TRPV1 regulates opioid analgesia during inflammation

Acute inflammation in humans or mice enhances the analgesic properties of opioids. However, the inflammatory transducers that prime opioid receptor signaling in nociceptors are unknown. We found that TRPV1−/− mice are insensitive to peripheral opioid analgesia in an inflammatory pain model. We report that TRPV1 channel activation drives a MAPK signaling pathway accompanied by the shuttling of β-arrestin2 to the nucleus. This shuttling in turn prevents: β-arrestin2-receptor recruitment, subsequent internalization of agonist-bound mu opioid receptor (MOR), and suppression of DAMGO-induced inhibition of N-type calcium current observed upon desensitization. Consequently, inflammation-induced activation of TRPV1 preserves opioid analgesic potency in a mouse model of opioid receptor desensitization. Overall, our work reveals a TRPV1-mediated signaling mechanism, involving β-arrestin2 nuclear translocation, that underlies the peripheral opioid control of inflammatory pain. Our data single out TRPV1 channels as modulators of opioid analgesia.

Pain modulation and inflammation are two integrated biological responses that function in cooperation following injury. Particularly known for their analgesic properties, opioids target receptors expressed throughout the afferent pain pathway, including central and peripheral nerve terminals of primary afferent neurons [1][2][3] . Indeed, local endogenously produced opioids, β-endorphin, enkephalins, and dynorphins exert efficient inhibitory control of pain at both spinal 1,2 and peripheral sites following inflammation [4][5][6][7][8] . At inflammatory sites, immune-derived opioids are released to counteract pain effectively and facilitate healing. Moreover, alteration in the endogenous opioid system in the spinal cord was recently proposed to prevent the transition from acute inflammatory to chronic pain 1 , particularly through tonic μ-opioid receptor (MOR) activity that suppresses hyperalgesia following resolution of inflammation. Thus, both endogenous and xenobiotic opioids are most effective immediately after tissue injury, suggesting that the early phases of inflammation primes opioid receptor signaling in primary afferent neurons 7,9 . Alterations in this priming might contribute to a decrease in the therapeutic window of opioids, requiring increased analgesic doses that can lead to side effects.
At sites of acute local inflammation in humans or mice, the analgesic effect of opioids is enhanced 10 , whereas in contrast, the application of opioids to a human peripheral nerve plexus does not elicit analgesia in non-inflamed tissue 11 . At these sites, TRPV1 is the main endpoint target of inflammatory mediators, including reactive-oxygen species, metabolites of polyunsaturated fatty acids, inflammatory acidosis or toxins released by pathogen 12 . Moreover, both the activity and trafficking of the channel are sensitized by GPCR-targeted mediators (BK, PGE2, proteases) [13][14][15][16] , which in turn causes hyperalgesia during tissue inflammation 17 and in many immune-associated diseases 12,[18][19][20][21] . Along these lines, recent findings suggest that MORs expressed in TRPV1 + nociceptors are essential contributors to analgesic tolerance 22 yet, the molecular mechanisms underlying these biological regulations remain unknown. We set out to test whether the TRPV1 channel, a central integrator of inflammatory signals 12 , could prime opioid receptor signaling or modulate opioid analgesia during inflammation. Here, using the complete Freund's adjuvant (CFA) inflammatory pain model, we show that TRPV1 -/mice exhibit alterations in both endogenous and exogenous opioid analgesia. We reveal further that, upon TRPV1 activation, -arrestin2, a critical regulator of opioid signaling, is routed to the nucleus through a calcium-dependent mitogen-activated protein kinase (MAPK) signaling pathway. As a result of this exclusion of -arrestin2 from the cytosol, activated MORs are unable to recruit β-arrestin2 and internalize during agonist-induced desensitization. This reduced MOR desensitization has major implications for opioidinduced analgesia and β-arrestin2 biased signaling in vivo. Overall, our data uncover a mechanism by which inflammation-mediated activation of TRPV1 channels minimizes opioid receptor desensitization to optimize analgesia. This TRPV1-opioid receptor interaction could be pharmacologically targeted in clinical settings to improve pain management by opioids.

Results: TRPV1 regulates endogenous opioid-mediated analgesia during inflammation
To determine the molecular link between inflammation and opioid analgesia we asked if, as a central integrator of inflammatory signals, TRPV1 channel could modulate intrinsic opioid receptor function during peripheral inflammation. Therefore, we assessed inflammation-induced endogenous opioid antinociception in TRPV1 -/mice. TRPV1 is likely the best characterized sensory transducer that contributes to thermal hyperalgesia in acute models of inflammation 23-25 , although the channel is expressed in polymodal C fibers that process both thermal and mechanical hyperalgesia during long-lasting inflammation 26 . Using the CFA model of chronic inflammatory pain, we evaluated peripheral immune-derived endogenous opioid analgesia 6,27-30 by measuring paw withdrawal threshold (PWT) during the later phase of inflammation. In WT animals, PWT was decreased by 50% at day 4 after intraplantar injection of CFA, and animals started to recover at day 6. Treatment with naloxone-methiodide (Nal-M), a peripherally restricted non-selective and competitive opioid receptor antagonist, delayed the recovery time as attested by the increased PWT from day 6 to day 10 in both male ( Fig. 1a and b) and female mice (Supplementary Fig. 1a and b). In TRPV1 -/mice, we found that PWT was decreased by 40% at day 4, and then recovered gradually until day 14. Strikingly, treatment with Nal-M in both TRPV1 -/male ( Fig. 1a and b) and female mice (Supplementary Fig. 1a and b), had no effect on PWT at any time of the treatment. This absence of endogenous opioid analgesia in TRPV1-/-mice was due neither to a difference in inflammation (edema was similar between WT and KO mice), nor the  Figure 2a-c). Thus, as previously reported, local opioid signaling is fine-tuned by inflammation 29 , and our results indicate that TRPV1 is a central contributor to this process.

TRPV1 activation disrupts MOR-β-arrestin2 interaction
To delineate how opioid modulation is impaired in TRPV1 -/mice, we focused on β-arrestin2, a central hub signaling protein that modulates both opioid receptor desensitization and recycling 31 . A crosstalk between β-arrestin2 and the TRPV1 channel has recently emerged, particularly in the context of opioid receptor sequestration 32,33 . To determine whether activation of TRPV1 redirects -arrestin2 away from the agonistbound MOR, we tested the effect of capsaicin using a BRET assay that monitored interaction between MOR-Rluc8 and -arrestin2-Venus. As shown in Fig. 2, DAMGOmediated activation of MOR induces the recruitment of -arrestin2 to MOR, indicated by the increase in BRET signals over time. In the absence of TRPV1, treatment with capsaicin does not induce a BRET response or affect the DAMGO-induced recruitment of -arrestin-2 to MOR. However, in TRPV1 expressing cells, capsaicin co-treatment along with DAMGO completely blocks the interaction between MOR and -arrestin2 ( Fig. 2b and c), at low and even saturating concentrations of DAMGO (Fig. 2d), while capsaicin alone had no effect. Use of the reversed pair of BRET biosensors (βARR2-Rluc + MOR-YFP) yielded to similar results (Supplementary Fig. 3), thus confirming the inhibitory effect of TRPV1 activation on MOR/β-arrestin2 interaction. Interestingly, As shown in Fig. 2, co-immunoprecipitation of MOR with TRPV1 indicates the formation of a receptor-channel signaling complex at steady state (Fig. 2e), and the saturating BRET signals obtained for MOR and TRPV1 co-expression suggests a direct interaction between MOR and TRPV1 in HEK cells, with or without DAMGO (Fig. 2f). Therefore, our findings suggest that proximity of the TRPV1 channel to MOR, may alter the -arrestin2-biased agonism of MOR by redirecting -arrestin2 or preventing its recruitment to the receptor. Interestingly, we also found that this mechanism applies to other neuronal GPCRs, as the Proteinase-activated receptor-2 (PAR2), which like the MOR is involved in inflammatory pain, showed impaired -arrestin2 recruitment following capsaicin cotreatment. Given that β-arrestin2 is an essential modulator of MOR desensitization and recycling, we next tested the functional outcome of TRPV1 activation on receptor internalization. Using confocal microscopy we found that, as extensively reported, exposure to DAMGO for 20 min promoted significant MOR internalization in HEK cells transfected with MOR-YFP and TRPV1-mCherry. In contrast, cells co-stimulated with DAMGO and capsaicin (or RTX, not shown), had significantly less internalized receptor (Fig 3a and b). Importantly, deletion of β-arrestin2 using a CRISPR β-arrestin2 knockout cell line ( Supplementary Fig. 4), eliminated DAMGO-evoked internalization, and thus mimicked the DAMGO+capsaicin condition in β-arrestin2 WT cells. (Fig. 3 c and d).

TRPV1 activation induces both β-arrestin2 nuclear localization and ERK activation
As Por et al. reported that β-arrestin2 could interact with TRPV1 upon channel stimulation 32 , we asked whether channel activation could, through competitive interaction, prevent the recruitment of β-arrestin2 to MOR. Using our BRET assay, we found no evidence for an interaction between βarr2-YFP and TRPV1-Rluc ( Fig. 4a and   b) either at steady state, or upon channel activation by the potent TRPV1 ligand RTX. In contrast, BRET signals indicate a strong interaction between the Rluc-TRPV1 and YFP-TRPV1, confirming the functional assembly of tagged-subunits into oligomeric channels.
As β-arrestin-2 does not interact with either MOR or TRPV1 upon capsaicin exposure, we examined the fate of -arrestin2-YFP using spinning disk confocal microscopy on TRPV1 transfected cells. We made the striking observation that channel activation promotes a rapid translocation of β-arrestin2 to the nucleus (Fig. 4c). Quantification of β-arrestin2 nuclear localization indicates that >70% of TRPV1mCherry-expressing cells display nuclear translocation of β-arrestin2 at 15 min. after RTX application ( Fig. 4d and   e). This effect was slowly reversed upon RTX wash out and did not occur in absence of the channel (Supplementary Fig. 5a and b). Nuclear Hoechst blue staining confirmed the sequestration of β-arrestin2 into the nucleus after RTX challenge (Fig. 4f), an effect that could also be observed with capsaicin, using confocal imaging (Fig. 4g), or assessed by bystander BRET with the renilla GFP fused to the nuclear localization sequence (NLS) PKKKRKVEDPKS targeting the nucleus (Fig. 4h) 34 . In contrast, activation of TRPA1, another calcium-permeant member of the TRP channel family, did not trigger -arrestin2 translocation (Supplementary Fig. 5c). As for the activation of TRPA1, increasing intracellular calcium concentrations using the ionophore, ionomycin, also failed to stimulate the nuclear translocation of β-arrestin2 (not shown). Importantly, immunostaining of -arrestin2 in TRPV1-expressing neurons identified using a TRPV1-YFP reporter mouse indicated similar results (Fig. 4i). Thus, activation of TRPV1 rapidly relocates -arrestin2 into the nucleus both in expression systems and in native sensory neurons. As a consequence,-arrestin2 that normally colocalizes with internalizing MOR-containing vesicles, is now sequestered within the nucleus upon channel  Fig. 6).

(Supplementary
As arrestins are important scaffolds for the mitogen-activated protein kinase (MAPK) signaling pathway, we tested the effect of TRPV1 stimulation on extracellular signal-regulated kinases (ERK1/2). In TRPV1 transfected cells, RTX induces a transient phosphorylation of ERK1/2 at 5 min. post-stimulation (Fig. 5 a and b). ERK1/2 activation is blocked by chelating extracellular calcium with EGTA (10 mM) (Fig. 5c), as is the -arrestin2 nuclear translocation (Fig. 5d), confirming that calcium influx through activated TRPV1 channels mediates both ERK1/2 activation and β-arrestin2 translocation. Several kinases have been implicated upstream of ERK1/2 activation, including Src and different isoforms of PKC 35,36 . We found that pharmacological inhibition of PKCβII by CGP53353 blocks ERK1/2 phosphorylation, whereas selective inhibition of the α or βI isoforms with the GF109203X compound did not (Fig. 5 e and f). Accordingly, TRPV1 activation by RTX directly activates PKCβII (Fig. 5g). Finally, to confirm that β-arrestin2 was a central chaperone in the ERK signaling cascade engaged by TRPV1 stimulation, we tested the effect of knocking out β-arrestin2 in capsaicinstimulated TRPV1 cells. Western blot analysis indicated an absence of ERK1/2 activation at 5 min. of TRPV1 stimulation on cells treated with β-arrestin2 but not scrambled siRNA (Fig. 5h). Altogether, these findings indicate that activation of the TRPV1 channel engages a signaling cascade involving both a Ca2 + and PKCβII-dependent ERK1/2 activation, which coincides with the translocation of β-arrestin2 into the nucleus. To assess the functional consequence of -arrestin2 nuclear translocation on MOR function and desensitization, we measured N-type (Cav2.2) voltage-gated calcium channel inhibition, as a surrogate of MOR-induced G protein coupling, following prolonged exposure to agonist. In control conditions, activation of MOR by DAMGO (1 µM) induces a rapid and robust inhibition of Cav2.2 N-type current in both transiently transfected HEK cells (47.23 ± 5.21 %, n=15) and native DRG neurons (55.57 ± 4.38 %, n=8) (Fig. 6). In contrast, absence of MOR or activation of TRPV1 alone does not promote calcium current inhibition (Supplementary Fig. 7a), confirming that agonistbound receptors mediate the current inhibition. Following DAMGO pre-incubation for an hour, desensitized MOR does not modulate Cav2.2 currents, in both HEK (4.86 ± 2.83 % (n=9) and neurons (5.72 ± 2.42 % (n=7)) ( Fig. 6), an effect relieved by a 60 min. wash out allowing recycling of internalized MOR (Supplementary Fig. 7b). We then tested whether activating TRPV1 rescues the N-type current inhibition by preventing β-arrestin-2 recruitment to MOR upon DAMGO exposure. As shown in Fig. 6, pre-incubation of cells with a combination of DAMGO and capsaicin (100nM) restores the MOR-driven Ntype inhibition (51.29 ± 9.42 % (HEK cells n=9 , Fig 6a and b) and 53.27 ± 4.5 % (neuron n=8, Fig. 6c and d), respectively), an effect lost in the absence of TRPV1 in both HEK cells (Supplementary Fig. 7c) and DRG neurons (Fig 6d) (Fig 6a and b).

TRPV1 activation prevents MOR desensitization
To determine whether TRPV1 contributes to the increased efficacy of exogenous opioids in inflammatory conditions 37 , we used an in vivo protocol of acute opioid receptor desensitization in which mice were treated with 10mg/kg of morphine i.p. twice daily for three days. This protocol causes a decrease in morphine efficiency measured with the tail flick latency test over the three day course of treatment in WT non-inflamed mice (Fig 7   a, blue bar graphs), and TRPV1 -/non inflamed mice (Fig 7 b, blue bar graphs). We then assessed analgesia in morphine-desensitized WT and TRPV1 -/mice subjected to CFA injection. Under these conditions, morphine analgesia was maintained in WT mice even after 3 days of repetitive morphine treatment (Fig 7 a, red bar graphs). This effect was dependent on the presence of TRPV1, as TRPV1 -/mice show a decreased morphine analgesia in inflammatory conditions (Fig 7 b, red bar graphs). These results suggested that TRPV1 is responsible for maintaining opioid receptor signaling efficacy in inflammatory conditions. To address whether this facilitatory process occurs centrally or at the periphery, we assessed the dose-response of morphine analgesia after intrathecal or local injection, in CFA-inflamed mice, treated or not with morphine for three days (Fig.   8a). Specifically, we compared morphine antinociceptive response in mice that were never exposed to morphine with mice that received 3 days of morphine injections. To determine morphine potency (ED50), these mice were given escalating doses of morphine via intrathecal (central) or local (peripheral) injection into the inflamed paw (Fig 8). When administered intrathecally in morphine-naïve mice, morphine dosedependently attenuated CFA-induced mechanical hypersensitivity: the ED50 was comparable in WT and TRPV1-/-mice, suggesting that TRPV1 did not alter morphine antinociceptive potency in mice without previous exposure to morphine (Fig. 8b and c).
In other words, acute morphine antinociception is not impacted by the absence of TRPV1. By contrast, following repeated morphine treatment, there was a notable reduction in morphine anti-nociceptive potency (i.e. increased ED50) (Fig. 8b and c), suggesting decreased capacity of MOR to respond to morphine. When administered locally in the inflamed paw of saline treated CFA-inflamed mice, morphine dosedependently reduced mechanical hypersensitivity (Fig. 8d), thus confirming previous work that peripheral opioid receptors mitigate inflammatory pain under basal conditions 6,38,39 . However, as opposed to what is observed centrally, peripheral morphine potency is maintained in chronic morphine treated WT mice (Fig. 8d and e). This result indicates that local inflammation prevents peripheral MOR desensitization upon chronic morphine treatment. Strikingly, this effect is dependent on the presence of TRPV1, as TRPV1 -/mice exhibit a decrease in peripheral morphine potency (> ED50) as observed for centrally administered morphine (Fig 8d and e). Thus, our findings show that TRPV1 is responsible for maintaining peripheral opioid receptor signaling in nociceptors in the setting of inflammation.
Overall, our results suggest a model wherein activation of TRPV1 during an inflammatory insult translocates β-arrestin2 to the nucleus, which in turn prevents opioidinduced recruitment of β-arrestin2 to MOR and ensuing receptor desensitization, thus increasing opioid potency at peripheral nociceptors. molecular underpinnings of this process involving the TRPV1-triggered translocation of β-arrestin2 to the nucleus. We found that TRPV1-/-mice are insensitive to endogenous opioid analgesia during resolution of inflammation. We then show that TRPV1 channel signaling overrides desensitization and maintains peripheral opioid receptor function, when using an in vitro or in vivo opioid receptor desensitization paradigm.
Agonists of the TRPV1 channel, including capsaicin containing cream have been extensively used as local analgesics. While Ca2 + dependent TRPV1 channel desensitization, inhibition of ion channels and nerve degeneration have been suggested to contribute to the analgesic effects of capsaicin, these mechanisms remain poorly defined, are reversible, and thus do not explain the long lasting pain relieving effects of capsaicin that last for weeks after treatment 45-47 . Previous work from Chen et al. 48 determined that RTX-induced ablation of TRPV1-expressing afferent neurons prolonged opioid analgesia despite a reduction in MOR expression. Nevertheless, our findings indicate that even when keeping the integrity of the neurons, TRPV1 channel activity regulates β-arrestin2 mobilization to fine-tune opioid receptor function when it's most needed. Additional work has described a direct inhibition of voltage-gated calcium channels by capsaicin in DRG neurons 49-51 which could contribute to decrease excitatory neurotransmission in the afferent pain pathway. We did not observe such effect in our hands when preincubating HEK cells or DRG neurons with DAMGO+capsaicin before wash out and recording Ntype current inhibition. Furthermore, our BRET analysis suggests a shift in MOR signaling likely caused by an absence of β-arrestin2 interaction upon channel activation.
Our electrophysiology data indeed indicate that Gβɣ modulation of N-type is restored by TRPV1 activation, yet a direct assessment of Gi/o coupling to MOR will ascertain that G TRPV1 controls opioid analgesia.

Basso et al.
protein biased signaling is favored upon TRPV1 channel stimulation. The main target of opioids used in the treatment of acute and chronic pain is represented by MOR and their effect is mediated partly by inhibition of high voltage gated N-type channels 52 . Sustained agonist activation of MOR, however, produces cellular opioid tolerance shown to be caused by β-arrestin2-dependent internalization of MOR which are trafficked to the endosomal compartments (for a review see 53 ). It is well established that receptor desensitization and internalization are two processes that participate, among other mechanisms (resensitization, downregulation or de novo receptor synthesis) to opioid tolerance 54-56 . Therefore, further work will test this TRPV1-MOR paradigm in mouse models of opioid tolerance.
Finally our findings that β-arrestin2 translocates to the nucleus following channel activation will raise some new lines of research on the genetic and epigenetic mechanisms of pain modulation by this TRPV1-β-arrestin2 pathway. In contrast to previous studies showing physical association of β-arrestin2 with TRPV1, we could not find evidence of direct interaction between TRPV1 and β-arrestin2 using a BRET assay, to the nucleus does not argue for apoptotic pathways to be engaged by the channel.
However, we believe that transcriptional regulation mediated by TRPV1-β-arrestin2 signaling could contribute to the analgesic effect of topical capsaicin formulations used for pain management. Future proteomic and transcriptional studies may advance our knowledge of the factors interacting with β-arrestin2 as well as the genes regulated by TRPV1 signaling. It is likely that β-arrestin2 has an important function in the plasticity of nociceptive circuits post-inflammation.
Our new data emphasize information that appeared upon completion of our study, showing that TRPV1 activation blocks the opioid-dependent phosphorylation of the MOR by G protein-coupled receptor kinase 5 (GRK5) 61 . Thus, not only would TRPV1 activation attenuate the GRK-mediated recruitment of β-arrestin2 to trigger MOR internalization-desensitization, but the TRPV1-mediated translocation of β-arrestin2 out of the cytosol would also diminish its impact on MOR and likely other GPCRs present in nociceptors. Taken together, our work, along with the data of Scherer and coworkers, support a new concept that TRPV1 can bias MOR signaling via a dual mechanism involving the inhibition of receptor phosphorylation and the shuttling of β-arrestin2 to the nucleus.
Of note, we found that, although preventing calcium influx via TRPV1 blocks β-arrestin2 translocation, raising intracellular calcium by activating the calcium-permeant TRPA1 channel, or by using the calcium ionophore, ionomycin, did not stimulate the nuclear trafficking of β-arrestin2. Therefore, the dynamics of these events will have to be investigated further to understand the spatial and temporal requirements for the calcium TRPV1 controls opioid analgesia.
Basso et al.
signaling to trigger β-arrestin2 translocation, and to determine if there is co-trafficking of other signaling constituents (e.g. GRK5) along with β-arrestin2. To sum up, the rapid dual effect of TRPV1 activation to block GRK5 phosphorylation and to stimulate β-arrestin2 translocation can work together to leave the MOR in a sensitized state.
Overall, our work has unraveled a TRPV1-MOR interplay that governs opioid analgesia during inflammation. These results establish a novel framework for understanding the regulation of opioid receptor function in response to inflammation and the maladaptive processes that could lead to pathological pain during tissue healing. While clinical observations suggest that opioids are effective analgesics following acute injury, prolonged treatment for chronic pain conditions is often accompanied by side effects that may relate to MOR desensitization and the subsequent requirement of higher opioid doses to achieve the same analgesic effect. Our proposed model points to the TRPV1 as a molecular target, which could tailor strategies to enhance opioid efficacy early on while reducing tolerance. Therefore, if used in conjunction, TRPV1 agonists could improve the management of chronic and severe morphine resistant pain by high jacking endogenous mechanisms of receptor desensitization 62 , potentiating opioid analgesia and thus limiting dose escalation.

Acknowledgements:
We would like to thank Dr. Graciela Pineyro for technical assistance with the BRET assay and Michael Bruchas for his generous gift of the β-arrestin2-Venus and pRLuc8.  PAR2-YFP has been previously described 63 . Rat mu opioid receptor 1 (MOR1)-YFP was a gift from Gerald Zamponi and was used to generate MOR1-HA by cloning stickyended oligos containing the HA tag in place of YFP. MOR-Rluc8 was produced by using PCR to introduce AgeI and NotI sites to RLuc8 (a gift from Michael Bruchas) to clone in place of YFP. β-arrestin2-Venus was generated as previously described 64 (supplementary fig 4).

Western blot assay
Two days after transfection, cells were treated with agonist, harvested, pelleted and lysed