Inflamm-Aging and Arachadonic Acid Metabolite Differences with Stage of Tendon Disease

The contribution of inflammation to the pathogenesis of tendinopathy and high prevalence of re-injury is not well established, although recent evidence suggests involvement of prostaglandins. We investigated the roles of prostaglandins and inflammation-resolving mediators in naturally occurring equine tendon injury with disease stage and age. Levels of prostaglandins E2 (PGE2), F2α (PGF2α), lipoxin A4 (LXA4) and its receptor FPR2/ALX were analysed in extracts of normal, sub-acute and chronic injured tendons. To assess whether potential changes were associated with altered PGE2 metabolism, microsomal prostaglandin E synthase-1 (mPGES-1), prostaglandin dehydrogenase (PGDH), COX-2 and EP4 receptor expression were investigated. The ability of tendons to resolve inflammation was determined by assessing FPR2/ALX expression in natural injury and IL-1β stimulated tendon explants. Alterations in the profile of lipid mediators during sub-acute injury included low PGE2 and elevated LXA4 levels compared to normal and chronic injuries. In contrast, PGF2α levels remained unchanged and were three-fold lower than PGE2. The synthetic capacity of PGE2 as measured by the ratio of mPGES-1:PGDH was elevated in sub-acute injury, suggesting aberrations in tendon prostaglandin metabolism, whilst COX-2 and EP4 receptor were unchanged. Paradoxically low tendon PGE2 levels in early injury may be attributed to increased local clearance via PGDH or the class switching of lipid mediators from the prostaglandin to the lipoxin axis. PGE2 is therefore implicated in the development of tendon inflammation and its ensuing resolution. Whilst there was no relationship between age and tendon LXA4 levels, there was an age-associated decline in FPR2/ALX receptor expression with concurrent increased PGE2 levels in injury. Furthermore, uninjured tendon explants from younger (<10 years) but not older horses (≥10 years) treated with IL-1β responded by increasing FPR2/ALX suggesting aged individuals exhibit a reduced capacity to resolve inflammation via FPR2/ALX, which may present a potential mechanism for development of chronic tendinopathy and re-injury.


Introduction
Tendinopathy of the human Achilles and the functionally equivalent equine superficial digital flexor tendon (SDFT) are significant causes of morbidity in athletic individuals [1,2]. Repetitive mechanical loading during exercise is cited as a major causative factor [3,4] with high risk of re-injury [5] due to the inferior mechanical properties of the poorly organised fibrous tissue following healing [6]. The importance of inflammation in tendinopathy is highly debated with the aetiology often cited as a degenerative mechanism [7,8]. However, this inference is influenced by analyses of injured human tendons that are often only available for examination at surgery, usually some time after the initial injury, by which time acute phase events are lost and chronic disease is well established.
The horse presents an attractive large animal model for the study of the equivalent human injury due to the shared characteristics of aging phenotypes [9,10] and elastic energy storing function common to the weight-bearing tendons of both species [11,12]. Equine tendons present a more readily attainable source than the human counterpart, permitting targeted investigation of disease throughout the injury phases as well as normal (uninjured) tendons of a wide age range for comparison. Furthermore, similar to the human injuries, tendon repair processes are frequently clinically classified into three phases in naturally occurring equine injury; the acute phase occurs immediately after the initial trauma lasting only a few days, followed by sub-acute (3-6 weeks) and chronic injury phases (.3 months after injury) [13].The tensile region of the equine SDFT is most susceptible to overstrain injury [14,15]. Injured tendons are enlarged compared to normal and exhibit a haemorrhagic granular central core during early stage injury. The histological appearance of injured equine SDFTs are shown in Fig. 1, illustrating increased cellularity soon after injury compared to normal tendons. During healing, the damaged tissue is remodelled and a fibrogenic scar repair forms and the highly organised arrangement of collagen fascicles are not restored (Fig. 1c) and [16], predisposing to re-injury due to diminished mechanical strength. The effects of age, exercise and mechanical loading are inextricably linked and are potentially synergistic factors in the development of tendinopathy. The frequency of tendon injury in sprint horses has been shown to increase with age from 6% in 2 year olds to 16% in horses aged 5 years and over [17]. Similarly, an increased incidence of Achilles tendon rupture has also been reported in middle aged athletes or aged non-athletic persons [18,19]. Hence the effects of ageing and cumulative microdamage can further exacerbate the risk of re-injury in diseased tendons.
The contribution of inflammation to the development of tendinopathy is not fully elucidated and there is a paucity of data reporting inflammatory processes, particularly during the early stages of injury. However, several studies support the involvement of prostaglandins such as prostaglandin E 2 (PGE 2 ) in the development of tendinopathy via inflammatory processes [20][21][22]. Indeed, prostaglandin lipid mediators are synthesised in response to tissue insult or injury and contribute to pain and inflammation in many connective tissues within the body [23]. PGE 2 levels are reported to increase in the peri-tendinous space of the Achilles of healthy exercising human subjects [24] and in murine patellar and Achilles tendons following treadmill exercise [25], suggesting exercise can also induce tendon inflammation. These observations are supported by in vitro experiments whereby tendon fibroblasts in culture release PGE 2 in response to repetitive cyclic strain [26][27][28]. Furthermore, prostaglandins regulate MMP production, partly via an IL-1b mediated mechanism in catabolism of cartilage, periodontal ligament [29,30] and tendon [20,22] contributing to degradation of the extracellular matrix (ECM). However the involvement of other prostaglandins such as those of the D series and their cyclopentanone metabolites to the development of tendinopathy are not known.
The receptors mediating prostaglandin effects are also cited as contributors to the pathogenesis of tendon injuries. A series of four EP receptor subtypes are responsible for the downstream effects of PGE 2 . The EP 4 receptor is reported to mediate the IL-1b-induced catabolic metabolism via the p38 MAPK pathway in human tendon fibroblasts, implicating its role in the development of tendinopathy [31]. Regulation of mPGES-1 and PGDH enzymes controlling prostaglandin synthesis and the clearance mechanisms associated with degradation have been described for burn related injuries and sepsis in human patients [32]. However, little is currently known about prostaglandin metabolism in flexor tendons that have sustained a natural injury, nor the effect of injury stage and age.
In addition to prostaglandins, other products of the arachadonic acid pathway exert important roles in regulating inflammation. Lipoxin A 4 (LXA 4 ) is a specialised pro-resolving mediator that selectively signals through the FPR2/ALX receptor providing endogenous stop signals for inflammation [33,34]. The ability to resolve inflammation after injury or sepsis is well documented for other body tissues [33,35,36], although knowledge of the anticipated roles of specialised pro-resolving mediators such as lipoxins is limited for tendon injuries. We recently described significantly increased expression of FPR2/ALX in sub-acutely injured equine tendons [16]; however expression appeared to be of insufficient duration and magnitude to suppress inflammation, which may potentiate development of chronic disease and fibrotic repair.
Taking all these observation together, it is likely that additional factors play a role in repair-processes during tendon injury. A reduced ability to respond to inflammation may be a contributing factor influencing the reduced efficacy of tendon repair. Inflammaging is a component of immunosenescence which is an age associated decline in immune function, whereby the major cell types of the immune system exhibit age-related changes, resulting in a diminished ability to cope with inflammation [37]. Although tendon pathology and incidence of injury are known to increase in aged individuals [18,38], the effect of age on the ability to resolve tendon inflammation and the contribution of immunosenescence to the development of disease are not understood. The aims of this study were to assess the temporal and differential alterations in prostaglandin and resolving lipid mediators in normal and naturally injured equine tendons throughout the stages of healing and to determine the effect of age and injury stage on the regulation of prostaglandin metabolism. We hypothesised that the production of PGE 2 increases with age in injured flexor tendons and that pro-resolving lipid mediators are activated during the early injury phase. We report altered PGE 2 metabolism and elevated LXA 4 levels occur during the early stage of tendon disease, and reduced expression of the inflammation resolving receptor FPR2/ALX with increasing age, which has implications for sustaining chronic injury.

Class Switching of Lipid Mediators Occurs in Early Stage
Tendon Injury PGE 2 concentrations were reduced in extracts prepared from sub-acutely injured tendons compared to normals and chronic injuries (P,0.001 and P,0.05 respectively) (Fig. 2a). In contrast, PGF 2a concentrations were similar in normal and injured tendons and were 3-fold less compared to PGE 2 (Fig. 2b). Furthermore, increased (,2-fold) level of LXA 4 was found in sub-acute injury compared to normal and chronic injured tendons (P,0.05; P,0.01 respectively) (Fig. 2c), although no correlation was seen between tendon LXA 4 levels and age within each group. The relationship between PGE 2 levels with age in normal and injured tendons was also assessed in these samples. In normal tendons, there was a significant negative correlation between PGE 2 levels and horse age (P#0.01, r 2 = 0.31) (Fig. 3a). In contrast, with injury there was a significant positive correlation between PGE 2 levels and increasing horse age (P,0.05, r 2 = 0.3) (Fig. 3b), although when separated for injury stage, neither sub-acute nor chronic injuries were significant in isolation.

Regulation of Tendon PGE 2 Metabolism
As PGF 2a levels were lower than PGE 2 and did not differ with injury stage, further analyses were focused towards PGE 2 . To assess whether the measured differences in PGE 2 were attributable to altered prostaglandin metabolism, we analysed gene expression of the key enzymes responsible for PGE 2 synthesis (COX-2, mPGES-1) and degradation (PGDH) based on their roles in prostaglandin metabolism. Normalized mPGES-1 and PGDH expression did not change significantly between normal and injured tendons (data not shown). To assess the balance between PGE 2 synthesis and degradation, we analysed the ratio of the two key enzymes involved in PGE 2 metabolism mPGES-1 and PGDH at the different stages of injury. This comparison revealed a ,3fold increase of mPGES-1:PGDH in sub-acute injury compared to normals (P,0.05) and chronic injury (P,0.01) (Fig. 4a normalized to GAPDH and 4b normalized to 18S ribosomal RNA). There was no relationship between mPGES-1:PGDH mRNA expression with age and no significant differences were observed in COX-2 or EP 4 receptor mRNA expression with age or between normal and injured tendons (data not shown).
PGDH and mPGES-1 proteins were also assessed in extracts of normal, sub-acute and chronic injured SDFTs. A representative Western blot of PGDH protein expression is shown in Fig. 5.
Protein bands indicate two forms of PGDH are present in tendons as previously reported in equine preovulatory follicles, showing a minor monomeric form (30 kDa) and a major dimeric form (60 kDa) [39]. Densitometric analysis of Western blots of PGDH normalised to b-actin showed significantly increased PGDH levels in sub-acutely injured tendon extracts compared to normals (P = 0.04) (Fig. 5), but this was not significantly different in the chronic injury group. mPGES-1 was detectable at very low level in normal and injured tendon extracts and was not quantifiable (data not shown).

FPR2/ALX Expression is Upregulated in Natural Tendon Injury and by IL-1b in vitro
Based on the temporal differences in PGE 2 levels, we next addressed whether alterations in the pro-resolution mediators FPR2/ALX and LXA 4 existed with age or disease stage and their response to inflammation. We previously reported FPR2/ALX protein expression was not detectable in uninjured tendons [16].
In the current study we focused on determining FPR2/ALX expression in natural tendon injury and its regulation in cytokine stimulated tendon explants in vitro. Linear correlation analysis of tendons from horses with injuries showed a significant negative correlation between FPR2/ALX protein expression and age (P,0.001, r 2 = 0.77) (Fig. 6). Interestingly, its expression was lowest in chronic injuries which mostly occurred in the older animals. To test the hypothesis whether the predominance of chronic injuries with age was related to a diminished ability of tendons to resolve inflammation, FPR2/ALX expression was determined in explant cultures of normal tendons stimulated with 5 ngml 21 IL-1b. FPR2/ALX expression could be upregulated by IL-1b in tendons derived from young horses (,10 years old) but its expression was significantly reduced in explants derived from horse's $10 years of age (mean ,10-fold reduction; P = 0.01) (Fig.7a). In contrast, FPR2/ALX expression was not detectable in the corresponding non-stimulated controls (Fig 7b). There was no correlation between media LXA 4 levels and age from tendon explants stimulated with IL-1b (data not shown).

LXA 4 Levels in Media after Combined Stimulation with IL-1b and PGE 2
Stimulation of tendon explants with either IL-1b or a combination of IL-1b and PGE 2 enhanced LXA 4 release in media after 24 hours compared to non-stimulated controls (P = 0.005). Combined stimulation with IL-1b and 1.0 mM PGE 2

Discussion
Prostaglandins such as PGE 2 are produced by tenocytes and other fibroblasts in response to injury and after stimulation with pro-inflammatory cytokines [21,22], initiating MMP mediated catabolism of tendon ECM [40]. Although seemingly destructive to the local tissue architecture, this process facilitates clearance of cellular debris and debridement of the affected ECM as described for wound healing in other connective tissues. Prostaglandins may also exert beneficial regulatory actions in healthy tissues maintaining normal physiologic processes such as local bone remodelling [41] and modification of renal blood flow [42]. Furthermore, their presence following injury signals the onset of lipoxin mediated resolution processes, such that the duration and magnitude of the inflammatory response can be regulated to restrict the degree of tissue damage [33]. Thus prostaglandins can be said to possess 'double-edged sword' properties in terms of their dichotomous roles in wound healing processes. The extent to which these properties play a role in tendinopathies with injury and repair stage remains unclear.
In the current study, PGF 2a levels were unchanged with injury and were substantially lower than PGE 2 levels in normal tendons. This may imply differential regulation of these prostaglandins in tendon, with PGF 2a less susceptible to changes with injury suggesting PGE 2 is the main prostaglandin operative in tendon injury. PGE 2 levels were found to decrease with aging in normal tendons. This could be a consequence of the reduction in tendon cellularity with increasing age [43,44] leading to a decreasing tendon prostaglandin synthetic capacity. Alternatively, it may be related to a lack of PGE 2 synthesising pro-inflammatory macrophages as we have described previously for uninjured tendons [16]. The relationship between age and the pattern of PGE 2 levels was difficult to determine in injured tendons because of the confounding issue that sub-acute injury predominated in younger horses compared to chronic injury, which occurred with greater frequency in older individuals. However, the positive correlation between increasing tendon PGE 2 levels with age in injured horses could be attributable to a greater PGE 2 synthetic capacity both by increased tendon fibroblast cellularity and infiltration of proinflammatory macrophages into injured regions of tendon [16]. This was supported by the increase in mPGES-1:PGDH ratio in sub-acute injury which suggests an interplay between PGE 2 synthesis and degradation could lead to an increased synthetic  capacity in the tissue. Furthermore, activated macrophages from aged humans and mice are reported to produce more PGE 2 than macrophages from younger individuals [45] which may contribute to the greater frequency of tendon injury in older individuals through sustained activation of proteolytic action on the ECM. Whilst there are no equine specific antibodies available to neutrophils or mast cells, precluding immunofluorescent analysis, we were not able to identify these cells by histology of injured tendons between 3-6 weeks post injury (data not shown). As we were unable to access tendons with injuries of less than 2 weeks duration, we cannot exclude the presence of these cells and their contribution to the synthesis of PGE 2 at this earlier phase of injury. However as macrophages are known to release PGE 2 and tendon injury has been shown to be associated with activation and recruitment of these cells [16], they represent an important source of PGE 2 during tendon injury.
Regulation of prostaglandin metabolism is not well documented for normal and pathologic tendons, although the majority of circulating prostaglandins are degraded in the pulmonary vasculature via PGDH [32]. However, tissue levels of PGE 2 are finetuned by locally produced PGDH [46] and the net balance between synthesis and degradation may be a mechanism for controlling the action of PGE 2 . In the present study, the ratio of mPGES-1: PGDH was increased in sub-acute compared to chronic disease or normal tendons, suggesting potential aberration of these genes with disease phase. We propose that the altered intracellular prostaglandin regulation is attributable to a proportionately greater increase in mPGES-1 transcription rather than reduced PGDH transcription due to our observation of an increase in PGDH protein in sub-acute tendon injuries. In further support of this, PGDH kinetics have shown it to be a short lived enzyme whose replacement is dependent upon de novo protein synthesis at the level of translation rather than that of transcription, due to the prolonged half life of PGDH mRNA [47]. Thus, PGDH mRNA is present in low abundance as a stable moiety, presumably as a mechanism for rapid and precise control of enzyme activity. Although these data suggest PGE 2 levels should be increased due to elevated mPGES-1 mRNA, the observed reduction in PGE 2 levels in sub-acute injury compared to normal and chronic injuries could be explained by the increased PGDH protein levels in sub-acute injury. This suggests a secondary (cellular) clearance mechanism for PGE 2 degradation whereby local PGE 2 levels can be regulated [46]. The increased vascularity of tendon that occurs after recent injury may also be a contributing factor to the paradoxical lower levels of PGE 2 after injury [48], as the increased vascular perfusion is likely to facilitate efficient systemic prostaglandin clearance. In addition to vascular clearance, the lower levels of PGE 2 in sub-acute tendon injury could also be attributable to lipid substrate re-routing towards the resolving pathways that are activated during inflammation. These recently discovered pathways demonstrate critical roles in the switching of lipid mediators from the prostaglandin to the lipoxin axis, returning injured tissues to their previous state [49] by depleting PGE 2 levels due to reduced arachadonic acid substrate availability. Indeed, the current study shows significantly increased LXA 4 levels in sub-acute injury compared to normal and chronic injured tendons, suggesting pro-resolving processes are active during the early stage of tendon injury. The alterations in the profile of lipid mediators during this time include low PGE 2 and elevated LXA 4 levels compared to normal and chronic injuries and suggest lipid mediator class switching is active in the early phase of tendon injury. We propose this class switching represents an endogenous protective mechanism to limit the degree of damage to tendon ECM and preserve tissue integrity. This concept is supported in part by the findings from this study, demonstrating combined stimulation of normal tendon explants with 5 ngml -1 IL-1b and 0.01 mM or 1.0 mM PGE 2 induced LXA 4 release, with greater production with the higher dose of PGE 2 . It has been previously shown in an identical experimental system that addition of 1.0 mM PGE 2 to normal tendon explants induced maximal LXA 4 release after 72 hours in tissue culture [16]. These observations suggest that PGE 2 may exert anti-catabolic effects on Samples were loaded on a volume basis and the ratio of PGDH normalised to b-actin was calculated for each sample using band densitometric analysis. Graph shows densitometric analysis of western blots for PGDH in protein extracts prepared from normal (n = 7) sub-acute (n = 5) and chronic injured SDFTs (n = 8). The densitometric values were normalized to levels of b-actin expressed in each sample. There was a significant increase in PGDH in sub-acutely injured tendon extracts compared to normals but this was not significantly different in the chronic injury group. * P,0.05, **P,0.01. Mean values are shown, error bars denote standard deviation. doi:10.1371/journal.pone.0048978.g005 Figure 6. FPR2/ALX protein expression in natural tendon injury. The relationship between FPR2/ALX levels with age is shown in injured flexor tendons (n = 10). Horse age ranged between 4 and 16 years (mean 1164 years). FPR2/ALX expression was significantly reduced with increasing age (P = 0.0008, r 2 = 0.77). Overlapping points are present for tendons derived from more than one 15  tendon ECM via the induction of pro-resolving LXA 4 and switching of lipid mediators from the prostaglandin to the lipoxin axis. Furthermore, in the setting of a pro-inflammatory environment, the presence of higher levels of PGE 2 may exert an autoregulatory feedback effect on IL-1 activity in order to modulate the inflammatory reaction [50]. Although the cell types responsible for lipid mediator class switching have not been identified in inflamed tendons, we hypothesise that the interaction between resident tendon cells and infiltrating pro-inflammatory macrophages during early stage injury initiates activation of pro-resolving processes. LXA 4 levels were reduced during the chronic injury phase where the tendon does not return to normal structure and function. As LXA 4 is a key determinant of pro-resolving processes [51] it is therefore plausible that incomplete resolution sustains a low level of inflammation, perpetuating chronic disease. Although the present study did not measure the multiple enzymes that synthesise the components of prostaglandin and lipoxin pathways, it is hypothesised that control of class switching involves the regulation of some of these enzymes.
The lipoxin A 4 receptor FPR2/ALX is reported to have a pivotal role in controlling the duration and magnitude of the inflammatory response, providing endogenous stop signals for inflammation [33,34]. Despite the anticipated importance of specialised pro-resolving mediators such as LXA 4 in healing, these resolving pathways are not widely studied in injured tendons. We recently identified significantly increased expression of FPR2/ ALX by tenocytes in early equine tendon injury [16] and studies in other inflamed connective tissues have emphasised the importance of resolution processes for regulating inflammation, including inhibition of leukocyte recruitment and modification of vascular permeability [33]. The current study provides novel data illustrating levels of FPR2/ALX are markedly diminished in the tendons of aged injured individuals. Because these mediators are essential for controlling the inflammatory cascade, this suggests an age-related deterioration of tendons to mount a counter-response to inflammation via FPR2/ALX. A component of immunosenescence is 'inflamm-aging' whereby aged individuals exhibit diminished ability to modulate inflammation [37,52]. Studies in humans and rodents report an age related decline in cutaneous wound repair [45,53], suggesting age contributes to dysregulated tissue repair. Our data of reduced FPR2/ALX together with the elevated PGE 2 levels with age suggests an inflamm-aging process is present in injured tendons. In support of this concept, IL-1b stimulated tendon explants derived from uninjured horses aged 10 years and above showed a diminished capacity to express FPR2/ ALX compared to individuals less than 10 years of age. Interestingly, tendon derived cells from older horses also had a reduced response to IL-1b induced PGE 2 production compared to  young horses (Fig. S1). Taken together these data suggest aged individuals are less capable of mounting a protective response to tendon inflammation by this mechanism compared to younger individuals. Immunosenescence may therefore represent a potential mechanism for tendon re-injury via this pathway and in the development of chronic injury and explain, at least in part, the high frequency of tendon injuries sustained by aged individuals [17,18]. We hypothesise that 'inflamm-aging' is an important component in the complex aetiology of tendinopathy, to which additional contributing factors such as load-induced proteolytic matrix disruption [9] and lower tendon metabolic activity with age [54] are superimposed. Furthermore, as we did not observe any relationship between LXA 4 levels with age in samples of natural disease or in vitro, it appears that it is the ability to generate the FPR2/ALX receptor that is impeded by age and not synthesis of the LXA 4 ligand, diminishing the ability to mediate downstream processes culminating in a deficit to resolve tendon inflammation.
From the findings in the present study, we conclude that PGE 2 is implicated in the development of tendon inflammation and its ensuing resolution. Temporal analysis suggests class switching of lipid mediators occurs in early stage injury, driving synthesis of LXA 4 . However, ageing individuals exhibit a reduced capacity to resolve tendon inflammation via the FPR2/ALX receptor, which may present a mechanism for the development of chronic tendinopathy. Improved understanding of injury pathogenesis throughout the phases of tendon repair will facilitate identification of novel therapeutic targets to modulate or prevent disease progression, such as the use of FPR2/ALX agonists.

Ethics Statement
Ethical approval for the collection of post mortem equine tendons from an abattoir or local equine veterinary referral hospital for this study was sought and approved from the Ethics and Welfare Committee at the Royal Veterinary College (URN 2011 1117).

Classification of Injury Stage
Equine forelimbs from Thoroughbred or Thoroughbred cross breed horses aged between 2-20 years were obtained from an abattoir or local equine referral hospital with known history of injury and the tensile region of the SDFT harvested within 4 hours of death. Tendons were grouped as sub-acutely injured (3-6 weeks post injury, n = 6, mean age 965 years) or chronically injured (.3 months post injury, n = 9, mean age 1364 years), as shown before [16]. Tendon injuries were aged based on historical information obtained from either the owner or referring veterinary surgeon prior to euthanasia of the horse. Tendons were classified as normal based on their macroscopic post mortem appearance which included lack of visible signs of swelling of the tendon body and a consistent pattern of fascicles on transverse sections (n = 19, mean age 865 years). The typical microscopic appearance of normal, sub-acute and chronic injured tendons are shown in Fig. 1.

Determination of Prostaglandin Levels in Tendons
After harvest, samples were stored at 280uC and assayed within 6 months. Tendon extracts were prepared as described by Zhang and Wang [55]. Briefly, approximately 1 g of the central portion of the tensile region of the SDFT was cut into 5 mm 3 cubes. Tissue was placed in 5 ml of homogenization buffer (0.1 M Phosphate Buffered Saline (PBS) (PAA, UK) pH 7.4, containing 1 mM EDTA and 10 mM Indomethacin) and homogenized using a dismembranator (Retsch MM2000, Germany). Acetone (2 ml) was added and the sample vortexed and allowed to stand at room temperature for 5 minutes. Tissue was pelleted by centrifugation at 16,0006g and the supernatant harvested. Acetone was removed by evaporation at room temperature for 18 hours and the residual supernatant used for determination of PGE 2 and PGF 2a measurement via radioimmunoassay as described by Cheng et al. [56].

Measurement of Lipoxin A 4 in Tendon Extracts and Tissue Culture Media
LXA 4 levels were determined as an indicator of resolution of inflammation in the supernatant prepared earlier to determine prostaglandin levels in extracts of normal (n = 8), sub-acute (n = 7) and chronic injured SDFTs (n = 6) and in samples of tissue culture media from cytokine stimulated tendon explants as described later. LXA 4 was separated from the supernatant used to determine prostaglandin levels by passage through Bond Elut C18 columns (Agilent Technologies, USA) followed by elution with methyl formate. Samples were evaporated to dryness in a stream of nitrogen and the resulting residues used to determine LXA 4

Quantitative RT-PCR Analysis of PGE 2 Synthesis and Degradation Enzymes
RNA was extracted from 200 mg of tendon tissue (normal n = 6, sub-acute, n = 8 and chronic injury n = 6) as described by Young et al. using the RNeasy kit (Qiagen, UK) [44]. RNA (22 mL) was used for cDNA synthesis using random primers (Promega, UK) and Superscript II reverse transcriptase (Invitrogen, UK) as described by Young et al. Gene specific primers (Table 1) were used to prepare amplicons which were cloned into a pGEMH-T Easy Vector (Promega) and plasmid DNA was used to prepare standard curves as described previously for subsequent absolute copy number evaluation [57]. We investigated expression levels of COX-2, mPGES-1 (inducible terminal enzyme in PGE 2 synthesis), PGDH and the PGE 2 receptor EP 4 which is reported to be implicated in the pathogenesis of tendinopathy [31]. Expression levels of GAPDH and 18S ribosomal RNA were used for normalization. Equine oligonucleotide sequences used for quantitative real-time PCR are shown in Table 1. For each gene of interest and housekeeping gene, 1 mg of cDNA and 1 mM of each forward and reverse primer were used per reaction (conducted in triplicate) and amplified using SYBRH Green JumpStart TM Taq ReadyMix TM (Sigma-Aldrich, UK) for quantitative PCR using an Opticon II DNA engine thermocycler (MJ Research, UK). Standard curves ranged from 1610 8 to 1610 1 copies (GAPDH) or 1610 10 to 1610 3 copies (18 S) such that absolute copy number could be calculated according to cycle threshold. PCR efficiency was tested in each experiment and confirmed to be approximately 2.0, indicating 100% amplification efficiency according to a previously described mathematical model [58].

Western Blot Analysis for mPGES-1 and PGDH
To assess if mRNA changes in PGE 2 metabolism were reflected at the protein level, Western blotting was performed to assess mPGES-1 and PGDH protein expression in tendon extracts. Protein extracts of normal (n = 7), sub-acute (n = 5) and chronic injured tendons (n = 8) were prepared by extraction of 100 mg of finely diced tendon in 15 volumes of 4 M Guanidine hydrochloride with protease inhibitor cocktail III (Merck, CalbiochemH, UK) for 48 hours at room temperature. Insoluble material was separated by centrifugation (13,0006g, 20 mins) and proteins precipitated by the addition of 9 volumes (v/v) ethanol buffer (ethanol with 50 mM sodium acetate) and chilling to 280uC for 2 hours. Pellets were washed twice with ethanol buffer and dried pellets resuspended in Laemmli sample buffer (Bio-Rad, UK). Samples were reduced by addition of DTT to 0.1 M and heated to 95uC prior to electrophoresis on 10% SDS-polyacrylamide gels. Proteins were transferred onto PVDF membrane (GE Healthcare) and blocked for 2 hours in Tris buffered saline in 1% Triton (TBST buffer) containing 8% powdered skimmed milk and 2% bovine serum albumin (BSA). Membranes were incubated with either anti-human PGDH (Santa Cruz Biotechnology Inc, USA) (diluted 1:200), anti-human mPGES-1 (Agrisera, Sweden) (diluted 1:500) or anti-mouse b actin (Sigma-Aldrich) (diluted 1:2000) antibodies for 2 hours at room temperature in TBST containing 5% BSA and 1% Tween-20H. Washed membranes were incubated with anti-rabbit (diluted 1:1000, Cell Signaling TechnologyH, USA) or anti-mouse (diluted 1:2000, GE Healthcare, UK) secondary antibodies conjugated to horseradish peroxidase for 2 hours to visualise proteins using ECL reagent and film (GE Healthcare, UK) according to manufacturer's instructions. Densitometry analysis of bands was performed using ImageJ software (NIH) for the protein of interest relative to bactin.

Analysis of FPR2/ALX Expression by Immunohistochemistry
Having assessed levels of secreted lipid mediators produced at different stages of tendon injury, we next investigated the ability of injured tissues to mount a resolution response to inflammation. This was accomplished by analyzing FPR2/ALX expression in samples of natural tendon injury and by IL-1b stimulation of normal tendon explants in vitro, such that the effect of injury stage and age on the ability to resolve tendon inflammation could be determined. Fresh tendon pieces were embedded in optimal cutting temperature compound (OCT, Sakura Tissue-TekH, The Netherlands) and snap frozen in pre-chilled (280uC) n-hexane and stored at 280uC until used. Serial sections (8-10 mm thickness) were cut on a cryostat (Bright, UK), mounted onto poly-L-lysine coated slides (VWR, UK) and allowed to dry for 2 hours at room temperature prior to immunofluorescent analysis as described below. To investigate the influence of horse age on the ability of tendon cells to resolve inflammation via this mechanism, we assessed the effect of IL-1b on FPR2/ALX expression in an explant culture model. Explant tissues from normal tendons were grouped according to horse age as ,10 (n = 5) or $10 years of age (n = 8). 0.4 cm 3 explants (300 mg 630 mg) were incubated at 37uC and 5% CO 2 under humidified atmosphere in 3 ml of Dulbecco modified Eagle's medium (PAA, UK) containing 1% Penicillin and Streptomycin without foetal calf serum. Samples were stimulated with 5 ngml -1 human recombinant IL-1b (Merck, CalbiochemH, UK) and non-stimulated (vehicle only) samples served as controls. At 72 hours after stimulation, explant tissues were embedded in OCT and snap frozen in chilled (280uC) nhexane, and cryosections cut as described above.
Subsequently, consecutive cryosections were blocked in 5% normal goat serum (Sigma-Aldrich) in PBS for 1 hour in a humid chamber and probed with a 1:100 diluted mouse monoclonal antibody to FPR2/ALX (Lipoxin A 4 receptor, IgG 1 , AbCam, UK) and secondary goat anti-mouse IgG 1 (Southern Biotech, USA) each for 2 hours at room temperature. To visualise nuclei, slides were incubated with 0.5 mgml 21 Hoechst 33342 (Invitrogen, UK) for 20 minutes and washed in PBS Tween-20H (PBS-T). To quench the background fluorescence, slides were incubated with 0.1% Sudan Black B (BDH, Poole, UK) in 70% ethanol for 20 minutes [59], washed in PBS-T and mounted under coverslips using a solution of 80% glycerol in 0.5 mM Tris buffer (pH 7.2). Slides were stored at 4uC in the dark until image acquisition. Cryosections of equine spleen were used as positive control tissue to validate suitability and dilution of antibodies for use on tendon sections. Negative controls consisted of spleen cryosections incubated with murine isotype matched primary control antibodies (Southern Biotech, USA) as previously reported [16]. To assess FPR2/ALX expression, images were recorded using a Leica SP5 confocal microscope (Leica Microsystems, UK) as reported elsewhere [16].

Measurement of LXA 4 in Media from Explants Stimulated with IL-1b and PGE 2
Our previous work has shown that stimulation of normal tendon explants with 5 ngml 21 IL-1b or 1.0 mM PGE 2 in vitro induced production of the pro-resolving ligand LXA 4 which binds to FPR2/ALX [16]. In the current study, LXA 4 levels in media were measured to ascertain if concurrent stimulation of macroscopically normal tendon explants with IL-1b (5 ngml 21 ) and low (0.01 mM) or high dose (1.0 mM) PGE 2 induced a dose-dependent increase in LXA 4 levels. Tendon explants were derived from 3 horses aged between 9 and 14 years and LXA 4 levels determined 24 hours after stimulation with the combination of pro-inflammatory mediators.

Statistical Analysis
Statistical analyses were performed using GraphPad Prism 5 (GraphPad Software Inc., San Diego, CA). Normality was tested using a Kolmogorov-Smirnov test. One-way ANOVA with Tukey's multiple comparison tests were performed to determine differences in PGE 2 , LXA 4 and the ratio of PGDH to b-actin protein between normal, sub-acute and chronic injured tendons. Kruskal-Wallis tests were performed to compare gene expression of mPGES-1, PGDH, COX-2 and the EP 4 receptor normalized to housekeeping genes in normal, sub-acute and chronic injured tendons. Kruskal-Wallis with post hoc Mann Whitney tests were used to compare gene ratios of mPGES-1 to PGDH in normal, sub-acute and chronic injured tendons. A Mann Whitney test was used to detect differences in FPR2/ALX expression in IL-1b stimulated tendon explants in vitro from horses ,10 or $ 10 years of age. Relationships between horse age and PGE 2 levels or FPR2/ALX expression in normal and injured tendons were assessed by linear correlation analysis. A linear mixed model using SPSS PASW Statistics 18 (SPSS Inc Illinois, USA) was used to analyse LXA 4 release from tendon explants stimulated with proinflammatory mediators to account for effects of horse and experimental condition. In all cases, the P value was considered significant if below 0.05. Figure S1 Prostaglandin E 2 (PGE 2 ) production by tendon derived cells stimulated with IL-1b (5 ngml -1 ) in vitro. Tendon cells derived from 8 year old horses (n = 3) had a reduced response to IL-1b induced PGE 2 production compared to 3 year old horses (n = 3). Median values are shown with maximum and minimum range. (TIF)