Altered specificity of 15-LOX-1 in the biosynthesis of 7S,14S-diHDHA implicates 15-LOX-2 in biosynthesis of resolvin D5

The oxylipins, 7S,14S-diHDHA and 7S,17S-diHDHA (RvD5), have been found in macrophages exudates and are believed to function as specialized pro-resolving mediators (SPM’s). Their biosynthesis is thought to proceed through sequential oxidations of docosahexaenoic acid (DHA) by lipoxygenase enzymes, specifically by h5-LOX first to 7S-HDHA, followed by h12-LOX to form 7S,14S-diHDHA or h15-LOX-1 to form 7S,17S-diHDHA (RvD5). In this work, we determined that oxidation of 7S-HpDHA to 7S,14S-diHDHA can be performed by either h12-LOX or h15-LOX-1, with similar kinetics. The oxidation at C14 of DHA by h12-LOX was expected, but the non-canonical reaction of h15-LOX-1 to make primarily 7S,14S-diHDHA was unexpected. Computer modeling suggests the alcohol on C7 of 7S-HDHA hydrogen bonds with the backbone carbonyl of I399, forcing the hydrogen abstraction from C12 to oxygenate on C14, and not C17. This result raised questions regarding synthesis of 7S,17S-diHDHA (RvD5). Strikingly, we find h15-LOX-2 oxygenates 7S-HDHA almost exclusively at C17, forming RvD5 with faster kinetics than h15-LOX-1. The presence of h15-LOX-2 in neutrophils and macrophages, suggests it may have a greater role in biosynthesizing SPM’s than previously thought. We also determined that the reactions of h5-LOX with 14S-HpDHA and 17S-HpDHA are kinetically slow compared to DHA, suggesting these may be minor biosynthetic routes in-vivo. Additionally, we show that 7S,14S-diHDHA and RvD5 have anti-aggregation properties with platelets at low micro-molar potencies, which could directly regulate clot resolution.


INTRODUCTION
Inflammation plays an essential role in protecting the body from tissue injury and foreign pathogens. In its acute form, it involves the up-regulation of localized inflammatory signaling molecules and cytokines that attract neutrophils to an area of injury (1). Over time, an area of injury begins producing specialized pro-resolving mediators (SPM's), which actively down-regulate the immune response (2), a process referred to as the resolution of inflammation. Mis-regulation of the transition to resolution can extend the early beneficial effects of acute inflammation into the damaging effects of chronic inflammation. This can contribute to cardiovascular disease (3)(4)(5), diabetes (6,7) and autoimmune disorders (8), to name a few examples.
In conjunction with the detection of MaR1 in cell extracts, Serhan Figure 1) (17). These three analogues were shown to affect macrophage chemotaxis, but were less potent than MaR1. In particular, 7S,14S-diHDHA appeared to be biosynthesized in a di-oxygenation fashion without an epoxide intermediate, distinct from the other analogues. Its structure suggests that it is synthesized through two sequential LOX oxygenation reactions at C7 and C14, without the requirement of a hydrolase to open the epoxide intermediate (19)(20)(21). Given these oxygenation sites, as well as the enzymatic preferences of specific LOX isozymes, it is reasonable to assume that the biosynthesis of 7S,14S-diHDHA involves h5-LOX oxidizing DHA at C7, with h12-LOX oxidizing at C14, but it is unclear which oxidation occurs first is a D-series resolvin made from DHA (22), and has been identified in human blood, hemorrhagic exudates and synovial fluid (23,24). RvD5 enhances phagocytosis in neutrophils and macrophages and reduces expression of the pro-inflammatory molecules, NF-κB and TNF-α (25). Like 7S,14S-diHDHA, the production of RvD5 appears to require two oxygenation events and is proposed to occur through the sequential reaction of two lipoxygenases (24,26). Since many cell types contain only a single lipoxygenase, oxylipins with multiple oxygenation sites are often created by the interaction of multiple cell types in an area of inflammation through a process called transcellular biosynthesis (27-29). RvD5 has been isolated from neutrophils (30) and is produced late in the process of blood coagulation, increasing over the lifetime of a clot (31). PMN (24), suggesting that h5-LOX reacts with 17-HDHA to produce RvD5.
As stated above, LOX isozymes play a key role in the biosynthesis of SPMs, however, the biosynthetic molecular mechanism for specific SPMs is poorly defined. For 7S,14S-diHDHA, the current understanding of its biosynthetic pathway is based on the substrate specificity of h5-LOX and h12-LOX with AA, and experimental evidence that the non-specific LOX inhibitor, baicalein, reduces the production of MaR1 and 14(S)-dihydroxy-4Z,7Z,10Z,12E,16Z,19Z-docosahexaenoic acid (14S-HDHA) (32). However, baicalein is not a selective inhibitor (33) and it is well recognized that LOXs can produce different products depending on the fatty acids or oxylipin used (34). For example, h5-LOX produces primarily 5(S)-hydroperoxy-6E,8Z,11Z,14Z-eicosatetraenoic acid (5S-HpETE) from AA, but it produces multiple products when DHA is the substrate (35), presumably due to the length and unsaturation differences between AA and DHA, which presumably affects substrate positioning for hydrogen atom abstraction. LOX's also vary widely in their ability to tolerate oxygenated substrates, as reflected in their kinetic parameters (19). In addition, the specific hydrogen atom abstracted can be affected by the nature of the substrate (34). h5-LOX abstracts a hydrogen atom from C7 of AA to produce 5S-HpETE, but it can subsequently abstract from C10 from 5S-HpETE to produce the 5,6-epoxide (36). This variability in product distribution is also seen with h15-LOX-1, where approximately 10% of the product made from AA is 12S-HpETE, the non-canonical product. Kuhn and coworkers proposed that this lack of specificity is linked to human evolution and specifically the human inflammatory response, with only higher primates having this LOX function (37). This ability of h15-LOX-1 to oxygenate either C15 or C12 on AA raises the possibility that h15-LOX-1 could generate 7S,14S-diHDHA from DHA (38), as shown in Scheme 1, Pathway 3.
Given the biological importance of SPMs, such as 7S,14S-diHDHA and 7S,17S-diHDHA (RvD5), and the multiple possible biosynthetic pathways for their production, we investigated two critical aspects of the biosynthetic routes to producing these two SPMs: the ease of which LOXs can perform the proposed reactions in vitro (i.e. kinetics), and the specificity of LOXs when the substrate is an oxylipin, as opposed to the un-oxidized fatty acid. We probed these effects in the production of both 7S,14S-diHDHA and 7S,17S-diHDHA (RvD5) and determined that not only are the rates greatly affected by the nature of the oxylipin substrate, but also the product profile is altered, raising the possibility that unrecognized, non-canonical LOXs may be involved in SPM production.

Expression and Purification of h15-LOX-1, h15-LOX-2, h12-LOX, and h5-LOX. Overexpression and
purification of wild-type h15-LOX-1 (Uniprot entry P16050), h12-LOX (Uniprot entry P18054), h5-LOX (Uniprot entry P09917) and h15-LOX-2 (Uniprot entry O15296) were performed as previously described (78)(79)(80). The purity of h15-LOX-1 and h12-LOX were assessed by SDS gel to be greater than 85%, and metal content was assessed on a Finnigan inductively-coupled plasma-mass spectrometer (ICP-MS), via comparison with iron standard solution. Cobalt-EDTA was used as an internal standard. The wt h5-LOX used in the kinetics of this work was not purified due to a dramatic loss in activity and was therefore prepared as an ammonium sulfate-precipitate.  . 7S,17S-diHDHA has an absorbance max of 245 nm, however, due to overlap with the substrate peak at 234 nm formation of this product was measured at 254 nm using an extinction coefficient of 21,900 M -1 cm -1 to adjust for the decreased rate of absorbance change at this peak shoulder (19). KaleidaGraph (Synergy) was used to fit initial rates (at less than 20% turnover), as well as the second order derivatives   (PDB). Therefore, we constructed a homology model of h15-LOX-1 from its sequence (Uniprot ID: P16050) using the substrate mimetic inhibitor bound, high-resolution structure of porcine 12-LOX (PDB ID: 3rde). The h15-LOX-1 sequence is 86% identical to the porcine 12-LOX sequence, with both being considered ALOX15 genes. The homology model was constructed using Prime software (version 5.4, Schrodinger Inc). During the modeling step we retained the co-crystallized inhibitor, metal ion (Fe 3+ ), and a hydroxide ion that coordinated the metal ion from the homolog structure. The model was subsequently energy minimized using Protein Preparation Wizard module of Maestro (version 11.8, Schrodinger Inc).
During this step, hydrogen atoms were added to the protein, co-crystallized ligand and the hydroxide ion.
Hydrogen atoms of the titratable residues were adjusted and side chains of Tyr, Thr, Ser, Asn and Gln were optimized so they could make better hydrogen bonding interactions. The structure was finally For h15-LOX-2, structure-based docking calculations were performed using the available crystal structure of human ALOX15b (PDB id: 4nre). From the crystal structure, we retained the protein, Fe 3+ ion, co-crystallized inhibitor, polyoxyethylene detergent (C8E), present in the active site and a hydroxide ion that co-ordinates the metal ion, all other atoms were deleted. Prior to docking, the structure was subjected to the protein-preparation step using Maestro's Protein-Preparation Wizard (Maestro version 11.8, Schrodinger Inc). Although the co-crystallized inhibitor bound in a U-shaped binding mode, it did not have a carboxylate group, and it had only 21 atoms in the main chain as opposed to 22 atoms present in DHA or 7S-HDHA. Therefore, we performed flexible-receptor flexible-ligand docking using InducedFit docking software (Schrodinger Inc). During induced-fit docking only the following active site residues were treated flexibly: Phe365, Val426, Val427, Arg329 and Asp602.

Biosynthesis of 7S,14S-diHDHA
As discussed in the introduction, there are three proposed routes by which 7S,14S-diHDHA can be synthesized by LOXs (Scheme 1, Pathways 1-3). These three pathways were therefore investigated in vitro to implicate possible in vivo biosynthetic routes. It should be emphasized that since these are dioxygenation pathways, the di-hydroperoxide is formed in each of the pathways, however the biologically isolated molecule, 7S,14S-diHDHA, is the reduced form due to cellular glutathione peroxidases.

Pathway 1: DHA with h12-LOX to 14S-HpDHA, then h5-LOX to 7S,14S-diHpDHA
The first possible route for biosynthesis of 7S,14S-diHDHA involves oxygenation of DHA by h12-LOX to form 14S-HpDHA (Scheme 1, Pathway 1). In order to study the ability of h12-LOX to form 14S-HpDHA, steady state kinetics and product profiles were measured. The kcat for h12-LOX with DHA was measured to be 14 sec -1 , while the kcat/KM was found to be 13 sec -1 µM -1 ( Table 1), demonstrating that DHA is a comparable substrate to AA for h12-LOX in vitro (43,44). The product profile indicates that the majority of oxylipin is oxygenated at C14 (greater than 80%), with minor amounts oxygenated at C11, as previously reported (45,46).
After DHA reacts with h12-LOX to form 14S-HpDHA, it may subsequently react with h5-LOX to form 7S,14S-diHpDHA (Scheme 1, Pathway 1), the oxidized form of 7S,14S-diHDHA. In order to study the ability of h5-LOX to react with 14S-HDHA and 14S-HpDHA, steady-state kinetic values and product profiles were assessed but no activity was observed by UV-vis spectroscopy (i.e. no change in absorbance at 254, 270 or 302 nm being detected). Using the more sensitive LC-MS/MS method, the Vmax values for 14S-HDHA and 14S-HpDHA were determined to be 0.00038 and 0.0015 (mol/sec -1 mol -1 ) at 10 µM substrate, respectively (Table 2). If we compare the Vmax at 10 µM of DHA (vide supra), the Vmax values of 14S-HDHA and 14S-HpDHA are 368-fold and 93-fold slower than that of DHA, indicating that 14-oxylipins are poor substrates of h5-LOX. Even with this lowered kinetic rate, the major product in both reactions was the 7S,14S-oxylipin, along with trace amounts of 8,14-oxylipin and tri-oxygenated products.

Pathway 2: DHA and h5-LOX to 7S-HpDHA, then h12-LOX to 7S,14S-diHpDHA (oxidized form of 7S,14S-diHDHA)
The second possible route for biosynthesis of 7S,14S-diHDHA begins with oxygenation of DHA by h5-LOX to form 7S-HpDHA (Scheme 1, Pathways 2 and 3). The ability of h5-LOX to form products from DHA was assessed through kinetic measurements and product profiles. The Vmax for DHA at 10 µM was determined to be 0.14 mol/sec -1 mol -1 (Table 2), similar to the reported value of 0.14 mol/sec -1 mol -1 for AA (converted from kcat and KM at 10 µM AA in reference (34)). Interestingly, the reaction of h5-LOX with DHA was found to produce 8 different products, with 7S-HpDHA being the major product at 52%, and 7 minor products comprising the remaining 48% (Table 3 and Supporting Information, Figure 1S).
The 4-product was not observed, consistent with results obtained from h5-LOX in neutrophils (35, 47) and monocytes (48). It is important to note that the increased non-specificity of h5-LOX with DHA relative to AA is a common observation for other LOX isozymes as well (vide infra). This appears to be a function of the structural difference between DHA and AA, with the increased length and unsaturation of DHA leading to a substrate-binding mode that promotes greater product diversity relative to AA.
To test this possibility, steady-state kinetic values were determined for the reaction of h12-LOX with 7S-HDHA and 7S-HpDHA. The kcat for 7S-HDHA and 7S-HpDHA were found to be 3.3 sec -1 and 0.68 sec -1 , respectively and the kcat/KM's for 7S-HDHA and 7S-HpDHA were found to be 1.8 sec -1 µM -1 and 0.25 sec -1 µM -1 , respectively (Table 1), which are markedly greater than the kinetic parameters of h5-LOX with the 14-oxylipins. The products of h12-LOX reacting with 7S-HDHA were exclusively dioxygenated oxylipins, with 82% of 7S,14S-diHDHA and 18% of 7S,17S-diHDHA being generated (Table 4 and Supporting Information, Figure 2S). The production of 7S,17S-diHDHA is a remarkable result since h12-LOX is known to produce mostly 14S-HpDHA and only a minor amount of 11-HpDHA from DHA. The formation of 11-HpDHA indicates that DHA inserts deeper into the cavity for hydrogen atom abstraction at C9. However, with 7S-HDHA as the substrate, the minor product is 7S,17S-diHDHA, not 7S,11S-diHDHA, suggesting that 7S-HDHA does not enter as deep into the active site, allowing for abstraction at C15 and the generation of 7S,17S-diHDHA. The reaction of 7S-HpDHA with h12-LOX produced comparable products as that from 7S-HDHA, indicating that no dehydration occurred to form the epoxide or its derivatives (Table 4).
h15-LOX-1 was reacted with DHA, 7S-HDHA and 7S-HpDHA, with 7S-HDHA demonstrating a comparable rate to that of DHA, but 7S-HpDHA showing a slower rate, with kcat/KM values of 0.68, 0.51 and 0.08 sec -1 µM -1 for DHA, 7S-HDHA and 7S-HpDHA, respectively (Table 1). Even more surprising, the kcat value for 7S-HDHA was over 3-fold greater than that for DHA. These results indicate that DHA is still the more efficient substrate at low concentration (i.e. the fastest rate of substrate capture, kcat/KM) (49), but 7S-HDHA is an alternative substrate to DHA at high concentration, having the fastest rate of product release (kcat) (49).
To determine the oxylipins generated by h15-LOX-1 from 7S-HDHA and 7S-HpDHA, the reaction products were isolated and analyzed via LC-MS/MS. Surprisingly, when h15-LOX-1 was reacted with 7S-HDHA, 90% of the products were 7S,14S-diHpDHA (i.e. the oxidized form of 7S,14S-diHDHA), with only 10% of 7S,17S-diHpDHA being produced (Table 5 and Supporting Information, Figure 3S). Along with the primary di-HDHA products, minor amounts of tri-HDHA secondary products were produced in the reaction, representing less than 4% of the total and comprising mainly 7,16,17-triHDHA, indicating that the low ratio of 7S,17S-diHDHA to 7S,14S-diHDHA was not due to formation of tri-HDHA products. When 7S-HpDHA was used as the substrate, the products were comparable, with

Chiral chromatography characterization of 7S,14S-diHDHA and 7S,17S-diHDHA (RvD5).
We predicted that the oxygen on C14 of 7S,14S-diHDHA is in the S-configuration because h15- Together, these two observations indicate that while the major product of h5-LOX is the S-configured product, a small amount of the R-product is also made.
Along with chiral retention times, UV spectra were compared to further differentiate the double bond geometry of the various products. All 7,14-diHDHA products contained a central peak at ~270 nm, surrounded by shoulders at ~260 nm and ~280 nm, consistent with the presence of a conjugated triene. In Mar1 and 7-epi-MaR1, the shoulders at 281 nm and 261 nm were of equal intensity, indicative of a conjugated triene with EEZ configuration (50)(51)(52). In contrast, the 7,14-HDHA produced enzymatically showed a more intense shoulder at 260 nm than at 280 nm, indicating the EZE configuration (53). It should be noted that we have attempted to determine the absolute stereochemistry of 7S,14S-diHpDHA by generating the double Mosher derivative, but unfortunately low yields from enzymatic synthesis and NMR peak overlap prohibits the assignment of the molecules' stereochemistry. To

Molecular Modeling of DHA and 7S-HDHA bound to h15-LOX-1
The shift in product profile seen between h15-LOX-1 reacting with DHA and 7S-HDHA suggests that 7S-HDHA has a different binding mode than DHA, and therefore molecular modeling was employed to assess this possibility. Extra-precision Glide scores were used to approximate ligand binding free energy, with lower negative scores representing tighter binding (54). Figure  Ile592. However, a key difference is that the C7 hydroxyl group of 7S-HDHA forms a hydrogen bond with the backbone carbonyl oxygen of Ile399. This difference in binding between DHA and 7S-HDHA is manifested in their distances between the iron-hydroxide oxygen atom and the hydrogen on the reactive carbons, C12 and C15. The modeling data indicate that for 7S-HDHA, the C12 pro-S hydrogen is markedly closer (2.6 Ang) to the iron-hydroxide moiety than C15 hydrogen (5.9 Ang) ( Table 9).
Considering that C12 hydrogen atom abstraction leads to 7S,14S-diHDHA, while C15 hydrogen atom abstraction leads to 7S,17S-diHDHA, these docking results are consistent with the enzymatic results. For DHA, the distance for the C15 pro-S hydrogen (3.8 Ang) is slightly shorter than that of C12 (4.1 Ang), consistent with the experimental results, but the distance difference is not as distinct as that for 7S-HDHA, possibly due to the more homogeneous hydrophobic nature of DHA compared to 7S-HDHA.
Interestingly, there is no difference between the pro-S and pro-R hydrogens indicating the limitation of the docking model.  (Table 10). These distances are shorter than the pro-S hydrogen of C12 and the oxygen atom of the hydroxide ion for DHA and 7S-HDHA (7.2 Å and 4.0 Å, respectively) (Table 10). These two binding modes are consistent with the enzymatic reaction involving abstraction of the pro-S hydrogen from C15 for both DHA and 7S-HDHA, to produce the 17-product exclusively (Table 10) (55-59) and not the 14-product from C12 hydrogen abstraction.

Effect of 7S,14S-diHDHA and RvD5 on Platelets
Previous studies have shown that DHA and the h12-LOX-derived oxylipins of DHA have antiplatelet effects (60). Therefore, to determine the effect of RvD5, 7S,14S-diHDHA and related maresin isomers on platelet activation, washed platelets were treated with oxylipins in half-log increments and

Platelet Lipidomics with 7S-HDHA
The reduction in aggregation seen in platelets incubated with 7S-HDHA could be caused by either the 7S-HDHA itself or by a biosynthetic product produced by platelets during incubation. To investigate the latter possibility, platelets were incubated with 7S-HDHA and analyzed for the presence of 7S-HDHA metabolites using LC-MS/MS. Although large amounts of unreacted 7S-HDHA were detectable in the samples, no levels of any di-HDHA's or tri-HDHA's were detectable. As a control, platelets incubated with 5S-HETE were also analyzed for the presence of 5S-HETE metabolites.
Although large levels of 12-HETE were detectable from endogenous AA, no appreciable levels of 5S-HETE/ h12-LOX metabolites were detectable down to 2 ng/mL. This is an unexpected result since h12-LOX is capable of reacting with 7S-HDHA in vitro to produce 7S,14S-diHDHA. The fact that no 7S,14S-diHDHA is produced indicates both that 7S-HDHA is most likely the bioactive species and that 7S-HDHA does not react appreciably with h12-LOX in the cellular milieu of the platelet.

Biosynthesis of 7S,14S-diHDHA
7S,14S-diHDHA is an analogue of MaR1 (17) and is proposed to have a distinct bio-synthetic pathway from MaR1. Its structure suggests three possible routes for its biosynthesis. Pathway 1 involves that oxidation of DHA at C14 by h12-LOX, followed by oxidation at C7 by h5-LOX, however, the in vitro data of this work suggests this is a very unfavorable biosynthetic route. Although DHA is a good substrate for h12-LOX, forming 14S-HpDHA at a rate similar to 12S-HpETE formation from AA (44), the second step in this pathway is kinetically unfavorable. Compared with DHA, the Vmax of h5-LOX with 14S-HDHA is ~368 fold slower (Table 2).
For Pathway 2 in Scheme 1, DHA is oxidized at C7 by h5-LOX followed by oxidation at C14 by h12-LOX. In comparison to Pathway 1, Pathway 2 is a kinetically favorable in-vitro biosynthetic route. These results indicate that, although the hydroxyl group on C7 has a large effect on the positional specificity of the enzyme, it has little effect on catalysis, leading to a favorable overall kinetic condition.
In summary, if we consider the product of both the rate and the percent product formation (i.e. the biosynthetic flux, Scheme 2), the data indicates that the most efficient pathway for making 7S,14S-diHDHA is thru h5-LOX and then h12-LOX or h15-LOX-1 and not vice versa since the reaction of h5-LOX with 14-HpDHA is markedly slower and hence rate-limiting.
Although changes in the positional specificity of LOXs have been demonstrated before (19), the finding that h15-LOX-1 primarily oxygenates C17 on DHA, but C14 on 7S-HDHA is surprising. In order to obtain a better understanding of this binding event, DHA and 7S-HDHA were docked to a model of h15-LOX-1. In the case of DHA, both C12 and C15 are at a reasonable distance for hydrogen abstraction (65), whereas in case of 7S-HDHA, C12 is closer to the metal ion than C15, supporting the experimental results ( Table 9

Biosynthesis of 7S,17S-diHDHA (RvD5)
The discovery of Pathway 3 as a relevant route to the biosynthesis of 7S,14S-diHDHA raises questions with respect to the biosynthetic route to making 7S,17S-diDHA (RvD5). Previously, 7S,17S-diDHA (RvD5) was proposed to be biosynthesized from DHA in two steps; initial oxygenation of C7 by h5-LOX, followed by oxygenation of C17 by h15-LOX-1 (24,26). This hypothesis was based on the prevalence of h15-LOX-1 in the inflammasome and the preference of h15-LOX-1 to oxygenate DHA at C17. However, our work shows that h15-LOX-1 shifts its positional specificity when reacting with 7S- than AA (15). Considering that hydrogen atom abstraction is generally the first irreversible ratedetermining step for kcat and kcat/KM for fatty acid substrates, these results suggest that the rate of oxylipin binding is markedly slower relative to fatty acid binding for h15-LOX-2 (15).
With respect to the biosynthetic pathway of RvD5, the fact that RvD5 contains two hydroxyls on C7 and C17 suggests three likely routes for the biosynthesis of RvD5 from DHA (Pathways 4, 5 and 6, Scheme 1). The first route could be the oxygenation at C7 by h5-LOX followed by oxygenation of C17 by a 15-LOX isozyme (Pathway 4 or 5), either h15LOX-1 or h15-LOX-2. The second route is the reverse, with oxygenation of C17 by a 15-LOX isozyme, followed by oxygenation of C7 by h5-LOX (Pathway 6).
In the current study, we find that h5-LOX reacts well with DHA, however, it reacts significantly slower with 7S-HDHA and 17S-HDHA. For a direct comparison, the Vmax at 10 µM substrate was determined to be 0.14 mol/sec -1 mol -1 for h5-LOX with DHA (Table 8), which is over 100 times greater than the rate observed with 17S-HDHA (Vmax = 0.0012 mol/sec -1 mol -1 at 10 µM). This is consistent with the low reactivity of h5-LOX and 14S-HDHA, indicating that h5-LOX does not react well with oxygenated derivatives of DHA and brings into question Pathway 6 as a viable in vivo biosynthetic route. Pathway 4 also appears unfavorable since the current work indicates that h15-LOX-1 reacts with 7S-HDHA to produce 90% 7S,14S-diHDHA (vide supra). It therefore appears that h15-LOX-1 may not contribute much to the in vitro biosynthesis of RvD5. h15-LOX-2, however, reacts well with 7S-HDHA (Vmax = 1.2 mol/sec -1 mol -1 at 10 µM) ( Table 8), indicating that the rate of oxygenation at C7 by h5-LOX followed by oxygenation of C17 by h15-LOX-2 is a markedly faster pathway than the rate of oxygenation of C17 by h15-LOX-2 followed by oxygenation of C7 by h5-LOX. Therefore, if we calculate the biosynthetic flux (i.e. the rates multiplied by the percent product formation (Scheme 2)), it is apparent that the preferred pathway of RvD5 production in vitro is thru h5-LOX and then h15-LOX-2.
Previously, it was shown that h15-LOX-2 oxygenated 5S-HpETE, but not 12S-HpETE or 15S-HpETE (15). Together with the current work, this suggests that the h15-LOX-2 active site tolerates substrates which are oxygenated close to the carboxylate end, but not those substrates oxygenated closer to the methyl end. This is consistent with the substrate-binding model, where the methyl-end of the substrate binds at the bottom of the active site (64,67,68), allowing for the abstraction of a hydrogen atom from C15 and the subsequent oxidation at C17. In order to understand this binding event in more detail, DHA and 7S-HDHA were modeled into the active site of h15-LOX-2. Both substrates bind in a tail first, U-shaped binding mode with their carboxylate groups forming a salt-bridge with R429 on the helix a12. Their hydrophobic tails are buried deep in a hydrophobic pocket created by residues F365, L420, I421, V427, F438 and L607 (Figure 4), and the reactive pro-S hydrogen of C15 of 7S-HDHA is 2.7Å from the active site hydroxide-ferric moiety. This binding mode is consistent with the abstraction of the pro-S hydrogen atom from C15 to produce 7S,17S-diHpDHA. Parenthetically, the pro-S hydrogen is also abstracted in the formation of 12-HETE, 5-HETE, and 15-HETE by h12-LOX, h5LOX and h15-LOX-1 respectively (55-59).

Biological Consequences
If we consider that the in vitro biosynthetic pathways outlined above for 7S,14S-diHDHA and biosynthesis. This is especially relevant since we did not observe platelets producing 7S,14S-diHDHA when given 7S-HDHA, therefore, the cellular milieu may affect h12-LOX activity since we observe 7S,14S-diHDHA in vitro. We are currently investigating these biosynthetic pathways in platelets and macrophages in more detail to identify both the pathway and the cell type that makes these oxylipins.
Second, if h15-LOX-2 is the primary enzyme involved in the in vivo RvD5 biosynthesis, then the role of h15-LOX-2 in human disease could be larger than expected. It is known that h15-LOX-2 converts the h5-LOX-derived pro-inflammatory oxylipins, 5S-HETE and 5,6-diHETE, into pro-resolving molecules, such as lipoxins, in vitro (75). With the knowledge that h15-LOX-2 also efficiently produces RvD5 from 7S-HDHA, it could be that h15-LOX-2 is important in mediating the switch between inflammation and resolution. For example, neutrophils express both h5-LOX and h15-LOX-2, so it is conceivable that neutrophils produce RvD5 independently, without the need for a transcellular synthesis mechanism. However, it should be noted that RvD5 production was observed by exposing isolated neutrophils to 17S-HpDHA (41), suggesting h5-LOX can produce RvD5 in neutrophils. This result is in contrast to our in vitro results and suggests that there may be a factor in the neutrophils, similar to the role of 5-lipoxygenase activating protein (FLAP), which could increase the h5-LOX activity with 17S-HDHA (76). We are currently investigating this discrepancy further by investigating neutrophils with our selective LOX inhibitors.
With respect to blood coagulation, this work shows that RvD5 has micromolar potency in inhibiting platelet activation, indicating that h15-LOX-2 may also play a role in hemostasis. RvD5 levels are reduced in blood treated with anticoagulants in vitro (77) and RvD5 formation does not occur during initial platelet activation, however, it is produced in later stages of clot progression (77). Given our result that RvD5 reduces platelet activation, the production of RvD5 could be a signal to diminish the clot size.
Given that platelets do not produce h15-LOX-2 and thus cannot make RvD5, these results suggest that RvD5 is formed by h15-LOX-2 in macrophages and/or neutrophils, which migrate into the area of latestage clots and increase their RvD5 production by increasing the h15-LOX-2 expression, and not that of h15-LOX-1 (4). We are currently investigating the role of h15-LOX-2 in more detail with our specific/potent h15-LOX-2 inhibitors (78) with the hope of understanding the correct biosynthesis pathway of RvD5 and hence its role in coagulation and atherosclerotic disorders.

CONCLUSION
7S,14S-diHDHA is synthesized in vitro by the sequential reactions of h5-LOX and h12-LOX with DHA. However, we have also discovered a novel alternative biosynthetic pathway for the production of 7S,14S-diHDHA, which involves h15-LOX-1. The alcohol on 7S-HDHA changes the binding position of the oxylipin in the active site, thus altering the product specificity. This non-canonical result has farreaching ramifications. It first indicates a possible alternative biosynthetic pathway for 7S,14S-diHDHA, but more importantly, the result indicates that our knowledge of LOX reactivities with fatty acids, may not extend to oxylipins, and therefore further study is needed to determine the exact biosynthetic pathways for each oxylipin, such as lipoxins, protectins and maresins. Importantly, these results suggest that LOX-products observed during in-vivo studies may originate from non-canonical biosynthetic routes involving previously overlooked cell types.
In addition, this result suggests that the biosynthesis of 7S,17S-diHDHA (RvD5) is achieved with h15-LOX-2 and thus its role in human disease may be more important than previously suspected. These   Table 3. Distribution of products created by reaction of h5-LOX with DHA. 7S-HpDHA is the major product, at 52% of the total and 7 minor products comprise the remaining 48%.
Note, the abbreviation, "14-product" etc., is used to simplify the complexity of the table labels. *A peak of approximately 3% the total area was observed that had a parent mass of a di-HDHA, however its exact nature was undetermined due to its small area.  Table 8. Vmax values of reaction steps in the biosynthesis of RvD5 were calculated at 10 µM substrate concentration for comparison. *Biosynthetic flux is calculated by multiplying each Vmax. by the percentage of total product from that reaction that serves as substrate for the next step in the synthesis of 7S,17S-diHDHA. Reactions with DHA represent the first of two biosynthetic steps, while reactions with oxylipins represent the second.