Identification of prostamides, fatty acyl ethanolamines, and their biosynthetic precursors in rabbit cornea.

Arachidonoyl ethanolamine (anandamide) and prostaglandin ethanolamines (prostamides) are biologically active derivatives of arachidonic acid. Although available through different precursor phospholipids, there is considerable overlap between the biosynthetic pathways of arachidonic acid-derived eicosanoids and anandamide-derived prostamides. Prostamides exhibit physiological actions and are involved in ocular hypotension, smooth muscle contraction, and inflammatory pain. Although topical application of bimatoprost, a structural analog of prostaglandin F2α ethanolamide (PGF2α-EA), is currently a first-line treatment for ocular hypertension, the endogenous production of prostamides and their biochemical precursors in corneal tissue has not yet been reported. In this study, we report the presence of anandamide, palmitoyl-, stearoyl-, α-linolenoyl docosahexaenoyl-, linoleoyl-, and oleoyl-ethanolamines in rabbit cornea, and following treatment with anandamide, the formation of PGF2α-EA, PGE2-EA, PGD2-EA by corneal extracts (all analyzed by LC/ESI-MS/MS). A number of N-acyl phosphatidylethanolamines, precursors of anandamide and other fatty acyl ethanolamines, were also identified in corneal lipid extracts using ESI-MS/MS. These findings suggest that the prostamide and fatty acid ethanolamine pathways are operational in the cornea and may provide valuable insight into corneal physiology and their potential influence on adjacent tissues and the aqueous humor.

contributes to corneal transparency that is essential for optimum vision and regulates ocular pressure ( 1 ). Although a healthy cornea is avascular of blood and lymph vessels, hypertension or glaucoma can cause injury through abrasion of the endothelial cell lining leading to neovascularization, which if uncontrolled causes scarring and can lead to blindness ( 2,3 ). Prostanoids are important regulators of corneal homeostasis with prostaglandin (PG) E 2 and thromboxane (TX) A 2 mediating corneal endothelial cell proliferation (4)(5)(6). While PGE 2 is also involved in angiogenesis and polymorphonuclear monocyte recruitment ( 7 ), PGD 2 can suppress corneal tumor growth and hyperpermeability ( 8 ). Furthermore, analogs of PGs are used as ocular hypotensive agents with bimatoprost, a structural analog of PGF 2 ␣ ethanolamide (PGF 2 ␣ -EA), being widely used to treat glaucoma ( 9 ).
Prostaglandin ethanolamines (prostamides; PG-EAs) are sequentially biosynthesized from the endocannabinoid arachidonoyl ethanolamine (anandamide; A-EA) by cyclooxygenase (COX)-2 followed by various prostaglandin synthases (PGSs) (10)(11)(12)(13). Anandamide is released from precursor phospholipids N -arachidonoyl phosphatidylethanolamines (NArPE) via N -acyl phosphatidylethanolamine (NAPE)specifi c phospholipase (PL) D (NAPE-PLD), although recent fi ndings have indicated the existence of other pathways mediated by either ␣ , ␤ -hydrolase 4 followed by cleavage of glycerophosphate to yield A-EA, or PLC and subsequent dephosphorylation of phosphoanandamide to A-EA [reviewed in ( 14 )]. The majority of studies investigating these Abstract Arachidonoyl ethanolamine (anandamide) and pros taglandin ethanolamines (prostamides) are biologically active derivatives of arachidonic acid. Although available through different precursor phospholipids, there is considerable overlap between the biosynthetic pathways of arachidonic acid-derived eicosanoids and anandamide-derived prostamides. Prostamides exhibit physiological actions and are involved in ocular hypotension, smooth muscle contraction, and infl ammatory pain. Although topical application of bimatoprost, a structural analog of prostaglandin F 2 ␣ ethanolamide (PGF 2 ␣ -EA), is currently a fi rst-line treatment for ocular hypertension, the endogenous production of prostamides and their biochemical precursors in corneal tissue has not yet been reported. In this study, we report the presence of anandamide, palmitoyl-, stearoyl-, ␣ -linolenoyl docosahexaenoyl-, linoleoyl-, and oleoyl-ethanolamines in rabbit cornea, and following treatment with anandamide, the formation of PGF 2 ␣ -EA, PGE 2  The cornea functions to refract light and protect the intraocular structures of the eye. While its outermost epithelial layer facilitates oxygen diffusion and acts to absorb UV radiation (UVR), the innermost endothelial layer while topical administration reduces intraocular pressure ( 27 ). However, its mode of action is mediated through the CB 1 and CB 2 cannabinoid and vanilloid subtype-1 (TRPV1) receptors that are not activated by prostamides ( 24,28 ). In addition, reports now indicate that NAPEs also have biological functions in their own right, such as membrane stabilization and inhibition of macrophage phagocytosis ( 29 ). Consistently, NAPEs have been detected in low abundance in mammalian systems but then accumulate under conditions of cellular stress (e.g., ischemia and infl ammation), leading to suggestions of putative protective roles [reviewed in ( 29,30 )].
Although A-EA has been found in the cornea as a minor lipid ( 30 ), neither its metabolism through COX-2 to form prostamides nor the prevalence of its biochemical precursor NArPE have been investigated. In this study, we explored the endogenous production of PGF 2 ␣ -EA, PGE 2 -EA, and PGD 2 -EA by the cornea and show its capability to form these prostamides when A-EA is added externally. We also present data detailing the levels of A-EA and its congeners, as well as NArPE and other fatty acyl NAPE species in rabbit corneal tissue. Given the pharmacological potency of prostamides in ocular health, detailed information on their profi le and biochemical precursors in cornea could provide valuable insight into ocular physiology and potential therapeutics.
Prostamides exhibit a range of activities in various systems. PGF 2 ␣ -EA is involved in infl ammatory pain and dorsal horn nociceptive neuron excitability, while PGE 2 -EA increases blood fl ow and reduces mean arterial pressure in the renal medulla, exhibits strong neuroprotective properties in cerebellar neurons, and along with PGD 2 -EA, induces apoptosis in an in vitro model of colorectal carcinoma (18)(19)(20)(21). Prostamides do not show potent interaction with prostanoid receptors, and studies using isolated feline iris cells have suggested the presence of prostamide-sensitive receptors different from the ones responding to PGs (22)(23)(24). The prostamide precursor A-EA has also been shown to exhibit neuroprotective and analgesic roles in infl ammation and pain models ( 25,26 ), Instrument control and data acquisition were performed using the MassLynx™ V4.0 software. For optimization of ESI/MS and ESI/MS/MS conditions, individual standards (10 ng/ l) were introduced into the spectrometer by direct infusion through a syringe pump (fl ow rate 10 l/min) through the HPLC solvent fl ow (rate 0.2 ml/min). All analytes were monitored on the positive ionization mode. Capillary voltage was set at 3,500 V, source temperature at 100°C, desolvation temperature at 400°C, cone voltage at 35 V, while the collision energy was optimized for each compound using argon as collision gas and was set to the follow- Chromatographic analysis of FA-EA species was performed on a Luna C18(2) column (5 m, 150 × 2.0 mm inner diameter) (Phenomenex, Macclesfi eld, UK) maintained at ambient temperature. Sample injections were performed with a Waters 2690 autosampler; the sample chamber temperature was set at 8°C. The injection volume was 10 l, and the fl ow rate 0.2 ml/min. Analytes were separated using an acetonitrile-based gradient system comprising two solvents; solvent A, acetonitrile-water-glacial acetic acid 2:97.5:0.5 (v/v/v); solvent B, acetonitrile-water-glacial acetic acid 97.5:2:0.

Extraction and LC/ESI-MS/MS analysis of prostanoids
Prostanoids were extracted and analyzed as previously described ( 34,35 ). Briefl y, individual corneas were homogenized in 500 l of ice-cold 15% methanol (v/v) using PGB 2 -d4 (40 l of a 1 ng/ l ethanol solution) as internal standard. The homogenates were acidifi ed to pH 3.0 with 1 M hydrochloric acid, semipurifi ed using SPE and eluted with methyl formate. The solvent was then evaporated under nitrogen, and the lipid residue reconstituted in 100 l ethanol and stored at Ϫ 20°C. LC/ESI-MS/MS analysis of prostanoids was based on MRM assays using the following transitions: 15

Extraction and ESI-MS/MS analysis of NAPE species
Two corneas were homogenized individually, using a glass Dounce tissue grinder (1 ml) in ice-cold chloroform-methanol (2:1, v/v) (0.5 ml aliquots to a volume of 3 ml per cornea). The sample was then kept on ice for 90 min with occasional vortexing. Water (0.5 ml) was added to each sample and the vials vortexed before being centrifuged at 5,000 rpm for 8 min to separate the organic and aqueous phases. The organic layer (bottom) from buffer (100 mM, pH 8 adjusted with 1 M HCl) containing EDTA (1 mM), FAAH inhibitor PF3845 (100 nM), and a protease inhibitor cocktail (1:100 dilution). During homogenization the tissue grinder and homogenate were kept on ice. When endogenous production of prostamides was monitored, corneal tissue homogenates (eight corneas) were pooled. Subsequent ex vivo investigations of prostamide formation were carried out by incubating corneal tissue homogenates (two corneas in 3 ml Tris-hydrochloride buffer) for 10 min at 37°C with exogenously added a ) A-EA ( Prostamides and fatty acyl ethanolamines (FA-EA) were extracted using chloroform-methanol (2:1, v/v) ( 32,33 ). Specifically, ice-cold chloroform-methanol (9 ml) was added to each corneal tissue homogenate followed by internal standard (A-EAd 8, 5 l as 1 ng/ l in ethanol). The resulting suspensions were kept on ice for 30 min with occasional vortexing. Each sample was vortexed and centrifuged at 4,000 rpm for 8 min to separate the organic and aqueous phases. The organic layer (bottom) from each sample was then removed into a clean wide-neck vial. The pooled supernatant was evaporated under a fi ne stream of nitrogen, and the remaining residue was reconstituted in 50-100 l ethanol and stored at Ϫ 20°C, for no more than 1 week, awaiting LC/ESI-MS/MS analysis.

LC/ESI-MS/MS analysis of prostamides
Analysis and characterization of PG-EA produced by corneal tissue was performed on an electrospray (ESI) tandem quadrupole Xevo TQ-S mass spectrometer (Waters, Elstree, Hertsfordshire, UK) coupled to an Acquity Ultrahigh Pressure Liquid Chromatography (UPLC) system. The system was controlled by MassLynx v4.1 Software. TargetLynx was used to construct calibration lines and calculate the concentration of analytes of interest. Optimized ESI-MS/MS conditions were achieved through use of Intellistart within MassLynx software. Individual standards (100 pg/ l) were introduced into the spectrometer by direct infusion via the Xevo TQ-S integrated syringe pump (fl ow rate 10 l/min) combined with UPLC solvent fl ow (rate 0.2 ml/min). All analytes were monitored on the positive ionization mode. Capillary voltage was set at 2,000 V, source temperature at 100°C, desolvation temperature at 400°C, and the cone voltage at 35 V. The collision energy was optimized for each compound to obtain optimum sensitivity using argon as collision gas and was set at 14 eV for PGE 2 -EA and PGD 2 -EA, and 16 eV for PGF 2 ␣ -EA.
Chromatographic analysis of PG-EA was performed on an Acquity UPLC® BEH Phenyl C18 column (1.7 m, 2.1 × 50 mm) (Waters) maintained at 25°C supported with Acquity UPLC® BEH Phenyl VanGuard precolumn (1.7 m, 2.1 × 5 mm) (Waters). Sample injections were performed with the Acquity sample manager (Waters); the sample chamber temperature was set at 8°C, and the injection volume was 3 l. Analytes were separated using a method comprising two solvents: solvent A, water-glacial acetic acid 99.5:0.5 (v/v); solvent B, acetonitrile-glacial acetic acid 99.5:0.5 (v/v). Prostamides were eluted using an isocratic method of 25.5% solvent B from 0 to 3 min with a fl ow rate of 0.4 ml/min. At 3.1 min, solvent B was increased to 80% and the fl ow rate to 0.6 ml/min to wash the column for a further 5 min before returning to the original conditions. Multiple reaction monitoring (MRM) assays were set up using the following transitions: Ϫ corresponding to NArPE and NAPE were further analyzed by ESI-MS/MS to confi rm their identity and obtain information on the sn-1 and sn-2 acyl chains.

LC/ESI-MS/MS analysis of prostamides in rabbit cornea
The ESI-MS, MS/MS spectra and fragmentation patterns of prostamides PGF 2 ␣ -EA, PGE 2 -EA, and PGD 2 -EA were studied using commercially available standards ( Fig. 2 ). All prostamide standards were found to form stable sodiated each sample was then removed and pooled into a clean wide-neck vial, and the solvent evaporated under a fi ne stream of nitrogen. The lipid residue was reconstituted in 100 l chloroformmethanol (1:4, v/v) and stored at Ϫ 20°C awaiting ESI-MS/MS analysis ( 36 ).
In order to optimize the ESI-MS and ESI-MS/MS conditions for NAPE analysis, commercially available N -arachidonoyl dipalmitoyl phosphatidyl ethanolamine was used. Using direct infusion (fl ow rate 10 l/min), the optimum collision energy was found to be 40 eV, using argon as collision gas. The analyte was monitored on negative ionization mode and was found to fragment in a similar way to previously published data ( 36 ). The corneal extract was diluted 1:10 (v/v) with chloroform-methanol-water-acetic acid (2: 6.95:1:0.05, v/v/v/v) and analyzed through direct infusion. production ( ‫ف‬ 65%) ( Fig. 6A , B ). For clarity, Figs. 5 and 6 show two of the four recorded transitions (i.e., m/z 380 > 62 and 380 > 283), the ones that gave the best signal when examining the A-EA-supplemented corneal extract and confi rm the presence of 2-amino ethanol ( Fig. 4 ).
The LC/ESI-MS/MS reconstructed ion chromatograms presented in Fig. 4F-I suggest that the A-EA-supplemented corneal extracts produce PGE 2 -EA and PGD 2 -EA, albeit at very low levels. The presence of these compounds is further supported by the product ion spectra corresponding to peaks eluted at 2.24 and 2.79 min, the retention times of PGE 2 -EA and PGD 2 -EA authentic standards, respectively ( Fig. 4J, K ). Inhibition of COX by indomethacin showed reduction of the putative PGE 2 -EA and PGD 2 -EA peaks ( ‫ف‬ 30-70%; peaks eluting at 2.23 and 2.79 min, respectively; Fig. 5C, D ), while inhibition of FAAH appeared to increase the corresponding signals ( ‫ف‬ 30-60%) ( Fig. 6C,  D ), further supporting the identifi cation.
Treatment with indomethacin reduced the relative production of peaks eluting at retention times earlier than that of PGF 2 ␣ -EA indicating the possible presence of 6-keto PGF 1 ␣ -EA, the stable metabolite of prostacyclin ethanolamine (PGI 2 ), in the corneal tissue ( 11 ). A single ion recording (SIR) for [M+H] + m/z 414 revealed two broad peaks at retention times 0.56 and 1.27 min, respectively (supplementary Fig. 1) Fig. 1D-H). These fi ndings indicate the putative formation of 6-keto PGF 1 ␣ -EA by corneal tissue, although a synthetic standard would be needed to further explore and confi rm this fi nding.

LC/ESI-MS/MS analysis of prostanoids in rabbit cornea
Although A-EA is the substrate for enzymatic conversion by COX-2 to PGH 2 -EA ( 13 ), it is the expression and activity of the individual PGSs that ultimately determines the tissue profi le of prostamides. We have, therefore, assessed the profi le of prostanoids in rabbit cornea as means of appreciating the range and relative activity of PGSs in this tissue. Prostacyclin (PGI 2 , measured as 6-keto PGF 1 ␣ ; Table 1 ) appeared to be the predominant prostanoid produced at 270.9 ± 109.8 pg/mg tissue, while PGF 2 ␣ , PGE 2 , and PGD 2 were produced at lower levels (40.5 ± 15.0, 151.9 ± 103.5, and 187.1 ± 73.4 pg/mg tissue, respectively) ( Table 2 ). Interestingly, PGE 1 and PGD 1 , and PGE 3 and PGD 3 , derived from COX metabolism of dihomo ␥ -linolenic acid (20:3) and EPA (20:5) respectively, were also detected albeit at very low levels (0.9-5.3 pg/mg tissue). Furthemore, the cyclopentanone PGs PGJ 2 , ⌬ 12 PGJ 2 , and 15 deoxy ⌬ 12,14 PGJ 2 were also detected at 3-47 pg/mg tissue, showing that PGD 2 may also act as precursor to anti-infl ammatory species in the cornea. ions [M+Na] + m/z 420 for PGF 2 ␣ -EA, and m/z 418 for both PGE 2 -EA and PGD 2 -EA, possibly refl ecting their storage in glass vials. Notably, the relative abundance of [M+H] + species ( m/z 398 for PGF 2 ␣ -EA, and m/z 396 for both PGE 2 -EA and PGD 2 -EA) was found to be very low ( Fig. 2A, D ( Fig. 2F, I ) 2C, F, I ). All these fi ndings are in agreement with previously published data on the prostamide formation and identifi cation in vitro and FAAH knockout mice ( 10,12,13,37 ).
Although sodium adducts have been used to analyze PGF 2 ␣ -EA ( 21 ), prostamide adducts proved to be very stable under our experimental conditions, and the collision energies required were found too high (>40 eV) to produce identifi able characteristic fragments. Furthermore, under our experimental setting the prostamide standards were readily dehydrated resulting in a very low abundance of the [M+H] + ions ( Fig. 2A, D, G ), and it was necessary to use very high concentrations of the commercially available standard (>40 ng on the column) in order to detect these parent ions. Therefore, in order to set up an LC/ESI-MS/ MS assay appropriate for detection and quantitation of prostamides found at low concentrations in biological samples, we followed the fragmentation of [M+H-H 2 O] + ions using four MRM transitions: PGF 2 ␣ -EA, m/z 380 > 362, 380 > 344, 380 > 283, and 380 > 62; PGE 2 -EA and PGD 2 -EA, m/z 378 > 360, 378 > 342, 378 > 299, and 378 > 62. Chromatographic separation of PGF 2 ␣ -EA, PGE 2 -EA, and PGD 2 -EA was achieved using a reverse phase C18 column with an acidifi ed acetonitrile-based gradient system ( Fig. 2J, K ). Analysis of rabbit corneal tissue extracts using this assay did not offer conclusive evidence for the presence of PGF 2 ␣ -EA, PGE 2 -EA, and PGD 2 -EA ( Fig. 3 ).
Further experiments were designed to explore the capability of rabbit cornea to form prostamides. For this, tissue homogenates were incubated with exogenously added A-EA (10 and 50 M) ( Fig. 4 ). The LC-MS/MS reconstructed ion chromatograms presented in Fig. 4A-D show increased production of PGF 2 ␣ -EA in all transitions recorded and following incubation with 50 M A-EA. The ESI-MS/MS spectrum of the compound eluted at 2.14 min (retention time of the PGF 2 ␣ -EA standard) confi rms the presence of PGF 2 ␣ -EA in the A-EA-treated corneal extract ( Fig. 4E ). Furthermore, when the tissue homogenate was incubated with A-EA (50 M) in the presence of the COX inhibitor indomethacin, PGF 2 ␣ -EA formation was inhibited by ‫ف‬ 70% ( Fig. 5A , B ), while the presence of the FAAH inhibitor PF3845 showed a clear increase in PGF 2 ␣ -EA DISCUSSION Prostamides and their biochemical precursor A-EA exhibit a range of pharmacological and physiological functions that make them an attractive basis for therapeutic intervention. Nevertheless, there is very little information available that describes endogenous prostamide levels and their biosynthetic formation pathway(s). In particular, there is a complete absence of any comprehensive analyses of biosynthetic pathways to the prostamides upstream of A-EA. The authors believe that this is the fi rst report of such an investigative lipidomic analysis of prostamide formation in a tissue. Beyond prostamides, these studies also incorporated analytic detection of other FA-EAs. The function of these species concurrently detected with A-EA may form a foundation for a more complete investigation of biolipid function in the cornea and anatomically adjacent ocular tissues.
LC/ESI-MS/MS lipidomic analysis of rabbit corneal lipid extract did not provide clear evidence for the presence of endogenous prostamides. The identifi cation was based on four fragment ions per compound, selected for increased sensitivity and corresponding to dehydrated and structure specifi c ions such as the diagnostic for 2-amino alcohols ion m/z 62 ( Fig. 2 ). Although some peaks with retention times similar to the ones of commercially available PGF 2 ␣ -EA, PGE 2 -EA, and PGD 2 -EA standards were detected, the presence of weak broad signals did not support their identifi cation ( Fig. 3 ). Therefore, evidence of the corneal tissue capability to produce prostamides was sought using externally added A-EA. The formation of PGF 2 ␣ -EA increased following incubation with A-EA, and this production was found to be reduced when COX was inhibited and stimulated by FAAH inhibition. PGE 2 -EA and PGD 2 -EA showed the same response although they were produced at lower levels than PGF 2 ␣ -EA.
Anandamide as well as L-EA and ST-EA were found at relatively low levels (10.7 ± 5 pg/mg, 13.8 ± 4.2 pg/mg, and 8.3 ± 4.9 pg/mg tissue, respectively) and were not the most abundant of the seven species of FA-EA identifi ed: O-EA and P-EA were found at higher levels (42.1 ± 26.8 and 32.7 ± 12.5 pg/mg tissue, respectively) ( Fig. 7 ). This could be attributed to the higher prevalence of their respective fatty acids at position sn -1 of the phosphatidylcholine (PC) precursor ( 39 ). DH-EA and AL-EA were minor congeners (0.4 and 0.7 pg/mg tissue, respectively), indicating very low levels of DHA and ␣ -linolenic incorporation at the sn -1 position of corneal PC available for transacylation to the amine terminal of phosphatidylethanolamine (PE) ( 40 ). This is in contrast to the high levels of di-docosahexanoyl-PC and -PE species reported in rat and bovine retinal phospholipids ( 41,42 ) and highlights Finally, the presence of 13,14 dihydro 15-keto metabolites of PGE 2 and PGF 2 ␣ shows the expression of 15-prostaglandin dehydrogenase (15-PGDH) and PG keto reductases in rabbit cornea, suggesting that the tissue actively controls the levels of PGs through their metabolism and deactivation ( 38 ). Overall, these data clearly show the presence of an active arachidonic acid cascade through COX, while the PG profi le suggests the prevalence of PGIS, PGES, and PGDS isoforms in rabbit cornea, suggesting that the tissue has the capability of forming the correspondent prostamide species.

LC/ESI-MS/MS analysis of FA-EAs in rabbit cornea
Tissue levels of prostamides depend on the availability of A-EA; thus, a low corneal A-EA level could explain the lack of detectable levels of PGF 2 ␣ -EA, PGE 2 -EA, and PGD 2 -EA in baseline corneal extracts. This was confi rmed by LC/ESI-MS/MS analysis of the corneal extract and showed A-EA at 10.7 ± 5.0 pg/mg tissue. Overall, seven species of FA-EA were detected in rabbit cornea; the level of A-EA was similar to ST-EA (8.3 ± 4.9 pg/mg tissue) and L-EA (13.8 ± 4.2 pg/mg tissue), but lower than P-EA (32.7 ± 12.5 pg/mg tissue) and O-EA (42.1 ± 26.8 pg/mg tissue), while AL-EA and DH-EA were detected at much lower concentrations (0.4 ± 0.1 and 0.7 ± 0.3 pg/mg tissue, respectively); data shown in Fig. 7 .

Analysis of NAPEs in rabbit cornea
The pool of A-EA and congeners is derived from membrane stores of the respective NAPE. Commercially available N -arachidonoyl dipalmitoyl phosphatidyl ethanolamine (978 Da) was used to optimize the experimental conditions for the ESI-MS/MS analysis of NAPE species by direct infusion. Fragmentation of [M-H] Ϫ m/z 977 resulted in an abundant product ion m/z 255, which was attributed to the carboxylate anion derived from the fatty acyl in position sn-2 (i.e., palmitate), while other product ions were identifi ed as the sn-2 N -acyl lysophospholipid m/z 739, N -arachidonoyl ethanolamide cyclic phosphate derivative m/z 482, and N -arachidonoylethanolamine phosphate m/z 426. These data were in agreement with results published by Astarita et al. ( 36 ). A general scan in the range of m/z 850-1200 indicated more than 25 potential NAPE species present in corneal lipid extracts. Following a focused MS/MS analysis of the species found in higher abundance (i.e., exceeding 10 7 ion intensity), fi ve NAPE precursor ions with m/z 1041, 1083, 1097, 1055, and 1027 were identifi ed ( Table 2 ). These ions are consistent with the expected masses of seen NAPE species (namely, P-EA, AL-EA, L-EA, O-EA, ST-EA, A-EA, and DH-EA NAPE) precursors of the main FA-EA identifi ed in rabbit cornea ( Table 2 and Fig. 7 ).  the tissue specifi c distribution of sn -1 docosahexanoyl species of PC ( 43 ).
Although A-EA is considered a minor lipid species, representing only 1-10% of total FA-EA in human membranes under basal conditions, studies carried out in brain and nervous tissue indicate A-EA levels are increased in response to injury ( 44,45 ), and that A-EA participates as an antiinfl ammatory agent of the immune response ( 25,46 ). Few studies have examined the actions of A-EA per se in the cornea; however, it is a highly innervated tissue, and CB 1 receptors expressed on corneal sensory nerves stimulated by an agonist were found to support a role in contributing to antinociception in the anterior eye ( 47 ). Also, in a wound-healing model, both CB 1 and TRPV1 receptor activation increased proliferation and migration in corneal epithelial cells ( 48,49 ), thereby indirectly associating A-EA with these physiological functions. The presence of O-EA, ST-EA, L-EA, and DH-EA was previously unreported in corneal tissue ( 50 ). Reports showing a difference in the levels and distribution of 2-AG, A-EA, and P-EA in normal and glaucomatous ocular tissue, from human donors, suggest a role of these fatty acyl moieties in this disease state. The function of other FA-EA species is under investigation and effects on sleeping pattern, appetite control, and depression have been published to date (51)(52)(53). EP-EA was not detected in the cornea, which was unsurprising as eicosapentaenoic acid is a enzymes possess no sequence homology and are optimally active under basic and acidic conditions respectively, as also refl ected in their intracellular localization with FAAH found in membrane fractions and N-AAA in lysosomes minor species and DHA is the predominant omega-3 fatty acid found in the brain and ocular tissues ( 41,54 ).
Inactivation of A-EA and other FA-EA occurs through hydrolysis via FAAH and N-AAA. Interestingly, these  represent only a minor class of lipids, making up just 0.01% of total animal membrane phospholipids, under physiological conditions ( 29 ). Nevertheless, NAPE species are reported to exert bioactive functions independent of being the precursor of the FA-EA [reviewed in ( 30 )]. Concentration of NAPE species can vary greatly depending on the tissue analyzed (e.g., levels described in the rat kidney are 6-fold higher than in rat brain cortex) ( 57,58 ). In the present study, we have identifi ed several NArPE and NAPE species present in corneal tissue ( Table 2 ), and further studies are required to address the pathway of FA-EA generation from these parent phospholipids. Their effects on corneal tissue and cells, per se, also remain to be studied.
The data presented herein suggest that PGF 2 ␣ -EA, PGE 2 -EA, and PGD 2 -EA can be produced by corneal tissue, a fi nding that could be attributed to the specifi c profi le of prostanoid synthases expressed in rabbit cornea. To assess this, we analyzed corneal PGs and found that PGI 2 was the most prevalent metabolite, followed by PGE 2 and PGD 2 in approximately equal concentrations, and, at even lower levels, PGF 2 ␣ . It would follow that the most prevalent PG-EA produced by the cornea would be PGI 2 -EA, and evidence for its biosynthesis from PGH 2 -EA has been presented by Kozak et al. ( 11 ). We have explored the possible production of PGI 2 -EA through formation of its stable metabolite 6-keto PGF 1 ␣ -EA, using SIR of [M+H] + m/z 414 and fragment ions predicted based on the fragmentation patterns of other prostamides (e.g., dehydration, ( 17 ). The therapeutic potential of increasing in situ levels of A-EA has been shown in several studies through the use of FAAH inhibitors ( 26,31 ). The use of nonsteroidal antiinfl ammatory drugs (NSAIDs) is also of interest in controlling A-EA levels, with reports suggesting that lower concentrations of NSAIDs are required for the inhibition of A-EA cyclooxygenation than those required for arachidonic acid cyclooxygenation ( 55,56 ). This would allow for a mechanism that modulates endocannabinoid levels without disrupting the effect of PGs.
While multiple pathways exist for generating free FA-EA from membrane stores ( 15 ), the precursor NAPE species Results are expressed as mean ± SD (n = 3 separate experiments). amino ethanol head group) and identifi ed an indomethacin sensitive peak with retention time ‫ف‬ 1.2 min. Although these observations suggest the formation of PGI 2 -EA, further work using a synthetic standard is needed to confi rm and further explore this fi nding. The prevalence of PGs PGF 2 ␣ , PGE 2 , and PGD 2 supports the identifi cation of prostamides PGF 2 ␣ -EA, PGE 2 -EA, and PGD 2 -EA and points to the presence of functionally active PGFS, PGES, and PGDS isoforms in the cornea. Although production of the PGF 2 ␣ was ‫ف‬ 4-fold lower than both PGE 2 and PGD 2 , prostamide PGF 2 ␣ -EA was identifi ed at relatively higher levels following external addition of A-EA substrate. This could be possibly attributed to the prevalence of prostamide/PGFS synthase in the corneal tissue. This synthase has been reported in mouse eye, although its exact location in the ocular tissues was not reported and the activity was found to be much lower compared with brain or heart tissue ( 12,59,60 ).
The inactivation of PGE 2 and PGF 2 ␣ as evidenced by the presence of 13, 14-dihydro 15-keto PGE 2 and 13, 14-dihydro 15-keto PGF 2 ␣ suggests the presence of functionally active 15-PGDH in the rabbit cornea. This new fi nding provides valuable information on corneal function, suggesting that this tissue actively controls the production of PGs, and may have implications for the modulation of pain and injury. It also raises the possibility that the low levels of prostamides PGE 2 -EA and PGD 2 -EA detected in the corneal tissue could be a consequence of their metabolism by 15-PGDH ( 61 ). Studies have showed that PGF 2 ␣ -glyceryl ester is a poor substrate for 15-PGDH compared with the free acid, and it is plausible that PGF 2 ␣ -EA may also be less effi ciently oxidized ( 61 ); this together with the potential prevalence of a prostamide/PGFS synthase in the cornea ( 59 ) could contribute to relatively higher levels of PGF 2 ␣ -EA, as reported in the present study.
In conclusion, the novel fi ndings presented herein provide evidence that the pathway for the biosynthesis of PG-EA is operational in the cornea and, as such, constitutes a distinct target for modulating pain perception through use of FAAH and COX-2 inhibitors, in a way that is independent from the classical PG pathway. In addition, the congeners of A-EA were detected and quantifi ed, which provides valuable insight into corneal physiology and those tissues that are anatomically adjacent. Thus, it is possible that corneal FA-EA and their biosynthetic precursors may infl uence a proximal region, such as the endothelial cells of Schlemm's canal. These studies provide rationale for such future investigations.