Regioselective epoxidation of a polyunsaturated fatty acid by a vanadyl complex with redox-active ligands Authors: Andrew G. Hill, Tiffany Y.-C. Tang, Christina M. Knight, Kaitlyn S. Engle, Maximilian G. Bernbeck, Jake D. Soper, Aditi Das Published: 2026 ABSTRACT Reaction of docosahexaenoic acid (DHA) with the oxovanadium complex [(Phisq)(Phibq)V(O)Cl] produces exclusively 19,20-epoxydocosapentaenoic acid (19,20-EDP). Computations reveal that the observed regioselectivity is not attributable to thermodynamics. New vanadyl analogs [(dippisq)(dippibq)V(O)Cl] and [(dippisq)2V(O)] were synthesized and showed no selectivity in DHA epoxidation. Epoxides of long-chain polyunsaturated fatty acids (PUFAs), produced in vivo by cytochrome P450 (CYP) enzymes, play an important role in cardiovascular health.1 19,20-Epoxydocosapentaenoic acid (19,20-EDP), in particular, is the main metabolite of docosahexaenoic acid (DHA) and has been demonstrated in preclinical studies to display cardioprotective, anti-inflammatory, and anti-angiogenic properties.2–6 Studies of DHA metabolites as lipid mediators and drug candidates are significantly hampered by the difficulty of obtaining sufficient quantities of pure regioisomers.7,8 The multiple C Created by potrace 1.16, written by Peter Selinger 2001-2019 C double bonds present in polyunsaturated fatty acids make most chemical synthesis neither regiospecific nor stereospecific (Fig. 1a).9 For instance, the epoxidation of DHA by, e.g., meta-chloroperbenzoic acid (m-CPBA) readily forms all EDP regioisomers and polyepoxidated products, which must be laboriously separated by chromatography.9 Cinelli and Lee obtained single stereoisomers of 19,20-EDP via a 13-step synthesis method using asymmetric epoxidation and Wittig condensation (Fig. 1b),8 but, to our knowledge, no synthetic method is known for selective conversion of DHA into a single EDP isomer in one step. Cytochrome P450 enzymes (CYPs) capable of catalyzing selective epoxidation of long-chain polyunsaturated fatty acids include, but are not limited to, CYP2C8, 2C9, and 2J2 metabolizing ω-6 arachidonic acid (AA) and ω-3 docosahexaenoic acid (DHA) and eicosapentaenoic acid (EPA).10–12 The CYP epoxidations are regio- and stereoselective, giving up to 90% selectivity for 19,20-EDP (Fig. 1c).7 Survey of reported methods for the preparation of 19,20-epoxydocosapentaenoic acid. (a) Conventional epoxidation by m-CPBA. (b) Total synthesis reported by Lee and coworkers. (c) CYP-catalyzed epoxidation reported by Lee and coworkers. (d) One-step method described in this study. Site-selective epoxidations of substrates that present inherent challenges to selectivity, such as polyenes13–15 and allylic alcohols,14,16–18 have been accomplished by tight steric control of the transition state in transition metal-mediated16–19 or intramolecular13,15 oxidations. All reported substrate-directed epoxidations are proposed to proceed through preorganization of the reactants.20–22 While there are now several examples of selective epoxidation proximal to a directing group, non-proximal CC linkages remain challenging targets with only a handful of published demonstrations.13,14,20 Vanadium compounds catalyze a broad range of organic transformations, including polymerizations23,24 and many oxidative transformations of small molecules,25–27 including epoxidations.28–31 We recently reported a vanadyl compound with an anomalously weak V Created by potrace 1.16, written by Peter Selinger 2001-2019 O bond, which is capable of stoichiometric and aerobic catalytic O-atom transfers to small molecule alkenes.32 We hypothesized that steric direction from the flanking aryl groups and distinct hydrophobic/hydrophilic regions might promote specific conformational interactions with long-chain PUFAs, leading to site-selective epoxidation of DHA. To that end, we prepared two additional vanadyl complexes with more sterically demanding ligands. Their reactivity towards DHA is presented herein, including a demonstration of regioselective direct epoxidation, forming exclusively 19,20-EDP. The previously reported vanadyl compound [(Phisq)(Phibq)V(O)Cl] (I)32 ([Phisq]2− = 2,4-di-tert-butyl-6-(phenylimino)semiquinonate, [Phibq˙]− = 2,4-di-tert-butyl-6-(phenylimino)benzoquinone) (Fig. 2, I) was reacted with 1 equiv. of docosahexaenoic acid (DHA) for 24 h at 50 °C in toluene under N2 (10 mM). The reaction mixture was purified by liquid chromatography. The chromatogram (Fig. 3, I) showed formation of a single product peak (32 min), followed by unreacted DHA (40 min). The 32-minute fraction was collected and characterized using NMR and mass spectrometry (Fig. S4–S7, S9 and S10). All data are consistent with assignment as 19,20-EDP based on NMR8 and MS2 fragmentation,9 in 46% isolated yield. Lengthening the reaction time to 1 week did not appreciably increase the yield of 19,20-EDP and led to formation of additional degradation products (Fig. S15). Structures of complexes I–IV. ORTEP plots (50% probability thermal ellipsoids) and selected bond lengths and metrical oxidation states of II and III. LC–UV chromatograms of the reaction mixtures of DHA with I–IV. The thermodynamics of DHA epoxidation were examined by DFT (B3LYP,33 D3BJ,34 and def2-TZVPP).35,36 DFT calculations on the six EDP regioisomers suggest that all epoxides are within 5 kcal mol−1 of each other (Table S1). Accordingly, I is sufficiently oxidizing to form all EDP isomers, and the selectivity observed in the reaction between I and DHA has no thermodynamic basis. We speculated that selectivity could instead derive from kinetic access to the vanadyl oxido functionality, potentially from an organized [I·DHA]‡ transition state. To probe this hypothesis, two analogues of I were prepared with varying steric constraints about the oxido ligand. Addition of two equiv. of in situ generated Na(dippisq) ([dippisq˙]− = 2,4-di-tert-butyl-6-(2,6-diisopropylphenylimino)semiquinonate) to a THF solution of VOCl3 resulted in an immediate color change to dark brown II. The solvent was removed in vacuo and the resulting solids were extracted with benzene and filtered through Celite. The solvent was removed again to give II as a brown-black solid (99%). Crystals of II suitable for X-ray diffraction were grown by vapor diffusion of pentane into a THF solution (Fig. 2, II). Complex II is pseudo-octahedral and isostructural with I. The V–O bond length of II (1.597(6) Å) is statistically indistinguishable from that of I. The metrical data support assignment of II as vanadium(iv) with one radical iminosemiquinonate ligand [dippisq˙]− (MOS = −1.08 ± 0.11) and one iminobenzoquinone ligand [dippibq]0 (MOS = 0.03 ± 0.04).37 Addition of 1 equiv. of Na[C10H8] to a THF solution of II effected a color change to dark green. After stirring for 2 h, the solvent was removed in vacuo. The solids were extracted with pentane, filtered through Celite, and then concentrated to give III as a green-black solid (99%). Crystals suitable for X-ray diffraction were grown from hexane (Fig. 2, III). III is pseudo-square pyramidal (τ5 = 0.09) with equatorial iminosemiquinonate ligands and an axial oxido ligand. The V–O bond length of III (1.602(1) Å) shows no significant elongation relative to II. Both of the crystallographically inequivalent ligands show bond metrics consistent with assignment as iminosemiquinonate radicals [dippisq˙]− (MOS = −1.17 ± 0.11, −0.91 ± 0.07).37 The solution magnetic moment of μeff = 1.9μB (C6D6, 25 °C) is consistent with one unpaired electron. The metrical and magnetic data suggest an electron configuration with three spin centers: one d electron and two radical ligands, where two of these three spins are strongly antiferromagnetically coupled. In contrast to the remarkable selectivity of I, reactions of II with DHA performed under the same conditions showed no significant EDP formation (Fig. 3, II). III + DHA showed only minimal (<1%) product formation under N2 (10 mM), but a reaction under air at a 1 : 10 ratio (10 mM III, 100 mM DHA) gave multiple products (Fig. 3, III, Fig. S16). The MS2 fragmentation of the 26-minute fraction closely resembles reported mass spectra of the 19,20-EDP regioisomer (Fig. S11 and S12).9 MS2 revealed that the peak at 30 minutes comprised two co-eluting species whose fragmentation corresponded to previously reported 16,17-EDP and 10,11-EDP (Fig. S13).9 I–IV are stable at 50 °C in toluene solutions under N2 in the absence of a substrate, as evidenced by a lack of changes in the UV-vis spectra. Reaction of [(Phap)(Phisq)VIVCl] (IV), which lacks an oxo ligand, with DHA under N2 produced no EDP products (Fig. 3, IV). However, the previously reported reaction of IV with air to produce I implies that catalytic epoxidation of DHA with I or IV, and O2 as the oxidant, should be possible (Scheme S1). I + DHA (10 mM I, 100 mM DHA, 10% loading of I) was reacted in air for 28 days at 50 °C. The LC chromatogram showed complete consumption of DHA and a single product peak at 19 minutes (Fig. S17). This retention time is consistent with a vic diol hydrolysis product of EDP,38 and the identity of the product as 19,20-dihydroxydocosapentaenoic acid (DiHDP) was confirmed by MS1 and MS2 (Fig. S14). This result demonstrates that I is a viable catalyst for regioselective aerobic generation of 19,20-DHDP (TON = 10), which has been shown to form from EDP upon gentle heating in the presence of H2O.38 To investigate the effects of PUFA length and ωCC bond position, experiments were carried out with I and arachidonic acid (AA) or eicosapentaenoic acid (EPA) under identical conditions. Neither of them showed regioselectivity in epoxide formation (Fig. S18 and S19). A reaction was also attempted using DHA ethyl ester, which yielded no epoxide products (Fig. S20). These results suggest that the acid group and the position of the terminal double bond in DHA are essential features of the observed reactivity. Reactions in CH3CN and CH2Cl2 also did not give epoxides, suggesting that the nonpolar solvent is important for selectivity (Fig. S21 and S22). To examine whether the observed selectivity is based solely on DHA pre-organization in toluene, a reaction of m-CPBA and DHA was performed using the conditions optimized for I. In contrast to typical unselective epoxidation of DHA in CH2Cl2,39 the reaction in toluene yielded no epoxide products (Fig. S23). 1H NMR of I + DHA in toluene-D8 (10 mM I and DHA) reveals no chemical perturbations of DHA other than broadening of the carboxylic acid peak into the baseline, indicating equilibrium protonation of I by DHA on the NMR timescale or hydrogen bonding (Fig. S8). One-step production of 19,20-EDP from DHA is, to our knowledge, without precedent. The remarkable selectivity in this system cannot be rationalized by thermodynamic considerations. I is a strong oxidant, and the olefinic sites in DHA are all comparable O-atom acceptors. The O-atom transfer thermodynamics of II and III have not been determined experimentally, but the properties that make I a strong oxidant are preserved in II. It follows that the divergent reactivity must reflect stereoelectronic differences in the vanadyl complexes. We previously reported I as an inseparable mixture of cis and trans isomers, which do not readily interconvert (Fig. S24).32 The moderate stoichiometric yields in reactions with I are potentially explicable by thermal instability of I, or by only one isomer being reactive toward DHA, especially considering the highly substrate-, oxidant-, and solvent-specific nature of the reaction. In this context, the bulky dipp groups of II should force a trans disposition of the N-donors; no isomerism has been observed in single crystals of II, and the 1H NMR spectrum is consistent with a single isomer, where the dipp groups crowd the oxo, which could prevent interaction with the substrate. The lack of reactivity between III and DHA can similarly be rationalized by increased bulk or by thermodynamic differences. Although all complexes herein are appropriately described as vanadium(iv), III is one formal oxidation state lower than I and II, making it less oxidizing. We have ascribed the ability of I to transfer its oxido ligand to the stabilization of the reduced product, IV, by the redox-active ligands, but a putative deoxygenated III may be inaccessible. It is possible that the presence of O2 in the air atmosphere during the reaction of III and DHA allows turnover by regenerating III, but the data presented here are insufficient to validate this hypothesis. The selective oxidation of DHA by I apparently requires highly specific conditions. The lack of reactivity with DHA ethyl ester could imply noncovalent tethering of the acid group to the vanadyl oxo, though it is unclear why this would promote distal reactivity, or acid co-catalysis may be necessary for the reaction. Additionally, no reaction was observed in polar organic solvents. We speculate that the reaction proceeds through a highly organized [I·DHA] transition state, which is sensitive to the polarity of the solution. Finally, reactions with other PUFAs were not selective for epoxidation at the terminal position, which suggests enzyme-like specificity of the oxidant to the substrate. Efforts to find selective oxidants for other lipids will require the development of new complexes, which is beyond the scope of this report. In sum, this report describes the first and only regio-exclusive one-step chemical epoxidation of docosahexaenoic acid (DHA). We propose that the selectivity could derive from a pre-organized DHA·vanadyl adduct that permits access to only one CC bond at the oxo ligand, which would constitute a rare demonstration of enzyme-like function without a protein scaffold. By enabling direct access to a single regioisomer, this method removes a major barrier to progress in lipid mediator research and opens new avenues for transition metal-mediated oxidations of complex lipids, which we are pursuing. Author contributions AGH: conceptualization, data curation, investigation, methodology, resources, validation, visualization, writing – original draft, and writing – review & editing; TY-CT: conceptualization, data curation, investigation, methodology, resources, validation, visualization, writing – original draft, and writing – review & editing; CMK: investigation, validation, visualization, writing – original draft, and writing – review & editing; KSE: investigation, resources, and writing – review & editing; MGB: investigation, resources, and writing – review & editing; JDS: conceptualization, funding acquisition, project administration, resources, supervision, validation, and writing – review & editing; AD: conceptualization, funding acquisition, project administration, resources, supervision, validation, and writing – review & editing. Conflicts of interest There are no conflicts to declare. Supplementary Material