A gram-scale synthesis of very-long chain polyunsaturated fatty acids (VLC-PUFAs)

Abstract

This manuscript describes our third generation, gram-scale synthesis of very long chain-polyunsaturated fatty acids (VLC-PUFAs), a unique and increasingly important class of lipids. Critical to this work and what makes it different from our previous approach to this family was the avoidance of oxidation sequences. Central to accomplishing this involved the use of a Negishi coupling reaction between the acid chloride derived from DHA and a saturated alkyl zinc reaction. Overall, the general approach required 6 synthetic transformations from DHA and was accomplished with an overall yield of 40%.

Introduction

Very long chain polyunsaturated fatty acids (VLC-PUFAs) are a unique class of lipids that were first reported by Aveldaño in 1987.¹ They combine polyunsaturated and saturated segments having alkyl chains whose lengths range from 24 to 38 carbons with between three and six skipped Z-alkenes. VLC-PUFAs are synthesized endogenously in vertebrates by the elongation of very long chain fatty acids protein (ELOVL) family of enzymes (specifically ELOVL4) starting from long chain polyunsaturated fatty acid precursors.² While information is still being collected, VLC-PUFAs appear to play important physiological roles. To date, they have been found in low concentrations in the eyes, the testes, and the brain and both their deficiency (Stargardt’s disease and age-related macular degeneration (AMD)) and their abundance (Zellweger spectrum of disorders) have been associated with disease.^3-6 As their connection to Stargardt’s disease and AMD imply, they appear to play an important role in visual health where they have been proposed to assist in the curvature of photoreceptor disks and to help stabilize and possibly translocate retinoids across the retinal membrane.⁷ Because they have not been available in sufficient quantity or purity until recently, the clarification of their role in health and disease has awaited the development of a more reliable supply chain.

Aware of the need for a scalable synthesis, we recently carried out and reported two syntheses of VLC-PUFA 32:6 n-3 (1). Both utilized the coupling of an aliphatic Grignard reagent with an aldehyde that was derived from docosahexaenoic acid (DHA).⁸ The use of DHA as a coupling partner was significant as it enabled us to avoid the purification of olefin isomers that would likely be present from any strategy that involved the de novo synthesis of the Z-alkenes.⁹ In our second-generation approach to 32:6 n-3 (Scheme 1), manipulation of the oxidation states of the VLC-PUFA carbon framework post-coupling allowed us to generate 1 in 8 steps from DHA (longest linear sequence) in 20% overall yield. This route was successfully employed to generate 32:6 n-3 in hundreds of milligram quantities. We were excited to find that studies that employed this material in mice demonstrated both the transport of 32:6 n-3 to the retina and that its presence led to enhanced visual acuity.¹⁰ In more fundamental studies, the synthetic material was also used to show that low concentrations of VLC-PUFA’s increased the fluidity of model membranes by increasing the rate of membrane flip-flop.¹¹

Scheme 1.

Previous synthesis of VLC-PUFA 32:6 n-3.

Discussion

With the successes outlined above in hand and the future needs of VLC-PUFAs in mind, our attention turned to generating gram quantities of VLC-PUFA 32:6 n-3. A hurdle to the use of our previous synthesis to accomplish this goal was the oxidation state manipulation that had been employed late in the synthesis. Not only were the yields often capricious but in our hands the product was unstable to the reaction conditions, particularly to workup where it was challenging to separate unreactive Oxone^® prior to concentration. In retrospect, if one considers the skipped alkenes present in VLC-PUFAs these problems were not surprising. We explored several alternatives to avoid this issue including the use of other solvents and workup procedures without notable success.

Considering these challenges, we became interested in pursuing a strategy to VLC-PUFAs that avoided the use of oxidants. To this goal, we opted to explore the use of an acid chloride Negishi reaction to couple a polyunsaturated acid chloride derived from DHA with an aliphatic organozinc reagent.¹² That the organozinc reagent would in theory be amenable to the presence of esters would enable us to incorporate the requisite carboxylic acid moiety in the desired oxidation state and thus avoid the aforementioned problematic oxidation reaction. It was appealing that this new strategy would rely on reductions rather than oxidations in the final stages. To this end, our new synthesis of VLC-PUFA 32:6 n-3 began with the coupling of methyl-10-bromodecanoate 9 with the DHA derived acid chloride 11 (Scheme 2). To our delight, the Negishi reaction allowed us to successfully couple 9 with 11 on gram scale to provide the entire VLC-PUFA skeleton as 12. Over the course of optimizing this transformation we found that the combination of freshly activated Zn and Pd(PPh₃)₄ in the presence of a mixture of THF, toluene, and DMA gave the best results. When the coupling reaction was carried out on gram scale, ketone 12 was contaminated with Pd by-products even after column chromatography. To our delight, subjecting the crude reaction mixture containing 12 to a sodium borohydride reduction not only provided the desired 2° alcohol 13 but it also helped in the removal of the Pd. This protocol enabled us to reliably generate 13 on multi-gram scale in 64% yield from DHA.

Scheme 2.

DHA acid chloride Negishi coupling to VLC-PUFAs.

From 13, it remained to chemoselectively reduce the alcohol in the presence of the ester. Our initial attempts to reduce the tosylate, mesylate, bromide, or iodide derivatives from 13 using NaBH₄ in THF were unsuccessful, either leading to recovered starting material or products that had been globally reduced.¹³ We eventually found that the conversion of 13 into the corresponding alkyl bromide using CBr₄ and PPh₃ followed by reduction of the bromide using freshly activated Zn in the presence of acid reliably delivered the desired VLC-PUFA ester 15 in 69% yield over the two steps (Scheme 3).

Scheme 3.

Completion of the gram scale synthesis of 32:6 n-3.

Finally, because of a combination of solubility issues and the instability of the product to some of the reaction conditions, the saponification of 14 turned out to be non-trivial. The solution to this problem turned out to involve the slow addition of aqueous NaOH to a solution of 14 in a mixture of THF and MeOH at 0 °C to give VLC-PUFA 32:6 n-3 in 89% yield.

As was mentioned above, a number of different VLC-PUFAs have been identified in the retina. Presently, it is not clear what physiological role the mixture plays or whether one or some VLC-PUFAs are more important than others. From an interest in shining a light on this, we applied the sequence used for 32:6 n-3 to the synthesis of the naturally occurring VLC-PUFA 34:6 n-3 20 (Scheme 4). The only surprise when compared to the synthesis of 1 was that the alkyl bromide that was derived from 17 was more sensitive to storage when compared to the alkyl bromide from 13. However, if the bromide was reduced immediately upon its preparation, we could isolate VLC-PUFA ester 19 in 41% yield from alcohol 17 over the two steps. Hydrolysis of 19 gave VLC-PUFA 34:6 n-3 20.

Scheme 4.

Synthesis of VLC-PUFA: 34:6 n-3.

During our preparation of this manuscript,¹⁴ Honzíková and co-workers reported the synthesis of the ethyl ester of VLC-PUFAs 28:6 n-3 and 26:6 n-3 using a related coupling strategy.¹⁵ In their work they carried out an sp³–sp³ coupling using the alkyl bromide from DHA (obtained from a LiAlH₄ reduction of DHA followed by conversion of the alcohol into the corresponding bromide) with the appropriate alkyl Zn reagent in the presence of a Pd-PEPPSI catalyst. The VLC-PUFA esters from their coupling reaction were generated in 15–21% yield and were contaminated with by-products that come from the decomposition of the DHA bromide. Although we also observed a small amount of a C-21 DHA decomposition product during our coupling reactions, this species was easily separated from 13. Thus, not only was our coupling much higher yielding but the product was purer.

Conclusions

To conclude, the use of a Negishi coupling strategy to couple the DHA acid chloride with an alkyl zinc reagent provided the entire VLC-PUFA carbon chain and allowed us to avoid a problematic late-stage oxidation reaction that had been problematic in our previous synthesis. This improvement resulted in a 6 step, gram scale synthesis of VLC-PUFA 32:6 n-3 in an overall yield of 40% from DHA. This protocol is amenable to the synthesis of other VLC-PUFAs as demonstrated by our synthesis of VLC-PUFA 34:6 n-3. We are confident these studies will enable us and others to better uncover how the structures of these interesting compounds relate to their physiological properties. These latter studies will be reported in due course.