2094
A. Tavassoli et al. / Tetrahedron Letters 46 (2005) 2093–2096
HO
CO2Me
HO
CO2Me
OH
HO
CO2Me
HO
CO2Me
HO
CO2Me
O
O
MeO2C
3
2
3
(iv)
(i)
(ii)
(iii)
S
2
+
5
CO2Me CO2Me
O
O
O
HO
CO2Me
O
N
O
O
O
OH
9a
N
10a
6
7
8
11a
Scheme 2. Reagents and conditions: (i) thiocarbonyldiimidazole/CH2Cl2/rt, 16 h (89%); (ii) Bu3SnH–AIBN/toluene/reflux, 3 h (83%); (iii) Amberlite
IR-120(H +)/MeOH/reflux, 3 h (70%); (iv) (a) NaIO4–silica gel/CH2Cl2, (b) H2O2/HCO2H/rt, 6 h, (c) Amberlite-120(H +)/MeOH/reflux,16 h (36%
10a + 15% 11a).
catalysed by citrate synthase has been shown to be
inverted, using (R)- and (S)-[2-2H1,2-3H1]-acetylCoA, 1
labelled derivatives by methods which would allow
unambiguous assignment of their stereochemistry. The
isopropylidene derivative 12a was therefore prepared
from shikimic acid by the method of Chahoua et al.10
and this was reduced in good yield to the 5-deoxy deriv-
ative 14a via 13a, as shown in Scheme 3, using the
method described above for the preparation of the qui-
nate analogue 8 (Scheme 2). Deprotection was now re-
quired so that epoxidation of the double bond would
be directed to the same face as the 3- and 4-hydroxyl
groups by the well known Sharpless directed epoxida-
tion of allylic and homoallylic alcohols.11,12 This was
achieved in 79% yield using Amberlite IR-120(H +) in
methanol, and epoxidation of the product 15a was car-
ried out using tert-butylhydroperoxide and vanadyl
acetylacetonate in dichloromethane, giving 16a in
57% yield. W-coupling between H-2 and H-4 was ob-
3
(HA = H, HB = 2H) and 1 (HA = 2H, HB = 3H), respec-
tively,7 and it is of interest to investigate this aspect of
the stereochemistry in the analogous reactions catalysed
by homocitrate synthase and by the protein from the
nifV gene. Unlike citric acid, homocitric acid 3 is asym-
metric and so the two hydrogens, HA and HB, arising
from acetate in this product are diastereotopic. Assign-
ment of stereochemistry to the chemical shifts arising
from these hydrogens will therefore enable the stereo-
3
chemistry of step (i) in Scheme 1 to be assessed by H
NMR spectroscopy when (R)- and (S)-[2-2H1,2-
3H1]-acetylCoA are used in the enzymatic reaction. We
now report a synthesis of trimethyl (2S,3R)- and
(2R,3R)-[2-2H1]-homocitrates 10 and dimethyl (2S,3R)-
and (2R,3R)-[2-2H1]-homocitric lactones 11, involving
reactions of unambiguous stereochemistry. Since homo-
citrate from the enzyme reactions can readily be con-
verted to these esters without racemisation,8 this
synthesis constitutes an assay for the stereochemistry
of the enzymic reaction.
1
served in the H NMR spectrum of 16a, suggesting that
these hydrogens are quasi-equatorial, as shown in Fig-
ure 1. This implies that the desired (1R,2S)-stereochem-
istry had been induced in the epoxidation reaction.
Reprotection of the secondary alcohol groups was now
necessary to allow for selective oxidation of the primary
alcohol which would result from reduction of the ester
when the epoxide was reduced from the re-face to create
the stereospecifically labelled centre at C-2. This was
achieved using 2,2-dimethoxypropane and camphorsulf-
onic acid, giving the isopropylidene derivative 17a.
Retrosynthetically, quinic acid might serve as the source
of the 3R centre of homocitrate if the hydroxyl group at
C-5 were to be specifically removed and the remaining
cis-diol cleaved. We therefore used the method of Shing
and Tang9 to prepare the quinate derivative 6, in which
the cis-3,4-diol moiety is protected. Conversion to the 5-
thiocarbonylimidazole derivative 7 in 89% yield fol-
lowed by reduction using Bu3SnH and AIBN afforded
the protected 5-deoxyquinate 8 in 83% yield, as shown
in Scheme 2. Deprotection using Amberlite IR-120(H +)
gave the diol 9a in 70% yield and this was cleaved using
periodate on silica gel followed by oxidation of the inter-
mediate dialdehyde with H2O2 and formic acid. Methyl-
ation then gave a mixture which was separated by
chromatography to afford, as oils, trimethyl (3R)-homo-
citrate, 10a, in 36% yield and dimethyl (5R)-homocitric
We were now able to introduce a deuterium label stereo-
specifically at C-2 whilst retaining the R stereochemistry
at C-1 (which was to become C-3 in homocitric acid) by
reacting 17a with lithium aluminium deuteride. The
product 18a was obtained in 83% yield. The unlabelled
analogue 18d of this compound was prepared indepen-
dently from the quinic acid derived 9a as shown in
Scheme 4 by protection followed by reduction using
LiAlH4. Comparison of the spectra of the shikimate-de-
rived 18a and the quinate-derived 18d confirmed the
stereochemistry at the centre C-1 in the former com-
pound. All attempts to oxidise the primary alcohol
group in 18a to an acid failed, although, when we oxi-
dised the unlabelled compound 18d using oxygen and
a platinum catalyst, followed by methylation and depro-
tection, we were able to obtain the desired methyl
5-deoxyquinate 9a. An isotope effect had evidently pre-
vented oxidation of 18a and indeed when the dideute-
1
lactone, 11a, in 15% yield. The H NMR spectra of
these compounds (Figs. 2a and 3a) showed good separa-
tion between the signals due to the diastereotopic hydro-
gens which would originate from acetate in the enzyme
catalysed reaction.
Having obtained the unlabelled target molecules, the
next step was to synthesise the two diastereoisomerically
rio-compound 18e
was obtained from methyl
deoxyquinate, as in Scheme 4, this was also resistant
to oxidation. The problem was circumvented by selec-
tively reducing the ester in 17a without affecting the
This compound had the required analytical and spectroscopic
properties.