Background
Untreated pain remains a major clinical problem, and there is a need for the identification of novel therapeutic approaches for both acute and chronic pain states. Sustained activation of the pain pathways is underpinned by the actions of peripheral and spinal inflammatory cells, peripheral sensitization of nerve terminals and the plasticity of the spinal neuronal circuits activated [
1,
2]. It is established that nociceptive afferent barrage leads to changes within the spinal cord, including the relatively quick up- and down-regulation of genes for various enzymatic pathways [
3‐
5], which may act to promote nociceptive responses and drive hyperalgesia or act to counteract these events.
Specialised proresolving mediators (SPMs) such as the essential fatty acid-derived lipoxins, resolvins, protectins and maresins [
6,
7] have robust inhibitory effects on inflammatory signalling pathways. This has been particularly well evidenced for the D-series resolvins, specifically resolvin D1 (RvD1 or 17S-RvD1) and its isomer, aspirin-triggered RvD1 (AT-RvD1 or 17R-RvD1), which are SPMs derived from the polyunsaturated fatty acid docosahexaenoic acid (DHA) [
8]. SPMs are generated after an overt inflammatory insult to promote resolution [
6], and this is achieved by the inhibition of pro-inflammatory cytokine production and shortening of the interval between inflammation and resolution by inhibiting neutrophil infiltration and stimulating macrophage phagocytosis [
6]. Endogenous synthesis of the D-series resolvins is via the 15-lipoxygenase (15-LOX)-mediated conversion of DHA to hydroperoxy intermediates (see Additional file
1: Figure S1). Synthesis of AT-RvD1 is enhanced by aspirin [
8] via the acetylation of a serine residue on cyclooxygenase-2 (COX-2), to permit enzymatic conversion of DHA to a 17R-hydroperoxy intermediate, prior to 5-lipoxygenase (5-LOX)-mediated RvD1/AT-RvD1 formation. RvD1 and AT-RvD1 are rapidly metabolised, predominantly by 15-prostaglandin dehydrogenase to 8-oxo and 17-oxo metabolites [
9]. A metabolically more stable analogue of resolvin D1 has been developed with similar potency in reducing neutrophil infiltration and increasing macrophage phagocytosis [
10]. The biological effects of RvD1 and AT-RvD1 have been attributed to the G-protein-coupled receptor GPR32 and formyl peptide receptor 2 (FPR2/ALX), utilising Gi and possibly Gq as signal transductions [
11,
12]. Both receptors are expressed in human tissue, but GPR32 is not yet identified in rodents [
13].
Despite their rapid degradation, locally administered RvD1 and AT-RvD1 have analgesic effects in various animal models of pain [
13‐
17]. In carrageenan-induced inflammatory pain, intraplantar administration of RvD1 attenuated paw oedema and heat hyperalgesia [
14]. Single intrathecal administration of RvD1 or AT-RvD1 rapidly decreased heat pain thresholds and reduced mechanical hypersensitivity in behavioural tests in the carrageenan model [
13,
14]. The spinal neuronal mechanisms underlying these neurophysiological effects on pain behaviour have yet to be elucidated.
The future exploitation of this potential novel class of analgesics requires a comprehensive understanding of their sites and mechanisms of action and the conditions under which inhibitory effects are evident. Here, we characterise the neurophysiological mechanism effects underpinning the effects of AT-RvD1 on spinal nociceptive processing using in vivo electrophysiology in two models of pain. Acute effects of spinal administration of AT-RvD1 on evoked spinal neuronal responses were characterised in the carrageenan-induced model of inflammatory pain and monosodium iodoacetate (MIA) model of chronic joint pain. To advance mechanistic understanding of this system in the spinal cord, gene expression of the known resolvin receptors in the rodent (FPR2/ALX and chemerin (ChemR23) receptor) and relevant enzymatic pathways, in particular 15-LOX and 5-LOX, were quantified in the model of inflammatory pain, compared to the relevant control group.
Discussion
Previous studies reported acute inhibitory effects of spinal resolvins on behavioural responses to painful stimuli. The results reported herein demonstrate that spinal administration of AT-RvD1 did not alter physiological spinal nociceptive responses but significantly attenuated evoked nociceptive responses of dorsal horn spinal neurones following a 24-h period of peripheral inflammation of the hind paw. At the timepoint of the novel effects of spinal AT-RvD1, we demonstrate that the model of hind paw inflammation is associated with changes in the dorsal horn gene expression of enzymes related to resolvin pathways. The novel inhibitory effect of spinal AT-RvD1 was not recapitulated in a model of chronic joint arthritis, suggesting specificity to conditions associated with overt peripheral inflammation.
The carrageenan model was associated with robust pain behaviour, altered weight bearing and decreased PWT, consistent with previous studies at early timepoints [
20,
35]. In the present study, we extended the period following carrageenan injection to allow the characterisation of spinal events at timepoints more relevant to injuries associated with acute inflammation and pain seen clinically and the timecourse of the activation of the endogenous resolution pathways [
36]. At the timepoint of spinal neuronal recordings, Aδ-evoked responses of WDR neurones significantly increased compared to the control group. All other evoked responses of WDR neurones were similar in saline- and carrageenan-treated rats. Clearly, a limitation of quantifying the behavioural pain response and recording the responses of WDR neurones at this later timepoint was that baseline responses of neurones pre-carrageenan injection could not be characterised, and therefore, only population changes in response can be reported. Previously, WDR neuronal responses characterised before and at 3 h post carrageenan injection were reported to exhibit increased C-fibre-evoked responses in around half the group, and the rest had decreased response [
26]. In our study, 22 (58 %) neurons in carrageenan rats displayed higher baseline C-fibre responses compared to the mean of baseline C-fibre responses in non-inflamed animals, similar to previously reported [
26].
Spinal administration of AT-RvD1 attenuated C-fibre-evoked responses of spinal neurones in carrageenan-treated rats but not in saline controls. The magnitude of the inhibitory effects of AT-RvD1 for both non-facilitated (input) and facilitated (post-discharge) C-fibre responses of WDR neurones were comparable. WU was the most sensitive to the effects of AT-RvD1 (53 % inhibited). These data provide for the first time a neurophysiological basis for the effects of a resolvin molecule on spinal nociception in vivo and reveal that these inhibitory effects are only evident under certain conditions. Our results are in agreement with other ex vivo spinal cord slice studies which demonstrated that RvE1 selectively reduces excitatory post-synaptic potential in the presence of the pro-inflammatory cytokine tumour necrosis factor-α (TNF-α), by inhibiting
N-methyl-
d-aspartate (NMDA) receptor activation [
14], and RvD2 selectively attenuates long-term potentiation of spinal neurones from inflamed animals [
37]. To date, the reported analgesic and anti-inflammatory effects of resolvins in animal pain models [
13‐
15,
17] were hypothesised to be mediated by both peripheral and central nervous system sites of action [
14]. Our demonstration of a direct spinal effect on neuronal responses in vivo provides a neuronal basis for the behavioural evidence that spinal AT-RvD1 reduces carrageenan-induced mechanical hypersensitivity at earlier timepoints [
13]. Timecourse analysis of the effects of AT-RvD1 revealed peak effects 30 min post administration, with responses returning to control levels within the hour (as shown in Additional file
1: Figure S3). Although AT-RvD1 is more resistant to metabolism compared to RvD1 [
9], the short-lived nature of action of these bioactive lipids is a disadvantage that needs to be overcome if they are to be harnessed as analgesics [
10,
38]. The inhibitory effects of AT-RvD1 on evoked responses of WDR neurones were blocked by BOC-2, an antagonist for the receptor FPR2/ALX which is thought to be coupled to Gi [
11] and in principle capable of reducing neuronal excitability. Spinal administration of BOC-2 alone did not alter evoked neuronal responses in inflamed rats, suggesting limited tonic inhibition of spinal neuronal activity.
Spinal AT-RvD1 did not inhibit nociceptive responses of WDR neurones in the absence of an earlier overt inflammatory stimulus (saline-treated) at the doses studied. Although it is possible that higher spinal doses of AT-RvD1 may alter physiological spinal nociception in control rats, the greater effectiveness of spinal AT-RvD1 following peripheral inflammation suggests that treatments targeting this mechanism may have a window of selectivity for inflammatory pain.
The MIA model of arthritis pain is rapidly developing (1–2 weeks post induction) and associated with knee joint features characteristic of human osteoarthritis and pain behaviour [
21,
22,
34,
39]. In the present study, the effects of spinal AT-RvD1 on evoked responses of spinal neurones were assessed at 28 days following intra-articular injection of MIA pain, a timepoint when spinal neuronal responses and weight bearing asymmetry are correlated [
21] and there is an increase in spinal glial fibrillary acidic protein (GFAP) immunofluorescence, a marker of spinal astrocyte sensitization. Reduction in PWT distal to the site of pathology in the MIA model is considered to represent centrally mediated receptive field expansion in osteoarthritis (OA) [
31,
40]. Thus, delivery of the electrical stimulation at the hind paw was used to investigate the effect of AT-RvD1 on central sensitisation in the MIA model. Spinal administration of AT-RvD1, at the higher dose, produced a modest inhibition of Aδ-fibre responses in MIA-treated rats, but all other evoked responses of WDR neurones were unaltered. However, our positive control spinal morphine given at the end of the experiment clearly inhibited the evoked responses of spinal neurones recorded in MIA rats.
The differential effect of AT-RvD1 on spinal nociceptive transmission following a period of overt peripheral inflammation is suggestive of changes in the resolvin system, compared to control conditions. Spinal expression of two well-characterised resolvin receptors FPR2/ALX and ChemR23 was quantified in the dorsal horn of the spinal cord in carrageenan- and saline-treated rats. FPR2/ALX is activated by RvD1, and AT-RvD1 and is predominantly localised with GFAP [
13,
30]. ChemR23 is expressed by neurones [
14] and localised with substance P in the central terminals of primary afferents in the superficial dorsal horn [
14]. There was no change in FPR2/ALX receptor mRNA level at 24 h post carrageenan injection, compared to controls. There was however a significant elevation in spinal expression of ChemR23 mRNA in the carrageenan model of inflammatory pain compared to controls (Fig.
4b). This observation is consistent with the report that ChemR23 gene expression is increased in chronic constriction model of neuropathic pain, but not in CFA-induced paw inflammation, at days 3 and 14 post model induction [
41]. Our observation that spinal expression of ChemR23 is increased, while FPR2/ALX was unaltered, in the carrageenan model does not necessarily account for the increased effect of AT-RvD1 in this group of rats. Previous studies have reported that RvD1 (structurally very similar to AT-RvD1 [
9]) has negligible effect at ChemR23, whereas resolvin E1 (RvE1) has an EC
50 ∼1.3 × 10
−11 M for ChemR23 [
42]. On this basis, we suggest that it is unlikely that the increased effectiveness of AT-RvD1 arises as a result of an increased expression of ChemR23 in the spinal dorsal horn in the carrageenan model of inflammation.
In parallel with these studies, we also sought evidence for potential changes in the spinal gene expression of enzymes involved with resolvin biosynthesis following the period of hind paw inflammation. We report a significant decrease in the gene expression of 15-LOX, the first step enzyme converting DHA to 17S-hydroperoxy DHA (17S-H(p)-DHA) [
43], in the dorsal horn of the spinal cord in the model of carrageenan inflammation. Expression of LOX-5, the final enzyme in the biosynthesis of RvD1 and RvE1, was however unaltered in the carrageenan model. Concomitantly, there was a significant increase in the expression of FLAP mRNA, a key enzyme mediating the generation of RvD1, in the dorsal horn of the spinal cord in carrageenan-treated rats compared to controls. Resolvins are endogenously generated during the resolution phase of inflammation [
7], which is consistent with the timing of our electrophysiological and gene expression studies. Although not proven, increased expression of 15-LOX and FLAP in the dorsal horn of the spinal cord is likely to increase baseline synthesis of endogenous resolvins, which may directly enhance the inhibitory effects of exogenous AT-RvD1, or reduce the rate of catabolism of exogenous AT-RvD1, leading to an increased inhibitory effect on evoked neuronal responses. Interrogation of this proposal is not readily achievable as spinal blockade of LOXs and COX-2 has analgesic effects [
44,
45] or inhibits neuronal firing [
46]. Collectively, these data provide new evidence for complex changes in key enzymes involved in the biosynthesis of the resolvins in the spinal cord under specific conditions, which require further interrogation.
Acknowledgements
The authors would like to thank Dr. Devi Sagar, James Spalton, Dr. Lilian Nwosu, Dr. Shahtaheri Seyed and Dr. Ian Devonshire for the assistance with these studies.