Targeting prostaglandin E2 receptor 2 in Schwann cells inhibits inflammatory pain but not inflammation.

The research conducted complies with all relevant ethical regulations. Behavioral studies followed Animal Research: Reporting of In Vivo Experiments (ARRIVE) guidelines 70 . Animal experiments and sample collections were carried out according to the European Union (EU) guidelines for animal care procedures and Italian legislation (DLgs 26/2014) application of the EU Directive 2010/63/EU. All animal studies were approved by the Animal Ethics Committee of the University of Florence and the Italian Ministry of Health (permits no. 765/2019-PR, 288/2021-PR). Male and female mice C57BL/6 J (Charles River, RRID: IMSR_JAX:000664) were used throughout (25–30 g, 6–8 weeks old). Littermate wild-type ( Trpv1 +/+ ) and TRPV1-deficient mice ( Trpv1 –/– , B6.129×1-Trpv1tm1Jul/J, 25–30 g, 6–8 weeks; The Jackson Laboratory) generated by C57BL/6 J background were used. To generate mice in which the Trpa1 gene was conditionally silenced in Schwann cells, homozygous 129S-Trpa1 tm2Kykw/J (floxed Trpa1, Trpa1 fl/fl , RRID:IMSR_JAX: 008649 Jackson Laboratory) were crossed with hemizygous B6.Cg-Tg(Plp1-CreERT)3Pop/J mice ( Plp-Cre ERT , RRID: IMSR_JAX:005975 Jackson Laboratory) expressing a tamoxifen-inducible Cre in Schwann cells (Plp1, proteolipid protein myelin 1) 18 . The progeny ( Plp-Cre ERT ;Trpa1 fl/fl ) was genotyped using PCR for Trpa1 and Plp-Cre ERT . Mice that were negative for Plp-Cre ERT ( Plp-Cre ERT– ;Trpa1 fl/fl ) were used as control. Both positive and negative mice for Cre ERT and homozygous floxed Trpa1 ( Plp-Trpa1 and control respectively) were treated with intraperitoneal (i.p.) 4-hydroxytamoxifen (4-OHT, 1 mg/100 μL in corn oil once a day consecutively for 3 days). To selectively delete Trpa1 in primary sensory neurons, Trpa1 fl/fl mice were crossed with hemizygous Advillin-Cre mice (Adv-Cre) 18 , 71 . Mice positive or negative for Cre and homozygous for floxed Trpa1 ( Adv-Trpa1 and control respectively) were used. The successful Cre-driven deletion of TRPA1 mRNA was confirmed using reverse transcription quantitative real-time PCR (RT-qPCR). Some Plp-Cre ERT + or Plp-Cre ERT - were treated with 4-OHT (i.p., 1 mg/100 μL in corn oil, once a day consecutively for 3 days) before the infection with AAV for selective silencing of the different genes in Schwann cells. Hemizygous Advillin-Cre mice ( Adv-Cre + ) or their control ( Adv-Cre - ) 18 were also used for the infection with AAV for selective silencing of the different genes in primary sensory neurons. The group size of n = 8 mice for behavioral experiments was determined by sample size estimation using G Power [v3.1 72 ] to detect the size effect in a post-hoc test with type 1 and 2 error rates of 5% and 20%, respectively. Allocation concealment of mice into the vehicle(s) or treatment groups was performed using a randomization procedure ( http://www.randomizer.org/ ). The assessors were blinded to the identity of the animals (genetic background) or allocation to treatment groups. None of the animals were excluded from the study. Mice were housed in a temperature- and humidity-controlled vivarium (12 h dark/light cycle, free access to food and water, 5 animals per cage). At least 1 h before behavioral experiments, mice were acclimatized to the experimental room and behavior was evaluated between 9:00 am and 5:00 pm. Animals were anesthetized with a mixture of ketamine and xylazine (90 mg/kg and 3 mg/kg, respectively, i.p.) and euthanized with inhaled CO 2 plus 10–50% O 2 . Primary human Schwann cells (hSCs) (# P10351 , Innoprot or #10HU-188, iXCells Biotechnologies) were cultured and maintained in Schwann cell medium (# P60123 , Innoprot or #MD0055, iXCells Biotechnologies) at 37 °C in 5% CO 2 and 95% O 2 19 . After 12 passages, cells were discarded and replaced. Human kidney epithelial (HEK293T) cells (#CRL-3216™, American Type Culture Collection) were cultured in DMEM supplemented with FBS (10%), L-glutamine (2 mM), penicillin (100 U/ml) and streptomycin (100 mg/ml) at 37 °C in 5% CO 2 and 95% O 2 . AAVpro 293T cells (#632273, Takara), were maintained in DMEM high glucose supplemented with 10% heat inactivated FBS, 4 mM L-glutamine,1 mM penicillin/streptomycin and 1 mM sodium pyruvate at 37 °C in 5% CO 2 and 95% O 2 . The day before transfection, cells were plated in DMEM supplemented with 2% FBS. Supplementary Tables  1 – 3 , reported in Supplementary Information, provide doses, routes of administration, and concentrations of all pharmacological agents and AAVs used in the study. Butaprost, L-902.688, Ro-1138452, L-798,106, ER-819762 and TG6-10-1 were purchased from D.B.A. S.r.l, phospholipase A 2 activating protein was purchased from Enzo Biochem Inc. If not otherwise indicated, the reagents were obtained from Merck Life Science SRL or Tocris Bioscience. Mice received intraplantar (i.pl., 10 μl/site) injection of PGE 2 (0.1, 0.5, 1.5, 5 nmol) or its vehicle (1% ethanol in 0.9% NaCl); (R)-Butaprost (0.1, 0.5, 1.5, 5 nmol), L-902,688 (0.5, 1.5 and 5 nmol), arachidonic acid (1, 10, and 100 nmol) or their vehicle (4% dimethyl sulfoxide, DMSO, 4% Tween 80 in 0.9% NaCl); phospholipase A 2 activating protein (0.5, 2, and 10 nmol), λ-carrageenan (300 μg), complete Freund’s adjuvant (CFA, 1 mg/ml) or their vehicle (0.9% NaCl). PF-04448948 (5 nmol), BGC 20-1531 (1, 5, 10 nmol), indomethacin (280 nmol), SC-51322 (5 nmol), Ro-1138452 (5 nmol), L-798,106 (5 nmol), ONOAE3-208 (5 nmol), CJ-42794 (5 nmol), TG6-10-1(5 nmol), L-NAME (1 μmole), ER-819762 (1, 5, 10 nmol), Dyngo-4a (500 pmol), PitStop2 (500 pmol) and SQ25536 (25 nmol), A967079 (300 nmol), H-89 (45 nmol), ML171 (250 nmol) or their vehicle (4% DMSO, 4% Tween80 in 0.9% NaCl), olgecepant (1 nmol) and st-Ht31 (17 nmol) or their vehicle (1% DMSO in 0.9% NaCl), PBN (1 μmol) or its vehicle (0.9% NaCl) were administered (10 μL, i.pl.) 30 min before the algesic stimuli (i.pl.). In the CFA model, drugs were administered at day 2 after CFA injection. In another set of experiment, diclofenac (25 mg/kg, i.p.) or PF-04448948 (10 mg/kg, intragastric, i.g.) or their vehicle (4% DMSO, 4% Tween80 in 0.9% NaCl and 0.5% carboxymethyl cellulose, respectively) were administered once a day from day 0 to day 6 after CFA injection. Perisciatic injection was made as previously reported 58 . Briefly, PGE 2 (1.5 nmol) or veh were injected in the region surrounding the sciatic nerve at high thigh level of the right hind limb (from ∼ 10 to ∼ 16 mm from the paw surface) without skin incision in a volume of 6 μl using a microsyringe with a 30-gauge needle. The position of sciatic nerve at high thigh level was chosen by using the femoral head as a landmark. Plp-Cre + and control or Adv-Cre + and control mice were infected with an intrasciatic (i.sc., 5 µL, 1×10 12  v/g) or intrathecal (i.th., 5 µL,1×10 12  v/g) injection of different AAVs. For intrasciatic injection mice were anesthetized and sciatic nerve was exposed. A volume of 3 µl of viral vectors was directly injected into the sciatic nerve through a 33-gauge needle and a Hamilton syringe connected to a manual micropump (World Precision Instruments). The needle remained in place at the injection site for 1 additional min, before it was slowly removed. Animals were used 3 weeks after AAVs infection. Sciatic nerve and DRGs were harvested for evaluating AAVs infection. In optogenetics experiments Plp-Cre + and control mice infected with an intraplantar (i.pl., 10 µL, 1×10 12  v/g) injection of different AAVs were used. After 3 weeks, the hind paw of infected mice was stimulated with a blue light (pulsed, 1 s/4 s, 450-nm light, 30 LED intensity) or red light (not pulsed, 650-nm light, 48 LED intensity) for 10 min before i.pl. CFA, Cg, or vehicle. The hind paw was harvested for evaluating AAV infection. Immediately after i.pl. injection, mice were placed inside a plexiglass chamber, and non-evoked nociception response was assessed for 15 min by measuring the time (sec) that the animal spent in lifting, biting, licking, shaking the injected paw. The non-evoked nociceptive response was also assessed in mice 3 h and 2 days after i.pl. Cg, CFA, respectively or vehicle. The mechanical paw-withdrawal threshold was measured using von Frey filaments of increasing stiffness (0.02–2 g) applied to the plantar surface of the mouse hind paw, according to the up-and-down paradigm 73 . The 50% mechanical paw-withdrawal threshold (g) response was then calculated from the resulting scores. Mechanical paw-withdrawal threshold was measured at baseline and at different time following treatments. Mice were placed on a hot plate (Ugo Basile) set at 50 ± 0.1 °C. The latency to the first hind paw licking/withdrawal was taken as an index of the nociceptive threshold and detected before (basal) and after treatments. Cutoff time was set at 30 s. Paw thickness (in millimeters) was measured using 0-12 mm stainless digital thickness gauge (Mitutoyo Corporation). Paw thickness was measured at baseline, and at different time following i.pl. Cg, CFA or vehicle and indomethacin or vehicle. Spontaneous pain was tested using the Mouse Grimace Scale 74 . Briefly, each mouse was placed in an individual chamber (9 × 5 × 5 cm) having transparent Plexiglas walls to allow experimental observation. Mice were habituated for 30 min before behavioral testing. Cameras directed at the front of the cubicle recorded 30 min of facial expressions. One clear facial image was taken for every 3 min interval, scrambled and scored blindly for facial grimacing 74 . The scorers assessed facial expressions such as orbital tightening (closing or narrowing of the eyelid and orbital area), ear position (outward or backwards rotation of the ears), scoring either a 0 (not present), 1 (moderately visible) or 2 (severe) depending on the magnitude of the expression. Nose and cheek bulging, and whisker change were not score because they were difficult to distinguish against the dark fur of the mice. The test was performed in mice 3 h and 2 days after i.pl. Cg, CFA, respectively or vehicle. Primary culture of mouse Schwann cells (mSCs) was obtained from sciatic nerves of C57BL/6 J mice 21 . Briefly, the epineurium was removed, and nerve explants were cut (1 mm segments) and dissociated in Hank’s Balanced Salt Solution (HBSS, 2 h, 37 °C) using hyaluronidase (0.1%) and collagenase (0.05%). After centrifugation (150 x  g , 10 min, RT), the pellet was resuspended and cultured in DMEM containing streptomycin (100 mg/ml), fetal calf serum (10%), forskolin (2 μM), L-glutamine (2 mM), penicillin (100 U/ml) and neuregulin (10 nM). Cytosine arabinoside (Ara-C, 10 mM) was added three days later, to remove fibroblasts. Cells were cultured at 37 °C in 5% CO 2 and 95% O 2 for 15 d before experiments by replacing the culture medium every 3 days. Mouse hindpaws were collected on ice cold HBSS, washed in HBSS and fat and blood vessels were gently removed from skin under a light microscope. This skin was cut into small cubes and digested in dispase (5 mg/ml) for 2 h at 37 °C in agitation (200 rpm). Epidermis was discarded and the dermis was minced and digested in collagenase Type IV (2 mg/ml) for 2 h at 37 °C in agitation (200 rpm). The tissues were centrifuged and washed twice in HBSS to remove debris and then dissociated by mechanical pipetting in 5 ml of HBSS containing DNase enzyme (400 µg/ml), and further incubated for 1 h at 37 °C in agitation (200 rpm). Samples were then centrifuged (300 x  g ) for 5 min and the pellet resuspended and cultured in DMEM supplemented with FBS (10 %), forskolin (2 μM), L-glutamine (2 mM), penicillin (100 U/ml), N2 supplement (1%). Cytosine arabinoside (Ara-C, 10 mM) was added three days later, to remove fibroblasts 75 , 76 . Cells were cultured at 37 °C in 5% CO 2 and 95% O 2 for 6 days. Using NanoBiT® MCS Starter System (#N2014, Promega), N- and C-terminal LgBiT and SmBiT fusions proteins were generated. Eight different constructs amplifying PKA and hTRPA1 synthetized by Vector Builder: pBit2.1-C PKA-SmBit (P1/P2 and NheI-HF/SacI); pBit2.1-N-SmBIT-PKA (P3/P4 with SacI/XbaI); pBit1.1-C-hTRPA1-LgBIT (P5/P6 with AsiSI/XhoI); pBit1.1-N-LgBIT- hTRPA1 (P7/P8 with XhoI/XbaI) were generated according to manufacturer’s instructions. Using a Gibson assembly strategy, pBit1.1-C-PKA-LgBit (P9/P10 for PKA fragment and P11/P12 for acceptor vector) and pBit2.1-N-SmBit-hTRPA1 (P13/P14 for hTRPA1 fragment and P15/P16 for acceptor vector) were cloned. pBit1.1-N-LgBit-PKA was generated by Gibson assembly strategy with P17/P18 for PKA fragment and P19/P20 for acceptor vector. pBit2.1-C-hTRPA1-SmBit was generated by amplifying hTRPA1 and cloned in pBit2.1 acceptor vector, P21/P22 and P23/P24 primers were used. Plasmids were cloned using Gibson Assembly Cloning Kit (#E5510S, New England Biolabs). pBit1.1-N-LgBit-PKA and pBit2.1-C-hTRPA1-SmBit was the only combination with a BRET signal after PGE 2 stimulation. In vitro studies . pCDNA3-AKAR4, pCDNA3-AKAP79-AKAR4, pCDNA3-Lyn11-AKAR4 and pCDNA3-AKAP79-(Ci-Ce)-Epac2 were used 48 . A blue light photo-activable AC from the Soil Bacterium Beggiatoa (bPAC) was amplified from pAAV-CMV-bPAC-T2A-mCherry (Vector Builder) and cloned into pCDNA3. Primers P25/P26 and NheI-HF/EcoRI-HF were used. pCDNA3-Lyn11-Lyn11-bPAC was generated by cloning the Lyn11-Lyn11 sequence into the pCDNA3-bPAC plasmid, P27/P28/P29/P30 oligos and NheI-HF/BamHI were used. pCDNA3-AKAP79-bPAC was generated by amplifying AKAP79 from pCDNA3-AKAP79-AKAR4 48 and cloned into pCDNA3-bPAC, P31/P32 primers and NheI-HF/BamHI were used. pCDNA4-AKAP79-R-FlincA was generated by replacing AKAP79 from pCDNA3-AKAP79-(Ci-Ce)-Epac2 in pCDNA4-R-FlincA ( R ed Fl uorescent in dicator for cA MP), kindly donated by Dr. Horikawa K 77 , BamHI-HF/XhoI were used. pCDNA4-Lyn11-Lyn11-R-FlincA was generated by replacing AKAP79 in the pCDNA4-AKAP79-R-FlincA with Lyn11-Lyn11 expressing adaptors P33/P34/P35/P36; sticky ends from HindIII-HF/BamHI were used. pCDNA3-LAPD was generated by replacing bPAC in the pCDNA3-bPAC with LAPD from pCG084 plasmid, gently donated by Dr. Möglich A 53 , P37/P38 primers and NheI-HF/XbaI were used. pCDNA-AKAP79-LAPD was generated by cloning AKAP79 from pCDNA3-AKAP79-bPAC into pCDNA3-LAPD, using NheI sticky ends. P39/P40 primers were used. pCDNA3-Lyn11-Lyn11-LAPD was generated by annealed Lyn11-Lyn11 expressing adaptors (P41/P42/P43/P44) and cloned with NheI-HF. pAAV EF1-alpha-MCS was generated by replacing the mCherry in the pAAV-EF1-alpha rev(mCherry) with a multicloning site (MCS), P45/P46 primers were used. EF1-alpha-promoter was replaced with a short-CMV to obtain pAAV-CMV-MCS that was used as acceptor plasmid for all the constructs. pAAV-CMV-bPAC-T2A-mCherry was generated by cloning loxP-bPAC-T2A-mCherry-loxP from pAAV-CAG-bPAC-T2A-mCherry (Vector Builder) into pAAV-CMV-MCS, P49/P50 primers and SpeI-HF/HpaI were used. pAAV-CMV-LAPD-FLAG and pAAV-CMV-FLAG-LAPD were generated by PCR amplification from pCG084 of LAPD-FLAG and FLAG-LAPD and cloning into pAAV-CMV-MCS, P51/P52 or P53/P54 primers and SpeI-HF/HpaI and NheI-HF/SnaBI were used respectively. pAAV-CMV-Lyn11-Lyn11-bPAC-T2A-mCherry and pAAV-CMV-Lyn11-Lyn11-LAPD-FLAG were generated by cloning a Lyn11Lyn11 expressing adaptors (P33/P34/P35/P36) into pAAV-CMV-bPAC-T2A-mCherry and pAAV CMV-LAPD-FLAG, AvrII/HpaI were used. pAAV-CMV-AKAP79-bPAC-T2A-mCherry and pAAV-CMV-AKAP79-LAPD-FLAG were generated by cloning AKAP79 from pCDNA3-AKAP79-AKAR4, into pAAV-CMV-bPAC-T2A-mCherry and pAAV-CMV-LAPD-FLAG, P31/P55 primers and NheI-HF/AvrII were used. To not exceed the cargo limit of AAVs, WPRE and BGH poly(A) were removed from pAAV-CMV-AKAP79-LAPD-FLAG (EcoRV/PmlI) and WPRE sequence was removed from pAAV-CMV-AKAP79-bPAC-T2A-mCherry (ClaI). All constructs were confirmed by Sanger sequencing. All primers (P) sequences are available in the Supplementary Data  1 . To obtain plasmids coding for short hairpin RNA targeting mPtger2, mPtger4, mAkap150 or scrambled shRNA as negative control, the plasmid pAAV[FLEXon]-CMV LL-rev(EGFP-5’ miR-30E-BfuAI ORF_44 bp BfuAI-3’ miR-30E)-rev(LL)-WPRE was designed and ordered (Vectorbuilder). shRNAs targeting the gene of interest (P56/P57 shRNA mPtger2; P58/P59 shRNA 2 mPtger2; P60/P61 shRNA mPtger4; P62/63 shRNA 2 mPtger4; P64/P65 shRNA mAkap150; Vectorbuilder) or scrambled shRNA (P66/P67 scrambled mPtger2; P68/P69 scrambled mPtger4; Invivogen) were cloned by replacing the 44bp ORF with preannealed and phosphorylated oligonucleotides by using BfuAI compatible bases. Recombinant AAV particles (rAAVs) were produced by using triple transfection strategy 19 . Briefly, AAVpro 293 T cells (#632273, Takara), were transfected with polyethylenimine (#23966, PEI, Polyscience) with a DNA:PEI ratio of 1:3 78 . To obtain rAAVs, AAVpro 293 T cells were transiently transfected with 2.5 mg total DNA (plasmid expressing genes of interest, pAdDeltaF6; #11287, Addgene and Rep/Cap, 1:1:1 molar ratio). To infect with high efficiency Schwann cells and primary sensor neurons, Rep/Cap 2/8 or 2/9n were used, respectively (pAAV2/8 #112864; pAAV 2/9n #112865, Addgene). rAAVs were extracted and isolated 72 h post-transfection, purified by iodixanol gradient ultracentrifugation, concentrated, and titrated using a RT-qPCR assay (#6233 AAVpro Titration Kit, Takara) according to the manufacturer instructions. HSCs were plated in 96-well poly-L-lysine-coated (8.3 μM) black clear bottom plates (5 × 10 5 cells/well; PerkinElmer) and transfected with cDNA (130-300 ng) using jetOPTIMUS® DNA transfection reagent (#55-250, Polyplus) for 16–24 h at 37 °C in 5% CO 2 and 95% O 2 . The day of experiments, hSCs were washed and added with HBSS at pH 7.4 at 37 °C. All the experiments were performed using a fluorescent microscope for recording Axio Observer 7; with a fast filter wheel and Digi-4 lens to record excitations and Ultra-fast Sutter Lambda DG4 Xenon excitation source (range 300–700 nM) (Zeiss) with 20x or 40x objectives or using FlexStation3 Multi-Mode Microplate Reader (Molecular Devices) with SoftMax® Pro7 software (Molecular Devices). To measure cAMP signaling in real time, cells were infected with a cADDIs BacMam virus (1.09 × 10 9  v/g/mL) encoding the green upward cAMP sensor (#U0200G, Montana Molecular). The fluorescent signal was recorded by microscopy ex/em 506/517 nm (filter set: FT 495, ex BP 470/40, em BP 525/50, interval 1 s) or using FlexStation3 (ex/em 485/515 nm, interval 1.5 s), a Red Membrane-Targeted fS15 cADDIs membrane sensor (#U0241R, Montana Molecular), ex/em 558/603 nm (filter set: FT 570, ex BP 550/25, em BP 605/70, interval 1 s) or a R-FlincA ex/em 568/592 nm (filter set: FT 495, ex BP 470/40, em BP 525/50, interval 1 s). HSCs transfected with pCDNA4-R-FlincA were cultured at 32 °C for 24 h before imaging. HSCs expressing the green upward cAMP sensor were stimulated with PGE 2 (100 fM-10 nM). The response to PGE 2 (100 pM) was evaluated in the presence of PF-04418948 (100 pM-1 µM) or BGC20-1531 (100 pM-1 µM). In another set of experiments, cells were stimulated with L-902,688 (1 nM-10 µM) or (R)-Butaprost (10 nM-100 µM). HSCs, mSCs from sciatic nerves and cutaneous derived SCs expressing Red Membrane-Targeted fS15 cADDIs membrane sensor, were stimulated with PGE 2 (10 nM) in the presence of PF-04418948 (1 µM) or BGC20-1531 (1 µM). In hSCs, single concentration of L-902,688 (100 µM) or (R)-Butaprost (100 µM) were tested. With R-FlincA sensor, hSCs were stimulated with PGE 2 (10 nM) in the presence of PF-04418948 (1 µM) or BGC20-1531 (1 µM). Single concentration of L-902,688 (100 µM) or (R)-Butaprost (100 µM) were tested. Signals were recorded for 360 s, the ΔF/F0 ratio was calculated for each experiment and the results were expressed as the area under the curve (AUC). bPAC was used to selectively induce an optogenetic cAMP increase in different subcellular compartments. HSCs were co-transfected with pCDNA-bPAC or pCDNA-Lyn11-Lyn11-bPAC and the red fluorescent cAMP sensor cADDIs (#U0200R for total cAMP, or #U0241R for Lyn11-Lyn11, Montana Molecular), or with pCDNA-AKAP79-bPAC and the pCDNA-AKAP79-R-FlincA sensor. HSCs were stimulated with a 450 nm light (pulsed, 1 s/5 s, 30% led intensity) for 10 min (led, #UHP-T-450-EP, controller #UHPTLCC-02-USB, Pulser USB for TTL pulse train generator, Prizmatik) and recorded with the fluorescent microscope (ex/em 558/603 nm; filter set: FT 570, ex BP 550/25, em BP 605/70, acquisition every 1 s for total and Lyn11-Lyn11-bPAC and 5 s for AKAP79-bPAC). Despite the presence of forskolin in the medium of cultured SCs (2 μM), a much higher concentration (100 μM) was able to produce a marked cAMP increase and therefore was used as positive control for cAMP increase. In some experiments the membrane cAMP formation after activation of Lyn11-bPAC was evaluated in the presence of PF-04418948 (1 µM) or BGC 20-1531 (1 µM). The ΔF/F0 ratio was calculated for each experiment and the results were expressed as AUC define. To further evaluate the role of cAMP, hSCs were co-transfected with pCDNA-Lyn11-Lyn11-LAPD and pCDNA-Lyn11-Lyn11-R-FlincA or, with pCDNA-AKAP79- LAPD (AKAP79/150-LAPD) and AKAP79/150-R-FlincA. HSCs were exposed to a 650 nm light continuously (48% light intensity) for 10 min (#UHP-T-650-EP, Prizmatikl) and then stimulated with PGE 2 (10 nM). To evaluate PKA activity, the FRET-based PKA activity sensor AKAR4 (pCDNA3-AKAR4) was used (filter set: channel1 FT 455, ex BP 436/25, em BP 535/30, interval 1 s, channel2 BS 420, ex BP 436/25, em BP 480/40, interval 1 s). HSCs were stimulated with PGE 2 (100 pM-10 nM). The response to PGE 2 (10 nM) was evaluated in the presence of PF-04418948 (1 µM) or BGC 20-1531 (1 µM). HSCs were exposed to different concentrations of (R)-Butaprost (10 µM - 500 µM) or L-902,688 (10 µM-500 µM). HSCs transfected with plasma membrane associated AKAR4 (pCDNA3-Lyn-AKAR4) or plasma membrane-localized scaffold protein AKAP79 (pCDNA3-AKAP79-AKAR4) were stimulated with PGE 2 (10 nM) in the presence of PF-04418948 (1 µM) or BGC 20-1531 (1 µM). HSCs were stimulated with single concentrations of (R)-Butaprost (100 µM) or L-902,688 (100 µM). FRET changes were measured as a ratio of the acceptor fluorophore emission (545 nm) to donor emission (480 nm). Signals were recorded for 360 s. The ΔF/F0 ratio was calculated for each experiment, and the results were expressed as the AUC. HSCs were plated on poly-L-lysine-coated (8.3 μM) 35 mm glass coverslips and maintained at 37 °C in 5% CO 2 and 95% O 2 for 24 h. Cells were loaded (40 min) with Fura-2 AM-ester (5 μM) added to the buffer solution (37 °C) containing (in mM) 2 CaCl 2 ; 5.4 KCl; 0.4 MgSO 4 ; 135 NaCl; 10 D-glucose; 10 HEPES and bovine serum albumin (BSA, 0.1%) at pH 7.4. Cells were washed and transferred to a chamber on the stage of a fluorescent microscope for recording (Olympus IX 81) and exposed to PGE 2 (10 nM) or vehicle (0.9% NaCl) in the presence of PF (1 μM), H-89 (1 μM) or A967079 (50 μM) and the calcium response was monitored. Results were expressed as percent increase in ratio340/380 over baseline normalized to the maximum effect induced by ionomycin (5 μM) added at the end of each experiment. A genetically encoded probe for H 2 O 2 -HyPer7.2 [HyPer7.2 79 , kindly donated by Dr. Emrah Eroglu, Harvard Medical School, Boston, US] was used on live cells. HSCs were plated on poly-L-lysine-coated (8.3 μM) 35-mm glass coverslips and transfected with DNA (2 μg) of HyPer7.2 using jetOPTIMUS ® DNA transfection reagent (#55-250; Polyplus, Lexington, MA, USA). After 24–48 h, the hSCs were washed and transferred to a chamber on the stage of a fluorescent microscope for recording (Axio Observer 7; with a fast filterwheel and Digi-4 lens to record excitations; ZEISS, Stuttgart, Germany). Cells were exposed to PGE 2 (10 nM), or its vehicle (0.9% NaCl), in the presence of PF (1 μM), H-89 (1 μM), A967079 (50 μM) or ML171 (1 μM) and H 2 O 2 variations were monitored for approximately 15 min. Results were expressed as the percentage increase in the ratio 408/455 over the baseline, the ΔF/F0 ratio was calculated for each experiment and the results were expressed as the AUC. To evaluate the dynamics of PKA-TRPA1 proximity in hSCs, a time-resolve luciferase re-complementation assay was used. HSCs were plated in 96-well poly-L-lysine-coated (8.3 μM) white clear bottom plates (5 × 10 5 cells/well; #6005181, PerkinElmer) the day before transfection. pBit2.1-C-hTRPA1-SmBit was co-transfected with pBit1.1-N-LgBit-PKA (130 ng DNA) with jetOPTIMUS® DNA transfection reagent (#55-250, Polyplus, Lexington). Forty-eight h after the transfection cells were treated for 10 min with Nano-Glo® Live Cell Reagent, according to manufacture instructions (#N2011, Nano-Glo ® Live Cell Assay System Promega). Then, hSCs were stimulated with PGE 2 (100 nM), CGRP (10 µM) or vehicle. The response to PGE 2 (100 nM) was evaluated in the presence of PF-04418948 (100 nM) or BGC 20-1531 (100 nM) or st-Ht31 (10 μM) or their vehicle (0.001% DMSO). To evaluate the dynamic of PKA-TRPA1 proximity in HEK293T cells, the same approach of hSCs was used. SmBit-TRPA1 was co-transfected with PKA-LgBit (130 ng DNA) with Ptger2 or Ptger4 expressing plasmids (6.5 ng of DNA) using PEI (#23966, Polyscience). Forty-eight h after the transfection HEK293T cells were stimulated with PGE 2 (100 nM). Luminescence of both experiments was measured for 6 min using a luminescence plate reader (FlexStation 3 Multi-Mode Microplate Reader; Molecular Devices) with SoftMax ® Pro7 software (Molecular Devices). The results are expressed as arbitrary units (AU) The ΔF/F0 ratio was calculated for each experiment, and the results were expressed as the AUC. cDNA encoding human EP2 and EP4 were cloned with a C-terminal monomeric (m) Venus tag on the C-terminus with a flexible linker (LRPLGSSGGGGGGSG). HSCs transfected with EP2-Venus or EP4-Venus (3 µg/10 cm dish), were plated in poly-D-lysine-coated 35 mm glass-bottomed dishes. After 48 h, cells were washed in HBSS. PGE 2 (10 µM) or Veh was added, and imaging obtained after 30 min incubation. Images were obtained using a Leica SPi8 confocal microscope (63X objective) and processed in ImageJ. Anesthetized mice were transcardially perfused with PBS and 4% paraformaldehyde. Lumbar DRGs and sciatic nerve tissues were collected, postfixed for 24 h, and paraffin-embedded or cryoprotected in 30% sucrose. Human DRGs (#0062-HP-240, Gentaur), mouse DRGs, human sciatic nerve (#0062-HP-261, Gentaur), mouse paw tissue, and mouse sciatic nerve were incubated with different primary antibodies: EP2 (#ab167171, rabbit monoclonal, Abcam, 1:250 or #APR-064, rabbit polyclonal, Alomone Labs, 1:50), EP4 (#BS-8538R, rabbit polyclonal, Bioss, 1:200), S100B (#MA1-26621, mouse monoclonal, Invitrogen, 1:50 or #ab196175, rabbit monoclonal, Abcam, 1:50), Sox10 (#AF2864, goat polyclonal, R&D Systems, 1:100), NeuN (#MAB377, mouse monoclonal, Merck, 1:250 or #MAB377X, mouse monoclonal, Merck, 1:250), GFP (#A6455, rabbit polyclonal, Invitrogen, 1:500) and glutamine synthetase (#ab64613, mouse monoclonal, Abcam, 1:1000), myelin basic protein (MBP) (#ab133620, rabbit monoclonal, Abcam, 1:500), diluted in fresh blocking solution (PBS, pH 7.4, 5% normal goat serum (NGS) or normal donkey serum (NDS)). We underline, as a limitation of the colocalization method, that detection of the biomarkers localized in two or more different intracellular compartments of the same cell may be affected by the slice orientation that may result in incomplete colocalization. For FFPE sections, the primary antibody was used after antibody retrieval preformed with tri-sodium citrate buffer (pH 6) for EP2 and tris-EDTA buffer (pH 9). Sections were then incubated with the fluorescent polyclonal secondary antibodies Alexa Fluor® 488 (#A32731, goat polyclonal anti-rabbit, Invitrogen, 1:600), Alexa Fluor® 488 (#A32790, donkey polyclonal anti-rabbit, Invitrogen, 1:600), Alexa Fluor® 594 (#A11005, goat polyclonal anti-mouse, Invitrogen, 1:600), Alexa Fluor® 594 (#A21207, donkey polyclonal anti-rabbit, Invitrogen, 1:600), Alexa Fluor® 594 (#A11058, donkey polyclonal anti-goat, Invitrogen, 1:600) and coverslipped using the mounting medium with DAPI (#ab104139, Abcam). All slides were visualized and analyzed using a Zeiss Axio Imager 2 microscope with Z-stacks in the Apotome mode (Zeiss). Pearson correlation (Rcoloc) values in the colocalization analysis were calculated using the colocalization Plugin of the ImageJ (v.1.54 f; National Institutes of Health, Bethesda, MD, USA). To exclude off-target effects of AAV infection: (1) the total number of DRG neurons (NeuN+ cells) was counted in 10 4 µm 2 boxes in Adv Cre and control mice; (2) the level of myelin basic protein (MBP) was measured as mean fluorescence value in sciatic nerve by Zeiss Zen imaging 2020, in Plp CreERT and control mice. Fluorescent in situ hybridization was performed using RNAscope™ Multiplex Fluorescent V2 Assay (ACDbio), according to manufacturer’s protocol. MSCs and hSCs were plated on fibronectin (#21370, SERVA) coated slides and cultured until confluence. Cells were fixed with paraformaldehyde 4% in PBS for 30 min at RT and dehydrated in 50%, 70% and 100% ethanol, for 1 min. Cells were rehydrated by subsequent steps in 70%, 50% ethanol and PBS followed by a permeabilization with PBS + 0.1% Tween 20 for 10 min at RT. Pretreatment was completed by adding 2–3 drops of H 2 O 2 , to inactivate endogenous peroxidases, followed by 2–3 drops of Protease III (ACDbio), diluted 1:15 in PBS for 10 min at RT. Slides were incubated at 40 °C in HybEZ II Oven (ACDbio). Cells were hybridized for 2 h with probes, to detect Ptger2 and Ptger4 in human (#406791, #406771, Biotechne) and mouse (#456481, #441461, Biotechne) cells. Negative control probe (#310043, Biotechne) was used as negative control. Signal amplification was obtained by consecutive incubations with AMP1, AMP2 and AMP3 reagents and HRP enzyme (ACDbio) for 15 min, followed by TSA Vivid fluorophore (#323271, ACDbio), diluted 1:1500 in TSA Buffer (#322809, ACDbio). Any further HRP activity was stopped by incubation with HRP blocker for 15 min. For immunofluorescence, hSCs cells were permeabilized in PBS + 0.1% Triton 100 for 15 min at RT. Cells were then blocked in in fresh blocking solution PBS, pH 7.4, 5% normal donkey serum (NDS) and incubated O/N at 4 °C with primary anti-SOX10 antibody (#AF2864, goat polyclonal, R&D Systems, 1:200). Cells were than incubated with secondary antibodies Alexa Fluor® 594 (#A11058, donkey polyclonal anti goat, Invitrogen, 1:600) for 2 h at RT and coverslipped using the mounting medium with DAPI. All slides were visualized and analyzed using a Zeiss Axio Imager 2 microscope with Z-stacks in the Apotome mode (Zeiss). HEK293T cells were transfected with pCDNA Lyn11-Lyn11-LAPD (2 µg) and CAAX-rGFP (2 µg) or Rab5a-LAPD (2 µg) and Rab5a-Venus (2 µg) using polyethylenimine (#23966, PEI, Polyscience) with a DNA:PEI ratio of 1:3, incubated overnight and plated on 12 mm glass coverslips. Cells were incubated overnight and fixed (4% paraformaldehyde in PBS) on ice for 20 min. Cells were washed 3 x in PBS and blocked in PBS + 0.3% saponin + 3% NHS for 1 h at RT. Cells were incubated with mouse monoclonal anti-6xHis antibody (1:1000, Takara) in PBS + 0.3% saponin + 1% NHS overnight at 4 °C. Cells were washed and incubated with donkey anti-mouseAlexa568 (1:1000, Invitrogen) for 1 h at RT, counterstained with DAPI (1 µg/ml), washed 3x in PBS and mounted with ProlongGlass hard set mounting medium (Thermofisher). Cells were imaged on a Leica SP8 confocal, and images were processed with ImageJ and Adobe Illustrator. HEK293T cells were transfected with pCDNA Lyn11-Lyn11-R-FlincA (0.2 µg) on 12 mm glass coverslips with polyethylenimine (#23966, PEI, Polyscience). After 24 h, transfected HEK293T cells were imaged on a Leica confocal microscopy (Stellaris 5), and images were processed with Adobe Illustrator. The total RNA from hSCs and mSCs or DRGs and sciatic nerve tissue of C57BL/6 J, Plp-Cre and Adv-Cre mice was extracted by standard TRIzol method together with RNeasy Mini Kit (#74106, QIAGEN) according to manufacturer’s protocol, while total RNA of human DRGs was commercially purchased (#636150, Clontech). RNA concentration and purity were assessed spectrophotometrically by measuring the absorbance at 260/280 nm. The purified RNAs were retro-transcribed using SuperScript IV One-Step RT-PCR System (#12595025, Thermo Fisher) according to manufacturer’s protocol. For relative quantification of mRNA compared to housekeeping gene, real-time PCR was carried out using Rotor Gene Q (Qiagen Spa). The chosen reference gene was GAPDH. The KAPA SYBR FAST Universal Kit (#KK4602) was used for amplification, and the cycling conditions were as follows: samples were heated to 95 °C for 10 min followed by 40 cycles of 95 °C for 10 sec, and 65 °C 95 °C for 20 sec. PCR reaction was performed in triplicate for each sample. Relative expression of mRNA was calculated using the 2 –Δ(ΔCT) comparative method, with each gene normalized against the GAPDH gene for the same sample. The sets of primers are reported in Supplementary Data  2 . PGE 2 IL-1β and TNF-α content was measured in the hind paw homogenates using a single-analyte enzyme-linked immunosorbent assay kit (#ab287802, #ab197742, #ab208348Abcam) according to the manufacturer’s protocol. Data are expressed as pg/mg of protein. MPO activity was measured in the hind paw homogenates using a fluorometric assay kit (#ab111749, Abcam) according to the manufacturer’s protocol. Data are expressed as μU/ml/mg of tissue. The results are expressed as the mean ± SEM. For multiple comparisons, a one-way ANOVA followed by a post-hoc Bonferroni’s test was used. The two groups were compared using Student’s t test. For behavioral experiments with repeated measures, a two-way mixed-model ANOVA followed by a post-hoc Bonferroni’s test was used. Statistical analyses were performed on raw data using GraphPad Prism 8 (GraphPad Software Inc.). P values less than 0.05 ( P   <  0.05) were considered significant. EC 50 values were determined from non-linear regression models using Graph Pad Prism 8. The statistical tests used and sample size for each analysis are shown in the Fig. legends. Further information on research design is available in the  Nature Portfolio Reporting Summary linked to this article.

The identity of the PGE 2 receptor subtype(s) involved in inflammatory pain remains elusive. Studies with global receptor knockout mice or systemically administered receptor antagonists have provided confounding results for many reasons, including opposing activities of central vs . peripheral receptors 14 and incomplete antagonist selectivity 27 , 28 . To find the peripheral EP subtype(s) implicated in PGE 2 -dependent inflammatory pain, agents were administered by local intraplantar (i.pl.) injection into the mouse hindpaw (unless otherwise specified), and, to identify the receptor cellular localization, EP subtypes were selectively silenced in SCs or DRG neurons. In C57BL/6 J (B6) mice injected with PGE 2 , we assessed the two temporally distinct responses reported in previous rodent studies 11 , 29 : a transient (<20 min) non-evoked nociception, consisting of lifting and licking, but not shaking and biting (Fig.  1a and Supplementary Fig.  1a ); and a sustained (∼4 h) hindpaw mechanical allodynia (allodynia) (Fig.  1b ), which were observed solely in the injected paw (Supplementary Fig.  1b ), indicating a locally confined action of PGE 2 . As no difference was found between male and female mice (Supplementary Fig.  1c, d ), to minimize the number of animals used, only key experiments were replicated in both sexes. Fig. 1 EP2 receptor activation in Schwann cells mediates PGE 2 -evoked sustained hindpaw mechanical allodynia (allodynia). a , d , f Dose-dependent non-evoked nociception and b , e , g dose- and time-dependent allodynia after intraplantar (i.pl.) injection of PGE 2 , L-902,688 (L-902), butaprost (Buta) or vehicle (Veh) in C57BL/6 J mice (B6) ( n  = 8 mice/group). c Schematic representation of agonists/antagonists targeting EP2 and EP4 receptors. h Non-evoked nociception and i allodynia after i.pl. PGE 2 (1.5 nmol) or Veh in B6 pretreated with PF-04448948 (PF, 5nmol), BGC 20-1531 (BGC, 5 nmol) or Veh ( n  = 8 mice/group). j RT-qPCR for Ptger2, Ptger4 Avil and S100 mRNA in mouse dorsal root ganglia (DRG) sciatic and cutaneous Schwann cells (SCs) (DRG and sciatic SCs n  = 4, cutaneous n  = 3 independent experiments). k , l Representative images of EP2, EP4, NeuN and S100B expression in mouse DRG and sciatic nerve tissue (scale bar: 20 μm) ( n  = 3 subjects). m Representative images of EP2, EP4 and SOX10 expression in mouse sciatic and cutaneous SCs ( n  = 3 independent experiments). n–r Non-evoked nociception (left panel) and allodynia (right panel) after i.pl. PGE 2 or Veh in Plp-Cre, Adv-Cre or Control mice infected with AAV for selective silencing of EP4 (-Ptger4 ) ( P1p-Ptger4 or Adv-Ptger4 ) ( n,o ) or EP2 (-Ptger2 ) ( P1p-Ptger2 or Adv-Ptger2 ) ( p,q ) or with AAV for a scrambled shRNA ( Plp-scrambled ) ( r ) ( n  = 8 mice/group). Data are mean ±  s.e.m. a , d , f , h , n , o , p , q , r 1-way or b , e , g , i , n , o , p , q , r 2-way ANOVA, Bonferroni correction. * P  < 0.05, ** P  < 0.01, *** P  < 0.001, **** P  < 0.0001 vs. Veh † P  < 0.05, †† P  < 0.01, ††† P  < 0.001, †††† P  < 0.0001 vs. PGE 2 /Veh, Control / PGE 2 . Source data are provided as a Source Data file. a , d , f Dose-dependent non-evoked nociception and b , e , g dose- and time-dependent allodynia after intraplantar (i.pl.) injection of PGE 2 , L-902,688 (L-902), butaprost (Buta) or vehicle (Veh) in C57BL/6 J mice (B6) ( n  = 8 mice/group). c Schematic representation of agonists/antagonists targeting EP2 and EP4 receptors. h Non-evoked nociception and i allodynia after i.pl. PGE 2 (1.5 nmol) or Veh in B6 pretreated with PF-04448948 (PF, 5nmol), BGC 20-1531 (BGC, 5 nmol) or Veh ( n  = 8 mice/group). j RT-qPCR for Ptger2, Ptger4 Avil and S100 mRNA in mouse dorsal root ganglia (DRG) sciatic and cutaneous Schwann cells (SCs) (DRG and sciatic SCs n  = 4, cutaneous n  = 3 independent experiments). k , l Representative images of EP2, EP4, NeuN and S100B expression in mouse DRG and sciatic nerve tissue (scale bar: 20 μm) ( n  = 3 subjects). m Representative images of EP2, EP4 and SOX10 expression in mouse sciatic and cutaneous SCs ( n  = 3 independent experiments). n–r Non-evoked nociception (left panel) and allodynia (right panel) after i.pl. PGE 2 or Veh in Plp-Cre, Adv-Cre or Control mice infected with AAV for selective silencing of EP4 (-Ptger4 ) ( P1p-Ptger4 or Adv-Ptger4 ) ( n,o ) or EP2 (-Ptger2 ) ( P1p-Ptger2 or Adv-Ptger2 ) ( p,q ) or with AAV for a scrambled shRNA ( Plp-scrambled ) ( r ) ( n  = 8 mice/group). Data are mean ±  s.e.m. a , d , f , h , n , o , p , q , r 1-way or b , e , g , i , n , o , p , q , r 2-way ANOVA, Bonferroni correction. * P  < 0.05, ** P  < 0.01, *** P  < 0.001, **** P  < 0.0001 vs. Veh † P  < 0.05, †† P  < 0.01, ††† P  < 0.001, †††† P  < 0.0001 vs. PGE 2 /Veh, Control / PGE 2 . Source data are provided as a Source Data file. To identify the contribution of individual EP receptors to PGE 2 pain-like behaviors, we used more recent and selective agonists and antagonists. Because of the robust evidence that PGE 2 pain-like responses are mediated by cAMP increases 10 , 11 , major attention was paid to EP2 and EP4, which are positively coupled to adenylyl cyclase (AC) activity 30 . The selective EP4 agonist, L-902,688 (L-902) 31 , elicited a transient non-evoked nociception (licking/lifting) and a moderate sustained allodynia (Fig.  1c–e ). In contrast, the selective EP2 agonist, butaprost 32 , elicited a robust allodynia, without detectable non-evoked nociception (Fig.  1f, g ). The EP2 antagonist, PF-04418948 (PF) 28 , reduced allodynia evoked by PGE 2 and butaprost without affecting non-evoked nociception to PGE 2 and L-902 (Fig.  1h, i ; Supplementary Fig.  1e, f ). The EP4 antagonist, BGC20-1531 (BGC) 33 , produced marginal or no reduction of PGE 2 - or butaprost-evoked allodynia, respectively, whereas it dose-dependently reduced non-evoked nociception to PGE 2 and L-902 (Fig.  1h, i ; Supplementary Fig.  1g–k ). The role of EP4 in PGE 2 -induced allodynia 34 , studied using the putative selective antagonist, ER819762 (ER81) 35 , was not confirmed here. Although ER81 reduced PGE 2 - and L-902-induced non-evoked nociception and allodynia, unlike BGC, higher doses of ER81 unselectively attenuated allodynia by the EP2 selective agonist butaprost (Supplementary Fig.  1l–p ). Therefore, previous confounding results are attributable to incomplete selectivity of early EP antagonists. To further strengthen the present results, we used additional EP2 and EP4 antagonists. Two distinct EP4 antagonists, ONOAE3-208 36 (ONOAE) and CJ-42794 37 (CJ-42), attenuated non-evoked nociception and allodynia to a selective EP4 agonist, L-902. They also reduced non-evoked nociception to PGE 2 and caused minimal decrease in allodynia to PGE 2 . Neither compound affected allodynia to the EP2 agonist, butaprost (Supplementary Fig.  2a–j ). Conversely, the EP2 antagonist TG6-10-1 38 (TG6) did not influence responses to L-902 or non-evoked nociception to PGE 2 , but effectively blocked allodynia to both PGE 2 and butaprost (Supplementary Fig.  2k–o ). We reasoned that the two additional PGE 2 receptors, EP1 and EP3, being coupled to G q11 (resulting in calcium mobilization) and G i protein (resulting in cAMP decrease), respectively 30 , are unlikely to be involved in the cAMP-mediated responses to PGE 2 . Furthermore, EP3 stimulation that reduces AC activity was found to inhibit PGE 2 -evoked mechanical allodynia 39 . Consistent with these findings, neither EP1 (SC-51322) nor EP3 (L-798,106) antagonists affected transient non-evoked nociception or sustained allodynia to PGE 2 (Supplementary Fig.  2p–s ). Similarly, prostacyclin receptor (IP) antagonism (Ro-1138452) neither reduced non-evoked nociception nor allodynia to PGE 2 (Supplementary Fig.  2t, u ). Thus, EP4 alone mediates the non-evoked and transient nociception, whereas EP2 primarily mediates the sustained allodynia induced by PGE 2 . Due to the prominent role of mechanical hypersensitivity in clinical inflammatory settings 1 , 40 and the major contribution of EP2 in PGE 2 -induced allodynia, more attention has been paid to this receptor subtype. The essential contribution of SCs to pain-like responses in preclinical models of neuropathic 41 , cancer 19 and migraine 21 pain was confirmed by optogenetic evidence 23 . Whereas both direct and indirect data support the presence and function of EP2 and EP4 in DRG neurons 42 , far less is known about expression and function of these receptors in peripheral glial cells 24 , 26 . We detected EP2 and EP4 mRNA and protein in human and mouse DRG neurons and sciatic nerve SCs and in mouse cutaneous SCs (Supplementary Fig.  3a–e ; Fig.  1j–m ). To identify the cell type that mediates PGE 2 -induced non-evoked nociception and allodynia, we used Plp CreERT or Adv Cre drivers to express short hairpin RNA (shRNA) in SCs of Plp CreERT mice, or in DRG neurons of Adv Cre mice to selectively down-regulate EP2 or EP4 expression. Viruses were packaged with the AAV2/8 or AAV2/9n serotypes for efficient infection of SCs or DRG neurons, respectively, and in both cases with a loxP flanked shRNA of Ptger2 (EP2) or Ptger4 (EP4). Downregulation of EP2 or EP4 transcripts in sciatic nerve and DRG homogenates, and of EP2 or EP4 protein immunoreactivity in S100B (SCs) and NeuN (DRG neurons) positive cells, confirmed the efficiency and selectivity of shRNA-mediated silencing after intrasciatic (i.sc.) AAV2/8 injection in Plp CreERT mice or intrathecal (i.th.) AAV2/9n injection in Adv Cre mice (Supplementary Fig.  4a–c and Supplementary Fig.  5a, b ). As Plp expression has been reported in satellite glial cells (SGCs) 43 , 44 , the presence of EP2 and EP4 was investigated by immunofluorescence in these cells. However, as no immunostaining for either EP2 or EP4 proteins was co-expressed with the SGCs marker, glutamine synthase (GS) (Supplementary Fig.  4d ), the possibility that Plp-induced modification may affect EP2 or EP4 expression in SGCs is unlikely. Expression of the floxed green fluorescence protein (GFP) reporter gene (control marker) in SCs or DRG cell bodies also supported cell-selective targeting in Plp CreERT or Adv Cre mice, respectively (Supplementary Fig.  4e–g and Supplementary Fig.  5c, d ). The absence of inflammatory responses in AAV-shRNA-treated Plp CreERT and control mice confirmed (Supplementary Fig.  4h ) that AAV-shRNA does not per se cause inflammation. The observation that no difference was found in the total number of DRG neurons (NeuN+ cells) in Adv Cre mice and in the levels of myelin basic protein (MBP) in the sciatic nerve trunk of Plp CreERT mice compared to their respective control mice (Supplementary Fig.  6a, b ) further supports the absence of non-specific effects of the silencing procedure. PGE 2 evoked similar non-evoked nociception and allodynia in control and SC EP4 silenced ( Plp-Ptger4 ) mice (Fig.  1n ). EP4 silencing in DRG neurons ( Adv-Ptger4 mice) inhibited markedly non-evoked nociception and only slightly allodynia in response to PGE 2 (Fig.  1o ). In contrast, SC EP2 silencing ( Plp-Ptger2 ) robustly inhibited PGE 2 -induced allodynia but not non-evoked nociception (Fig.  1p ). Non-evoked nociception and allodynia were not reduced in mice with EP2 silenced in DRG neurons ( Adv-Ptger2 mice) (Fig.  1q ). Mice infected with a scrambled shRNA ( scr-shRNA) were used as an additional control. In Plp-scr-shRNA mice, PGE 2 evoked similar non-evoked nociception and allodynia as in control mice (Fig.  1r ). To test efficiency of scrambled-shRNA expression, EP2 and EP4 and GFP protein were evaluated in S100B and NeuN positive cells (Supplementary Fig.  6c-e ). To exclude further off-target effects of shRNA, Plp CreERT mice were infected with a shRNA directed to a different Ptger2 region (Plp-Ptger2-bis mice ) . In Plp-Ptger2-bis mice, PGE 2 -dependent allodynia, but not non-evoked nociception, was reduced (Supplementary Fig.  6f ). Non-evoked nociception and moderate allodynia elicited by L-902 (EP4 agonist) were attenuated in Adv-Ptger4 mice (Supplementary Fig.  6g, h ), whereas the marked butaprost (EP2 agonist)-induced allodynia was attenuated in Plp-Ptger2 mice (Supplementary Fig.  6i ). Non-evoked nociception and/or allodynia to L-902 or butaprost were not affected in any other silenced mouse strain (Supplementary Fig.  6j–t ). Thus, genetic silencing indicates that neuronal EP4 mediates non-evoked nociception whereas EP2 in SCs largely mediates mechanical allodynia induced by PGE 2 . To investigate the contribution of neuronal or glial EP4 and EP2 in pain-like responses elicited by endogenous PGs, we administered arachidonic acid (AA) and phospholipase A 2 activating protein (PLAA), which are known to release eicosanoids 45 . AA and PLAA induced allodynia (∼4-6 h), without producing any measurable non-evoked nociceptive behavior (Fig.  2a–d ). AA and PLAA injection in B6 mouse hindpaw produced a slowly-developing increase in PGE 2 hindpaw levels that was abated by the COX inhibitor, indomethacin (Fig.  2e, f ). The PGE 2 peak level was ~50% of that produced by exogenous PGE 2 injection (Fig.  2e, f ). The differential response to exogenous vs . endogenous PGE 2 suggests that the non-evoked nociception by exogenous PGE 2 is a pharmacological response that is not replicated by endogenously released PGE 2 . SC EP2 silencing ( Plp-Ptger2 mice) (Fig.  2g, h ) or EP2 antagonism attenuated allodynia to AA or PLAA (Supplementary Fig.  7a, b ). Indomethacin treatment did not further enhance the attenuation produced by SC EP2 silencing (Fig.  2g, h ). All other interventions, including EP4 antagonism, SC and DRG neuron EP4 silencing and DRG EP2 silencing were ineffective (Supplementary Fig.  7c–j ), indicating a key role of the SC EP2 in sustaining allodynia by endogenous PGE 2 . Fig. 2 EP2 in Schwann cells mediates carrageenan and CFA hindpaw mechanical allodynia (allodynia) but not inflammation. a , c Dose-dependent non-evoked nociception and b , d dose- and time-dependent allodynia after intraplantar (i.pl.) phospholipase A 2 activating protein (PLAA), arachidonic acid (AA) or vehicle (Veh) in C57BL/6 J mice (B6) ( n  = 8 mice/group). e Time dependent PGE 2 assay in B6 paw tissue homogenates after i.pl. PGE 2 (1.5 nmol), PLAA (2 nmol), AA (10 nmol) or Veh ( n  = 6 mice/group) and f pretreated with indomethacin (Indo, 280 nmol) ( n  = 8 mice/group). g , h Allodynia after i.pl. PLAA, AA or Veh in P1p-Cre or Control mice infected with AAV for selective silencing of EP2 ( P1p-Ptger2 ) and pretreated with Indo or Veh ( n  = 8 mice/group). i , j Allodynia, paw thickness and myeloperoxydase (MPO) activity assay after i.pl. carrageenan (Cg, 300 μg), complete Freund’s adjuvant (CFA) or Veh in Plp-Ptger2 or Control mice and pretreated with Indo or Veh ( n  = 8 mice/group). Data are mean ± s.e.m. a , c , e , f and MPO activity in i , j , 1-way or b , d , g , h , allodynia and paw thickness in i , j 2-way ANOVA, Bonferroni correction. *** P  < 0.001, **** P  < 0.0001 vs. Veh or Control/Veh/Veh †† P  < 0.01, ††† P  < 0.001, †††† P  < 0.0001 vs. Control/Veh/PLAA, AA, Cg, CFA. Source data are provided as a Source Data file. a , c Dose-dependent non-evoked nociception and b , d dose- and time-dependent allodynia after intraplantar (i.pl.) phospholipase A 2 activating protein (PLAA), arachidonic acid (AA) or vehicle (Veh) in C57BL/6 J mice (B6) ( n  = 8 mice/group). e Time dependent PGE 2 assay in B6 paw tissue homogenates after i.pl. PGE 2 (1.5 nmol), PLAA (2 nmol), AA (10 nmol) or Veh ( n  = 6 mice/group) and f pretreated with indomethacin (Indo, 280 nmol) ( n  = 8 mice/group). g , h Allodynia after i.pl. PLAA, AA or Veh in P1p-Cre or Control mice infected with AAV for selective silencing of EP2 ( P1p-Ptger2 ) and pretreated with Indo or Veh ( n  = 8 mice/group). i , j Allodynia, paw thickness and myeloperoxydase (MPO) activity assay after i.pl. carrageenan (Cg, 300 μg), complete Freund’s adjuvant (CFA) or Veh in Plp-Ptger2 or Control mice and pretreated with Indo or Veh ( n  = 8 mice/group). Data are mean ± s.e.m. a , c , e , f and MPO activity in i , j , 1-way or b , d , g , h , allodynia and paw thickness in i , j 2-way ANOVA, Bonferroni correction. *** P  < 0.001, **** P  < 0.0001 vs. Veh or Control/Veh/Veh †† P  < 0.01, ††† P  < 0.001, †††† P  < 0.0001 vs. Control/Veh/PLAA, AA, Cg, CFA. Source data are provided as a Source Data file. Next, we investigated two preclinical models of inflammatory pain, carrageenan and CFA, that evoke PGE 2 -dependent mechanical allodynia 46 , 47 . In control mice, i.pl. carrageenan or CFA increased PGE 2 levels that were attenuated by indomethacin and did not produce detectable acute or delayed (assessed 3 h and 2 days after stimulus, respectively) non-evoked nociception (lifting and licking) (Supplementary Fig.  8a, b ). EP2 but not EP4 antagonism reduced the robust and sustained allodynia produced by carrageenan or CFA similarly in male and female mice (Supplementary Fig.  8c, d ). An EP1 antagonist failed to reduce mechanical allodynia by carrageenan and CFA (Supplementary Fig.  8e ). Allodynia evoked by both carrageenan and CFA was reduced in mice with EP2 silencing in SCs ( Plp-Ptger2 and Plp-Ptger2-bis mice) (Fig.  2i,j and Supplementary Fig.  8f ), whereas no attenuation was detected in any other silenced mouse strains (Supplementary Fig.  8g–m ). Importantly, allodynia was similarly inhibited in mice treated with indomethacin and in Plp-Ptger2 mice, and indomethacin did not further increase the attenuation produced by SC EP2 silencing (Fig.  2i, j ). However, whereas indomethacin inhibited carrageenan- and CFA-induced increases in a series of inflammatory responses, including paw edema and leukocyte myeloperoxidase (MPO), these responses were unchanged in Plp-Ptger2 mice (Fig.  2i, j ) or in any other of the DRG neuron or SC silenced mouse strains (Supplementary Fig.  8h–m ). Pharmacological studies corroborated these findings as EP2 (PF and TG6), but not EP4 (BGC, ONOAE and CJ-42) antagonism, reduced carrageenan- or CFA-induced allodynia (Supplementary Fig.  9a–j ). Neither EP2 nor EP4 antagonism reduced carrageenan- and CFA evoked paw edema (Supplementary Fig.  9a–j ) and increase in additional inflammatory responses, including increases in paw tissue homogenates of interleukin 1-β (IL-1β) and tumor necrosis factor-α (TNFα) (Supplementary Fig.  9k, l ). Carrageenan or CFA also increased a non-evoked pain-like response (grimace behavior) that was similarly attenuated by indomethacin, EP2 antagonism (PF and TG6), and in EP2 Plp-Ptger2 and Plp-Ptger2-bis mice, but not in Plp-scr-shRNA mice or by EP4 antagonism (BGC, ONOAE and CJ-42) (Supplementary Fig.  9m ). Intraplantar PGE 2 (1.5 nmol/paw) failed to produce grimace behavior (Supplementary Fig.  9n ). These results in prototypical inflammatory pain models highlight the unexpected yet critical role of SC EP2 in PGE 2 -mediated non-evoked and evoked pain-like responses, which is totally independent from the PG-mediated inflammatory response. cAMP is a primary intracellular mediator of PGE 2 -induced pain 11 . EP2 and EP4 couple to Gαs, which activates AC 13 . cAMP and downstream cAMP-dependent PKA constitute a temporally and spatially organized nanodomain that controls different compartmentalized signaling pathways 48 . We recently reported that CGRP-stimulated periorbital mechanical allodynia is mediated by a sustained cAMP increase in SCs, which requires clathrin- and dynamin-dependent internalization and endosomal signaling of the CGRP-CLR/RAMP1 complex 21 . In contrast, we report here that clathrin and dynamin inhibitors (Pitstop2 and Dyngo-4a, respectively) failed to reduce PGE 2 , carrageenan- and CFA-evoked allodynia (Supplementary Fig.  10 ). Thus, EP2 and cAMP signaling at endosomes of SCs is unlikely to mediate allodynia produced by inflammation. To probe the contributions of compartmentalized cAMP signaling in SCs to EP2-dependent allodynia, we used blue light Beggiatoa photo-activable AC (bPAC) 49 , 50 , which enables optogenetic stimulation of focal cAMP production. Human SCs (hSCs) were co-transfected with bPAC and a genetically-encoded red fluorescent global cAMP sensor, cADDIs, to monitor cAMP production. Blue light (pulsed 1 s/5 s for 10 min) stimulated a forskolin-like increase in cADDIs fluorescence (ΔF/F0) peaking at 60 s and returning to baseline within 10 min (Fig.  3a ), as previously reported 51 . To study this SC signaling pathway in vivo, we used a Plp CreERT driver that functions as a lineage tracer to express bPAC-T2A-mCherry in SCs. To induce the bPAC-selective expression in SCs, a Cre-dependent virus packaged with the AAV2/8 serotype was used for efficient infection of SCs. A loxP flanked bPAC-T2A-mCherry construct was used to exclusively express bPAC and mCherry in SCs (Fig.  3b ). Intraplantar virus injection in Plp CreERT mice labeled cutaneous SCs (Fig.  3c ). Blue-light evoked cAMP-dependent allodynia in Plp CreERT mice but not in control mice (Fig.  3d ). These data confirm the proalgesic role of cutaneous SCs 23 which contribute to mechanical allodynia. Fig. 3 Schwann cells cAMP nanodomains promote hindpaw mechanical allodynia (allodynia) by PGE 2 . a Global cAMP formation in human Schwann cells (hSCs), after blue light stimulation (450 nm, 10 min) ( n  = 121 cells, n  = 3 independent experiments) or forskolin (FSK 100 μM) ( n  = 78 cells, n  = 3 independent experiments). b Schematic representation of Cre recombinase dependent expression of Beggiatoa photo-activable AC (bPAC) and mCherry separated by T2A self-cleaving peptide sequence. c Representative images of mCherry expression in S100B+ cells in mouse paw tissue after intraplantar (i.pl.) infection with AAV for selective expression of bPAC in SCs (scale bar: 20 μm) ( n  = 4 subjects). d Illustration of bPAC activation after blue light stimulation and allodynia induced by blue light stimulation (pulsed 1 s/5 s for 10 min) in Plp-Cre or Control mice infected with AAV- bPAC ( Plp-bPAC ) ( n  = 8 mice/group). e Representative images and cumulative data of membrane and intracellular localization of EP2 and EP4 in hSCs cells, with or without pre-incubation with unlabeled PGE 2 (10 µM, 30 min). Scale bar, 20 µm. ( n  = 4 independent experiments). f Concentration-dependent global cAMP formation in hSCs induced by PGE 2 ( n  = 8 replicates). g Membrane confined cAMP formation induced by PGE 2 (10 nM) in hSCs in the presence of PF-04418948 (PF,1 μM), BGC 20-1531 (BGC, 1 μM) or vehicle (Veh) (cells number: PGE 2  = 80, PF = 90, BGC = 88, n  = 3 independent experiments). h Illustration of membrane tagged Lyn11-bPAC membrane confined cAMP formation and cAMP formation in hSCs after blue light stimulation ( n  = 47 cells, n  = 3 independent experiments) or FSK (100 μM) ( n  = 46 cells, n  = 3 independent experiments). i Allodynia induced by blue light stimulation in Plp-Cre or Control mice infected with AAV- Lyn11-bPAC ( Plp-Lyn11-bPAC ) ( n  = 8 mice/group). j Concentration-dependent PKA activation induced by PGE 2 in hSCs ( n  = 8 replicates). k Membrane confined PKA activation in hSCs induced by PGE 2 (10 nM) in the presence of PF (1 μM), BGC (1 μM) or Veh (cells number: PF = 64, BGC = 63, PGE 2  = 64, n  = 3 independent experiments). Data are mean ±  s.e.m. a , e , h ,2-tailed Student’s t test, g , k 1-way or d , i 2-way ANOVA, Bonferroni correction. AUC area under curve. **** P  < 0.0001 vs. Control/blue light. Source data are provided as a Source Data file. a Global cAMP formation in human Schwann cells (hSCs), after blue light stimulation (450 nm, 10 min) ( n  = 121 cells, n  = 3 independent experiments) or forskolin (FSK 100 μM) ( n  = 78 cells, n  = 3 independent experiments). b Schematic representation of Cre recombinase dependent expression of Beggiatoa photo-activable AC (bPAC) and mCherry separated by T2A self-cleaving peptide sequence. c Representative images of mCherry expression in S100B+ cells in mouse paw tissue after intraplantar (i.pl.) infection with AAV for selective expression of bPAC in SCs (scale bar: 20 μm) ( n  = 4 subjects). d Illustration of bPAC activation after blue light stimulation and allodynia induced by blue light stimulation (pulsed 1 s/5 s for 10 min) in Plp-Cre or Control mice infected with AAV- bPAC ( Plp-bPAC ) ( n  = 8 mice/group). e Representative images and cumulative data of membrane and intracellular localization of EP2 and EP4 in hSCs cells, with or without pre-incubation with unlabeled PGE 2 (10 µM, 30 min). Scale bar, 20 µm. ( n  = 4 independent experiments). f Concentration-dependent global cAMP formation in hSCs induced by PGE 2 ( n  = 8 replicates). g Membrane confined cAMP formation induced by PGE 2 (10 nM) in hSCs in the presence of PF-04418948 (PF,1 μM), BGC 20-1531 (BGC, 1 μM) or vehicle (Veh) (cells number: PGE 2  = 80, PF = 90, BGC = 88, n  = 3 independent experiments). h Illustration of membrane tagged Lyn11-bPAC membrane confined cAMP formation and cAMP formation in hSCs after blue light stimulation ( n  = 47 cells, n  = 3 independent experiments) or FSK (100 μM) ( n  = 46 cells, n  = 3 independent experiments). i Allodynia induced by blue light stimulation in Plp-Cre or Control mice infected with AAV- Lyn11-bPAC ( Plp-Lyn11-bPAC ) ( n  = 8 mice/group). j Concentration-dependent PKA activation induced by PGE 2 in hSCs ( n  = 8 replicates). k Membrane confined PKA activation in hSCs induced by PGE 2 (10 nM) in the presence of PF (1 μM), BGC (1 μM) or Veh (cells number: PF = 64, BGC = 63, PGE 2  = 64, n  = 3 independent experiments). Data are mean ±  s.e.m. a , e , h ,2-tailed Student’s t test, g , k 1-way or d , i 2-way ANOVA, Bonferroni correction. AUC area under curve. **** P  < 0.0001 vs. Control/blue light. Source data are provided as a Source Data file. Observations in transfected HeLa cells show that PGE 2 induces EP4 but not EP2 internalization 52 , suggesting that EP2-induced cAMP signaling in SCs is confined to the plasma membrane. We confirmed these findings in HEK293T cells and hSCs expressing EP2 or EP4 C-terminally fused to Venus. PGE 2 (10 µM) stimulated robust endocytosis of EP4 after 30 min, whereas EP2 remained at the plasma membrane (Fig.  3e and Supplementary Fig.  11a ). To identify the cAMP subcellular compartment implicated in allodynia, we studied total and nanodomain-specific cAMP signaling in SCs transfected with global cellular or plasma membrane-targeted cADDIs sensor. PGE 2 concentration- and time-dependently increased global cADDIS fluorescence in hSCs that was attenuated by selective EP2 (PF) and EP4 (BGC) antagonism (Fig.  3f and Supplementary Fig.  11b, c ). Increases in plasma membrane-associated cAMP evoked by PGE 2 in hSCs, and in mouse SCs from sciatic nerve trunks or cutaneous nerve fibers, were attenuated by EP2 but not EP4 antagonism (Fig.  3g and Supplementary Fig.  11d, e ). Peri-sciatic PGE 2 failed to cause non-evoked nociception but elicited sustained allodynia measured in the hind paw, ipsilateral to the injection, which was attenuated by EP2 antagonism and in EP2 Plp-Ptger2 and Plp-Ptger2-bis mice, but not by EP4 antagonism or in Plp-scr-shRNA mice (Supplementary Fig.  11f–k ). Thus, in vivo experiments support the hypothesis that all SCs, both cutaneous and those wrapping more proximal sciatic nerve trunks, sustain allodynia. The selective EP2 (butaprost) and EP4 (L-902) agonists increased global cAMP but only butaprost stimulated plasma membrane-associated cAMP (Supplementary Fig.  11l–n ). To determine the contribution of plasma membrane cAMP signaling to allodynia, we fused bPAC with the membrane targeting sequence, Lyn11 (Lyn11-bPAC). Plasma membrane localization of bPAC was confirmed in HEK293 cells (Supplementary Fig.  11o ). hSCs were cotransfected with Lyn11-bPAC and the plasma membrane-targeted cADDIS sensor (Fig.  3h ). Blue light increased plasma membrane cADDIs fluorescence similarly to forskolin (100 μM), which was unaffected by EP2 or EP4 antagonism (Supplementary Fig.  11p ). Paw exposure to blue light in Plp CreERT mice infected with i.pl. AAV-Lyn11-bPAC ( Plp-Lyn11-bPAC mice) time-dependently enhanced allodynia (Fig.  3i ). Thus, a plasma membrane-associated cAMP nanodomain mediates allodynia. cAMP and cAMP-dependent PKA constitute a spatiotemporally organized circuit that is strictly compartmentalized to form signaling nanodomains that control distinct biochemical and functional pathways 48 . The role of PKA in PGE 2 and inflammatory mechanical hypersensitivity has been previously documented 11 . However, to the best of our knowledge, increases in membrane-associated cAMP and the ensuing intracellular pathway have not been previously associated with pain signaling. To determine if membrane-associated cAMP activates discrete compartmentalized complexes through spatially sequestered PKA, hSCs were transfected with global (AKAR4) or plasma membrane-targeted (Lyn-AKAR4) fluorescent resonance energy transfer (FRET)-based PKA activity biosensors. PGE 2 , butaprost (EP2 agonist) and L-902 (EP4 agonist) induced a concentration- and time-dependent increase in AKAR4 FRET signals (Fig.  3j , Supplementary Fig.  12a, b ). Global PKA activation by PGE 2 was similarly prevented by both EP2 (PF) and EP4 (BGC) antagonists (Supplementary Fig.  12c ). However, PF but not BGC inhibited PGE 2 -stimulated plasma membrane-associated PKA activity detected with the Lyn-AKAR4 biosensor (Fig.  3k ). Selective EP2 and EP4 activation confirmed these results; butaprost but not L-902 stimulated plasma membrane PKA activity detected with Lyn-AKAR4 (Supplementary Fig.  12d ). To examine whether the SC plasma membrane-localized cAMP/PKA activity mediates PGE 2 -induced allodynia, we used a light-activated phosphodiesterase (LAPD) 53 , 54 fused with the plasma membrane targeted sequence Lyn11 (Lyn11-LAPD). hSCs were co-transfected with Lyn11-LAPD and the membrane targeted cAMP FRET sensor Lyn11-R-FlincA. To confirm the plasma membrane localization of LAPD and R-FlincA HEK293 cells were used (Supplementary Fig.  12e, f ). PGE 2 increased the FRET signal that was prevented by the red-light activation of the Lyn11-LAPD (Fig.  4a ). These findings were recapitulated in vivo as paw stimulation of Plp CreERT mice, infected with i.pl. AAV(Lyn11-LAPD) ( Plp-Lyn11-LAPD mice), with red-light prevented allodynia evoked by PGE 2 (Supplementary Fig.  12g ). Exposure to red-light of Plp-Lyn-LAPD mice also reduced allodynia evoked by AA, PLAA, carrageenan and CFA (Fig.  4b, c , Supplementary Fig.  12h, i ). Red light did not produce any effect in control mice (Fig.  4b, c, m, n , Supplementary Fig.  12g, h, i ). These results confirm the essential role of plasma membrane-associated cAMP/PKA nanodomains in SCs in mediating mechanical allodynia induced by PGE 2 and inflammation. Fig. 4 Schwann cells membrane-bound PKA/AKAP drives hindpaw mechanical allodynia (allodynia) by PGE 2 and inflammation. a Membrane cAMP in hSCs after red light (650 nm/10 min) activation of membrane-tagged light-activated phosphodiesterase (Lyn11-LAPD) and PGE 2 (10 nM) (cells: light = 61, no light = 99). b , c Allodynia by carrageenan (Cg, 300 μg) or CFA in Plp-Cre or Control mice after AAV- Lyn11-LAPD ( Plp-Lyn11-LAPD , i.pl.) and red light stimulation ( n  = 8 mice/group). d A-kinase anchor protein 79 (AKAP79) associated cAMP formation and e PKA activation by PGE 2 (10 nM) in the presence of PF-04418948 (PF, 1 μM), BGC 20-1531 (BGC, 1 μM) or vehicle (Veh) in hSCs ( d , cells: PGE 2  = 62, PF = 98, BGC = 57, e , PF = 95, BGC = 47, PGE 2  = 52, n  = 3 independent experiments). f AKAP79 associated cAMP formation in hSCs after blue light stimulation ( n  = 43 cells, n  = 3 independent experiments) or forskolin (FSK, 100 μM) ( n  = 46 cells, n  = 3 independent experiments). Allodynia by ( g ) blue light stimulation (1 s/5 s) in Plp-Cre or Control mice after AAV- AKAP150-bPAC ( Plp-AKAP150-bPAC , i.pl.) ( n  = 8 mice/group); h , i ,Cg, CFA or Veh in C57BL/6 J mice pretreated with st-Ht31 (17 nmol) or Veh ( n  = 8 mice/group). j AKAP79 associated cAMP formation in hSCs after red light activation of AKAP79-LAPD and PGE 2 (10 nM) (cells: light = 102, No light = 100, n  = 3 independent experiments). Allodynia induced by Cg, CFA or Veh in ( k , l ) Plp-Cre or Control mice infected with AAV for AKAP150 silencing ( Plp-AKAP150 ) ( n  = 8 mice/group); m , n , Plp-Cre or Control mice after AAV- AKAP150-LAPD ( Plp-AKAP150-LAPD , i.pl.) and stimulated with red light ( n  = 8 mice/group); o , p , Plp-Trpa1 or Control mice ( n  = 8 mice/group). q Illustration of BRET assay. r Catalytic subunit of PKA (CA) and TRPA1 interaction after PGE 2 (100 nM) in the presence of BGC (100 nM) PF(100 nM), st-Ht31 (10 μM) or CGRP (10 μM) (replicates: PGE 2  = 8, BGC = 8, PF = 8, st-Ht31 = 8, CGRP = 8). Data are mean ± s.e.m. a , f , j 2-tailed Student’s t test d , e , r , 1-way or b , c , g , h , i , k–p , 2-way ANOVA, Bonferroni correction. AUC area under the curve. ** P  < 0.01, *** P  < 0.001, **** P  < 0.0001 vs. Veh, Control/Veh, blue light, red light † P  < 0.05, †† P  < 0.01, ††† P  < 0.001, †††† P  < 0.0001 vs. Cg, CFA, PGE 2 /Veh, Control/Cg, CFA, Plp-AKAP150-LAPD/no light/Cg, CFA. Source data are provided as a Source Data file. a Membrane cAMP in hSCs after red light (650 nm/10 min) activation of membrane-tagged light-activated phosphodiesterase (Lyn11-LAPD) and PGE 2 (10 nM) (cells: light = 61, no light = 99). b , c Allodynia by carrageenan (Cg, 300 μg) or CFA in Plp-Cre or Control mice after AAV- Lyn11-LAPD ( Plp-Lyn11-LAPD , i.pl.) and red light stimulation ( n  = 8 mice/group). d A-kinase anchor protein 79 (AKAP79) associated cAMP formation and e PKA activation by PGE 2 (10 nM) in the presence of PF-04418948 (PF, 1 μM), BGC 20-1531 (BGC, 1 μM) or vehicle (Veh) in hSCs ( d , cells: PGE 2  = 62, PF = 98, BGC = 57, e , PF = 95, BGC = 47, PGE 2  = 52, n  = 3 independent experiments). f AKAP79 associated cAMP formation in hSCs after blue light stimulation ( n  = 43 cells, n  = 3 independent experiments) or forskolin (FSK, 100 μM) ( n  = 46 cells, n  = 3 independent experiments). Allodynia by ( g ) blue light stimulation (1 s/5 s) in Plp-Cre or Control mice after AAV- AKAP150-bPAC ( Plp-AKAP150-bPAC , i.pl.) ( n  = 8 mice/group); h , i ,Cg, CFA or Veh in C57BL/6 J mice pretreated with st-Ht31 (17 nmol) or Veh ( n  = 8 mice/group). j AKAP79 associated cAMP formation in hSCs after red light activation of AKAP79-LAPD and PGE 2 (10 nM) (cells: light = 102, No light = 100, n  = 3 independent experiments). Allodynia induced by Cg, CFA or Veh in ( k , l ) Plp-Cre or Control mice infected with AAV for AKAP150 silencing ( Plp-AKAP150 ) ( n  = 8 mice/group); m , n , Plp-Cre or Control mice after AAV- AKAP150-LAPD ( Plp-AKAP150-LAPD , i.pl.) and stimulated with red light ( n  = 8 mice/group); o , p , Plp-Trpa1 or Control mice ( n  = 8 mice/group). q Illustration of BRET assay. r Catalytic subunit of PKA (CA) and TRPA1 interaction after PGE 2 (100 nM) in the presence of BGC (100 nM) PF(100 nM), st-Ht31 (10 μM) or CGRP (10 μM) (replicates: PGE 2  = 8, BGC = 8, PF = 8, st-Ht31 = 8, CGRP = 8). Data are mean ± s.e.m. a , f , j 2-tailed Student’s t test d , e , r , 1-way or b , c , g , h , i , k–p , 2-way ANOVA, Bonferroni correction. AUC area under the curve. ** P  < 0.01, *** P  < 0.001, **** P  < 0.0001 vs. Veh, Control/Veh, blue light, red light † P  < 0.05, †† P  < 0.01, ††† P  < 0.001, †††† P  < 0.0001 vs. Cg, CFA, PGE 2 /Veh, Control/Cg, CFA, Plp-AKAP150-LAPD/no light/Cg, CFA. Source data are provided as a Source Data file. In excitable cells, the plasma membrane-localized scaffold protein, AKAP79 (or AKAP5), the human ortholog of the rodent AKAP150, tightly controls the phosphorylation of pain signaling proteins 12 , 55 . Due to its multivalent nature, we hypothesized that AKAP79/150 scaffolds cAMP/PKA at the plasma membrane of SCs and thereby spatiotemporally regulates PGE 2 -evoked allodynia. To determine whether AKAP79/150 recruits signaling effectors to distinct compartments of SCs that fine-tune PGE 2 responses, we fused the cAMP FRET sensor R-FlincA or the PKA FRET sensor AKAR4 to the full-length AKAP79 scaffold (gene AKAP5 ). PGE 2 induced a time-dependent increase in AKAP79-R-FlincA fluorescence in hSCs that was prevented by EP2 (PF) but not EP4 (BGC) antagonism (Fig.  4d ). Similarly, PGE 2 stimulated AKAR4-AKAP79 fluorescence that was attenuated by EP2 but not EP4 antagonism (Fig.  4e ). The EP2 agonist, butaprost, but not the EP4 agonist, L-902, increased the AKAP79-R-FlincA fluorescence and AKAR4-AKAP79 FRET signals in hSCs, further supporting the critical contribution of the AKAP79 compartment in SC EP2 signaling (Supplementary Fig.  13a, b ). To determine the contribution of the SC AKAP79/150-associated cAMP increase to allodynia, we fused bPAC to full-length AKAP79 (AKAP79-bPAC). In hSCs coexpressing AKAP79-bPAC and the AKAP79-R-FlincA FRET sensor, blue light stimulation increased FRET signal (Fig.  4f ). To translate these findings in vivo, blue light was applied to the paw of Plp CreERT mice infected with i.pl. AAV AKAP150-bPAC ( Plp-AKAP150-bPAC mice), leading to time-dependent allodynia (Fig.  4g ). To determine the contribution of AKAP79/150 scaffold to inflammatory pain in vivo, we used the peptide inhibitor of the AKAP/PKA interaction, st-Ht31 56 . st-Ht31 pretreatment inhibited carrageenan- or CFA-evoked allodynia in B6 mice (Fig.  4h, i ). In hSCs coexpressing the cAMP FRET sensor AKAP79-R-FlincA and AKAP79-LAPD, red-light activation of the AKAP79-LAPD prevented PGE 2 -stimulated cAMP production at the AKAP79 compartment (Fig.  4j ). Selective silencing of SC AKAP150 in Plp CreERT+ mice infected, via intrasciatic administration, with an AAV-shRNA selective for AKAP150 ( Plp-AKAP150 mice), reduced allodynia evoked by carrageenan and CFA (Fig.  4k, l ). Furthermore, paw stimulation of Plp-AKAP150-LAPD mice with red-light attenuated allodynia evoked by carrageenan or CFA (Fig.  4m, n ), underlining a mechanistic role of the SC PKA/AKAP interaction in inflammatory mechanical allodynia. The transient receptor potential ankyrin 1 (TRPA1) cation channel has been implicated in rodent pain models 57 . A specific role of Schwann cell TRPA1 has been reported in mouse models of neuropathic, cancer and migraine pain, elicited through diverse intracellular pathways 18 , 19 , 21 , 58 . CGRP signals from endosomes of SCs to activate cAMP and PKA, resulting in endothelial nitric oxide synthase (eNOS) phosphorylation and nitric oxide (NO) release 21 . NO was found to activate TRPA1, leading to a NADPH oxidase-1 (NOX1)-dependent formation of reactive oxygen species (ROS) that target TRPA1 on adjacent nociceptors to signal mechanical allodynia 21 . We investigated whether plasma membrane-associated cAMP/PKA nanodomains in SCs similarly promote mechanical hypersensitivity. NOS inhibition attenuates PGE 2 -evoked mechanical hyperalgesia in rats 59 . We confirmed that the unselective NOS inhibitor, Nω-nitro-L-arginine methyl ester hydrochloride (L-NAME), partially reduced PGE 2 -induced allodynia in mice (Supplementary Fig.  14a ). The CGRP receptor antagonist, olcegepant, and the combination of L-NAME and olcegepant produced a similar inhibition of PGE 2 -induced allodynia (Supplementary Fig.  14a ), suggesting that a component of allodynia evoked by PGE 2 is due to CGRP release from peptidergic nerve terminals and the ensuing NO release. In contrast to the data obtained with PGE 2 , L-NAME or olcegepant failed to reduce AA-, PLAA-, carrageenan- or CFA-evoked allodynia (Supplementary Fig.  14b–i ). Thus, endogenous PGE 2 released during inflammation is unable to target peptidergic nerve terminals and release CGRP. However, selective TRPA1 silencing in SCs ( Plp - Trpa1 mice) markedly inhibited PGE 2 -evoked allodynia (Fig.  4o, p and Supplementary Fig.  14j ). Despite the finding that endogenous PGE 2 and CGRP activate signaling pathways that converge on TRPA1 as a common final pathway to signal allodynia, the following results indicate that the underlying mechanism for channel activation elicited by these mediators is different. PKA anchoring to AKAP79/150 regulates hypersensitivity to inflammatory mediators, including sensitization of the TRP vanilloid 1 (TRPV1) channel by PGE 2 12 . We hypothesized that a direct PKA/TRPA1 phosphoactivation in SCs elicits the L-NAME-independent allodynia in response to EP2 stimulation. A time-resolved NanoLuc complementation reporter approach was used to measure the dynamics of PKA/TRPA1 interaction in hSCs. Large (LgBiT) and small (SmBiT) subunits of the Nanoluc luciferase were fused to the C- and N-terminal extremities of TRPA1 and PKA catalytic subunit, respectively (SmBiT-TRPA1, PKA-LgBiT) (Fig.  4q ). PGE 2 stimulated an increased luminescence in HEK293T cells coexpressing SmBiT-TRPA1 and PKA-LgBiT with EP2 but not with EP4 (Supplementary Fig.  14k ). In hSCs, PGE 2 , but not CGRP, increased luminescence. Antagonism of EP2 (PF) but not the EP4 (BGC) blocked this response (Fig.  4r ). Disruption of AKAP79/PKA interaction with the inhibitory peptide, st-Ht31, also attenuated PGE 2 -stimulated luminescence (Fig.  4r ). Thus, a first component of PGE 2 -induced allodynia results from a direct PKA/TRPA1 interaction while a second component depends on CGRP release and the ensuing NOS-induced NO release. In agreement with these data, allodynia by the EP2 agonist, butaprost, was completely attenuated by st-Ht31 and in Plp - Trpa1 mice but was unchanged in mice treated with the CGRP receptor antagonist, olcegepant (Supplementary Fig.  14l–n ). We confirmed the previous observation 11 that pre-treatment, but not post-treatment (30 min after stimulus) with the AC inhibitor, SQ22536, reduced PGE 2 -evoked allodynia (Supplementary Fig.  15a, b ), indicating that membrane-associated cAMP increase initiates but does not sustain mechanical hypersensitivity, which is maintained by PKA activation. Several findings suggest that PKA/TRPA1 phosphoactivation in SCs by EP2/cAMP sustains calcium-dependent oxidative stress, which eventually targets neuronal TRPA1 to signal pain. In cultured human SCs, PGE 2 elicited a calcium-response and H 2 O 2 production, which were attenuated by PF (EP2 antagonist), H-89 (PKA inhibitor), ML171 (NOX-1 selective inhibitor), and A967079 (TRPA1 antagonist) (Supplementary Fig.  15c, d ). In mice, PGE 2 -evoked mechanical allodynia was also attenuated by H-89, ML171, and A967079 (Supplementary Fig.  15e–g ), confirming previous evidence 18 , 21 that SCs express receptors, enzymes and channels that amplify the oxidative stress signal to sustain allodynia. Furthermore, the spin-trap agent, N-tert-butyl-alpha-phenylnitrone (PBN), which captures free-radicals, reduced allodynia to PGE 2 , carrageenan, or CFA, but did not affect transient nociception (licking and lifting) to PGE 2 (Supplementary Fig.  15h–k ). Selective silencing of TRPA1 in DRG neurons ( Adv-Trpa1 mice) did not reduce the transient nociception to PGE 2 , confirming previous data observed in global TRPA1 knock-out mice 29 but attenuated allodynia (Supplementary Fig.  15l, m ). Allodynia produced by carrageenan or CFA was also attenuated in Adv-Trpa1 (Supplementary Fig.  15n, o ). These findings suggest that under inflammatory circumstances, intracellular SC pathways, via oxidative stress generation, sensitize TRPA1 in DRG neurons to innocuous mechanical stimuli (Supplementary Fig.  15p ). The major role of the EP1/TRPV1 signaling pathway in PGE 2 -induced heat hypersensitivity was identified with global knock-out mice for either the receptor or channel 60 . We confirmed these findings by using the EP1 antagonist, SC-51322, and global TRPV1 deleted mice: these two interventions failed to affect mechanical hypersensitivity, but abrogated heat hypersensitivity to PGE 2 (Supplementary Fig  1q and Supplementary Fig  16a, c ). In contrast, the EP2 antagonist, PF, that attenuated allodynia, and EP3 and EP4 antagonism failed to reduce heat hypersensitivity (Supplementary Fig.  16d–f ). These data underline the mechanistically different responses to PGE 2 with respect to heat (EP1/TRPV1) vs . mechanical (EP2/TRPA1) hypersensitivity and represent an independent control of the specificity of the EP2/TRPA1 pathway to sustain mechanical hypersensitivity. Although NSAIDs transiently inhibit inflammatory pain in patients, they markedly delay its resolution in the CFA mouse model of inflammatory pain 6 . This delayed resolution of pain, observed following treatment with the commonly used non-selective COX inhibitor diclofenac 61 or the corticosteroid, dexamethasone 6 , is attributed to disruption of the protective role of PG-mediated inflammation that promotes recovery from the painful condition 6 . We also observed that daily treatment (intraperitoneal, for 6 days after CFA) with diclofenac, attenuated CFA-evoked allodynia as well as the increases in paw edema and MPO levels (Fig.  5a–c ). After cessation of diclofenac treatment (at day 7 after CFA), allodynia persisted, and recovery was not attained within the following 25 days (Fig.  5a ). Although an identical treatment schedule with the EP2 antagonist, PF, also initially inhibited allodynia (like diclofenac), PF neither decreased inflammation (paw edema, MPO levels) nor delayed allodynia recovery (Fig.  5a–c ). Furthermore, selective silencing of SC EP2 ( Plp-Ptger2 ) in mice produced an early and sustained attenuation of CFA-evoked allodynia without reducing inflammation (Fig.  5d–f ). The observation that, like in Plp-Ptger2 mice, in SC AKAP150 silenced ( Plp-AKAP150 ) mice CFA-induced allodynia was constantly attenuated without any reduction in inflammation (Fig.  5g–i ) supports the unique role of the EP2-dependent intracellular pathway in SCs in the control of inflammatory allodynia. Fig. 5 EP2 in Schwann cells blockade abates hindpaw mechanical allodynia (allodynia) but does not delay pain resolution. a–c Allodynia, paw thickness and myeloperoxidase (MPO) activity assay after complete Freund’s adjuvant (CFA) or vehicle (Veh) in C57BL/6 J mice treated for 6 consecutive days with diclofenac (25 mg/kg, i.p.), PF-04448948 (PF, 10 mg/kg, i.g.) or Veh ( n  = 8 mice/group). d–i allodynia, paw thickness and MPO activity assay after CFA or Veh in P1p-Cre or Control mice infected with AAV for selective silencing of d–f , EP2 (-Ptger2 ) ( Plp-Ptger2 ) or g–i , AKAP150 ( Plp-Akap150 ) ( n  = 8 mice/group). Data are mean ± s.e.m. a , b , d , e , g , h , 2-way or c , f , i , 1-way ANOVA, Bonferroni correction. * P  < 0.05, *** P  < 0.001, **** P  < 0.0001 vs. Veh, Control/Veh † P  < 0.05, †† P  < 0.01, †††† P  < 0.0001 vs . CFA/Veh, Control/CFA, §§ P  < 0.01, §§§§ P  < 0.0001 vs. CFA/diclofenac. Source data are provided as a Source Data file. a–c Allodynia, paw thickness and myeloperoxidase (MPO) activity assay after complete Freund’s adjuvant (CFA) or vehicle (Veh) in C57BL/6 J mice treated for 6 consecutive days with diclofenac (25 mg/kg, i.p.), PF-04448948 (PF, 10 mg/kg, i.g.) or Veh ( n  = 8 mice/group). d–i allodynia, paw thickness and MPO activity assay after CFA or Veh in P1p-Cre or Control mice infected with AAV for selective silencing of d–f , EP2 (-Ptger2 ) ( Plp-Ptger2 ) or g–i , AKAP150 ( Plp-Akap150 ) ( n  = 8 mice/group). Data are mean ± s.e.m. a , b , d , e , g , h , 2-way or c , f , i , 1-way ANOVA, Bonferroni correction. * P  < 0.05, *** P  < 0.001, **** P  < 0.0001 vs. Veh, Control/Veh † P  < 0.05, †† P  < 0.01, †††† P  < 0.0001 vs . CFA/Veh, Control/CFA, §§ P  < 0.01, §§§§ P  < 0.0001 vs. CFA/diclofenac. Source data are provided as a Source Data file.

Since discovery of the mechanism of action of aspirin 62 , the analgesic effect of NSAIDs has been intimately associated with their prominent anti-inflammatory activity. Herein, we challenge the dogma that for relief of PGE 2 -mediated inflammatory pain it is necessary to block PG-induced inflammation. In fact, we provide pharmacological and genetic evidence that PGs released during inflammation act locally to stimulate two mechanistically independent pathways, one of which mediated by PGE 2 and SC EP2 produces pain-like behaviors without any implication in the inflammatory response. The observations that SCs taken from cutaneous nerve fibers or sciatic nerve trunks responded similarly to EP2 agonism, antagonism, and silencing, and that hindpaw allodynia produced by either intraplantar or peri-sciatic PGE 2 injection was mechanistically similar, suggest that the EP2 proalgesic role is not limited to SCs localized to cutaneous terminal nerve fibers. The SC proalgesic pathway initiated by EP2 results in TRPA1-dependent ROS release that targets neuronal TRPA1, leading to sustained hypersensitivity to mechanical stimuli (Fig.  6 ). Fig. 6 Hypothesized mechanism of action of prostaglandins. Prostaglandins (PGs) released during inflammation promote inflammation and pain by distinct mechanisms. a PGs cause inflammation by an EP2-independent inflammatory process, which is characterized by edema, leukocyte infiltration and release of MPO. b PGs cause pain by activating EP2 on SCs. EP2 generates cAMP, leading to AKAP79-associated activation of PKA in plasma membrane-delimited nanodomains. PKA phosphorylates and activates TRPA1 in SCs, which elicits a calcium-dependent release of reactive oxygen species (H 2 O 2 ). H 2 O 2 targets TRPA1 on adjacent nociceptors, which results in sustained hypersensitivity to mechanical stimuli. Prostaglandins (PGs) released during inflammation promote inflammation and pain by distinct mechanisms. a PGs cause inflammation by an EP2-independent inflammatory process, which is characterized by edema, leukocyte infiltration and release of MPO. b PGs cause pain by activating EP2 on SCs. EP2 generates cAMP, leading to AKAP79-associated activation of PKA in plasma membrane-delimited nanodomains. PKA phosphorylates and activates TRPA1 in SCs, which elicits a calcium-dependent release of reactive oxygen species (H 2 O 2 ). H 2 O 2 targets TRPA1 on adjacent nociceptors, which results in sustained hypersensitivity to mechanical stimuli. Our results obtained with a series of different selective antagonists, including three EP4 and two EP2 antagonists, exclude the involvement of neuronal EP4 in mechanical hypersensitivity and grimace behavior at the site of inflammation. These findings align with our results obtained through AAV silencing strategies. Together, these observations support the proalgesic function of Schwann cell EP2 and provide evidence that the proalgesic actions of EP2 are not limited to the central nervous system 16 . Recently, EP4 has been implicated in mechanical allodynia induced by carrageenan and CFA injected in the mouse hind paw 63 . However, the inhibitory effects of EP4 antagonism were modest compared to markedly higher inhibition by cyclooxygenases inhibitors (see the present study and 46 ). Therefore, the receptors, responsible for the major component of the prostaglandin-mediated response, remain unidentified. One possible explanation for the differing results may relate to the routes of antagonist administration and their respective sites of action: local (intraplantar) in our study vs . systemic (oral or intraperitoneal) in the previous study 63 . The small effect of systemic EP4 antagonism 63 on mechanical thresholds and paw thickness may be mediated through systemic mechanisms, that are not engaged by local interventions in the inflamed paw. Our findings do not support the currently accepted view that the PGE 2 injection in peripheral tissues recapitulates the painful response elicited by endogenous PGE 2 , since we found that the actions of exogenous and endogenous PGs differ significantly. On one hand, we confirm that PGE 2 injection elicits both a robust and transient non-evoked nociception and a moderate delayed allodynia by directly targeting EP4 in DRG nociceptive nerve terminals. On the other hand, we reveal that SC EP2 sustains both evoked (mechanical allodynia) and non-evoked (grimace behavior) pain-like responses elicited by endogenous PGE 2 , released under inflammatory circumstances. It is possible that the higher PGE 2 tissue level achieved locally after the injection of the exogenous compound may more likely target the dense network of sensory nerve terminals in the mouse hind paw. The observation 64 that subcutaneous injection of a small dose of PGE 2 in human skin elicits long-lasting hyperalgesia but not acute spontaneous pain is consistent with the present findings and may be due to the lower density of cutaneous nerve fibers in human as compared to mouse skin 65 and the increased probability for injected PGE 2 to target sensory nerve terminals in mice vs . men. The unique role of SC EP2 to signal pain-like responses is highlighted by the receptor-mediated engagement of membrane cAMP nanodomains and membrane-associated AKAP/PKA to elicit pain-like responses in vivo. Our results expand the repertoire of SC intracellular mediators implicated in pain signaling, adding to endosomal cAMP increased by CGRP 21 , the plasma membrane cAMP compartment increased by PGE 2 as essential pain signals. A limitation of the present study is that currently available tools do not allow identification of the SC subtype, myelinated or Remak, mechanistically implicated in inflammatory pain-like responses, an issue that should be investigated by future research. An additional limitation is intrinsically linked to the incomplete efficiency of the silencing procedures that may confound interpretation of the results. An alternative approach for future studies may involve generating mouse lines with floxed alleles of Ptger2 or Ptger4 , which could allow for more efficient but still incomplete 66 , 67 gene deletion. However, the combination of AAV-induced silencing and selective pharmacological blockade with five different antagonists support our contention that SC EP2 is the major mediator of local PGE 2 -induced pain-like responses, but not inflammation, in the mouse hind paw. Other limitations include biases associated with translating from in vitro to in vivo as well as differences between species. Additionally, mouse models may not accurately represent specific human inflammatory diseases. However, we purposefully used the carrageenan and CFA tests that produce prostaglandin-dependent inflammation and pain-like responses aiming to decouple pain from inflammation. The present results highlight the role of the ROS and TRPA1 pathway in these inflammatory models. However, other possibilities might be considered, including TRPA1 overexpression in DRG neurons, which has been documented after inflammation and injury 68 , 69 . We found that TRPA1 mediates PGE 2 or carrageenan-induced allodynia within 30–60 min, which precludes TRPA1 overexpression. Further studies will be required to identify other mechanisms of PGE 2 or carrageenan-induced allodynia Recent evidence has highlighted the dual and opposing roles of NSAIDs. Although NSAIDs provide immediate relief from inflammatory pain, they substantially delay its resolution 6 . With the inherent limitations of the translational value of animal models of inflammatory pain, we were able to reproduce the delayed and insufficient recovery of allodynia provoked by diclofenac in the CFA model of inflammatory pain. More importantly, we found that an EP2 antagonist and selective silencing of EP2 and one of its downstream intracellular signaling mechanisms in SCs provided complete and sustained attenuation of inflammatory allodynia. Together, our results highlight the unique role of SC EP2 in inflammatory mechanical hypersensitivity, decoupling the PGE 2 -mediated proalgesic pathway from the concomitant PG-dependent inflammatory protective response that appears necessary for a timely and efficient pain resolution. Further preclinical and clinical studies may prove whether selective blockade of SC EP2 would provide better and safer treatments for inflammatory pain in patients and particularly in the elderly.

The protective role of inflammation includes the development of sustained pain to prevent the detrimental use of the injured tissue and allow for tissue healing. Inflammatory pain is treated with non-steroidal anti-inflammatory drugs (NSAIDs) that inhibit prostaglandin (PG) synthesis by blocking cyclooxygenase (COX) activity 1 , 2 . However, systemic PG inhibition can cause life-threatening gastrointestinal, renal, and cardiac toxicity 3 – 5 and, by reducing inflammation, delays pain resolution 6 . An ideal strategy to treat pain mediated by PGs would be to selectively reduce the pain without affecting their inflammatory and protective actions. Most studies on PG-induced pain-like responses have been based on the subcutaneous injection in rodents of PGE 2 , a major proalgesic and proinflammatory PG isoform 7 , which evokes a transient spontaneous nociceptive response and a sustained mechanical hypersensitivity, attributed to direct stimulation of terminals of dorsal root ganglia (DRG) nociceptors, via cyclic adenosine monophosphate (cAMP) formation and protein kinase A (PKA) activation 8 – 11 . A-kinase anchor protein (AKAP79/150)-dependent PKA also contributes to PGE 2 -induced nociceptor hypersensitivity 12 . Of the four PGE 2 receptor subtypes (EP1-4), EP4 has been implicated in PGE 2 -induced excitation and sensitization of nociceptors 13 . However, conflicting data on the contribution of EP4 to carrageenan- and CFA-evoked mechanical hypersensitivity 7 , 14 , the failure of an EP4-selective antagonist to provide benefit in migraine pain 15 , and absence of successful clinical trials with EP4 antagonists in inflammatory pain cast doubts on the PGE 2 receptor subtype and its cellular and molecular pathways implicated in inflammatory pain. Previous studies implicated EP2 in pain signaling in the periphery, including reports that EP2 deletion attenuates mechanical allodynia evoked by intradermal PGE 2 or zymosan 16 , and that a selective EP2 antagonist attenuates primary peripheral hyperalgesia in an endometriosis mouse model 17 . Schwann cells (SCs) make a major contribution to the maintenance of mechanical allodynia in murine models of neuropathic 18 and cancer 19 pain. The promigraine neuropeptide, calcitonin gene-related peptide (CGRP) 20 released from periorbital trigeminal nerve terminals elicits local mechanical allodynia by activating the calcitonin-like receptor and receptor activity-modifying protein 1 (CLR/RAMP1) on adjacent SCs, where sustained cAMP production in endosomes elicits persistent pain 21 . Disruption of low-density lipoprotein-receptor related protein-1 (LRP1) in SCs enhanced mechanical allodynia caused by nerve injury in mice 22 . Optogenetic stimulation of mouse cutaneous mouse SCs elicits mechanical hypersensitivity 23 . However, scant attention has been paid to the existence of PGE 2 receptors in peripheral glial cells and to their contribution to pain. EP1 and EP4 expression was reported in myelinated SCs 24 and EP2 was detected in dental pulp SCs 25 and oligodendrocytes 26 . Herein, we explored whether SCs contribute to PGE 2 -induced inflammatory pain. We observed that adeno-associated virus (AAV) cell-selective silencing of SC EP2 abrogates PGE 2 -dependent mechanical allodynia and grimace behavior without affecting inflammation. EP2 in human and murine SCs activates a proalgesic signaling pathway encompassing plasma membrane-compartmentalized cAMP/PKA that, via AKAP79/150 anchoring to specific subcellular nanodomains, signals mechanical allodynia elicited by carrageenan and CFA. This pathway is mechanistically independent from the PG-mediated inflammatory responses produced by these stimuli. In the CFA model of inflammatory pain in mice, in contrast with NSAIDs, SC EP2 antagonism or silencing reduces inflammatory pain without delaying its resolution.

Source Data Source Data