Fourier transform infrared investigation of non-heme Fe(III) and Fe(II) decomposition of artemisinin and of a simplified trioxane alcohol

Article abstract of DOI:10.1021/jm010848d

Fourier transform infrared spectra are reported for the Fe(III)- and Fe(II)-mediated activation of the antimalarial agents artemisinin I and its simplified synthetic analogue, trioxane alcohol 2. By monitoring the frequencies of the newly established marker lines in the FTIR spectra, the products of the Fe(II) and Fe(III) reactions have been characterized. In both reactions, artemisinin is activated giving a product mixture of a ring-contracted tetrahydrofuran acetatal 3, C₄-hydroxy deoxyartemisinin 4, and deoxyartemisinin 5. These data illustrate that the oxidation state of the iron places no restrictions on the endoperoxide reduction mechanism. The FTIR difference (light - dark) spectra indicate that the endoperoxide moiety of artemisinin is photolabile and that the resulted products have the same vibrational characteristics as those observed in the reactions with Fe(II) and Fe(III). The use of ¹⁸O-¹⁸O enriched endoperoxide in 2 has allowed us to identify two oxygen sensitive modes in the reactions with Fe(II). The reduction of the peroxide bond by Fe(II) in trioxane alcohol 2 follows both the C - C cleavage and 1,5-H shift pathways and produces a ring-contracted tetrahydrofuran acetal 6 which is converted to tetrahydrofuran aldehyde 7 and C₄-hydroxy deoxytrioxane alcohol 8, respectively. The cleavage of the O - O bond in 1 and 2 by iron and the ability to correlate vibrational properties of the reaction products with structural properties of the isolated products suggest that infrared spectroscopy is an appropriate tool to study the mode of action of antimalarial endoperoxides.

Full text of DOI:10.1021/jm010848d

3
150
J . Med. Chem. 2001, 44, 3150-3156
F ou r ier Tr a n sfor m In fr a r ed In vestiga tion of Non -Hem e F e(III) a n d F e(II)
Decom p osition of Ar tem isin in a n d of a Sim p lified Tr ioxa n e Alcoh ol
Sofia Kapetanaki and Constantinos Varotsis*
Department of Chemistry, University of Crete, 71409 Heraklion, Crete, Greece
Received February 12, 2001
Fourier transform infrared spectra are reported for the Fe(III)- and Fe(II)-mediated activation
of the antimalarial agents artemisinin 1 and its simplified synthetic analogue, trioxane alcohol
2
. By monitoring the frequencies of the newly established marker lines in the FTIR spectra,
the products of the Fe(II) and Fe(III) reactions have been characterized. In both reactions,
artemisinin is activated giving a product mixture of a ring-contracted tetrahydrofuran acetatal
3
4
, C -hydroxy deoxyartemisinin 4, and deoxyartemisinin 5. These data illustrate that the
oxidation state of the iron places no restrictions on the endoperoxide reduction mechanism.
The FTIR difference (light - dark) spectra indicate that the endoperoxide moiety of artemisinin
is photolabile and that the resulted products have the same vibrational characteristics as those
1
8
18
observed in the reactions with Fe(II) and Fe(III). The use of O- O enriched endoperoxide in
has allowed us to identify two oxygen sensitive modes in the reactions with Fe(II). The
2
reduction of the peroxide bond by Fe(II) in trioxane alcohol 2 follows both the C-C cleavage
and 1,5-H shift pathways and produces a ring-contracted tetrahydrofuran acetal 6 which is
4
converted to tetrahydrofuran aldehyde 7 and C -hydroxy deoxytrioxane alcohol 8, respectively.
The cleavage of the O-O bond in 1 and 2 by iron and the ability to correlate vibrational
properties of the reaction products with structural properties of the isolated products suggest
that infrared spectroscopy is an appropriate tool to study the mode of action of antimalarial
endoperoxides.
Artemisinin 1 (Chart 1) is a novel natural sesquiter-
pene lactone (1,2,4-trioxane), isolated from Artemisia
annua, with an endoperoxide function essential in the
chemotherapy of chloroquine-resistant strains of Plas-
modium falciparum malaria. The mode of action of
artemisinin and its simplified trioxane analogues is at
the intraerythrocytic stage.1 The heme groups re-
leased from the digestion of hemoglobin of infected blood
cells are aggregated into an insoluble substance called
Ch a r t 1
-4
5
hemozoin. Recently, the crystal structure of â-hematin,
which was shown to be essentially hemozoin, was
4
reported. The spread of P. falciparum has initiated a
tremendous interest in the mechanism of action, drug
development, and biochemistry behind the antimalarial
activity of artemisinin, and other structurally related
synthetic 1,2,4-trioxanes. AÄ number of tetracyclic arte-
misinin derivatives and tricyclic 1,2,4-trioxanes have
been developed as potent and rapidly acting anti-
7
-17
malarials.
Despite intensive research efforts, neither
the mechanism of activating these drugs into cytotoxic
species nor the centers and their redox state associated
with the linkage have yet been identified. It has been
accepted, however, that the different iron pools, such
as free iron and heme, within the malarial parasite, are
responsible for activating the 1,2,4-trioxane antimalari-
als to form cytotoxic radical intermediates.18,19
The mode of action of artemisinin has received a great
deal of attention in recent years as it allows for the
clarification of the molecular mechanisms in this class
of drugs and the rationalization of antimalarial activity
in others. Despite complexities apparently caused by
differences in the choice of experimental approach by
several research groups, some common properties were
found. On the basis of these experiments, structures for
many intermediates have been proposed, and the rates
for the associated O-O cleavage steps have been
determined. Some of the structures were inferred based
*
To whom correspondence should be addressed. Telephone: +30-
8
1-393653. Fax: +30-81-393601. E-mail: varotsis@edu.uoc.gr.
1
0.1021/jm010848d CCC: $20.00 © 2001 American Chemical Society
Published on Web 08/14/2001

Non-Heme Fe(III) and Fe(II) Decomposition of Artemisinin
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 19 3151
on observed final products, and others were proposed
based on spectroscopic properties. On the basis of Fe(II)-
induced decomposition of artemisinin and other triox-
anes, it has been suggested that the cleavage of the
O-O bond of the endoperoxide moiety is the initial key
step in the activation of the drug.7-17 In such a way,
the cleavage of the endoperoxide generates oxygen-
centered radicals that immediately rearrange to carbon
centered radicals. The transformation from the oxygen-
to carbon-centered radicals can occur either from a
homolytic cleavage of the C3-C4 bond or an intramo-
lecular 1,5-H atom shift. In recent studies of Posner and
co-workers, the major products of the Fe(II)-induced
The Fourier transform infrared (FTIR) technique has
been applied to the study of some 1,2,4-trioxanes and
has been found to be powerful in probing the bonds of
the endoperoxide moiety and the bonds of the rings
owing to the presence of the O-O, the C-O, the O-O-
C, as well as the CdO modes in the spectrum. The
endoperoxide moiety is especially useful in this regard
because the homolytic cleavage of the O-O bond can
be characterized and hence can be used to assess the
vibrational properties of the O- and C-centered radicals
and subsequently that of the C3-C4 bond cleavage.
In the work presented here, we have applied the FTIR
approach to further characterize the reactions of arte-
misinin and its synthetic analogue trioxane alcohol with
Fe(II) and Fe(III). Compounds 3, 4, and 5 are detected
in the Fe(II)/artemisinin reaction. They are readily
activation of artemisinin were found to be compounds
4
3
, 4, and 5 in 29%, 15%, and 54% yields, respectively.
Compounds 3 and 4 are pyrolysis products of artemisi-
nin.2 Compound 4 is also a known microbial metabolite
0
-1
characterized by the -COOCH3 mode at 1717 cm
2
1,22
-1
of artemisinin,
and compound 5 is both a microbial
(compound 3), the hydroxyl mode at 3440 cm (com-
-
1
and mammalian metabolite.22 More recently, Wu and
co-workers in similar experiments isolated and charac-
pound 4), and the six-membered lactone at 1755 cm
compound 5). Although the Fe(III)/artemisinin reaction
(
is slower, we show cleavage of the O-O bond and the
formation of three products which have vibrational
characteristics similar to those found in the products
of the Fe(II)/artemisinin reaction. The product ratios in
the Fe(II) and Fe(III) reactions depends on the oxidation
state of the iron and thus on whether the activation
follows the 1,5 H-shift or the C3-C4 cleavage pathway.
In the case of the Fe(II)-induced cleavage of trioxane
alcohol 2, three major products are detected, and their
characteristic CdO vibrations are located at 1717 and
terized compounds 3 and 4 in 25% and 67% yields,
respectively.23 In addition, they were able to isolate an
unstable epoxide that was postulated in earlier studies
by Posner et al. as the species responsible for the
antimalarial activity of the drug,24 and also they trapped
and characterized a secondary C-4 radical. The latter
observation provides the first direct evidence for the
involvement of radicals in the reductive cleavage of the
trioxane-type antimalarial compounds. Very interest-
ingly, deoxyartemisinin 5 was not obtained at all, and
its absence was attributed to the major difference
between the reaction conducted in aqueous medium and
that in THF. They also reported that no reaction
-
1
-1
1
740 cm and that of the hydroxyl (OH) at 3433 cm
.
From consideration of determined structures, we infer
these frequencies as being associated with compounds
6, 7, and 8. We have also applied laser photolysis FTIR
techniques to monitor the FTIR spectra of artemisinin
photoproducts and found that the O-O bond is photo-
labile. The artemisinin photoproducts have vibrational
characteristics that are similar to those observed in
their corresponding Fe(II)- and Fe(III)-induced reductive
cleavage products. From the different product ratios
observed in the Fe-induced cleavage of the endoperoxide
and that in the photolytic cleavage, we infer that there
is a mechanistic difference between the two processes.
These results also provide important links between
metabolites and chemical reaction products that have
been observed in the antimalarial mode of action of
trioxanes.
2
3
occurred when FeSO4 was replaced by FeCl3. Haynes
and Vonwiller,25 however, isolated and characterized
compounds 3 and 4 in the reactions with both FeSO4
and FeCl3. These studies led to a consensus on the
structures of compounds 3, 4, and 5 and puzzling
disagreements on the main product ratios and the
mechanism for the endoperoxide reduction by Fe(II).
The analogous reactions with Fe(III), however, have
received little attention, and a number of questions
concerning the structures of the activation products and
the mechanism of the critical O-O bond cleavage
process are ill answered.
Several important properties of artemisinin have been
determined from studies of simpler trioxane analogues.
It has been established by Posner and co-workers that
deoxygenation of 1,2,4-trioxanes into corresponding 1,3-
Ma ter ia ls a n d Meth od s
Artemisinin, ferrous bromide, and ferric chloride (Aldrich
Chemical Co.) were used as purchased. Trioxane alcohol was
4
dioxolanes occurs via an unzipping-zipping process. The
Fe(II)-induced cleavage of the peroxide bond in trioxane
tosylate leads through the C-C bond cleavage to a ring-
contracted tetrahydrofuran acetal 6 and then produces
a stable electophilic tetrahydrofuran aldehyde 7. On the
other hand, the 1,5-H atom abstraction process produces
a stable C4-hydroxy deoxytrioxane alcohol 8 as a mix-
ture of two diastereomers. These findings led them to
conclude that the reduction of the endoperoxide by Fe(II)
follows a different mechanistic course that leads to
different products than the endoperoxide cleavage by
7
synthesized according to Posner et al. The oxygen-18 labeled
1
8
trioxane alcohol was synthesized by introducing O during
2
the photooxygenation step. The 18
All reactions were carried out in acetonitrile (CH
ing 1 equiv of either FeBr or FeCl . The FTIR spectra were
O
was a product of Isotec.
CN) contain-
2
3
2
3
-
1
recorded from thin film samples at 2 cm resolution with a
Bruker Equinox 55 FTIR spectrometer equipped with a liquid
nitrogen cooled mercury cadmium telluride detector. The
photolysis experiments were performed with a Nd:YAG laser,
532 nm, ten 10 mJ flashes, 7 ns duration.
8
18
Resu lts a n d Discu ssion
non Fe(II) reducing agents. Recently, the use of the O2
enriched trioxane alcohol allowed us to detect the heme
Fe(IV)dO intermediate in the artemisinin/hemin dimer
The FTIR spectra presented here and the frequencies
of the modes are fully consistent with earlier studies of
artemisinin and with the vibrational data reported for
2
6
reaction.

3
152 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 19
Kapetanaki and Varotsis
those vibrations found in other 1,2,4-trioxanes.17 The
O-O and C-O stretching modes of the O-O-C moiety
-1
are located at 724, 831, and 883 cm , respectively, and
-
1
the CdO of δ-lactone is located at 1737 cm . Figure
B,C shows that extensive changes have occurred in the
1
vibrational properties of artemisinin in the reactions
with Fe(II) and Fe(III). In the low-frequency region of
-
1
the spectrum, the lines at 831 and 883 cm
have
disappeared, an expected characteristic of the endo-
peroxide cleavage. We find five new bands located at
-
1
1
596, 1717, 1755, 2938, and 3440 cm in the high-
frequency region of the FTIR spectra. The line at 1717
is assigned as the CdO stretching mode of the -COOCH3
moiety of compound 3. We assign the line at 1755 as
originating from the CdO of compound 5. Finally, the
-1
line at 3440 cm is attributed to OH mode of compound
. These results illustrate the independence of the
4
reaction products on the oxidation state of the iron and
hence indicate that all the features in the spectra result
from the cleavage of the O-O bond. However, the
-
1
relative proportions of the 1755 and 1717 cm species
in the Fe(III)/artemisinin reaction are different from
that detected in the Fe(II)/artemisinin reaction. Thus,
two separate pathways, as discussed below, are involved
in the activation products of artemisinin by iron.
To further investigate the artemisinin activation
process we have monitored the FTIR spectra of arte-
misinin after 532 nm laser pulsed excitation. The FTIR
difference spectra (light - dark) presented as an inset
F igu r e 1. FTIR spectra of artemisinin (trace A), Fe(II)/
artemisinin reaction (trace B), and Fe(III)/artemisinin reaction
(
trace C). The inset shows the “light - dark” FTIR spectrum
-1
of Figure 1 show two peaks at 1711 and 1758 cm as
of artemisinin.
-
1
well as three troughs at 724, 832, 885, and 1732 cm
.
1
7
This observation suggests that the O-O bond is photo-
labile and that new species are formed after the endo-
peroxide cleavage. The observed spectrum is character-
istic of compounds 3 and 5. This is indicated by the lines
some 1,2,4-trioxanes by J efford and co-workers. The
-
1
FTIR modes in the 800-1200 cm range have been
shown to be sensitive indicators to the O-O and C-O
modes of the O-O-C unit, and at 1737 cm-1 to the
-
1
at 1711 and 1758 cm . Thus, in the absence of either
Fe(II) or Fe(III), the photolytic cleavage of the O-O
bond is also capable to convert artemisinin to com-
pounds 3 and 5. Interestingly, the intensity ratio of the
δ-lactone carbonyl ν(CdO) mode. It has been shown
from previous experiments that addition of Fe(II) to
artemisinin leads to the formation of compound 3 and
either compound 4 or 5 which have been isolated and
-
1
1
758 /1711 cm modes suggests that product ratio is
characterized by their CdO and OH vibrations. The
FTIR spectra23 of the isolated product 3 exhibit the
similar to that measured in the Fe(II) reaction.
-1
δ-lactone carbonyl, ν(CdO) mode at 1760 cm while the
To gain further insight into the endoperoxide chem-
istry and thus in the mode of action of artemisinin, we
have studied the artemisinin analogue trioxane alcohol
2
3
14
analogous ν(CdO) of the isolated products 4 and 5
are located at 1730 and 1742 cm-1, respectively. In
-
1
28
compound 4, the ν(OH) is located at 3050 cm and the
CdO mode of the -COOCH3 moiety of compound 3 at
The isolated products from the Fe(II)/
trioxane tosylate reaction exhibit the CdO mode of the
(IC50 ∼ 34 ng/mL, concentration of the drug required
for killing 50% of P. falciparum in culture). The use of
this compound and other related trioxanes has allowed
Posner and co-workers to propose the 1,5-H atom shift
and C-C cleavage processes that have been widely
accepted. In contrast to artemisinin, which has a
-
1 23
1
746 cm
.
-
1
-
COOCH3 moiety of compound 6 at 1726 cm , the Cd
O mode of compound 7 at 1735 cm-1, and the OH mode
-
1 8
-1
of compound 8 at 3400 cm . The FTIR spectra we
report here allow for determination of the final products
and characterization of their vibrational properties. We
assign the bands in the 1700-1760 and 3400 cm
regions as originating from CdO modes (1700 cm for
ketones and 1755 cm for lactones or saturated esters)
and OH modes, respectively, and thus utilize them as
marker lines for product formation since spectral over-
lap with other sharp contributions in this spectral range
is not large.
Figure 1A shows the FTIR spectra of artemisinin, and
those of the Fe(II)- and Fe(III)-induced reaction products
are shown in Figure 1B,C, respectively. The principal
bands seen in Figure 1A are assigned as analogous to
characteristic vibration at 1737 cm due to the carbonyl
group of the lactone moiety, the trioxane alcohol 2 lacks
vibrations in this region because of the absence of any
CdO groups. The FTIR spectra of the Fe(II)-reductive
-
1
-
1
16
18
cleavage of 2 were measured for both O2 and O2
-1
enriched endoperoxides. The FTIR spectra (Figure 2A)
1
6
16
of the final products of the O- O enriched endoper-
oxide reaction with Fe(II) exhibit peaks at 1717 and
-
1
-1
1740 cm . Figure 2B shows that the 1717 cm mode
1
6
16
in the O- O has lost intensity and is downshifted to
-1
18
18
-1
1712 cm in the O- O derivative while the 1740 cm
mode remains unshifted. These results indicate that the
-
1
1717 cm mode originates from species that contain
groups that have resulted from the O-O cleavage. It

Non-Heme Fe(III) and Fe(II) Decomposition of Artemisinin
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 19 3153
stretching vibrations of compounds 6 and 7 and that at
-
1
1
015 cm to compound 8.
On the basis of the above results, it now appears very
likely that the electron transfer from Fe(II) to the
endoperoxide in 1 has cleaved the O-O bond and
initiated the formation of radical intermediates which
are rapidly converted to compounds 3, 4, and 5 as the
8
major products. Similar conclusions were previously
4
drawn by Posner and co-workers. However, Wu et al.
reported the formation of compounds 3 and 4, only.23 It
is clear from these experiments that the formation of
compound 5 is dependent on the experimental condi-
tions. It is detectable in the protocol used by Posner and
co-workers or by the present authors but not in that
2
3
used by Wu and co-workers. Up to now, the endoper-
oxide cleavage of 1,2,4-trioxanes is agreed by most
researchers in the field to be induced by Fe(II), without
addressing the main products in the analogous reaction
with Fe(III). The spectra shown in Figure 1C demon-
strate that compounds 3, 4, and 5 are also formed in
the Fe(III)/artemisinin reaction. This is in sharp con-
trast to the conclusions drawn by Wu et al.23 They
reported that no reaction occurs when the FeSO4 is
replaced with FeCl3. Thus, their interpretation that the
2
+
n+
replacement of Fe
by other metal ions (M ) will
retard both the â-scission and the radical substitution
at the oxygen atom is inconsistent with our data. The
main products reported by Haynes and Vonwiller,25
however, who detected compounds 3 (85% yield) and 4
(
8% yield) as the main products of the Fe(III)/artemisi-
F igu r e 2. FTIR spectra of the Fe(II) reaction with trioxane
nin reaction, are in partial agreement with our data,
and this further supports our assignment of the 1717
cm being associated with compound 3. The 1717 cm
1
6
16
alcohol in O- O enriched endoperoxide (traces A and A′) and
1
8
18
O- O enriched endoperoxide (trace B and B′).
-1
-1
-
1
band is much stronger than the 1755 cm band, which
suggests that compound 3 is the major reaction product.
For a complete vibrational analysis of the Fe/artemisi-
nin reaction products, a variety of isotopic substituted
derivatives would be necessary, but the present data
set is adequate for our purposes. It is clear from the
experiments reported here that both Fe(II) and Fe(III)
react with artemisinin to produce the same final prod-
ucts. There is general agreement that artemisinin
activation products arise from the homolytic fission of
the endoperoxide bond initiated by single electron
donation from Fe(II). In the absence of electron transfer
from Fe(III) to the endoperoxide, we postulate that
precomplexation, rather than direct cleavage of the
peroxide leads to formation of products (see below).
has been reported that the CdO modes of the tosylate
derivatives of compound 6 and 7 are at 1726 and 1735
-
1
8
cm , respectively. The carbonyl mode of compound 7
is not a consequence of the endoperoxide cleavage and
thus is not expected to show isotope frequency shift. On
-1
the other hand, the 1726 cm carbonyl mode originates
from the endoperoxide cleavage and therefore is ex-
pected to show isotope frequency shift. We adopt the
assignment and mechanism offered by Posner and co-
4
-1
workers that the 1717 and 1740 cm modes we detect
are associated with compounds 6 and 7, respectively.
We interpret the small isotope frequency shift at 1717
-1
cm as a result of vibrational coupling between the Cd
O mode and the modes of compound 6. The broad shape
and intensity of the OH mode at 3433 cm of compound
in the spectra of the final products makes it difficult
-1
The formation of new products in the photolytic
activation of artemisinin is not a surprising result since
it is clear from our data as well as those of others that
the activation of 1 progresses rapidly to the formation
of compounds 3, 4, and 5. The data reported by Lin and
2
to draw simple conclusions concerning the formation of
compound 8. Thus the OH mode is unable to serve as a
probe of compound 8 since trioxane alcohol 2 possesses
already a hydroxyl group. However, the observed low-
frequency data shown in the inset exhibit an oxygen
20
1
co-workers, who fully characterized by IR, H NMR,
13
C NMR, and X-ray crystallography the heat decom-
-
1
isotope shift pattern in the 900-1000 cm region. The
position (190 °C) products of artemisinin, were inter-
preted as the formation of compounds 3 and 4. They also
proposed that the mechanism for the formation of these
products involves the homolytic cleavage of the endo-
peroxide to generate a free radical intermediate that
rearranges or decomposes to give the observed products.
The comparison of our photolytic data to the reference
data clearly illustrates that species are produced with
spectra very similar to the spectrum shown in the inset
-
1
oxygen-sensitive lines at 987 and 1015 cm are not
unique features only seen in the products of 2 but are
common C-O-C modes found in other ring complexes
containing the C-O-C unit. It has been demonstrated
that as the ring becomes smaller, the frequency of the
band involving the asymmetric stretch of the C-O-C
bonds decreases. We interpret the observed shift pattern
-1
at 987 cm as originating from the asymmetric C-O-C

3
154 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 19
Kapetanaki and Varotsis
for high antimalarial activity.27,28 Recently, Gu et al.29
in a DTF study showed that the rate constant for the
8
-1
-35 -1
classical 1,5-H shift is 1.3 × 10 s and 2.9 × 10
s
at 298 and 30 K, respectively, and concluded that the
corresponding O-centered radical is detectable experi-
mentally at low temperature. Moreover, they calculated
8
-1
the C-C cleavage rate to be 1.9 × 10 s and 1.2 ×
-
33 -1
1
0
s
at 298 and 30 K, respectively, and reported
that the C-centered radical is 5.2 kcal/mol more stable
than the O-centered radical. Since there is experimental
evidence for the formation of a secondary C-4 radical
2
3
and of an unstable epoxide by Wu et al. and for the
3
0
formation of a primary C-3 radical, their structures
were used in the scheme. The epoxide was characterized
by its IR spectrum giving rise to ν(OH) and ν(CdO) at
-
1
3
500 and 1728 cm , respectively.
In the reaction scheme depicted in Figure 3, two
separate pathways are indicated for the decay of the
primary O-centered radicals. In the first pathway, if iron
associates with O-1, then compound 9 is formed which
is converted via the C3-C4 scission to the C-4 radical
compound 10. The latter compound is converted to
compound 3. On the other hand, if iron associates with
O-2, then through the 1,5-H shift compound 11 is
converted to the C-4 radical compound 12. The later
compound can be either converted to the unstable
epoxide 13 with the concomitant formation of Fe(IV)d
O or to the stable compound 5 and the formation of
Fe(III)-OH. Compound 13 can be further converted to
the final product 4. Under all experimental conditions
we used there is no evidence for the build-up of either
a secondary radical or of the unstable epoxide, due to
-
1
the absence of a peak at 1728 cm ; thus we assume
that their lifetimes are very short. In the case of Fe(II)
reaction, the ratio of the I(1755)/I(1717) is 1.2 while in
the photolysis products is 2.8. Consequently, under the
conditions in which the O-O bond is cleaved photolyti-
cally, and in the absence of bound Fe to neither of the
oxygens of the endoperoxide bridge, the dominant
pathway is that in which compound 5 is formed. The
reaction of Fe(III) with artemisinin suggests precom-
plexation of Fe(III) with artemisinin, rather than a
direct, reductive cleavage of the endoperoxide. This way,
the first step in the reaction is displacement of a ligand
from Fe(III) and Lewis acid complexation to O1 or O2.
The scheme depicted in Figure 4 shows that the com-
plexation of Fe(III) is associated with either O1 or O2.
In both cases, compounds 3, 4, and 5 are formed. The
F igu r e 3. Proposed mechanism for the Fe(II)-induced activa-
tion of artemisinin.
of Figure 1, namely, compounds 3, 4, and 5. We detect
no change in the position of the lines in the 1700-1750
-
1
cm region, as would be expected, from the transition
of the Fe-induced decomposition to the photolytic de-
composition of artemisinin. This observation suggests
that the formation of the final products in artemisinin
is independent of the way the O-O bond is cleaved.
In an effort to understand the progression of the
intermediates in the O-O cleavage of artemisinin, we
first present a reaction scheme for artemisinin in Figure
-
1
strong line we detect at 1717 cm as compared to the
3
. The structures of the intermediates and those of the
-
1
1
755 cm indicates, in contrast to the Fe(II)-induced
final products were taken from the literature and from
the present experiments. The data also show that the
chemical conditions of the reactions determine which
of the intermediates can accumulate. Although in the
scheme depicted in Figure 3 we show O- and C-centered
radicals, there is no evidence for the build-up of such
intermediates in our measurements. The characteriza-
tion of the O-centered radicals produced in the cleavage
of the O-O bond as well as the detection of the produced
C-centered radicals formed either through a 1,5-
hydrogen atom shift or the homolytic C3-C4 bond
cleavage remains problematic. It has been shown,
however, in several 1,2,4-trioxanes such as artemisinin
that, of the two pathways, only the one involving a C4
radical intermediate through a 1,5-H shift is important
cleavage products and photolysis products, that the
dominant species is compound 3.
The scheme depicted in Figure 5 is based on the
4
mechanism proposed by Posner and co-workers for the
activation of trioxane tosylate and is used to compare
the mechanism and structures of the final products with
those found in the activation of artemisinin. It shows
the stages in the reaction of Fe(II) with compound 2.
The reductive cleavage of the O-O bond signals the
onset of the Fe coordination to either O1 or O2. At
present, the sequence of intermediates that occur at this
stage of the reaction are the least well resolved, and
considerable research effort is aimed at elucidating the
processes that occur as the drug passes in to and out of

Non-Heme Fe(III) and Fe(II) Decomposition of Artemisinin
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 19 3155
F igu r e 4. Proposed mechanism for the Fe(III)-induced activa-
tion of artemisinin.
the Fe(III)-O level. Issues that are under active scrutiny
at present include the timing of the O-O bond cleavage
chemistry and the structure of the Fe(III)-OH and
Fe(IV)dO intermediates. The fate of compound 6 in the
reaction of compound 2 with Fe(II) contrasts sharply
with that of compound 1. The reaction in 2 proceeds to
form aldehyde 7, whereas no such product has been
identified in the reaction of 1 with Fe(II). Thus, it
appears that the lactone in 1 prevents conversion of the
compound to an aldehyde. We suggest, in agreement
F igu r e 5. Proposed mechanism for the Fe(II)-induced activa-
tion of trioxane alcohol.
resulting in the same products as those obtained in the
Fe(II) and Fe(III) reactions lays the foundation for
identifying the structures of the oxygen and carbon
radicals and thereby unraveling the mode of action of
this fascinating class of antimalarials.
Refer en ces
4
with the initial proposal by Posner et al., that the 1,5-H
(1) Klayman, D. L. Qinghaosu (Artemisinin): An Antimalarial Drug
shift process is operative in 2 with the release of Fe(IV)d
O that characterizes the analogous reaction in 2. Thus,
even in the reaction of 2 with Fe(II), we suggest that
there will be a branch at the initial stage and that the
products distribution will be determined by the rate
constant ratios, i.e., that a proportion of the reaction
proceeds through the 1,5-H shift with release of Fe(IV)d
O, and the remainder through the C3-C4 pathway.
In summary, the data reported here demonstrate that
the reaction of Fe(II) with artemisinin is very similar
to that of its synthetic analogue trioxane alcohol. In
artemisin, the electron transfer from Fe(II) to initiate
a cascade of radical intermediates would generate
products via the 1,5-H and C3-C4 pathways. Although
it was originally accepted that no Fe(III)/artemisinin
reaction occurs, the measurements we have made here
unequivocally demonstrate the formation of products
after mixing Fe(III) with artemisinin. Finally, the
observation that the endoperoxide bond is photolabile
from China. Science 1985, 228, 1049-1055.
(2) Meshnick, S. R. The mode of action of antimalarial endoperox-
ides. Trans. R. Soc. Trop. Med. Hyg. 1994, 88, 31-31.
(
3) Meshnick, S. R.; Taylor, T. E.; Kamchonwongpaisan, S. Arte-
misinin and the Antimalarial Endoperoxides: from Herbal
Remedy to Targeted Chemotherapy Microbiol. Rev. 1996, 60,
3
01-315.
(
4) Cumming, J . N.; Ploypradith, P.; Posner, G. H. Antimalarial
Activity of Artemisinin (Qinghaosu) and Related Trioxanes:
Mechanism(s) of Action. Adv. Pharmacol. 1997, 37, 2253-297.
5) Slater, A. F. G.; Swiggard, W. J .; Orton, B. R.; Flitter, W. D.;
Goldberg, D. E.; Cerami, A.; Henderson, G. B. An iron-carboxy-
late bond links the heme units of malaria pigment. Proc. Natl.
Acad. Sci. U.S.A. 1991, 88, 325-329.
(
(
6) Pagola, S.; Stephens, P. W.; Bohle, D. S.; Kosar, A.; Madsen, S.
K. The structure of malaria pigment â-haematin. Nature 2000,
404, 307-310.
(7) Posner, G. H.; Oh, C. H.; Gerena, L.; Milhous, W. K. Extraor-
dinarily Potent Antimalarial Compounds: New, Structurally
Simple, Easily Synthesized, Tricyclic 1,2,4-Trioxanes. J . Med.
Chem. 1992, 35, 2459-2467.
(
8) Posner, G. H.; Oh, C. H. A Regiospecifically Oxygen-18 Labeled
1
,2,4-Trioxane: A Simple Chemical Model System To Probe the
Mechanism(s) for the Antimalarial Activity of Artemisinin
(Qinghaosu). J . Am. Chem. Soc. 1992, 114, 8328-8329.

3
156 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 19
Kapetanaki and Varotsis
(
9) Peters, W.; Robinson, B. L.; Rossier, J . C.; J efford, C. W. The
chemotherapy of rodent malaria. XLVIII. The activities of some
synthetic 1,2,4-trioxanes against chloroquine-sensitive and chlo-
roquine-resistant parasites. Part 1: Studies leading to the
development of novel cis- fused cyclopenteno derivatives. Ann.
Trop. Med. Parasitol. 1993, 87 (1), 1-7.
specific parasite proteins. Antimicrob. Agents Chemother. 1994,
38, 1854-1858.
(20) Lin, A. J .; Klayman, D. L.; Hoch, J . M.; Silverton, J . V.; George,
C. Thermal Rearrangement and Decomposition Products of
Artemisinin (Qinghaosu). J . Org. Chem. 1985, 50, 4504-4508.
(21) Lee, I.-K.; Elsohly, H. N.; Croom, E. M.; Hufford, C. D. Microbial
Metabolism Studies of the Antimalarial Sesquiterpene Arte-
misinin. J . Nat. Prod. 1989, 52, 337-341.
(
10) Peters, W.; Robinson, B. L.; Rossiter, D.; Misra, D.; J efford, C.
W. The chemotherapy of rodent malaria. XLIX. The activities
of some synthetic 1,2,4-trioxanes against chloroquine-sensitive
and chloroquine-resistant parasites. Part 2: Structure-activity
studies on cis-fused cyclopenteno-1,2,4-trioxanes (fenozans)
against drug-sensitive and drug-resistant lines of Plasmodium
berghei and P. yoelii ssp. NS in vivo. Ann. Trop. Med. Parasitol.
(
22) Lee, I.-K.; Hufford, C. D. Metabolism of Antimalarial Sesqui-
terpene Lactones. Pharm. Ther. 1990, 48, 345-355.
23) Wu, W. M.; Wu, Y.; Wu, Y.-L.; Yao, Z.-J .; Zhou, C.-M.; Li, Y.;
Shan, F. Unified Mechanistic Framework for the Fe(II)-Induced
Cleavage of Qinghaosu and Derivatives/Analogues. The First
Spin-Trapping Evidence for the Previously Postulated Secondary
C-4 Radical. J . Am. Chem. Soc. 1998, 120, 3316-3325.
24) Posner, G. H.; Cumming, J . N.; Ploypradith, P.; Oh, C.-H.
Evidence for Fe(IV)dO in the Molecular Mechanism of Action
of the Trioxane Antimalarial Artemisinin. J . Am. Chem. 1995,
(
1
993, 87 (1), 9-16.
(
11) Peters, W.; Robinson, B. L.; Tovey, G.; Rossier, J . C.; J efford, C.
W. The chemotherapy of rodent malaria. L. The activities of some
synthetic 1,2,4-trioxanes against chloroquine-sensitive and chlo-
roquine-resistant parasites. Part 3: Observation on ‘Fenozan-
(
5
0F’, a difluorinated 3,3’-spirocyclopentane 1,2,4-trioxane. Ann.
1
17, 5885-5886.
25) Haynes, R. K.; Vonwiller, S. C. The Behaviour of Qinghaosu
Artemisinin) in the Presence of Non-Heme Iron (II) and (III).
Trop. Med. Parasitol. 1993, 87 (2), 111-123.
(
(
(
(
12) J efford, C. W.; Favarger, F.; Vicente, M.; J acquier, Y. The
decomposition of cis-fused cyclopenteno-1,2,4-trioxanes induced
by ferrous salts and some oxophilic reagents. Helv. Chim. Acta
(
Tetrahedron Lett. 1996, 37, 257-260.
26) Kapetanaki, S.; Varotsis, C. Ferryl-oxo heme intermediate in the
antimalarial mode of action of artemisinin. FEBS Lett. 2000,
1
995, 78, 452-458.
(
13) Posner, G. H.; Tao, X.; Cumming, J . N.; Klinedinst, D.; Shapiro,
T. Antimalarially Potent, Easily Prepared, Fluorinated Endo-
peroxides Tetrahedron Lett. 1996, 37, 7225-7228.
14) J efford, C. W.; Vicente, M. G. H.; J acquier, Y.; Favarger, F,;
Mareda, J .; Millasson-Schmidt, P.; Brunner, G.; Burger, U. The
Deoxygenation and Isomerization of Artemisinin and Artemether
and Their Relevance to Antimalarial Action. Helv. Chim. Acta
4
74, 2/3, 238-241.
27) Posner, G. H.; Wang, D.; Cumming, J . N.; Oh, C. H.; French, A.
N.; Bodley, A. L.; Shapiro, T. Further Evidence Supporting the
Importance of and the Restrictions on a Carbon-Centerd Rdical
for High Antimalarial Activity of 1,2,4-Trioxanes Like Artemisi-
nin. J . Med. Chem. 1995, 38, 2273-2275.
(
1
996, 79, 1475-1487.
(28) Posner, G. H.; Oh, C. H.; Wang, D.; Gerena, L.; Milhous, W. K.;
Meshnick, S. R.; Asawamahasadka, W. Mechanism-Based De-
sign, Synthesis, and in Vitro Antimalarial Testing of New
4-Methylated Trioxanes Structurally Related to Artemisinin:
The Importance of a Carbon-Centerd Radical for Antimalarial
Activity. J . Med. Chem. 1994, 37, 1256-1258.
(29) Gu, J .; Chen, K.; J iang, H.; Leszczynski, J . The Radical
Transformation in Artemisinin: A DFT Study. J . Phys. Chem.
A 1999, 103, 9364-9369.
(
15) Posner, G. H.; Park, S. B.; Gonzalez, L.; Wang, D.; Cumming, J .
N.; Klinedinst, D.; Shapiro, T. A.; Bachi, M. D. Evidence for the
Importance of High-Valent FedO and of a Diketone in the
Molecular Mechanism of Action of Antimalarial Trioxane Ana-
logues of Artemisinin. J . Am. Chem. Soc. 1996, 118, 3537-3538.
16) Posner, G. H.; Cumming, J . N.; Woo, S.-H.; Ploypradith, P.; Xie,
S. Orally Active Antimalarial 3-Substituted Trioxanes: New
Synthetic Methodology and Biological Evaluation. J . Med. Chem.
(
1
998, 41, 940-951.
(30) Butler, A. R.; Gilbert, B. C.; Hulme, P.; Irvine, L. R.; Renton,
L.; Whitwood, A. C. EPR Evidence for the Involment of Free
Radicals in the Iron-Catalysed Decomposition of Qinghaosu
(
(
(
17) Monhaupt, M.; Hagemann, H.; Perler, J .-P.; Bill, H.; Boukouv-
alas, J .; Rossier, J .-C.; J efford, C. W. A Vibrational Study of Some
1
18) Hong, Y.; Yang, Y.-Z.; Meshnick, S. R. The interaction of
artemisinin with malarial hemozoin. Mol. Biochem. Parasitol.
1
19) Asawamahasakda, W.; Ittarat, I.; Pu, Y.-M.; Ziffer, H, H.;
Meshnick, S. R. Reaction of antimalarial endoperoxides with
,2,4-Trioxanes. Helv. Chim. Acta 1988, 71, 992-999.
(
Artemisinin) and Some Derivatives; Antimalarial Action of
Some Polycyclic Endoperoxides. Free Radical Res. 1998, 28,
71-476.
4
994, 63, 121-128.
J M010848D