we used 22 adult mice (4 females and 17 males). For the recordings shown in Figure 4 and the recordings from TRPM8^−/− nociceptors, please refer to the details outlined in Vetter et al. (2013). The data underlying Figure 6 were recorded from five adult male TRPM8^−/− and three adult male C57BL/6J mice and include the recordings of C57BL/6J CC-fibers shown in Toro et al. (2015). Recordings from 129S1/SvImJ mice were subjected to new analysis but were all from Zimmermann et al. (2011). A-fiber nociceptors in C57BL/6J mice were new and recorded from 22 adult mice (9 females and 13 males).

We utilized the large and high thermal resolution ring configuration of our previously introduced circular temperature gradient assay (Touska et al., 2016). We quantified cold avoidance in TRPM8^−/− and DKO in the temperature range from 30 to 5°C. The data were from the same mice previously measured in the 15–40°C environment (Touska et al., 2016). During these experiments, all mice were adapted to the equipment on day 1 with the ring adjusted to room temperature for 30 min. Mice were measured for 60 min on day 2 using 15–40°C and on day 3 using 5–30°C. To achieve a symmetric gradient with a midpoint temperature of 17.5°C, we air-conditioned the room to 17–18°C, otherwise the room was heated to 26°C using a convector heater. Behavior was videotaped with a CCD-camera and analyzed with our custom-designed software described and validated in Touska et al. (2016). The protocol for in-vivo experiments in animals was reviewed by the local animal ethics committee (University of Erlangen) and approved by the local district government.

We performed electrophysiological recordings as previously described (Zimmermann et al., 2009, 2011; Vetter et al., 2013; Toro et al., 2015). Briefly, single-fiber activity was recorded from teased fibers via platinum or gold wire electrodes in an adjacent chamber overlaid with paraffin oil. Receptive fields were isolated from the surrounding fluid by means of a teflon ring. Cold sensitivity was tested by superfusion of the receptive field with precooled synthetic interstitial fluid (SIF). Well-defined cold stimuli were realized with a custom-designed counter-current temperature exchange system and they reached in nociceptors from bath temperature (30–32°C) to 8.3 ± 2.8°C and in thermoreceptors from bath temperature (33–35°C) to 7.3 ± 2.4°C in 60 s. Cold nociceptors were identified by searching mechanosensitive receptive fields with a glass rod and isolation of the receptive field with the teflon ring. Specifically, we recorded populations of C-fibers from DKO, TRPM8^−/− and A- and C-fibers from C57BL/6J. Fibers from several previous studies were analyzed retrospectively to specifically characterize the properties of TRPM8+ and TRPM8− cold nociceptors in the C57BL/6J and the 129S1/SvImJ mouse strains. Thermoreceptors were identified by splitting nerve bundles until C-fiber activity was noted and at least one and maximal five mechanosensitive receptive fields could be identified on the skin. This served to adjust the amplification range for each and every split to a level that undoubtedly detects C-fiber activity with a high enough signal-to-noise ratio. This is crucial, because the thermoreceptors stem from the smallest diameter dorsal root ganglia neurons and are expected to require close contact with the recording electrode. Subsequently, to recognize cold sensitive spots, a small ice cube (1 cm³) was slowly moved very closely over the corium side of the skin-nerve preparation without touching the receptive areas of the skin. Thermoreceptor activity can be recognized by immediate onset brisk firing at a high rate that ceases shortly after the ice cube is taken away from the respective area. To account for a faster equilibration of the bath temperature, we increased the flow and the bath temperature to 33–35°C. The respective area is then isolated with a teflon ring and the receptive field is subjected to controlled thermal stimulation (Zimmermann et al., 2009). Adjustments of the ring location are made if necessary.

Comparisons between two groups were performed using T-test for paired or unpaired samples, respectively; for more than two groups one-way ANOVA' planned comparison with Fisher LSD post-hoc test were computed using IBM SPSS Statistics version 21. Differences were considered significant at p < 0.05 and are marked by asterisks, hashtags, or dollar signs as indicated in figure legends. For skin nerve data we used the Grubbs's outlier test to identify outliers. Outliers were marked in Figures 4, 5 with stars. The outlier values were not included in the calculation of the mean and quartiles and were also removed for the statistical calculation. Error bars in figures are displayed as SEM or SD, as indicated in figure legends. In figures “^***” symbolizes p < 0.001, “^**” signifies p < 0.01 and “^*” represents p < 0.05.

In our previous study, we measured the thermal preference behavior of TRPM8^−/− and DKO in a thermal gradient assay equilibrated between 15 and 40°C. We found that TRPM8-deficient mice show a remarkable lack of cold avoidance in the assay configuration with low thermal resolution and steep gradient (0.5°C/cm; 3.6°C between individual fields), but nevertheless they still do recognize warmer zones as preferable to colder areas, but they require longer to locate to the warmer zones. Remarkably, in an environment with a shallower gradient and increased thermal resolution (0.3°C/cm, 2.3°C between fields), the lack of TRPM8 is well-compensated and the cold avoidance deficit is no longer apparent. Here, the only difference to the wildtype is a broader range of preferred temperatures visible in a less negatively skewed distribution in the last 15 min of the acquisition (Touska et al., 2016). Remarkably, additional lack of TRPA1 created a similarly large lack of cold avoidance in the DKO as previously recognized in the TRPM8^−/− strain in the smaller ring with the steep gradient. To compare these data with data acquired in a 5–30°C environment of the large ring with the shallow gradient, we calculated a cumulative response function for the early, explorative behavior in the first half hour (Figure 1A) and the late, thermal preference behavior (Figure 1B) acquired in the last 15 min. Remarkably, in the 15–40°C environment, the DKO show random exploratory behavior below 32°C devoid of any cold avoidance (p > 0.2), but intact avoidance of the warmer fields >34°C (p = 0.03, p = 0.000001, and p = 0.00004). In contrast, the TRPM8^−/− are indifferent from wildtype. In this temperature range, with a minimum temperature of 15°C, and with this thermal resolution, obviously, presence of either TRPA1 and TRPM8 is sufficient to produce acute cold avoidance, because lack of each single channel does not make any cold avoidance deficits apparent—in the case of TRPA1, mice are even more sensitive to cold than control mice (at 32°C, Figure 1A and Touska et al., 2016). Cold avoidance vanishes only when both channels are deficient. With longer exposure, i.e., in the last 15 min of the 1 h observation, in the DKO, a third, probably slow onset cold detection mechanism, primary afferent or central, unfolds and a considerable amount of cold avoidance becomes apparent although the DKO still have an outstandingly large tolerance for colder ring areas; they spend significantly less time than C57BL/6J between 26 and 34°C and, when compared to TRPM8^−/−, this concerns the entire temperature range between 15 and 34°C (Figure 1B). When we compared the early and late behaviors, as illustrated in Figures 1A,B, we found it to be different in all strains for the temperatures below 28°. These findings underline the assumption that time is a significant factor in a temperature gradient setup (Table 1) and that the early and late bins should be regarded separately.

We next analyzed the behavior of the same groups of mice in the circular running track with temperatures between 5 and 30°C (Figures 1C,D). Remarkably, exploratory behavior was guided by cold avoidance in all strains. Yet, the DKO showed least cold avoidance and were significantly different from TRPM8^−/− at all temperatures (p < 0.05, ANOVA), except for 30°C (p = 0.092, ANOVA). When compared to the wildtype littermates, TRPM8^−/− showed less cold avoidance only >24°C and the cumulative response function for TRPA1^−/− seemed shifted to higher temperatures, but this was significant only for 30°C (p < 0.0001, ANOVA; Figure 1C). In the last 15 min of the 1 h observation, the DKO showed a large tolerance for colder ring areas and they occupied all fields ≤29°C longer than TRPM8 or control mice (Figure 1D). For all temperatures ≤24°C TRPM8^−/− were again indifferent to control mice but they showed—similar to the 15–40°C environment—a broader range of thermal preference spending significantly less time at temperatures above 27° (Figure 1D). Comparison of the early and late behaviors was different in all strains with higher levels of significance than observed for the 15–40°C environment (Figures 1C,D and Table 1, bottom columns).

The preference temperature time course shown in Figure 2A further illustrates the large differences in the cold avoidance phenotypes of TRPM8^−/− and DKO. Virtually all timepoints of the timecourse of DKO locate to significantly lower weighted preferred temperatures in comparison to both TRPM8^−/− and wildtype (Figure 2A). In addition the DKO had significantly larger SD, even in the last 15 min of the measurement, than wildtype and TRPM8^−/−, which is indicative of a more erratic behavior and a larger tolerance for the lower temperatures (Figure 2A and Table 1). Again, similar to the 15–40°C environment, TRPA1^−/− appeared to be more sensitive to cold and seemed to recognize the warmer areas faster (not shown in Figure, compare parameters listed in Table 2). In the final 15 min of the experiment, TRPM8^−/− located at 1.2°C lower than wildtype (p = 0.2, ANOVA, Table 2), but only the DKO displayed a significant reduction of preferred temperature choosing 4.3°C lower than wildtype and 3.1°C lower than TRPM8^−/− (p = 0.0001, ANOVA, Table 2).

A more detailed view of the evolution of the cold avoidance phenotype over time provides the histogram chart of the first and last 15 min bins. It becomes apparent that in the first 15 min DKO avoided the colder zones ≤15°C much less than both wildtype and TRPM8^−/− (Figure 2B, lower panel). In the last bin of measurement time, the phenotype became more obvious to include all ring areas ≤24°C. Much in contrast, both TRPA1^−/− (not shown) and TRPM8^−/− showed adequate cold avoidance for these ring zones and were insignificantly different from wildtype. The lack of TRPM8 distinctively affected only temperatures above 24°C, where the TRPM8^−/− spent less time on the 30°C field and located instead rather between 25 and 28°C. Of all strains, the DKO spent the least time on the 30°C field (Figure 2B, upper panel). The temperatures of the largest phenotypic difference between TRPM8^−/− and wildtype was the range of 28–30°C and for the DKO and wildtype the range reached from 24–30°C. Therefore, we calculated time courses of thermal selection for these two surface sections which cover the warmest 25 and 8% of the ring area. For both time courses, DKO and TRPM8^−/− are significantly segregated at almost all time points, while the TRPM8^−/− displayed its largest differences to the wildtype for discrimination of 28°C. In numbers, the C57BL/6J mice required 16 min to spend at least 2/3rd or more of their time >24°C, the TRPM8^−/− required 22 min while the DKO never showed such a large selection for the temperature >24°C. The differences between genotypes became larger when regarded for the temperature >28°C, where C57BL/6J mice required 34 min to spend 2/3rd or more of their time and both TRPM8^−/− and DKO never showed such a large preference (Figures 2C,D).

Last but not least, we analyzed the general activity or locomotion of the strains in the ring assay as amount of zone transitions in the warm and cold thirds (Figure 3). For the cold third this includes the large field that is required for automated analysis (Touska et al., 2016) which is why counts include transitions between five fields, while for the warm third the data includes transitions between six fields (see sketches in Figure 3). We therefore compared transitions in cold and warm thirds between strains and between the 15–40°C (Figures 3A,B) and 5–30°C environments (Figures 3C,D). In all strains, and in both environments, the number of transitions was higher in the first half as compared to the second half of the measurement. In the 5–30°C environment, remarkably, TRPA1^−/− had the least number of zone transitions in the cold third and avoided entering the cold areas in the second half of the experiment (p < 0.0001, 41.7 early and 2.5 late transitions in the cold third; Figure 3C). While TRPM8^−/− (73.1 early and 18.5 late transitions in the cold third) were remarkably similar to C57BL/6J (66.5 early and 19.5 late), DKO showed more than twice as many zone transitions (154.3 early and 90.5 late) and, in contrast to all other strains, the cold avoidance was not much increased in the second half hour (p < 0.001, paired T-test; Figure 3C). With respect to the zones >22°C, again DKO showed the largest number of transitions in both the early and late halves (early: p < 0.0001 vs. TRPA1 and p = 0.02 vs. TRPM8^−/−; late: p < 0.001 vs. TRPA1, p = 0.001 vs. TRPM8, and p = 0.047 vs. C57BL/6J, ANOVA) while TRPA1^−/− moved significantly less than C57BL/6J (p = 0.048 ANOVA; Figure 3D). When comparing the 15–40°C and 5–30°C environments, we found that the DKO behaved indifferent in the respective cold thirds, but different in the warm thirds for both early (cold: p = 0.8 and warm: 0.002) and late (cold: p = 0.3 and warm: 0.009). Remarkably, TRPA1^−/− had significantly less zone transitions in the 5°C cold third as compared to the 15°C cold third for both early (p = 0.0004) and late (p = 0.02, Student's t-test; Figures 3A,C).

These large behavioral phenotypic differences of the DKO made apparent that lack of TRPA1 may cause larger deficits in primary afferent cold detection than previously assumed. To define the exact primary afferent electrophysiological correlates contributed by TRPA1 and TRPM8, we quantified cold responses of cold-sensitive fiber types in the mouse saphenous nerve for TRPM8-positive and TRPM8-negative fibers of C57BL/6J, TRPM8^−/−, and DKO.

We first assessed cold nociceptors (C-mechano-cold, CMC, and C-mechano-cold-heat, CMCH) and revisited our previous sampling studies performed mainly in the background C57BL/6J strain, but also in littermates of TRPM8^−/− (Vetter et al., 2013) where a distinction was made between TRPM8-positive and TRPM8-negative cold nociceptors based on pharmacological evidence. In the respective studies, all fibers were searched randomly using a glassrod and the effects of menthol 50 or 500 μM, camphor 2 mM and XE991 10 and 100 μM on the cold response were quantified after previous superfusion of the receptive field for 4–5 min with these compounds at bath temperature. All named compounds sensitize to cold only in presence of a functional TRPM8 receptor (Zimmermann et al., 2011; Vetter et al., 2013; Toro et al., 2015). One hundred and eight C-mechano fibers were included and subjected to a new analysis where the cold sensitive fibers (>3 spikes per 60 s cold stimulus) were segregated into the two populations. In contrast to our previous publications, we classified fibers now as CMC if they had zero spikes during a heat ramp and as CMCH if they had at least one spike during the heat stimulation, discharged close to or at ramp's peak, at least >44°C. In CMC fibers, a burst in the middle of the heat ramp around 40–44°C is interpreted as paradoxic heat response. With these criteria, 19 CMC and 29 CMCH were identified and, thereof 15 CMC and 3 CMCH were positive for a functional TRPM8. When we characterized the properties of the cold responses in these four groups, we found that TRPM8-positive CMC fibers (n = 26) generate 2.2-fold larger responses and 2-fold higher peak firing rates than TRPM8 negative (n = 4) CMCs (see Table 3; p > 0.05, Student's t-test). In CMCH-fibers, TRPM8 caused a 3.5-fold larger response and increased the peak discharge by 1.2-fold (p < 0.001 Student's t-test; compare Table 3). Irrespective of the fiber type, the difference a functional TRPM8 makes to a cold nociceptor's response reflects in an almost 4-fold difference in response magnitude (24.7 ± 17.5 vs. 6.4 ± 3.5, p < 0.001, Student's t-test, n = 30 each, Figures 4A,B) and leads to higher peak (3.4 ± 2.8 vs. 2.0 ± 3.5, p = 0.03; Figure 4C) and mean discharge rates (0.8 ± 0.6 vs. 0.5 ± 0.4, p = 0.003). The presence of TRPM8 has also unique effects on the shape of the response as it enables the fibers to produce a sustained static response with much less adaptation in contrast to TRPM8-negative fibers. With TRPM8 present, more than 84% of the response is discharged within seconds 16–60 of the stimulus, where the temperature has already reached below 12°C and is only changing at a very slow rate. The static response is reduced to 55% in absence of TRPM8 (p = 0.00005, Student's t-test, n = 30 each, Figures 4A,D). Interestingly, in the C57BL/6J strain, TRPM8-positive nociceptors have 5°C lower threshold temperatures and started to discharge when the temperature reaches 18.4°C ± 4.7, while the TRPM8-negative nociceptors activated already at 23.5°C ± 5.7 (p = 0.0001, Student's t-test; Figure 4E).

We compared the basic fiber types and their encoding characteristics to the 129S1/SvImJ mouse strain, because we wanted to know how well they were maintained across inbred strains. Both strains are evolutionary distant and belong to different groups, the Castle's mice and the C57/58 group, of the mouse family tree (Petkov et al., 2004). This or other 129-related strains are frequently used as donor strain for the generation of transgenic mice used in studies of the sensory system, including TRP-channel knockouts and backcrossing over variable generations involves a variable mixture of the genomes of both strains. In the past, assumptions about primary afferent differences in these strains were discussed as possible influence on the lack of a warm-sensing phenotype of transgenic mice lacking TRPV3 (Huang et al., 2011). The respective parameters of these fiber types for these two strains are compared in Table 3. In these strains the overall amount of cold nociceptors in the two samples was rather similar and added to 44% in C57BL/6J and 40% in 129S1/SvImJ. Interestingly, the occurrence of the fiber types CMC and CMCH was inverse, although functional TRPM8 was present in a comparable fraction of cold-sensitive fibers (p > 0.05, Chi-Square-test); remarkably, in the 129S1/SvImJ strain, TRPM8-positive nociceptors had almost 6°C higher threshold temperatures and started to fire at 25.5°C ± 3.8 (n = 8) than the TRPM8-negative nociceptors which had thresholds at 19.9°C ± 5.3 (n = 6, p = 0.04, Student's t-test; Table 3) which is also inverse to the C57BL/6J strain. Other parameters, such as the magnitude of the response and peak discharge rates were insignificantly different between both strains (p = 0.23 and p = 0.53, Student's t-test, Table 3).

In two previous sampling studies conducted in TRPM8-deficient mice, which were both guided by a mechanical search stimulus, we recorded from 54 mechanosensitive C-fibers (Vetter et al., 2013) of which 16 were found to be cold sensitive. Virtually all of these fibers were CMCH and the magnitude, mean and peak discharge and the activation threshold temperature of the cold response in these fibers was indifferent from the cold response pattern of the TRPM8-agonist insensitive C57BL/6J-based group (p > 0.5, Student's t-test, Table 3, Figures 5A–E). In addition, we searched for cold sensitivity in the knockouts by slowly passing an ice cube over the skin—a search stimulus we use to identify mechanoinsensitive cold receptors as described in detail in Section Methods (Zimmermann et al., 2011). In TRPM8^−/− we were able to identify another population of mechanosensitive fibers with larger cold responses than the CMCH. Their response pattern was characterized by a high threshold temperature around 30°C and a vivid, dynamic cold response. Remarkably, these cold responses had a comparable magnitude as the TRPM8+ cold nociceptors in the wildtype (p = 0.8, Student's t-test, n = 7), but higher peak firing rates (p = 0.02, Student's t-test). In addition the static response was reduced to as much as 23% as compared to the TRPM8+ cold nociceptors (p < 0.001, Student's t-test, Table 3, Figures 5A–E. We located additional four receptive areas where at least two fibers responded vividly to cold in a similar dynamic pattern as observed with the seven single receptive fields. These fibers could not be separated to single fiber level and were therefore not included in the analysis, but they further illustrate that this type of cold response appears in relative abundance in the TRPM8^−/−.

We performed a new sampling study guided by a mechanical search stimulus in the DKO. We recorded from 38 mechanosensitive C-fibers of 15 DKO mice. In this sample, we identified six cold sensitive fibers (see Table 3), which is about half as many as we found in TRPM8^−/− (Figure 5F). We also encountered 11 receptive fields were more than one fiber responded to the thermal stimulation and four of these multi-fiber recordings discharged in response to cold. Because these fibers could not be separated to single fiber level they were not included in the analysis of the cold response properties shown in Figures 5A–E. In additional eight preparations from seven mice, we used ice cubes and located additional eight receptive fields with cold sensitivity. However in none of them, we found responses similar to the properties of the mechanosensitive dynamic cold nociceptors identified in TRPM8^−/− and shown in Figure 5A. All cold nociceptors in DKO, except one, had low von Frey thresholds and belonged to the group of multimodal cold nociceptors. Table 3 illustrates the properties of these fibers, which were indifferent in all characteristics (response magnitude, peak discharge, and static response, p > 0.3) from the regular cold nociceptors identified in TRPM8^−/−. Apparently a considerable amount of cold nociceptors remains active in DKO and produces remarkable responses although they are increasingly difficult to find.

In the mouse, unimodal cold receptors were first characterized in the cornea to encode constant temperatures around 34°C with continuous, graded impulse activity and small temperature reductions with rapidly accelerating impulse activity. Corneal cold thermoreceptors are sensitive to menthol and virtually absent in the TRPM8-deficient mice (Parra et al., 2010). Depending on the subtype, they exhibit cooling thresholds around 32 or 28° and silence at 23 or 17°C, respectively (Gonzalez-Gonzalez et al., 2017). We first characterized a similar fiber type in skin-nerve preparations of 129S1/SvImJ mice (Zimmermann et al., 2011), albeit with a dynamic range that includes noxious temperatures. In the C57BL/6J strain, when compared to TRPM8-positive cold nociceptors, unimodal cold receptors have 16-fold larger responses, 18-fold higher peak, and 23-fold higher average firing rates (Figure 6A, Tables 3, 4). These properties are similar to the 129S1/SvImJ strain (Table 4). In the skin CC-fibers cover a large dynamic range with thresholds between 37 and 15°C and impulse activity encoding noxious temperatures below 10°C. Only a small fraction CC-fibers showed silencing above 17°C as it is regularly observed in the cornea (n = 1 of 22, marked with arrowhead; Figure 6B, Table 4). Menthol was found to show an unequivocal sensitizing effect only in a minority of fibers with high threshold temperature as outlined in supplement of Zimmermann et al. (2011); in many fibers menthol application caused a decreased response with reduced peak firing rate and increased adaptation with some cold responses being entirely blocked (Zimmermann et al., 2011; Toro et al., 2015). Given the encoding deficit of CC-fibers in TRPM8-deficient mice (Toro et al., 2015), the lack of apparent sensitization is not a lack of effect of menthol, but likely due to excessive depolarization and inactivation of sodium channels.

To estimate the relative frequency of these fibers in the saphenous nerve, we counted the number of cold spots (using ice cube test) per thinly split nerve bundle. Cold spots where then isolated with the Teflon ring and systematically evaluated whether they were single fibers (for details see Section Methods). In the C57BL/6J mice, we tested 54 bundles (see Table 4, Figure 6C). In 10, we identified single or multiple unimodal cold spots by slowly passing small ice cubes close to the skin surface. None of them were sensitive to application of high mechanical pressure with the glass rod. In TRPM8^−/−, we tested 106 bundles in preparations from four mice and found only 3 cold spots. They were all sensitive to mechanical stimulation, but shared more similarity with CC-fiber responses than to the previously described multimodal nociceptors. In DKO mice we tested 106 bundles from seven mice and identified one CC-fiber like response; also this fiber was mechanosensitive, albeit with a very high mechanical threshold. Presumably the relative frequency of cold spots is strongly reduced in the absence of TRPM8 and, TRPA1 may be involved in the encoding of some of the remaining responses. In addition to the reduced incidence, absence of TRPM8 reduced response magnitude, mean, and peak discharge to one-fifth of the average wildtype cold spot response (Figures 6D,E) and further increased adaptation (Figure 6F) and led to a different threshold temperature (Figure 6G).

The pattern of encoding cold stimulus intensity in murine Aδ-fibers is different from that in C-fiber mechanocold nociceptors and from Aδ-fibers in cats or primates; this makes the significance of this fiber type for cold temperature sensing in the mouse somatosensory system difficult to judge. Cold-sensitive A-fibers in both the C57BL/6J and the 129S1/SvImJ strain are infrequent. In contrast to this appears the fact that stimulation with temperatures below 0°C excites all Aδ-fibers in the rat (Simone and Kajander, 1997). With a stimulus temperature ramp between 30 and 5°C, we found that Aδ-fibers have smaller cold responses, lower peak and mean firing rates as compared to the C-fibers and menthol has an almost negligible sensitizing effect. Aδ-fiber cold responses also cover a variable range of thresholds (Table 5). In addition, while cold responses in C-fibers nociceptors are steady and reproducible (Zimmermann et al., 2009; Vetter et al., 2013), in A-fibers repeatedly applied cold stimuli in intervals of 3–5 min do often create highly variable responses or are not reproducible. In Table 5, the encoding properties are therefore enlisted as average of two consecutive cold stimuli. In our C57BL/6J sample of 47 A-fibers, 25% (n = 12) of the fibers responded to cold, but only 2 (17%) showed sensitization in response to 100 μM menthol, i.e., increased their cold response or peak discharge rate after application of menthol. Menthol may also act through TRPA1 and a desensitizing effect due to depolarization shunt may also be a sign of TRPM8 or TRPA1 expression, but the small responses and the lack of reproducibility, make it difficult to investigate molecular mechanisms. Nevertheless, in the 129S1/SvImJ strain a similar small fraction of cold responses were sensitized by 500 μM menthol (13%) and the A-fibers had similar properties (Table 5). Because of these difficulties, we did not search for this type of cold nociceptor in the knockout strains.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.