TRPV1 feed-forward sensitisation depends on COX2 upregulation in primary sensory neurons

Increased activity and excitability (sensitisation) of a series of molecules including the transient receptor potential ion channel, vanilloid subfamily, member 1 (TRPV1) in pain-sensing (nociceptive) primary sensory neurons are pivotal for developing pathological pain experiences in tissue injuries. TRPV1 sensitisation is induced and maintained by two major mechanisms; post-translational and transcriptional changes in TRPV1 induced by inflammatory mediators produced and accumulated in injured tissues, and TRPV1 activation-induced feed-forward signalling. The latter mechanism includes synthesis of TRPV1 agonists within minutes, and upregulation of various receptors functionally linked to TRPV1 within a few hours, in nociceptive primary sensory neurons. Here, we report that a novel mechanism, which contributes to TRPV1 activation-induced TRPV1-sensitisation within ~ 30 min in at least ~ 30% of TRPV1-expressing cultured murine primary sensory neurons, is mediated through upregulation in cyclooxygenase 2 (COX2) expression and increased synthesis of a series of COX2 products. These findings highlight the importance of feed-forward signalling in sensitisation, and the value of inhibiting COX2 activity to control pain, in nociceptive primary sensory neurons in tissue injuries.

Increased activity and excitability (sensitisation) of a series of molecules including the transient receptor potential ion channel, vanilloid subfamily, member 1 (TRPV1) in pain-sensing (nociceptive) primary sensory neurons are pivotal for developing pathological pain experiences in tissue injuries. TRPV1 sensitisation is induced and maintained by two major mechanisms; post-translational and transcriptional changes in TRPV1 induced by inflammatory mediators produced and accumulated in injured tissues, and TRPV1 activation-induced feed-forward signalling. The latter mechanism includes synthesis of TRPV1 agonists within minutes, and upregulation of various receptors functionally linked to TRPV1 within a few hours, in nociceptive primary sensory neurons. Here, we report that a novel mechanism, which contributes to TRPV1 activation-induced TRPV1-sensitisation within ~ 30 min in at least ~ 30% of TRPV1-expressing cultured murine primary sensory neurons, is mediated through upregulation in cyclooxygenase 2 (COX2) expression and increased synthesis of a series of COX2 products. These findings highlight the importance of feed-forward signalling in sensitisation, and the value of inhibiting COX2 activity to control pain, in nociceptive primary sensory neurons in tissue injuries.
Tissue injuries are followed by an inflammatory reaction, which aims to restore tissue integrity 1 . This inflammatory process is associated with pathological pain experiences including hypersensitivity to mechanical and heat stimuli known respectively as mechanical allodynia and heat hyperalgesia in humans [2][3][4] . The non-selective cationic channel, transient receptor potential ion channel, vanilloid subfamily, member 1 (TRPV1) is expressed in a major group of pain-sensing (nociceptive) primary sensory neurons, the first neurons detecting tissue pathology-it plays a pivotal role in the development of heat hyperalgesia and significantly contributes to the development of mechanical allodynia in inflammation following tissue injury 5,6 .
The development of hypersensitivities depends on plastic changes, known as sensitisation (a use-dependent increase in the activity and excitability), in neurons involved in nociceptive processing 7 . Upregulation of TRPV1 expression, TRPV1 translocation from the cytoplasm to the cytoplasmic membrane and post-translational modification-mediated increases in TRPV1 responsiveness and activity significantly contribute to the sensitised state of nociceptive primary sensory neurons and underlay the pivotal role of TRPV1 in the development of hypersensitivities to heat and mechanical stimuli in tissue inflammation [8][9][10][11][12][13][14][15][16][17] . Inflammatory mediators including nerve growth factor, bradykinin, prostaglandins (PGs) and ligands of various protease activated receptors (PARs), such as thrombin, mast cell-derived tryptase or kallikrein acting on their cognate receptors on TRPV1-expressing primary sensory neurons, initiate those changes in TRPV1 10,12,13,[18][19][20] . In addition, TRPV1 activation itself may also increase its own responsiveness and activity through feed-forward mechanisms occurring at various levels of signalling in primary sensory neurons. Firstly, TRPV1 activation through Ca 2+ influx induces the synthesis of the TRPV1 endogenous ligand anandamide in TRPV1-expressing primary sensory neurons within minutes 21,22 . Secondly, TRPV1 activation also induces, within hours, upregulation of the expression of genes whose products are implicated in post-translational modification-mediated increases in the responsiveness and activity of TRPV1 23 . Previously, Zhang and colleagues have reported a third type of feed-forward TRPV1 sensitisation that develops within ~ 20-30 min following TRPV1 activation by the archetypical exogenous ligand capsaicin and is mediated through extracellular signal-regulated kinase 1/2 (ERK1/2) and the Ca 2+ -calmodulin-dependent protein The development of capsaicin-induced sensitisation of subsequent capsaicin-evoked responses depends on COX2. The 30-min delay in the development of sensitised capsaicin-evoked responses following capsaicin application suggests that the sensitisation depends on changes in the transcription of immediate early genes ( 24 ; current data). A search for immediate early gene products that could induce TRPV1 sensitisation through ERK1/2 and CaMKII, the signalling molecules shown being involved 24 , suggested that upregulation in prostaglandin-endoperoxidase synthase 2 gene (ptgs2) and subsequent upregulation in COX2 expression and increased synthesis of COX2 products might underlie the development of capsaicin-induced sensitisation of subsequent capsaicin-evoked responses. Therefore, next we applied indomethacin (10 μM), a non-selective COX inhibitor, to the cells during the interim period between the two capsaicin applications. Indomethacin significantly reduced the proportion of cells exhibiting sensitised responses to ~ 3% (8 of 244; p < 0.0001; Fischer's exact test; Fig. 1D). Interestingly, while the degree of sensitisation was significantly greater in the remaining sensitizers (2.63 ± 0.9, n = 8 in indomethacin and 1.26 ± 0.03, n = 81 in control; p < 0.0001; ANOVA followed by Tukey's posthoc test; Fig. 1F), the ratio of the second and first normalised capsaicin-evoked responses in the non-sensitizer cells was significantly smaller in the presence of indomethacin than in control (0.41 ± 0.2, n = 236 in indomethacin and 0.84 ± 0.01, n = 204 in control; p < 0.0001; ANOVA followed by Tukey's post-hoc test; Fig. 1F).
Next, we applied the selective COX2 inhibitor NS-398 (3 μM) to the cells to specifically explore whether COX2, whose products have been reported to contribute to TRPV1 sensitisation 13,19,20 , is responsible for the sensitisation of the second capsaicin-evoked responses. As seen with indomethacin, NS-398 also significantly reduced the proportion of the sensitizing neurons (~ 3%, 1 of 37; p = 0.0002; Fischer's exact test; Fig. 1D). Again, the size of sensitisation in the single sensitizer neuron appeared greater (2.24 ± 0.9, n = 1 in NS-398), whereas the ratio of the second and first normalised capsaicin-evoked responses in the non-sensitizers was significantly smaller in the presence of NS-398 than in control (0.3 ± 0.03, n = 36; p < 0.0001; ANOVA followed by Tukey's post-hoc test; Fig. 1F). Collectively, these data suggested that capsaicin induced upregulation of ptgs2, COX2 and COX2 products could indeed significantly contribute to capsaicin-evoked sensitisation of subsequent capsaicininduced responses in primary sensory neurons. Thus, we further tested this hypothesis.
Capsaicin application to mouse cultured primary sensory neurons induces ptgs2 and COX2 upregulation. Conventional RT-PCR was used to assess ptgs2 expression in cultured primary sensory neurons after treating them with capsaicin or buffer for control. The size of the RT-PCR product (214 bp) was indistinguishable from the predicted size ( Fig. 2A; Supplementary Fig. 3). Analysis of gel images revealed that (A-C) shows typical recordings of changes in the intracellular Ca 2+ concentrations in three cultured murine primary sensory neurons in control extracellular buffer. Capsaicin (500 nM) was applied for 15 s at 3 (CAPS-1) and 33 min (CAPS-2). Applications of capsaicin were followed by the applications of KCl (50 mM) 2 min after capsaicin applications (KCl-1 and KCl2). The neurons whose responses are shown in (A,B) exhibited absolute (A), or relative to the KCl-evoked response (B), sensitised second capsaicin-evoked responses when the amplitudes are assessed. The cell whose response is shown in (C) exhibited a non-sensitised response. (D) Relative number of neurons exhibiting sensitizer (red) or non-sensitizer (blue) phenotype in control extracellular buffer (C), and in the presence of indomethacin (10 μM; I) or NS-398 (3 μM; NS). Both indomethacin and NS-398 significantly reduced the proportion of sensitizers (p < 0.0001; Fischer's exact test for indomethacin and p = 0.0002; Fischer's exact test for NS-398). The numbers above the columns indicate the total number of capsaicin responding neurons. (E) Averaged normalised amplitudes (to the amplitudes of KCl-evoked responses) of first (green) and second (red for sensitizers (Sens), blue for non-sensitizers (Nonsens)) capsaicin application-evoked responses. In sensitizers and non-sensitizers the second capsaicinevoked responses were, respectively, significantly greater (p < 0.0001; Student's paired t-test; n = 81) and smaller (p < 0.0001, Student's paired t-test; n = 204) than the amplitudes of the first capsaicin-evoked responses. The numbers at the bottom of the columns indicate the number of neurons. (F) Ratio of first and second normalised (to KCl-evoked responses) capsaicin-evoked responses in control buffer (C) and in the presence of indomethacin (10 μM; I) or NS398 (3 μM; NS) in sensitizer (Sens) and non-sensitizer (Non-sens) neurons. Both indomethacin and NS-398 increased or reduced the ratio in sensitizer or non-sensitizer neurons, respectively (indomethacin: p < 0.0001; Student's t-test for sensitizers and p < 0.0001; ANOVA followed by Tukey's post-hoc test for non-sensitizers; NS398: p < 0.0001; ANOVA followed by Tukey's post-hoc test for non-sensitizers). The numbers at the bottom of the columns indicate the number of neurons. www.nature.com/scientificreports/ www.nature.com/scientificreports/ capsaicin treatment indeed significantly upregulated ptgs2 expression (p = 0.0004; Student's two samples t-test, n = 3) in cultured primary sensory neurons ( Fig. 2A,B). Next, we used Western blotting and immunostaining to confirm that the de novo synthesised ptgs2 is translated into protein. Western blot images confirmed that cultured primary sensory neurons exhibit basal COX2 expression ( Fig. 2C; Supplementary Fig. 4; [25][26][27][28]. Image analysis also revealed that capsaicin exposure induced a significant increase in that expression (p = 0.0025; Student's t-test, n = 4; Fig. 2C,D).
Capsaicin application to mouse cultured primary sensory neurons increases the concentration of a series of COX2 products. To confirm that the upregulated COX2 expression is associated with increased PG synthesis, we collected the cells and superfusate of mouse cultured primary sensory neurons following their exposure to capsaicin or buffer for control, and used ultra-performance liquid chromatography with tandem mass spectrometry (UPLC-MS) for targeted analysis of the samples for a group of arachidonic acid-derived compounds ( Table 1). Our measurements showed that only a few fatty acids, lipoxygenase (LOX) products and one COX product were above the detection level in the first experiment (Table 1). Therefore, in order to increase the abundance of the products above the detection threshold, we repeated the experiments by exposing the cultures to arachidonic acid (10 µM), thus providing a substrate, during the 30-min incubation period. In these cultures, 32 of the 48 compounds included in the panel were above the detection level (Table 1). Among the COX products, four exhibited significant increase following capsaicin treatment; PGD2, 8-iso-PGFa, 15dPGJ2 and 11d-tromboxane2 (11dTBX2; Fig. 2H).

Discussion
Eicosanoids and related compounds, including PGE2, 15dPGJ2 and PGD2, PGI2 have been implicated in inducing TRPV1 sensitisation 13,19,20 . It is also reported that activation of primary sensory neurons leads to COX2 upregulation [25][26][27][28] . Further, it is reported that a group of primary sensory neurons exhibits sensitisation to capsaicin within 20-30 min after a previous capsaicin application 24 . However, this is the first report which shows that increased ptgs2 and COX2 expression and synthesis of a series of COX2 products are triggered by TRPV1 activation and thus underlie the development of sensitisation of TRPV1 in ~ 30 min after the first TRPV1 activation.
When capsaicin is repeatedly applied within a short period of time (i.e. tens of seconds -minutes), subsequent responses exhibit rapid desensitisation (for refs see 29 ). However, when capsaicin application is repeated only in tens of minutes, responses are sensitised 24 . Here we report that, similarly to rat cultured primary sensory neurons 24 , murine cultured primary sensory neurons also exhibit this type of sensitisation, though there are some differences between the two species. In mouse primary sensory neurons, the estimated proportion of cells exhibiting sensitisation is lower and the correlation between the amplitude of the first and second responses is less pronounced than in rat primary sensory neurons 24 . Regarding the proportion of sensitizer neurons, however, it should be noted that our estimation may be conservative as only neurons which exhibited at least a 5% increase in the amplitude of the second capsaicin-evoked responses were considered sensitizers. The assumption that we underestimated the proportion of sensitizers is supported by the findings that while ~ 1/3 of the capsaicin-responsive neurons exhibited at least 5% increase in responses to the second capsaicin application, about 1/2 of the TRPV1-expressing neurons appeared to express COX2-both in the control environment and after capsaicin exposure.
While Zhang and co-workers found involvement of ERK1/2 and CaMKII signalling 24 here we have identified further down-stream elements of the signalling pathway(s) which lead to capsaicin-evoked delayed sensitisation of TRPV1mediated responses. We found that blocking COX activity-either by the nonselective COX inhibitor indomethacin, or COX2 activity by the selective COX2 inhibitor NS-398-almost completely eliminated the delayed sensitisation. These data suggest that COX2 activity is pivotal in delayed TRPV1 sensitisation. However, the few sensitizer neurons remaining in our samples after blocking COX or COX2 indicate that in addition to COX2, other mechanisms may also contribute to TRPV1 feed-forward sensitisation, albeit to a seemingly lesser extent than COX2.
Results of the RT-PCR, Western blotting and immunostaining together confirmed the role of COX2 in the capsaicin-induced delayed TRPV1 sensitisation as we detected upregulated synthesis of ptgs2 and COX2. In accordance with findings that a group of primary sensory neurons expresses COX2 25-28 , we found basal ptgs2 and COX2 expression in our neuron samples. Our data indicate that COX2 basal expression, in agreement with previous reports on inducible COX2 expression in a group of primary sensory neurons 25,26 , was significantly increased after capsaicin exposure. Further, here we report that capsaicin application to cultured primary sensory neurons also results in increased synthesis of COX2 products-though only when cells were presented by the COX2 substrate arachidonic acid. The failure in finding COX2 products with the absence of arachidonic acid in the buffer indicates the concentration of COX2 products without the presence of exogenous arachidonic acid is below the detection threshold when dissociated primary sensory neurons derived from one mouse are divided into control and treated cultures. Interestingly, in addition to several COX products, the concentration of several fatty acids and lipoxygenase products were also increased after arachidonic acid supplementation. Importantly, several of those upregulated lipoxygenase products, for example LTB4 or 15-(S)-HETE, might also contribute to delayed TRPV1 sensitisation; for example, in cells unaffected by the COX and COX2 inhibitors 30,31 .
Several COX2 products, including PGE2, 15dPGJ2 and PGD2, PGI2, have been implicated in acting on TRPV1-expressing primary sensory neurons 13 www.nature.com/scientificreports/ COX products included in the panel we used (PGD2, PGF2a, 15dPGJ2 and 11dTBX2), showed significantly increased synthesis after capsaicin exposure of cultured murine primary sensory neurons. This finding suggests that enzymes downstream of COX2 may exhibit differential expression and/or activation in primary sensory neurons after capsaicin exposure. Among the upregulated prostaglandins we found PGD2, which previously has been shown to activate PKA type II in TRPV1-expressing primary sensory neurons 18 . As shown previously, PKA-mediated TRPV1 phosphorylation results in TRPV1 sensitisation 15 indicating that PGD2 might indeed sensitise TRPV1. However, Isensee and co-workers 18 found prostaglandin D synthase expression in primary sensory neurons that lack TRPV1 activity. Therefore, at present it is not clear whether, in our cultures, TRPV1expressing neurons themselves or non-TRPV1-expressing cells, activated by some agents released from TRPV1-expressing neurons by capsaicin, constitute the source of PGD2.
Another COX-related product we found upregulated, after exposing our cultures to capsaicin, is 15d-PGJ2. This product is synthesised from PGD2 through PGJ2 32 . Previously, 15d-PGJ2 has been reported to activate TRPV1 through covalently binding to the ion channel 33 .
Interestingly, even though the concentration of PGE2 (the best known, most powerful TRPV1 sensitizer, and widely studied COX product) was increased following capsaicin exposure-this change failed to reach the level of significance. This result might suggest that PGE2 has little role in the development of feed-forward TRPV1 sensitisation. Instead, PGE2 released from monocytes or differentiated macrophages 34 may have a greater contribution to a quick sensitisation of TRPV1 during inflammation 35 .
The novel downstream effects of TRPV1 activation we present here were induced by the exogenous TRPV1 activator capsaicin. These downstream effects can be biologically meaningful only if endogenous TRPV1 activators are also able to induce it. To the best of our knowledge, no data on the TRPV1 activation-evoked TRPV1 sensitisation by endogenous TRPV1 ligands are currently available. However, similarly to capsaicin, endogenous TRPV1 activators also induce Ca 2+ influx, which activates ERK1/2 and CaMKII, the signalling molecules that mediate the TRPV1 activation-evoked TRPV1 sensitisation 24 . Hence, while it must yet be elucidated, it is highly likely that endogenous TRPV1 activators evoke similar COX2-dependent TRPV1 activation-induced TRPV1 sensitisation than capsaicin.
In summary, here we report the phenomenon of feed-forward TRPV1 sensitisation in murine primary sensory neurons, which depends largely on COX2 activity. We also report that COX2 activity is accompanied by upregulated synthesis of the enzyme itself and a significant increase in the synthesis of a number of COX-related products. While COX products released from immunocompetent cells sensitise TRPV1 seconds after acting on TRPV1-expressing primary sensory neurons, agents produced by COX2 in TRPV1-expressing primary sensory neurons themselves appear to prolong the sensitisation state. Hence, the COX2-mediated feed-forward TRPV1 sensitisation represents another 'layer' of PG-mediated TRPV1 sensitisation. While the development of this second layer requires ~ 30 min, due to the development of a TRPV1 activation-COX2 upregulation-upregulated PG synthesis-TRPV1 sensitisation-TRPV1 activity reverberating circuit, it may maintain the sensitised state of TRPV1 for a prolonged period of time even when the concentration of PGs are reduced in the interstitial milieu. Interestingly, our data suggest that COX2-dependent sensitisation through peripheral and primary sensory neuron mechanisms may involve different COX-related agents. The relative contribution of TRPV1 sensitisation through immunocompetent and primary sensory neuron mechanisms is not known at present. Further, the receptors involved in the delayed sensitisation process and the signalling pathways have not been elucidated. Nevertheless, these findings, in addition to the "immediate" (within minutes) and "late" (within hours) feedforward TRPV1 sensitisation (mediated respectively by TRPV1 activators such as anandamide 21,22 and secondary gene products such as PARs 23 reveal a novel, delayed (~ 30-min) mechanism of feed-forward autocrine signalling leading to TRPV1 sensitisation through the upregulation of immediate early gene and gene products. Therefore, these findings also indicate that, in addition to TRPV1 sensitisation by inflammatory mediators, feed-forward autocrine TRPV1 sensitisation ("autosensitisation") is multifaceted and that COX2 inhibition in primary sensory neurons is important in reducing inflammatory heat and mechanical hypersensitivity.

Materials and methods
Experiments were conducted in accordance with guidelines published by the International Association for the Study of Pain (IASP) 36 , UK Animals (Scientific Procedures) Act 1986, revised National Institutes of Health Guide for the Care and Use of Laboratory Animals, Directive 2010/63/EU of the European Parliament and of the Council on the Protection of Animals Used for Scientific Purposes, Good Laboratory Practice and ARRIVE guidelines. A total of 24 wild type C57BL/6 mice were used. Mice were killed humanely following Isoflurane anaesthesia. Experiments were designed and approved by Imperial College London's Animal Welfare and Ethical Review Body in compliance with Animals (Scientific Procedures) Act1986 (amended in 2012). Every effort was taken to minimize the number of animals.
Culturing primary sensory neurons. Primary sensory neuron cultures were prepared as described 37 .
Briefly, dorsal root ganglia (DRG) from the 3rd cervical to the 6th lumbar segments were harvested, placed into Ham's Nutrient Mixture F12 supplemented with Ultraser G (2% final concentration), l-glutamine (0.2 mM final concentration), penicillin (100 IU/ml final concentration) and streptomycin (100 μg/ml final concentration) and incubated at 37 °C for 2.5 h with collagenase type IV (312.5-625 IU/ml final concentration) followed by an additional 30 min incubation in the presence of trypsin (2.5 mg/ml final concentration). Ganglia were then washed in F12, triturated then plated onto poly-dl-ornithine-coated coverslips. Cultures were incubated in the supplemented F12 at 37 °C for 24-48 h before being used for further experiments. www.nature.com/scientificreports/ Calcium imaging. Imaging was conducted as described previously 38 . Briefly, cells were loaded with 1 μM FURA-2 (Invitrogen) and 2 mM probenecid (Invitrogen) for 45-60 min in extracellular buffer (NaCl, 150 mM; KCl, 5 mM; CaCl 2 , 2 mM; MgCl 2 , 2 mM; glucose, 10 mM and HEPES, 10 mM; pH7.4) at 37 °C. Coverslips were superfused (~ 1.5 ml/min) with the extracellular buffer at 37 °C in a flow chamber. The outlet of the perfusion system was positioned next to the group of cells from which the recordings were obtained. Only one field of view was tested on each coverslip. Cells were exposed to excitation light alternating between the wavelengths of 355 and 380 nM (OptoLED Light Source, Cairn Research, UK). The fluorescent signals emitted in response to each excitation were captured with an optiMOS camera (QImiging, Canada) connected to a PC running the WinFluor software package (John Dempster, Strathclyde University, UK). Images were analysed offline with the WinFlour and Clampfit (Molecular devices, USA) software packages. Briefly, after establishing the regions of interest, the time course of changes in the fluorescent ratios were examined and exported to Clampfit where the standard deviation (SD) of the baseline activity and the maximum amplitudes of responses were measured. Change in the fluorescent ratio was accepted as a response if the generation of the response corresponded to the drug application and the amplitude of the change was greater than 3 times the SD of the baseline activity. The number of cells in various groups (i.e. sensitizer and nonsensitizer neurons) and the amplitude of responses of cells from at least three independent cultures were pooled and averaged.

Reverse transcriptase polymerase chain reaction (RT-PCR).
RT-PCR was prepared as described previously 38  Western blotting. Cells were incubated in 500 nM capsaicin or in HBSS at 37 °C at 37 °C for 5 min then in HBSS supplemented with 5 μg/mL selective proteasome inhibitor MG-132 in DMSO supplemented for 30 min. 50 μL of NP40 lysis buffer (Invitrogen, UK) supplemented with 1 mM phenyl methane sulfonyl fluoride (PMSF) and 1 × protease inhibitor cocktail (Sigma-Aldrich, USA) was used for protein extraction. The lysates were transferred to Eppendorf tubes and spun down at 14,800 rpm at 4 °C for 10 min and the supernatant collected. The protein concentrations of cell lysates were measured before the SDS-PAGE. The supernatant was mixed with 6 × reducing sample buffer (500 mM Tris pH6.8, 0.1 mg/mL SDS, 34% glycerol, 0.1 mg/mL bromophenol blue, 5% β-mercaptoethanol) and heated at 98 °C for 2 min. An equal mass of proteins was loaded onto 9% polyacrylamide gels. Proteins were separated and transferred onto nitrocellulose membrane (GE Healthcare Life Sciences). The membrane was washed in tris-buffered saline containing 0.1% Tween (Sigma; (TBST) then blocked in 5% non-fat milk or bovine serum albumin (BSA; Sigma) in TBS-T at room temperature for 1 h. The membrane was then probed with 1:2500 diluted anti-COX-2 (ab15191, Abcam) and 1:2000 diluted anti-β-tubulin III (neuronal) (T8578, Sigma-Aldrich), supplemented in 5 mL of 5% non-fat milk or BSA in TBS-T at 4 °C overnight. The reaction was visualised using HRP-conjugated anti-rabbit IgG (7074S, Cell Signalling) and anti-mouse IgG (7076S, Cell Signalling) and the Pierce ECL 2 Western blotting substrate following the manufacturer's instructions. The membranes were then imaged by the imaging machine GeneGnome XRQ SynGene. Analysis of immunoblots was performed using ImageJ. The intensity of the COX2 signal was normalised to the intensity of the β-tubulin III signal. No modifications on membrane images other than cropping were done. The membrane image from one culture is shown in Supplementary Fig. 4. Immunostaining. Cells were incubated in 500 nM capsaicin or in HBSS at 37 °C for 5 min, followed by an incubation in HBSS supplemented with 5 μg/mL selective proteasome inhibitor MG-132 in DMSO at 37 °C for 30 min and fixing for 15 min in 4% paraformaldehyde. Cells were permeabilised (0.3% Titon X), blocked in 5% normal goat serum and incubated in anti-COX-2 M19 (sc-1747, Santa Cruz) antibody (1:200)  www.nature.com/scientificreports/ 2000R digital camera (QImaging, USA) controlled by Leica Application Suite. All the images of randomly taken fields were taken with the same camera settings and conditions. Immunofluorescence images without any processing were analysed using ImageJ. The average COX2 and TRPV1 immunofluorescent signals were measured in a group of neurons in randomly selected areas in capsaicintreated and control coverslips prepared from 3 cultures (> 100 cells/culture). Intensity values from each coverslips were plotted, the threshold for positive immunofluorecent signals and the average signal intensity and the proportion of COX2-and/or TRPV1-expressing cells was established 39 . No modifications other than cropping were done on images shown in Fig. 2. Measuring capsaicin-induced eicosanoid production. Cells were incubated at 37 °C for 5 min in capsaicin (500 nM) or HBSS buffer for control. The wells were then incubated for an additional 30 min in HBSS or in HBSS with 10 µM arachidonic acid. Equivolume of 100% methanol was added to wells. Then, cells were scraped, collected in Eppendorf tubes, vortexed and 50µL of the lysate were removed from each sample for protein quantification. The samples were stored at − 20 °C.
Protein quantification was performed using the Pierce BCA protein assay kit (Thermo Fisher Scientific, UK) according to the instructions. Standards and samples were measured in duplicates. Reading was performed at a wavelength of 540 nm in a BioTek ELx800 spectrophotometer (Fisher Scientific, UK) and duplicate readings were averaged.
The rest of the samples were analysed at the MRC-NIHR National Phenome Centre (UK) using a novel targeted assay developed for the quantification of 48 lipid mediators of inflammation 40 . Briefly, separation was performed using a Waters HSS T3 column. Samples were then analysed in negative ion mode in a Waters Xevo TQ-S triple quadrupole mass spectrometer (MS) with an electrospray ionisation (ESI) source. Concentrations detected by UPLC-MS were first normalised to the protein content of the samples. Then, normalised compound concentrations of the capsaicin-treated cells were normalised to values established in the respective control.
Statistical analysis. Normal distribution of data was assessed using the Shapiro-Wilk test. Significant change in the number of sensitising/non-sensitising and immunostained cells was assessed using Fischer's exact test. Amplitudes in Ca 2+ imaging were compared using Student's t-test or ANOVA followed by Tukeys post-hoc test as appropriate. Changes in both expression values and the concentration of lipid mediators were compared using Student's t-tests. In the case of the concentration of lipid mediators, Student's test values were corrected for false discovery rate using the Benjamini-Hochberg approach. Values are expressed as average ± standard error of mean (SEM). Differences were regarded as significant at p/q < 0.05. p values above 0.0001 are given as exact values, whereas below 0.0001, they are given as p < 0.0001. www.nature.com/scientificreports/ Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/.