Cox-Dependent Fatty Acid Metabolites Cause Pain Through Activation of the Irritant Receptor TRPA1

Cox-dependent fatty acid metabolites cause pain through activation of the irritant receptor TRPA1

Serena Materazzi*, Romina Nassini*, Eunice Andre` †, Barbara Campi†, Silvia Amadesi‡, Marcello Trevisani†, Nigel W. Bunnett‡, Riccardo Patacchini§, and Pierangelo Geppetti*†¶

*Department of Preclinical and Clinical Pharmacology, University of Florence, Florence, Italy; †Center of Excellence for the Study of Inﬂammation, University of Ferrara, Ferrara, Italy; ‡Departments of Surgery and Physiology, University of California, San Francisco, CA 94143; and §Pharmacology Department, Chiesi Pharmaceuticals, Parma, Italy

Edited by Susan E. Leeman, Boston University School of Medicine, Boston, MA, and approved June 6, 2008 (received for review March 8, 2008)

Prostaglandins (PG) are known to induce pain perception indirectly EP1–4 and IP receptors, respectively (7–9). Accordingly, the by sensitizing nociceptors. Accordingly, the analgesic action of analgesic property of NSAIDs and selective COX-2 inhibitors nonsteroidal anti-inﬂammatory drugs (NSAIDs) results from inhi- (Coxibs) has been assigned to their ability to suppress nociceptor bition of cyclooxygenases and blockade of PG biosynthesis. Cyclo- sensitization by COX metabolites (7–9). pentenone PGs, 15-d-PGJ2, PGA2, and PGA1, formed by dehydration Transient receptor potential A1 (TRPA1) is a recently identified excitatory ion channel (10, 11) that is coexpressed with the of their respective parent PGs, PGD2, PGE2, and PGE1, possess a highly reactive ␣,␤-unsaturated carbonyl group that has been ‘‘capsaicin receptor’’ TRPV1 by a subpopulation of primary proposed to gate the irritant transient receptor potential A1 afferent neurons containing substance P (SP) and calcitonin gene-related peptide (CGRP), which mediate pain and neuro- TRPA1) channel. Here, by using TRPA1 wild-type (TRPA1؉/؉)or) ؊/؊ genic inflammation (12–15). Agonists of TRPA1 include the deﬁcient (TRPA1 ) mice, we show that cyclopentenone PGs pungent ingredients of various spices, including mustard, garlic, produce pain by direct stimulation of nociceptors via TRPA1 acti- and cinnamon (11, 16), environmental irritants such as acrolein vation. Cyclopentenone PGs caused a robust calcium response in ؉ ؉ (14), formaldehyde (17), and 4-hydroxy-2-nonenal (4-HNE), an dorsal root ganglion (DRG) neurons of TRPA1 / , but not of endogenous ␣,␤-unsaturated aldehyde generated in response to .(TRPA1؊/؊ mice, and a calcium-dependent release of sensory neu- oxidative stress after inflammation and tissue injury (18, 19

ropeptides from the rat dorsal spinal cord. Intraplantar injection of Cyclopentenone prostaglandins, including PGA2, PGA1, and PHARMACOLOGY cyclopentenone PGs stimulated c-fos expression in spinal neurons PGJ2, are PG metabolites formed by dehydration within the of the dorsal horn and evoked an instantaneous, robust, and cyclopentane ring of PGE2, PGE1, and PGD2, respectively. The transient nociceptive response in TRPA1؉/؉ but not in TRPA1؊/؊ biological actions of cyclopentenone PGs are not mediated by mice. The classical proalgesic PG, PGE2, caused a slight calcium activation of the classical G protein-coupled prostanoid receptors, response in DRG neurons, increased c-fos expression in spinal neu- but rather through interaction with other target proteins (20). For ⌬12,14 rons, and induced a delayed and sustained nociceptive response in example, 15-deoxy- -PGJ2 (15-d-PGJ2), a PGJ2 metabolite, is ؉/؉ ؊/؊ a potent natural activator of the nuclear receptor peroxisome both TRPA1 and TRPA1 mice. These results expand the mech- ␥ anism of NSAID analgesia from blockade of indirect nociceptor sen- proliferator-activated receptor (PPAR)- (20). Cyclopentenone PGs are characterized by the presence of a highly reactive ␣,␤- sitization by classical PGs to inhibition of direct TRPA1-dependent unsaturated carbonyl group within their cyclopentenone ring, a nociceptor activation by cyclopentenone PGs. Thus, TRPA1 antago- moiety that confers the ability to adduct thiol-containing com- nism may contribute to suppress pain evoked by PG metabolites pounds, via Michael addition (20), and through this mechanism without the adverse effects of inhibiting cyclooxygenases. they have been proposed to stimulate TRPA1 (21). We hypothesized that cyclopentenone PGs stimulate TRPA1 cyclopentenone isoprostane ͉ cyclopentenone prostaglandins and thereby directly activate nociceptors to cause pain, in contrast to the classical PGs that indirectly sensitize nociceptors. rostanoids are a group of bioactive compounds that include We have obtained genetic evidence that cyclopentenone PGs excite nociceptors, release sensory neuropeptides, activate spinal Pprostaglandins (PGD2, PGE2, PGF2␣), prostacyclin (PGI2), and thromboxane A . They originate from arachidonic acid, nociceptive neurons, and induce acute pain by a mechanism that 2 is completely and exclusively dependent on TRPA1. Cyclopen- which is released intracellularly from plasma membrane phos- tenone PGs may represent a new class of COX-dependent algesic pholipids upon tissue damage and inflammation. Constitutively agents, and their novel mechanism of action adds a new dimen- expressed cyclooxygenase (COX)-1 and inducible COX-2 con- sion to our understanding of the mechanisms of the analgesic vert arachidonic acid to the precursors PGG2 and PGH2, from effect of NSAIDs and Coxibs. which all prostanoids are generated by tissue-specific synthases. The C20 polyunsaturated fatty acid, dihomo-␥-linolenic acid, is Results also a substrate of COX for eicosanoid production, from which Cyclopentenone PGs Activate TRPA1 in Dorsal Root Ganglion (DRG) prostaglandins of the 1-series, including PGE1, derive (1). Inhi- Neurons. To assess whether cyclopentenone PGs directly excite bition of COX activity is the main mechanism of action of aspirin DRG neurons by a TRPA1-dependent mechanism, we tested the and related nonsteroidal anti-inflammatory drugs (NSAIDs), as discovered by the pioneering work of Vane and colleagues (2). At least eight G protein-coupled receptor subtypes for prosta- Author contributions: N.W.B., R.P., and P.G. designed research; S.M., R.N., E.A., B.C., S.A., and M.T. performed research; S.M., R.N., E.A., B.C., S.A., M.T., N.W.B., R.P., and P.G. noids have been identified, which differ in tissue distribution, analyzed data; and N.W.B., R.P., and P.G. wrote the paper. signal transduction pathways, and sensitivity to the various The authors declare no conﬂict of interest. agonists (3, 4). This article is a PNAS Direct Submission. Pain relief, by far, is the most frequent therapeutic indication ¶To whom correspondence should be addressed at: Department of Preclinical and Clinical of COX inhibitors, and the contribution of prostanoids and their Pharmacology, University of Florence, Viale Pieraccini 6, 50139 Florence, Italy. E-mail: receptors to pain has been an area of intense investigation (5, 6). pierangelo.geppetti@uniﬁ.it. PGs do not directly excite primary sensory neurons but rather This article contains supporting information online at www.pnas.org/cgi/content/full/ indirectly sensitize the peripheral terminals of nociceptors, with 0802354105/DCSupplemental. PGE2 and PGI2 inducing a robust sensitization by activating © 2008 by The National Academy of Sciences of the USA

www.pnas.org͞cgi͞doi͞10.1073͞pnas.0802354105 PNAS ͉ August 19, 2008 ͉ vol. 105 ͉ no. 33 ͉ 12045–12050 Downloaded by guest on September 24, 2021 2ϩ Fig. 1. Cyclopentenone PGs and isoprostane increase [Ca ]i in mouse DRG neurons. (A) Concentration-response curves of cyclopentenone PGs or isoprostane and 2ϩ PGE2 in DRG neurons of Swiss mice (n ϭ 521 neurons). (B) Representative traces of [Ca ]i evoked by 15-d-PGJ2 (30 ␮M) or PGE2 (30 ␮M), followed by capsaicin (CPS, ϩ/ϩ Ϫ/Ϫ 1 ␮M) and potassium (KCl, 50 mM). A calcium response was induced by 15-d-PGJ2 in DRG neurons from TRPA1 but not from TRPA1 mice. PGE2, capsaicin or ϩ/ϩ Ϫ/Ϫ 2ϩ potassium evoked a similar response in neurons from TRPA1 or TRPA1 mice. (C) Pooled data of [Ca ]i evoked by cinnamaldehyde (CNM, 30 ␮M), capsaicin (1 ␮M), ϩ/ϩ Ϫ/Ϫ 15-d-PGJ2, PGA1, PGA2, 8-iso-PGA2, PGD2, and PGE2 (all 30 ␮M) in DRG neurons from TRPA1 (gray or orange columns) or TRPA1 (ﬁlled or red columns) mice. Among CPS-sensitive neurons from wild-type animals, a subpopulation of CNM-responsive cells was observed. In DRG neurons from TRPA1-deﬁcient mice CNM, 15-d-PGJ2, ϩ/ϩ Ϫ/Ϫ PGA1, PGA2, or 8-iso-PGA2 did not cause any response. PGE2 evoked a small response in DRG neurons from TRPA1 or TRPA1 mice. PGD2 failed to cause any response. ϩ ϩ Ϫ Ϫ The robust response evoked by capsaicin was similar in neurons from TRPA1 / and TRPA1 / mice. Each column represents the mean Ϯ SEM of n Ն 23 cells; *, P Ͻ 0.01 compared with TRPA1Ϫ/Ϫ DRG neurons.

Ϫ/Ϫ ability of 15-d-PGJ2, PGA2, and PGA1 to mobilize intracellular neurons from TRPA1 mice were almost completely unrespon- 2ϩ calcium ([Ca ]i) in DRG neurons from Swiss mice and from sive to 15-d-PGJ2, PGA2, PGA1, and 8-iso-PGA2 (Fig. 1 B and C), TRPA1 wild-type (TRPA1ϩ/ϩ) and deficient (TRPA1Ϫ/Ϫ) mice. and the ratio between the number of neurons responding to We also evaluated the effects of the isoprostane, 8-iso-PGA2, cyclopentenone compounds (30 ␮M) and neurons responding to which can also stimulate TRPA1 (21). Isoprostanes are PG-like capsaicin was: PGA2 (2/34, 5%), PGA1 (3/46, 2%), 15-d-PGJ2 (2/30, substances whose generation does not require COX activation 6%), and 8-iso-PGA2 (3/48, 6%). Moreover, the magnitude of the 2ϩ Ϫ/Ϫ (22). A cyclopentenone isoprostane isomer of PGA2, 8-iso- increase in [Ca ]i in the few neurons from TRPA1 mice that PGA2, like cyclopentenone PGs, is produced by dehydration of responded to cyclopentenone compounds was always Ͻ5% of that ϩ/ϩ E-isoprostane within the cyclopentane ring. It (8-iso-PGA2) of ionomycin, whereas responses of neurons from TRPA1 mice contains an electrophilic ␣,␤-unsaturated carbonyl moiety that were 30–50% of the ionomycin response (Fig. 1C). No response to ϩ/ϩ rapidly adducts cellular thiols by Michael addition (23). PGD2 was observed in neurons from either TRPA1 or Ϫ/Ϫ 15-d-PGJ2, PGA2, PGA1, and 8-iso-PGA2 evoked a concen- TRPA1 mice (Fig. 1C). In cultured DRG neurons from Swiss 2ϩ tration-dependent increase in [Ca ]i in DRG neurons of Swiss mice PGE2 caused a small response (maximal response was 10.9 Ϯ mice. EC50s (confidence interval, C.I.) were 8.9 (7.3–10.7); 24.0 1.2% of ionomycin, n ϭ 84 cells), with an EC50 of 7.6 (C.I., 2.7–21.4) (20.4–27.9); 15.1 (9.4–24.2); 22.4 (10.8–46.7) ␮M, respectively ␮M. The small calcium response to PGE2 was identical in (Fig. 1A). All neurons activated by cyclopentenone PGs or TRPA1ϩ/ϩ and TRPA1Ϫ/Ϫ mice (Fig. 1 B and C), and the propor- 8-iso-PGA2 also responded to capsaicin, while Ϸ40% of the tion of neurons responding to both PGE2 and capsaicin was capsaicin-sensitive neurons (n ϭ 437) responded to the various identical in TRPA1ϩ/ϩ (8/25, 32%) and TRPA1Ϫ/Ϫ (12/40 30%) cyclopentenone PGs or isoprostane. These findings are consis- mice. As expected, the TRPV1 agonist capsaicin produced similar tent with the report that cyclopentenone PGs and isoprostane responses in neurons from TRPA1ϩ/ϩ and TRPA1Ϫ/Ϫ mice (Fig. 1 excite nociceptive, capsaicin-sensitive trigeminal neurons of B and C), whereas the selective TRPA1 agonist, cinnamaldehyde C57BL6 mice (21). (25), evoked calcium responses exclusively in neurons from In DRG neurons from TRPA1ϩ/ϩ mice (C57 background), TRPA1ϩ/ϩ mice (Fig. 1C). Thus, cyclopentenone PGs and isopros- 15-d-PGJ2, PGA2, PGA1, and 8-iso-PGA2 (all 30 ␮M) increased tane strongly excite nociceptive neurons by a mechanism that is 2ϩ [Ca ]i, comparable to responses of neurons from Swiss mice (Fig. entirely and exclusively dependent on TRPA1, whereas the minor 1 B and C). The ratio between the number of neurons responding response to PGE2 is unrelated to TRPA1. to cyclopentenone compounds and neurons responding to capsaicin was: PGA2 (23/47, 49%), PGA1 (28/72, 38%), 15-d-PGJ2 (36/76, Cyclopentenone PGs Release Neuropeptides from Central Endings of 53%), and 8-iso-PGA2 (19/43, 44%). These percentages are entirely Nociceptors. To test whether TRPA1-evoked calcium mobiliza- consistent with the report that TRPA1-expressing, mustard oil- tion by cyclopentenone compounds stimulated the release of responsive neurons represent a subset (Ϸ50%) of TRPV1- neuropeptides from central endings of nociceptors, we chal- expressing, capsaicin-responsive DRG neurons (16, 24). In contrast, lenged slices of rat dorsal spinal cord with cyclopentenone PGs

12046 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0802354105 Materazzi et al. Downloaded by guest on September 24, 2021 Fig. 2. Cyclopentenone PGs and isoprostane release neuropeptides in the dorsal spinal cord. Release of CGRP-like immunoreactivity (LI) (upper panel) and SP-LI (bottom panel) from slices of rat spinal cord after exposure to cyclopentenone PGs or isoprostane (30 ␮M). Desensitization by pretreatment with capsaicin (CPS), or omission and chelation of extracellular Ca2ϩ (1 mM EDTA, ϪCa) abolished the evoked neuropeptide release. Each column represents mean Ϯ SEM; n Ն 4 experiments per condition. *, P Ͻ 0.05 compared with cyclopentenone agonist alone.

and measured release of CGRP and SP. 15-d-PGJ2, PGA2, a nocicefensive response and examined the contribution of PHARMACOLOGY PGA1, and 8-iso-PGA2 (30 ␮M) strongly stimulated CGRP and TRPA1 to this response. Intraplantar injection (20 ␮l/paw) of SP release (Fig. 2). Desensitization of nociceptors by prolonged capsaicin (0.1 nmol) or cinnamaldehyde (30 nmol) caused a rapid exposure to a high concentration of capsaicin (10 ␮M, 20 min) and intense nociceptive response that terminated after 2–4 min abolished these responses. This procedure renders capsaicin- in TRPA1ϩ/ϩ mice (Fig. 4). In TRPA1Ϫ/Ϫ mice the response to sensitive nerve terminals unresponsive to substances that excite capsaicin was unaltered, whereas that to cinnamaldehyde was nociceptors through TRPV1 and other mechanisms (26, 27). completely absent (Fig. 4). Intraplantar injection of 15-d-PGJ2 Depletion of extracellular calcium also abolished responses to (15 nmol) caused a robust nociceptive behavior in TRPA1ϩ/ϩ cyclopentenone PGs (Fig. 2). Thus, cyclopentenone PGs or mice that was maximal at 1 min and sustained for 5–7 min (Fig. isoprostane release CGRP and SP, two major neuropeptides 4). 15-d-PGJ2 failed to produce any significant increase in Ϫ/Ϫ associated with pain transmission by capsaicin-sensitive nocicep- nociceptive behavior in TRPA1 mice (Fig. 4). PGA2, PGA1, tive nerve terminals in the dorsal spinal cord. and 8-iso-PGA2 (15 nmol) evoked nociceptive behaviors similar ϩ/ϩ to that of 15-d-PGJ2 in TRPA1 mice (Fig. 4). Responses were Cyclopentenone PGs Activate Spinal Nociceptive Neurons by a TRPA1- all rapid in onset and lasted Ͻ10 min. These responses were Dependent Mechanism. We examined whether cyclopentenone practically absent in TRPA1Ϫ/Ϫ mice (Fig. 4). PGs and 8-iso-PGA2 activate a nociceptive pathway in the spinal Intraplantar injection of PGD2 (15 nmol) did not evoke cord by stimulating TRPA1. PGs were administered to the detectable nociceptive behavior either in TRPA1ϩ/ϩ or Ϫ/Ϫ mouse paw and c-fos was localized in neurons of the dorsal horn TRPA1 mice (Fig. 4). PGE2 (15 nmol) produced a slowly of the spinal cord as an indicator of neuronal activation. In- developing and mild-to-moderate nociceptive behavior in ϩ/ϩ traplantar injection (20 ␮l/paw) of vehicle stimulated a modest TRPA1 mice. The response to PGE2, that was absent or increase in the number of c-fos containing neurons in the lumbar minimal during the first 4–5 min, increased with time, becoming spinal cord that was similar in TRPA1ϩ/ϩ or TRPA1Ϫ/Ϫ mice more evident after 8–10 min and was still present at 15 min. (Fig. 3B). Intraplantar injection of capsaicin (0.1 nmol) or Nociceptive behavior evoked by PGE2 was indistinguishable in cinnamaldehyde (30 nmol) produced a 2- to 3-fold increase in TRPA1ϩ/ϩ and TRPA1Ϫ/Ϫ mice (Fig. 4). The vehicle did not ϩ/ϩ c-fos expression in TRPA1 mice, respectively (Fig. 3B) [see produce nociceptive behavior. Thus, 15-d-PGJ2, PGA2, PGA1, supporting information (SI) Fig. S1]. While the effect of capsa- and 8-iso-PGA2 cause nociception by TRPA1 activation. The icin was maintained, the response to cinnamaldehyde was com- time course of cyclopentenone PGs- and isoprostane-evoked pletely absent in TRPA1Ϫ/Ϫ mice (Fig. 3B and Fig. S1). In- nociception is similar to that caused by other agents that directly traplantar injection of 15-d-PGJ2, PGA2, PGA1, and 8-iso-PGA2 activate nociceptors, such as capsaicin or cinnamaldehyde, and (15 nmol) produced a robust increase in c-fos expression in different from that evoked by agents that indirectly sensitize ϩ/ϩ Ϫ/Ϫ TRPA1 mice, but not in TRPA1 mice. Intraplantar PGE2 nociceptors, such as PGE2. (15 nmol) increased c-fos expression in a similar manner in both TRPA1ϩ/ϩ and TRPA1Ϫ/Ϫ mice (Fig. 3 A and B). Thus, cyclo- Discussion pentenone PGs and isoprostane activate a spinal nociceptive Injury and inflammation are associated with increased prosta- pathway by a mechanism that requires TRPA1, but PGE2 noid synthesis and pain hypersensitivity. Prostanoids mediate activates that pathway by a TRPA1-independent mechanism. inflammation and immune responses, as their administration reproduces the major signs of inflammation, including hyperal- Cyclopentenone PGs Induce Pain by a TRPA1-Dependent Mechanism. gesia (7–9). Prostanoids induce pain by indirect mechanisms of We investigated whether cyclopentenone PG- and isoprostane- nociceptor sensitization. Thus, prostanoids activate neuronal PG evoked nociceptor stimulation, release of sensory neuropep- receptors, mainly of the EP subtype, that couple to downstream tides, and c-fos expression in the spinal cord were associated with mediators including cAMP and protein kinases A and C (28, 29),

Materazzi et al. PNAS ͉ August 19, 2008 ͉ vol. 105 ͉ no. 33 ͉ 12047 Downloaded by guest on September 24, 2021 Fig. 3. Cyclopentenone PGs and isoprostane evoke c-fos expression in L4–L5 Fig. 4. TRPA1 mediates cyclopentenone PG- and isoprostane-induced no- spinal cord lumbar sections. Photomicrographs (A) of transverse sections of cifensive behavior in mice. Agonists were injected into the hind paws of the lumbar spinal cord and pooled data (B) showing the effects of intraplantar TRPA1ϩ/ϩ (empty circles or gray and orange columns) or TRPA1Ϫ/Ϫ (filled circles injection of cinnamaldehyde (CNM, 30 nmol) capsaicin (CPS, 0.1 nmol), cyclo- or filled and red columns) mice, and time spent licking or lifting after the pentenone prostaglandins or isoprostane (15 nmol each) on the expression of injection was recorded for 15 min. Insets show the total time spent licking or ϩ/ϩ c-fos-like immunoreactivity (LI) in TRPA1 (gray or orange columns) or lifting over the 15 min of observation. 15-d-PGJ2, PGA1, PGA2, or 8-iso-PGA2 (all TRPA1Ϫ/Ϫ (filled or red columns) mice 90 min after injection in the hind paw 15 nmol) caused an early and transient nociceptive response in TRPAϩ/ϩ mice, (4–5 animals for each treatment). Each photomicrograph is a representative an effect that was completely absent in TRPA1Ϫ/Ϫ littermates. In TRPAϩ/ϩ mice, example of one section (L4–L5) of dorsal horn ipsilateral to the injection. CNM, the TRPA1 agonist cinnamaldehyde (30 nmol) and the TRPV1 agonist capsaicin 15-d-PGJ2, PGE2, PGA1, PGA2, and 8-iso-PGA2 and CPS increased c-fos-LI in (0.1 nmol) produced a response with a time course similar to that of cyclo- ϩ/ϩ Ϫ/Ϫ TRPA1 mice. In TRPA1 mice the effects of CPS and PGE2 were maintained, pentenone compounds. Whereas the response to capsaicin was maintained, Ϫ/Ϫ whereas the effects of CNM, 15-d-PGJ2, PGE2, PGA1, PGA2 were abolished. The that to cinnamaldehyde was abolished in TRPA1 mice. PGD2 (15 nmol) did vehicle induced a moderate increase in c-fos-LI expression, identical in not evoke any measurable nociceptive behavior either in TRPA1ϩ/ϩ or in ϩ/ϩ Ϫ/Ϫ Ϫ/Ϫ TRPA1 and TRPA1 mice. Each column represents the mean Ϯ SEM of TRPA1 mice. PGE2 (15 nmol) produced a slowly developing and protracted ϩ ϩ Ϫ Ϫ c-fos immunoreactive cells; n Ն 10 slices per condition in L4–L5 segments; *, nociceptive behavior that was similar in TRPA1 / and TRPA1 / mice. Error ϩ ϩ ϩ ϩ P Ͻ 0.05 compared with TRPA1 / mice. bars indicate mean Ϯ SEM of 6–9 animals per trial; *, P Ͻ 0.05 vs. TRPA1 / mice.

which activate voltage-dependent sodium channels (30), or inhibit voltage-dependent potassium channels (31), inducing ids during oxidative stress (14, 18), directly gate TRPA1 by neuronal hypersensitivity. Prostanoids similarly sensitize other covalent binding to cysteine and lysine residues of the intracel- channels on sensory neurons to cause hyperalgesia, as is the case lular N terminus of the channel (18, 35, 36). The relevance of the for TRPV1, where PGs lowered the threshold temperature for TRPA1 channel in pain has been further highlighted by the channel activation (32). There is no evidence that PGs cause pain finding that nociception produced by formalin injection in mouse by direct stimulation of nociceptors (7–9). However, nonenzy- paw, a widely used assay for testing analgesic compounds, is matic pathways of PG degradation produce metabolites, includ- entirely due to the action of formaldehyde on TRPA1 (17, 19). ing cyclopentenone PGs, whose known biological functions are Our results show that TRPA1 mediates the pro-algesic effects not mediated by G protein-coupled prostanoid receptors. For evoked by COX-dependent (cyclopentenone PGs) and COX- example, 15-d-PGJ2, which has been extensively investigated, exerts delayed anti-inflammatory and antihyperalgesic effects independent (cyclopentenone isoprostane) fatty acid metabo- via stimulation of the PPAR-␥ receptor (33, 34). We investigated lites. We observed that the cyclopentenone compounds, 15-d- whether these metabolites cause pain by directly activating PGJ2, PGA2, PGA1 and 8-iso-PGA2, directly excited nociceptors nociceptors. of isolated DRG neurons to increase intracellular calcium, Recently, because of the presence of a highly reactive ␣,␤- resulting in the release of SP and CGRP, mediators of neuro- unsaturated bond in their structure, cyclopentenone PGs and genic inflammation and pain transmission. When injected into isoprostane have been proposed as TRPA1 agonists (21). The the paw, these PGs induced c-fos expression in spinal neurons of TRPA1 channel, expressed in TRPV1-positive somatosensory the dorsal horn, indicating activation of spinal nociceptive neurons, is targeted by metabolic irritants generated during pathways and caused a robust nociceptive response. The obser- inflammation. In this manner, bradykinin-dependent activation vation that TRPA1-deficient mice were completely unresponsive of the B2 receptor stimulates a phospholipase C␤-dependent to cyclopentenone PGs and isoprostane provides conclusive pathway that activates TRPA1 and causes pain (14). Alterna- evidence that expression of this channel is necessary and suffi- tively, bioactive aldehydes including acrolein and 4-HNE, which cient for the algesic action of cyclopentenone metabolites of originates from peroxidation of plasma membrane phospholip- arachidonic or dihomo-␥-linolenic acid.

12048 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0802354105 Materazzi et al. Downloaded by guest on September 24, 2021 In contrast to their cyclopentenone metabolites, PGD2 (15- Materials and Methods

d-PGJ2 precursor) and PGE2 (PGA2 precursor), either did not Reagents. PGD2, PGA1, PGA2, 15-d-PGJ2, and 8-iso-PGA2 were from Cayman produce any nociceptive effect (PGD2) or evoked neuronal Chemicals; PGE2, cinnamaldehyde, and capsaicin from Sigma; rabbit anti-c-fos from Santa Cruz Biotechnology; goat anti-rabbit biotinylated ABC and DAB responses and nociceptive effects (PGE2) that were completely kits from Vector Laboratories; and Fura-2-AM-ester from Alexis Biochemicals independent from TRPA1 activation. PGD2 is the most abun- dant PG released from inflammatory cells at sites of inflamma- or Invitrogen. tion (37). Although PGD can inhibit postspike hyperpolariza- 2 Animals. Institutional Animal Care and Use Committees of the University of tion of nociceptors in vagal afferents (38), we and others (39, 40) Florence and the University of California, San Francisco approved all proce- reported that it is unable to directly excite DRG or trigeminal dures. Sprague–Dawley rats (male, 250 g), Swiss mice, TRPA1Ϫ/Ϫ mice (14), and neurons. Thus, although its relevance in nociception remains per wild-type littermates (male, 30 g) were housed in a temperature- and humid- se questionable, the role of all PGD2-derived cyclopentenone ity-controlled vivarium (12 h dark/light cycle, free access to food and water). metabolites (20) should be considered in pain transmission. It is The TRPA1 heterozygote male mice had been backcrossed with C57BL/6 female mice for two generations to obtain the C57BL/6 genetic background. well established that PGE2 indirectly sensitizes nociceptors through activating prostanoid receptors that regulate ion chan- TRPA1-wild-type and -deﬁcient mice used in the present experiments derived from breeding heterozygous animals. Experiments with TRPA1-wild-type and nels and lower pain thresholds (7–9). We observed that the time -deﬁcient littermates were blinded to the genotype. Animals were killed with course of the nociceptive behavior evoked by intraplantar PGE2 sodium pentobarbital (200 mg/kg i.p.) and thoracotomy. was dissimilar from that produced by its cyclopentenone metabolite, PGA2. PGE2-evoked pain was delayed and protracted, DRG Neuron Isolation and Culture. DRG from thoracic and lumbar spinal cord suggesting indirect sensitization, whereas PGA2-evoked pain was of mice were removed and minced in cold HBSS as described in ref. 45. Ganglia immediate and transient, indicating direct nociceptor stimula- were digested using 1 mg/ml of collagenase type 1A and 1 mg/ml of papain in tion. Indeed, all cyclopentenone PGs and isoprostane evoked HBSS (25 min, 37°C). Neurons were pelleted and resuspended in Ham’s-F12 similar immediate and transient nociceptive responses, compa- containing 10% FBS, 100 U/ml of penicillin, 0.1 mg/ml of streptomycin, and 2 mM glutamine, dissociated by gentle trituration, and plated on glass cover- rable to those elicited by other direct algesic compounds, such as slips coated with poly-L-lysine (8.3 ␮M) and laminin (5 ␮M). Neurons were cinnamaldehyde and capsaicin. Thus, our results indicate that cultured for 3–4 d. the cyclopentenone PGs, derived from PGs that either sensitize or do not target nociceptors, elicit pain responses by directly Calcium Imaging. DRG neurons were loaded with Fura-2-AM (5 ␮M) in HBSS

stimulating TRPA1. containing CaCl2 1.6 mM, KCl 5.4 mM, MgSO4 0.4 mM, NaCl 135 mM, D-glucose PHARMACOLOGY An important question is whether cyclopentenone PGs or 5 mM, Hepes 10 mM, BSA 0.1%, pH 7.4 (40 min, 37°C). Cells were washed and isoprostane are produced in sufficient amount to stimulate the transferred to a chamber for recording changes by the F340/F380 ratio in TRPA1 at sites of inflammation or injury. Although formation individual neurons using a dynamic image analysis system as described in ref. 45. DRG were challenged with capsaicin (1 ␮M) and by KCl (50 mM) to identify of cyclopentenone PGs from their parent compounds has been nociceptive neurons and at the end of each experiment with ionomycin (5 demonstrated both in vitro and in vivo (34, 37, 41), the extent to ␮M). Results are expressed as the percentage of the maximum response to which cyclopentenone PGs are formed in vivo is unclear. The ionomycin. unsaturated carbonyl groups of cyclopentenone PGs rapidly react with thiol-containing proteins and peptides, and measure- Neuropeptide Release. Slices (0.4 mm) of the lumbar enlargements of the dorsal spinal cord from rats were stabilized for 60 min in an oxygenated Krebs ments of 15-d-PGJ2 (or other cyclopentenone PGs) in vivo underestimates true concentrations. Thus, 15-d-PGJ2 is detect- solution at 37°C. Fractions (4 ml) were collected at 10-min intervals into able in vivo primarily as a Michael conjugate to gluthatione (42). ethanoic acid (final concentration 2 N) before, during, and after administration of cyclopentenone PGs or isoprostane. Fractions were freeze-dried, re- The same consideration applies to the measurement of 8-iso- constituted with assay buffer, and analyzed for CGRP and SP immunoreactiv- PGA2 levels at sites of inflammation or tissue injury (22). Thus, ities as described in ref. 45. Neither cyclopentenone PGs nor cyclopentenone more appropriate methods are required to assess actual levels of isoprostane (30 ␮M) cross-reacted with SP or CGRP antisera. Exposure to ␮ PGA1, PGA2, 15-d-PGJ2, and 8-iso-PGA2 adducts and thereby capsaicin (10 M, 20 min) was used to produce a specific and complete assess their contribution to inflammatory pain. desensitization of TRPV1-expressing sensory nerve terminals. Further details NSAIDs and Coxibs are considered to produce analgesia via are reported in a SI Text section. their anti-inflammatory action because of inhibition of PG C-fos Localization. The following agents were injected into the plantar surface generation. Our results extend the mechanism of the antinoci- of one hind paw (20 ␮l/paw) of mice: cyclopentenone PGs or isoprostane (15 ceptive action of NSAIDs/Coxibs from preventing indirect no- nmol), cinnamaldehyde (30 nmol), capsaicin (0.1 nmol) or vehicle (7.5% DMSO ciceptor sensitization by classical PGs, to inhibiting direct noci- in PBS). After 90 min, animals were transcardially perfused with 0.1 M of PBS ceptive activation by cyclopentenone PGs and TRPA1. The followed by 4% paraformaldehyde in PBS. Spinal segments (L4–L5) were beneficial actions of NSAIDs and Coxibs are offset by their well immersion fixed in paraformaldehyde overnight, and 20-␮m coronal sections known and severe adverse effects in the gastrointestinal (7–9) prepared using a freezing microtome. Serial sections were incubated with and cardiovascular systems (43). TRPA1-mediated pain is also c-fos antibody (1:5000, overnight, 4°C), washed, incubated with goat anti- induced by other irritant COX-independent fatty acid metabo- rabbit biotinylated antibody (1:500, 1 h, room temperature), and stained using avidin-biotin peroxidase complex and chromogen nickel/3.3Ј- lites, such as 8-iso-PGA2 or 4-HNE (18, 22, 23, 44). Should these diaminobenzidine solution. Ten sections for each mouse were randomly compounds play a significant role in inflammatory pain, their selected and the expression of c-fos immunoreactivity was evaluated by action would be clearly unaffected by NSAIDs or Coxib, whereas counting the number of positive neurons in the superficial dorsal horn region it could be blocked by TRPA1 antagonists. Thus, TRPA1 (lamina I–II). inhibitors could represent a novel therapeutic strategy for re- ducing inflammatory pain, with consistent advantages at the Nocifensive Responses. Agonists were injected into the mouse paw as de- level of both safety and efficacy. Compared with NSAIDs and scribed (46, 47). Immediately after injection, mice were placed in a Plexiglas chamber. The total time spent licking and lifting of the injected hind paw was Coxibs, their use should not be associated with adverse effects recorded for 15 min. Further details are reported in a SI Text section. in the gastrointestinal (7–9) and cardiovascular systems (43), where PG biosynthesis would be preserved. Moreover, TRPA1 Statistical Analysis. Data are expressed as mean Ϯ SEM. Statistical significance antagonists could reduce pain evoked by COX-independent was determined using Student’s t test for unpaired data or the two-way nociceptive fatty acid metabolites. ANOVA, followed by Bonferroni’s post hoc analysis. P Ͻ 0.05 was considered

Materazzi et al. PNAS ͉ August 19, 2008 ͉ vol. 105 ͉ no. 33 ͉ 12049 Downloaded by guest on September 24, 2021 signiﬁcant. Agonist potency was expressed as the molar concentration pro- obtained in HEK293 cells expressing the human TRPA1 and mouse DRG ducing 50% of the maximal effect (EC50). neurons, respectively.

NOTE. Contemporary to the submission of the present study a paper has been ACKNOWLEDGMENTS. We thank David Julius (University of California San published (48) reporting that reactive oxygen species and 15-d-PGJ2 stimu- Francisco, San Francisco, CA) for providing TRPA1-deficient mice. This work lated TRPA1 and that the nociceptive behavior evoked by 15-d-PGJ2 was was supported by grants from Ente Cassa di Risparmio di Firenze (Florence); absent in TRPA1-deficient mice. However, this study failed to detect a calcium Ministry for University and Scientific Research (MiUR) Rome, Italy Grants FIRB response by PGA2 in Chinese hamster ovary cells expressing the mouse TRPA1. RBIP06YM29 and PRIN 2006069824࿝004 (P.G. and R.P.); and National Institutes This observation is at variance with previous (21) and present (Fig. 1) data of Health Grants DK 57840, DK 39957, and DK 43207 (N.W.B.).

1. Kirtland SJ (1988) Prostaglandin E1: A review. Prostaglandins Leukot Essent Fatty Acids 27. Szolcsanyi J (1977) A pharmacological approach to elucidation of the role of different 32:165–174. nerve fibres and receptor endings in mediation of pain. J Physiol (Paris) 73:251–259. 2. Vane JR (1971) Inhibition of prostaglandin synthesis as a mechanism of action for 28. Aley KO, Levine JD (1999) Role of protein kinase A in the maintenance of inflammatory aspirin-like drugs. Nat New Biol 231:232–235. pain. J Neurosci 19:2181–2186. 3. Bos CL, Richel DJ, Ritsema T, Peppelenbosch MP, Versteeg HH (2004) Prostanoids and 29. Khasar SG, et al. (1999) A novel nociceptor signaling pathway revealed in protein prostanoid receptors in signal transduction. Int J Biochem Cell Biol 36:1187–1205. kinase C epsilon mutant mice. Neuron 24:253–260. 4. Narumiya S, Sugimoto Y, Ushikubi F (1999) Prostanoid receptors: Structures, proper- 30. Gold MS, Reichling DB, Shuster MJ, Levine JD (1996) Hyperalgesic agents increase a ties, and functions. Physiol Rev 79:1193–1226. tetrodotoxin-resistant Naϩ current in nociceptors. Proc Natl Acad Sci USA 93:1108– 5. Ferreira SH (1972) Prostaglandins, aspirin-like drugs and analgesia. Nat New Biol 1112. 240:200–203. 31. Evans AR, Vasko MR, Nicol GD (1999) The cAMP transduction cascade mediates the 6. Ferreira SH, Nakamura M, de Abreu Castro MS (1978) The hyperalgesic effects of PGE2-induced inhibition of potassium currents in rat sensory neurones. J Physiol prostacyclin and prostaglandin E2. Prostaglandins 16:31–37. 7. Tilley SL, Coffman TM, Koller BH (2001) Mixed messages: Modulation of inflammation 516:163–178. and immune responses by prostaglandins and thromboxanes. J Clin Invest 108:15–23. 32. Moriyama T, et al. (2005) Sensitization of TRPV1 by EP1 and IP reveals peripheral 8. Zeilhofer HU (2007) Prostanoids in nociception and pain. Biochem Pharmacol 73:165– nociceptive mechanism of prostaglandins. Mol Pain 1:3. 174. 33. Ricote M, Li AC, Willson TM, Kelly CJ, Glass CK (1998) The peroxisome proliferator- 9. Samad TA, Sapirstein A, Woolf CJ (2002) Prostanoids and pain: Unraveling mechanisms activated receptor-gamma is a negative regulator of macrophage activation. Nature and revealing therapeutic targets. Trends Mol Med 8:390–396. 391:79–82. 10. Jaquemar D, Schenker T, Trueb B (1999) An ankyrin-like protein with transmembrane 34. Gilroy DW, et al. (1999) Inducible cyclooxygenase may have anti-inflammatory prop- domains is specifically lost after oncogenic transformation of human fibroblasts. J Biol erties. Nat Med 5:698–701. Chem 274:7325–7333. 35. Hinman A, Chuang HH, Bautista DM, Julius D (2006) TRP channel activation by revers- 11. Nilius B, Owsianik G, Voets T, Peters JA (2007) Transient receptor potential cation ible covalent modification. Proc Natl Acad Sci USA 103:19564–19568. channels in disease. Physiol Rev 87:165–217. 36. Macpherson LJ, et al. (2007) Noxious compounds activate TRPA1 ion channels through 12. Caterina MJ, et al. (1997) The capsaicin receptor: A heat-activated ion channel in the covalent modification of cysteines. Nature 445:541–545. pain pathway. Nature 389:816–824. 37. Shibata T, et al. (2002) 15-deoxy-delta 12,14-prostaglandin J2. A prostaglandin D2 13. Story GM, et al. (2003) ANKTM1, a TRP-like channel expressed in nociceptive neurons, metabolite generated during inflammatory processes. J Biol Chem 277:10459–10466. is activated by cold temperatures. Cell 112:819–829. 38. Greene R, Fowler J, MacGlashan D, Jr., Weinreich D (1988) IgE-challenged human lung 14. Bautista DM, et al. (2006) TRPA1 mediates the inflammatory actions of environmental mast cells excite vagal sensory neurons in vitro. J Appl Physiol 64:2249–2253. irritants and proalgesic agents. Cell 124:1269–1282. 39. Hwang SW, et al. (2000) Direct activation of capsaicin receptors by products of 15. Geppetti P, Holzer P (1996) Neurogenic Inflammation (CRC Press, Boca Raton). lipoxygenases: Endogenous capsaicin-like substances. Proc Natl Acad Sci USA 97:6155– 16. Jordt SE, et al. (2004) Mustard oils and cannabinoids excite sensory nerve fibres through 6160. the TRP channel ANKTM1. Nature 427:260–265. 17. McNamara CR, et al. (2007) TRPA1 mediates formalin-induced pain. Proc Natl Acad Sci 40. Zhang XC, Strassman AM, Burstein R, Levy D (2007) Sensitization and activation of USA 104:13525–13530. intracranial meningeal nociceptors by mast cell mediators. J Pharmacol Exp Ther 18. Trevisani M, et al. (2007) 4-Hydroxynonenal, an endogenous aldehyde, causes pain and 322:806–812. neurogenic inflammation through activation of the irritant receptor TRPA1. Proc Natl 41. Kikawa Y, Narumiya S, Fukushima M, Wakatsuka H, Hayaishi O (1984) 9-Deoxy-delta 9, Acad Sci USA 104:13519–13524. delta 12–13,14-dihydroprostaglandin D2, a metabolite of prostaglandin D2 formed in 19. Macpherson LJ, et al. (2007) An ion channel essential for sensing chemical damage. human plasma. Proc Natl Acad Sci USA 81:1317–1321. J Neurosci 27:11412–11415. 42. Brunoldi EM, et al. (2007) Cyclopentenone prostaglandin, 15-deoxy-Delta12,14-PGJ2, 20. Straus DS, Glass CK (2001) Cyclopentenone prostaglandins: New insights on biological is metabolized by HepG2 cells via conjugation with glutathione. Chem Res Toxicol activities and cellular targets. Med Res Rev 21:185–210. 20:1528–1535. 21. Taylor-Clark TE, et al. (2008) Prostaglandin-induced activation of nociceptive neurons 43. Spektor G, Fuster V (2005) Drug insight: Cyclo-oxygenase 2 inhibitors and cardiovas- via direct interaction with transient receptor potential A1 (TRPA1). Mol Pharmacol cular risk–where are we now? Nat Clin Pract Cardiovasc Med 2:290–300. 73:274–281. 44. Benedetti A, Comporti M, Esterbauer H (1980) Identification of 4-hydroxynonenal as a 22. Milne GL, Musiek ES, Morrow JD (2005) The cyclopentenone (A2/J2) isoprostanes– cytotoxic product originating from the peroxidation of liver microsomal lipids. Biochim unique, highly reactive products of arachidonate peroxidation. Antioxid Redox Signal Biophys Acta 620:281–296. 7:210–220. 45. Trevisani M, et al. (2002) Ethanol elicits and potentiates nociceptor responses via the 23. Chen Y, Zackert WE, Roberts LJ, 2nd, Morrow JD (1999) Evidence for the formation of vanilloid receptor-1. Nat Neurosci 5:546–551. a novel cyclopentenone isoprostane, 15-A2t-isoprostane (8-iso-prostaglandin A2) in 46. Sakurada T, Katsumata K, Tan-No K, Sakurada S, Kisara K (1992) The capsaicin test in vivo. Biochim Biophys Acta 1436:550–556. mice for evaluating tachykinin antagonists in the spinal cord. Neuropharmacology 24. Nagata K, Duggan A, Kumar G, Garcia-Anoveros J (2005) Nociceptor and hair cell transducer properties of TRPA1, a channel for pain and hearing. J Neurosci 25:4052– 31:1279–1285. 4061. 47. Ferreira J, da Silva GL, Calixto JB (2004) Contribution of vanilloid receptors to the overt 25. Bandell M, et al. (2004) Noxious cold ion channel TRPA1 is activated by pungent nociception induced by B2 kinin receptor activation in mice. Br J Pharmacol 141:787– compounds and bradykinin. Neuron 41:849–857. 794. 26. Szallasi A, Blumberg PM (1999) Vanilloid (Capsaicin) receptors and mechanisms. Phar- 48. Andersson DA, Gentry C, Moss S, Bevan S (2008) Transient receptor potential A1 is a macol Rev 51:159–212. sensory receptor for multiple products of oxidative stress. J Neurosci 28:2485–2494.

12050 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0802354105 Materazzi et al. Downloaded by guest on September 24, 2021