Improved synthesis of enantiomerically pure Etomoxir and its 2,4-dinitrophenyl analogue

A generally applicable synthesis has been developed for 2-substituted oxirane-2-carboxylic esters, which have attracted interest as inhibitors of carnitine palmitoyltransferase-1 (CPT-1) for the treatment of non-insulin-dependent diabetes mellitus (NIDDM). The route involves alkylation of the dianion of 2-methyl-2-propen-1-ol (methallyl alcohol) followed by Sharpless epoxidation. The utility of the method has been demonstrated by the synthesis of ( R )-Etomoxir, the best known member of this class of compounds, and ethyl ( R )-2-[6-(2,4-dinitrophenoxy)hexyl]oxirane-carboxylate, previously synthesised only as a racemate. The high enantiomeric purity of the compounds has been demonstrated by formation of Mosher esters of the products of Sharpless epoxidation and analysis by 1 H NMR spectroscopy.


Introduction
The enzyme carnitine palmitoyltransferase-1 (CPT-1) catalyses the conversion of long-chain fatty-acid CoA esters, principally the CoA ester 1 of palmitic acid, to carnitine esters such as 2. The latter can enter the mitochondria, where they are converted back to CoA esters by another carnitine palmitoyltransferase, CPT-2 (Figure 1). 1 Inside the mitochondria, CoA esters are metabolised by iterative β-oxidations to acetyl CoA, which enters the citric acid cycle, ultimately being degraded to CO 2 with the production of ATP.
By blocking entry of fatty-acid CoA esters into the mitochondria, production of acetyl CoA, ATP and NADH are reduced, thus attenuating gluconeogenesis and hepatic glucose production. 2 Further, the body is forced to rely on glycolysis for the production of acetyl CoA. 3 Therefore, CPT-1 inhibitors have hypoglycaemic (blood-sugar-lowering) effects and have attracted attention as potential treatments for type-2, or non-insulin-dependent diabetes mellitus (NIDDM). 4ne class of CPT inhibitors is the substituted oxirane carboxylates, of which the best known is probably Etomoxir 3a (Figure 2). 5 These compounds, of which only the (R) enantiomer is active, are converted to CoA esters in the body, mimicking fatty-acid CoA esters.They bind to CPT-1, whereupon the enzyme is covalently alkylated by the epoxide, probably at a histidine residue. 6Hence, these compounds are irreversible inhibitors.Etomoxir was never brought to market however, due to the very large doses needed for efficacy, 7 as well as side-effects such as cardiac hypertrophy. 8These side effects are due in part to the lack of selectivity (of Etomoxir 3a and other similar compounds) for the liver isoform of CPT-1 (L-CPT-1) over the skeletal muscle isoform (M-CPT-1).
M-CPT-1 is the major isoform in the heart, and unwanted inhibition of this isoform is a major factor in the side-effects.To overcome this, the L-CPT-1-selective inhibitor ethyl 2-[6-(2,4-dinitro-phenoxy)-hexyl]oxiranecarboxylic 4 (Figure 2) was developed, although only as a racemate. 9,10(The free acid form, prepared via a silyl ester rather than the ethyl ester, was used for biological testing).Although Etomoxir 3a has not been accepted for use treating NIDDM, it has more recently shown potential as a treatment for chronic heart failure. 11Combined with the demonstration with compound 4 that the side-effects of substituted oxirane-2-carboxylates can be overcome, it seems likely that these compounds and analogues will continue to attract attention.Hence, we have developed a new synthetic methodology for substituted oxirane-2-carboxylates, exemplified by syntheses of 3a and monochiral 4 [(R)-enantiomer 4a].

Results and Discussion
The key transformations in our method are (i) alkylation of the dianion of methallyl alcohol (2methyl-2-propen-1-ol) with a bromide corresponding to the desired side-chain of the target, (ii) Sharpless epoxidation 12 of the resulting allyl alcohol, and (iii) oxidation and esterification of the Sharpless product.Various other published syntheses introduce the chirality by enzymatic resolution 13,14 and via a chiral auxiliary. 15Sharpless epoxidation has been reported before in the synthesis of Etomoxir 3a, but the synthesis of the necessary precursor 7 was long and inconvenient. 16Our synthesis of precursor 7 is just two steps, and the complete synthesis of Etomoxir 3a just four steps.(Scheme 1.) 4-Chlorophenol 5 was alkylated with an excess of 1,5-dibromopentane, sodium hydroxide and a phase-transfer catalyst to give the bromide 6 in a slightly modified literature procedure. 17his was used to alkylate methallyl alcohol, doubly deprotonated with butyl lithium in the presence of TMEDA.We found that this reaction can be facilitated by using 2.5 M butyl lithium diluted to 1.6 M with diethyl ether, and also by allowing the reaction mixture to warm from -78 °C to room temperature over one hour after addition of the bromide.The resulting allyl alcohol 7 was epoxidised under modified Sharpless conditions using L-(+)-diethyl tartrate as the ligand to give compound 8a, or using D-(-)-diethyl tartrate to give the (R)-enantiomer 8b (not pictured).These stereochemical outcomes are based on the accepted model. 18Alternatively, epoxidation with mCPBA gave the corresponding racemate (compound 8c, not pictured).The final step in the synthesis was ruthenium-catalysed periodate oxidation of 8a and esterification of the resulting crude acid.We initially used diethyl sulfate but found that ethyl iodide was superior, being less toxic and easier to remove.The product (R)-Etomoxir 3a (or (S)-Etomoxir 3b, not pictured, from (R) epoxy alcohol 8b) was easily purified by chromatography on silica.Finally, (R)-Etomoxir 3a could if desired be hydrolysed using sodium hydroxide to give (R)-Etomoxir, sodium salt 10a (not pictured), and likewise (S)-Etomoxir 3b gave (S)-Etomoxir sodium salt 10b (not pictured).
The high enantiomeric purity of the Etomoxir 3a was proved by formation of an ester 9a of its precursor epoxy alcohol 8a with (R)-α-methoxy-α-trifluoromethylphenylacetic acid (Mosher's acid), 19 using diisopropylcarbodiimide as the coupling reagent and DMAP as catalyst (Scheme 2).Both 8a and its enantiomer 8b were converted to (R)-Mosher esters for comparison (compounds 9a and 9b respectively).The crude samples were analysed by 1 H NMR before purification, and the peaks of the AB system corresponding to the diastereotopic CH 2 group indicated in Figure 3 were carefully integrated.In both cases, a diastereoisomeric ratio of 98 : 2, corresponding to an e.e. of at least 96 % for 8a and 8b, was observed.The 1 H NMR spectra for these CH 2 groups are shown in Figure 3.The synthesis of ethyl (R)-2-[6-(2,4-dinitrophenoxy)-hexyl]-oxiranecarboxylate 4a required a partly different strategy.We began by attempting an exactly analogous synthesis to that described for (R)-Etomoxir, and accordingly alkylated 2,4-dinitrophenol 11 with an excess of 1,5-dibromopentane in the presence of sodium hydroxide and a phase-transfer catalyst to give the bromide 12 in good yield (Scheme 3).Unfortunately, trying to use 12 to alkylate doubly deprotonated methallyl alcohol resulted only in the formation of a black tar containing none of the desired product 13.5-Bromo-1-pentanol 14 was prepared according to a literature procedure, 20 and protected using triisopropylsilyl chloride and imidazole in DMF to give bromide 15.This underwent coupling with deprotonated methallyl alcohol satisfactorily in the presence of TMEDA to give allyl alcohol 16.This was epoxidised under modified Sharpless conditions using L-(+)-diethyl tartrate as the ligand to give compound 17a, D-(-)-diethyl tartrate to give the (R) enantiomer 17b (not pictured), or mCPBA to give racemate 17c (not pictured).Enantiomeric purity (e.e. ≥96 %) was again proved by the formation of (R)-Mosher esters as above (not pictured, compounds 18a and 18b for the (S) and (R) enantiomers respectively in the experimental section).
Epoxy alcohol 17a was oxidised and esterified under the same conditions which yielded Etomoxir 3a to give ester 19, and the protecting group was removed using tetrabutylammonium fluoride (TBAF) in THF to give alcohol 20.This alcohol should be used promptly as it appears to go off on standing.
The final step was the coupling of 20 with a source of the 2,4-dinitrophenoxy group to yield the target compound 4a.To achieve this, two approaches were examined: nucleophilic aromatic substitution with 2,4-dinitrofluorobenzene, and Mitsunobu reaction 21 with 2,4-dinitrophenol.The Mitsunobu reaction, which has been used in the synthesis of 2-substituted-oxirane-2carboxylates previously, 15,22 was the better of the two, 2,4-dinitrofluorobenzene giving difficultto-reproduce results and sometimes yielding compound 4a contaminated with an unidentified impurity difficult to remove by chromatography.
Overall, this route is slightly longer than the route which yielded Etomoxir 3a, but has the advantage that the use of the Mitsunobu reaction allows for the dereferal of the coupling of the aromatic group until after the formation of the oxirane-carboxylate portion of the molecule.This opens up the possibility of the synthesis of a wide variety of chemically sensitive or nonphenolic analogues.

Conclusions
In conclusion, we have developed a versatile and efficient route to 2-substituted oxirane-2carboxylates, which should allow the synthesis of a broad range of analogues in enantiomerically pure form.In doing so, we have developed the shortest synthesis yet published of Etomoxir.

Experimental Section
General Procedures.Anhydrous solvents were purchased from Fluka, and n-butyllithium from Sigma-Aldrich.TMEDA and methallyl alcohol were distilled immediately before use.tert-Butyl hydrogen peroxide in anhydrous toluene was prepared according to the method of Sharpless. 23ll other reagents were used as received.Column chromatography on silica refers to medium pressure chromatography using Davisil 40 -63µ silica gel, Fluorochem, Derbyshire, UK.Optical rotations were recorded at 589 nm.