Regulation of TRPM8 channels by PKCβ mediates morphine-induced cold hypersensitivity. CURRENT STATUS:

Postoperative shivering and cold hypersensitivity are major side effects of acute and chronic opioid treatments respectively. TRPM8 is a cold and menthol-sensitive channel found in a subset of dorsal root ganglion (DRG) nociceptors. Deletion or inhibition of the TRPM8 channel was found to prevent the cold hyperalgesia induced by chronic administration of morphine. Here, we examined the mechanisms by which morphine was able to promote cold hypersensitivity in DRG neurons and transfected HEK cells. Mice daily injected with morphine for five days developed cold hyperalgesia. Treatment with morphine did not alter the expressions of cold sensitive TREK-1, TRAAK and TRPM8 in DRGs. However, TRPM8-expressing DRG neurons isolated from morphine-treated mice exhibited hyperexcitability. Sustained morphine treatment in vitro sensitized TRPM8 responsiveness to cold or menthol and reduced activation-evoked desensitization of the channel. Blocking phospholipase C (PLC) as well as protein kinase C beta (PKCβ), but not protein kinase A (PKA) or Rho-associated protein kinase (ROCK), restored channel desensitization. Identification of two PKC phosphorylation consensus sites, S1040 and S1041, in the TRPM8 and their site-directed mutation were able to prevent the MOR-induced reduction in TRPM8 desensitization. Our results show that activation of MOR by morphine 1) promotes hyperexcitability of TRPM8-expressing neurons and 2) induces a PKCβ-mediated reduction of TRPM8 desensitization. This MOR-PKCβ dependent modulation of TRPM8 may underlie the onset of cold hyperalgesia caused by repeated administration of morphine. Our findings point to TRPM8 channel and PKCβ as important targets for opioid-induced cold hypersensitivity. USA). Voltage- and current-clamp protocols were applied using pClamp 10.4 software (Axon Instruments). Data were filtered at 1 kHz (8-pole Bessel) (whole cell voltage clamp) and 5 kHz (current clamp) and digitized at 10 kHz with a Digidata 1550 A converter (Axon Instruments). Average DRG neurons capacitance was 11.26 ± 0.69 pF for naive and 10.59 ± 0.98 pF for morphine-treated animal. HEK cells had an average capacitance of 26.78 ± 2.15 pF. Only the cells that exhibited a stable voltage control throughout the recording were used for analysis.


Introduction
Opioids are widely used analgesics for the treatment of moderate to severe pain but their use often leads to the development of severe side effects, including opioid induced cold hyperalgesia [1][2][3]. The clinically used analgesic morphine acts on MOR expressed on the primary afferent pathway. At the cellular level, MOR activates Gα i/o proteins leading to inhibition of cyclic adenosine monophosphate (cAMP) generation, increased extracellular signal-regulated kinase (ERK) phosphorylation, inhibition of presynaptic voltage-gated calcium channels at primary afferent central terminals and activation of postsynaptic G protein-regulated inwardly rectifying potassium channels (GIRKs) in projecting 7 the Supplementary Table 1 and were designed to amplification of DNA.

Calcium imaging
Transfected HEK cells or isolated DRGs were loaded with Fura-2-AM (0.5μM for 30 min, Invitrogen) before imaging. The recordings were performed at 37ºC. We perfused either menthol (100 µM,Sigma) or extracellular solution at 25ºC, at a rate of ~1 ml/min, to examine TRPM8 activity in control conditions or in cells treated with morphine (500 nM) for 16 hrs. Images of cell-field were continuously recorded every 100 ms using 340 and 380 nm excitation wavelengths with emission measured at 520 nm with a microscope based imaging system (Olympus IX73) on CellSense software. Images were processed using ImageJ by drawing discrete regions of interest around cells that responded to menthol or cold. We expressed a change in fluorescence as a percentage change from the amplitude of the first response to menthol or cold.

Behavior analysis
6-week old C57BL/6 male mice were obtained from Jackson Laboratory (USA) and were acclimated for 2 days prior to behavioral experiments. Mice were housed with free access to food and water, with a 12/12 light dark cycle. All experiments were conducted on age-matched animals, under protocols approved by the University of Calgary Animal Care Committee and in accordance with the international guidelines for the ethical use of animals in research and guidelines of the Canadian Council on Animal Care.
Behavioral assessment of cold sensitivity in mice was done using the cold plate test. Mice received intraperitoneal injection of escalating doses of morphine (from 10 mg/kg to 50 mg/kg) or saline, twice daily for 5 days. Cold sensitivity was measured using the cold plate test (Bioseb). Briefly, mice were placed on the surface of a metal plate cooled at 0° C (with an ambient temperature of 21° C). The time taken for each mouse to show the first nociceptive response (paw withdrawal, shaking, liking of the rear paw, or jumping to try to escape) was monitored on the first and last day of the morphine injection. An experimenter blinded to the treatments performed the behavioral assessment. Data points represent each individual mouse.

Electrophysiological measurements
Electrophysiological recordings were conducted using an external solution containing (in mM): 140.0 NaCl, 1.5 CaCl 2 , 2.0 MgCl 2 , 5.0 KCl, 10.0 HEPES, 10.0 D-glucose, pH 7.4 adjusted with NaOH. HEK cells expressing the transfected TRPM8 channel and MOR were identified via GFP fluorescence using an inverted epi-fluorescence microscope (Olympus IX51, Olympus America Inc., USA). DRG neurons were recorded based on size, knowing that TRPM8 is expressed in a subpopulation of small neurons [21]. Membrane currents were measured using conventional whole-cell patch clamp and action potentials were recorded using current clamp. Borosilicate glass (Harvard Apparatus Ltd., UK) pipettes were pulled and polished to 2-5 MΩ resistance with a DMZ-Universal Puller (Zeitz-Instruments GmbH., Germany). For the voltage clamp experiments pipettes were filled with an internal solution containing (in mM): 120.0 CsCl, 10.0 EGTA, 10.0 HEPES, 3.0 MgCl 2 , 2.0 ATP Na2, 0.5 GTP, pH 7.2 adjusted with CsOH whereas for the current clamp experiments the internal solution contained (in mM): 140.0 KCl, 5.00 NaCl, 1 CaCl 2 , 1.0 EGTA, 10.0 HEPES, 1.0 MgCl 2 , 3.0 ATP Na2, pH 7.3 adjusted with KOH. All solutions were prepared and used at room temperature (22±2ºC) and their osmolarity adjusted to 310 mOsm. Data obtained with different types of internal solutions were not pooled. For the current clamp experiments the spontaneous activity of the DRG neurons was recorded at room temperature (~ 22⁰C) for three minutes before application of the first cold (10⁰C) or menthol challenge (applied to the bath at ~ 1000 µm from the cell at a rate of 500 µl/min). Only the neurons in which the resting membrane potential was more negative than -40 mV were used. Recordings were performed using an Axopatch 200B amplifier (Axon Instruments, Foster City, CA, USA). Voltage-and current-clamp protocols were applied using pClamp 10.4 software (Axon Instruments). Data were filtered at 1 kHz (8-pole Bessel) (whole cell voltage clamp) and 5 kHz (current clamp) and digitized at 10 kHz with a Digidata 1550 A converter (Axon Instruments). Average DRG neurons capacitance was 11.26 ± 0.69 pF for naive and 10.59 ± 0.98 pF for morphine-treated animal. HEK cells had an average capacitance of 26.78 ± 2.15 pF. Only the cells that exhibited a stable voltage control throughout the recording were used for analysis.
For prolonged morphine treatment, neurons were incubated with a low concentration of morphine evoked action potential discharge after morphine treatment (Fig. 1b). The percentage of mentholsensitive small diameter neurons exhibiting spontaneous activity was enhanced (Fig. 1c) as was the frequency of spontaneous activity (Fig. 1d). Chronic morphine did not change significantly the neuronal resting membrane potential but lowered the action potential threshold ( Fig. 1e and f). To assess whether morphine signaling was affecting voltage-gated ionic conductances, we measured the neuronal activity evoked by ascending ramps of injected current. Neurons from morphine treated mice exhibited an increase in the frequency of AP compared to control ( Fig. S1a and b). Overall, these data indicate that chronic exposure to morphine enhances the excitability of TRPM8-expressing neurons and leads to a modulation of voltage-gated ion channels.

Chronic morphine reduces both cold and menthol-evoked desensitization of TRPM8
TRPM8 is the primary cold receptor of the somatosensory system and its desensitization has been reported to account for the adaptation to environmental cold [23]. We thus assessed TRPM8 desensitization after morphine treatment by measuring TRPM8 calcium responses to two repeated stimulations with either cold or menthol. In DRG neurons from saline injected animals, TRPM8 calcium signals were markedly desensitized in response to a second stimulation by cold or menthol [23,24].
In contrast, DRG neurons from morphine treated mice showed a loss of TRPM8 desensitization to both stimuli ( Fig. 2a and b). As the desensitization of TRPM8 in response to the second stimulation is a calcium sensitive process that is directly contingent on the intracellular calcium load evoked by the first stimulation, we asked whether morphine was able to affect the first response to cold or menthol.
As shown in Fig. S1d, the response to cold or menthol did not differ in cultured sensory neurons exposed to morphine or in neurons isolated from morphine-treated mice, relative to control. In accordance with previous published results [12,21,25,26], we found that cold sensitive DRG neurons were a rare subset of small diameter neurons (< 20 µm in diameter). In control mice, 3.06% of total neurons (37/1200) responded to cold and 3.31% (37/1138) to menthol (100 µM). Interestingly, this number increased in morphine-treated groups, with 16.71% of total neurons (304/1822) responding to cold and 16.95% (325/1916) to menthol (Fig. 2c). These results indicate that chronic morphine caused a subset of menthol-insensitive neurons to acquire de novo responsiveness to menthol and cooling.
Nonetheless, no significant change in mRNA expression was found for either TRPM8 or other cold sensitive TREK-1 and TRAAK channels (Fig. S1c), implying a functional rather than transcriptional regulation of these neurons [11,[26][27][28].

Sustained morphine treatment enhances TRPM8 responsiveness to cold and menthol in cultured DRG neurons
We then concentrated our work on TRPM8 since both cold and menthol responses were affected equally by morphine treatment. To determine the signaling pathways by which morphine was able to reduce TRPM8 desensitization, we tested the effect of morphine on cultured DRG neurons. As found with morphine-treated mice, in vitro exposure of neurons to morphine (500 nM, 16 hours) increased both cold-and menthol-evoked depolarization and AP discharge ( Fig. 3a and b). Importantly, application of the TRPM8 specific blocker AMTB (30µM) was able to induce membrane potential hyperpolarization and block menthol-evoked AP discharge, as previously described [29] ( Fig. 3a and   b). Strikingly, our in vitro chronification model corroborated our chronically treated animals as morphine exposure suppressed TRPM8 desensitization to cold or menthol and appeared to even potentiate TRPM8 responsiveness in a small proportion of cells. (Fig. 3d and e). In agreement with a sensitization of TRPM8 to cold or menthol following morphine treatment, the temperature-response curve of TRPM8 was shifted (~4 °C) towards warmer temperatures in both DRG neurons from morphine-sensitized animals and neurons from naïve mice exposed to morphine (Fig. 3f). Likewise, morphine induced a leftward shift of the dose-response curve of TRPM8 sensitivity to menthol (Fig. 3g). A similar pattern of sensitization was found in HEK cells transfected with TRPM8+MOR and exposed to morphine overnight ( Fig. S2a and b).

Morphine-induced reduction in TRPM8 desensitization requires PKCβ activation
We next interrogated the MOR signaling pathway that regulated TRPM8 desensitization. In transfected HEK cells, sustained application of morphine negated the TRPM8 desensitization as observed in neurons (Fig. 4a). Reduction of TRPM8 desensitization by morphine depended on the expression of MOR and was lost upon application of its competitive antagonist Naloxone. These data suggested a selective MOR-mediated signaling pathway that regulated channel desensitization (Fig. 4b). In addition, the effect of morphine on TRPM8 desensitization required long lasting treatment, as acute exposure failed to prevent menthol-or cold-evoked desensitization of TRPM8 measured by whole cell patch clamp (Fig.S3a) or calcium imaging (Fig. S3b). Strikingly, the effect appeared to be triggered by morphine-bound MOR signaling since [D-Ala2, N-MePhe4, Gly-ol]-enkephalin (DAMGO), a highly specific MOR agonist, failed to inhibit TRPM8 desensitization (Fig. S2c). Further investigation with other Gq-or Gi-coupled receptors showed that sustained activation of the protease activated receptor (PAR2), the α2 or α1 adrenergic receptor, or the cannabinoid receptor CB1 also failed to prevent TRPM8 desensitization in response to their specific agonists (Fig. S2d). Altogether, our findings reveal the sensitization of TRPM8 by virtue of morphine-induced activation of MOR.
MOR is coupled to the Gαi/o class of Gα proteins. While activation of MOR by DAMGO induces βarrestin (βARR) recruitment and receptor internalization, morphine was shown to exhibit a different signaling profile with preference to PLC activation, mitogen-activated protein kinase (MAPK) and PKCdependent desensitization [4,30,31]. PLC signaling mediates hydrolysis of phosphatidyl inositol 4,5phosphate 2 (PIP2), a membrane-associated phospholipid that maintains the open probability of TRPM8 channels [23,32], and generates diacylglycerol (DAG) [33,34]. We next used pharmacological blockers to test the contribution of this pathway on TRPM8 desensitization. Morphine-induced inhibition of TRPM8 desensitization was reversed by using the PLC inhibitor U73122 (1 μM), the broad spectrum PKC inhibitor GF109203X (3 μM) as well as by the specific PKCβ blocker enzastaurin (1 μM) ( Fig. 4c). In contrast, blocking PKA or ROCK was not able to prevent MOR-induced reduction in TRPM8 desensitization. Importantly, in the absence of morphine, activation of PKC with an acute PMA treatment was able to prevent desensitization of TRPM8 (Fig. 4c). These results suggest that activation of MOR by morphine stimulates a MOR-PLC-PKCβ signaling that dictates TRPM8 desensitization.

Site-specific regulation of TRPM8 at Ser 1040 and 1041 mediates morphine-induced reduction in TRPM8 desensitization
Phosphorylation of ion channels by PI3K or PKC is known to regulate cell surface expression of ion channels in sensory neurons [30,35] and opioids have been proposed to modulate thermosensation by internalizing the TRPM8 channel [18]. To test whether MOR sensitized channel function by modulating channel trafficking, we used biotinylation assay in HEK cells to assess the cell surface expression of TRPM8 following morphine treatment. In the absence or presence of the MOR, insertion of TRPM8-YFP at the plasma membrane was unaffected by morphine treatment (Fig. 5a and b). These data confirm that morphine augments TRPM8 sensitivity to cold or menthol, without altering its expression at the cell surface.
In order to examine how MOR-PKCβ signaling was able to regulate TRPM8 desensitization, we asked if TRPM8 associated with the kinase using co-immunoprecipiation. As shown in Fig. 6a, PKCβ-GFP immunoprecipitated a TRPM8-HA channel in transfected HEK cells, indicating the existence of a TRPM8-PKCβ signaling complex that could respond to MOR activation. Due to a lack of a specific P-Ser/Thr TRPM8 antibody we were unable to examine the level of phosphorylation of TRPM8 in response to morphine (see discussion), however, we examined the phosphorylation consensus amino acid sequence for PKC phosphorylation (RXXS/TXRX; where X is any amino acid) using PhosphoSitePlus. This in sillico analysis revealed several possible phosphorylation sites with only two highly conserved consensus motifs: one at Ser850 ( 843 LIHIFTVSRNLGPKI 857 ) at the end of the S4 segment, the second in the C terminus ( 1034 KEKNMESSVCCFKNE 1048 ) of TRPM8. We focused on the Cterminal region of the channel that contains the TRP domain involved in menthol binding and cold sensing [36][37][38][39]. Serines were replaced with alanines at sites 1040 and 1041. Expression of channel mutants in HEK cells did not indicate alterations in the current-voltage relationship, and the current density evoked by menthol, compared to wild type TRPM8 (Fig. S2e). Nevertheless, expression of the mutant channel completely eliminated the morphine-induced sensitization of both cold and mentholevoked calcium response ( Fig. 6b and c). These results strongly suggest that activation of MOR by morphine elicits a PKCβ-mediated phosphorylation of TRPM8 at Ser 1040 and 1041, which in turn prevents channel desensitization.

Discussion
Cold hypersensitivity is an important behavioral manifestation of chronic morphine treatment. Here we report that morphine-induced cold hyperalgesia in mice is associated with 1) an increase in neuronal excitability of TRPM8-expressing DRG neurons and 2) a loss of TRPM8 desensitization evoked by cold or menthol. We showed that mice chronically exposed to morphine exhibit cold hyperalgesia, and neurons isolated from these mice are more excitable than neurons from naïve mice. Importantly, we found that morphine enhances the sensitivity of TRPM8 to cold or menthol and reduces channel desensitization. These changes in TRPM8 activity seem to account for the increased neuronal excitability induced by morphine as the TRPM8 blocker AMTB was able to hyperpolarize menthol sensitive neurons and inhibit AP discharge. We also found that prolonged activation of MOR by morphine contributes to the more positive resting membrane potential and increased both the frequency of spontaneous action potential and the AP firing in response to menthol. Despite the hyperpolarizing effect of AMTB, we thus cannot rule out that alterations in the activity of voltage gated calcium or potassium channels occur following morphine treatment. Overall, alterations in voltage-dependent ionic conductances leading to enhanced excitability, along with changes in TRPM8 activity, may together promote cold hypersensitivity induced by morphine.
Both morphine and endogenous enkephalin peptides are potent analgesics that act on MOR, however morphine displays partial agonism at the MOR and predominantly exerts both analgesic effects and . This agonist specific activation pattern may also explain why DAMGO was not able to elicit a loss of TRPM8 desensitization as found with morphine.
The activity of the thermosensor TRPM8 in primary sensory neurons is unique in that it senses cool temperatures but also mediates cold-induced analgesia [10][11][12]. TRPM8 was previously found to be involved in the development of cold hyperalgesia following chronic morphine administration, although the underlying mechanisms of this effect involved an upregulation of the channel expression [19].
Shapovalov et al. found a distinct mechanism of TRPM8 regulation in rat DRG neurons with morphine administration leading to an increase in TRPM8 internalization [18]. In contrast to what was reported by Gong et al. [19] we did not observe an increase in TRPM8 mRNA expression in response to morphine. In addition, assessment of plasma membrane biotinylated fraction did not reveal an alteration in TRPM8 surface expression. These negative results support the idea that regulation of TRPM8 intrinsic activity rather than change in channel expression account for the altered thermosensitivity observed upon escalating dose of morphine. Furthermore, the expression of the receptor at the mRNA or protein level has been reported to be relatively stable, with a few exceptions, following morphine treatment [42,43]. Accordingly, despite a central role of MOR in the development of morphine-induced cold hyperalgesia, we did not see a change in the expression of MOR mRNA in our experimental conditions. Both cold and menthol sensitivities were found to be enhanced by morphine. Previous studies described menthol as being promiscuous in its targets with the possibility of modulating the activity of other thermosensors (TRPA1, TRPV3, TRPC5, Nav1.9 and two pore potassium channels) [11,26,28,44]. Yet, menthol inhibits the activity of most of these channels at the concentrations used, including TRPA1 which was considered to be a noxious cold sensor [11,[26][27][28]44]. At the molecular level, we found that MOR activation by morphine leads to a PKCβ-induced reduction of TRPM8 desensitization likely via phosphorylation of the two Ser residues 1040 and 1041. While MOR acutely inhibits cAMP generation via Gi/o proteins, blocking PKA with H89 was not able to prevent TRPM8 regulation. In addition, we examined the RhoA/ROCK pathway that is involved in many cellular functions, including thermal hyperalgesia [45]. Our calcium signaling experiments using the Y27632 suggests that ROCK signaling does not contribute to TRPM8 regulation downstream of MOR.
The reduction in channel desensitization may be relevant to the transmission of cold signaling in primary afferent neurons and our findings suggest a crucial role of TRPM8 phosphorylation in this process. Activation of MOR by morphine was previously found to activate PLC and ERK [4,5] Our experiments however indicate a role of PKC downstream of morphine-induced MOR activation.
Application of a PKC blocker was able to reverse the loss of TRPM8 desensitization produced by morphine. In silico identification of PKC consensus sites and site-directed mutagenesis of Ser 1040 and 1041 completely blocked the morphine-induced inhibition of TRPM8 desensitization. The site of the mutation is in the vicinity of the S6-TRP box linker which seems to be a central determinant in channel gating by voltage and menthol [38]. This C-terminal region is also important in cold sensing and has several PIP2 binding sites involved in channel gating [36][37][38][39]. Using co-immunoprecipiation we showed that PKCβ was able to associate with TRPM8 in HEK cells, however we were not able to demonstrate whether TRPM8 was phosphorylated by PKCβ. We recognize that such data set would link our pharmacological experiments in Fig. 4 and the site directed mutagenesis study that identifies Ser 1040/41 (Fig. 6), however there are no phospho Ser/Thr TRPM8 antibodies available to perform immunostaining or western blotting analysis in the chronic morphine model. Moreover, a Pan-Ser/Thr antibody can be used on immunoprecipitated TRPM8, but since a very small proportion of neurons express TRPM8, the likelihood of measuring an increase in Ser/Thr phosphorylation of TRPM8 channels     of DRG neurons collected from naïve mice and exposed to morphine (MS) overnight versus control. (b) Whisker plots of the membrane potential of menthol sensitive neurons exposed to vehicle or morphine, before (dark) or after (red) menthol stimulation (-59.6 ± 1.4 mV vs.

Figure 7
Schematic representation of site-specific regulation of TRPM8 by MOR-PKCβ signaling.
Sustained morphine treatment acting on MOR induces PLC activation (1) which regulates PKCβ activity (2). The activated PKCβ phosphorylates the C terminus domain of the TRPM8 channel at S1040 and S1041 (3). This leads to a reduction of activity-induced channel desensitization (4). Both, increase in excitability of TRPM8-expressing neurons and reduction in activity-induced desensitization promotes morphine-induced cold hypersensitivity that is associated with chronic opioid treatment.

Supplementary Files
This is a list of supplementary files associated with this preprint. Click to download.