Epimerization in acid degradation products of artemisinin

Article abstract of DOI:10.1039/a702714a

Treatment of artemisinin 1 with acid leads to either a cyclohexane dione degradation product 10, which is a useful intermediate for biosynthetic studies of artemisinin, or to a decalin system which has undergone epimerization 8. It is shown by NMR spectro

Full text of DOI:10.1039/a702714a

View Article Online / Journal Homepage / Table of Contents for this issue
Epimerization in acid degradation products of artemisinin
Shi-Man Hui, Koon-Sin Ngo and Geoffrey D. Brown*
Chemistry Department, The University of Hong Kong, Pokfulam Rd., Hong Kong
Treatment of artemisinin 1 with acid leads to either a cyclohexane dione degradation product 10, which is
a useful intermediate for biosynthetic studies of artemisinin, or to a decalin system which has undergone
epimerization 8. It is shown by NMR spectroscopy, chemical reactions and molecular modelling that the
bulky 7-substituent in the epimerized decalin series (8, 17, 14) adopts an axial solution conformation
and that this is thermodynamically favoured over the natural configuration for which this substituent is
equatorial (11, 15, 13). Conversely, for the cyclohexane dione series, the natural configuration in which the
7-substituent is equatorial is more favoured. Reasons for the differing conformational preferences in the
two series, which are ultimately responsible for promoting epimerization, are discussed and a simple
spectroscopic procedure for identification of epimerized products is presented.
accompanied by loss of formaldehyde to give a variety of prod-
ucts including the nor-sesquiterpenes 10 and 87–12 (Scheme 1).
Introduction
Artemisinin (qinghaosu) 1, a natural product from the Chinese
plant Artemisia annua,1 is one of the most promising new anti-
malarial drugs, which is currently exciting great interest both in
respect of its biological activity and its unusual chemical struc-
ture.2 We are particularly interested in delineating the bio-
synthetic route to 1, and have adopted a strategy which first
involves synthesis of a variety of postulated biosynthetic inter-
mediates such as 2a–4b3,4 incorporating C/H isotopic labels at
H
H
H2SO4
AcOH
O
O
O
O
H
O
O
O
O
14
1
8
H
10
H
H
O
H+/MeOH
4
14
14
1
O
15
O
H
H
H
10
10
O
6
H+
Ba(OH)2
O
5
H
1
1
11
15
O
H
O
O
4
4
O
6
6
O
13
12
5
R
O
R
MeO
HO
O
11
O
H
O
12
11
HO
O
13
MeO
MeO
1
2a R = Me
3a R = Me
13
12
O
2b R = =CH2
3b R = =CH2
O
O
9
10
11
14
Scheme 1 Reported acid-degradation reactions of artemisinin
H
H
H
10
The cyclohexane dione 10 would seem to have good potential
for conversion into the biosynthetic intermediates required for
our project; Robinson annulation of 10 under mild conditions
has been reported to give the decalin acid 11 in the natural
configuration.12 We were interested to note that the decalin
lactone 8, produced directly by treatment of 1 with sulfuric
acid–acetic acid, was reported to have undergone epimerization
at the 7-position.7,9–11 Because control of absolute chemistry
in the preparation of isotopically labelled 2a–4b is obviously
critical if these products are to be useful in establishing a bio-
synthetic pathway, we have set out to determine the reasons for
such epimerization and establish a simple means for recognizing
its occurrence.
1
6
4
5
O
O
O
O
O
H
5
11
12
O
HO
R
13
MeO
O
MeO
O
O
4a R = Me
5
6
4b R = =CH2
H
O
O
Results and discussion
O
5′
Treatment of the endoperoxide 9 (obtained from 1 by a liter-
ature procedure8) with acid led exclusively to the expected
cyclohexane dione 10. An alternative direct procedure, involv-
ing treatment of 1 with 30% MeOH–H2SO4 also yielded pre-
dominantly 10, contaminated by small amounts of products 5
and 6, which had undergone epimerization at the 1- and/or
7-positions. 13C and 1H NMR assignments (Table 1) for
these three isomers (and for all compounds cited herein)
were rigorously determined by 2D-NMR (HSQC, HMBC and
2′
1′
3′
7
the 15- and/or 5-positions, and then feeding of these intermedi-
ates to the plant. Some of these target compounds such as
artemisinic acid 2b, arteannuin B 3b and the seco-cadinane 4b
have also been reported as natural products from Artemisia
annua.3,5,6
Compound 1 has been reported to undergo acid degradation
J. Chem. Soc., Perkin Trans. 1, 1997
3435

View Article Online
Table 1 NMR data for diketone cyclohexane compounds used in Robinson annulations
δC
δH
Atom
5
6
7
10
56.8
12
56.6
23
57.0
5
6
7
10
12
23
1
2
56.5
25.3
53.8
21.3
56.8
20.2
2.12
2.00
1.99
2.41
2.41
2.56
1.97
1.39
2.49
2.29
1.96
1.80
1.75
2.49
2.28
2.08
1.85
1.80
2.54
2.35
2.08
1.84
1.77
2.58
2.39
2.02
1.80
1.74
2.50
2.32
20.2
20.2
20.2
3
40.7
41.7
41.4
41.2
41.3
41.3
4
5
6
208.4
30.2
214.9
49.2
209.0
29.9
212.1
53.2
209.2
29.9
213.2
51.1
209.0
29.8
211.5
53.5
209.1
29.9
211.2
58.3
209.0
29.9
213.1
54.0
2.13
2.12
2.10
2.12
2.12
2.10
7
2.78
1.96
1.58
1.49
2.06
2.06
2.67
2.68
2.03
1.56
1.72
2.02
2.38
2.68
2.45
1.94
1.32
1.46
1.82
1.50
2.34
3.30
3.30
0.87
1.06
2.61
2.00
1.55
1.52
1.90
1.60
2.77
3.00
2.04
1.77
1.58
1.92
1.61
2.38
2.05
1.45
1.50
1.85
1.55
2.15
3.45
3.37
1.00
1.05
8α
8β
9α
9β
10
11
12
25.7
26.2
28.1
31.0
31.6
30.6
27.0
32.2
34.6
34.5
34.5
34.8
36.9
38.8
177.3
37.4
38.8
177.0
40.1
31.3
72.9
40.2
39.3
176.1
40.0
143.5
112.5
40.5
32.7
72.9
4.92
4.70
1.74
1.10
13
14
OMe
1Ј
14.4
19.5
51.7
14.5
13.3
51.7
13.4
20.7
15.1
20.5
51.6
21.5
20.7
16.0
20.6
1.12
0.95
3.67
1.12
0.73
3.69
1.18
1.08
3.67
73.4
73.0
4.48
4.48
—
7.32
7.32
7.32
4.49
4.45
—
7.31
7.31
7.31
2Ј
138.6
128.4
128.3
127.6
138.7
128.3
127.5
127.4
3Ј/7Ј
4Ј/6Ј
5Ј
1H-1H COSY). Following complete assignments for all protons,
relative stereochemistry and conformations were determined
from NOESY correlations, supplemented by the results of
1H-1H J-resolved spectra. Critical NOESY correlations used in
determining the configurations and conformations of 10, 5
and 6 are shown in Scheme 2. The axial nature of the 14-methyl
retention of configuration at both 1- and 7-positions. As for the
cyclohexane diones, this could be clearly demonstrated by use
1
of NMR techniques such as NOESY and H-1H J-resolved
spectroscopy which showed the 7-isopropanoic acid substitu-
ent to be equatorial (NOESY correlations for the correspond-
ing methyl ester 13 were identical; Scheme 3). As a solution in
CDCl3 the free acid 11 was converted slowly into the cis-lactone
15, in which the 7-substituent remained equatorial and the
decalin system was trans-fused.
The epimeric cis-lactone 8 (obtained directly from degrad-
ation of 1 in acetic–sulfuric acid) was shown by NOESY to
adopt a conformation with the 7-substituent axial and the
decalin system cis-fused (Scheme 3). This was confirmed by the
Me
O
Me
H
H
O
CH2
Me
Me
Me
CO2Me
H2C
H
H
Me
CO2Me
H
O
H
H
O
H
H
H
H
H
H
10
5
Me
Me
CO2R
Me
H
H
H
O
H
O
H
H
Me
H
Me
H
H
H
Me
H
H
H
H
CO2Me
CO2Me
O
H
14
11 R = H
H
Me
H
H
O
H2C
13 R = Me
O
CH2
Me
O
H
Me
O
H
O
H
H
O
H
H
H
H
H
O
6
12
H
H
H
Me
O
H
Scheme 2 Critical NOESY correlations used in assigning the config-
uration and solution conformations of 10, 5, 6 and 12 indicated by
double-headed arrows
H
Me
Me
H
O
H
H
H
Me
group in the doubly epimerized compound 6 was confirmed
by a large upfield shift in the 13C NMR spectrum (ca. 7 ppm)
for this resonance when compared with 10 or 5, which is the
result of gauche interactions; in the 1H NMR spectrum of 6 the
equatorial 10-proton had undergone a 0.5 ppm downfield shift
relative to its axial counterpart in compound 10. The bulky 7-
substituent remained equatorial or pseudo-equatorial for all
three cyclohexane dione epimers.
15
8
Scheme
3
Critical NOESY correlations used in assigning the
configuration and solution conformations of 11, 13, 14, 15 and 8
indicated by double-headed arrows
results of 1H-1H J-resolved spectra which demonstrated a
W-coupling (2 Hz) between H-5α (dd, J 16, 2 Hz) and the
equatorial H-7 (ddd, J 13, 7, 2 Hz). [No such coupling was
observed in J-resolved spectra of 15, for which the H-5β proton
shared a W-coupling (2 Hz) with H-3β (ddd, J 13, 7, 2 Hz) but
As expected, Robinson annulation of compound 10 in the
presence of barium hydroxide gave the decalin acid 11 with
3436
J. Chem. Soc., Perkin Trans. 1, 1997

View Article Online
H
H
O
CDCl3
O
O
HO
O
O
11
15
MeOH
DCC
H
H
H
NaBH4
HO
O
MeO
MeO
O
O
13
16
H
O
H
H
KOH
CDCl3
CH2N2
O
O
O
HO
MeO
O
O
O
8
17
14
NaBH4
H
H
H
H
H
NaBH4
HO
O
–O
H
MeO
O
Fig. 1 Rate of lactonization of 11 into 15 determined by ratio of
integrals in 1H NMR at δ 1.27 (H-13, 11) and δ 1.13 (H-13, 15) as
O
O
18
9.7 × 10Ϫ3 Ϫ1; rate of lactonization of 17 into 8 determined by ratio
h
of integrals in 1H NMR at δ 5.93 (H-5, 17) and δ 2.66 (H-11, 8) as
Scheme 4 Differing reactions of 7-epimeric degradation products 11
and 8
1.2 × 10Ϫ2 minϪ1
not with the axial H-7.] Base-catalysed ring opening of 8 gave
the free acid 17, which quickly reverted to the parent lactone in
CDCl3 solution and was best derivatized as the methyl ester 14
for NMR analysis. The results of NOESY clearly indicated that
the 7α-methyl isopropanoate ester substituent in 14 was also
axial.
system; reduction at the ketone group is then only possible from
the α-face). The relative stereochemistry of 18 was determined
1
from H-1H couplings: H-1 (dddd, J 11.4, 11.4, 11.4, 2.5 Hz)
shows three large couplings to its neighbours and therefore
requires a trans-fused decalin system, whilst H-5 (dd, J 6.3, 6.0
Hz) is equatorial and H-4 (ddd, J 11.7, 5.8, 6.0 Hz) is axial
(these predictions were also confirmed by NOESY). The
cis-decalin system of 16 was demonstrated by NOESY spectra
(H-6 correlates both with H-4 and H-1) and the axial nature of
H-4 (dddd, J 11.7, 11.7, 4.9, 4.9 Hz) was confirmed by observ-
ation of two large coupling constants in 1H NMR. The
unexpected reaction pathway followed by 14 and the stereo-
chemistry of the resulting decalin system is simply explained as
a trapping of the enolate intermediate, generated during conju-
gate reduction of 14, by the axially disposed methyl ester group.
Such trapping is much less likely for the equatorial methyl ester
of 13.
One useful consequence of the preferred solution conform-
ations for epimeric acid degradation products of artemisinin is
that it is possible to distinguish products in which epimerization
has occurred (such as 8, 17 and 14) from those in which it has
not (such as 11, 15 and 13) by simple inspection of the 1H NMR
spectrum. A diagnostic coupling pattern (dq, J ≈ 12, 7 Hz) for
the well-resolved H-11 resonance (δ 2.6–3.1 ppm) was noted for
all epimerized products that was obviously different from the
pattern for products of natural configuration (dq, J ≈ 7, 7 Hz).
This difference is a consequence of the axial or equatorial
The differing conformations predicted by NMR for the two
decalin 7-epimeric series were confirmed by their chemical reac-
tions (Scheme 4). Thus, free acid 11 underwent lactonization to
15 when stored in CDCl3 some 75 times more slowly than the
7α-epimer 17 (Fig. 1). The rapid lactonization of 17 is con-
sistent with the axial disposition of the carboxy group in this
epimer. In the axial conformation the carboxylic acid nucleo-
phile can easily participate in an intramolecular Michael reac-
tion since it is well positioned to interact axially with the
π-system of the enone: good interaction is much more difficult
for the equatorial carboxy group in 11 (cf. Scheme 3). (Inciden-
tally, this difference in reactivity also accounts for the observ-
ation that the decalin 7α-epimer is normally isolated as a
lactone 8 from acid degradation of artemisinin, whilst the
product with natural configuration 11 is obtained as the free
acid.) Sodium borohydride reduction of the α-epimeric methyl
ester 14 yielded a tricyclic product incorporating a trans-decalin
ring 18, whereas reduction of the corresponding β-epimer
13 gave the expected reduction product 16 (the equatorial
7-substituent in 13 blocks initial hydride attack at the conju-
gated double bond from the β-face resulting in a cis-decalin
J. Chem. Soc., Perkin Trans. 1, 1997
3437

View Article Online
Table 2a 13C NMR data for decalin products from Robinson annulations
Atom
8
11
45.9
24.6
34.2
199.9
121.4
168.0
48.3
33.0
35.0
38.6
40.5
13
46.0
24.5
34.0
199.6
121.3
168.1
48.7
33.4
35.0
38.5
40.6
14
40.8
26.2
36.1
199.8
125.6
167.1
47.9
26.8
29.1
39.4
41.6
15
48.1
17
40.7
26.1
35.9
200.5
125.5
167.3
47.6
26.7
29.1
39.5
41.5
20
45.0
26.1
35.8
201.0
123.2
167.7
52.4
31.7
34.6
39.0
144.7
113.8
21.4
22
41.6
24
46.1
25
1
2
3
4
5
6
7
8
9
44.2
21.6
35.9
41.0
25.6
35.5
200.0
125.1
169.8
48.1
28.6
29.7
39.4
33.2
74.2
15.8
20.2
24.0
40.3
207.0
50.7
85.9
43.3
24.4
32.5
31.2
39.9
178.0
9.2
26.2
36.1
25.0
34.5
199.9
121.6
169.8
48.7
30.9
35.3
38.9
32.9
72.6
17.1
20.3
207.2
47.2
87.6
48.9
22.5
29.6
30.7
36.4
177.7
13.3
18.9
*
126.8
*
49.9
28.5
30.2
*
10
11
12
13
14
OMe
1Ј
*
180.2
15.8
20.2
176.1
16.0
20.2
175.4
15.0
20.0
179.4
14.9
20.0
112.6
22.3
20.2
19.9
20.3
51.7
51.4
73.2
138.6
128.4
127.5
127.5
73.2
138.3
128.4
127.7
127.6
2Ј
3Ј/7Ј
4Ј/6Ј
5Ј
* Resonances not clearly resolved from those of 20.
nature of the 7-substituent in the two epimeric series. For the
α-epimers which adopt an axial conformation for the 7-
substituent, H-11 must reside below the plane of the decalin
ring in order to allow the more bulky methyl and carboxy
groups to project away from the ring system. This leads to a
dihedral angle of 180Њ between H-7 and H-11 and a relatively
H
O
H
H
Ba(OH)2
or H+
Ba(OH)2
H+
11
O
O
O
O
O
H
MeO
O
MeO
1
large 3-bond H-1H coupling constant (≈12 Hz) as predicted
O
O
by the Karplus equation. For the β-epimeric series, the 7-
substituent is equatorial and a smaller dihedral angle arises
between H-7 and H-11 (Scheme 3), the H-7/H-11 coupling
constant is therefore correspondingly smaller (≈7 Hz). It is also
evident from inspection of Table 2a, that 13C chemical shifts
for the B-ring are generally upfield (by up to 7 ppm) for com-
pounds belonging to the epimerized series as compared with
those of natural configuration. This is the result of gauche
effects arising from the axial nature of the 7-substituent in the
O
10
1
13
H+
H
H
H
H+
H+
8
O
O
HO
HO
MeO
MeO
α-epimer. Consequently, inspection of either the 1D- 1H or 13
C
O
O
O
19
spectrum is generally considered sufficient to recognize the
occurrence of epimerization for the decalin series of acid deg-
radation products.
17
14
Scheme 5 Postulated mechanism for epimerization in acid-catalysed
degradation of artemisinin
The 7α-epimeric lactone 8, obtained from acid treatment of
1, must have arisen from Robinson annulation of a cyclohexane
dione intermediate such as 10. Since we have shown that there is
little tendency for epimerization in the formation of 10 (the
configuration of 10 is such that all substituents around the
cyclohexanone ring are able to adopt equatorial conformations:
compound 10 was always the predominant product accom-
panied by only small amounts of 5 and 6), epimerization at the
7-position must occur following formation of the new carbo-
cyclic ring (Scheme 5). The most natural mechanism would
involve the intermediacy of an extended enol, 19, which is gen-
erated in acidic medium: reversible reprotonation/deproton-
ation of 19 from either the α or β face allows formation of
epimers at C-7. (Presumably such a mechanism does not oper-
ate under the heterogeneous reaction conditions when Ba(OH)2
is employed as the cyclization catalyst.)
Some evidence for the intermediacy of an extended enol
intermediate such as 19 comes from study of Robinson annul-
ation of the closely related compound 12 (prepared in two steps
from isopulegol). Compound 12 was obtained exclusively as the
1β,7β-epimer as proven by NOESY correlations and chemical
shift values which compared well with those of 10 (see Table 1
and Scheme 2). The major product of annulation of the cyclo-
hexane dione 12 is the highly conjugated compound 21 which
has undergone isomerization of the terminal double bond
(Scheme 6). Only small quantities of the two 7-epimers 20 and
22 were isolated. (NOESY spectra used to assign conform-
ations for 20 and 22 gave comparable results to those shown in
Scheme 3 for 11 and 14, e.g. correlations were observed from
H-7 to H-9α and H-1α in compound 20; but only from H-7 to
H-5 in compound 22. 1H and 13C chemical shift assignments for
20 and 22 also compared favourably with those for 11/13 and
17/14 shown in Tables 2a and 2b.) Such double bond isomeriz-
ation is naturally explained as a consequence of formation of
an extended enolic intermediate analogous to 19.
Exclusive formation of the 7α-epimer 8 in the acid degrad-
ation of artemisinin, via this mechanism of extended enol for-
mation, must then imply that the epimerized decalin product is
thermodynamically favoured over that of the natural conform-
ation since both epimers exist in equilibrium. Further evidence
for this conclusion came from Robinson annulation of the
related compound 23 [prepared as the 1β,7β-epimer together
with a little 1α,7β-epimer 7 (see Table 1) in four steps from
isopulegol]. The stereochemistry and conformation of the
cyclohexane dione 23 was determined by 1H-1H J-resolved
spectroscopy as being all equatorial because both H-7 (ddd,
J 12.4, 5.8, 5.8 Hz) and H-1 (ddd, J 10.9, 8.5, 3.2 Hz) were
clearly axial and shown to be involved in large couplings with
H-8α (12.4 Hz) and H-10 (10.9 Hz), respectively. That 7 is the
1-epimer was demonstrated by similar means [H-7 (ddd, J 12.4,
3438
J. Chem. Soc., Perkin Trans. 1, 1997

View Article Online
Table 2b 1H NMR data for decalin products from Robinson annulations
Atom
8
11
13
14
15
17
20
22
24
25
1
1.48
2.12
2.08
2.30
1.89
2.15
1.93
2.32
2.37
5.80
1.89
2.14
1.96
2.30
2.36
5.76
2.21
2.25
1.59
2.37
2.20
5.76
1.49
2.19
1.85
2.33
2.23
2.25
1.56
2.26
2.37
5.93
1.94
2.25
1.72
2.28
2.39
5.77
2.10
2.24
1.69
2.20
2.40
5.91
1.85
2.15
1.85
2.25
2.35
5.88
2.03
2.09
1.63
2.12
2.32
5.82
2α
2β
3α
3β
5
2.30
2.54
2.85 (β)
2.35 (α)
1.85
1.84
1.80
1.30
1.73
1.68
2.66
2.79 (β)
2.35 (α)
2.15
1.76
1.12
1.10
1.72
1.37
3.02
7
8α
8β
9α
9β
10
11
12
2.35
2.02
1.38
1.32
1.86
1.64
2.75
2.35
1.89
1.31
1.31
1.85
1.64
2.73
2.47
1.97
1.65
1.35
1.62
1.36
2.82
2.52
1.98
1.62
1.34
1.59
1.35
2.75
2.79
1.89
1.60
1.33
1.87
1.45
—
4.98
4.79
1.71
1.05
3.02
2.14
1.69
1.40
1.58
1.42
—
4.96
4.74
1.71
1.01
2.12
2.05
1.12
1.25
1.82
1.56
2.08
3.48
3.35
1.07
1.02
2.16
1.98
1.55
1.40
1.58
1.40
2.05
3.24
3.11
1.05
1.01
13
14
OMe
1Ј
1.21
1.04
1.27
1.05
1.22
1.04
3.69
1.20
1.05
3.54
1.13
1.00
1.22
1.02
4.51
4.45
7.32
7.32
7.32
4.39
4.39
7.32
7.32
7.32
3Ј/7Ј
4Ј/6Ј
5Ј
Table 3 Calculated energies (kJ molϪ1) for most stable conformers
of 7-epimeric artemisinin degradation products*
H
H
H
Ba(OH)2
Class of
compound
Energy
Energy
(kJ molϪ1
)
7α-series
(kJ molϪ1
)
O
–O
7β-series
O
O
Cyclohexane
diones
Decalins
(free acid)
Decalins
(methyl ester)
10
11
13
123.0
78.2
5
127.4
66.9
12
20
17
14
141.4
128.1
H
H
* Calculated using MM2 force field model.
O
O
H
H
H
Ba(OH)2
O
12 months
O
22
21
O
O
Scheme 6 Evidence for participation of an extended enolic inter-
mediate in epimerization
OBz
OBz
OBz
5.5, 5.2 Hz) shared a large coupling with H-8α (12.4 Hz) and
therefore remained axial, whereas H-1 (ddd, J 7.6, 5.2, 1.7 Hz)
is equatorial since it shares a small coupling (5.2 Hz) with
H-10]. Robinson annulation of 23 produced the decalin prod-
uct of natural configuration 24 as expected when Ba(OH)2 was
employed as catalyst at room temperature (significantly, a 1:1
mixture of 7-epimers was obtained at reflux, however). Follow-
ing storage at room temperature for 1 year, it was discovered
that 24 had been largely converted into the 7α-epimer 25 which
must therefore be thermodynamically more stable (Scheme 7).
Epimerization in 25 was determined by results of 1H-1H
J-resolved spectroscopy [H-7, δ 2.16 (ddd, J 11.3, 5.3, 3.0 Hz)
now shared an 11.3 Hz coupling with H-11 which, as com-
mented on previously, is diagnostic for epimerization at the
7-centre] and confirmed by NOESY correlations (as for com-
23
24
25
3′
1′
CH2
2′
5′
Bz =
Scheme 7 Evidence for 7α-epimer being more thermodynamically
stable
for the 7-epimeric cyclohexane diones 10 and 5 have also con-
firmed that, conversely, the natural configuration is of lower
energy than its 7-epimer and that both isomers prefer the
equatorial or pseudo-equatorial conformations predicted by
NMR.
The relative stability and conformations adopted by the
cyclohexane diones used in Robinson annulation conforms
with a simplistic analysis in which bulky ring-substituents are
assumed always to prefer an equatorial conformation, which is
normally of lower energy. The conclusion from both modelling
and experiment that the 7α-epimers of the cyclized decalin
series are more stable and that the bulky 7-substituent of the
epimerized series preferentially adopts an axial conformation
does not accord with such a simplistic analysis. Furthermore,
following epimerization, it is possible for either the 7- or the 10-
substituent to adopt the energetically disfavoured axial con-
1
pound 14 in Scheme 3). H and 13C chemical shifts for 25 also
agreed well with those for epimerized products 17, 14 and 22.
Molecular modelling studies have confirmed all the preceed-
ing experimental results. The 7α-epimers of the decalin series of
compounds are indeed expected to be significantly more stable
than their 7β-counterparts by more than 10 kJ molϪ1 (Table 3).
The predicted lowest energy conformations demonstrate an
axial conformation preference for the α-epimers and an equa-
torial preference for the β-epimers. Molecular modelling studies
J. Chem. Soc., Perkin Trans. 1, 1997
3439

View Article Online
formation by flipping in the B-ring and, given a choice, the
simplistic model would predict that the less bulky 10-methyl
group should become axial (as observed for compound 6 for
example) rather than the 7-substituent.
[Mϩ Ϫ O2, ∆ = 0.0 mmu for C17H28O4] (3), 264 (3), 251 (4), 225
(11), 193 (15), 179 (40), 165 (24), 151 (29), 133 (18), 101 (47) and
86 (100).
From the point of view of the 7-substituent, the major differ-
ence between the two series of compounds is the presence of
either a vicinal double bond to O in the cyclohexane diones or a
vicinal double bond to CH(R) in the decalins. It seems that the
overriding factor in destabilizing the 7β-equatorial substituent
in the decalin series is an interaction with the proton of the
double bond, which is absent from the cyclohexane dione series.
Such destabilizing influences, designated A1,3 strain,13 have been
proposed previously for 2-substituted methylenecyclohexane
systems, for which bulky substituents have been observed to be
axial rather than equatorial.
We therefore propose that the tendency to epimerize in the
acid degradation of artemisinin is attributable to the appear-
ance of A1,3 strain following formation of a decalin ring by
Robinson annulation. Whilst products of natural configuration
are unable to overcome such strain (flipping of the B-ring
would require both 10- and 7-substituents to become axial,
which is evidently even more energetically disfavoured than
allowing A1,3 strain), epimerization at the 7-centre allows only
the 7-substituent to become axial and the ensuing relief of A1,3
strain is thereby energetically favoured. Given that the natural
tendency in preparation of postulated biosynthetic intermedi-
ates 2a–4b is expected to be towards epimerization, such a
possibility is best guarded against by employing mild conditions
or displacing the 5,6 double bond at an early stage. Fortunately,
as a consequence of our studies, simple 1D-NMR procedures
now exist to detect such unwanted epimerization when it does
occur.
Direct conversion of 1 into the cyclohexane dione 10
Compound 1 (5.0 g) was added to H2SO4 (12 )–MeOH (4:10,
v/v; 140 ml). After being rapidly stirred, the reaction mixture
was poured into cold water (100 ml). Unchanged 1 was filtered
off and the filtrate was then extracted with CHCl3 (3 × 100 ml).
The combined extracts were washed with water (3 × 300 ml),
dried (MgSO4) and concentrated to give a pale yellow liquid
which was subjected to column chromatography [hexane–
EtOAc (3:1, v/v)] to yield 10 (3.6 g) together with small quan-
tities of the epimers 5 and 6.
Compound 10. Oil, δH(CDCl3) 3.67 (3H, s), 2.77 (1H, dq,
J 7.0, 7.0), 2.61 (1H, ddd, J 11.9, 7.0, 5.4), 2.54 (1H, ddd, J 17.3,
9.4, 5.3), 2.38 (1H, ddd, J 17.3, 9.2, 6.1), 2.12 (3H, s), 1.18 (3H,
d, J 7.0) and 1.08 (3H, d, J 6.1); HREIMS m/z (rel. int %):
268.1672 [Mϩ, ∆ = 0.2 mmu for C15H24O4] (2), 250 (100), 235
(10), 209 (30), 180 (50), 163 (28) and 151 (28).
Compound 5. Oil, [α]D Ϫ11.2 (c 0.46 in CHCl3); νmax/cmϪ1
(CHCl3) 2957, 2871, 1732 and 1711; δH(CDCl3) 3.67 (3H, s),
2.78 (1H, ddd, J 11.5, 9.2, 6.2), 2.67 (1H, dq, J 9.1, 7.1), 2.13
(3H, s), 1.12 (3H, d, J 7.1) and 0.95 (3H, d, J 7.0); HREIMS m/z
(rel. int %) 250.1567 [M Ϫ H2O, ∆ = 0.2 mmu for C15H22O3]
(19), 209 (50), 208 (45), 180 (100), 163 (80) and 119 (95).
Compound 6. Oil, [α]D ϩ61.5 (c 1.03 in CHCl3); νmax/cmϪ1
(CHCl3) 2955, 2899, 1732 and 1715; δH(CDCl3) 3.69 (3H, s),
2.68 (2H, m), 2.49 (1H, ddd, J 17.4, 8.8, 3.0), 2.29 (1H, ddd,
J 17.4, 8.5, 6.7), 2.12 (3H, s), 1.56 (1H, dddd, J 12.9, 12.9, 12.9,
4.0), 1.12 (3H, d, J 6.8), 0.74 (3H, d, J 7.1); HREIMS m/z (rel.
int %) 268.1681 (Mϩ, ∆ = Ϫ0.6 mmu for C15H24O4) (0.5), 250
(100), 218 (45), 208 (50), 190 (65), 180 (70), 178 (75), 153 (80)
and 136 (60).
Experimental
Conversion of 9 into 10
General details
Compound 9 (1 g) was dissolved in dilute aq. HCl (10%; 10 ml)
and the solution stirred for 30 min, after which it was extracted
with CHCl3. The extract was washed, dried (MgSO4) and evap-
orated under reduced pressure to yield compound 10 (0.3 g).
All NMR experiments were run on Bruker DPX 300 or DRX
500 instruments with CDCl3 as solvent. The chemical shifts
were recorded relative to TMS as an internal standard and all
coupling constants, J, are reported in Hz. Two-dimensional
1
spectra such as HSQC, HMBC, H-1H COSY, NOESY and
1H-1H J-resolve were normally recorded with 1024 data points
in F2 and 256 data points in F1 (occasional high resolution
experiments had 4096 data points in F2 and 1024 data points in
F1). High resolution MS were recorded in EI mode (70 eV) on a
Finnigan-MAT 95 MS spectrometer. FTIR spectra were
recorded in CH2Cl2 on a Shimadzu FTIR-8201 PC instrument.
TLC plates were developed using p-anisaldehyde. Column
chromatography was performed using silica gel 60–200 µm
(Merck). HPLC separations were performed using a PREP-SIL
20 mm × 25 cm column, flow rate 8 ml minϪ1. Optical rotations,
[α]D, are recorded as 10Ϫ1 deg cm2 gϪ1.
Acid conversion of 1 into the decalin lactone 8
Compound 1 (30.0 g) was added to H2SO4 (12 )–AcOH
(glacial) (1:10, v/v; 330 ml). After being stirred overnight at
room temperature, the reaction mixture was poured into cold
water (500 ml), and extracted with CHCl3 (3 × 300 ml). The
combined extracts were then washed with water (5 × 600 ml),
dried (MgSO4) and evaporated. The residual dark-brown gum
was subjected to recrystallization twice (EtOH) to yield 8 (10.2
g) as crystals, mp 151–152 ЊC; [α]D Ϫ15.9 (c 1.1 in CHCl3); νmax
/
cmϪ1 (CHCl3) 2965, 2934, 1771 and 1720; δH(CDCl3) 2.85 (1H,
d, J 15.9), 2.66 (1H, dq, J 13.8, 6.9), 2.35 (1H, dd, J 15.9, 2.0),
1.21 (3H, d, J 7.0) and 1.04 (3H, d, J 6.2); HREIMS m/z (rel. int
%) 236.1419 (Mϩ, ∆ = Ϫ0.7 mmu for C14H20O3), 208 (65), 192
(70), 163 (70), 134 (50) and 119 (100).
Acid conversion of 1 into the endoperoxide 9
Compound 1 (7.0 g) was added to ice-cooled H2SO4 (12 )–
MeOH (3:10, v/v; 130 ml). After being stirred for 3 h with ice-
cooling, the reaction mixture was poured into cold water (200
ml), and extracted with CHCl3 (3 × 100 ml). The combined
extracts were then washed with water (3 × 300 ml), dried
(MgSO4), and evaporated to yield 9 (3.5 g) as a liquid, [α]D
ϩ156 (c 1.22 in CHCl3); νmax/cmϪ1 (CHCl3) 2955, 2941, 2871
and 1732; δH(CDCl3) 9.92 (1H, d, J 2.7, H-5), 3.65 (3H, s,
12-OMe), 3.35 (3H, s, 4-OMe), 3.18 (1H, dq, J 3.5, 6.5, H-11),
2.32 (1H, m, H-10), 1.23 (3H, d, J 7.2, H-13), 1.19 (3H, s, H-4)
and 0.89 (3H, d, J 6.5, H-14); δC(CDCl3) 200.7 (CH, C-5), 175.3
(C, C-12), 108.5 (C, C-4), 94.1 (C, C-6), 59.5 (CH, C-1), 51.4
(CH3, 12-OMe), 49.1 (CH3, 4-OMe), 48.7 (CH, C-7), 40.6 (CH2,
C-3), 37.9 (CH, C-11), 35.5 (CH2, C-9), 32.0 (CH, C-10), 25.3
(CH2, C-8), 22.1 (CH2, C-2), 20.5 (CH3, C-14), 19.6 (CH3, C-15)
and 17.8 (CH3, C-13); HREIMS m/z (rel. int %): 296.1988
Robinson annulation of 10
Ba(OH)2ؒ8H2O (4.5 g) was added to a solution of 10 (3.6 g) in
EtOH (110 ml). After being stirred for 2 h at room temperature
the reaction mixture was acidified with aq. HCl (3 ). The
resulting precipitate was filtered off and dissolved in CH2Cl2,
and the solution was washed with water, dried (MgSO4) and
evaporated to yield 11 (1.1 g) as a solid, mp 158–159 ЊC; [α]D
ϩ13.9 (c 0.17 in CHCl3); νmax/cmϪ1 (CHCl3) 3400–2400br, 2959,
2931, 2874, 1709 and 1669; δH(CDCl3) 5.80 (1H, s), 2.75 (1H,
dq, J 8.2, 7.0), 1.27 (3H, d, J 6.9) and 1.04 (3H, d, J 6.5);
HREIMS m/z (rel. int %) 236.1412 [Mϩ, C14H20O3 ∆ = 0.0
mmu] (75), 191 (70) and 163 (100). When left in CDCl3, com-
pound 11 was slowly converted into compound 15, a solid, mp
171–172 ЊC; [α]D Ϫ60.9 (c 0.36, CHCl3); νmax/cmϪ1 (CHCl3)
3440
J. Chem. Soc., Perkin Trans. 1, 1997

View Article Online
2938, 2862, 1772 and 1717; δH(CDCl3) 3.02 (1H, dq, J 6.9, 6.9),
2.79 (1H, dd, J 14.9, 2.0), 2.54 (1H, ddd, J 15.4, 2.5, 2.5), 2.35
(1H, d, J 14.9), 1.13 (3H, d, J 7.2) and 1.00 (3H, d, J 6.5);
HREIMS m/z (rel. int %) 236.1412 (Mϩ, ∆ = 0.3 mmu for
C14H20O3) (93), 208 (100), 180 (75), 179 (72), 163 (67) and 134
(35).
NaBH4 reduction of 13
Compound 13 was reduced under similar conditions to those
employed with compound 14 to yield 16 as an oil, [α]D Ϫ14.6 (c
0.58 in CHCl3); νmax/cmϪ1 (CHCl3) 3502br, 2924, 2854 and 1720;
δH(CDCl3) 3.66 (3H, s, OMe), 3.59 (1H, dddd, J 11.7, 11.7, 4.9,
4.9, H-4), 2.30 (1H, dq, J 11.1, 6.8, H-11), 1.10 (3H, d, J 6.9,
H-13) and 0.83 (3H, d, J 6.2, H-14); δC(CDCl3) 177.7 (C, C-12),
71.8 (CH, C-4), 51.4 (CH3, OMe), 43.7 (CH, C-7), 42.9 (CH,
C-1), 42.3 (CH, C-11), 36.0 (CH, C-6), 35.6 (CH2, C-9), 30.4
(CH2, C-3), 30.1 (CH2, C-5), 27.4 (CH, C-10), 26.5 (CH2, C-8),
26.3 (CH2, C-2), 19.6 (CH3, C-14) and 15.0 (CH3, C-13);
HREIMS m/z (rel. int %) 254.1888 [Mϩ, ∆ = Ϫ0.6 mmu for
C15H26O3] (1), 236 (2), 205 (1), 177 (2), 149 (100), 121 (2), 107 (3),
93 (3) and 88 (52).
Conversion of 8 into the free acid 17
Compound 8 (9.0 g) was dissolved in aqueous KOH (10% w/v;
180 ml), after which sufficient aq. HCl (3 ) was added to the
solution to bring it to pH ca. 3. The resulting solid was filtered
off, washed with a small amount of ice-cold water and then
subjected to suction to dry it. Compound 17 as a white solid
(4.5 g) resulted; δH(CDCl3) 5.93 (1H, s), 2.75 (1H, dq, J 11.2,
6.8), 2.52 (1H, dd, J 11.1, 3.3), 2.37 (1H, ddd, J 15.0, 4.0, 4.0),
1.22 (3H, d, J 6.7) and 1.02 (3H, d, J 5.9); HREIMS m/z (rel.
int %) 236.1409 [Mϩ, ∆ = 0.3 mmu for C14H22O3] (93), 208 (100),
180 (75), 179 (72), 163 (67) and 134 (35).
Preparation of the cyclohexane dione 12 from isopulegol
A solution of (Ϫ)-isopulegol (8.0 g) in acetone (30 ml) was
cooled to 0–5 ЊC and to it was added cooled Jones oxidation
reagent at a rate sufficient to maintain the temperature of the
reaction mixture at ca. 20 ЊC. The reaction mixture was stirred
for a further 1.5 h, after which it was extracted with light petrol-
eum (bp 40–60 ЊC; 3 × 30 ml). The combined extracts were
washed successively with saturated brine (2 × 30 ml), saturated
aq. NaHCO3 (2 × 30 ml) and then again saturated brine (30
ml), after which they were dried (MgSO4) and evaporated under
reduced pressure to give 2-isopropenyl-5-methylcyclohexanone
(5.5 g) as an oil, [α]D Ϫ3.10 (c 3.8 CHCl3); νmax/cmϪ1 (CHCl3)
3079, 3024, 3013, 2930, 2872, 1703 and 1647; δH(CDCl3) 4.94
(1H, s), 4.72 (1H, s), 2.96 (1H, dd, J 12.9, 5.4), 2.41 (1H, ddd, J
13.0, 3.5, 2.2), 1.75 (3H, s) and 1.03 (3H, d, J 6.1); δC(CDCl3)
209.8 (C), 143.6 (C), 112.8 (CH2), 57.9 (CH), 50.7 (CH2), 35.3
(CH2), 34.0 (CH2), 31.3 (CH2), 22.3 (CH3) and 21.3 (CH3). A
solution of 2-isopropenyl-5-methylcyclohexanone (100 mg) in
THF (1 ml) was added dropwise to a solution of lithium diiso-
propylamine (LDA) in THF at Ϫ78 ЊC [prepared from BuLi
(1.6 ; 0.5 ml) diisopropylamine (0.5 ml) and THF (3 ml)].
After being stirred for 30 min, the mixture was treated with 3-
trimethylsilylbut-3-en-2-one14 (140 mg) in THF (1 ml), added
dropwise, after which stirring was continued at Ϫ78 ЊC for 1 h.
The reaction mixture was then allowed to warm to 0 ЊC (2.5 h)
after which it was acidified with 10% aq. HCl (pH ca. 3) and
stirred for a further 15 min. It was then neutralized with 5% aq.
NaHCO3 and extracted with EtOAc. The extract was washed,
dried and concentrated to give 12 as a crude product (55 mg).
This was purified by column chromatography (15% EtOAc–
hexane) to afford an oil, [α]D ϩ27.2 (c 1.6 in CHCl3); νmax/cmϪ1
(CHCl3); 2959, 2934, 2860 and 1708; δH(CDCl3) 4.92 (1H, s),
4.70 (1H, s), 3.00 (1H, d, J 13.0, 5.2), 2.58 (1H, ddd, J 17.2, 9.2,
5.4), 2.39 (1H, ddd, J 17.2, 9.1, 6.3), 2.12 (3H, s), 1.74 (3H, s)
and 1.10 (3H, d, J 5.9); HREIMS m/z (rel. int %) 222.1617 (Mϩ,
∆ = 0.3 mmu for C14H22O2) (85), 207 (20), 164 (70) and 109
(100).
Methylation of 11
To a mixture of 11 (0.20 g), MeOH (0.14 g) and 4-dimethyl-
aminopyridine (0.04 g) in CH2Cl2 (2 ml), was added a solution
of dicyclohexylcarbodiimide (DCC) in CH2Cl2 (1.446 ; 0.4
ml). After the mixture had been stirred overnight, a precipitate
formed which was filtered off. The filtrate was concentrated to
give 13 as a colourless liquid (0.08 g). This was purified by
preparative TLC (hexane–EtOAc, 7:3) to afford an oil, [α]D
ϩ4.9 (c 0.12 in CHCl3); νmax/cmϪ1 (CHCl3) 2955, 2855, 1732
and 1668; δH(CDCl3) 5.76 (1H, s), 3.69 (3H, s), 2.73 (1H, dq, J
8.5, 6.8), 1.22 (3H, d, J 6.8) and 1.04 (3H, d, J 5.9); HREIMS
m/z (rel. int %): 250.1570 [Mϩ, C15H22O3 ∆ = Ϫ0.1 mmu] (55),
191 (100) and 163 (95).
Methylation of 17
To a mixture of 17 (6.6 g) and Diazald (30.0 g) in MeOH (153
ml), was added dropwise a methanolic solution of KOH (7.8
g/241 ml) with ice–salt cooling. The reaction mixture was
stirred overnight at room temperature to afford a white preci-
pitate which was filtered off. The filtrate was concentrated to
give 14 as a pale yellow liquid (3.3 g). This was purified by
column chromatography (hexane–EtOAc, 16:9); δH(CDCl3)
5.76 (1H, s), 3.54 (3H, s), 2.82 (1H, dq, J 11.2, 6.8), 2.37 (1H,
ddd, J 15.9, 5.1, 4.0), 1.20 (3H, d, J 6.8) and 1.05 (3H, d, J 5.9);
HREIMS m/z (rel. int %) 250.1570 [Mϩ, ∆ = Ϫ0.1 mmu for
C15H22O3] (20), 191 (10), 163 (70) and 162 (100).
NaBH4 reduction of 14
A solution of 14 (575 mg) in pyridine (9 ml) was added to a
suspension of NaBH4 (504 mg) in pyridine (7 ml). The reaction
mixture was stirred at room temperature for 6 h, after which it
was acidified with aq. HCl (3 ) and then extracted with Et2O
(60 ml). The extract was washed with dil. aq. HCl and water,
dried (MgSO4), and concentrated to give 18 as an oily crude
product (200 mg). This was purified by column chrom-
atography (hexane–EtOAc, 7:3 v/v) to afford an oil, [α]D Ϫ63.9
(c 1.1 in CHCl3); νmax/cmϪ1 (CHCl3) 3503, 2924, 2854 and 1720;
δH(CDCl3) 4.19 (1H, br s, OH), 3.85 (1H, ddd, J 11.7, 6.0, 5.8,
H-4), 2.68 (1H, dd, J 6.3, 6.0, H-5), 2.28 (1H, dq, J 11.5, 7.2,
H-11), 2.08 (1H, m, H-3β), 1.61 (1H, ddd, J 13.4, 6.8, 3.6, H-
9β), 1.25 (1H, dddd, J 12.7, 12.7, 11.7, 3.6, H-3α), 1.09 (3H, d, J
7.2, H-13), 0.87 (3H, d, J 6.3, H-14), 0.83 (1H, dddd, J 12.8,
12.8, 12.8, 3.0, H-2β) and 0.55 (1H, dddd, J 11.4, 11.4, 11.4, 2.5,
H-1); δC(CDCl3) 225.3 (C, C-12), 70.2 (CH, C-4), 54.8 (CH,
C-5), 45.0 (CH, C-6), 44.5 (CH, C-11), 43.3 (CH, C-1), 42.8
(CH, C-7), 35.3 (CH, C-10), 34.5 (CH2, C-3), 31.7 (CH2, C-9),
26.0 (CH2, C-2), 25.0 (CH2, C-8), 19.3 (CH3, C-14) and 15.1
(CH3, C-13); HREIMS m/z (rel. int %) 222.1617 [Mϩ, ∆ = 0.2
mmu for C14H22O2] (28), 204 (60), 173 (40), 163 (50), 149 (100)
and 148 (98).
Robinson annulation of 12
To a solution of the diketone 12 (50 mg) in EtOH (4 ml) was
added Ba(OH)2ؒ8H2O. The mixture was stirred at room temper-
ature for 1.5 h and then neutralized with 10% aq. HCl and
concentrated under reduced pressure. The residue was extracted
with CH2Cl2 and the extract was washed, dried and evaporated
under reduced pressure. The crude product was purified by
HPLC (23% EtOAc–hexane) to give 21 (20 mg) (Rt 13.34 min)
and 20/22 as a mixture (5 mg) (Rt 13.07 min).
Compound 21. Oil, [α]D Ϫ205.4 (c 0.28 in CHCl3); νmax/cmϪ1
(CHCl3) 3011, 2961, 2930 and 2874; δH(CDCl3) 5.76 (1H, d,
J 2.0, H-5), 2.76 (1H, ddd, J 13.7, 3.7, 3.7, H-8), 2.42 (1H, m,
H-3), 1.83 (1H, ddd, J 12.8, 7.5, 3.7, H-9), 1.77 (3H, s, H-12),
1.76 (3H, s, H-13), 1.25 (1H, dddd, J 13.0, 13.0, 13.0, 3.9, H-9)
and 1.03 (3H, d, J 6.4, H-14); δC(CDCl3) 200.3 (C, C-4), 164.0
(C, C-6), 133.1 (C, C-7), 129.6 (C, C-11), 126.5 (CH, C-5), 45.4
(CH, C-1), 39.6 (CH, C-10), 36.4 (CH2, C-3), 34.7 (CH2, C-9),
J. Chem. Soc., Perkin Trans. 1, 1997
3441

View Article Online
30.1 (CH2, C-8), 27.0 (CH2, C-2), 22.8 (CH3, C-13), 20.9 (CH3,
C-12) and 20.1 (CH3, C-14); HREIMS m/z (rel. int %) 204.1512
(Mϩ, ∆ = 0.2 for C14H20O) (10), 176 (10), 161 (100), 134 (55) and
122 (60).
Compound 20/22 (mixture of isomers). Oil, HREIMS m/z
(rel. int %) 204.1511 [Mϩ, ∆ = 0.3 mmu for C14H20O] (100), 189
(44), 161 (47), 147 (55), 133 (74), 105 (63), 91 (72) and 77 (37).
Compound 20 δH(CDCl3) 5.77 (1H, s), 4.98 (1H, d, J 1.7), 4.79
(1H, s), 2.79 (1H, dd, J 12.5, 2.5), 1.71 (3H, s), 1.05 (3H, d,
J 6.3). Compound 22 δH(CDCl3) 5.91 (1H, s), 4.96 (1H, s), 4.74
(1H, s), 3.02 (1H, m), 1.71 (3H, s) and 1.01 (3H, d, J 6.3).
aq. HCl to bring it to pH ca. 3. After this the reaction mixture
was stirred at 0 ЊC for 15 min, and then extracted with EtOAc.
The extract was washed with water to neutrality, dried (MgSO4)
and evaporated to afford 23 as a pale yellow liquid (1.01 g). This
was purified by means of column chromatography (hexane–
EtOAc, 7:3, v/v). Compound 7 was also obtained as a minor
product from the column. Compound 23 δH(CDCl3) 7.31 (5H,
m), 4.49 (1H, d, J 12.8), 4.45 (1H, d, J 12.8), 3.45 (1H, dd, J 9.1,
4.9), 3.37 (1H, dd, J 9.1, 5.7), 2.10 (3H, s), 1.05 (3H, d, J 5.9)
and 1.00 (3H, d, J 6.8). Compound 7 δH(CDCl3) 7.32 (5H, m),
4.48 (2H, s), 3.30 (2H, m), 2.10 (3H, s), 1.06 (3H, d, J 5.8) and
0.87 (3H, d, J 7.0).
Preparation of the diketone cyclohexanones 7 and 23 from
isopulegol
Robinson annulation of 23
Compound 23 (0.27 g) was added to a suspension of
Ba(OH)2ؒ8H2O (0.25 g) in EtOH (8 ml). After being stirred at
room temperature for 2.5 h, the reaction mixture was acidified
(10% aq. HCl) and evaporated under reduced pressure. The
residue was taken up in CH2Cl2 and the solution washed with
water, dried (MgSO4) and evaporated to afford a colourless
liquid (0.07 g). A mixture of the colourless liquid (0.07 g) and
oxalic acid (0.08 g) in EtOH (3.2 ml) was refluxed for 2.5 h and
then neutralized (10% aq. NaHCO3) and concentrated under
reduced pressure. This residue was extracted with CH2Cl2. The
extract was washed with water and evaporated to afford 24 as a
pale yellow oil (0.04 g). This was purified by means of HPLC
[hexane–EtOAc (3:1, v/v)]; δH(CDCl3) 7.32 (5H, m), 5.88 (1H,
s), 4.51 (1H, d, J 12.1), 4.45 (1H, d, J 12.1), 3.48 (1H, dd, J 9.1,
3.3), 3.35 (1H, dd, J 9.1, 6.0), 1.07 (3H, d, J 6.4) and 1.02 (3H,
d, J 6.4); HREIMS m/z (rel. int %) 312.2088 [Mϩ, ∆ = 0.1 mmu
for C21H28O2] (46), 254 (5), 221 (30), 191 (100), 164 (70), 119 (9)
and 91 (33).
To a solution of (Ϫ)-isopulegol (1.08 g) in THF (10 ml), was
added dropwise a solution of BH3ؒTHF complex in THF soln.
(1 ; 7 ml) with ice-cooling. After being stirred for 1 h at room
temperature, the mixture was treated with THF–water (1:1; 4
ml), followed by a mixture of aq. KOH (10%; 5 ml) and aq.
H2O2 (50%; 2 ml). After this, the reaction mixture was stirred
for 5 min, and then filtered and evaporated. The residue was
washed with water and taken up into Et2O and the ethereal
solution was dried (MgSO4) and evaporated to give a white
solid. This was purified by sublimation (ca. 60 ЊC, 10 mmHg) to
yield 9-hydroxyisopulegol (0.85 g) as crystals, mp 88–89 ЊC;
δH(CDCl3) 4.58 (2H, br s), 3.63 (1H, dd, J 11.8, 5.2), 3.57 (1H,
dd, J 11.5, 3.4), 3.42 (1H, ddd, J 10.5, 10.5, 3.9), 1.95 (1H, br d,
J 17), 1.21 (1H, ddd, J 12.7, 12.7, 3.5), 0.95 (3H, d, J 7.3) and
0.91 (3H, d, J 6.3); δC(CDCl3) 69.8 (CH), 66.8 (CH2), 48.6
(CH), 44.4 (CH2), 38.6 (CH), 34.6 (CH2), 31.4 (CH), 29.6 (CH2),
22.1 (CH3) and 11.9 (CH3); HREIMS m/z (rel. int %) 172.1458
[Mϩ, ∆ = 0.5 mmu for C10H20O2] (5), 154 (10), 139 (17), 124 (42),
112 (49), 95 (47) and 81 (100). A solution of 9-hydroxyiso-
pulegol (2.74 g) in THF–DMF (4:1; 35 ml) was added slowly
to a suspension of NaH (60% suspension in paraffin) (2.45 g) in
THF–DMF (70 ml). The reaction mixture was stirred for 1 h
after which it was treated with a solution of benzyl chloride
(3.02 g) in THF–DMF (10 ml); stirring was then continued for
3 h at ca. 10 ЊC. After this, the reaction mixture was filtered,
added to ice–water (100 ml) and extracted with Et2O (2 × 100
ml). The combined extracts were evaporated to give 9-benzyl-
oxylisopulegol (2.76 g) as a yellow liquid; δH(CDCl3) 7.33 (5H,
m), 4.55 (1H, d, J 12), 4.48 (1H, d, J 12), 3.64 (1H, br s), 3.50
(1H, dd, J 9.1, 6.2), 3.40 (1H, dd, J 9.1, 3.7), 1.13 (1H, dddd, J
12.7, 12.7, 12.7, 3.4), 0.96 (3H, d, J 7.1) and 0.91 (3H, d, J 6.6);
δC(CDCl3) 137.7 (C), 128.3 (CH × 2), 127.6 (CH × 3), 74.3
(CH2), 73.3 (CH2), 70.4 (CH), 48.9 (CH), 44.9 (CH2), 35.4
(CH), 34.6 (CH2), 31.4 (CH), 27.8 (CH2), 22.1 (CH3) and 13.5
(CH3). A solution of 9-benzyloxylisopulegol (1.34 g) in CH2Cl2
(5 ml) was added to a suspension of pyridinium chloro-
chromate (4.83 g) in CH2Cl2 (20 ml) and the mixture was stirred
for 24 h. It was filtered, washed with water (30 ml) and 10% aq.
Na2SO3 (30 ml), dried (MgSO4) and evaporated to yield
9-benzyloxylisopulegone (1.13 g) as a yellow liquid; [α]D Ϫ 8.3Њ
(c 1.4 in CHCl3); νmax/cmϪ1 (CHCl3) 2959, 2930, 2872 and 1705;
δH(CDCl3) 7.31 (5H, m), 4.48 (1H, d, J 12.3), 4.46 (1H, d,
J 12.3), 3.47 (1H, dd, J 9.0, 5.4), 3.38 (1H, dd, J 9.0, 6.1), 1.01
(3H, d, J 6.8) and 0.99 (3H, d, J 6.1); δC(CDCl3) 212.0 (C),
138.7 (C), 128.2 (CH), 127.5 (CH), 73.0 (CH2), 72.9 (CH2), 52.2
(CH), 50.9 (CH2), 35.3 (CH), 34.1 (CH2), 32.6 (CH), 29.4
(CH2), 22.3 (CH3) and 15.3 (CH3); HREIMS m/z (rel. int %)
260.1774 [Mϩ, ∆ = 0.2 mmu for C17H24O2] (6), 202 (35),
169 (31), 153 (21), 151 (17) and 91 (100). A solution of
9-benzyloxyisopulegone (2.16 g) in THF (3 ml) was added
dropwise to a solution of LDA in THF at Ϫ78 ЊC [prepared
from BuLi (1.6  in hexane; 7.1 ml), diisopropylamine (1.6 ml)
and THF (3 ml)]. After being stirred for 30 min, the mixture
was treated with a solution of 3-trimethylsilylbut-3-en-2-one14
(2.26 g) in THF (3 ml), added dropwise, and then stirred for 1 h
at Ϫ78 ЊC and then for 2 h at 0 ЊC. It was then treated with 10%
On storage for 1 year 24 was converted into 25, which was
purified by HPLC as above; δH(CDCl3) 7.32 (5H, m), 5.82 (1H,
d, J 1.3), 4.39 (2H, s), 3.24 (1H, dd, J 9.2, 4.3), 3.11 (1H, dd,
J 9.2, 6.5), 1.05 (3H, d, J 6.6) and 1.01 (3H, d, J 6.1); HREIMS
m/z (rel. int %) 312.2071 [Mϩ, ∆ = 1.8 mmu for C21H28O2] (7),
221 (26), 209 (17), 182 (42), 163 (29),124 (49) and 91 (100).
Acknowledgements
We thank the CRCG for funding this research and The
University of Hong Kong for provision of a postgraduate
studentship to Mr Hui and Mr Ngo.
References
1 J.-M. Liu, M.-Y. Ni, Y.-F. Fan, Y.-Y. Tu, Z.-H. Wu, Y.-L. Wu and
W.-S. Zhou, Acta Chimica Sinica, 1979, 37, 129.
2 Y.-L. Wu and Y. Li, Med. Chem. Res., 1995, 5, 569.
3 G. D. Brown, Phytochemistry, 1994, 36, 637.
4 R. K. Haynes and S. C. Vonwiller, Acc. Chem. Res., 1997, 30, 73.
5 L. N. Misra, A. Ahmad and R. S. Thakur, J. Nat. Prod., 1993, 56,
215.
6 D. Jeremic, A. Jokic, A. Behbud and M. Stefanovic, Tetrahedron
Lett., 1973, 3039.
7 Y.-Z. Gai, Y.-P. Zheng and L.-Q. Li, Huaxue Xuebao, 1984, 42, 259.
8 Y. Li, P. Yu, Y. Chen, J. Zhang and Y. Wu, Kexue Tongbao, 1986, 31,
1038.
9 J. F. Mi and C.-Y. Shen, Yaoxue Xuebao, 1984, 19, 184.
10 X.-X. Zu, Z.-H. Wu, J.-M. Shen, C.-H. Cheng, Y.-L. Wu and
W.-S. Zhou, Huaxue Xuebao, 1984, 42, 333.
11 L. Li, Y. Zheng and Y. Gai, Kexue Tongbao, 1984, 29, 396.
12 Y. Li, Y.-X. Chen, P.-L. Yu and R.-Y. Ji, Acta Pharmaceutica Sinica,
1986, 21, 899.
13 E. L. Eliel and S. H. Wilen, Stereochemistry of organic compounds,
VCH, Weinheim, 1994, p. 737.
14 X.-X. Xu, J. Zhu, D.-Z. Huang and W.-S. Zhou, Tetrahedron, 1986,
42, 819.
Paper 7/02714A
Received 21st April 1997
Accepted 17th July 1997
3442
J. Chem. Soc., Perkin Trans. 1, 1997