Inflammatory pain control by blocking oxidized phospholipid-mediated TRP channel activation

Phospholipids occurring in cell membranes and lipoproteins are converted into oxidized phospholipids (OxPL) by oxidative stress promoting atherosclerotic plaque formation. Here, OxPL were characterized as novel targets in acute and chronic inflammatory pain. Oxidized 1-palmitoyl-2-arachidonoyl-sn-glycero-3-phosphocholine (OxPAPC) and its derivatives were identified in inflamed tissue by mass spectrometry and binding assays. They elicited calcium influx, hyperalgesia and induced pro-nociceptive peptide release. Genetic, pharmacological and mass spectrometric evidence in vivo as well as in vitro confirmed the role of transient receptor potential channels (TRPA1 and TRPV1) as OxPAPC targets. Treatment with the monoclonal antibody E06 or with apolipoprotein A-I mimetic peptide D-4F, capturing OxPAPC in atherosclerosis, prevented inflammatory hyperalgesia, and in vitro TRPA1 activation. Administration of D-4F or E06 to rats profoundly ameliorated mechanical hyperalgesia and inflammation in collagen-induced arthritis. These data reveal a clinically relevant role for OxPAPC in inflammation offering therapy for acute and chronic inflammatory pain treatment by scavenging OxPAPC.


Results
Thermal and mechanical hyperalgesia by local OxPAPC injection in rats. As OxPAPC is generated by ROS in the inflamed tissue in other diseases we evaluated whether OxPAPC elicits hyperalgesia when injected in the rat paw. Intraplantar injection of OxPAPC dose-dependently lowered mechanical thresholds for up to 6 h with a maximal effect at 1-3 h at the dose of 500 µg (Fig. 1a). Thermal thresholds declined from 3 to 6 h after injection (Fig. 1b). Mechanical or thermal nociceptive thresholds remained unchanged in uninjected contralateral paws ( Supplementary Fig. 1a,b), after local application of non-oxidized PAPC (500 µg; Fig. 1c,d, Supplementary  Fig. 1c,d) or solvent control (0.9% NaCl).

Detection of OxPAPC in CFA-induced inflammation.
To evaluate a possible contribution of endogenous OxPAPC to inflammatory hyperalgesia, we examined the presence of OxPAPC in inflamed tissue compared to saline-injected or untreated controls. We used a model of localized inflammation by injection of CFA into rat's hindpaws. Inflammation including swelling, leukocyte invasion, mechanical and thermal hyperalgesia develops within 2 h (see below; 23 ).
For quantitation of OxPAPC in vivo we established a competitive binding assay using the E06 antibody (Fig. 1f). We measured concentrations after local treatment of CFA in rats' hindpaws compared to rats treated with OxPAPC (500 µg/paw) as well as in untreated controls. In CFA-induced inflammation, the paw tissue contained three to five times more E06-binding OxPAPC compounds than in untreated controls (Fig. 1f).
OxPAPC as an activator of TRPA1 in vitro. To characterize the impact of OxPAPC on different TRP channel family members, we performed calcium imaging-based screening of recombinant TRP channels expressed in HEK-293 cells in a fluorescence imaging plate reader assay ( Fig. 3a-f). Characteristic agonists for each TRP channel served as a reference. OxPAPC was applied at concentrations ranging from 0.05-100 µM. OxPAPC-induced calcium influx was prominent in hTRPA1-transfected HEK-293 cells already at concentrations of ≥6 µM (Fig. 3a,g). In this assay, OxPAPC neither activated recombinant TRPV2, TRPV3, or TRPV4 nor untransfected HEK-293 cells, but weakly stimulated recombinantly expressed TRPV1 at higher OxPAPC concentrations (Fig. 3b,g). Next, we created a HEK-293 cell line stably expressing hTRPA1 (HEK-293 TRPA1 ) and performed single cell ratiometric calcium measurements to investigate the characteristics of TRPA1 activation by OxPAPC ( Fig. 4a-g). In HEK-293 TRPA1 OxPAPC evoked a robust and sustained increase in intracellular calcium. However, in comparison to TRPA1-mediated calcium influxes evoked by the TRPA1 agonist AITC, OxPAPC-induced calcium influx was temporally delayed at equimolar concentrations, in particular in a subpopulation of cells (Fig. 4a,b). Acute and delayed activation kinetics could best be distinguished in single traces of OxPAPC responses. In addition, the number of cells responding to an OxPAPC stimulus was lower in comparison to AITC stimulation (Fig. 4c). Calcium amplitudes (ratio F(340/380)) of cells with a pronounced OxPAPC-induced response were comparable with AITC-induced responses, however, as less cells were reactive in case of OxPAPC stimulation, the overall potency of OxPAPC was lower ( Fig. 4a-d). Specific components of OxPAPC, such as PGPC and POVPC, also activated TRPA1-induced calcium influx. In comparison to OxPAPC, the number of responding cells and the area under curve were slightly increased when all reagents were compared at equimolar concentrations (Fig. 4c,d). The TRPA1 inhibitor HC-030031 (Fig. 4a,b) immediately blocked both, OxPAPC-and AITC-induced activation of TRPA1. Neither OxPAPC nor AITC stimulated control HEK-293 cells ( Supplementary Fig. 5a).
The effect induced by OxPAPC suggests the potential involvement of ROS generation. Ascorbic acid (vitamin C), an antioxidant, only modestly but not significantly reduced the OxPAPC-induced activation of HEK-293 TRPA1 ( Supplementary Fig. 6a,b). In HEK-293 TRPA1 cells loaded with dihydroethidium 24 , an oxidation-sensitive dye, the integrated fluorescence intensity rose significantly after treatment with carbonyl cyanide-4-(trifluoromethoxy) phenylhydrazone (FCCP, positive control) 25 . FCCP elicited intracellular ROS production by cellular stress due to uncoupling the ATP synthesis. After treating the cells with OxPAPC no changes were detectable during 30 min incubation time ( Supplementary Fig. 6c).
TRPA1 is a polychemoreceptor 21 and carries three N-terminal intracellular cysteines responsible for electrophilic binding of some TRPA1 agonists such as AITC 26 . To prove that these intracellular residues are involved in OxPAPC-induced effects via TRPA1, a TRPA1 mutant was investigated. In this TRPA1 mutant (hTRPA1-3C) three critical cysteines were replaced by serine (C621S/C641S/C665S) 27 . Indeed, OxPAPC and AITC failed to activate hTRPA1-3C. In contrast, carvacrol, a TRPA1 agonist, which acts through a non-electrophilic binding site in the TRPA1 receptor, still caused calcium influx, thus verifying that the TRPA1 ion channel function was intact ( Fig. 4e-g). However, as the OxPAPC compounds PGPC and POVPC are not electrophilic 4 , this experiment also suggests that other mechanisms than electrophilic interaction with these critical cysteine residues contributes to TRPA1 activation by OxPL 26,28,29 .
To further assess the mechanism by which OxPAPC activates TRPA1 we used a model peptide of the intracellular N-terminal part of hTRPA1 (amino acids 607-670) 30,31 , which includes three cysteine residues responsible for hTRPA1 activation by noxious stimuli 32 . No direct reaction products between OxPAPC and TRPA1 peptide were detected. Instead a dose-dependent oxidation was observed (Fig. 4h,i), with lower dose leading to formation of one disulfide bond (noticeable by leftward shift of deconvoluted mass spectrum by m/z = 2; Fig. 4i). Higher doses led to a more pronounced oxidation, with formation of two disulfide bonds (leftward shift of deconvoluted mass spectrum by m/z = 4; Fig. 4j). This indicates that oxidation of individual OxPAPC compounds with the intracellular N-terminal domain of TRPA1 might contribute to TRPA1 activation. However, for instance PGPC is an end-stage oxidation product with TRPA1 activation activity, indicating that more mechanisms than electrophilic interaction or TRPA1 oxidation contribute to OxPAPC function. OxPAPC elicited a modest and delayed activation of recombinant TRPV1. In order to attain this objective, higher concentrations of OxPAPC (30 µM instead of 10 µM) were needed. The calcium-influx through TRPV1 was blocked by BCTC ( Fig. 4k-n). The number of activated HEK-293 TRPV1 cells and the AUC of OxPAPC (10 µM) compared to capsaicin-activated HEK-293 TRPV1 was significantly lower (Fig. 4m,n). A substantial proportion of HEK-293 TRPV1 cells could not be activated even with a higher concentration of OxPAPC, as well as PGPC and POVPC (Fig. 4m,n). In summary, in vitro reconstitution showed that OxPAPC activates TRPA1 (see also 33 ) and to a lesser extend TRPV1.
OxPAPC evokes currents in HEK-293 TRPA1 and HEK-293 TRPV1 . In whole cell patch clamp experiments ( Fig. 5) HEK-293 TRPA1 and HEK-293 TRPV1 cells as well as the parental control cell line were stimulated with OxPAPC (1, 10 and/or 30 µM). Measurements on HEK-293 TRPA1 were performed in calcium-free extracellular bath solution to prevent calcium-induced desensitization of TRPA1 34 . In addition, 50% of sodium ions were replaced by N-methyl-D-glucamine to limit OxPAPC-evoked inward currents of TRPA1 35 . OxPAPC evoked an activation of large TRPA1 currents in HEK-293 TRPA1 cells (Fig. 5a,b,e), which were absent in the presence of HC-030031 (10 µM; Fig. 5c,d) and in the parental control cell line ( Fig. 5l-n). Application of voltage ramps after addition of OxPAPC revealed a linear current voltage relationship (Fig. 5b). Furthermore, TRPA1-related currents were recorded after stimulation with 1 µM and 10 µM OxPAPC as well as PGPC, POVPC and AITC (10 µM each; Fig. 5e). Here, agonists were applied to the cell by a stimulation pipette. For comparison, current/time courses and current/voltage ramps were measured in HEK-293 TRPA1 cells stimulated with AITC (10 µM; Fig. 5f,g). Current densities were calculated by taking the peak currents 2-10 min upon stimulation with OxPAPC, OxPAPC in the presence of HC-030031 or AITC (Fig. 5h). Analysis of the onset of TRPA1-mediated currents evoked by OxPAPC revealed a smaller current within the first 2 min compared to AITC (Fig. 5i) confirming a delayed activation kinetic of TRPA1 by OxPAPC (Fig. 5e). In HEK-293 TRPV1 only a non-significant, delayed shift of the current could be induced by 3-fold higher concentrations of OxPAPC compared to capsaicin (Fig. 5j,k). In parental control cells, no current was observed after stimulation with either 10 µM (Fig. 5l) or 30 µM OxPAPC (Fig. 5m,n). Experiments on HEK-293 TRPV1 and parental cells were performed in buffer containing 100% sodium chloride and calcium chloride. It has been shown that cytosolic inorganic polyphosphates support the activation of TRPA1 by pungent chemicals such as AITC. For this reason, we added polytriphosphate (PPP i ) to the intracellular solution and performed whole-cell patch clamp recording of HEK-293 TRPA1 cells. Local application of AITC, OxPAPC, or PGPC induced long-lasting and stable inward currents for up to 3 min ( Supplementary Fig. 5b). Intracellular calcium increase in native DRG neurons by OxPAPC. Next, we wanted to determine whether OxPAPC was also able to stimulate native TRPA1 in cultured sensory neurons. For this, dissociated DRG   neurons of adult C57BL/6 mice and Wistar rats were used for calcium imaging experiments. Stimulation of DRG neurons with OxPAPC, AITC or capsaicin increased intracellular calcium levels ( Fig. 6, Supplementary Fig. 7b). Mouse and rat neurons activated by OxPAPC also responded to AITC and capsaicin, while only few neurons were solely responsive to AITC or capsaicin ( Fig. 6a-d, Supplementary Fig. 7a,b). When DRG neurons were first stimulated with AITC, OxPAPC failed to boost an increase in cytosolic calcium ion concentrations, suggesting desensitization of TRPA1 after AITC stimulation ( Supplementary Fig. 7a). Pre-incubation of DRG neurons with the TRPA1 antagonist HC-030031 completely blocked the OxPAPC-induced calcium influx (Fig. 6e) while pre-incubation with the TRPV1 antagonist BCTC only diminished OxPAPC-induced activation (Fig. 6f). The number of responding cells and the AUC were significantly reduced by HC-030031, both by 25%. In contrast, BCTC rather increased the number of reacting cells and did not reduce the AUC (Fig. 6g,h). These data indicate that TRPA1 is the preferential direct target of OxPAPC in cultured adult DRG neurons of mice and rats.
E06, a monoclonal antibody, interferes with OxPAPC function in vitro and in vivo. E06 mAb, a monoclonal IgM antibody, originally isolated from apolipoprotein-deficient mice 18 , is known to bind and neutralize OxPAPC 36 . Therefore, we asked whether E06 mAb functions as an inhibitor of OxPAPC-induced hyperalgesia in vivo and whether the antibody also interferes with OxPAPC-induced activation of TRPA1 in vitro. Intraplantar injection of OxPAPC evoked mechanical and thermal hyperalgesia (Fig. 7a,b). Hyperalgesia could be reduced by preincubation of OxPAPC with E06 mAb for 30 min before injection of the mixture into rats' hind paws. For control, IgM isotype antibodies did not alter nociceptive thresholds. Mechanical hyperalgesia induced by injection of CFA was abrogated by co-injection of CFA with E06 mAb but not with the IgM isotype control antibody (Fig. 7c). In addition, E06 mAb reversed mechanical hyperalgesia, when injected 3 h after induction of CFA-evoked inflammation (Fig. 7e). However, E06 mAb treatment did not alter thermal CFA-induced hyperalgesia ( Fig. 7d). No change in mechanical or thermal nociceptive thresholds was observed after treatment with E06 mAb per se without CFA or OxPAPC ( Supplementary Fig. 8). Extracted lipids from the rats' paw tissues were analyzed by MALDI-TOF MS, 3 h after CFA treatment, in presence or absence of either E06 mAb or control IgM antibodies (Fig. 7f). All samples contained non-oxidized phosphatidylcholines including PAPC metabolites. Characteristic oxidation products of PAPC (in particular m/z 496.3, the H + adduct of lysophosphatidylcholine (LPC) 16:0) were not detectable in E06-treated animals (Fig. 7f), but remained after treatment with control IgM or in inflamed paws without treatment. In collagen-induced arthritis, a single intraplantar injection of E06 mAb on day 13 reversed mechanical hyperalgesia in contrast to IgM isotype treatment ( Fig. 7g) with no change in paw edema ( Supplementary Fig. 8c).
In vitro, stimulation of HEK-293 TRPA1 with a mixture of OxPAPC and E06 mAb significantly diminished the intracellular increase in calcium. E06 mAb reduced the number of OxPAPC sensitive cells by 75% and reduced the AUC by 85%. IgM isotype control Ab did not change OxPAPC-evoked calcium influx via TRPA1 (Fig. 8a-c,g,h). In contrast, AITC-induced activation of TRPA1 was marginally reduced by E06 mAb and no significant changes in the number of responding cells and the AUC were detected when compared with the IgM isotype control Ab ( Fig. 8d-f,i,j).
In mouse DRG neurons, preincubation with E06 mAb, but not the IgM control antibody, reduced the number of OxPAPC responsive neurons by 48%, while the AUC of the remaining reacting neurons was not significantly altered by either treatment (Fig. 8k-n). Similar results were obtained in rat DRG neurons ( Supplementary  Fig. 7d). E06 mAb did not block AITC or capsaicin-induced DRG activation. Overall, E06 mAb, but not the control IgM antibody, reduced OxPAPC-induced Ca 2+ influx in HEK-293 TRPA1 cells and cultured DRG neurons.

Blockade of OxPAPC-induced TRPA1 activation by ApoA-I mimetic peptide D-4F in vitro and
in vivo. D-4F, a peptide deduced from the ApoA-I protein, is known to bind OxPAPC and to moderate the development of atherosclerotic plaque lesions 37,38 . For this reason, we tested whether D-4F could reverse OxPAPC-induced hyperalgesia. In rats, D-4F prevented OxPAPC-induced mechanical and thermal hyperalgesia either after local intraplantar co-injection or following a five day period of systemic intraperitoneal pretreatment (Fig. 9a,b). No alterations in mechanical and thermal nociceptive thresholds were seen following intraplantar injection of D-4F alone ( Supplementary Fig. 9a,b) or in contralateral paws ( Supplementary Fig. 9c,d).
In vitro, pre-treatment of HEK-293 TRPA1 with D-4F and subsequent stimulation with OxPAPC or with a mixture of OxPAPC and D-4F (preincubated for 30 min), both, lowered the efficiency of OxPAPC to induce TRPA1-mediated calcium influx ( Fig. 9c-g). With either treatment, the number of responding cells decreased by 70% and the AUC by 80%. No change in the number of reacting cells or the area under the curve (AUC) was observed when TRPA1-expressing cells were preincubated with D-4F and activated with AITC ( Fig. 9h-k).
D-4F also reduced the OxPAPC-induced stimulation of mouse DRG neurons ( Fig. 9l-n). Indeed, D-4F significantly diminished the number of responding cells by 64% and the AUC by 62% when DRG neurons were treated with a mixture of OxPAPC and D-4F. No change was seen in AITC-induced Ca 2+ -influx, when D-4F was present. Data were confirmed in rat DRG neurons ( Supplementary Fig. 7e).

D-4F for treatment of inflammatory and arthritis hyperalgesia. The therapeutic potential of D-4F
for treatment of inflammatory hyperalgesia was assessed in models of acute and chronic inflammation, meaning CFA-induced hindpaw inflammation 23 and collagen-induced arthritis 39,40 . Mechanical hyperalgesia due to CFA was significantly reduced for up to 3 h, when rats were locally treated with a single shot of D-4F (Fig. 10a). No change was observed in thermal hyperalgesia ( Supplementary Fig. 9e). In MALDI-TOF MS, PAPC oxidation fragments (i.e. particularly the LPC peak at m/z 496.3) were not detectable in paw tissue taken 3 h after injection of CFA plus D-4F compared to CFA alone (Fig. 10b). Peaks of non-oxidized phospholipids (m/z about 750-840) remained unaffected.
Scientific RepoRts | 7: 5447 | DOI:10.1038/s41598-017-05348-3 Repeated daily intraperitoneal injections of D-4F significantly reduced the development of CFA-induced paw edema by 40% from day 2 to day 5 after CFA injection ( Supplementary Fig. 10a). Similarly, daily treatments lead to a sustained amelioration of mechanical hyperalgesia when paw pressure thresholds were measured before each daily injection of D-4F (Fig. 10c). On top of this, each daily intraperitoneal D-4F injection completely reversed mechanical hyperalgesia with a maximum effect after 15 min (Fig. 10d, day 1-3: Supplementary Fig. 10b-d). The antinociceptive effect of D-4F persisted for 2 d after the last administration on day 4 after CFA injection (Fig. 10e).
Finally, we assessed whether D-4F was also effective as treatment of chronic inflammatory pain in collagen-induced arthritis (Fig. 10f-h). Immunization of adult female Lewis rats with collagen, emulsified in CFA, induced mechanical hyperalgesia, which developed within one week (Fig. 10f). Thermal hyperalgesia was observed after 10 d (Fig. 10g) as previously described 39,40 . Hyperalgesia preceded the onset of inflammation as paw thickness (Fig. 10h) and clinical signs of arthritis ( Supplementary Fig. 10e) developed 12-14 d after collagen immunization. Daily treatment with D-4F from day 6 to day 15 after immunization, reduced paw edema significantly and clinical scores non-significantly ( Fig. 10h and Supplementary Fig. 10e). Importantly, development of both mechanical and thermal hyperalgesia was prevented by D-4F treatment (Fig. 10f,g). Thus, D-4F has potent

Discussion
Our results show experimental evidence that antagonism of the action of OxPL is a potent strategy to target inflammatory pain in rodents. OxPAPC is locally formed in inflamed tissue, and it is a potent pro-nociceptive mediator, which releases the proinflammatory neuropeptide CGRP. OxPAPC species including POVPC and PGPC act predominantly via TRPA1 ion channels, a sensor for chemical stimuli on primary afferent nociceptors. Most relevantly, the pro-nociceptive function of OxPL in acute and chronic inflammation can be effectively blocked either locally by a monoclonal antibody (E06 mAb) or locally and systemically by the ApoA-I mimetic peptide D-4F. These results demonstrate in vivo function of OxPAPC as pro-nociceptive mediator and two potent therapeutic strategies, which interfere with OxPAPC-induced pain.
Phospholipids are major constituents of cellular membranes and lipoproteins. During inflammation phagocytes generate ROS 3 . ROS oxidize phospholipids leading to augmented formation of OxPAPC. OxPAPC is a complex mixture of different oxidized phospholipids known to be formed under certain pathophysiological situations such as atherosclerosis, acute lung injury and sepsis as well as neurodegenerative diseases such as Alzheimer's and Parkinson's disease 10,11 . Here, OxPAPC (100-500 µg ∼ 125-625 nmol) was used in a dose comparable to other in vivo treatments, e.g. LPS-induced peritonitis 41 or the air pouch model 42 . Several lipid TRP  agonists have previously been used in comparable doses, e.g. hydroxynoneal (150 nmol) 43 , HODE derivatives (100 µg ∼ 687 nmol) 22 or 15d-PGJ 2 (20 µmol) 44 . In vitro, OxPAPC concentrations range from 10 to 100 µg/ml (12.5-125 µmol/ml) 12, 45-48 , which were the basis of our dose selection. The commercially obtained OxPAPC contained PAPC, POVPC, PGPC and PEIPC. POVPC and PGPC both simulate TRPA1 and likely elicit hyperalgesia. 15d-PGJ 2 , a remote metabolite of OxPAPC, also activates TRPA1. Hence, all of these OxPL species could probably play a role in inflammatory pain. OxPAPC metabolites are not stable for long periods since they rapidly react with other molecules in vivo. Typically, some OxPAPC components such as the aldehyde POVPC covalently form Schiff 's bases or Michael adducts with proteins and, therefore, they are only transient pathophysiological mediators 49,50 . In vitro data indicate that enzymes are necessary to metabolize OxPAPC since the composition of OxPAPC in ambient air for up to 5 h barely changes. However, non-oxidized PAPC is highly reactive on ambient air as observed before 12 .
Quantification of the OxPAPC species in vivo after injection of either OxPAPC or CFA was accomplished by an E06-based competitive binding assay 51,52 . We observed higher amounts of E06-immunoreactivity in CFA-inflammation compared to E06-immunoreactivity after local injection of 500 µg OxPAPC (approx. 230 µg/g wet tissue). For comparison, 40-100 µg/g wet tissue OxPAPC species were detected in rabbit aorta 7, 52 and 5-20 ng/ml POVPC/10-30 ng/ml PGPC in human atheroma plaques 53 . The validity of the competitive binding assay is limited by the specificity of E06 binding to phosphocholine head group of oxidized phospholipids. E06 does not bind to LPC 36 . This suggests that LPC, a hydrolysis end-product of OxPAPC, cannot be the major mediator of the E06-sensitive nociceptive behaviour.
We detected POVPC and PGPC after immediate retrieving tissue by MALDI-TOF while OxPAPC metabolites like LPC were found regularly 15 min after OxPAPC injection. Absolute quantification of individual OxPAPC species by mass spectrometry techniques depends on the use of internal standards for which defined, deuterated oxidized phospholipids would be preferable 54,55 . Albeit some oxidized phospholipids with a lower reactivity, such as PGPC or POVPC, are available as pure substances and can be used for comparative or semi-quantitative analysis 56,57 , absolute quantification of our phospholipid samples was technically not yet possible. This was due to the lack of availability of deuterated oxidized phospholipids, or techniques which allow the absolute quantification of individual phospholipid compounds in complex lipid preparations 54,55 . Rapid metabolism has been observed for other lipids in vivo before 58 . In CFA inflammation, OxPAPC species, e.g. POVPC, were endogenously produced in the inflammatory tissue, but peaks differed between animals. Several aspects might explain these findings: (1) increased and continuous ROS production and thereby OxPAPC formation in the course of inflammation, (2) high reactivity of OxPAPC including complex lipid oxidation processes in vivo yielding transient products and secondary reactions or rearrangements to generate more stable metabolites (e.g. reaction of the aldehyde with the amino groups of proteins as observed with the TRPA1 protein) 49 as well as (3) facilitated metabolization of OxPC species by phospholipase A 2 particularly under inflammatory conditions 59 . Similarly, 15 min after local OxPAPC injection, PGPC and POVPC were not present any more, while LPC was routinely found and decreased in E06 or D4F treated animals. Whether short carbon chain aldehydes and oxidized fatty acids, oxidation fragments of PAPC, also contribute to OxPAPC-induced pain behavior cannot be excluded yet.
OxPAPC was able to induce long-lasting hyperalgesic behavior in rats and mice suggesting that it has not only a fast action as a stimulant of pain behaviour. It might act as a key component in a cascade of proalgesic mediators leading to long-lasting inflammatory pain and sensitization. It was observed that acute OxPAPC-mediated activation of TRP channels was sufficient to release CGRP, thus indicating that OxPAPC triggers elements of the molecular cascade leading to inflammatory pain. Antagonizing OxPAPC function was an efficient way to reduce immediate actions of OxPAPC stimuli such as calcium responses in vitro but also long-lasting hyperalgesia in vivo.
In nociception, TRPA1 and TRPV1 are of outstanding importance 1 . These ion channels act as detectors of noxious chemical agents on pain-signaling neurons, activating pain pathways to trigger avoidance behavior and even promoting inflammation by release of neuropeptides. Both channels are polymodal receptors and are also sensitive to ROS and peroxidized lipids 21,22,60 . For this reason, TRPA1 and TRPV1 are preferential candidate channels for an excitatory action of OxPL on nociceptive neurons.
Five lines of evidence suggest that OxPAPC is an activator of TRPA1: (1) In vitro reconstitution of TRPA1 channels in HEK-293 cells shows that OxPAPC and components of OxPAPC such as PGPC and POVPC stimulate TRPA1-mediated ion influx to the cytosol. (2) The TRPA1 blocker HC-030031 inhibits acute OxPAPC-mediated calcium fluxes in DRG neurons. (3) OxPAPC action in vivo is antagonized by the TRPA1 blocker HC-030031 and, is attenuated in TRPA1 KO. (4) In vitro, OxPAPC-mediated CGRP release is significantly reduced in tissue from TRPA1 KO. (5) OxPAPC dose-dependently oxidizes the TRPA1 peptide, which represents an intracellular N-terminal domain of the ion channel. Recently it was described that exogenous OxPAPC can induce acute pain and persistent mechanical hyperalgesia. In vivo application of the TRPA1 inhibitor HC-030031 reversed this effect 33 . This group also showed that OxPAPC can activate TRPA1 in vitro, and therefore suggested that targeting the OxPAPC/TRPA1 signaling pathway may be promising in inflammatory pain treatment. This group found no contribution of TRPV1 on OxPAPC function.
Oxidation products of PAPC such as PEIPC share electrophilicity as a common feature with many TRPA1 ligands. Three critical cysteine residues (C621, C641 and C665 in human TRPA1), located in the N-terminal region of the channel, have been proposed to mediate electrophile sensitivity by covalent binding of the ligand and Ab ( 1/300, l-n). (m,n) Summary data are displayed as percentages of reacting cells and AUC analysis of OxPAPC-induced calcium responses. (a-n: OxPAPC: n = 7-13; AITC: n = 5-7; DRGs: n = 7-8, boxplot, median; one-way ANOVA post hoc Holm-Sidak, *p ≤ 0.05).  26 . In calcium imaging experiments, using the hTRPA1 mutant channel lacking the three critical cysteine residues 27, 32 both, the partly electrophilic mixture of OxPAPC as well as the typical TRPA1 agonist AITC lost their ligand functions, but the channel was still responsive to the non-electrophilic agonist carvacrol. However, the OxPAPC compounds PGPC or POVPV are not electrophilic. This indicates that more than one mechanism leads to OxPAPC-induced activation of TRPA1 and TRPV1. The physiological function of OxPAPC compounds can be direct, e.g. via electrophilic interaction or oxidation of the ion channel, but may also include indirect effects. PGPC and POVPC are known to change the lipid and protein environment of cell plasma membranes 50,61 and it might be that the compounds induce mechanical perturbations in the plasma membrane and activate TRPA1 indirectly. This is in accordance with the mechanosensing function of TRPA1 [62][63][64] . Such a mechanism has been proposed for other lipophilic activators of TRPA1 such as bacterial lipopolysaccharides 28 . Furthermore, it would be possible that for instance the carbon acid PGPC, which is known to integrate into the cellular membrane and thought to act only physically with other biomolecules 65 adds electrostatic charges to the membrane by its polar sn-2 chain. The polar group would be exposed to the aequeous phase 50 . These electrostatic forces might be a trigger for TRPA1 activation 66 . How OxPLs act on TRPV1 is puzzling, but might include electrophilic, cysteine-dependent or cysteine-independent interactions with OxPAPC compounds 67 . Additional experiments will be needed to find out how the single OxPAPC compounds act. Reconstitution of functional TRPA1 or TRPV1 in artificial lipid bilayers and subsequent acute application of OxPAPC oxidation end-products such as PGPC may offer deeper insights in the mechanisms how OxPL interact with membranes and whether this indirect mechanism contributes to the activation of TRPA1 and TRPV1 in nociceptors.
OxPAPC, PGPC and POVPC activate recombinant human TRPA1. OxPAPC induces calcium influx in rat and mouse DRG neurons, and OxPAPC-induced pain behavior was observed in the mouse and rat and was also observed in TRPV1 KO mice. This indicates that TRPA1 activation by OxPL is species-independent, albeit we cannot exclude quantitative differences between species.
In the hTRPA1 peptide binding study, no adduct formation and direct thiol oxidation was detected, suggesting that OxPAPC acts as a two-electron oxidant. Kelch-like ECH-associated protein 1 (Keap1), the oxidative stress sensor, is often activated by the same electrophilic compounds as TRPA1. Keap1 activation by OxPAPC has already been documented and it probably occurs by the same mechanism that we observe here 68 .
Previously, it has been observed that 12.5-fold higher concentrations of OxPAPC (125 µM) induce ROS production in human pulmonary artery endothelial cells while 12.5 µM did not cause this response 69 . Here we show, that 10 µM OxPAPC activates TRPA1 channels. It is unlikely that OxPAPC reacts indirectly through acute ROS production since the activation occurs within seconds. We show that high concentrations of OxPAPC result in an increased oxidizing activity on a peptide mimicking the electrophilic binding site of TRPA1. For these reasons, it is plausible that very high concentration of OxPAPC can also induce ROS production.
Pharmacological TRPA1 and TRPV1 blockade reversed OxPAPC-induced mechanical hyperalgesia, while thermal hyperalgesia was only affected by TRPV1 in rats. In previous studies, TRPA1 and TRPV1 agonists equally triggered CGRP release from mouse skin 70 . In line with this, TRPA1 and TRPV1 antagonists, TRPA1 KO, TRPV1 KO and TRPA1/TRPV1 double KO mice showed reduced OxPAPC-triggered CGRP release to a similar extent. TRPA1 preferentially mediated mechanical hyperalgesia in rats (as shown by the TRPA1 antagonist HC-030031) but mechanical hyperalgesia was only partly reduced in TRPA1 gene deficient mice. Thermal OxPAPC-induced hyperalgesia was completely abolished in TRPV1 deficient mice. Both receptors contributed to OxPAPC-induced CGRP release. In the cellular models, however, TRPA1 channels were the preferential direct targets of OxPAPC. One explanation for this finding could be that the responsiveness of TRPV1 depends on its phosphorylation status, which is raised by inflammatory mediators such as PGE 2 and bradykinin, and also by TRPA1 activation itself, which stimulates protein kinase A-mediated phosphorylation 71 . TRPV1 activation might be also facilitated by intracellular phosphorylation induced by other OxPAPC-binding receptors (CD36, PAF receptor, toll like receptor 4, peroxisome proliferator-activated receptors, prostaglandin E2 receptor) 48 on TRPV1 expressing neurons. It needs to be considered that in vivo inflammatory mediators are abundant in the CFA-injected paw. OxPAPC could induce Toll-like receptor 4 activation possibly in CCR2-expressing monocytes or macrophages, which then mediate indirect, sensitizing actions on TRPV1 function 9,42,72 . Furthermore, TRPA1 function is increased via TRPV1-induced Ca 2+ -influx 73 . In summary, our data suggest that in vivo the interaction of TRPA1 and TRPV1 is necessary for OxPAPC action.
Besides pro-inflammatory functions, OxPAPC has anti-inflammatory purposes. Recently, it was shown that the prostanoid-like OxPL epoxycyclopentones are the potent anti-inflammatory OxPL activating transcription factor nuclear factor E2-releated factor 2 (nrf2) 12 . OxPL species with high anti-inflammatory properties include PEIPC or PECPC, but not POVPC or PGPC. OxPAPC has several known receptors including CD36, PAF receptor, toll like receptor 4, peroxisome proliferator-activated receptors, prostaglandin E2 receptor 48 or VEGF receptors 7 . Since none of these are expressed in HEK-293 cells we think that this mechanism in not relevant in vitro, however, we cannot completely rule out that in vivo other mechanisms play a role besides TRPA1. Apparently, the net outcome of pro-or anti-inflammatory effects depends on the biological context.
We identified two strategies to treat inflammatory hyperalgesia by blocking OxPAPC function. Strategy 1 involved anti-OxPAPC antibodies. OxPL are immunogenic and can be regarded as "neo"-self antigens 7 . OxPLs can initiate both innate and adaptive immune responses. This is of major relevance in atherosclerosis 7 .
Scientific RepoRts | 7: 5447 | DOI:10.1038/s41598-017-05348-3 Autoantibodies to epitopes of OxLDL are highly abundant in the plasma of humans and other mammals 36 . These autoantibodies recognize epitopes of OxLDL in atherosclerotic lesions where they can form immune complexes with OxLDL 74 . In a key study by Palinski et al. 18 , multiple monoclonal antibodies (called E0 antibodies) against various epitopes of OxLDL were isolated from apoE-deficient mice. One of these monoclonal antibodies, the E06 mAb, attracted considerable attention, because it recognizes a variety of OxPL metabolites and is therefore useful in diagnostics of atherosclerosis and exhibits promising therapeutic properties 75 . Indeed, the phosphorylcholine head group is essential for E06 binding, but its binding also depends on the oxidized phospholipid conformation including fatty acyl chain length 36,76 . In our study, we now show that E06 mAb has the potential to reduce hyperalgesic behavior, most likely because it acts as a function-blocking antibody against OxPAPC, which is formed in inflammatory tissue. When E06 is pre-incubated with OxPAPC the agonist function of OxPAPC on TRPA1 channel activation is reduced, indicating that E06 is a potent antagonist to functional OxPAPC metabolites. This suggests that humanized monoclonal anti-OxPAPC antibodies might be novel therapeutics for the treatment of  Supplementary Fig. 9b-d). (e) After the last D-4F injection on day 4 mechanical nociceptive thresholds are presented for up to 72 h. (f-h) Collagen-induced arthritis was induced by immunization with collagen in Lewis rats. Rats were treated daily by intraperitoneal injections of D-4F (12.5 mg/kg/d i.p., ), starting at day 6.
Mechanical allodynia (f) and thermal nociceptive thresholds (g) were determined after treatment with D-4F ( ) compared to vehicle treatment (PBS; □). Paw thickness/edema was measured in mm (h). n = 8/group, mean ± SEM, two-way ANOVA RM post hoc Holm-Sidak (a,c-e,g,h,f), *p ≤ 0.05. inflammatory pain. Since anti-OxPL antibodies naturally rise in the plasma of humans 74 , it can be assumed that therapeutic antibodies against OxPAPC are well tolerated if given locally at the site of inflammation.
Strategy 2 used D-4F peptide to reduce inflammatory hyperalgesia. In recent years, ApoA-I mimetic peptides have received high attention as potential therapeutic agents, because these synthetic peptides demonstrated anti-inflammatory and anti-atherosclerotic activity in vivo 77 . Furthermore, ApoA-I mimetic peptides are just emerging for the treatment of autoimmune disease, endotoxemia and cancer 77 . Among the best-studied ApoA-I peptides is D-4F, which is already tested in clinical phase II studies 78,79 and is well tolerated by patients 80 . D-4F is synthesized from D-amino acids and shows a remarkably high affinity to OxPL 37 . For instance, D-4F and its homolog L-4F bind pro-inflammatory OxPL at low nanomolar concentrations. The binding affinity of this interaction is orders of magnitude higher than the affinity of OxPL to human ApoA-I, the natural binding partner 37 . Apo-mimetic peptides, however, are not specific for OxPAPC, but the affinity of D-4F to PGCP, POVPC, PEIPC and 1-(palmitoyl)-2-(5-keto-6-octene-dioyl) phosphatidylcholine (KOdiaA-PC) is substantially higher than for PAPC or other nonoxidized lipids, cholesterol and oxidized lipids like 5(S)-hydroperoxy-5Z,8Z,10E,14Z-eicosatetraenoic acid (HPETE) 37 . In this study, we tested whether systemic administration of D-4F has an antinociceptive effect in animal models of acute and chronic inflammation. In collagen-induced arthritis in rats, a model for chronic inflammation, intraperitoneal injections of D-4F raised mechanical and thermal nociceptive thresholds. In acute CFA-induced inflammation, D-4F treatment showed long-term reduction of mechanical hyperalgesia. Mechanistically we suggest that the antinociceptive property of D-4F is due to a scavenger-like inhibition of OxPAPC function, thus antagonizing the pro-inflammatory and pro-nociceptive properties of OxPAPC metabolites. This proposal is based on the following results: D-4F (1) reduces the abundance of OxPAPC fragments, (2) inhibits OxPAPC-induced hyperalgesic behavior, and (3) interferes with OxPAPC ligand function on TRPA1, and to a lesser degree TRPV1 ion channels (heterologously expressed in HEK cells and native TRPA1 in rat and mice DRG neurons).
In conclusion, we provide experimental evidence that OxPL are activators of TRPA1 and TRPV1 in vitro and in vivo and that antagonism of OxPL and specifically OxPAPC is beneficial for the treatment of acute and chronic inflammatory pain. Potential pharmacotherapies can either be based on monoclonal antibodies with E06-like properties, or ApoA-I mimetic peptides, such as D-4F.

Materials and Methods
Reagents and Chemicals. The  Animals. Male Wistar rats (Janvier, Saint-Berthevin Cedex, France) and female Lewis rats (Charles River, Margate, Kent, UK) were used in the experiments. For injection of agents to rats, a brief isoflurane anesthesia was performed. Rats were euthanized in deep isoflurane anesthesia by intracardiac injection of a solution of T61 (embutramide, mebezonium and tetracaine) or an overdose of pentobarbital. C57BL/6 J wild type mice were bred in Institute of Clinical Neurobiology, University of Wuerzburg, Germany. Initial TRPA1 KO mice were a generous gift from Drs Kwan and Corey 63 and TRPV1 KO mice were donated by Dr. Davis 81 . All mutant mice were bred in the Institute of Physiology and Pathophysiology, University of Erlangen, Germany. Mice were asphyxiated in CO 2 .
All animals were kept under a nonsterile housing environment in accordance with the local Animal Care and Use Guidelines. Animals were randomly assigned to treatment groups.

Lipid extraction and MALDI-TOF MS.
Paw tissue from treated animals (see below) was shock-frozen after dissection from bone and skin followed by homogenization in a mixture of chloroform, methanol and water (2:1:1 volume) with a lab mixer (IKA ® RW 14, IKW ® -Werke, Staufen, Germany). After centrifugation (1500 g, 4 °C, 30 min), the chloroform (lower) phase was recovered, evaporated under nitrogen and stored at −20 °C. MALDI-TOF MS was performed immediately after extraction using 2,5-dihydroxybenzoic acid as matrix, as described earlier 82 . Briefly, chloroform extracts or reagents (OxPAPC, PAPC; CFA) 83 were re-dissolved in 20-30 µl of a 2,5-dihydroxybenzoic acid matrix solution (in methanol) prior to deposition onto the MALDI target. MALDI-TOF mass spectra were acquired on a Bruker Autoflex mass spectrometer, which is equipped with a pulsed nitrogen laser, emitting at 337 nm (Bruker Daltonics, Bremen, Germany). The extraction voltage was 20 kV. Gated matrix suppression was applied to prevent the saturation of the detector by matrix ions. For each mass spectrum, 128 single laser shots were averaged. The laser fluence was kept about 10% above threshold (i.e. the minimum laser fluence required to achieve detectable signals) to obtain optimum signal-to-noise (S/N) ratios. In order to enhance the spectral resolution all spectra were acquired in the reflector mode using delayed extraction conditions. E06 competitive binding assay. The competitive binding assay was designed according to Cox et al. 84 .
OxPAPC-coated Snooper (Avanti Polar Lipids, Alabaster, AL, USA) were blocked with BSA (2%) and 0.27 mM EDTA in PBS for 45 min. Meanwhile, lipid extracts as well as OxPAPC in different concentrations as a standard curve was emulsified in PBS and mixed with 2 µg/ml TopFluor-labeled E06 mAb dissolved in 1% BSA/PBS containing 0.135 mM EDTA 51,52 . After 4x washing the snoopers with PBS, the mixture of lipid plus antibody was added and incubated for 1 h at room temperature. Wells were washed 3 times with PBS. Fluorescence intensity was measured in distilled water at 485 nm with a Tecan Genios pro Fluorescence plate reader (Tecan group ltd., Männedorf, Switzerland). Concentrations were calculated from a standard curve obtained with OxPAPC (0.006-20 µg). hTRPA1 peptide binding of OxPAPC. A model peptide of the intracellular N-terminal part of hTRPA1 including amino acids 607-670 (UniProt database, O75762, Thermo Fischer scientific) was used to evaluate the effect of OxPAPC on cysteine residues of TRPA1 30 . The peptide (50 M) was dissolved in 20 mM ammonium bicarbonate buffer, pH 7.4, and treated with 50 and 500 M OxPAPC for 15 min. The samples were injected using a syringe pump at a flow rate of 240 mLh −1 into a maXis 4G, a ultra-high-resolution ESI-TOF mass spectrometer (Bruker, Bremen, Germany). Nitrogen was used as the nebulizing gas at a pressure of 10 psi and as the drying gas at a temperature of 180 °C and a flow rate of 5 L min −1 . All experiments were carried out in the positive-ion mode and obtained spectra deconvoluted and further processed in a data analysis software provided by Bruker Daltonics. Analytical samples were prepared by mixing peptide samples in water with a mixture of 0.1% formic acid in water/acetonitrile (1:1, v/v). CGRP release. CGRP release was measured from skin flaps from both hindpaws as described earlier 85 .
Samples were directly placed in test tubes, washed for 30 min in carbogene gassed synthetic interstitial fluid (SIF) containing (in mM) 108 NaCl, 3.48 KCl, 3.5 MgSO 4 , 26 NaHCO 3 , 1.7 NaH 2 PO 4 , 1.5 CaCl 2 , 9.6 sodium gluconate, 5.5 glucose und 7.6 sucrose. CGRP release was measured in a shaking bath at 37 °C basally and several incubation steps designed in an intraindividual application scheme. Chemical stimulation with OxPAPC (30 µM) with/ without antagonists occurred in a third incubation step and was followed by two incubations in SIF. To block TRPV1 and TRPA1 receptors, respectively, antagonists were added from the second to the fourth incubation period. CGRP was quantified by an enzyme immunoassay kit (Bertin Pharma, Montigny le Bretonneux, France) and read out photometrically (Dynatech, Channel Islands, UK). The antibodies used are directed against human α/β-CGRP but are cross-reactive against mouse CGRP. DRG preparation. Preparation of DRG from adult wildtype mice and adult rats were carried out as described 34 . DRG neurons were grown at a density of 7 × 10 3 cells per glass cover slip coated with poly-L-lysine (20 µg/ml) and cultured at 37 °C, 5% CO 2 atmosphere for one day. Measurements were performed the following day. Medium contained 100 ng/ml nerve growth factor (Sigma-Aldrich, Taufkirchen, Germany).
In all statistical analysis, the mean of basal fluorescence intensity was determined for each measurement. Number of reacting cells (%) was calculated by 1.5-fold increase of mean basal fluorescent intensity after stimulation. The area under curve (AUC) was taken from the mean. Intervals correspond to the stimulation period of each substance. Mean of basal fluorescence was set as zero.
Whole cell patch clamp recordings. HEK-293 cells (10 5 ) cells were placed in a recording chamber.
Hindpaw inflammation and pain behavior testing. Randomly assigned male Wistar rats received intraplantar injections of either 150 µl CFA 23 or 100-500 µg OxPAPC/PAPC/paw (125-625 nmol dissolved in 100 µl 0.9% saline) under isoflurane anesthesia. The doses were chosen in accordance with previous publications of OxPAPC injection or applications of other oxidized lipids 22,42,43 . In selected experiments rats were treated with inhibitors as described in the legends either by local intraplantar or systemic (intravenous or intraperitoneal) injection: BCTC (4 mg/kg i.v.), HC-030031 (30 mg/kg i.p.), E06 mAb (1/100: 0.01 µg/paw), D-4F (2.5 mg/kg i.pl.: 500 µg/ paw or 12.5 mg/kg i.p.). In TRPA1 and TRPV1 KO mice, 100 µg of OxPAPC was suspended in 25 µl 0.9% saline and injected i.pl. Mechanical nociceptive thresholds in rats were determined using the paw pressure algesiometer (modified Randall-Selitto test; Ugo Basile, Comerio, Italy) as described 88 . The pressure required to elicit paw withdrawal, the paw pressure threshold, was determined in rats. In mice, mechanical allodynia was measured by the upand down method using von Frey hair filaments 40 . First filament used was number five. Vigorous paw withdrawal or licking was considered as a positive trial. Each filament was applied at least three times. When two out of three trials were positive the response was considered positive, the next filament with the higher number was applied. In case of a negative response, the next filament with the lower number was used. The sequence was continued for four more responses after the first positive response. The resulting pattern of the responses was used to calculate the 50% threshold using the following formula: 50% g threshold = (10 [X f +kδ] )/10,000. (X f = value of the last applied filament; κ = tabular value of the pattern of the responses and δ = mean difference between the filaments).
Thermal nociceptive thresholds for rats and mice were obtained by measuring the paw withdrawal latency according to Hargreaves using the plantar test apparatus (Ugo Basile) as previously described 3 . Here, rats were habituated for 15 min to individual plexiglas chambers placed on a glass floor at room temperature. A beam of radiant heat was applied onto the plantar surface of each hindpaw and paw withdrawal latencies (s) were recorded. Withdrawal latencies were monitored three times in each paw alternately and the latencies of both paws were average on each measurement day. At least one minute was given between consecutive measurements in the same paw and a cut-off latency time of 20 sec was used to avoid tissue damage. Averages from three measurements per treatment were calculated. Baseline measurements were obtained before agent injections and at the indicated time points thereafter. Paw edema was determined by placing hindpaws in a water-filled Perspex cell of a plethysmometer (Ugo Basile, Comerio, Italy).
Collagen-induced arthritis and pain behavior testing. Induction of arthritis was performed as described previously 40 . Briefly, bovine type II collagen (4 mg/ml; MD Bioproducts, Zürich, Switzerland) was dissolved in acetic acid (0.1 M) and then emulsified with CFA (1 mg/ml). Under isoflurane anesthesia female Lewis rats were injected intradermally at the base of the tail with 200 µl of the collagen (400 µg)/CFA emulsion. The onset of arthritis scores was around day 12 (see Fig. 10f-h). Rats were scored on a scale of 0-3 per hindpaw; 0-6 per rat. Clinical signs of arthritis and body weight were monitored prior to immunization and then during the disease process. Mechanical hindpaw withdrawal thresholds were assessed by applying a series of calibrated von Frey filaments (0.4-15.0 g; North Coast Medical, Gilroy, CA, USA) to the plantar surface of the hindpaw according to the "up-down" method. Heat hyperalgesia was assessed as described above. Paw swelling was determined by measuring the thickness of each hindpaw using a thickness gauge (Mitutoyo, Leonberg, Germany; expressed in mm).
Statistics. Data are presented as mean SD/S.E.M., median and boxplot, defined as followed: box = upper/ lower quartile, whisker = 1x SD, line = median, square = mean, diamonds = upper/lower outliner. Multiple measurements with one or two variables were analyzed by one-way ANOVA or by two-way ANOVA repeated measurements (RM), post hoc test as indicated. Differences were considered significant when p ≤ 0.05. Statistical analysis for in vitro data was performed with OriginPro 9.0 (OriginLab Corporation, Northampton, MA, USA) and behavior studies with SigmaStat (Systat Software GmbH).

Study approval.
All experiments and study protocols were performed in accordance with the European Union guidelines, and were approved by our institutional Animal Care and Utilization Committee (Regierung von Unterfranken, Wuerzburg, Germany) and by the Home Office, London, UK, and in accordance with the International Association for the Study of Pain.