Organic & Biomolecular Chemistry Stereoselective Synthesis of Protectin D1: a Potent Anti-inflammatory and Proresolving Lipid Mediator †

Stereoselective synthesis of protectin D1: a potent anti-infl ammatory and proresolving lipid mediator A convergent stereoselective synthesis of the potent anti-inflammatory, proresolving and neuroprotective lipid mediator protectin D1 (2) has been achieved in 15% yield over eight steps. The key features were a stereocontrolled Evans-aldol reaction with Nagao's chiral auxiliary and a highly selective Lindlar reduction of internal alkyne 23, allowing the sensitive conjugated E,E,Z-triene to be introduced late in the preparation of 2. The UV and LC/MS–MS data of synthetic protectin D1 (2) matched those obtained from endogenously produced material.


Introduction
Polyunsaturated fatty acids (PUFAs), such as docosahexaenoic acid (1, DHA), play a major role in the physiology of living organisms. 1 Recent efforts by the Serhan research group have established that DHA (1) is a substrate for the biosynthesis of several potent anti-inflammatory proresolving mediators, such as protectin D1 (2), 2 maresin 1, 3 resolvin D1 and resolvin D3. 2a,4 All of these compounds have enabled new research areas related to many disease states associated with inflammation. 5 It was reported that protectin D1 (2) is biosynthesized from DHA (1) via a lipoxygenase-mediated pathway that converts 1 by 15-lipoxygenase (15-LO) to the 17S-hydroperoxide intermediate (3), which is rapidly converted into the 16,17epoxide (4), followed by enzymatic hydrolysis to the antiinflammatory and proresolving oxygenated lipid 2 (Fig. 1). 6 This compound has been reported to exhibit strong in vivo protective activity in several inflammatory 6 as well as many other disease models. [7][8][9][10] For example, the oxygenated polyunsaturated fatty acid 2 protects the retina and the brain from oxidative stress with very potent agonist activities. 7 It is noteworthy that 2 was observed to be several orders of magnitude more potent in vivo than its precursor DHA. 2c Moreover, additional biological effects have recently been reported for this C22-oxygenated metabolite. 11 Hence, protectin D1 (2) is very interesting as a lead compound for the development of potential new anti-inflammatory drugs. 12 The prefix neuro is added when this oxygenated PUFA is formed by neural tissues. 2a As of today, two syntheses of protectin D1 (2) have appeared. 6,13 In connection with our interest in the synthesis of biologically active PUFA-derived natural products, 14 as well as the many interesting biological activities of protectin D1 (2), we decided to prepare the DHA derived product 2. A common structural feature for several of the lipid mediators isolated by the Serhan group 2-4 is the chemically unstable E,E,Z-triene connected to either one or two secondary allylic alcohols. In the retrosynthetic analysis of 2, Fig. 2, the aldehyde 6 is a key intermediate.
Aldehyde 8 was prepared by a slightly modified and improved literature protocol. 16 Commercially available pyridinium-1-sulfonate (14) was treated with aqueous potassium hydroxide at −20°C to yield glutaconaldehyde potassium salt 15 that was transformed further with the Br 2 /PPh 3 complex to (2E,4E)-5-bromopenta-2,4-dienal (8) in 41% yield over the two steps. This sensitive aldehyde was then reacted with thiazolidinone 9a, developed by Nagao and co-workers, 17 in an Evansaldol 18 type reaction using conditions developed by Olivo and co-workers (TiCl 4 , Et(i-Pr) 2 N, CH 2 Cl 2 , −78°C). 19 This smoothly produced the intermediate 16a in a 15.3 : 1 diastereomeric ratio as determined by HPLC and 1 H NMR analyses. We also investigated reactions using thiazolidinones 9b and 9c, with the phenyl and the benzyl group, respectively, which afforded 16b and 16c with lower diastereoselectivity (4.5 : 1 and 9.8 : 1). Purification by chromatography yielded diastereomeric pure 16a in 86% isolated yield. Protection of the alcohol functionality in 16a to compound 17 was achieved using standard conditions. 20 Then DIBAL-H-reduction of 17 in CH 2 Cl 2 at −78°C afforded the sensitive aldehyde 6 (Scheme 1).
Next, the Wittig-salt 7 was synthesized. The dianion of 4-pentynoic acid (11) in HMPA, 21 prepared by treatment with excess n-BuLi, was reacted with ethylene oxide (10). This afforded 7-hydroxy-hept-4-ynoic acid which was directly esterified to 18 (MeOH, catalytic H 2 SO 4 ), see Scheme 2. Reduction of the internal alkyne in 18 using the Lindlar reaction gave (Z)methyl 7-hydroxyhept-4-enoate (19) with high stereochemical purity as determined by 1 H NMR analyses. Then an Appel reaction 22 provided the iodide 20 which was treated with PPh 3 in acetonitrile to provide the Wittig-salt 7 in a total yield of 42% from 11. Conditions for the Z-stereoselective Wittig reaction between the key aldehyde 6 and the salt 7 were then investigated. Different bases, i.e. LiHMDS, KHMDS, NaHMDS, temperatures as well as altering the concentrations of 6 and 7, with or without different amounts of HMPA in THF, all resulted in lower Z-selectivity. The best result was obtained when aldehyde 6 and the ylide of 7, the latter obtained after treatment with NaHMDS in THF, were reacted at −78°C. This afforded the bromo-E,E,Z,Z-tetraene ester 21 (Scheme 2).
Chromatographic purification on silica gel yielded stereochemically pure product 21 (HPLC, 1 H-NMR) in 47% yield over two steps. Then alkyne 5 was reacted with 21 in a Sonogashira reaction 23 at ambient temperature in the presence of Pd-(PPh 3 ) 4 and CuI using diethyl amine as a solvent. This afforded the bis-hydroxyl-protected methyl ester 22 in 95% yield. Deprotection of the two TBS-groups in 22 was achieved with an excess of five equivalents of TBAF in THF at 0°C to afford 81% yield of the diol 23. 24 The internal conjugated alkyne in 23 was reduced to the methyl ester 24 in 65% yield after chromatographic purification on silica. A modified Lindlar hydrogenation reaction 25 produced triene 24 with high stereoselectivity, while the diimide reduction 26 or the standard Lindlar hydrogenation reaction 27 of 23 failed to give a high conversion to 24. The Boland reduction 28 gave in our hands a large amount of elimination of water from 23. Finally, lenient saponification of the methyl ester 24 at 0°C with dilute aqueous LiOH in methanol followed by mild acidic work-up (aqueous NaH 2 PO 4 ) afforded a 78% yield of protectin D1 (2) in the last step (Scheme 3). The chemical purity of synthetic 2 and 24 was determined to be >95% and >98%, respectively, by HPLC analyses (see ESI †). The UV spectrum of synthetic protectin D1 (2) showed absorbance peaks (λ MeOH max ) at 262, 271 and 282 nm, which is in excellent agreement with the literature. 6 In order to obtain evidence that synthetic 2 and 24 matched that of authentic protectin D1 (2), protectin D1 (2) was obtained from endogenous murine self-resolving exudates. 29 Fig. 3 shows that the synthetic 2 was matched with endogenously produced 2.
In Fig. 3A authentic protectin D1 (2) obtained in vivo from exudates is displayed amongst its stereoisomers. 30 Fig. 3B shows the chromatographic behaviour of endogenously produced 2 (T R = 13.2 min) and Fig. 3C demonstrates that synthetic 2 co-elutes with endogenous 2. In addition, the MS-MS spectra for both biosynthesized 2 and synthetic 2 displayed essentially identical MS-MS fragmentation spectra with the following fragments assigned: (2). The chromatographic properties of synthetic 2 and the free acid of 24, the latter obtained by hydrolysis with aqueous LiOH in THF, 6 were matched with data of endogenously formed protectin D1 (2). These results demonstrated that hydrolyzed 24 co-elutes with authentic 2. Furthermore, the MS-MS spectra for both the free acid obtained from 24 and biosynthesized 2 displayed essentially identical MS-MS fragmentation spectra (see ESI †). Our NMR spectral data of synthetic 2 were in accord with those published by Petasis, Serhan and co-workers, 13b but not with the spectra published by others. 13a