Blocking microglial pannexin-1 channels alleviates morphine withdrawal in rodents

Opiates are essential for treating pain, but termination of opiate therapy can cause a debilitating withdrawal syndrome in chronic users. To alleviate or avoid the aversive symptoms of withdrawal, many of these individuals continue to use opiates. Withdrawal is therefore a key determinant of opiate use in dependent individuals, yet its underlying mechanisms are poorly understood and effective therapies are lacking. Here, we identify the pannexin-1 (Panx1) channel as a therapeutic target in opiate withdrawal. We show that withdrawal from morphine induces long-term synaptic facilitation in lamina I and II neurons within the rodent spinal dorsal horn, a principal site of action for opiate analgesia. Genetic ablation of Panx1 in microglia abolished the spinal synaptic facilitation and ameliorated the sequelae of morphine withdrawal. Panx1 is unique in its permeability to molecules up to 1 kDa in size and its release of ATP. We show that Panx1 activation drives ATP release from microglia during morphine withdrawal and that degrading endogenous spinal ATP by administering apyrase produces a reduction in withdrawal behaviors. Conversely, we found that pharmacological inhibition of ATP breakdown exacerbates withdrawal. Treatment with a Panx1-blocking peptide (10panx) or the clinically used broad-spectrum Panx1 blockers, mefloquine or probenecid, suppressed ATP release and reduced withdrawal severity. Our results demonstrate that Panx1-mediated ATP release from microglia is required for morphine withdrawal in rodents and that blocking Panx1 alleviates the severity of withdrawal without affecting opiate analgesia.

Opiates are essential for treating pain, but termination of opiate therapy can cause a debilitating withdrawal syndrome in chronic users. To alleviate or avoid the aversive symptoms of withdrawal, many of these individuals continue to use opiates [1][2][3][4] . Withdrawal is therefore a key determinant of opiate use in dependent individuals, yet its underlying mechanisms are poorly understood and effective therapies are lacking. Here, we identify the pannexin-1 (Panx1) channel as a therapeutic target in opiate withdrawal. We show that withdrawal from morphine induces long-term synaptic facilitation in lamina I and II neurons within the rodent spinal dorsal horn, a principal site of action for opiate analgesia. Genetic ablation of Panx1 in microglia abolished the spinal synaptic facilitation and ameliorated the sequelae of morphine withdrawal. Panx1 is unique in its permeability to molecules up to 1 kDa in size and its release of ATP 5,6 . We show that Panx1 activation drives ATP release from microglia during morphine withdrawal and that degrading endogenous spinal ATP by administering apyrase produces a reduction in withdrawal behaviors. Conversely, we found that pharmacological inhibition of ATP breakdown exacerbates withdrawal. Treatment with a Panx1-blocking peptide ( 10 panx) or the clinically used broad-spectrum Panx1 blockers, mefloquine or probenecid, suppressed ATP release and reduced withdrawal severity. Our results demonstrate that Panx1-mediated ATP release from microglia is required for morphine withdrawal in rodents and that blocking Panx1 alleviates the severity of withdrawal without affecting opiate analgesia.
We investigated the mechanisms underlying opiate withdrawal by administering an opioid receptor antagonist (naloxone, 2 mg per kg body weight (mg/kg) intraperitoneally (i.p.)) to rats chronically treated for 5 d with systemic morphine sulfate (MS, Supplementary  Fig. 1a). Naloxone administration precipitated a spectrum of autonomic and somatic withdrawal signs in morphine-but not saline-treated rats (Supplementary Fig. 1b), the overall severity of which was expressed as a cumulative withdrawal score (Fig. 1a). In morphine-treated rats, CD11b immunoreactivity was significantly elevated within the spinal dorsal horn, indicating that spinal microglia are activated by morphine treatment (Fig. 1b and Supplementary  Fig. 2). Depletion of spinal lumbar microglia by intrathecal injections of a saporin (Sap)-conjugated antibody to Mac1 (Mac1-Sap; 20 µg) decreased withdrawal behaviors (Supplementary Fig. 1b) and attenuated the severity of withdrawal (Fig. 1a) without affecting morphine antinociception (Fig. 1c). By contrast, intrathecal injection of a nontargeting Sap (20 µg) control had no effect on spinal CD11b immunoreactivity (Fig. 1b) or naloxone-induced morphine withdrawal (Supplementary Fig. 1b).
In the central nervous system, Panx1 channels are expressed on neurons and glia 7,8 . Flow cytometric analysis of adult lumbar spinal cord homogenates revealed that morphine treatment resulted in elevated Panx1 expression on CD11b-positive cells (i.e., microglia) but not CD11b-negative cells (i.e., neurons and astrocytes) ( Fig. 1d and Supplementary Fig. 3). We acutely isolated CD11b-positive cells from the spinal cord of morphine-or saline-treated rats and assessed Panx1 channel function by measuring BzATP-evoked uptake of YO-PRO-1 dye, an indicator of Panx1 channel activation 9,10 . YO-PRO-1 uptake was significantly greater in CD11b-positive cells isolated from morphine-treated rats (Fig. 1e). This response was blocked by 10 panx (10 µM) 5,11 but was not affected by a scrambled peptide ( scr panx) (Fig. 1e). Both the increase in Panx1 protein expression (Supplementary Fig. 4a) and YO-PRO-1 uptake (Fig. 1f) were recapitulated in a BV-2 microglial-like cell line 12 exposed to morphine for 5 d in vitro. Dye uptake was further potentiated in the presence of naloxone (10 µM) (Fig. 1f) and blocked by 10 panx (10 µM) (Fig. 1g). Application of extracellular solution (ECS) as a vehicle control for BzATP did not evoke dye uptake (Fig. 1f). Moreover, treatment of cultured BV-2 cells with morphine and CTAP (a selective µ-receptor antagonist) together prevented the increase in Panx1 expression and function (Supplementary Fig. 4a,c), whereas repeated administration of the synthetic µ-receptor agonist peptide DAMGO mimicked the effects of morphine on Panx1 (Supplementary Fig. 4b,d).
Cell turnover rates differ between central and peripheral CX 3 CR 1expressing populations 13 . Therefore, in another cohort of Cx3cr1-Cre ERT2 ::Panx1 flx/flx mice, morphine injections were initiated 28 d after tamoxifen treatment to allow for repopulation of peripheral but not central CX 3 CR 1 -expressing cells. In this cohort of Panx1-deficient mice, the reduction in morphine withdrawal was comparable to that **** *** a b d S / s c r p a n x C T R / 1 0 p a n x C T R / s c r p a n x observed in mice receiving morphine injections 7 d after tamoxifen treatment (Fig. 2d), suggesting that Panx1 on microglia, not peripheral CX 3 CR 1 -expressing cells, is required for morphine withdrawal. Furthermore, morphine withdrawal was attenuated in both male and female Panx1-deficient mice (Supplementary Fig. 7), indicating that microglial Panx1 critically contributes to opiate withdrawal in both sexes. As another control, we administered tamoxifen to mice lacking inducible Cre recombinase (Panx1 flx/flx ) and found that the severity of naloxone-precipitated withdrawal was indistinguishable from vehicle-treated Panx1-expressing mice (Cx3cr1-Cre ERT2 ::Panx1 flx/flx ) (Fig. 2d). Thus, tamoxifen per se does not affect morphine withdrawal. To further investigate the requirement for microglial Panx1 in morphine withdrawal, we compared the actions of 10 panx (10 µg) in Panx1-deficient and Panx1-expressing mice. We reasoned that if the effects of 10 panx were mediated by blocking Panx1 on microglia, they would be lost in the absence of microglial Panx1. Indeed, intrath-ecal injection of 10 panx before naloxone challenge significantly attenuated morphine withdrawal in Panx1-expressing mice ( Fig. 2e) but did not further ameliorate withdrawal in Panx1-deficient mice (Fig. 2e). This loss of 10 panx effect, together with the amelioration of withdrawal in Panx1-deficient mice, implicates microglial Panx1 in morphine withdrawal.
To investigate the mechanisms underlying Panx1-mediated withdrawal, we first measured spinal cFos expression as a cellular correlate of neuronal activation 14 . We determined that naloxone-precipitated withdrawal increased the number of cFos-positive neurons within the spinal dorsal horn of Panx1-expressing mice; this increase was suppressed in mice lacking microglial Panx1 (Fig. 3a). To assess neuronal function more directly, we used a whole lumbar spinal cord preparation with intact dorsal roots 15 prepared from Panx1-expressing or microglial Panx1-deficient adult mice that had been treated for 5 d with saline or morphine. We electrically stimulated the dorsal root to evoke postsynaptic field potentials (fPSPs) recorded from laminae I-II of the spinal dorsal horn and found that bath application of naloxone (10 µM) produced a slow-rising facilitation of fPSPs that persisted for at least 60 min in morphine-but not saline-treated Panx1-expressing mice (Fig. 3b,c). By contrast, this response to naloxone did not occur in morphine-treated Panx1-deficient mice (Fig. 3b,c). The absence of synaptic facilitation in these mutant mice was not due to a general defect in spinal synaptic facilitation, because electrically stimulating the dorsal roots of Panx1-deficient mice at low frequency (2 Hz) produced a robust and long-lasting increase in fPSPs (Fig. 3d,e).
Because ATP release is a key consequence of Panx1 activation 6,16 , we asked whether Panx1-mediated ATP release occurs during withdrawal. To test this, we bath-applied naloxone (10 µM) to spinal cord slices isolated from Panx1-expressing and Panx1-deficient mice that had been treated for 5 d with saline or morphine. We measured the amount of ATP in spinal superfusates and found that the level of ATP in response to naloxone was significantly higher in morphine-than in saline-treated Panx1-expressing mice (Fig. 4a). Naloxone-induced ATP release was not observed in slices prepared from Panx1-deficient mice (Fig. 4a). To test whether ATP is directly released from microglia, we applied naloxone to microglia in culture. Supernatants of morphine-treated microglia showed increase in ATP, and this increase was blocked by 10 panx (Fig. 4b). Thus, naloxone induces the release of ATP from morphine-treated microglia via a Panx1-dependent mechanism.
To test whether ATP contributes to morphine withdrawal, we administered ATPγS (100 µM) intrathecally to Panx1-deficient mice, which display attenuated morphine withdrawal behaviors. In these mice, local delivery of the ATP analog together with naloxone challenge restored a spectrum of withdrawal behaviors; these behaviors were not observed when ATPγS was administered to saline-treated Panx1-deficient mice (Fig. 4c). We reasoned that if ATP is required for morphine withdrawal, then altering endogenously released ATP in the spinal cord might influence withdrawal behaviors. In morphine-treated rats, we tested this possibility by intrathecal administration of the ATP-degrading enzyme apyrase (10 units) 17 , which produced a notable reduction in withdrawal score (Fig. 4d). Conversely, inhibiting ATP breakdown by intrathecally administering the ecto-ATPase inhibitor ARL67156 (10 nmol) 18,19 exacerbated morphine withdrawal (Fig. 4d). We therefore conclude that ATP is a key substrate for morphine withdrawal. P2X7R activation is a core mechanism for opening Panx1 channels 5,10 . We determined that P2X7R expression was selectively increased in spinal microglia isolated from morphine-dependent rats (Supplementary Fig. 8a), and blockade of these receptors by intrathecal administration of the selective P2X7R antagonist A-740003 attenuated withdrawal (Supplementary Fig. 8b). These findings position P2X7R as an upstream signaling node through which morphine potentiates Panx1 function, defining P2X7R-Panx1 as a microglial signaling ensemble required for morphine withdrawal. Specifically targeting Panx1 while leaving P2X7R function intact could, however, be the preferred strategy for treating opiate withdrawal.
Having established that Panx1 is crucially involved in morphine withdrawal, we tested as proof of concept two clinically used broadspectrum Panx1 inhibitors: probenecid, an anti-gout medication 20 , and mefloquine, an antimalarial drug 21 . In morphine-treated rats, systemic administration of probenecid or mefloquine 60 min before naloxone challenge significantly ameliorated morphine withdrawal ( Fig. 4e and Supplementary Fig. 9a). These compounds also blocked BzATPevoked potentiation of Panx1 activation ( Fig. 4f and Supplementary  Fig. 9b) and suppressed ATP release in morphine-treated microglia cultures ( Fig. 4g and Supplementary Fig. 9c) without altering P2X7Rmediated calcium responses (Supplementary Fig. 10a,b).
To further extend the potential clinical utility of probenecid, we tested the actions of this compound in a model of spontaneous morphine withdrawal that does not rely on naloxone challenge. Mice given a 5-d morphine treatment showed robust withdrawal behaviors 24 h after morphine cessation, and these behaviors were attenuated by a systemic injection of probenecid (50 mg/kg, i.p.) (Fig. 4h). Spontaneous withdrawal was also reduced in Panx1-deficient mice (Fig. 4i), suggesting a common microglial Panx1-dependent mechanism in both naloxone-precipitated and spontaneous withdrawal. Finally, we investigated the effects of probenecid on fentanyl, another clinically used and widely abused opioid. In fentanyl-treated mice, naloxone precipitated robust withdrawal that was ameliorated by probenecid (Fig. 4j). Taken together, the effects of probenecid and mefloquine in alleviating opiate withdrawal open the possibility that these, and other clinically available broad-spectrum Panx1 inhibitors, have potential application in the treatment of opiate withdrawal.
In summary, we have discovered that a substrate within the spinal cord-microglial Panx1-mediated ATP release-underlies the cellular and behavioral correlates of morphine withdrawal. We propose that Panx1 activation is a fundamental mechanism through which microglia cause long-term synaptic facilitation in spinal lamina I-II neurons during withdrawal. This form of neuronal hyperexcitability is a cardinal feature of withdrawal 14,22 , and we demonstrate here that this central tenet of withdrawal occurs directly in the spinal cord of morphine-treated animals. Because opiate withdrawal engages both spinal and supraspinal mechanisms, including the periaqueductal gray 14 , locus coeruleus 23 , ventral tegmental area [24][25][26] and nucleus accumbens 27 , descending inputs from these supraspinal regions may influence spinal facilitation in the intact system. However, our isolated spinal cord preparation suggests that intrinsic spinal mechanisms are sufficient for synaptic facilitation during withdrawal, and as further evidence for a spinal locus of opiate action, we show that spinal manipulations targeting Panx1 channels or ATP levels effectively block or recapitulate the behavioral and cellular corollaries of withdrawal. Although we do not rule out contributions from supraspinal mechanisms, the potent alleviation of opiate withdrawal by clinically used Panx1-blocking drugs (probenecid and mefloquine) highlights the potential of our findings for immediate therapeutic translation. Targeting Panx1 channels may therefore represent a new clinical strategy for alleviating withdrawal symptoms without impairing morphine analgesia.

MeTHODs
Methods, including statements of data availability and any associated accession codes and references, are available in the online version of the paper.

ONLINe MeTHODs
Animals. Adult male rats and mice were used (rats aged 7-8 weeks; mice aged 8-12 weeks for behavioral experiments and 6-10 weeks for electrophysiology experiments). Mice and rats were housed under a 12-h/12-h light/dark cycle with ad libitum access to food and water and were randomly allocated to different test groups. All experiments were approved by the University of Calgary, Université Laval and University of California Irvine Animal Care Committees and are in accordance with the guidelines of the Canadian Council on Animal Care and the US National Institutes of Health Guide for the Care of Use of Laboratory Animals.
Morphine dosing paradigm and nociceptive behavioral models. Morphine sulfate (PCCA) prepared in 0.9% sterile saline solution was injected i.p. twice daily (8 a.m. and 5 p.m.) into male Sprague-Dawley rats (weight ranging 200-250 g, escalating doses from 10 to 45 mg/kg) 28 , C57BL/6 mice, male and female Cx3cr1-Cre ERT2 ::Panx1 flx/flx or Panx1 flx/flx mice (weight ranging 16-30 g, escalating doses from 7.5 to 50 mg/kg). Thermal nociceptive thresholds were assessed using the tail-flick test (rats) and the tail-immersion test (mice). For rats, an infrared thermal stimulus (Ugo Basile) was applied to the ventral surface of the tail, and time latency for tail removal from the stimulus was recorded. For mice, the distal portion of the tail was submerged in a 50 °C water bath, and time latency for tail removal from the stimulus was recorded. A maximum cutoff time was set at 10 s to prevent tissue damage. Nociceptive measurements were taken before and 30 min after morphine injections, and values were normalized to daily baseline. In a subset of mouse experiments, a time course of morphine-induced antinociception was performed at 30, 45, 60, 90 and 120 min after a single acute injection of morphine (7.5 mg/kg).
Behavioral assessment of morphine withdrawal. Rats and mice received ascending doses of morphine i.p. at 8-h intervals (rats: day 1, 10 mg/kg; day 2, 20 mg/kg; day 3, 30 mg/kg; day 4, 40 mg/kg; mice: day 1, 7.5 and 15 mg/kg; day 2, 20 and 25 mg/kg; day 3, 30 and 35 mg/kg; day 4, 40 and 45 mg/kg). On day 5, rats and mice received a morning injection of morphine (rats, 45 mg/kg; mice, 50 mg/kg) and 2 h later naloxone (2 mg/kg, naloxone hydrochloride dihydrate, Sigma) to rapidly induce opiate withdrawal. Control rats and mice received equivalent volumes of 0.9% saline and were challenged with naloxone on day 5. Signs of withdrawal were recorded as previously described 29 . Jumping, teeth chattering, wet-dog shakes, headshakes and grooming behaviors were evaluated at 5-min intervals for a total test period of 30 min, and a standardized score of 0 to 3 was assigned to each behavior. Allodynia, piloerection, salivation, ejaculation and tremors or twitching were also evaluated, with 1 point given to the presence of the behavior during each 5-min interval. All signs were counted and compiled to yield a cumulative withdrawal score. Rats and mice were weighed before and after naloxone challenge to calculate weight loss. In all behavioral studies, experimenters were blind to the drug treatment group and genetic profile of rats and mice. In a subset of experiments, spontaneous withdrawal behaviors were assessed in mice 24 h after last morphine or control injection (on day 6).

Intrathecal drug administration.
In a subset of experiments, rats and mice were subject to drug administration by intrathecal injection under 2% isoflurane (vol/vol) by lumbar puncture method as previously described 30 . Mac1-Sap or Sap (20 µg, Advanced Targeting Systems) was injected on days 1 and 3 before morning morphine injections. Intrathecal injections of 10 panx (10 µg, WRQAAFVDSY, New England Peptide), scr panx (10 µg, FSVYWAQADR, New England Peptide) and A-740003 (10 µL of 10 µM, Sigma) were delivered 60 min before naloxone-precipitated withdrawal, and intrathecal apyrase (10 units, Sigma) and ARL67156 (10 nmol, Sigma) were delivered 15 min before naloxone. ATPγS (100 µM, Roche) was delivered intrathecally immediately before naloxone-precipitated withdrawal without the use of isoflurane. All control animals received an equivalent volume of intrathecal 0.9% sterile saline on a timeline identical to that used for treated animals from the same experiment.
Assessment of sensorimotor coordination. Sensorimotor coordination was evaluated using the accelerating rotarod test (Panlab, Harvard Apparatus). All mice underwent two conditioning trials before testing. Motor behavior was analyzed 7 d after last tamoxifen or vehicle injection. In brief, mice were placed on a rotating rod set to gradually accelerate from 4 to 40 r.p.m. over 5 min. The trial was terminated when the mouse fell off the apparatus, and both time and terminal speed (r.p.m.) were recorded.
Behavioral assessment of fentanyl withdrawal. Fentanyl citrate (West Ward Pharmaceuticals) prepared in 0.9% sterile saline solution was injected i.p. at ascending doses (day 1, 5 and 25 µg/kg at 8-h intervals; day 2, 50 µg/kg at 8-h intervals; day 3, 75 µg/kg at 8-h intervals; day 4, 100 µg/kg at 8-h intervals; day 5, 150 µg/kg at 8-h intervals; day 6 150 µg/kg at 4-h intervals). On day 7, mice received two injections of fentanyl 4 h apart (150 µg/kg) and 30 min after the second dose naloxone (10 mg/kg, Sigma) to rapidly induce opiate withdrawal. Control mice received an equivalent volume of 0.9% saline and were challenged with naloxone on day 7. Signs of withdrawal were recorded after naloxone injection for a 20-min test period as described above for morphine withdrawal. The experimenter was blind to the drug treatment groups of the mice.
Isolation of adult mixed neuron-glia culture. Rats and mice were anaesthetized and perfused transcardially with PBS. Spinal cord and (for mice only) brain tissues were isolated and placed in HBSS. After blunt dissociation, spinal cord contents were filtered through a 70 µm cell strainer into DMEM containing 10 mM HEPES and 5% FBS. Isotonic Percoll (density 1.23 g/mL) was added to the cell suspension, followed by a 1.08 g per mL Percoll underlay 31 . Samples were spun at 3,000 r.p.m. for 30 min at 20 °C. Following centrifugation, myelin debris was removed, and the interface between Percoll gradients was collected and transferred into fresh medium. Samples were centrifuged again at 1,350 r.p.m. for 10 min at 4 °C, and the pellet was reconstituted in PBS containing 10% FBS (for flow cytometry) or DMEM containing 10% FBS and 1% penicillin-streptomycin (for live-cell imaging).
BV-2 microglia culture. BV-2 microglial-like cells were provided by M. Tsuda and K. Biber 12 . Cells were maintained in DMEM (Gibco) containing 5% FBS and 1% penicillin-streptomycin (Pen-Strep) at 37°C with 5% CO 2 and treated with morphine (10 µM), morphine and CTAP (1 µM), DAMGO (1 µM), saline and CTAP, or saline once a day for 5 d. BV-2 microglial-like cells were found to be free of mycoplasma contamination using MycoFluor Mycoplasma Detection Kit (Molecular Probes).  32 . In a subset of experiments, mefloquine (40 µM, Sigma or 100 nM, Bioblocks) or probenecid (1 mM, Life Technologies) were bath applied in ECS after loading of Fura-2 a.m. and incubated at room temperature for 30 min before BzATP (100 µM) stimulation. All experiments were conducted at room temperature using an inverted microscope (Nikon Eclipse Ti C1SI Spectral Confocal), and the fluorescence of individual microglia was recorded using EasyRatioPro software (PTI). Excitation light was generated from a xenon arc lamp and passed alternatingly through 340-or 380-nm band-pass filters (Omega Optical). The 340/380 fluorescence ratio was calculated after baseline subtraction.
Generation of Cx3cr1-Cre ERT2 ::Panx1 flx/flx mice. Mice with microglial-specific deletion of Panx1 were generated using a Cre-loxP system. Panx1 flx/flx homozygote mice 33 containing loxP sequences flanking exon 2 of the Panx1 gene were crossed with C57BL6/J mice expressing Cre-ERT2 fusion protein and enhanced yellow fluorescent protein (eYFP) under the Cx3cr1 promoter (Jax mice: B6.129P2(Cg)-Cx3cr1 tm2.1(cre/ERT)Litt /WganJ, stock number 021160). Progeny genotype was screened by PCR, and homozygous Panx1 flx/flx and Cx3cr1-Cre ERT2 mice were bred and backcrossed for 8 generations to yield the conditional knockout Cx3cr1-Cre ERT2 ::Panx1 flx/flx mice. To induce Cre recombination, mice were injected i.p. with 1 mg per day tamoxifen (Sigma) for 5 d. Control mice were littermates that received vehicle injections (sunflower oil with 10% ethanol) for 5 d, while tamoxifen-related effects were controlled for using Panx1 flx/flx littermate mice that received 5 d of tamoxifen injections. In the majority of experiments, testing occurred 7 d after final tamoxifen injection. In a subset of experiments, behavioral assessment of morphine withdrawal was conducted 28 d after final tamoxifen injection to control for effects of peripheral Cx3cr1-expressing cells.
YO-PRO dye uptake. After 5 d of morphine or saline treatment, cell culture medium was removed, and the BV-2 cells were incubated in ECS containing either YO-PRO-1 or YO-PRO-3 dye (2.5 µM, Invitrogen). After a 5-min baseline recording, cells were stimulated with BzATP (150 µM, Sigma), and dye uptake was recorded for 30 min. Cell viability was assessed immediately after 30 min recording by application of ionomycin (1 µM, Sigma). Cells that did not respond to ionomycin were excluded from the analysis. YO-PRO-1 dye fluorescence emission (491/509 nm) or YO-PRO-3 dye fluorescent emission (612/631 nm) was detected at 37 °C using a FilterMax F5 plate reader (Molecular Devices). Drugs used included 10 panx (10 µM, New England Peptide), scr panx (10 µM, New England Peptide), naloxone (10 µM, Sigma), probenecid (1 mM, Life Technologies), and mefloquine (40 µM, Sigma or 100 nM, Bioblocks). Drugs were bath applied in ECS containing YO-PRO-1 dye (in the absence of morphine) and incubated at 37 °C for 10 min before BzATP stimulation. For the BV-2 cell experiments, fluorescence emission at 30 min after BzATP application was calculated as an average of the entire imaging field as percentage change from baseline, and responses at 30 min were averaged over multiple independent experiments. Representative images of BV-2 YO-PRO-1 dye uptake were taken using EasyRatioPro software (PTI) at room temperature using an inverted microscope (Nikon Eclipse Ti C1S1 Spectral Confocal). Individual traces of BV-2 YO-PRO-1 dye uptake were taken at 10-min intervals and represent the average response of all BV-2 cells in the recorded view from a subset of experiments. To assess Panx1 function in adult microglia, neuron-glia mixture culture was isolated from adult Sprague-Dawley rats treated with saline or morphine for 5 d, or from Cx3cr1-Cre ERT2 ::Panx1 flx/flx mice treated with tamoxifen or vehicle for 5 d as described above. Once isolated, cells were plated in DMEM containing 10% FBS and 1% Pen-Strep and incubated overnight at 37 °C with 5% CO 2 . Cells were washed and incubated with DAPI (1:10,000) for 10 min, and then incubated in YO-PRO-1 dye (rats) or YO-PRO-3 dye (mice). In rats, microglia were identified from mixed adult neuronal-glial culture using a fluorophore-conjugated antibody CD11b/c-PE (1:500, eBioscience), and in mice microglia were identified by endogenous expression of eYFP. Fluorescence of individual microglia were recorded for a 5-min baseline period and then for 30 min after BzATP stimulation (300 µM). Individual traces were taken at 10-min intervals and represent responses from a subset of individual microglia. Fluorescence emission at 30 min was expressed as percentage change from the baseline recorded from the same cell.
Naloxone-induced ATP release. ATP levels were detected using bioluminescence by combining samples with recombinant firefly luciferase and its substrate d-luciferin (ATP Determination Kit, Life Technologies). The ATPase inhibitor (ARL67156, 1 µM, Sigma) was added to the ECS or aCSF (in mM: 126 NaCl, 1.6 KCl, 1.1 NaH 2 PO 4 , 1.4 MgCl 2 , 2.4 CaCl 2 , 26 NaHCO 3 , 11 glucose) to minimize breakdown of ATP, and samples were incubated in medium for 30 min before naloxone challenge to reduce mechanically induced ATP release. Samples were exposed to naloxone (10 µM) for 30 min, then the supernatant was collected and samples read immediately using a FilterMax F5 plate reader at 28 °C. For experiments with BV-2 microglial-like cultures, cells were incubated in 10 panx (10 µM, New England Peptide), scr panx (10 µM, New England Peptide), probenecid (1 mM, Life Technologies), mefloquine (40 µM, Sigma or 100 nM, Bioblocks) or ECS for 10 min before naloxone exposure. Final ATP measurement was expressed relatively to control samples (saline-treated BV-2 cells exposed to ECS) from the same plate. For ATP release in spinal cord slices, tamoxifen-or vehicle-treated Cx3cr1-Cre ERT2 ::Panx1 flx/flx mice were perfused with ice-cold oxygenated sucrose-substituted aCSF (s-aCSF in mM: 252 sucrose, 2.5 KCl, 1.5 CaCl 2 , 6 MgCl 2 , 10 d-glucose), and the spinal cord rapidly dissected. The lumbar segment of the spinal cord was sliced into 300 µm sections, and incubated in oxygenated 32 °C aCSF for 1 h. Spinal cord slices were then transferred to room temperature oxygenated aCSF and exposed to naloxone (10 µM). For quantification, ATP release was normalized to total protein of each sample.