Towards a Synthesis of Vigabatrin Using Glycal Chemistry

Application of the Ferrier rearrangement led to a novel carbohydrate based synthetic route to 4-aminohexenoic acid viz. (R) and (S)-vigabatrin. The potential of D- glucose or D-galactose as the chiral starting materials for the synthesis of (R) and (S)- vigabatrin has been explored.


Introduction
γ-Aminobutyric acid (GABA) is the vital inhibitory neurotransmitter in the mammalian central nervous system [1]. It was found that 4-amino-5-hexenoic acid (γ-vinyl GABA, vigabatrin, 1) is one of the most effective and selective catalytic inhibitors of GABA-T.
Inhibition of this enzyme results in an increase in the levels of GABA and could have therapeutic applications in neurological disorders including epilepsy [2], Parkinson's disease [3], Huntington's chorea [4] and Alzheimer's disease [5]. Recently it has been found that an increase in GABA also blocks the effect of drug addiction to nicotine and cocaine [6]. Vigabatrin (1), not a complex chiral molecule, is already marketed in its racemic form as Sabril ® in Europe [7], and has evoked considerable interest among organic chemists and medicinal chemists as an attractive synthetic target.
Since vigabatrin has a chiral center at C-4, two enantiomers are possible viz. (R)-vigabatrin and (S)-vigabatrin. It has been found that the (S)-enantiomer is pharmacologically active. Several methods have been reported for the synthesis of enantiomerically pure 1 and most of these employ natural αamino acids viz., L-glutamic acid, regioselective ring opening of an enantiomerically enriched epoxy alcohol (obtained through Sharpless epoxidation), the Pd(0)-catalyzed deracemization reaction and Sharpless asymmetric aminohydroxylation [8]. The vinyl moiety was then introduced by pyrolysis of an N-oxide or an ester or through a Wittig reaction. While many syntheses of racemic and enantiomerically pure vigabatrin are known, the potential of D-glucose or D-galactose as the chiral starting materials for the synthesis of (R) and (S)-vigabatrin has not been explored. Herein we disclose our results in this area. The synthesis of the target molecule vigabatrin (1) was visualized as starting from the 2,3unsaturated methylglycoside (3), obtained via Ferrier rearrangement of tri-O-acetyl-D-glucal (2) [9]. The 5-amino functionality could be introduced by a S N 2 displacement of the mesylate by azide ion and subsequently the vinyl group could be constructed by fragmentation of 6-bromo-6-deoxyglycosides with Zn or n-BuLi -the Vasella reaction [10]. The complete retrosynthetic analysis of (R)-vigabatrin is outlined in Scheme 1.

Results and Discussion
Methyl-4,6-di-O-acetyl-2,3-dideoxy-α-D-erythro-hex-2-enopyranoside (3), obtained from 3,4,6-tri-O-acetylglucal 2 [9a] was hydrogenated using Pd/C in MeOH for 1h to afford the saturated derivative 4 in 90% yield. After purification and subsequent hydrolysis of 4, the diol obtained was directly converted to the benzylidene derivative 5 in 76% yield using benzaldehyde dimethylacetal/cat. PTSA in DMF. The benzylidene derivative was fully characterized by spectroscopic techniques. The signal due to the benzylidene proton appeared as a singlet at δ 5.55 in the 1 H-NMR spectrum and that of the acetal carbon at δ 101.66 in the 13 C-NMR spectrum. Compound 5 was subjected to the Hannesian-Huller reaction [11] using NBS in refluxing CCl 4 , to furnish the bromobenzoate 6 in 70% yield (Scheme 2). The benzoate 6 was hydrolyzed using Na 2 CO 3 /MeOH to give methyl-6-bromo-2,3-dideoxy-Derythro-hexopyranoside (7) in 80% yield as a homogeneous gum (by TLC). The 1 H-NMR and 13 C-NMR spectral data were in accordance with the proposed structure. An attempt was made to introduce the azide group at C-4 with an inversion by a modified Mitsonobu reaction (Scheme 3) using bis-(p-nitrophenyl)phosphorazidate as reagent [12], however, there was no reaction at either ambient or elevated temperature (110 o C) and the starting material 7 was recovered.
Having failed to introduce the C-4 azido group by the above methodology, the bromo derivative 7 was converted to the mesylate using mesyl chloride in Et 3 N. The mesylate 8 was obtained as a viscous liquid in 82% yield. Again considerable efforts were directed to effect exclusive displacement of the mesylate group with azide ion to obtain 16, but all were in vain. These reactions were attempted at different temperatures ranging from 0 o C to 110 o C, and in different solvents viz. DMF and DMSO, as well as under phase transfer catalyst conditions, but without any success, as invariably, a mixture of monoazide, diazide and other products was obtained (Scheme 3).  From the above observations, it was apparent that there was not much difference in the reactivity of bromo as a leaving group at C-6 and mesylate at C-4 towards nucleophilic substitution with azide ion. We then considered the synthesis of the C-4 epimer of the mesylate 8, in the hope that there would be some steric hindrance for substitution of the Br at C-6 due to the axial mesylate group and this would result in the selective displacement of the mesylate by azide furnishing the desired 17, a potential precursor for pharmacologically active (S)-vigabatrin. We thus synthesized the epimeric mesylate 15 via a similar synthetic sequence as described for the mesylate 8 (Scheme 2). The synthesis proceeded well in an overall yield of 42% over six steps. All the intermediates leading to 15 were characterized by IR, 1 H-NMR and 13 C-NMR spectroscopy. Unfortunately, our attempts to bring about a selective displacement of the mesylate group in 15 by azide ion to form 17 were also unsuccessful. Though TLC analysis revealed that the product was homogeneous, giving a close moving non-polar spot but the 1 H-NMR and 13 C-NMR spectra indicated it to be a mixture of 4,6-diazide and 6-azide derivatives.
In retrospect, we realized that selective displacement of the secondary mesylate with azide ion at C-4, be it axial or equatorial, in the presence of the primary bromide was neither practical nor feasible. Initial incorporation of the azido group at C-4 and later appropriate functionalisation for the Vasella reaction [10] should serve as a viable alternative. Accordingly, the synthetic strategy was slightly modified (Scheme 4).
In the proposed alternative route diacetate 4 was hydrolyzed and the C-6 hydroxyl was selectively protected as the corresponding trityl ether using Ph 3 CCl/Et 3 N in DMF, thus furnishing compound 18 in 70% yield. The trityl ether 18 was again characterized using IR, 1 H-and 13 C-NMR spectroscopy. Since we had learned from previous experience that bis (p-nitrophenyl)phophorazidate was not successful in replacing the C4-OH by an azido group, we did not attempt this reaction on 18. Instead 18 was directly converted to the mesylate 19 in 80% yield using methanesulfonyl chloride/Et 3 N in CH 2 Cl 2 . The mesylate 19 underwent a smooth nucleophilic substitution reaction upon treatment with NaN 3 in DMF at 110 o C and afforded the crystalline (m.p. 69 o C), methyl 4-azido-6-triphenylmethyl-2,3-dideoxy-Dthreo-hexopyranoside (20)  While the anomeric proton resonated at δ 4.67(d), the signal due to H-4 of 20 was shifted upfield δ 3.87(td). The stereochemistry at C-4 was established from the coupling constants of H-4. The signal due to H-4 in 19 appeared as a triplet of doublets with J = 9.99Hz and 5.51Hz, whereas the signal due to H-4 in the azide 20 appeared as a triplet of doublets with smaller coupling constants of 6.1 Hz and 1.47 Hz. The large coupling constant between J 4,5 = J 3a,4 = 9.99 Hz for H-4 in 19, due to the diaxial nature of H 3a and H 5 at C-3 and C-5 disappeared, giving rise to a coupling constant J 4,5 and J 3a,4 = 6.1 Hz due to the vicinal axial-equatorial relationship existing in 20. Based on this, we infer that nucleophilic substitution by the azide has taken place with complete inversion at C-4, thus leading to the threo series.
After successful incorporation of the protected amino functionality at C-4, the conversion of 22 to the bromo derivative 23 was necessary for effecting the Vasella reaction and arrive at (R)-vigabatrin (1). This appears to be a straightforward reaction, but various attempts to convert the 6-OH to the corresponding bromide by CBr 4 /PPh 3 or PBr 3 /pyridine were all in vain. The lack of reactivity of the 6-OH in the above reaction may be due to the presence of a bulky axial 4-(N-t-butoxycarbonyl) group at C-4. The same steric hindrance should be absent in the erythro series, which therefore offers a rationale and motivation to pursue the synthesis of pharmacologically active (S)-vigabatrin. These synthetic efforts are currently underway.

Experimental
General 1 H-and 13 C-NMR spectra were obtained using a Varian Gemini 200 NMR and were recorded at 200 and 50 MHz respectively. All reagents and chemicals were obtained from Aldrich Chemical Company (USA) and were used as received unless otherwise noted.
Compounds 3 or 10 (1 mmol) was placed in a 25 mL round bottomed flask. The substrate was dissolved in methanol (5 mL) and a catalytic amount of 10% Pd/C catalyst was added. Hydrogen gas was passed through the solution using a balloon. When there was no more uptake of hydrogen (5 h to 6 h), the reaction was stopped and the catalyst filtered. The solvent was evaporated in a rotary evaporator under reduced pressure to afford methyl-4,6-di-O-acetyl-2,3-dideoxy-D-erythrohexopyranoside (4) and methyl-4,6-di-O-acetyl-2,3-dideoxy-D-threo-hexopyranoside (11).