The mechanism of the spontaneous autoxidation of dihydroartemisinic acid

Article abstract of DOI:10.1016/S0040-4020(01)01193-0

Dihydroartemisinic acid undergoes slow spontaneous autoxidation to artemisinin and other natural products, which have been reported from the medicinal plant Artemisia annua. The mechanism of this complex transformation is shown to involve four steps: (i) initial reaction of the Δ^4.5-double bond of dihydroartemisinic acid with molecular oxygen, (ii) Hock cleavage of the resulting tertiary allylic hydroperoxide; (iii) oxygenation of the enol product from Hock cleavage; and (iv) cyclization of the resulting vicinal hydroperoxyl-aldehyde to the 1,2,4-trioxane system of artemisinin.

Full text of DOI:10.1016/S0040-4020(01)01193-0

TETRAHEDRON
Pergamon
Tetrahedron 58 -2002) 897±908
The mechanism of the spontaneous autoxidation of
dihydroartemisinic acid
Lai-King Sy and Geoffrey D. Brownp
Department of Chemistry and Open laboratory of Chemical Biology of the Institute of Molecular Technology of Drug Discovery and
Synthesis, Area of Excellence Scheme of University Grants Committee #Hong Kong), The University of Hong Kong, Pokfulam Road,
Hong Kong, People's Republic of China
Received 23 July 2001; revised 6November 2001; accepted 29 November 2001
AbstractÐDihydroartemisinic acid undergoes slow spontaneous autoxidation to artemisinin and other natural products, which have been
reported from the medicinal plant Artemisia annua. The mechanism of this complex transformation is shown to involve four steps: -i) initial
reaction of the D4,5-double bond of dihydroartemisinic acid with molecular oxygen, -ii) Hock cleavage of the resulting tertiary allylic
hydroperoxide; -iii) oxygenation of the enol product from Hock cleavage; and -iv) cyclization of the resulting vicinal hydroperoxyl-aldehyde
to the 1,2,4-trioxane system of artemisinin. q 2002 Elsevier Science Ltd. All rights reserved.
1. Introduction
2 to yield the tertiary allylic hydroperoxide 3. Compound 3
then undergoes Hock cleavage, leading to the enolic inter-
mediate 4, which reacts with molecular oxygen to form the
vicinal hydroperoxyl-aldehyde 5. The reaction terminates
when compound 5 undergoes cyclization to artemisinin
-Scheme 1). This proposal was based on previous experi-
mental evidence which had shown that essentially the same
mechanism was involved in the formation of peroxo-
fabianane, a very close analog of artemisinin, from the
natural product 4-amorphen-11-ol found in Fabiana
imbricata.7
It has been known for some time that the 1,2,4-trioxane
ring1,2 in the natural product artemisinin -1) from the
medicinal plant Artemisia annua is responsible for the
biological activity of this increasingly important anti-
malarial drug. It has been shown both by ourselves3,4 and
by others5 that the natural product dihydroartemisinic acid
-2) from this same species can undergo spontaneous
autoxidation either as a solid or in organic solution yielding
artemisinin -1) as well as other natural products which have
been reported from A. annua -compounds 6±8 in Fig. 1).
These reactions are slow but seem to proceed in the absence
of photosensitizer, which would be required by classical
theory for the generation of singlet oxygen -1O2).
It is interesting to note that, without exception, all chemical
syntheses of the 1,2,4-trioxane system in 1 have also
involved ring-closure from
a hydroperoxyl-aldehyde
precursor identical or equivalent in structure to that of
compound 5 shown in Scheme 1, which is in turn always
derived from oxygenation of an enol identical or equivalent
in structure to that of compound 4.8±16 In some cases,
Recently, we have proposed a mechanism6 for the spon-
taneous formation of the 1,2,4-trioxane ring of 1 in vitro
which begins with the oxygenation of the double bond in
Figure 1. Products of the spontaneous autoxidation of dihydroartemisinic acid -2) which have been reported in the literature.
Keywords: terpenes and terpenoids; autoxidation; enols and derivatives; mechanisms.
Corresponding author. Tel.: 1852-2859-7915; fax: 1852-2857-1586; e-mail: gdbrown@hkucc.hku.hk
p
0040±4020/02/$ - see front matter q 2002 Elsevier Science Ltd. All rights reserved.
PII: S0040-4020-01)01193-0

898
L.-K. Sy, G. D. Brown / Tetrahedron 58 #2002) 897±908
Scheme 1. Proposed mechanism for the spontaneous conversion of dihydroartemisinic acid -2) to artemisinin -1) via -i) spontaneous autoxidation to the tertiary
allylic hydroperoxide 3; -ii) Hock cleavage of 3 to an enol 4; -iii) autoxidation of enol 4 to a hydroperoxyl-aldehyde 5; and -iv) closure of the 1,2,4-trioxane
ring.
1
dihydroartemisinic acid -2) itself -or its derivatives) has
been converted directly into 1 by photo-oxygenation
-presumably via the presumed intermediate 4).17±19 Haynes
and Vonwiller20±22 have undertaken studies of the
mechanism for the sensitized photo-oxygenation and subse-
quent CuII-catalyzed rearrangements of the tertiary allylic
hydroperoxide which is derived from arteannuic acid, the
11,13-dehydro analog of 2 -arteannuic acid is also a natural
product from A. annua). Their results for the chemically
induced transformations of arteannuic acid favor the
mechanism which we propose for the spontaneous
autoxidation of dihydroartemisinic acid -2) shown in
Scheme 1. However, Roth and Acton23±25 have also
independently studied this same chemically induced trans-
formation of 2, and although they also found that a tertiary
allylic hydroperoxide was involved, they have arrived at a
quite different mechanism for carbon±carbon cleavage to
that proposed by Haynes and Vonwiller.
parison with H NMR data reported in the literature, as
belonging to the following natural products which have
been isolated previously from A. annua: artemisinin -1),28
dihydroartemisinic acid tertiary hydroperoxide -3),3,26
dihydro-epi-deoxyarteannuin B -6)3,29 and arteannuin H
-7).3,6 In particular, it was apparent that peaks belonging
to compound 3 appeared early on in the spontaneous trans-
formation of 2 in CDCl3 solution and then declined in
intensity, whilst peaks corresponding to compound 1 con-
tinued to grow steadily with time. Such behavior is entirely
consistent with the role suggested in Scheme 1 for 3 as an
intermediate derived from the spontaneous autoxidation of 2
en route to 1.
From other precedents in the literature,3,4,6 we propose that
compound 6 is formed by SN20 attack of the carboxylic acid
group at the allylic hydroperoxide resulting in elimination
of hydrogen peroxide from 3. The natural product 7 is
almost certainly derived from an alternative secondary
allylic hydroperoxide product -compound 9 in Scheme 2),
which is also produced by spontaneous oxygenation of the
D4,5-double bond in 2, but which does not accumulate itself
under the conditions of reaction, and is instead trapped by
the 12-carboxylic acid group -Scheme 2).6 Good evidence
for this came from the detection of very small peaks in the
1H NMR spectrum of the mixture corresponding to trace
amounts of the primary allylic hydroperoxide 11, which
has been reported previously as a product from the 3,2-
allylic rearrangement of 9 that competes with the lactoniza-
tion reaction involved in the formation of 7 in CDCl3
solution.6 Several peaks also appeared in the extreme down-
®eld region of the 1H NMR spectrum -9±12 ppm) but
could not be assigned by comparison with the literature.
Some of these peaks appeared early on in the experiment,
and disappeared towards the end of the experiment, when
the conversion of 2 into 1 was complete, and they may
represent intermediates in the cyclization of 5 to 1 -see
later).
In this study we set out to determine whether the mechanism
shown in Scheme 1 for the remarkable spontaneous trans-
formation of dihydroartemisinic acid -2) to artemisinin -1) is
correct. The answer to this question is important since it
may have a bearing on the biogenesis of compound 1 in
vivo. In this regard, we note that Pras et al.26 have recently
isolated the tertiary allylic hydroperoxide of dihydro-
artemisinic acid -compound 3 in Scheme 1) as a natural
product from A. annua. These authors suggested that this
compound may be formed in vivo by oxidation of dihydro-
artemisinic acid and that it is then transformed into
artemisinin, although they presented no direct experimental
evidence to support this. The same authors have even
suggested that dihydroartemisinic acid might be a natural
anti-oxidant in this species.27
2. Results and discussion
2.1. The ®rst spontaneous autoxidation of
dihydroartemisinic acid #2)
It is interesting to note that the spontaneous formation of all
of compounds 1, 6 and 7 was signi®cantly accelerated in
samples of 2 contaminated by stearic acid -which is dif®cult
to separate from dihydroartemisinic acid from the natural
sourceÐsee Ref. 3). The rates of transformations were in
general slower for more concentrated solutions of 2, which
may be due to the rate of dissolution of oxygen in the NMR
tube becoming a limiting factor.30 In support of this, we
found that signals due to compound 2 generally disappeared
completely from 1H NMR spectra within a few weeks when
A CDCl3 solution of dihydroartemisinic acid -2) -isolated
from A. annua) was kept in an NMR tube under laboratory
conditions for several weeks and H NMR spectra were
1
recorded at intervals of every few days. Several new
peaks began to appear within a few days in these H NMR
spectra and after ®ve weeks the starting material accounted
for only around half of the mixture. Although these spectra
were complex, some peaks could be identi®ed by com-
1

L.-K. Sy, G. D. Brown / Tetrahedron 58 #2002) 897±908
899
Scheme 2. Products 3, 9 and 10 from photo-oxygenation of compound 2 and further transformations of the secondary allylic hydroperoxide 9 in CDCl3 which
are reported in the literature.6
CDCl3 solutions were left in open vessels rather than NMR
tubes. No transformations were observed when NMR solu-
tions of 2 were kept in the dark indicating that the initial
conversion of 1 into 3 requires light and therefore almost
certainly involves singlet oxygen -1O2).
experiment, reached a maximum after 3 days, and then
gradually declined in intensity -Fig. 2). We suspect that
this resonance corresponds to the aldehyde proton -H-5)
of the vicinal hydroperoxyl-aldehyde intermediate 5
shown in Scheme 1 -an aliphatic resonance at dH 2.16
-3H, s) ppm for the H-15 methyl group was also clearly
seen and followed the same pattern of change in intensity
with time). However, this assignment must remain tentative
since we were unable to isolate this compound by chromato-
graphy. Additional strongly down®eld resonances -dH 10.19
-1H, dd, Jˆ1.5, 1.5 Hz), 9.77 -s) and 9.75 -s); not shown in
Fig. 2) also reached a maximum intensity after 3±5 days
-later than the maximum for the resonance at dH 9.93 ppm,
which was tentatively assigned to 5) when the conversion of
3 was ca. 50% complete and then declined. These reso-
nances may correspond to more advanced intermediates in
the cyclization of 5 to 1, although once again this could not
be con®rmed because none of these components could be
In con®rmation of the suggested intermediacy of 3 in the
spontaneous transformations of 2, it was found that when a
pure sample of this tertiary allylic hydroperoxide -obtained
by HPLC puri®cation of the crude product from photo-
oxygenation of 2Ðsee Scheme 2 and Ref. 6) was left in a
solution of CDCl3 for several days, the formation of
1
compounds 1 and 6 was again noted in H NMR spectra
-Fig. 2) and the starting material disappeared within 1±2
weeks. Several other resonances were observed in these
spectra in addition to those associated with compounds 1,
3 and 6. In particular, a very obvious resonance appeared at
dH 9.93 -d, Jˆ2.5 Hz) within a few hours of the start of the
Figure 2. Spontaneous transformations of the tertiary allylic hydroperoxide of dihydroartemisinic acid -3) into compounds 1 and 6 in CDCl3 solution, via
vicinal hydroperoxyl aldehyde 5 -tentative identi®cation only). The percentage of each of these components in the mixture was estimated from integration of
clearly resolved characteristic resonances for each compound in the alkene region of 1H NMR spectra acquired at each time point: compound 1 -dH 5.86, s,
H-5); compound 3 -dH 5.25, s, H-5); compound 6 -dH 5.64, s, H-5); compound 5 -dH 9.93, d, Jˆ2.5 Hz, H-5)Ðtentative identi®cation only.

900
L.-K. Sy, G. D. Brown / Tetrahedron 58 #2002) 897±908
Scheme 3. Proposed transformation pathways -i)±-vii) for the allylic hydroperoxide 3 in CDCl3.
isolated by chromatography. The presence or absence of
light had no effect on these further transformations of 3,
and it seems that the second autoxidation reaction required
transformations of this allylic hydroperoxide were rapid and
simpli®ed; after the starting material had disappeared, only
two products were isolated, compound 1 -43%) and
compound 6 -5%). Such facile conversion of 3 into 1 in
the presence of strong acid is also in agreement with the
®ndings of others.25
3
for the spontaneous formation of 1 therefore involves O2
rather than O2 -see also Section 2.3).
1
Clearly both the spontaneous autoxidation of 2 to 3 and the
subsequent rearrangement reactions of 3 in organic solution
-as well as those of the alternative secondary allylic hydro-
peroxide autoxidation product, compound 9) are complex
processes, producing a variety of products in addition to
compounds 1 and 6. When tri¯uoroacetic acid -TFA) was
added to a petroleum ether solution of 3 -cf. Ref. 7), the
2.2. Spontaneous Hock cleavage of tertiary allylic
hydroperoxide #3)
In order to further investigate the mechanism for the
carbon±carbon cleavage reaction at C-4/C-5 which is
clearly required for the spontaneous transformation of

L.-K. Sy, G. D. Brown / Tetrahedron 58 #2002) 897±908
Table 1. 13C NMR data for novel compounds 4, 12 and 15±25 in CDCl3
901
Position
4a
12
15
16
17
18
19
20
21
22
23
24
25
Position
17
20
1
2
3
4
36.8
28.1
40.8
46.7
20.1
26.1
50.0
21.6
26.7 106.2
47.3
26.3
52.0
23.8
33.0
42.3
26.7
38.1
47.3
20.9
41.2
38.5
28.0
41.2
44.5
22.0
33.9
46.1
24.4
41.0
45.2
23.0
41.0
43.8
23.9
38.6
46.1 10
24.7 20
41.3 300
44.4
25.9
32.0
47.1
21.6
43.5
214.6142.3 123.5 155.9 109.6208.3 208.3 210.3 109.2 208.6209.1 210.0 209.0
140.2 120.5 134.8 136.7 131.2 205.9 131.9 133.3
±86
40.3
23.1
22.8
30.630.7
41.9
4
80.7 208.8
5
611.17.9
7
8
9
10
11
12
13
14
15
5-OMe
99.6 206.1 104.3 181.4 184.5 50
82.4
42.3
123.0
137.0
43.3
32.1
35.2
39.8
42.1
183.4 172.2
9.3
20.0
97.1
80.1
44.4
23.2
34.5
35.1
32.7
b
124.7 130.1
45.9
33.2
56.1 124.6 119.4
49.9
43.1
43.1
40.5
48.1
41.4
45.5 60
44.0 700
49.8
22.0
35.3
49.7
21.5
43.7
32.9
41.0
23.5
35.5
41.7
41.2
42.7
41.5
33.4
41.4
28.624.0
22.9
30.3
23.5
35.3
27.3
33.4
23.1
32.8
31.1
33.5
23.7
8
0
34.634.8 .136 34.7
37.4 39.3
38.1
35.5
32.7
34.9
42.9
34.7
40.8
35.69
43.4
0
31.1
33.4
31.2 10
39.0
41.1
41.8
36.6
43.5 1100
16.3 1300
184.7 180.6181.2 179.2 183.0 177.4 171.6179.0 171.9 181.0 173.8 181.4 181.612
17.0
20.0
30.624.1
±
13.2
19.9
12.7
18.5
16.3
18.8
9.7
19.4
14.4
19.620.0
30.1
±
11.6
17.2
12.6
18.620.2
29.7
15.1
20.2
29.4
±
13.2
19.7
29.9
56
14.5
20.3 14
30.0 15
12.7
21.2
19.9
30.623.9
±
0
20.9
20.7
23.9
30.0
24.629.7
±
±
±
±
±
±
.4
±
±
a
d values recorded at 233 K in CDCl3/TFA solution.
Not resolved.
b
tertiary allylic hydroperoxide 3 into compound 1 in organic
solution, we attempted to prevent the occurrence of the next
step, the second autoxidation reaction shown in Scheme 1
-which forms compound 5), by preparing a CDCl3 solution
of 3 under an atmosphere of nitrogen. Changes observed
with time in the 1H NMR spectra obtained from this experi-
ment bore some parallels with the results summarized in
Fig. 2 -e.g. the formation of compound 6 was once again
clearly evident), although rather unexpectedly there was
now a considerable degree of additional complexity asso-
ciated with the exclusion of molecular oxygen from the
system, even though a single hydroperoxide was employed
as the starting material. When all the resonances corre-
sponding to the starting material -3) had disappeared from
these 1H NMR spectra, the mixture was separated by HPLC
and 12 products could then be characterized by 2D NMR
-compounds 1, 6 and 12±21; Scheme 3 and Tables 1 and 2).
These compounds have been grouped together in Scheme 3
according to the mechanisms which, we believe, are
involved in their formation. We have identi®ed seven
distinct pathways for the transformations of the tertiary
allylic hydroperoxide 3 in this Scheme 3: -i) SN20 attack
of the 12-carboxylic acid group, resulting in the elimination
of hydrogen peroxide -forming 6 and 12); -ii) E1 elimination
of the hydroperoxide group and H-7 -forming 13);
-iii) rearrangement to a b-epoxy alcohol -forming 14);
-iv) 3,2-allylic rearrangement -forming 15); -v) Hock
cleavage and tautomerization of the resultant enol -forming
16±19); -vi) Hock cleavage and radical dimerization of the
resultant enol -forming 20); and -vii) oxygenation of the
enol from Hock cleavage -forming 1 and 21).
Although, the seventh pathway is clearly the dominant one
for the transformation of compound 3 in the presence of
oxygen -resulting predominantly in conversion to 1, as
discussed in Section 2.1) it is much less signi®cant in this
experimentÐin which compound 18, believed to be formed
by Hock cleavage and tautomerization of the resultant enol
4 to an aldehyde in the absence of molecular oxygen -path-
way v) in Scheme 3), was found as the predominant product.
Given that great care was taken to exclude atmospheric
oxygen from the CDCl3 solution it was a little surprising
1
Table 2. H NMR data for novel compounds 4, 12 and 15±25 in CDCl3
Position
4a
12
15
16
17
18
19
20
21
22
23
24
25
Position
17
20
0
1
2.21
1.61
1.61
2.57
2.68
6.31
1.53
2.03
1.80
2.03
1.93
5.53
1.75
1.92
1.72
1.81
2.35
5.79
1.79
2.57b
2.04b
4.91
±
6.07
± 2.36 ±
2.04
1.85
1.22
1.58
1.4861.6 2.34
1.562.061.78
1.28
1.92
1.561.28
2.60
2.48
1.25
2.40
1.17
1.97
2.02
1.47
1.82
1.21
1.15
2.03
1.58
1
2
1.32
1.57
1.17
2.05
1.53
1.24
5.47
±
2.74
1.41
1.59
1.75
1.24
1.05
2.58
1.23
0.92
1.30
1.56
1.82
2.10
2.52
2.79
5.69
±
1.92
1.25
1.89
1.88
1.13
1.64
2.74
1.12
1.07
2.07
2a
2b
3a
3b
5
1.74b
1.48b
1.64b
1.72b
5.88
a0
1.22
2.32
2.45
9.58
±
1.49
2.45
2.63
6.08
2
b0
1.63
1.78
5.70
2.66
2.66
9.97
2.75
2.33
5.02
2.46
2.33
±
2.73
2.38
±
3a0
3b0
500
6.19
6±
7
±
±
±
±
2.452.212.162.88
6
2.15
1.77b
1.34b
1.28b
1.83b
1.74
2.62
1.12
0.93
2.32
±
2.19
1.68
1.44
1.161.19
1.82
1.55
2.12
1.80
1.61
2.64
1.63
1.44
1.93
1.87
1.47
2.16
1.83
1.10
2.23
1.79
1.41
2.00
1.92
1.01
1.81
1.67
1.30
1.73
1.67
1.03
2.42
1.66
1.38
1.69
1.97
1.54
70
8a
8b
9a
9b
10
11
13
14
15
5-OMe
8a0
8b0
a0
1.17
1.19
1.11
1.19
1.24
1.09
1.81
1.19
1.90
1.03
1.78
1.19
1.74
0.97
1.78
9
9
1.90
2.05
2.63
1.32
1.77
1.57
2.58
1.261.19
0.89
1.43
±
1.78
1.45
2.67
1.81
1.27
2.48
1.861.91
1.24
2.95
b0
1.73
2.75
1.28
3.18
1.23
0.98
2.15
±
1.65
2.41
1.21
0.97
2.14
±
1.34
2.61
1.19
0.92
2.14
1.19
2.51
1.29
0.90 14
2.1615
±
1.80
2.49
100
1100
2.74
1.35
1.18
0.94
2.14
1.22
0.99
1.15
0.860.94
1.20
13
0
0
0.960.94
1.70 01.6 1.77
±
0.87
2.162.18
±
1.53
±
±
±
±
3.46±
a
d values recorded at 233 K in CDCl3/TFA solution.
a and b assignments uncertain -due to extensive chemical exchange in NOESY spectra, which are dominated by EXSY-type peaks in the case of 4).
b

902
L.-K. Sy, G. D. Brown / Tetrahedron 58 #2002) 897±908
that compounds such as 1 and 21 -from pathway vii) in
Scheme 3) were isolated at all: we believe that the small
amounts of these compounds obtained are the result of the
occurrence of oxidation by hydrogen peroxide, which is
formed in solution as a by-product from the SN20 and E1
pathways, rather than by molecular oxygen.
alcohol, has been well documented in the oxidation
chemistry of fatty acids,36 and we have also recently isolated
the b-epoxy alcohol, which is now proposed to be an inter-
mediate in pathway -iii) of Scheme 3, as a natural product
from this species -unpublished results). In further support of
this mechanism, we note that a more abundant natural
product from A. annua, arteannuin B -the 11,13-dehydro
analog of compound 14), has previously been obtained,
together with epi-deoxyarteannuin B -the 11,13-dehydro
analog of 6), from photo-oxygenation of arteannuic acid
-the 11,13-dehydro analog of 2).37
Although the complex nature of the varying transformations
undergone by 3 in this experiment were unexpected -and
undesired), they are of considerable interest because several
of the reaction products have themselves been reported as
natural products from A. annua -note that 3 is also itself
claimed as a natural product).26,27 These include the
lactones dihydro-epi-deoxyarteannuin B -6)3,29 and dihydro-
deoxyarteannuin B -12).31 6,7-Dehydro-dihydroartemisinic
acid -13) is the 11,13-dihydro analog of the natural
product 6,7-dehydroartemisinic acid from this species.32,33
We have recently provided experimental evidence from the
4-hydroxyl analog of compound 34 for both of the
mechanisms proposed in the formation of these natural
product/natural product analogsÐi.e. SN20 attack of the
12-carboxylic acid group at the 6-position of the allylic
hydroperoxide which forms both 6 and 12; and E1 elimina-
tion of the allylic hydroperoxyl group with H-7 which is
responsible for forming diene 13. Similar results have
been obtained from photo-oxygenation of arteannuic acid,
the 11,13-dehydro analog of 2.34
The remaining novel compounds 15±20, which were
isolated from transformations of 3 in organic solution in
the absence of molecular oxygen have not, as yet, been
reported as natural products from A. annua. Evidence for
the mechanism proposed for the formation of the lactone
endoperoxide 15 by 3,2-allylic rearrangement and subse-
quent `trapping' of the rearranged allylic hydroperoxide
by the 12-carboxylic acid group comes from close analogies
with our recently reported biomimetic synthesis of
arteannuin H -7) from secondary allylic hydroperoxide 9
-Scheme 2).6 There are several precedents in the
literature20±22 to suggest that the C-4/C-5 carbon±carbon
bond scission which has occurred in all of the products
16±20 is the result of Hock cleavage of the tertiary allylic
hydroperoxide 3, as shown in pathway -v) of Scheme 3. The
structure of compound 16 is particularly revealing as it is
only easily accounted for by the elimination of H1 from the
tertiary carbocation which is expected to be formed from the
1,2-shift of the D5,6-double bond of 3 to the internal oxygen
atom of the hydroperoxide -with the accompanying loss of
Dihydroarteannuin B -14) has also been described as a
natural product from A. annua3,35Ðthe mechanism
proposed for its formation from 3, by rearrangement of
the tertiary allylic hydroperoxide group to a b-epoxy
Scheme 4. Con®rmation of the stereochemistry for the 6-aldehyde substituent in compound 18 produced from Hock cleavage of 3, by ozonolysis of compound
2 which yielded the epimeric aldehyde 22. Both 6-aldehydes undergo slow spontaneous autoxidation to carboxylic acids, not 1,2,4-trioxanes.

L.-K. Sy, G. D. Brown / Tetrahedron 58 #2002) 897±908
903
the external oxygen atom as water) which is the initial step
in Hock cleavage. Nucleophilic attack at this cation by a
second molecule of 3, instead of the elimination of H1,
would then explain the formation of the dimer 17. Alter-
natively, re-addition of water to this same carbocation
would result in the enol 4, which is the starting point for
the further transformations shown in pathways -v)±-vii) of
Scheme 3. Although not isolated in this experiment, the
existence of compound 4 as the immediate product of
Hock cleavage might be inferred from the formation of all
of the products 18±20. Thus, the enol functional group in
proposed intermediate 4 has clearly undergone tautomeriza-
tion to an aldehyde in compound 18 which was the major
reaction product isolated from this experiment. The stereo-
chemistry of the aldehyde group in 18 was proven to be
6a-by correlations observed in NOESY sepctroscopy
12±16, 18, 19 and 21 probably represent end-points in the
varied transformations of 3, and that only 17 and 20 are
intermediates. Having thus obtained good evidence that
Hock cleavage is the mechanism by which spontaneous
cleavage occurs at C-4/C-5 in 3 -from the isolation of
compounds 16 and 17) and some indications for the
transient formation of the enol intermediate 4 which
would be expected from such a Hock cleavage mechanism
-by the isolation of end products 18 and 19), we next set out
to obtain more direct evidence for the intermediacy of
compound 4, which is proposed to undergo a second
autoxidation reaction in the transformation of 3 into 1 in
Scheme 1.
2.3. The second spontaneous autoxidation of enolic
intermediate 4
1
-following assignment of all H and 13C resonances by the
2D NMR experiments HSQC, HMBC and 1H±1H COSY, as
for all other novel compounds reported herein). Ozonolysis
of dihydroartemisinic acid -2) -Scheme 4) yielded the
epimeric 6b-aldehyde, compound 22, which had substan-
tially different NMR spectra to those of 18 -Tables 1 and
2), particularly in the vicinity of the 6-aldehyde substituent,
When TFA was added to a CDCl3 solution of 3 maintained
under an atmosphere of nitrogen at room temperature there
was an almost immediate conversion into a single product,
the enol lactone 19, which was proposed to be derived via
Hock cleavage of intermediate 4 in pathways -v)/-vi) of
Scheme 3. Based on this ®nding, we reasoned that it
might be possible to actually characterize the presumed
enol intermediate 4, before it became trapped by the
12-carboxylic acid group as a lactone in compound 19,
and without interference from the plethora of other products
which can be formed by alternative elimination, rearrange-
ment, dimerization and oxygenation reactions of the hydro-
peroxide 3 -i.e. pathways -i)±-iv) and -vii) in Scheme 3), by
using this same treatment under more controlled conditions.
There were some precedents in the literature for this
expectation from the work of Haynes and Vonwiller who
had previously obtained evidence from 1D 1H NMR spectra
to support the existence of the 11,13-dehydro 12-methyl
ester analog of 4, which was reported to be unexpectedly
stable at low temperature.20,22
²
thereby tending to con®rm this assignment. Enol lactone 19
may be formed either by attack of the 12-carboxylic acid at
the aldehyde group in 18 with ensuing loss of water from a
hemi-acetal intermediate, or it may be the result of geo-
metrical isomerism of the E-enolic double bond in
compound 4 to the Z-form, perhaps occurring by a radical
process as shown in pathway -vi), prior to lactonization.
Dimer 20 may also be derived from such a radical, which
attacks the enolic double bond of a second molecule of 4, as
shown in the self-condensation pathway -vi) in Scheme 3.
Obtaining con®rmation of the molecular weight of dimers
17 and 20 by mass spectroscopy was dif®cult, as they frag-
mented very easily under all ionization techniques which
were available -in addition, compound 17 was also unstable
when puri®ed and decomposed to deoxyartemisinin -21) in
a matter of days under normal laboratory conditions).
However, the presence of a correlation between the two
halves of the molecule in the HMBC spectrum of 20
-from dC 80.1 -C-60) ppm to dH 6.19 -H-5) ppm) clearly
showed that it was a dimer. In addition, we were able to use
diffusion ordered NMR spectroscopy -DOSY) to con®rm
that the molecular weight for 20 was approximately twice
that of the starting material 3.38,39
Gratifyingly, when performed under an atmosphere of
nitrogen at 233 K, treatment of a CDCl3 solution of tertiary
allylic hydroperoxide 3 with TFA did indeed result in
conversion into a single novel compound -contaminated
by small amounts of the expectable side-products 1, 6 and
1
19) in less than 30 min as observed by H NMR -Fig. 3).
When kept in the bore of the magnet of the NMR spectro-
meter at low temperature under nitrogen, the CDCl3 solution
of this compound was stable for several hours, permitting its
characterization by the 2D NMR experiments HSQC,
HMBC and 1H±1H COSY. This data unambiguously
indicated the planar structure of enol 4, although the stereo-
chemistry of this compound, in particular the geometry of
the enolic double bond, could not be determined from
nuclear Overhauser enhancements, as the NOESY spectrum
of 4 was dominated by positive exchange-type -EXSY)
cross-peaks,40 making observation of through-space
connections from negative nOe correlation cross-peaks
impossible. The E-geometry depicted for 4 in Schemes 1,
3 and 5 is therefore only an assumption based on the
expected mechanism for the Hock cleavage reaction.
With the bene®t of hindsight, resonances corresponding to
all of the compounds 1, 6 and 12±21 now isolated by HPLC
from the various possible reactions of 3 when left to stand in
CDCl3 solution under an atmosphere of nitrogen, could now
be identi®ed in the 1H NMR spectra which had been
acquired over the three-week period of the reaction -e.g.
dH 9.58 -br) and 2.14 -3H, s) for the 5-aldehyde and
15-methyl groups of compound 18, which accounted for
1
more than 50% of the ®nal crude reaction product by H
NMR). With the exception of dimers 17 and 20, all these
characteristic resonances were seen to grow steadily in
intensity with time, con®rming that all of compounds 1, 6,
When the temperature of the solution was raised to 298 K,
whilst maintaining an atmosphere of nitrogen, compound 4
was converted predominantly into the enol lactone 19, as
²
Compound 23 was also isolated as a minor product from ozonolysis.

904
L.-K. Sy, G. D. Brown / Tetrahedron 58 #2002) 897±908
Figure 3. Percentage of compounds 1, 3, 4 and 6 -estimated by integration of characteristic alkene resonances in 1H NMR) as a function of time following
treatment of a CDCl3 solution of 3 with TFA at 233 K under an atmosphere of nitrogen.
expected -smaller quantities of compounds 1 and 6 were
also isolated by HPLC). The reaction was completed within
1 h as shown by 1H NMR spectra of the CDCl3/TFA
solution, which were acquired at intervals of every few
minutes. By contrast, when the temperature was raised to
283 K, whilst at the same time allowing oxygen in the
atmosphere access to the solution, there was instead an
almost quantitative conversion of enol 4 into artemisinin
-1) -Scheme 5) within a few hours. This conversion of 4
It is worth pointing out that although the Hock cleavage
product 4 has thus been shown to be extremely susceptible
to spontaneous autoxidation by molecular oxygen, yielding
essentially only a single product, artemisinin -1), its
aldehyde tautomer, compound 18, was much less reactive
towards oxygen and was not converted to a 1,2,4-trioxane
ring system in organic solution. Instead, a solution of this
6a-aldehyde in CDCl3 underwent spontaneous autoxidation
to the corresponding 6a-carboxylic acid, compound 24
only after several weeks at room temperature and with no
detectable formation of artemisinin. The same was true of
the 6b-aldehyde 22, obtained from ozonolysis of dihydro-
artemisinic acid, which underwent slow oxidation to the
corresponding 6b-carboxylic acid, compound 25
-Scheme 4).
3
into 1 is presumably a radical process involving O2 -path-
way -vii) in Scheme 3) since the reaction could be studied in
the darkened environment of the bore of the magnet of an
NMR spectrometer -cf. comments in the ®rst section regard-
ing the spontaneous transformation of 3 into 1 in CDCl3
solution which was unaffected by light).
Scheme 5. Differing transformations of enol 4, obtained from Hock cleavage of 3, under atmospheres which either contain or exclude oxygen.

L.-K. Sy, G. D. Brown / Tetrahedron 58 #2002) 897±908
905
3. Conclusion
measuring the decrease in the integral for the H-5 resonance
-dH 5.11 ppm).3 Formation of natural products 1, 3, 6 and 7
was detected by identi®cation of characteristic peaks
associated with these products in the H NMR spectra of
the mixture.
Based on the preceding experimental results: -i) dihydro-
artemisinic acid -2) undergoes spontaneous autoxidation to
the tertiary allylic hydroperoxide 3 in CDCl3 solution; -ii)
this allylic hydroperoxide in turn undergoes transformations
into a wide variety of products, several of which appear to
be derived from the transient enol intermediate 4, which is
expected from Hock Cleavage; -iii) it was actually possible
to characterize compound 4 by 2D NMR at low temperature
and in the absence of molecular oxygen, following treat-
ment of 3 with acid; and -iv) compound 4 was converted
rapidly and exclusively into artemisinin -1) in the presence
of molecular oxygen; we believe that we now have good
evidence to support the mechanism for the spontaneous
autoxidation of 2 to 1 which has been suggested in Scheme
1. Based on our experience with the spontaneous autoxida-
tion of other natural products,30 we suspected that the
proximity of the 12-carboxylic acid group to the D4,5-double
bond in 2 might be responsible for the apparent ease of this
complex series of transformations. This hypothesis is
explored in the companion paper.
1
Compound 1:28 dH 5.86-1H, s, H-5), 3.40 -1H, dq, Jˆ12.6,
7.2 Hz, H-11) ppm.
Compound 3:3,26 dH 5.25 -1H, s, H-5), 2.73 -1H, dq, Jˆ7.0,
7.0 Hz, H-11) ppm.
Compound 6:3,29 dH 5.64 -1H, s, H-5), 3.15 -1H, dq, Jˆ6.7,
6.7 Hz, H-11) ppm.
Compound 7:3,6 dH 5.02 -1H, d, Jˆ11.0 Hz, H-5), 4.91 -1H,
s, H-15a/b), 4.84 ppm -1H, s, H-15a/b).
Peaks at dH 4.38 -m), 4.35 -m) and 5.56-s) ppm were
identi®ed as belonging, respectively, to H-15a/H-15b and
H-5 of the primary allylic hydroperoxide 11 formed by
3,2-allylic rearrangement of secondary allylic hydroperox-
ide 9 -not detectedÐbut see Ref. 6).
4. Experimental
4.1. General
4.3. Spontaneous transformations of dihydroartemisinic
acid tertiary hydroperoxide 3 in CDCl3 solution
Dihydroartemisinic acid tertiary allylic hydroperoxide -3)
was prepared by photo-oxygenation of dihydroartemisinic
acid -2)Ðsee Ref. 6for experimental details and puri®ca-
tion. A solution of 3 -1 mg) in CDCl3 -0.6ml) was left in an
NMR tube under laboratory conditions and 1H NMR spectra
were acquired once or twice a day over a period of two
weeks. Formation of known compounds 1 and 6 was
con®rmed by identi®cation of characteristic peaks asso-
All novel compounds reported were fully characterized by
1
the 2D NMR experiments HSQC, HMBC, H±1H COSY
and NOESY. Chemical shifts are expressed in ppm -d)
relative to TMS as internal standard. Proton chemical shifts,
multiplicities, coupling constants and integral reported in
1
this section are those which are clearly resolved in 1D H
NMR without recourse to 2D NMR analysis -see Tables in
the main text for full assignments by 2D NMR). All NMR
experiments were run on a Bruker DRX 500 instrument.
NMR experiments under an atmosphere of nitrogen were
performed using 5 mm NMR tubes equipped with a te¯on
valve which isolates the NMR tube from the atmosphere
1
ciated with these products in the H NMR spectra of the
mixture as above. A minor peak at dH 6.12 ppm seen in
some experiments was assigned to the H-5 proton of
compound 13b,4 and a more signi®cant peak at dH
8.02 ppm was assigned to formaldehyde -possibly formed
as a by-product of Hock cleavage reactions of 5).
1
-J. Young, 528-VL-7). HSQC, HMBC H±1H COSY and
NOESY spectra were recorded with 1024 data points in F2
and 256data points in F 1. High-resolution MS were
recorded in EI mode at 70 eV on a FinniganÐMAT 95
MS spectrometer. IR spectra were recorded in CHCl3 on a
Shimadzu FTIR-8201 PC instrument. Column chromato-
graphy was performed using silica gel 60±200 mm
-Merck). HPLC separations were performed using a Varian
chromatograph equipped with RI star 9040 and UV 9050
detectors and a Prep-Sil 20 mm£25 cm column, ¯ow rate
8 ml/min. Optical rotations were measured by a Perkin±
Elmer 343 polarimeter -Na 589 nm). [a]D values are
given in 1021 deg cm2 g21 and CHCl3 was used as a solvent.
4.4. Spontaneous transformations of the tertiary allylic
hydroperoxide 3 in CDCl3 solution under an atmosphere
of nitrogen
Dissolved oxygen was removed from a solution of 3
-30 mg) in CDCl3 -1 ml) in a NMR tube equipped with a
valve by the freeze±thaw method. The sample was left
under an atmosphere of N2 under laboratory conditions
1
and H NMR spectra were acquired every one or two days
over a period of three weeks. Formation of compounds 1
and 6 was con®rmed by identi®cation of characteristic peaks
associated with these products in the 1H NMR spectra of the
mixture, as earlier. The mixture was separated by repeated
HPLC after 3 weeks.
4.2. Spontaneous transformations of dihydroartemisinic
acid 2 in CDCl3 solution
Samples of dihydroartemisinic acid were obtained from A.
annua plants as described previously.3 A solution of
dihydroartemisinic acid -2) -1 mg) in CDCl3 -0.6ml) was
left in an NMR tube under laboratory conditions and H
NMR spectra were acquired every 3±7 days over a period
of 2 months. The disappearance of 2 was monitored by
Compound 1:28 -1.6mg, 5%; Rt 41.7 min, 15% EtOAc/n-
hexane).
1
Compound 6:3,29 oil. -1.2 mg, 4%; Rt 29.8 min, 15%
EtOAc/n-hexane).

906
L.-K. Sy, G. D. Brown / Tetrahedron 58 #2002) 897±908
Compound 12:31 oil. -2.1 mg, 6%; Rt 10.2 min, 50%
EtOAc/n-hexane; Rt 35.1 min, 15% EtOAc/n-hexane). IR
s), 1.18 -3H, d, Jˆ6.9 Hz), 0.94 -3H, d, Jˆ6.4 Hz)Ðsee
also Table 2; 13C NMR: see Table 1; HREIMS m/z -rel.
int.) 250.1570 [M12H2O, C15H22O3 requires 250.1569]
-22), 232 -23), 222 -7), 192 -16), 167 -36), 149 -100);
CIMS: 269 [M11] -35), 251 -28), 233 -100), 205 -52).
nmax -CHCl3): 3026, 2934, 2874, 1751, 1456 cm21 1H
;
NMR d -CDCl3) ppm: 5.53 -1H, s), 2.74 -1H, dq, Jˆ9.0,
8.0 Hz), 2.19 -1H, ddd, Jˆ12.5, 9.0, 2.5 Hz), 1.70 -3H, s),
1.35 -3H, d, Jˆ8.0 Hz), 0.96-3H, d, Jˆ6.2 Hz)Ðsee also
Table 2; 13C NMR: see Table 1; HREIMS m/z -rel. int.)
234.1613 [M1, C15H22O2 requires 234.1620] -19), 219 -4),
190 -91), 161 -100).
Compound 19: oil. -2.3 mg, 8%; Rt 16.6 min, 50%
EtOAc/n-hexane; Rt 60.0 min, 22% EtOAc/n-hexane); H
1
NMR d -CDCl3) ppm: 6.08 -1H, s), 2.95 -1H, dq, Jˆ7.2,
7.0 Hz), 2.63 -1H, ddd, Jˆ17.8, 9.2, 4.8 Hz), 2.45 -1H, ddd,
Jˆ17.8, 8.5, 8.5 Hz), 2.16-3H, s), 1.22 -3H, d, Jˆ7.0 Hz),
0.99 -3H, d, Jˆ6.1 Hz)Ðsee also Table 2; 13C NMR: see
Table 1; HREIMS m/z -rel. int.) 250.1575 [M1, C15H22O3
requires 250.1569] -3), 232 -3), 211 -1), 192 -100).
Compound 13:4 oil. -0.9 mg, 3%; Rt 11.5 min, 50%
EtOAc/n-hexane; Rt 21.0 min, 15% EtOAc/n-hexane); H
1
NMR d -CDCl3) ppm: 6.13 -1H, s, H-5), 3.84 -1H, q,
Jˆ6.9 Hz, H-11), 1.79 -3H, s, H-15), 1.23 -3H, d,
Jˆ6.9 Hz, H-13), 0.99 -3H, d, Jˆ6.0 Hz, H-14); HREIMS
m/z -rel. int.) 234.1614 [M1, C15H22O2 requires 234.1620]
-44), 189 -30), 161 -100).
Compound 20: oil. -1.0 mg, 3%; Rt 42.3 min, 50%
EtOAc/n-hexane; Rt 42.5 min, 15% EtOAc/n-hexane). IR
nmax -CHCl3): 3400±2600 -br), 2930, 2856, 1740,
Compound 14:3,35 oil. -2.4 mg, 8%; Rt 18.0 min, 50%
EtOAc/n-hexane; Rt 28.8 min, 15% EtOAc/n-hexane); H
1715 cm21; H NMR d -CDCl3) ppm: 6.19 -1H, s), 5.92
1
1
-1H, d, Jˆ13.2 Hz, 50-OH), 5.69 -1H, d, Jˆ13.2 Hz), 2.18
-3H, s), 2.07 -3H, s), 1.15 -3H, d, Jˆ6.9 Hz), 1.12 -3H, d,
Jˆ7.2 Hz), 1.07 -3H, d, Jˆ6.3 Hz), 0.86 -3H, d,
Jˆ7.1 Hz)Ðsee also Table 2; 13C NMR: see Table 1;
CIMS: m/z -rel. int.) 515 [M11] -62), 281 -100).
NMR d -CDCl3) ppm: 3.02 -1H, s, H-5), 2.77 -1H, dq,
Jˆ7.5, 7.8 Hz, H-11), 2.25 -1H, ddd, Jˆ13.4, 7.5, 2.9 Hz,
H-7), 1.38 -3H, d, Jˆ7.8 Hz, H-13), 1.36-3H, s, H-15), 0.96
-3H, d, Jˆ6.6 Hz, H-14); 13C NMR: 179.5 -C, C-12), 82.9
-C, C-6), 59.5 -CH, C-5), 58.1 -C, C-4), 50.2 -CH, C-7),
45.6-CH, C-1), 38.6-CH, C-11), 34.8 -CH 2, C-9), 30.5
-CH, C-10), 24.1 -CH2, C-3), 23.1 -CH3, C-15), 21.6
-CH2, C-8), 18.6-CH , C-14), 16.3 -CH2, C-2), 12.6-CH ,
Compound 21:28,41 oil. -1.4 mg, 4%; Rt 10.8 min, 50%
EtOAc/n-hexane; Rt 12.3 min, 30% EtOAc/n-hexane); H
1
NMR d -CDCl3) ppm: 5.70 -1H, s), 3.18 -1H, dq, Jˆ12.1,
7.2 Hz), 2.00 -1H, ddd, Jˆ12.1, 4.4, 4.4 Hz), 1.53 -3H, s),
1.20 -3H, d, Jˆ7.2 Hz), 0.94 -3H, d, Jˆ5.6Hz)Ðsee also
Table 2; 13C NMR: see Table 1; HREIMS m/z -rel. int.)
266.1517 [M1, C15H22O4 requires 266.1518] -7), 224 -28),
195 -18), 165 -100), 151 -60).
3
3
C-13).
Compound 15: oil. -0.9 mg, 3%; Rt 9.2 min, 50% EtOAc/n-
hexane; Rt 13.1 min, 15% EtOAc/n-hexane). IR nmax
-CHCl3) 3013, 2927, 2854, 1759 cm21
;
1H NMR d
-CDCl3) ppm: 5.79 -1H, s), 2.63 -1H, dq, Jˆ8.4, 7.7 Hz),
2.35 -1H, m), 2.12 -1H, ddd, Jˆ12.8, 8.4, 3.4 Hz), 2.05 -1H,
m), 1.60 -3H, s), 1.32 -3H, d, Jˆ7.7 Hz), 0.94 -3H, d,
Jˆ6.6 Hz)Ðsee also Table 2; 13C NMR: see Table 1;
HREIMS m/z -rel. int.) 250.1571 [M1, C15H22O3 requires
250.1569] -11), 222 -8), 192 -2), 179 -100), 151 -26); CIMS:
251 [M11] -100), 233 -34), 223 -11), 205 -35).
4.5. Ozonolysis of dihydroartemisinic acid 2
A cooled -2788) solution of dihydroartemisinic acid -2)
-57 mg, 0.24 mmol) in CH2Cl2/CH3OH -1:1, 40 ml) was
subjected to ozonolysis to give an oil -52 mg, 90% w/w),
which was separated by HPLC -35% EtOAc/n-hexane).
Compound 16: oil. -1.8 mg, 6%; Rt 14.2 min, 50%
EtOAc/n-hexane; Rt 22.0 min, 15% EtOAc/n-hexane); H
Compound 22: oil. -20 mg, 0.07 mmol, 31%; Rt 31.0 min).
[a]Dˆ246.7 -c 0.57, CHCl3); IR nmax -CHCl3): 3400±2600
-br), 2936, 2855, 1706 cm21; 1H NMR d -CDCl3) ppm: 9.97
-1H, br s), 2.67 -2H, br s), 2.15 -3H, s), 1.23 -3H, d,
Jˆ7.0 Hz), 0.98 -3H, d, Jˆ6.3 Hz)Ðsee also Table 2; 13C
NMR: see Table 1; HREIMS m/z -rel. int.) 250.1574
[M12H2O, C15H22O3 requires 250.1569] -3), 227 -31),
209 -100), 181 -100).
1
NMR d -CDCl3) ppm: 6.07 -1H, s), 4.91 -1H, dd, Jˆ6.8,
6.8 Hz), 2.58 -2H, m), 1.77 -3H, s), 1.26 -3H, d, Jˆ6.8 Hz),
0.87 -3H, d, Jˆ6.4 Hz)Ðsee also Table 2; 13C NMR: see
Table 1; HREIMS m/z -rel. int.) 250.1569 [M1, C15H22O3
requires 250.1569] -35), 232 -4), 177 -100), 159 -58), 133
-70).
Compound 23: oil. -5 mg, 0.02 mmol, 7%; Rt 23.5 min); 1H
NMR d -CDCl3) ppm: 5.02 -1H, d, Jˆ6.4 Hz), 3.46 -3H, s),
2.75 -1H, ddd, Jˆ16.9, 9.6, 3.8 Hz), 2.61 -1H, dq, Jˆ3.8,
7.1 Hz), 2.33 -1H, ddd, Jˆ16.9, 8.0, 5.2 Hz), 2.14 -3H, s),
1.21 -3H, d, Jˆ7.1 Hz), 0.97 -3H, d, Jˆ6.4 Hz)Ðsee also
Table 2; 13C NMR: see Table 1.
Compound 17: oil. -1.0 mg, 3%; Rt 11.1 min, 50%
EtOAc/n-hexane; Rt 17.3 min, 15% EtOAc/n-hexane). The
sample decomposed to 21 after puri®cation under laboratory
conditions within a few days. 1H NMR d -CDCl3) ppm: 5.88
-1H, s), 5.47 -1H, s), 2.74 -1H, m), 2.67±2.64 -2H, m), 2.58
-1H, dq, Jˆ4.4, 7.1 Hz), 1.43 -3H, s), 1.30 -3H, s), 1.23 -3H,
d, Jˆ7.1 Hz), 1.19 -3H, d, Jˆ7.0 Hz), 0.92 -3H, d,
Jˆ6.4 Hz), 0.89 -3H, d, Jˆ7.0 Hz)Ðsee also Table 2; 13C
NMR: see Table 1; CIMS: 477 -100), 281 -52).
4.6. Conversion of a CDCl3 solution of 3 into enol lactone
19 by treatment with TFA under an atmosphere of
nitrogen
Compound 18: oil. -9.1 mg, 30%; Rt 30.8 min, 50%
EtOAc/n-hexane; Rt 68.0 min, 30% EtOAc/n-hexane); H
NMR d -CDCl3) ppm: 9.58 -1H, d, Jˆ4.9 Hz), 2.14 -3H,
1
Dissolved oxygen was removed from a solution of 3
-10 mg) in CDCl3 -0.6ml) as earlier and TFA -2 ml) was

L.-K. Sy, G. D. Brown / Tetrahedron 58 #2002) 897±908
907
added at room temperature. Analysis by 1H NMR indicated
the complete transformation of 3 into 19 in less than 5 min
int.) 266.1510 [M12H2O, C15H22O4 requires 266.1518] -4),
248 -19), 227 -23), 209 -60), 181 -100).
1
-the time than it took to lock, shim and acquire a H NMR
spectrum).
Acknowledgements
4.7. Characterization of enol intermediate 4 from
treatment of 3 with TFA under an atmosphere of
nitrogen
We thank the Generic Drug Programme of the Chemistry
Department of the University of Hong Kong for providing a
postdoctoral fellowship to Dr Jaclyn Sy. This project was
funded by a grant from the CRCG.
Dissolved oxygen was removed from a solution of 3 -25 mg,
0.09 mmol) in CDCl3 -1 ml) as earlier and TFA -ca. 1 ml)
was added to a just-thawed solution. The NMR tube was
immediately transferred to the NMR spectrometer -probe-
head pre-cooled to 233 K) and formation of compound 4
References
1
was complete within 30 min as determined by H NMR
1. Cumming, J. N.; Wang, D.; Park, S. B.; Shapiro, T. A.; Posner,
G. H. J. Med. Chem. 1998, 41, 952±964.
2. Rong, Y.-J.; Wu, Y.-L. J. Chem. Soc., Perkin Trans. 1 1993,
2147±2148.
-disappearance of dH 5.21 ppm -H-5) for the starting
material). Some 13C resonances of products were assigned
by 2D NMR -all dC and dH values are in CDCl3/TFA at
233 K -ppm)).
3. Sy, L.-K.; Brown, G. D.; Haynes, R. Tetrahedron 1998, 54,
4345±4356.
Compound 4: dH 6.31 -1H, s), 2.68±2.57 -3H, m), 2.32 -3H,
s), 1.12 -3H, d, Jˆ5.9 Hz), 0.93 -3H, d, Jˆ6.9 Hz)Ðsee
4. Sy, L.-K.; Zhu, N.-Y.; Brown, G. D. Tetrahedron 2001, 57,
8495±8510.
1
also Table 2; almost all H resonances showed positive
5. Kim, N.-C.; Kim, S.-U. J. Korean Agric. Chem. Soc. 1992, 35,
106±109.
EXSY-type correlations in NOESY spectra: e.g. dH 6.31
-H-5) correlated with 2.15 -H-7), 1.12 -H-13) ppm etc.;
13C NMR: see Table 1.
6. Sy, L.-K.; Ngo, K.-S.; Brown, G. D. Tetrahedron 1999, 55,
15127±15140.
7. Ngo, K.-S.; Brown, G. D. Tetrahedron 1999, 55, 15109±
15126.
Compound 1: dH 5.95 -1H, s, H-5), 3.46-1H, dq, Jˆ7.0,
7.0 Hz, H-11), 1.49 -3H, s, H-15); 13C NMR 105.7 -C, C-4),
94.1 -CH, C-5), 79.5 -C, C-6), 32.8 -C-11), 12.6 -CH3, C-13).
8. Xu, X.-X.; Zhu, J.; Huang, D.-Z.; Zhou, W.-S. Tetrahedron
1986, 42, 819±828.
9. Zhou, W.-S. Pure Appl. Chem. 1986, 58, 817±824.
10. Schmid, G.; Hofheinz, W. J. Am. Chem. Soc. 1983, 105, 624±
625.
Compound 6: dH 5.67 -1H, s, H-5), 3.29 -1H, dq, Jˆ7.0,
6.9 Hz, H-11), 1.72 -3H, s, H-15), 1.18 -3H, d, Jˆ6.9 Hz,
H-13); 13C NMR 182.0 -C, C-12), 143.4 -C, C-4), 120.9
-CH, C-5), 42.5 -C-7), 40.2 -C-11), 30.4 -C-3), 23.8 -C-8),
9.5 -CH3, C-13).
11. Avery, M. A.; Chong, W. K. M.; Jennings-White, C. J. Am.
Chem. Soc. 1992, 114, 974±979.
12. Nowak, D. M.; Lansbury, P. T. Tetrahedron 1998, 54, 319±
336.
Compound 19: dH 6.08 -1H, s, H-5), 3.02 -1H, dq, Jˆ7.0,
7.0 Hz, H-11), 2.17 -3H, s, H-15), 1.25 -3H, d, Jˆ7.0 Hz,
H-13), 1.00 -3H, d, Jˆ6.8 Hz, H-14); 13C NMR 212.4 -C,
C-4), 173.4 -C, C-12), 131.4 -CH, C-5), 124.9 -C, C-6), 46.8
-CH, C-1), 36.3 -CH, C-11), 28.5 -C-8), 11.7 -CH3, C-13).
13. Avery, M. A.; Bonk, J. D.; Mehrotra, S. J. Labelled Compd
Radiopharm. 1995, 48, 249±254.
14. Zhou, W.-S.; Xu, X.-X. Acc. Chem. Res. 1994, 27, 211±216.
15. Jung, M.; Li, X.; Bustos, D. A.; ElSohly, H. N.; McChesney,
J. D. Tetrahedron Lett. 1989, 30, 5973±5976.
16. Ravindranathan, T.; Anil Kumar, M.; Menon, R. B.; Hiremath,
S. V. Tetrahedron Lett. 1990, 31, 755±758.
17. Ye, B.; Wu, Y.-L. J. Chem. Soc., Chem. Commun. 1990, 726±
727.
Further conversion of 4 in the presence of O2 at 283 K or
under an atmosphere of nitrogen at 298 K: the valve of the
NMR tube was opened to admit O2 or kept closed to main-
tain an atmosphere of N2 as the temperature of the probe-
head was raised.
18. Liu, H.-J.; Yeh, W.-L.; Chew, S. Y. Tetrahedron Lett. 1993,
34, 4435±4438.
19. Constantino, M. G.; Beltrame, M.; da Silva, G. V. J. Synth.
Commun. 1996, 26, 321±329.
4.8. Compounds 24 and 25 from spontaneous
autoxidation of aldehydes 18 and 22 in CDCl3 solution
20. Vonwiller, S. C.; Warner, J. A.; Mann, S. T.; Haynes, R. K.
J. Am. Chem. Soc. 1995, 117, 11098±11105.
21. Haynes, R. K.; Vonwiller, S. C. J. Chem. Soc., Chem.
Commun. 1990, 451±453.
Compound 24: Oil. -Rt 30.9 min, 50% EtOAc/n-hexane); 1H
NMR d -CDCl3) ppm: 2.14 -3H, s), 1.19 -3H, d, Jˆ7.2 Hz),
0.92 -3H, d, Jˆ5.7 Hz)Ðsee also Table 2; 13C NMR: see
Table 1; CIMS: 285 [M11] -40), 267 -100), 221 -46).
22. Haynes, R. K.; Vonwiller, S. C. Acc. Chem. Res. 1997, 30, 73±
79.
23. Acton, N.; Roth, R. J. J. Org. Chem. 1992, 57, 3610±3614.
24. Roth, R. J.; Acton, N. J. Nat. Prod. 1989, 52, 1183±1185.
25. Roth, R. J.; Acton, N. J. Chem. Ed. 1991, 68, 612±613.
26. Wallaart, T. E.; Pras, N.; Quax, W. J. J. Nat. Prod. 1999, 62,
1160±1162.
Compound 25: Oil. [a]Dˆ236.5 -c 1.0, CHCl3); IR nmax
-CHCl3): 3400±2600 -br), 2930, 1709, 1456 cm21 1H
;
NMR d -CDCl3) ppm: 2.88 -1H, br), 2.73 -1H, m), 2.16
-3H, s), 1.29 -3H, d, Jˆ8.4 Hz), 0.90 -3H, d, Jˆ6.2 Hz)Ð
see also Table 2; 13C NMR: see Table 1; HREIMS m/z -rel.
27. Wallaart, T. E.; van Uden, W.; Lubberink, H. G. M.;

908
L.-K. Sy, G. D. Brown / Tetrahedron 58 #2002) 897±908
Woerdenbag, H. J.; Pras, N.; Quax, W. J. J. Nat. Prod. 1999,
62, 430±433.
36. Gardner, H. W.; Weisleder, D.; Nelson, E. C. J. Org. Chem.
1984, 49, 508±515.
28. Agrawal, P. K.; Bishnoi, V. J. Sci. Indian Res. 1996, 55, 17±
26.
37. Jung, M.; Yoo, Y.; ElSohly, H. N.; McChesney, J. D. J. Nat.
Prod. 1987, 50, 972±973.
29. Brown, G. D. J. Nat. Prod. 1992, 55, 1756±1760.
30. Sy, L.-K.; Brown, G. D. Tetrahedron 1999, 55, 119±132.
31. Sy, L.-K.; Brown, G. D. Phytochemistry 2001, 58, 1159±
1166.
38. Morris, K. F.; Johnson, C. S. J. Am. Chem. Soc. 1992, 114,
3139±3141.
39. Barjat, H.; Morris, G. A.; Smart, S.; Swanson, A. G.;
Williams, S. C. R. J. Magn. Reson. B 1995, 108, 170±172.
40. Braun, S.; Kalinowski, H.-O.; Berger, S. 100 and More Basic
NMR Experiments; VCH: Weinheim, 1996pp 341±343.
41. Lee, I.-S.; El-Sohly, H. N.; Croom, E. M.; Hufford, C. D.
J. Nat. Prod. 1989, 52, 337±341.
32. El-Feraly, F.-S.; Al-Meshal, I. A.; Khalifa, S. I. J. Nat. Prod.
1989, 52, 196±198.
33. Sy, L.-K.; Brown, G. D. J. Chem. Soc., Perkin Trans. 1 2001,
2421±2429.
34. Acton, N.; Roth, R. J. Phytochem. 1989, 28, 3530±3531.
35. ElMarkaby, S. A.; El-Feraly, F.-S.; ElSohly, H. N.; Croom,
E. M.; Hufford, C. D. J. Nat. Prod. 1987, 50, 903±909.