Soluble epoxide hydrolase and epoxyeicosatrienoic acids modulate two distinct analgesic pathways

Inhibitors of sEH Suppress the Induction of Spinal COX2 Message.

In mice during sepsis or in rats during local inflammation, increased plasma PGE₂ levels were consistently reduced after sEHI treatment (13, 15). However, peripheral inflammation and noxious stimuli are known to evoke a robust increase in the spinal cord COX2 gene expression and prostanoid production (17–19). Given the ability of sEHIs to reduce plasma levels of PGE₂ we hypothesized that sEHI would block spinal prostaglandin production. Relative spinal COX2 mRNA levels after LPS-elicited pain and sEHI treatment were monitored as a measure of spinal prostanoid production. Similar to previous reports we observed a highly significant increase in spinal COX2 mRNA after i.pl. LPS administration (Fig. 1C), although this increase was different from that produced by complete Freund's adjuvant where the resulting slower induction is more prolonged but less efficacious (19). Two structurally different sEHIs, AEPU and TPAU, administered peripherally markedly attenuated COX2 up-regulation in the rat spinal cord (Fig. 1C). We found that both sEHIs used efficiently penetrated into the brain, and thus these compounds are capable of direct action in the CNS (Fig. 1D). The suppression of spinal COX2 message is in parallel to an earlier report using another sEHI in which we showed a reduction in cox-2 protein level in livers of inflamed mice (20). The potent activity of intraspinal sEHI, the spinal repression of COX2 induction by sEHIs, along with detection of both sEHIs in the brain strongly supports a centrally mediated antihyperalgesic mechanism of action for sEHIs.

Inhibitors of sEH Have COX2-Independent Antihyperalgesic Effects.

Given the lack of effect of sEHIs in the absence of facilitated pain states and the suppression of the COX2 induction in the spinal cord during inflammation, the inhibitors seemed to target transcriptional regulation of the COX2 gene. To test this hypothesis we asked whether COX2 message levels correlated with pain behavior. Neither the 2 sEH inhibitors nor LPS treatment displayed a direct correspondence between spinal COX2 expression and antihyperalgesia (Fig. S2A). It is not unusual in the case of LPS to observe a weak linear relationship between spinal COX2 and pain scores because inflammation evokes a cascade of reactions including the release of numerous pronociceptive mediators with overlapping yet distinct temporal and spatial occurrence. However, sEHIs were antihyperalgesic while COX2 message was induced, displaying a counterintuitive correspondence between increasing spinal COX2 and antihyperalgesia in these animals. Whereas glucocorticoids are well-known repressors of COX2 expression and display a linear relationship between decreased pain-related behavior and suppressed COX2 message (21), sEHIs apparently lack this correlation (Fig. S2A). As a control we evaluated sEHIs using a neuropathic pain model, streptozocin-induced diabetic neuropathy (22), that does not involve extensive COX2 up-regulation. Surprisingly, we observed a significant decrease in mechanical allodynia of diabetic rats using the 2 structurally different sEHIs (Fig. 1E).

These results led us to look for an alternative mechanism of action. We hypothesized that EETs are the major mediators of the antihyperalgesic activity and screened the binding of EETs to a small set of cellular receptors. Given that EETs are highly hydrophobic and significantly similar in structure to ubiquitous fatty acids we did not anticipate that they would have affinity to only 3 of 48 targets tested (14). Of these potential targets we focused on translocator protein (TSPO), formerly known as the peripheral benzodiazepine receptor (23). The mixture of synthetic EETs or their methyl ester analogs (EET-me) displaced a high-affinity radioligand, [³H] PK 11195, from the TSPO with an IC₅₀ of 4.6 μM without affecting [³H]flunitrazepam binding (Fig. S3). The TSPO is proposed to translocate cholesterol from the outer to the inner mitochondrial membrane for downstream synthesis of all steroids in the peripheral tissues, but in the CNS the end products are primarily neurosteroids (24–26). Earlier, TSPO ligands were shown to have antinociceptive and antiinflammatory effects (27, 28).

Steroid Synthesis Is Required for sEHI-Mediated Analgesia.

The dose-dependent displacement of [³H] PK 11195 from its binding site by EETs, while demonstrating a probable interaction of EETs and TSPO or a component of the steroidogenic machinery, did not reveal whether EETs are agonistic or antagonistic in regard to the activity of this receptor. Additionally, the observed effective concentration values (IC₅₀ of EETs mixture = 4.6 μM) were far higher than what would be considered a tight receptor–ligand interaction. However, this assay is not an EET binding assay; rather, it measures displacement of a high-affinity ligand. In addition, EETs were shown to stimulate cortisol production in bovine adrenal fasciculata cells and estradiol and progesterone production in cultures of human luteinized granulose cells at similar concentrations (29–31). Accordingly, we surmised that EETs activate TSPO and that the effects of synthetic sEHIs and natural EETs were mediated partially through an increase in the production of analgesic neurosteroids in the CNS. We postulated that inhibition of acute steroidogenesis would partially antagonize sEHIs and tested this hypothesis using 2 steroid synthesis inhibitors that penetrate into the CNS (32, 33). As predicted, the antihyperalgesic activity of AEPU was abolished when aminoglutethimide (AGL, 10 mg/kg), a general steroidogenesis inhibitor, or finasteride (FIN, 20 mg/kg), a 5α-reductase inhibitor, were coadministered (Fig. S4 A and B). These antagonists had no significant effect on the development of LPS-induced thermal hyperalgesia, nor did they change the responses of vehicle-treated animals (Fig. S4 C and D). Aminoglutethimide, a selective inhibitor of cytochrome P450scc (side chain cleavage of cholesterol) did not change the plasma EET/DHET ratio in LPS- and AEPU-treated rats, indicating that antagonism by this compound could not be attributed to reduced EET production. This observation is in contrast to AEPU treatment, which decreased plasma PGE₂ and DHET levels (Fig. S5). Furthermore, aminoglutethimide did not antagonize the ability of the sEHI to reduce PGE₂, reiterating the presence of multiple mechanisms for the antihyperalgesic effects of inhibiting sEH (Fig. S5).

Next, we took a 2-pronged approach to test whether sEHI activity required the activation of nuclear steroid hormone receptors or whether sEHIs influenced circulating steroid levels. None of the tested steroid receptor antagonists (10 mg/kg) significantly reversed the sEHI-mediated antihyperalgesia (Fig. S6). Interestingly, peripheral inflammation increased circulating progesterone levels with no change in testosterone levels among treatments (Fig. S7A). Circulating hormone levels in animals treated with steroid synthesis inhibitors displayed the expected changes, but the hormones were not completely depleted during the course of the experiment (Fig. S7). Although AEPU treatment did not alter the levels of testosterone with or without LPS treatment it decreased plasma progesterone levels (Fig. S7B). We also quantified a steroidogenesis marker gene, steroidogenic acute regulatory protein (StARD1), to confirm the plasma hormone assays. The mRNA levels of StARD1 in testis and adrenals were 5,000- and 37,000-fold higher than that of spinal cord, which was used as the calibrator. Changes in expression level of StARD1 in 2 major peripheral steroidogenic tissues, the testis and adrenal glands, corresponded well with circulating progesterone and testosterone levels. There was no further enhancement of these levels by sEHI although the sEHI led to a minor decrease in adrenal StARD1 message level in parallel to the decrease observed in plasma progesterone level (Fig. S7C). These findings implicate a selective pattern of regulation of steroidogenesis by sEH inhibitors and/or EETs in addition to supporting the absence of an effect through classical steroid-mediated gene expression or a general increase in steroidogenesis.

EETs and sEHIs Selectively Enhance Spinal StARD1 (Steroidogenic Acute Regulatory Protein) Expression.

In contrast to the above in vivo findings with sEH inhibitors, the in vitro stimulating effect of AA, its lipoxygenase, and cytochrome P450-generated metabolites on steroidogenesis were recognized as early as the 1980s (34, 35). At least part of this effect was traced to EETs, which stimulate cortisol production (30). Recently, EETs were shown to directly increase StARD1 gene expression and thus steroid synthesis in cell lines from reproductive tissues (36). It is proposed that acute steroidogenesis is largely dependent on rapid production and degradation of StARD1 message and protein and that TSPO and de novo StARD1 cooperatively facilitate the rate-determining, finely tuned, on-demand transport of cholesterol into the mitochondria (37–39). In the CNS, however, the parallel steroid synthesis cascade produces a group of endogenous molecules termed neurosteroids that potentiate inhibitory GABA currents in neurons (40). We therefore asked whether increasing the level of EETs in the CNS by inhibiting sEH would enhance the expression of StARD1 mRNA. Interestingly, spinal StARD1 expression was already increased, although briefly, during inflammation (Fig. 2A) in parallel to the increase in adrenal StAR message. The 2 chemically dissimilar sEHIs greatly enhanced the increase in spinal StARD1 message in inflamed animals but not in noninflamed controls that received AEPU alone. The increase in StARD1 message was positively correlated with the temporal occurrence of antihyperalgesia after administration of AEPU and TPAU (Fig. S2B). Notably, TPAU, the stronger repressor of COX2 message (Fig. 1C), displayed a shallower slope, possibly because of a ceiling effect or superior down-regulation of COX2. In brain, baseline StARD1 message levels were identical to those quantified from the spinal cord. Neither local inflammation nor AEPU alone elicited an increase in StARD1 message in the brain, although a 2-fold increase was evident in inflamed animals treated with AEPU (Fig. 2B). Given that the calculated half-life of StAR protein is ≈5 min and that each StAR molecule is estimated to turn over ≈400 cholesterol molecules per minute in adrenal cells, we expect that the brief and minor expression changes mediated by sEHIs that are detected here can significantly amplify neurosteroid synthesis in the CNS and thus lead to antihyperalgesia (38, 41).

EETs and sEHIs Redirect Elevated cAMP to an Analgesic Pathway.

An important requirement for the interaction between EETs, TSPO activity, and StARD1 expression may be the presence of elevated cAMP because expression and phosphorylation of StARD1 is greatly enhanced upon gonadotropic hormone stimulation, which increases intracellular cAMP levels (42, 43). Separately, the maintenance of hyperalgesia in inflammatory and neuropathic pain states is known to be largely regulated by the activation of the cAMP signaling pathway (44–46). In the brain, intracellular cAMP level is known to rise rapidly in response to inflammation mainly because the cox-2 product PGE₂ activates E-prostanoid receptors and initiates a cascade of events beginning with stimulation of adenylate cylase (47). The resulting inflammatory pain can be blocked by an inactive cAMP analogue, which prevents PKA activation (48). Here we confirmed that peripheral inflammation led to an increase in spinal cord levels of intracellular cAMP by quantifying 2 cAMP-responsive genes, both of which were significantly induced during the course of inflammation (Fig. S8).

The prevailing outcome of elevated intracellular cAMP appears to be a sustained pain state. However, we hypothesized that increasing the level of endogenous EETs in the CNS in the presence of elevated cAMP may favor neurosteroid production by up-regulating StARD1 expression. This should reduce nociceptive activity. Inferring that the concurrent presence of cAMP and EETs may be required for neurosteroid-based antihyperalgesia we tested whether StARD1 expression in the brain or the spinal cord of noninflamed animals could be increased by direct spinal administration of 8-Br cAMP, EETs, and sEHI. Because these animals were not inflamed and were under anesthesia, we predicted that changes in StAR expression would stem from injected cAMP and EETs/sEHI. Nociceptive thresholds of these animals were not determined because this assay was done under isoflurane anesthesia. As predicted, in noninflamed rats 30 min after compound administration only coadministration of 8-Br cAMP (100 μg) plus EETs (5 μg) and 8-Br cAMP plus AEPU (1 μg) significantly increased spinal StARD1 levels (Fig. 2C). The EETs alone suppressed basal StARD1 expression whereas AEPU alone or cAMP alone were without effect. In the brain of the same animals again only the group that received cAMP and AEPU displayed an increase in StARD1 expression. Because AEPU in vivo is many-fold more stable than EETs, this sEHI elicited a parallel increase in brain StARD1 whereas intraspinal EETs alone had no affect on brain StARD1 mRNA (Fig. 2D). Neither brain nor spinal StARD1 expression changed in response to saline or 8-Br cAMP administration. This observation is in contrast to cultured adrenal or testis cells where cAMP analogues are able to induce StARD1 expression and steroidogenesis. It is plausible that regulation of StARD1 in the CNS differs from that in reproductive and endocrine tissues. Overall these observations may explain the lack of efficacy of sEHIs in the absence of inflammation or neuropathy when intracellular cAMP levels are inadequate to drive neurosteroid production. Equally, during inflammation when EETs are not elevated or stabilized the influence of such an endogenous neurosteroid-based antihyperalgesic mechanism may be marginal because EET levels in this case could become rate-limiting. Interestingly, the expression levels of sEH message in spinal cord or brain were identical throughout the treatments in this study (data not shown), but inflammation caused a clear decrease in plasma oxylipins implying that spinal EETs may also be decreased during peripheral inflammation. As shown earlier, sEHI restored the plasma EET/DHET ratio (Fig. S5) (20). The hypothesis that AA release is required for sEHI-mediated antihyperalgesia remains to be tested.

Details.

Details of the experimental protocols are given in SI Text.

Animals, Treatments, Pain Models, and Nociceptive Testing.

The study was approved by the University of California Davis Animal Care and Use Committee. Two models of inflammatory pain and 1 model of diabetic neuropathic pain were used to test the effects of sEHI. The main inflammatory pain model used involved i.pl. LPS (10 μg per animal) administration as described previously (13). Diabetic neuropathy was induced as described by Aley and Levine (22). After baseline thermal withdrawal latency and mechanical withdrawal threshold determination LPS or carrageenan (1% in saline, 50 μL) was administered into 1 hind paw and nociceptive thresholds were monitored over time. Thermal withdrawal latencies and mechanical withdrawal thresholds were corrected to baseline responses and are reported as percentage control latency or threshold as described previously (13). In experiments in which EETs, sEHI, and cAMP analogue were administered intraspinally animals were maintained under deep anesthesia, and therefore nociceptive thresholds were not determined.

Oxylipin Analysis.

Oxylipins were analyzed as described previously (15).

Quantitative Real-Time RT-PCR.

Changes in gene expression were quantified by using the relative (C_Τ) method according to the manufacturer's instructions (Applied Biosystems).