Nociceptive afferent phenotyping reveals that TRPA1 promotes cold pain via neurogenic inflammation upstream of the neurotrophic factor receptor GFRα3 and the menthol receptor TRPM8

1. Introduction

The Transient Receptor Potential Melastatin 8 ion channel (TRPM8), a receptor for cooling compounds such as menthol, is the principle mammalian cold receptor on peripheral somatosensory afferents [35]. Similarly, the Transient Receptor Potential Ankyrin 1 (TRPA1) channel is a receptor for exogenous and endogenous irritants and proalgesics, but is also a candidate noxious cold transducer [47]. However, if TRPA1 or the afferents expressing this channel directly contribute to cold pain in vivo is controversial [40]. Specifically, reports on the behavioral phenotype of TRPA1-deficient mice have been inconclusive with some studies finding deficits in the acute cold behaviors tested, whereas others have not [5; 7; 24; 25; 27]. Cold hypersensitivity associated with inflammatory and neuropathic pain are reduced in TRPA1-nulls, and cold responsiveness is potentiated by exogenous or endogenous TRPA1 agonists in vivo [16; 17; 19; 38; 50]. Thus, under pathological conditions that upregulate endogenous channel modifiers, TRPA1 may augment cold sensitivity, consistent with its role as a “gatekeeper” for chronic pain [6; 50].

TRPA1 and TRPM8 are non-selective cation channels that permeate small ions, as well as larger positively-charged molecules [23; 44]. The permeation of large cations via these channels has been used to selectively target local anesthetics to specific afferents subtypes, allowing their functional phenotyping in vivo [44]. For example, co-injection of the positively-charged and cell-impermeant sodium channel blocker QX-314 (N-ethyl-lidocaine) with agonists for the Transient Receptor Potential Vanilloid 1 (TRPV1), a noxious heat-gated channel [14], have been used extensively to phenotype pain and itch afferents [8; 45]. Similarly, we have shown that blocking nerve conduction via TRPM8 channels produces inhibition of cold and cold pain in mice [33; 39].

As TRPA1 also permeates QX-314 [36], we sought to determine if nerve conduction block of TRPA1 afferents alters cold sensitivity in mice. Surprisingly, we find that only robust activation of TRPA1 leads to a QX-314-dependent inhibition of cold behaviors which, in contrast, are unperturbed in TRPA1-deficient mice. Similar results were produced with capsaicin. As stimulation of TRPA1 or TRPV1 induces neurogenic inflammation, we reasoned that robust channel activation, even in the presence of QX-314, releases proalgesics that may act on TRPM8 channels to promote permeation of QX-314. In support of this hypothesis, we find QX-314 inhibition of cold is neurogenic and mediated sexually dimorphically by either the pro-inflammatory calcitonin gene-related peptide (CGRP) or signaling via Toll-like Receptor-4 (TLR4). Our data shows that both function upstream of the glial cell line-derived neurotrophic factor receptor GFRα3 and its specific neurotrophin artemin, which we previously showed produces cold allodynia in a TRPM8-dependent manner [29; 30]. Thus, our results indicate that activation of nociceptive afferents neurogenically promotes the release of artemin, which in turn sensitizes TRPM8 via GFRα3. Increased TRPM8 channel activity permeates QX-314 and blocks of nerve conduction in TRPM8 afferents. In support of this model, we find that increased cold sensitivity induced by TRPA1 activation is GFRα3-dependent, suggesting a mechanism whereby TRPA1 acts upstream of artemin/GFRα3 signaling to promote cold sensitization via stimulation of TRPM8 afferent fibers.

2. Experimental procedures

2.1. Animals

All experiments were conducted in compliance with the University of Southern California Institutional Animal Care and Use Committee and in accordance with National Institutes of Health guidelines. Mice greater than 8 weeks old were used in all experiments and equal numbers of male and female mice were used unless otherwise indicated. All mice had access to food and water ad libitum and were maintained on a 12 hour light and dark cycle at room temperature, with all experiments performed during the light cycle. TRPM8, TRPA1, and TRPV1 null mice were purchased from Jackson Laboratories and were in the B57Bl/6 background unless otherwise specified. Gfrα3 null mice are maintained in house and heterozygous knockout mice were bred to obtain wildtype littermate controls. The TRPM8 antagonist PBMC was used to validate the role of TRPM8 instead of TRPM8 nulls mice as these animals are basally cold-insensitive (see [39]).

2.2. Pharmacologic agents

WS-12, QX-314, allyl isothiocyanate, cinnamaldehyde, and capsaicin were purchased from Sigma Aldrich. Artemin was purchased from R&D Systems, CGRP(8–37) receptor antagonist from Tocris, TRPM8 antagonist PBMC from Focus Biomolecules, and TLR4 antagonist TAK-242 from Calbiochem.

Ethanol was used as vehicle throughout this study except for injections of CGRP(8–37), TAK-242 and artemin in which 0.9% saline was used as the vehicle. All injections were at a volume of 20μl and intraplantar in one hind paw for all experiments, except for Cold Plate where both fore paws were injected with 10μl, as previously described [30; 39]. Amounts injected of each chemical were: cinnamaldehyde 120ng and 50μg, capsaicin 1μg, allyl isothiocyanate 0.15%, CGRP(8–37) 6.4mmol, TAK-242 0.2μg, WS-12 20μg, PBMC 100μg, artemin 0.2μg, and QX-314 1%.

2.3. Cold Plantar Assay

The Cold Plantar assay was used to measure innocuous cold sensitivity as previously described [30; 39]. Mice were habituated in plexiglass chambers for 2 hours on a surface maintained at 30°C unless otherwise stated. Initially a 20 second cut off was used to limit potential damage (Figs. 1–2), however later changed to 30 seconds to observe the full effects of QX-314 mediated block (Figs. 3–7). Mice were returned to the Cold Plantar surface after injection and assayed at 30-, 60- or 90-minutes post-injection. CGRP(8–37) and TAK-242 were given as pretreatment 30 minutes prior to QX-314/agonist injections. PBMC pre-treatment was administered 3 hours before QX-314/agonist injections. In experiments that utilized habituations post-injection at 10°C, 24°C, or 36°C, mice were placed on an ITTC chamber set at the desired temperature for 30 minutes, returned to the 30°C Cold Plantar surface an additional 30 minutes, then latencies were recorded.

Figure 1: Cold sensitivity is attenuated by presumptive conduction block of TRPA1 positive sensory afferents.

(A) Latencies to lift measured by the Cold Plantar assay in wildtype and Trpa1^−/− mice were similar at adapting temperatures of 24° and 30°C. (n=6, ^nsp>0.05 for 24° vs. 30°C and for wildtype vs. Trpa1^−/− at either temperature; t-test with multiple comparisons with Bonferroni-Dunn method). (B) Lift latencies on the Cold Plantar assay were unaltered 90 minutes post intraplantar injection of 1% QX-314 alone or QX-314 with 120ng (Cinn-low) of the TRPA1 agonist cinnamaldehyde (n=8, ^nsp>0.05; baseline (BL) vs. post-inj., paired t-test), but were increased with co-injection of QX-314 and 50μg of cinnamaldehyde (Cinn-high) or 0.15% allyl isothiocyanate (AITC, n=8, ***p<0.001; paired t-test). (C) There were no statistical differences between ipsilateral (Ipsi) and contralateral (Contra) responses in Trpa1^−/− mice injected with QX-314/cinnamaldehyde (n=8, ^nsp>0.05; unpaired t-test with Welch’s correction) in the Cold Plantar assay. (D) Mechanical behaviors were inhibited when QX-314 and cinnamaldehyde were co-injected (BL vs. post-inj, n=4, **p<0.01; paired t-test). (E) Heat-evoked behaviors were unaffected by co-injection of QX-314 and cinnamaldehyde (n=5, ^nsp>0.05; paired t-test).

Figure 2: Cold sensitivity is attenuated by presumptive conduction block of TRPV1 positive sensory afferents.

(A) QX-314 co-injection with capsaicin (Cap) induced significant increases in lift latencies on the Cold Plantar assay in wildtype but not Trpv1^−/− mice (n=6–8, ***p<0.001, ^nsp>0.05 BL vs. post-inj.; paired t-test). (B) Mechanical behaviors were inhibited when QX-314 and capsaicin were co-injected (BL vs. post-inj, n=4, ***p<0.001; paired t-test). (C) Heat-evoked behaviors were inhibited by co-injection of QX-314 and capsaicin (n=4, ***p<0.001; paired t-test). (D) Duration of cold inhibition on the Cold Plantar assay by co-injection of QX-314 with TRPA1, TRPV1, or TRPM8 agonists (n=8).

Figure 3: Nocifensive responses to noxious cold behaviors are not dependent on TRPA1 channels but are inhibited by presumptive block of TRPA1 afferents.

(A) Both wildtype (*p<0.05, n=5) and Trpa1^−/− mice (***p<0.001, n=8; paired t-test) exhibit increased nocifensive responses on a plate set to 0°C compared to 30°C. No differences were observed between wildtype and Trpa1^−/− mice at 0°C (^nsp>0.05 unpaired t-test). (B) Nocifensive responses on the 0°C Cold Plate during a 3 minute period measured 30 minutes post QX-314/Cinn (n=8) injection were significantly reduced in comparison with QX-314 (n=7, *p<0.05) alone; unpaired t-test with Welch’s correction t). (C) A noxious cold stimulus (10°C) induced QX-314-dependent inhibition of innocuous cold responses in Trpa1^−/− mice (n=8, ***p<0.001 ipsi vs. contra, paired t-test and vehicle vs. ipsi, unpaired t-test).

Figure 7: Cold allodynia induced by TRPA1 agonists requires GFRα3.

(A) Model for TRPA1-mediated cold allodynia/hyperalgesia via artemin/GFRα3 and TRPM8. (B) ARTN induced cold allodynia in both wildtype (vs. BL, n=8, ***p<0.001; paired t-test) and Trpa1^−/− mice (vs. BL, n=8, ***p<0.001; paired t-test). (C) Cinnamaldehyde injection induces decreased hind paw lift latencies in wildtype mice (vs. BL, n=7, ***p<0.001; paired t-test) but not in Gfrα3^−/− mice where a slight yet significant increase in latencies were observed (vs. BL, n=7, p=0.013; paired t-test) in the Cold Plantar assay. (D) WS-12 induced cold sensitization in both wildtype (vs. BL, n=4, **p<0.01; paired t-test) and Gfrα3^−/− mice (vs. BL, n=5, ***p<0.001; paired t-test).

2.4. Cold Plate Assay

Experiments to assess nocifensive responses to noxious cold were performed as previously described [26; 39]. Briefly, 3 minute videos were recorded after mice were placed on a Cold Plate apparatus (ITTC) set to 0°C or 30°C and quantified later for number of fore paw shakes or latency to lick. Videos were quantified and averaged by two individuals blind to the temperature of the plate and the injection parameters.

2.5. Von Frey Assay

Mechanical sensitivity was determined 60 minutes post treatment using the up-down von Frey method and 50% withdrawal threshold was measured with filaments as previously described [39].

2.6. Hargreaves Assay

Experiments were performed 60 minutes post treatment as previously described to assess responses to heat [39].

2.7. Statistical Analysis

All data represented as mean and +/− SEM error bars and statistical analyses were performed in Prism using the statistical methods described in the Figure Legends.

3. Results

3.1. Co-injection of QX-314 with TRPA1 agonists inhibits cold-induced behaviors in mice.

To determine the involvement of TRPA1 channels in cold sensation, we first determined if TRPA1-deficient mice (Trpa1^−/−) [5] exhibit a phenotype in the Cold Plantar assay, generally considered a measure of innocuous cold sensitivity [10; 11]. As previously reported, wildtype mice show a consistent hind paw latency to lift near 10–11 seconds regardless if they have been adapted to a surface temperature of 24° or 30°C (Fig. 1A) [11; 39]. Similar latencies were observed in Trpa1^−/− mice at either surface temperature and there were no differences between the two genotypes (Fig. 1A, p>0.05 24° vs. 30°C for Trpa1^−/−; wildtype vs. Trpa1^−/− at both T°C). Thus, TRPA1-deficient mice have normal basal innocuous cold sensitivity regardless of the habituation temperature.

Next, we sought to determine if selectively blocking nerve conduction in TRPA1 afferents using the cell impermeant, voltage-gated Na⁺ channel blocker QX-314 altered cold behaviors in mice. QX-314 permeates agonist-gated non-selective cation channels and produces cell-specific nerve conduction block [8], an approach used to functionally phenotype select afferent subtypes in vivo [39; 44]. We have shown that habituating mice at 24°C is a sufficient stimulus to engage TRPM8 channels to take up QX-314 and block cold responses [39]. Thus, unless otherwise stated, all studies were performed on mice habituated to a surface temperature of 30°C, a temperature demonstrated to prevent TRPM8-mediated uptake of QX-314 (Fig. 1B) [6]. Using single unilateral intraplantar hind paw injections, we initially tested a concentration (120ng; Cinn-low) of the TRPA1-specific agonist cinnamaldehyde shown to produce QX-314-dependent inhibition of mechanical hyperalgesia [28]. However this dose produced no change in innocuous cold behaviors (Fig. 1B), whereas a concentration of cinnamaldehyde (50μg; Cinn-high) that induces pain responses in mice [3] attenuated cold behaviors when injected with QX-314 (Fig. 1B). Similarly, the potent TRPA1 agonist allyl isothiocyanate (0.15%, AITC) [22] also inhibited cold behaviors when injected with QX-314 (Fig. 1B). The cold phenotype induced by the higher dose of cinnamaldehyde was TRPA1-dependent as it was absent in Trpa1^−/− mice (Fig. 1C). Consistent with the role of TRPA1 in mechanical but not heat pain [23; 41], cinnamaldehyde selectively inhibited only mechanical responses when injected with QX-314 (Figs. 1D,E). Taken together, robust activation of TRPA1 is required to induce QX-314-mediated block of cold in mice, a phenotype not observed in TRPA1-deficient animals.

3.2. Activation of TRPV1 channels produces QX-314-mediated cold inhibition.

The distinct phenotypes of mice lacking TRPA1 channels and those in which TRPA1 afferents are putatively inhibited prompted us to ask if stimulation of other nociceptive afferents produced similar effects. The heat-gated and nociceptive channel TRPV1 is expressed with TRPA1 and also permeates QX-314 when activated by its ligand capsaicin, leading to inhibition of both heat and mechanical behaviors [8; 9; 14; 47]. Thus, we co-injected capsaicin and QX-314 in wildtype mice, finding that cold was also inhibited in a TRPV1-dependent manner, a surprising result as Trpv1^−/− mice lack a cold phenotype (Fig. 2A, p>0.05 baseline (BL) wildtype vs. Trpv1^−/−) [13]. This injection strategy attenuated both mechanical (Fig. 2B) and heat responses (Fig. 2C), an effect previously reported using this approach [8; 9]. Intraplantar injection of the TRPM8 agonist WS-12 with QX-314 produces a complete and long lasting inhibition of cold behaviors in mice [39], prompting us to compare this with the phenotypes produced by agonists for TRPA1 and TRPV1. As shown in Fig. 2D, while the effect of WS-12 was longer lasting, the other agonists were as effective in producing reduced cold behaviors. Thus, injection of QX-314 with TRPA1 or TRPV1 agonists leads to inhibition of cold-evoked behaviors, a phenotype not observed in mice lacking these channels.

3.3. Putative block of TRPA1 afferents inhibits noxious cold.

The absence of an innocuous cold phenotype in Trpa1^−/− mice is consistent with prior data suggesting that TRPA1 functions solely as a detector of noxious cold [10; 47]. However, the phenotype produced by co-injection of TRPA1 agonists with QX-314 is inconsistent with this role for the channel, prompting us to determine if there was a similar effect with noxious cold stimuli. Results using classical Cold Plate hind paw latencies to lift or flinch have been inconsistent between studies [34], whereas nocifensive fore paw behaviors have proven to be a reliable measure of noxious cold sensitivity [26; 39; 49]. However, to the best of our knowledge this approach has yet to be examined in TRPA1-deficient mice. As shown in Fig. 3A, wildtype mice exhibit a significant number of nocifensive fore paw behaviors when on a 0°C plate for 3 minutes compared to a plate held at 30°C. Trpa1^−/− mice exhibit a similar phenotype and we observed no differences in nocifensive behaviors between the two genotypes (Fig. 3A). Thus, as with the Cold Plantar assay, TRPA1-deficient mice have normal sensitivity to noxious cold.

Next, we asked if injection of cinnamaldehyde with QX-314 in both fore paws altered noxious cold responses. As with the divergent phenotypes observed with the Cold Plantar assay, we found that wildtype animals injected with QX-314 and cinnamaldehyde had a significantly reduced number of nocifensive behaviors compared to animals injected with QX-314 (Fig. 3B). These data are consistent between the two cold assays and highlight a disconnect between cold phenotypes of TRPA1-deficient mice and those in which these afferents are presumably inhibited.

Lastly, stimulating a QX-314-injected hind paw with noxious cold results in the inhibition of subsequent cold behaviors induced by innocuous cold stimuli, presumably via thermally activated TRPM8 channels [39]. To determine if TRPA1 contributes to this effect of noxious cold on QX-314, we performed unilateral hind paw injections of only QX-314 in Trpa1^−/− mice and placed them on a 10°C plate for 30 minutes. Animals were then habituated as described above, then tested using the Cold Plantar assay. Similar to the phenotype of wildtype mice [39], Trpa1^−/− mice had significant reductions in lift latencies compared to vehicle injected mice or to the contralateral hind paw (Fig. 3C). Thus, TRPA1 does not mediate QX-314-induced cold inhibition produced by stimulation with noxious cold.

3.4. Nerve conduction block induced by activation of nociceptors is TRPM8-dependent.

The presumptive phenotypes of TRPA1 and TRPM8 afferent block are indistinguishable in magnitude (Fig. 2D), which is surprising as the two channels are not known to be co-expressed [47]. Additionally, our results with presumptive block of TRPV1 afferents are inconsistent with prior studies finding that mice lacking functioning TRPV1 afferents have no deficiencies in cold behaviors [15; 42]. TRPM8 is known solely as a receptor for cold [26; 42], thus we asked if these phenotypes were TRPM8-dependent by pharmacologically blocking TRPM8 using the antagonist PBMC, an approach used previously to validate cellular QX-314 entry via TRPM8 [39]. Hind paw PBMC injection inhibits cold for over 4 hours (see [39]). Using the temporal difference between the duration of PBMC inhibition and that of QX-314 with cinnamaldehyde (Fig. 2D), we injected QX-314 and cinnamaldehyde (time 0) in mice given either a PBMC or vehicle injection 3 hours prior (time −3 hours; Fig. 4A). Using this dual-injection approach, we find that 2 hours after the QX-314/agonist injections, cold inhibition was significantly reduced in mice injected with PBMC, but not vehicle (Fig. 4A). Thus, these data suggest that inhibiting TRPM8 channels prevents cinnamaldehyde-induced uptake of QX-314 and the subsequent inhibition of cold sensitivity. Lastly, we repeated this experiment in mice injected with capsaicin and QX-314, observing a similar, if not more pronounced, phenotype in mice pre-injected with PBMC (Fig. 4B).

Figure 4: TRPA1 and TRPV1 agonists induce QX-314-mediated cold inhibition via TRPM8.

(A) Pre-treatment with the TRPM8 antagonist PBMC (100μg) attenuated QX-314/cinnamaldehyde inhibition of cold behaviors in the Cold Plantar assay (vs. vehicle or PBMC alone, n=6, ***p<0.001; ANOVA with Tukey’s multiple comparison test). (B) PBMC pre-treatment inhibited QX-314/capsaicin block of cold behaviors (vs. vehicle or PBMC alone, n=6, ***p<0.001, **p<0.01; ANOVA with Tukey’s multiple comparison test). (C) Habituation at 36°C prevented the reduced sensitivity to cold stimuli induced by injection of QX-314 and cinnamaldehyde (pre vs. post, n=8, ^nsp>0.05; paired t-test).

Previously, we have shown that permeation of QX-314 via TRPM8 is inhibited by warmth [39]. Therefore, to further determine if active TRPM8 channels are required for the effects of TRPA1 agonists, we tested the temperature-sensitivity of cold inhibition induced by cinnamaldehyde and QX-314. When mice were placed on a plate held at an inhibitory temperature (36°C) immediately after injection, cinnamaldehyde and QX-314 had no effect on cold responses (Fig. 4C). When taken as a whole, these results demonstrate that activation of nociceptive afferents, either by TRPA1 or TRPV1 agonists, produces QX-314 mediated cold inhibition via TRPM8 channels.

3.5. Nociceptor activation promotes QX-314 uptake via CGRP and TLR4.

To understand the TRPM8-dependency of this phenotype, we reasoned that nociceptor activation produces neurogenic inflammation via release of proalgesic peptides such as CGRP [4]. Thus, we asked if antagonizing CGRP receptors affected this phenotype. To test the effect of CGRP receptor antagonism on cold behaviors, we first examined cold sensitivity 30 minutes after intraplantar injection of the antagonist CGRP(8–37) [1]. Under these conditions, we observed no differences in basal cold responses (Fig. 5A), establishing that cold was unaffected by CGRP receptor antagonism. However, subsequent cinnamaldehyde and QX-314 injections in these same mice produced a modest reduction to the degree of cold inhibition produced in the absence of CGRP(8–37) (Fig. 5A), suggesting that this effect was partly dependent on CGRP. However, further analysis of this data and an increase in the number of animals tested showed distinctive spread into two cohorts; one that was inhibited and a second that was not. When these results were analyzed by sex, the QX-314 effect was absent in female (Fig. 5B), yet preserved in male mice (Fig. 5C), a difference not found in any of our prior data.

Figure 5: Antagonism of CGRP and TLR4 signaling prevents cinnamaldehyde/QX-314 cold block and is sexually dimorphic.

(A) In the Cold Plantar assay, pre-injection of the CGRP receptor antagonist CGRP(8–37) (6.4mmols) had no effect of cold behaviors (^nsp>0.05, paired t-test) but attenuated QX-314/cinnamaldehyde mediated cold inhibition (n=14, equal number of males and females, **p<0.01 vs. BL or CGRP(8–37); paired t-test). (B) In females, CGRP(8–37) prevented QX-314/cinnamaldehyde inhibition (vs. antagonist only, n=7, ^nsp>0.05; paired t-test ). (C) Cold inhibition was unaltered in males treated with CGRP(8–37) (vs. antagonist only, n=7, **p<0.01; paired t-test). There were no differences between baseline and antagonist only (n=7, ^nsp>0.05; paired t-test). (D) Pre-injection with 0.2μg of the TLR4 antagonist TAK-242 attenuated QX-314/cinnamaldehyde mediated cold inhibition (n=16, equal number of males and females, **p<0.01 vs. BL or CGRP(8–37); paired t-test). (E) In females, QX-314 block was maintained (vs. antagonist only, n=8,***p<0.001; paired t-test), while in males (F) QX-314 inhibition was prevented (vs. antagonist only, n=8, ^nsp>0.05; paired t-test). There were no differences between baseline and antagonist only (n=8, ^nsp>0.05; paired t-test).

Pursuing this intriguing sexually dimorphic result further, we examined the male-specific Toll-like receptor 4 (TLR4) pain signaling pathway [51]. Pre-injection of the antagonist TAK-242 had no effect on basal cold sensitivity (Fig. 5D), but prevented cold inhibition after cinnamaldehyde and QX-314 injections in male (Fig. 5E) but not female mice (Fig. 5F). Thus, our results suggest that nociceptor activation produces a sexually dimorphic neurogenic response that promotes QX-314 uptake via TRPM8 to inhibit cold.

3.6. Artemin and GFRα3 mediate nociceptor-induced QX-314 inhibition of cold.

Artemin is the only known endogenous molecule capable of directly inducing cold sensitization, and its receptor GFRα3 is required for inflammatory and neuropathic cold pain [29; 30]. Thus, we asked if the QX-314 effect induced by nociceptor activation requires GFRα3. Remarkably, cold inhibition induced in wildtypes after co-injection of QX-314 with either cinnamaldehyde (Fig. 6A) or capsaicin (Fig. 6B) was absent in GFRα3-null (Gfrα3^−/−) mice. GFRα3 is not required for QX-314 uptake as cold inhibition was induced in both wildtype and Gfrα3^−/− by the TRPM8 agonist WS-12 (Fig. 6C). Thus, QX-314 inhibition of cold after nociceptor activation requires GFRα3, prompting us to test its ligand artemin, reasoning that peripheral artemin should promote QX-314-induced inhibition of cold. To test this, we co-injected artemin with QX-314 in wildtype mice and found the degree of cold inhibition was similar to both nociceptor and TRPM8-afferent activation (Fig. 6D). Thus, artemin alone is able to promote QX-314-mediated inhibition of cold sensitivity.

Figure 6: Artemin and GFRα3 mediate QX-314 inhibition of cold behaviors after nociceptor activation.

(A) Cold inhibition induced by co-injection of QX-314 and cinnamaldehyde was present in wildtype (n=7, ***p<0.001; paired t-test) but not Gfrα3^−/− (n=7, p>0.05; paired t-test) in the Cold Plantar assay. (B) Cold inhibition induced by co-injection of QX-314 and capsaicin was present in wildtype (n=6, ***p<0.001; paired t-test) but not Gfrα3^−/− (n=6, p>0.05; paired t-test). (C) In comparison, co-injection of QX-314 and WS-12 inhibited cold behaviors in both wildtype (n=4, ***p<0.001; paired t-test) and Gfrα3^−/− (n=6, ***p<0.001; paired t-test). (D) Artemin (ARTN, 0.2μg) co-injected with QX-314 inhibited cold responses in the ipsilateral (vs. BL, n=8, ***p<0.001; paired t-test) but not the contralateral hind paw (vs. BL, n=8, ^nsp>0.05; paired t-test).

3.7. Cold sensitization via TRPA1 is GFRα3-dependent.

We have shown that GFRα3 is required for inflammatory and neuropathic cold pain, which can be ameliorated by artemin sequestration [30]. Moreover, artemin induces cold allodynia in a TRPM8-dependent manner [29]. Data supporting the involvement of TRPA1 in cold pain includes increased cold sensitivity in vivo after TRPA1 activation, and reduced inflammatory or neuropathic cold pain in Trpa1^−/− mice and with TRPA1 antagonism [16; 19; 37; 40]. Based on our results, we reasoned that the phenotypes in these pathologies were not due to TRPA1 afferents signaling cold, but were a result of TRPA1-mediated neurogenic release of proalgesics that activated either the CGRP or TLR4 pathways (Fig. 7A). This in turn promotes localized artemin release that sensitizes TRPM8 channels in a GFRα3-dependent manner. In this model, TRPA1-induced cold sensitivity should require GFRα3.

To test this, we first asked if artemin-induced cold allodynia requires TRPA1 by examining artemin-induced cold sensitization in wildtype and Trpa1^−/− mice, observing no differences between the two genotypes (Fig. 7B). Next, we asked in cold allodynia produced by TRPA1 agonists required GFRα3, finding that cinnamaldehyde-induced cold sensitization observed in wildtypes was absent in Gfrα3^−/− mice (Fig. 7C). Note that our data shows a significant difference (p=0.013, n=8) between baseline and post-injection in Gfrα3^−/−, but this is a slight decrease in cold sensitivity, not the expected increased response observed in wildtype mice. The lack of a cold allodynic phenotype was not due to an overall inability of these Gfrα3^−/− mice to mount a nociceptive response as the TRPM8 agonist WS-12 still sensitized these animals to cold (Fig. 7D). Therefore, TRPA1 activation does not directly transmit nociceptive cold signals, but likely works upstream of artemin and GFRα3 mediated sensitization of TRPM8.

4. Discussion

Here, we show that activation of peripheral nociceptive afferents with TRPA1 or TRPV1 agonists in the presence of QX-314 leads to not only inhibition of heat and mechanical sensation, but also attenuates cold. Consistent with prior studies [8; 9; 28; 43; 48], the former is likely due to gradual inhibition of either TRPA1- or TRPV1-expressing cells which are established mediators of mechanical or heat stimuli, respectively. Of note is the inhibition of mechanical responses with presumptive block of TRPV1+ afferents with co-injection of capsaicin and QX-314 in this assay. Mechanically-evoked behaviors are unaltered in animals in which TRPV1+ afferents are ablated genetically or chemically [15; 42]. However, consistent with our results, activation of TRPV1 channels in the presence of QX-314 inhibits some mechanical behaviors [8; 9]. For cold, our data indicates that the observed phenotype is not due to inhibition of either TRPA1 or TRPV1 afferents, but rather downstream inhibition of TRPM8-expressing cells. Taken together, our results suggest caution when interpreting the behavioral phenotypes observed by agonist induced nerve block using QX-314 as we have seen effects beyond the targeted receptor. Additionally, it is difficult to accurately quantify either the number or type of afferents that take up QX-314 in vivo, nor determine the level of nerve conduction block in affected nerve fibers. Nonetheless, cell-targeted nerve conduction block is a powerful tool for phenotypic analyses and a potentially modality selective therapeutic approach.

Historically, our understanding of the role of TRPA1 in cold transduction in vivo has been problematic due to conflicting results in various published studies [34]. For example, the reports on TRPA1-deficient mouse phenotypes have been contradictory [5; 18; 27], as have studies characterizing in vitro cold sensitivity of TRPA1 channels [19; 22; 47]. Here we show that mice lacking TRPA1 channels retain normal cold acuity to either an innocuous (Cold Plantar assay) or noxious cold stimulus (Cold Plate assay), consistent with prior findings [5; 12; 42]. However, studies that have reported cold-deficiencies in mice lacking TRPA1 channels [27] primarily compared cold phenotypes in various injury models [16; 37; 38; 52; 53], or in animals treated with TRPA1 agonists [19; 20]. Moreover, studies showing a cold-deficient phenotype in non-pathological conditions used assays in which the animals were exposed long-term to tissue damaging noxious cold stimuli [24; 27]. Regardless of the experimental paradigm, to the best of our knowledge, all of these in vivo assays were performed in the presence of normal TRPM8 channel expression and activity.

Our results provide a plausible rationale for these prior ambiguous findings in that activation of TRPA1 channels, either by exogenous agonists as done herein, or endogenous proalgesics such as those produced during inflammation, leads to a neurogenic response that sensitizes TRPM8 channels and neurons (Fig. 7A). We propose that robust TRPA1 (or TRPV1) activation induces neurogenic inflammation, which appears to utilize TLR4 or CGRP pathways. This leads to the release of artemin and activation of its receptor GFRα3, which then sensitizes TRPM8 channels and afferents. This hypothesis is supported by our prior studies observing TRPM8-dependent artemin-induced cold sensitization [29]. Moreover, under neuropathic and inflammatory conditions, cold sensitization is dependent on GFRα3 and can be ameliorated by systemic treatment with artemin-neutralizing antibodies [30]. Thus, artemin signaling via GFRα3 and TRPM8 is required for cold allodynia and hyperalgesia.

In the context of decreased expression or activity of TRPA1 as has been tested in the multitude of studies described above, neurogenic inflammation would be reduced and therefore lessen or prevent increased artemin, which in turn would not lead to cold sensitization via TRPM8. In support of this general neurogenic inflammatory mechanism, we find identical results with TRPV1 activation which generally produces a larger effect on cold than TRPA1 activation, consistent with higher levels of TRPV1 channel expression in sensory ganglia [47]. Furthermore, the effects we see are not specific for either channel, coupled with our intriguing observations on the requirement of the CGRP receptor and TLR4 activity, demonstrates the general influence of nociceptor channel activation leading to neurogenic cold allodynia and hyperalgesia.

The sexual dimorphic nature of QX-314 inhibition of cold after nociceptor activation is intriguing and consistent with new evidence of sex-specific pathways that underlie pain [32]. CGRP release is a well-established mediator of neurogenic inflammation and more recently it has been reported that CGRP induces migraine-related pain only in female and not male mice [2]. While many studies do not account for female and male differences [21], together with our finding that antagonism of the CGRP receptor prevented TRPA1 induced QX-314 block of cold sensation in females, only highlights a role for CGRP in female specific peripheral pain. Similarly, our findings that antagonism of TLR4 prevents TRPA1-induced QX-314 block of cold sensation in males only is consistent with this signaling pathway’s involvement in pain transduction in males but not females [46; 51]. TLRs are primarily expressed in immune cells, are well known for their role in recognizing pathogen or endogenous associated molecular patterns, and have been implicated in male specific signaling pathways [31]. The mechanisms that underlie these sexual dimorphic pathways in acute neurogenic inflammation and their relation to cold will require further investigation.

Nonetheless, while these sexual dimorphisms are present, ultimately GFRα3 and artemin are required for cold sensitization regardless of sex, further supporting GFRα3 and artemin signaling as promising therapeutic targets for inflammation induced cold allodynia and hyperalgesia [29; 30]. When taken as whole, our data suggests that TRPM8 is the primary mediator of innocuous and painful cold, and that TRPA1 serves an important pro-modulatory role in cold pain, consistent with its involvement in other pain modalities [4].