A new synthesis of the F 5 furan fatty acid and a first synthesis of the F 6 furan fatty acid

Total syntheses have been achieved of the two furan fatty acids F 5 and F 6 1 , in which the central step features generation of the furan ring using either iodine or silver-induced 5-endo -dig cyclisations of suitably functionalized 3-alkyne-1,2-diols.


Introduction
The furan fatty acids F 5 and F 6 , 1a and 1b, are unusual in that they contain respectively a tri-and a tetra-substituted furan core; most furanoid fatty acids are 2,5-disubstituted compounds. 1First identified in Exocarpus cupressiformis seed oil, 2 they occur in fish oils where they may be associated with reproductive cycles, 3 and are also found in marine sponges, 4 and in many plant species, which inevitably means they are present in many foods. 5While their precise role remains unclear, urofuran acids [structure 1, but CO 2 H in place of the 3-methyl group], which occur in human urine are likely metabolites of these natural products. 6here have been three previous syntheses of F 5 furan fatty acid 1a, the first starting from methyl 3-methyl-2-furoate delivered the methyl ester and served to confirm the proposed structure but, while brief, was rather inefficent. 7Subsequently, two rather different approaches were reported much later, in 1998.In Marson and Harper's route, 8 1-acetyl-1-cyclododecene is homologated by the sequential addition of heptynyl lithium and alkene epoxidation.Furan formation was then achieved using catalytic acid and mercury(II) to provide an excellent yield of the aldehyde corresponding to furan F 5 1a.In a conceptually very different approach, Bach and Krüger set out with 4,5-dibromofuran-2-carboxaldehyde and employed the greater reactivity of the α-bromide to attach the necessary methoxycarbonyldecyl chain using a regiospecific Sonogashira coupling, followed by Wittig homologation to establish the pentyl chain.The βmethyl was then introduced by a second Pd(0)-catalysed coupling with tetramethyltin acting as the carbon source and the synthesis was completed by hydrogenation of the two unsaturated linkages and finally ester hydrolysis. 9As far as we are aware, there has been no report of a synthesis of furan fatty acid F 6 1b As a result of our studies of the 5-endo cyclisation mode, 10 we have recently defined two novel approaches to poly-substituted furans 3 using "Baldwin-favoured" 5-endo-dig cyclisations 11 of 3-alkyne-1,2-diols 2 (Scheme 1).In the first method, exposure of the precursors 2 to excess iodine in the presence of a mild base led to 70~85% yields of the iodofurans 3a. 12 In a much more environmentally friendly process, isomerisation and dehydration of the alkynediols 2 to the trisubstituted furans 3b has been achieved using catalytic quantities of silver nitrate on silica gel; yields using this heterogeneous and easily recycled and reused catalyst are essentially quantitative. 13It occurred to us that both furan fatty acids 1 would provide significant tests of these two methodologies, both of which, in principle, should be suitable for the synthesis of both acids 1.
Scheme 1 Herein, we report in preliminary form on the successful applications of these two approaches, which have nevertheless revealed a few surprises along the way.

Results and Discussion
Our first approach was aimed at the F 5 furan fatty acid methyl ester using our iodocyclisation methodology 12 and featuring generation of the required alkyne-diol by selective bis-hydroxylation of the corresponding enyne.We began by converting commercially available 11bromoundecanoic acid 4a into the corresponding iodide 4b 14 using a Finkelstein reaction and then coupling this with lithium acetylide-ethylenediamine complex [LAEDA] using HMPA as solvent, which gave the alkynoic acid 5a routinely in ca.85% yields (Scheme 2). 15This was then smoothly esterified using acidic methanol to give the methyl ester 5b.Although many alkynes have been prepared successfully from both chlorides and bromides using LAEDA, 16 in the case of the bromo-acid 4a, in our hands, the reaction proved difficult to drive to completion and the use of more forcing conditions and/or greater amounts of LAEDA resulted in significant amounts of isomerisation of the initial alkynoic acid 5a to the corresponding 11-ynoic acid.The (E)-iodoalkene 6 was prepared in excellent yield from 1-heptyne by sequential hydroalumination using Dibal-H and halogenation by N-iodosuccinimide 17 ; a subsequent Sonogashira coupling 18 with iodo-ester 5b then gave the (E)-enyne 7.As we clearly did not require homochiral material, this was then bis-hydroxylated regioselectively at the alkene function, as originally reported by the Sharpless group, 19 but using the simpler achiral Warren procedure, 20 which delivered an excellent 86% yield of the required alkyne-diol 8.We were dismayed to find that our initial attempts to carry out a 5-endo-dig iodocyclisation 12 using the alkene-diol 8 met with almost complete failure.Using the 'standard' conditions consisting of three equivalents each of iodine and sodium hydrogen carbonate in either dry acetonitrile or dichloromethane gave at most 10% yields of the desired and indeed expected iodofuran 9a (Scheme 3).Alterations to the base (potassium carbonate, triethylamine), the electrophile (bis(sym-collidine)iodine(I) hexafluoro-phosphate, phenylselanyl chloride) and working at lower temperatures were equally ineffective.Some NMR and mass spectrometric evidence pointed towards formation of the di-iodide by direct addition of iodine to the alkyne group.
Reasoning that the lengthy carbon chain was perhaps coiling in solution and preventing cyclisation in the expected manner, we carried out a solvent study to try and rescue the situation.Tetrahydrofuran proved completely ineffective and the use of very non-polar petrol ether was precluded because the precursor 8 was not appreciably soluble.In an effort to pursue the theme of using a non-polar solvent system, a small percentage of acetonitrile was added to increase the substrate solubility but this also proved ineffective.Remarkably, when a small amount of ethyl acetate was used to increase the solubility, NMR analysis of the crude product showed an appreciable amount of iodofuran to have formed.Even better, when neat ethyl acetate was used as the solvent, the conversion after 24h at 40 o C was even better, but the product 9a was always accompanied by a second, as yet unidentified by-product.A final optimization showed that the reaction in ethyl acetate was unexpectedly both very rapid and clean: work-up after one hour at ambient temperature secured a 93% yield of the iodofuran 9a (Scheme 3), uncontaminated by the by-product, which is evidently formed from the initial product 9a.Although we could not gain firm evidence to support the suggestion, we speculate that this might have been the 3,4-di-iodide derived from the iodofuran 9a.In similar fashion to some related attempts to introduce the final methyl group using Stilletype methodology, 9 we found it difficult to complete this transformation without the formation of the desmethyl derivative [9; R = H].The conditions shown in Scheme 3 were amongst the best we found and the poor yield was more associated with losses during the difficult chromatographic separation of the target 9b than with substrate decomposition.Unfortunately, the presence of the ester group rather precluded using what would probably be the most efficient way to effect this transformation, that of halogen-metal exchange at low temperature and reaction with iodomethane.We therefore modified the approach to incorporate both the opportunity to use the latter chemistry for incorporation of the necessary β-methyl group and also to exploit our newer silver(I)-catalysed cyclisation methodology (Scheme 1).The initial steps are shown in Scheme 4. Beginning with the mono-TBS derivative 10 of 1,12dodecanediol, 21 PCC oxidation followed by condensation delivered the alkynol 11.Temporary protection of the alcohol group as the corresponding acetate was followed by alkyne hydration using an excellent gold(III)-catalysed method 23 and a basic work-up with aqueous carbonate to arrive at the acyloin 12 in good overall yield.Reaction with two equivalents of lithio-heptyne then delivered the required alkyne-diol 13 in excellent yield.This direct approach was favoured over the more usual tactic of alcohol protection, alkyne addition and deprotection on the grounds of both expediency and atom efficiency: the equivalent of 1heptyne present in the final product was easily removed by rotary evaporation and about the same number of atoms or less was wasted in this way, relative to using a protection-deprotection method, which in most cases would require additional reagents and solvents.
We were delighted to find that the silver(I)-catalysed method was effective in converting the alkyne-diol 13 into the F 5 furan fatty alcohol 14 in essentially quantitative yield (Scheme 5).Final deprotection and oxidation 8 then provided the target 1a in good yield.Using the crucial information gained during out first synthesis of furan F 5 1a (Scheme 3), the same precursor 13 was also treated with iodine in ethyl acetate, when the iodofuran 15 was formed in excellent yield.Once again, Stille and related Pd(0)-catalysed displacements of iodine by methyl proved difficult, mainly because of formation of variable amounts de-iodinated material which again was very difficult to separate from the desired product.However, working at this lower oxidation level allowed us to use the older method of halogen-metal exchange and alkylation, which proved to be much more reliable (Scheme 5).The crude product was then deprotected and oxidized in similar fashion to provide the first synthetic sample of F 6 furan fatty acid 1b. 24Hence, these last two approaches may be competitive with alternatives featuring Pd(0)-catalysed couplings, despite the necessity for including the additional oxidation steps.