Sie sind auf Seite 1von 24

Current Medicinal Chemistry 2001, 8, 1803-1826

1803

Why Artemisinin and Certain Synthetic Peroxides are Potent Antimalarials.


Implications for the Mode of Action
Charles W. Jefford*
Department of Organic Chemistry, University of Geneva, CH-1211 Geneva, Switzerland
Abstract: The discovery that the sesquiterpene peroxide yingzhaosu A (13 ) and 1,2,4trioxane artemisinin (14 ) are active against chloroquine-resistant strains of Plasmodium
falciparum, has opened a new era in the chemotherapy of malaria. In vitro and in vivo
tests with synthetic structurally simpler trioxanes clearly demonstrate that much of the
skeleton of 14 is redundant and that chirality is not required for activity. In addition,
structure-activity relations and the search for the pharmacophore reveal that high
antimalarial activity can be displayed by molecules which do not resemble the geometry
of 13 and 14 at all. The possible mode of action of 13 , 14 , and synthetic peroxides is examined. They are believed
to kill intraerythrocytic Plasmodium by interacting with the heme discarded by proteolysis of ingested
hemoglobin. Complexation of heme with the peroxide bond followed by electron transfer generates an oxy
radical that evolves to the ultimate parasiticidal agent. Experiments with ferrous reagents indicate that active
peroxides including 14 and its congeners kill the parasite by alkylation with a sterically non-encumbered Ccentered radical. However, another possibility is the involvement of a Fe(IV)=O species as the toxic agent. The
review covers our own and other contributions to this timely topic and evaluates the different mechanisms
proposed for the mode of action of peroxidic antimalarials.

INTRODUCTION
Background to the Disease
Malaria is an age-old disease and one of the commonest
causes of illness in the world. It is estimated that about 2.5
billion people living in malarious zones are at risk and that
300-500 million clinical cases of malaria occur each year [1].
The disease is transmitted in a two-stage process by the bite
of an infected female anopheline mosquito which can transfer
to the human host four species of parasite, Plasmodium
malariae, P. ovale, P. vivax and P. falciparum [2]. The
parasites first invade the liver and then as merozoites
penetrate erythrocytes where they proliferate and eventually
burst the parasitized cells releasing about 25 daughter
merozoites which then invade more erythrocytes so initiating
a new infectious cycle. Some merozoites evolve into male
and female gametes which undergo fertilization after being
drawn into the gut of a second mosquito when it bites the
infected host. The resulting sporozoites are then injected via
the salivary glands into a fresh host through another bite
thereby maintaining transmission.
The regions most affected by malaria are the northern part
of South America, Central America, Africa below the Sahara,
the Indian subcontinent, South East Asia, Vietnam,
Indochina, Indonesia and the southern rim of the Pacific
basin. Most patients suffer from uncomplicated malaria,
characterized by fever, anemia and debilitation. However,

*Address correspondence to this author at the Department of Organic


Chemistry, University of Geneva, CH-1211 Geneva, Switzerland; Ph.:
+41-22-7763316; Fax: +41-22-7763601;
e-mail: Charles.Jefford@chiorg.unige.ch
0929-8673/01 $28.00+.00

chronic infections can lead to kidney failure, hyperactive


malarial splenomegaly and Burkitts carcinoma. Most
malaria is caused by P. vivax and is essentially benign, but
Plasmodiun falciparum is virulent. It is the most dangerous
because the level of parasitemia is much higher than for the
other three species. Also more than 80% of the hemoglobin
within the cell gets degraded to heme by the parasite.
Moreover, the parasitized red blood cells in the late
maturation stage lose their platelet-like form and no longer
freely circulate in the blood. They become knobbly and stick
together blocking the microvasculature in vital organs such
as the brain; with coma and death being the usual sequel. In
tropical Africa alone falciparum malaria claims each year the
lives of 1.5-3 million children most of whom are under five
years of age.
Treatment and Prevention of Malaria
The prevention and treatment of malaria today constitutes
an acute challenge for modern medicine and public health
management because many of the traditional quinoline-based
drugs are becoming increasingly ineffective in certain parts of
the world owing to multi-drug resistance [3]. In order to put
the problem in perspective a short overview of the
development of traditional antimalarials is presented.
Jesuit missionaries in Peru around 1630 discovered that
the bark of the cinchona tree allayed fever. A few years later
exportation of the bark to Europe at Romes behest led to its
inclusion in the pharmacopoeia as a cure for fever [4]. In
1820 the active principle was found in to be the alkaloid
quinine (1). This finding biased subsequent drug
development in the years prior to 1939 towards chemically
related remedies such as chloroquine (2), amodiaquine (3),
2001 Bentham Science Publishers Ltd.

1804

Current Medicinal Chemistry, 2001, Vol. 8, No. 15

HO

Charles W. Jefford

HO

Me
Et2N

NH

NH
Et2N

MeO
N

Cl
1

Cl

2
Me

Et2 N

MeO

NH
N

OMe
HN
Cl

N
4

NEt 2
Me
5

Fig. (1). Some traditional quinoline antimalarials.

mepaquine (4) and pamaquine (5). 4-Aminoquinoline and 9acridine derivatives (2-4) and 1 act as blood schizonticides,
whereas the 8-amino derivative 5 acts in the liver [5] Fig.
(1). It should be mentioned that the mode of action of these
quinoline derivatives was not known at the time.
Consequently, the design of antimalarial drugs in the early

HN

NH

Cl

Et

CHMe 2
NH
H2N

HN
O

NH

As a stopgap measure, mefloquine or Lariam (10), either


used alone or in conjunction with 7 and 9 (Fansimef), was
O

Cl

HN

Unfortunately, in 1960 or thereabouts, the incidence of P.


falciparum started to alarmingly increase. At the same time
new chloroquine-resistant strains arose [9].

H2N
N

NH 2
7

NH 2
MeO

NH 2

O
N

MeO

Fig. (2). Some traditional inhibitors of dihydrofolate reductase and synthetase.

post-World War II years, based purely on mechanism, was a


notable advance [6,7]. Proguanil (6), pyrimethamine (7),
dapsone (8) and sulfadoxine (9) were rationally developed to
curtail the growth of the malarial parasite in the liver and
blood stages by virtue of their selective inhibition of
dihydrofolate reductase and synthetase Fig. (2).
Disadvantages of Traditional
Drugs

Nitrogen

Containing

With this stock of cheap drugs available for prevention


and cure, malaria was regarded as a vanquished disease [8].

developed Fig. (3). However, mefloquine, since it too is a


quinoline derivative, soon elicited resistant strains.
Originally, the designers of Fansimef believed that 7 and 9
(termed Fansidar) would prevent resistance developing
against 10, but the validity of this idea has been questioned
[10].
In any event, the risk of severe cutaneous reaction
(Stevens-Johnson syndrome) by Fansidar rules out its
prophylactic use against falciparum malaria [11].
Furthermore, Fansidar is inefficacious against P. vivax and
carries a significant risk of mortality. In an endeavor to fill
the niche left by progressively obsolescent chloroquine,
Cl

HO

OH
Bu
N
Bu

Cl

H
N

CF 3

OH
O

CF 3
10
Fig. (3). Some successor antimalarials.

11

CF 3

Cl
12

Implications for the Mode of Action

Current Medicinal Chemistry, 2001, Vol. 8, No. 15

recourse was recently made to antique drugs such as


atovaquone (11) and proguanil (6), combined together as
Malarone, and a derivatized phenanthrene isostere
halofantrine (12) [12].

documented, being highly active against P. vivax and


chloroquine-resistant P. falciparum. Rather than taking 13
and 14 for what they are, namely prototypes, tremendous
efforts were deployed to devise syntheses and to make semisynthetic first generation derivatives as potential drug
candidates [21, 22]. The different aspects of artemisinin and
its congeners have been well covered in many reviews [2328]. A recent review is devoted to peroxidic antimalarials in
general [29].

Apart from the diminished effectiveness of the


aforementioned nitrogen heterocycles due to resistance by P.
falciparum, there are certain personal disadvantages and risks
for the patient or user. Chloroquine is limited in its
geographical use, only working in the Middle East, Mexico
and Central America. Mefloquine is expensive, 100 times
mores so than chloroquine, and has resulted in seizures and
psychiatric disorders. Halofantrine is equally expensive,
unsuitable for prophylaxis, and has led to cases of
cardiotoxicity. Even quinine is never totally effective, and its
toxic side effects deter its usage [13].

Semi-synthetic Derivatives
In the case of yingzhaosu A (13), a synthetic route to the
natural product based on (-)-carvone (15) as starting material
[30] was adapted for preparing in some 10 or so steps
essentially the surrogate molecules arteflene or Ro 42-1611
(16) and Ro 41-3823 (17) as pure enantiomers in addition to
other analogues [31] Fig. (5). In comparison, derivatives of
artemisinin were much easier to come by, merely entailing
the initial reduction of the freely available and crystalline
natural product to the epimeric and -lactols (18) which
were then converted to esters and ethers by standard
procedures Fig. (6). Scores of semi-synthetic derivatives of
18 were produced of which the most important are -arteether
(19), -artemether (20), sodium -artelinate (21) and artesunic acid (22). All have antimalarial activities
commensurate with that of the parent 14 [32, 33].

THE ADVENT OF PEROXIDIC ANTIMALARIALS


Yingzhaosu and Artemisinin
Against this disheartening backdrop of increasingly
unsatisfactory N-heterocyclic drugs coupled with the rising
incidence of the deadly falciparum malaria, the advent of two
non-nitrogenous lead compounds was not only timely, but
signaled a new era of antimalarial chemotherapy [14, 15]. It
was discovered in 1979 that two sesquiterpene peroxides
obtained from Chinese medicinal plants, yingzhaosu A (13)
and qinghaosu or artemisinin (14), possessed powerful
antimalarial activity [16] Fig. (4). Yingzhaosu A (13), a
typical 1,2-dioxane, occurs as a decomposition product
arising in the stored roots of a sparsely growing vine,
Artabotrys uncinatus (Lam.) Merr. [17]. Artemisinin (14) is
a complex tetracyclic 1,2,4-trioxane found as the active
principle of Artemisia annua Linn., a wild shrub of widespread habitat [18-20].
O
Me
Me

Me

Me

O
O

OH

Me

OH

11 B
6

O
C

Me

Despite many successful syntheses, those presently


devised for 13 and 14 are impractical either in terms of
length or cost. For example, the surrogate arteflene has not
been further promoted as a commercial pharmaceutical
product, mainly for reasons of cost. Even the extra couple of
steps required to convert 14 into 19-22 probably add too
much to the price of the products for their intended use in
third world markets. The idea of using the lead or even a
first generation product as the actual commercial entity is
shortsighted. The usual proceeding for creating a new drug is
first to identify the pharmacophore of the providentially
provided natural product by stripping away the extraneous
bits. Once the essential structural elements are laid bare, the
next step is to elucidate the mode of action. With such
information in hand, it should be possible to design more
potent, cheaper, synthetically accessible, and more
biologically suitable antimalarials, and to gain, into the
bargain, mechanistic insights that might reveal other
therapeutic applications.

Me

OO
D

13

14

Fig. (4). Yingzhaosu A and artemisinin.

Although the evidence is largely anecdotal, 13 is active


against P. berghei, while the activity of 14 is securely

CF 3

10 or so steps

Me

O
Me

Me
15

CF 3
and

Me

1805

O
16

Fig. (5). Synthesis of arteflene (Ro 42-1611) and Ro 41-3823 from carvone.

Me
O

Me
O

O
17

(CH2) 10Me

1806

Current Medicinal Chemistry, 2001, Vol. 8, No. 15

Charles W. Jefford

HO

Me

EtO

OO

Me
Me

Me

Me

18

Me

20

Me

19

COOH
OO

Me

OO

Me

OO

Me

C6H 4CO2Na
O
O
O

Me

MeO

OO

Me

14

Me

OO

Me

22

21

Me

Fig. (6). Some semisynthetic derivatives of artemisinin (14 ).

LOOKING FOR THE PHARMACOPHORE

The 1,2,4-trioxane ring per se is or rather was a


relatively unknown chemical entity. Before 1957 none was
reported in the literature. Subsequently and especially since
the discovery of artemisinin, many methods for their
synthesis were devised [36]. As a result, numerous trioxanes
of simplified, but varied structure were prepared and screened
for in vitro activity against P. falciparum using the
Desjardins test [37, 38]. In this test concentrations of the
drug (expressed in ng/ml) which inhibit growth of the
parasite by 50 and 90% are estimated (IC50 and IC 90).

Tricyclic Trioxanes
On inspecting the artemisinin skeleton several questions
arise. How many and which of the four rings (A,B,C, and D)
are necessary for high activity? Is the boat conformation of
the trioxane ring really necessary? Would a peroxide do as
well? Is chirality important? What sort of peripheral
substituents are required to bolster activity? It should be
mentioned straightaway that the peroxide linkage in 14 is a
critical element, since its metabolite deoxyartemisinin (23)
having one O-atom less is without activity [34] Fig. (7).
Moreover, the lactone function is a double liability in that it
offers a site for hydrolytic degradation and also constitutes an
inherent therapeutic weakness as attested by the 8-fold greater
activity of deoxoartemisinin (24) [35].
O

Me
Me

Me

Me
Me

23

The synthesis and testing of many simpler tricyclic


trioxanes reveal that certain rings in 14 are redundant. The
high activity of the dimethyl derivative 25 shows that ring D
can be dispensed with altogether or rather that the spirocyclic
pentane ring serves in its stead Fig. (8). Clearly, the lactone
function in ring B is superfluous. Further evidence that the B
ring is unnecessary is attested by the high activity of the
endo-methoxy trioxanes 26 and 27. The pair of exo-methoxy
epimers 28 and 29 performs less well, some five-fold. In
both instances, the phenyl group at the bridgehead boosts
the activity over the methyl derivative. The peroxide 30 is
devoid of activity; so there are limits to what extent the
basic sesquiterpene skeleton can be modified [39-41].

Me

O O

24

Several other active, tricyclic trioxanes e.g. 31-33, have


been synthesized, tested, and confirm that the full tetracyclic

Fig. (7). Deoxyartemisinin and deoxoartemisinin.

O
B

Me

Me

MeO

OO

MeO

OO

31
Fig. (8). Some tricyclic trioxanes.

Me

MeO

30

28 R = Me
29 R = Ph

OTs
O
OO
32

Me

MeO

O Me
O O OMe
D

26 R = Me
27 R = Ph

OO

O
C

25

PhCH2 O

OH
O
OO
33

Me
Ph

Implications for the Mode of Action

Current Medicinal Chemistry, 2001, Vol. 8, No. 15

array of 14 is not required for high activity [42-44]. In the


foregoing examples where tricyclic portions of the
artemisinin architecture are mimicked, it could be argued
that artemisinin-like activity is retained, because only the
useless bits of the molecule were jettisoned. But as it will be
seen later, molecules far removed from artemisinin (14) in
shape and geometry can surpass it in activity. The same is
true for yingzhaosu A (13); its structure does not need to be
imitated in order for its effectiveness to be equalled.

[41, 45-46]. However, switching the parent ring from


dihydronaphthalene to cyclohexene brought about a change
in the right direction; the second set, 37-40, gave IC50
values going from 729 to 104 ng /ml. The antimalarial
activity, poor to start with, improved progressively on
attaching spirocyclic rings at the C3 position. A dramatic
jump was achieved in the case of the cyclopentene
homologues 41-44; their IC50 values against the W2 clone
ranged from 5.31 to 0.49 ng/ml, demonstrating
unequivocally that the carbocyclic framework of artemisinin
is simply not required for activity. It is worth mentioning by
way of reference that IC50 values for artemisinin are usually
about 1-2 ng/ml, so the preceding values are indicative of
worthy contenders.

Bicyclic Trioxanes
As the foregoing tricyclic trioxanes are not really so easy
to prepare, a program was undertaken to devise syntheses of
simpler bicyclic versions. For reasons of procedural
convenience, three types of bicyclic trioxane were prepared,
exemplified by the
cis-fused dihydronaphthalene,
cyclohexene and cyclopentene compounds 34, 37, and 41
Fig. (9). In spite of apparent structural similarities, the first
two, and their derivatives 35, 36, 38-40, created something
of a surprise by displaying little activity. The first set, 34-

R1

Me
Me

O O
O
R2

Unlike artemisinin, the cis-fused bicyclic structures of 4144 are flexible, permitting the trioxane ring to undergo
inversion between two chair conformations. Clearly, the cisfused, almost planar cyclopentene and the spirocyclopentane
rings are crucial features of the pharmacophore in these fully
synthetic analogues (v. infra). Another structural requirement
which proved of supreme importance for attaining high

Ph

Ph
Ph

O O
O

R1

R1

R2
R1 = R 2 = Me
R1, R2 = (CH 2)O(CH2)2
R1, R2 = (CH 2)5
R1, R2 = (CH 2)4

Ph

O O
O

R2

34 R1 = R 2 = Me
37
35 R1 = H, R2 = CH2 C6H4 OMe-4
38
1
2
36 R = H, R = (CH 2)6C 6H3Cl 2-3,4 39
40

1807

R1 = R 2 = Me
R1, R2 = (CH 2)O(CH2)2
R1, R2 = (CH 2)5
R1, R2 = (CH 2)4

41
42
43
44

Fig. (9). Some bicyclic trioxanes.

36, was essentially inactive in the in vitro screen against the


chloroquine-resistant W2 clone of P. falciparum, indicating
that the trioxane ring is necessary, but not sufficient in itself

activity in the 41 series is the presence of two alkyl


substituents at the C3 position and preferably spirocyclic,
spirocyclopentyl being the best. Monosubstituted methyl,
O

O
O

H HN

N
O
C6H 4F-p

O
O
C6H 4F-p
O
C6H 4F-p

OH
O

C6H 4Me
48

Fig. (10). Some synthetic trioxanes substituted at C5.

OH

O
O

C6H 4Me
49

C6H 4Me
47

C6H 4Me
O

C6H 4Me

46

C6H 4Me
O

C6H 4F-p

45
O

50
51
52

OR
C6H 4F-p

C6H 4F-p

R = CH2C 6H4CO2Na
R = CO(CH2 )2CO2Na
R = P(O)(ONa)2

1808

Current Medicinal Chemistry, 2001, Vol. 8, No. 15

Table 1.

Charles W. Jefford

In vitro and In vivo Activity of Peroxides 16 and 17 and Trioxanes 44 and 53

O
O
O
Me

CF 3

P. falciparum
K1 strain, ng/ml
ED50

CF 3

C6H4 F-p

Ph

Me

Me

O
O
Me

Me(CH 2) 10

16

Ph
O O rac.

C6 H4F-p
O O rac.

Controls

17

44

ART CLQ MFQ

53

27.6

31.1

P. berghei,
mg/kg/day x 4
ED50 p.o.
ED90 p.o.

10.4
18.0

5.9
7.4

25.0
75.0

2.5
6.0

5.0
14.0

1.9 1.3
2.3 1.8

ED50 s.c
ED90 s.c.

2.7
3.9

3.4
4.1

4.0
8.0

2.5
6.0

0.9
2.5

1.0 1.2
1.2 1.8

ethyl, propyl or t-butyl derivatives regardless


configuration at C3, are all poorly active in vitro.

9.51

of

As all the synthetic trioxanes discussed so far are racemic


mixtures, only one enantiomer being depicted for the sake of
clarity, it might be asked if their IC50 values should be
halved to double their potency to what it appears to be on
the assumption that only one of the enantiomers is the actual
therapeutic agent. However, it can be concluded with
confidence that in every case both enantiomers are equally
parasiticidal on the basis of the proof offered by the exo 5hydroxy enantiomers 54 and 55 Fig. (11). Each turned out
Me
Me

Me

O
O

Ph

O O
(+)

54

O
O

Ph

O O
O

Ph
OH

(-)

97.9 7.6

Chirality and Configuration

As yingzhaosu A (13) is a fugitive species semi-synthetic


derivatives have not been prepared. So arteflene (Ro 421611) (16) and Ro 41-3823 (17) have to serve as indices of
activity for the natural product [50]. It is therefore worth
noting that the wholly synthetic cyclopentene-trioxanes 44
and its p-fluorophenyl analogue 53 exhibit ED50 values of

OH
S Ph

4.6

9.5 and 5.6 against the chloroquine-resistant K1 strain of P.


falciparum, values which are superior to those of 27.6 and
31.1 ng/mls shown by 16 and 17 respectively (Table 1).
The p-fluorophenyl trioxane 53 also performs 2-4 times
better than 16 against P. berghei both at the ED 50 and ED90
levels by oral administration. A brief description of the in
vivo test protocol is given later (see Table 2). The in vitro
and in vivo results demonstrate that the yingzhaosu skeleton
does not have to be copied to confer high activity. It
demonstrates as well that 53 is a potential competitor to 16.

The activity of the best candidate 44 is also sensitive to


substitution in other parts of the ring. Replacing the phenyl
by tolyl or p-fluorophenyl substituents alters activity
slightly. Substitution on the double bond, e.g. 45, and
substitution at the C5 position in either the exo or endo
configuration with neutral groups, e.g. 46-49, leads to EC50
values of about 9-14 ng/ml, again for the W2 clone, which
are about 3-4 times less active than the parent cyclopent-5,6enes Fig. (10). When ionic water soluble groups are
appended at the C5 position, as illustrated by trioxanes 5052, activity disappears entirely [48, 49]. The conclusion is
that the configuration at C5 does not impinge strongly on
activity, whereas the ionic nature of the substituent does.

Ph

5.57

O O

55

Fig. (11). Pairs of enantiomeric and diastereomeric synthetic trioxanes.

Me
S

CO
O
Ph
S

(-)

O O
O
O

56

O
O S
Ph CO
O

Ph

(+)

57

Me
Me

Implications for the Mode of Action

Current Medicinal Chemistry, 2001, Vol. 8, No. 15

to have the same activity. Even the diastereomeric


camphanoate derivatives 56 and 57 have similar activities to
each other and also to 54 and 55, all compounds having
IC50 values lying between 1.5 and 2.4 ng/ml. In other
words, they are all indistinguishable from artemisinin in
their potency [51].
Further confirmation is tellingly provided by the in vivo
activities of the pure enantiomers of the p-fluorophenyl
derivative of cyclopentene-trioxane 58 and 59 when
compared to that of the racemic mixture 53 (Table 2) [52]. In
the Peters 4-day test against murine malaria, batches of mice
are initially infected with chloroquine-sensitive P. berghei N
or chloroquine-resistant P. yoelii NS lines respectively and
dosed each day for 4 days subcutaneously (sc) or orally (po)
with the test antimalarial [53]. On the 5th day the
parasitemia is read and the effective dose which suppresses
50% and 90% estimated. It is immediately seen that the
racemate 53, the (+) and (-)-enantiomers, 58 and 59, have
essentially identical ED50 values against the sensitive strain
in both the sc and po modes of administration (Table 2).
The same is true for the ED 90 values. A similar alignment of
ED50 and ED90 values is also evident for the resistant line. It
is also to be noted that the synthetic trioxanes (53, 58 and
59) are superior to artemisinin in the resistant line.
Moreover, 53 offers a bonus in being gametocytocidal and
sporontocidal [54].
The foregoing results from the two sets of
enantiomerically pure synthetic trioxanes convey an
unambiguous message. Configuration and chirality play no
role in the mode of action. The trioxane ring in the synthetic
bicyclic molecules have interconverting chair conformations,
so a fixed boat conformation is not a requirement.
Table 2.

Nonetheless, other molecular features are obviously


important and must intervene during the parasiticidal event.
However, before they can be identified and their significance
ascertained, the mode of action needs to be understood.

MODE OF ACTION
The above mentioned natural and synthetic trioxanes and
peroxides are chemically very different from the traditional
aminoalkyl substituted quinoline remedies such as quinine
and chloroquine. The former are neutral species whereas the
latter are bases. Both classes of drug are schizonticides acting
on the parasite within the infected erythrocyte, but they do
so by entirely different mechanisms. Curiously enough, the
receptor in both cases has been identified as heme. Since
heme is achiral it is understandable that the configuration of
an enantiomerically pure antimalarial has no bearing on
potency.
In order to be effective, an antimalarial must first of all
diffuse into the red blood cell and interact with heme. It is
now known from a large body of evidence that artemisinin
(14), its lactol derivatives 18-22 and peroxidic antimalarials
kill the malarial parasite in the blood of the host by a
stepwise process of chemical induction Fig. (6). During the
trophozote stage of the intraerythrocytic cycle the parasites
invade the red blood cells and ingest hemoglobin to provide
amino acids for growth. After proteolysis the spent prosthetic
group, heme, because of its solubility and toxicity to the
parasite, is eliminated immediately by oxidation and
polymerization to the insoluble malarial pigment, hemozoin
[55]. When the host is treated with quinine or chloroquine,
it seems that polymerization of heme is pre-empted by

In vivo Antimalarial Activity of Some Cyclopentenotrioxanes

C6H4 F-p
C6H4 F-p
RR + SS

(racemic)

R
O O
O

C6H 4F-p
R (+)

S
S
C6 H4F-p
O
O O ( _)

58

53

59

P. berghei N sc
mg/kg/day x 4 po

ED50
2.5
2.5

ED90
6.0
6.0

ED50
2.1
2.6

ED90
3.6
4.8

ED50
1.8
2.1

ED90
3.2
3.6

P. yoelii NS sc
mg/kg/day x 4 po

4.5
5.6

7.6
10.0

1.8
1.4

3.4
5.0

1.5
1.1

2.8
3.1

Quinine

Chloroquine
P. berghei N
P. yoelii NS

ED50
sc 1.8
sc 2.4

ED90
3.1
56

1809

ED50
65
128

ED90
170
290

Artemisinin
ED50
0.9

ED90
2.3

5.8

10.0

1810

Current Medicinal Chemistry, 2001, Vol. 8, No. 15

Charles W. Jefford

O
O

HO 2C
HO 2C

53

Me

O
Ar

Ar
O
62 Ar

Ar

Me

Me
Fe

N
N

HO 2C
HO 2C

Me

Me

Me

NO N
N Fe N

Me

Me
60

61
O
64

63

Fig. (12). Oxygen atom transfer from a trioxane to heme, and then to cyclohexene.

chelation and its toxic effect potentiated so that the parasite


is killed [56-58]. During the mortal event chloroquine
remains chemically unchanged neither reacting with nor
killing the parasite. Instead, it lets heme do the job.
Treatment of the host with an antimalarial peroxide also
interrupts the aforementioned detoxification process. But it
does it in a different way from the quinoline-type drugs. It
closely coordinates with heme and then reacts with it to
produce a lethal agent.

dihydroartemisinin, and arteflene to P. falciparum-infected


erythrocytes resulted in the transfer of label to six malarial
proteins [64]. It was therefore concluded that in some
undefined manner, not necessarily involved with parasite
death, reaction with specific malarial proteins had occurred.
Nonetheless, signs were seen that the parasiticidal process
somehow involved transient radicals. Gradually, by devising
model experiments with simpler peroxides and 1,2,4trioxanes, the nature of the lethal agent and the mechanistic
details of the mode of action became clearer and more
complete.

The nature of the latter, and how it forms took some time
to be delineated. Experiments performed with hemin and
artemisinin confirmed their interaction to form an unknown
adduct after generating an unidentified oxy radical [59-62].
Artemisinin was thought to alkylate heme and parasite
proteins, but no structures were proposed [63]. It was also
discovered that the administration of radiolabeled -arteether,
O
O

In recent years numerous model experiments have been


performed and even today some of the results and
interpretations presented as the final word are open to
question as they are at variance with the canons of
contemporary organic chemistry.

FeCl2 .4H2 O
(0.35 equiv)
MeCN, 22 o, 2h

Ar

R(CH2) 4

Ar
53 Ar = C6 H4F-p

Ar
65 R = Cl 66%
66 R = OH 8%
67 R = H 2%

2+
Fe
_
Cl, H2O

3+
Fe
O
O Ar
73

+
H

O
Ar

2+
Fe
3+
Fe

O_+
O

Ar
Ar

Ar

O
O

Fig. (13). Ferrous ion-induced rearrangement of trioxane 53 to propionates.

O Ar
77
Ar

Ar

74

Ar

O_

75

. O_ O

Ar

O
HO

. O_

Ar
76

Implications for the Mode of Action

Current Medicinal Chemistry, 2001, Vol. 8, No. 15

Are Antiparasitic Trioxanes Oxygen Atom Donors?


Originally it was thought, erroneously as it transpired,
that racemic synthetic trioxanes such as 53 and artemisinin
(14), as they appear to be loaded with excess oxygen, might
behave as oxygen atom transfer agents. The idea was fostered
by the fact that deoxyartemisinin (23) was the main
metabolite and that heme might behave like a P450-type
monooxygenase [65, 66]. It was postulated that active
trioxanes would first transfer an oxygen atom to heme (60) to
form the corresponding iron-oxene intermediate (61) which
in turn would destroy a neighboring parasite by
monooxygenating its protein Fig. (12). Heme might be
expected to convert 14 into 23 during the parasiticidal event.
Similarly, 53 would give the dioxolane 62. The idea was
tested by using ferrous chloride tetrahydrate in acetonitrile as
a model for heme. As a simple chemical model for the
parasite cyclohexene (63) was chosen as it should be easily
epoxidized to 64 by the ferryl equivalent of 61. Actual
treatment of 53 and 14 with ferrous chloride caused rapid
unraveling of the trioxane rings [67, 68] Fig. (13). Three
products were formed from 53; the cyclopentenyl esters of chloropentanoic (65), -hydroxypentanoic (66) and pentanoic
acids (67) in yields of 66, 8 and 2% respectively [62]. The
racemic diphenyl cyclopentene-trioxane 44 under the same
conditions gave analogous results. Repetition of both
experiments in the presence of cyclohexene (63) led to
basically the same result. Neither dioxolane 62 nor epoxide
64 was formed.
In the case of artemisinin (14), only isomerization
occurred [68]. Exposure to FeCl2 in MeCN for no more than
O

Me

O-O

Me

14
Fe

Me
FeCl 2.4H2 O, MeCN Me
25o , 5-15 min
Me
without cyclohexene
cyclohexene (1.18 equiv.)

2+

O O

Fe
O O

Me

Me

C-C scission

15 min gave furan-acetate 68 and the hydroxydeoxyartemisinin 69 in 78 and 17% yield Fig. (14).
Repeating the experiment in the presence of cyclohexene (63)
merely altered the yield of the two products to 84 and 1-8%
respectively. No traces of 64 or deoxyartemisinin (23) were
detected. -Artemether (19) gave a similar result [68] Fig.
(15). It reacted in 5 min. isomerizing to the corresponding
furan-acetate 70, the hydroxy-deoxyartemsinin derivative 71,
and an epimeric mixture of formyl diketones 72 in yields of
32, 23, and 16%. Running the reaction in cyclohexene
improved the yields of the same three products to 36, 30 and
19%.
The Intermediacy of Carbon Radicals
The foregoing product compositions are strongly
indicative of the intermediacy of radicals in their formation.
In fact, there are many precedents for the redox behavior of
ferrous ion towards cyclic peroxides which induces
decomposition by forming radicals [69-72]. In the present
instance, ferrous ion functions in the customary manner
complexing with the O-O bond of 53 with concomitant
single electron transfer Fig. (13). Scission of the resulting
oxy radical 73 is driven by the formation of the ester function
creating simultaneously the terminal pentanoate radical 74.
Recuperation of the electron by ferric ion furnishes 75, which
probably for reasons of entropy is unable to cyclize to the
lactone 76. Instead, ambient chloride ion and to a lesser
degree water, attack 75 giving the chloro and hydroxy
products 65 and 66. Overall, the conversion of 53 to 65 and
66 can be considered as an isomerization with adjunction of
O

O
O

O
Me

2+

. Fe
OO
3

Me

H
79

2+

Me

Fe
O
OH

Me

81

Me

1,5 shift

Me
2+
Fe

O
69 OH
17%
1-8%

Me

Me

O 2+
O Fe
80

Me

Me

OO

Me

68

Me H 78

Me

78%
84%

2+

1811

2+
O=Fe O

Me
Me

OH

Me

82

Fig. (14). Ferrous ion-induced isomerization of artemisinin via radical intermediates.

Me
2+
O=Fe

O O
Me

Me
Me

OH
O
83
Fe 2+

1812

Current Medicinal Chemistry, 2001, Vol. 8, No. 15

MeO
Me
Me

Charles W. Jefford

FeCl2.4H2O MeO
Me MeCN
Me
25o , 5 min

O-O

Me

19
2+
Fe

MeO

Me
Me

Fe2+
97

Me

Me

.
OO

Me
Me

Fe

Me

_
+
MeO, CO, H
O O

Me

Fe2+

Me

98

72
16%
19%

23%
30%

2+ MeO

CHO
O

Me

OH

Me 71

MeO

Me

Fe 2+
O

Me

Me

without cyclohexene
32%
cyclohexene (1.18 equiv.) 36%

O
OO

Me MeO

70

Fe 2+
O

O
O

Me

Me
Me

99

Me

100

Fig. (15). Ferrous ion-induced rearrangement of -artemether.

a molecule of hydrogen chloride or water. The origin of the


pentanoate 67 can be attributed to the abstraction of a
hydrogen atom by 74 from the solvent.
Treatment of 53 with ferrous bromide in tetrahydrofuran
(THF) as solvent gave mainly the bromo analogue of 65 [73,
74] Fig. (13). No dioxolane 62 was observed. Indirect
confirmation of the radical 74 was obtained by carrying out
the same experiment with ferrous sulfate and cupric acetate in
methanol. The terminal olefin 77 was the exclusive product.
As soon as 74 forms it is oxidized by cupric ion to the
corresponding masked primary cation 75 which promptly
eliminates a proton.
In just the same way as before ferrous ion complexes with
the peroxide bond of artemisinin (14) Fig. (14). Although
not specified in the preceding example, the ferrous ion after
single electron transfer forms a covalent bond with the
oxygen anion of the sundered peroxide bond giving a pair of
equilibrating ferric monodentate oxygen radicals 78 and 79
which are formed in different proportions. They evolve
differently too. The main course is followed by 78 which
cleaves to the pendent primary ethyl radical 80; the driving
force being the acquisition of thermodynamic stability by
formation of the acetate group. Expulsion of the contiguous
ferric ion as ferrous ion and union of the two radical centers
creates the tetrahydrofuran 68.
The evolution of the minor oxy radical 79 was originally
puzzling Fig. (14). A 1,5 hydrogen atom shift to produce the

O
78

Me

secondary C-centered radical 81 was proposed [75].


Thereafter, the cyclic enol ether 82 was supposed to arise by
the excision of O=Fe2+ [76]. The latter, if not dispersed or
destroyed, could then react again with 82 to afford the
epoxide 83. Alternatively, 83 could be formed directly by
loss of ferrous ion from 81. Finally, opening of the epoxide
83 by internal nucleophilic attack by the tertiary hydroxyl
group accounts for the formation of the minor product 69
[77]. The intermediacy of O=Fe2+ may have been invented
to account for deoxyartemisinin (23), the human metabolite
of 14, which is formed in addition to 68 and 69 when
artemisinin was exposed to ferrous bromide in THF (v.
infra). Although C-centered radicals like 78 and 79 had been
previously invoked [75] as possible agents responsible for
the high antimalarial activity of the tricyclic tosylate (32),
the later predilection for O=Fe2+ as the toxic agent may have
been influenced by the large body of research carried out with
P450-monooxygenases.
The 1,5 H Shift
Despite their plausibility, some of the foregoing steps
seemed unsatisfactory and deserve comment. The 1,5
hydrogen transfer, although favored over 1,3 and 1,4
transfers, must meet certain geometric criteria for it to occur
[78]. It appeared from an examination of the calculated and
X-ray structures of 14 taken as a model for the Fe3+ -bound
complex that the interatomic distance between H-C(3) and
O-C(6) in 79 is 2.478-2.803 [68]; a distance greater than

O
2+
Fe
OO

O
Me

Me

O O
2+
Fe
O

H
Me

84

Me

Fig. (16). Incorrect rearrangement of oxy radical 78 via 1,3 and 1,2 shifts.

2+
Fe

OH
85

Me
69

Implications for the Mode of Action

Current Medicinal Chemistry, 2001, Vol. 8, No. 15

the critical distance of 2.1 above which migration is


thought not to be possible [79]. The distance between HC(3) and O-C(4) in 78 was less (2.467-2.560 ), but still
exceeded 2.1 . As there was no clear-cut preference for 1,5
over 1,3-H shift, the latter was favored for plotting a path
from 78 to 69 Fig. (16). Consequently, a 1,3-H shift
followed by a 1,2 shift of a hydroxyl group (78 84 85)
were proposed to give the tertiary radical which then was
supposed to close to 69 by loss of ferrous ion.

stable than its acetal precursor 89 by 12.2 kcal/mol. These


calculations of course strictly apply to the artificial oxy
radicals 87 and 89, but nonetheless they are relevant to the
1,5 H shift 79 81 and the C-C scission 78 80. The
conclusion is ineluctable. Converting an oxy radical to a Ccentered radical by shifting a H-atom or cleaving a C-C bond
is going to be easy. More importantly, the C-centered
radicals 80 and 81 derived from artemisinin, once formed,
will be unable to regress to their oxy radical predecessors
because the energy of activation in the back reaction is now
far higher than it was in the forward reaction. In other words,
these C-centered radicals are kinetic products and their
formation is irreversible.

Unfortunately, the preceding path is the wrong one as


attested by the subsequent isolation of epoxide 83 and the
trapping of the secondary radical 81 (v. infra). The findings
of a recent study using density functional theory also
provided a mechanistic corrective [80]. Taking 6,7,8trioxybicyclo[3.2.2]nonane (86) as the model for 14, formal
addition of a hydrogen atom affords the pair of oxy radicals
87 and 89 Fig. (17). Calculation confirmed that the
interatomic distance between the oxygen radical and the
contiguously oriented 1,5 disposed hydrogen atom in the
ground state of 87 is 2.340 , a distance definitely greater
than the critical value of 2.1 . However, the activation
energy for 1,5 H-transfer to give the secondary carbon radical
O
O
O

A further indication that ester formation is the driving


force for producing an active primary C-centered was
provided by calculations on the 1-methoxycyclopentyl-1oxyl radical (91) and its scission product the -radical (92)
[52]. Their geometries were optimized by semi-empirical
unrestricted Hartree-Fock calculations according to the PM3
method Fig. (18). The respective heats of formation, -59.5
and -76.6 kcal/mol, confirm that the conversion 91 92 is
strongly exothermic by 17.1 kcal/mol. A precedent for the

OH

.O

HO

87
O

OH

1,5 shift

86

1813

88
O.

HO

C-C scission

HO
89

90

Fig. (17). Rearrangement and cleavage of oxy radicals derived from 86 .

88 turned out to be low, only 6.4 kcal/mol. Moreover, the


geometry of the reacting atoms in the transition state leading
to 88 revealed that a collinear arrangement is not attainable,
which means that it is evidently not a requirement. The
carbon radical 88 was found to be more stable than the oxy
radical 87 by 4.5 kcal/mol.
Carbon Radicals as Kinetic Products
The evolution of the acetal-type radical 89 was similarly
favored. The activation energy for its cleavage to the formyl
primary carbon radical 90 was computed to be 7.8 kcal/mol.
Again, the C-centered radical 90 was predicted to be more

In similar fashion, the hydroxy oxy radical 93 derived by


adding a hydrogen atom to artemisinin serves as a model for
both ferric derivatives 78 and 79 Fig. (19). The heats of
formation for 93 and its rearranged products the primary and
secondary radicals 94 and 95 were found to be -199.4,
-216.3, and 219.3 kcal/mol respectively [68]. In other
words, the act of creating the primary and secondary radicals
releases 16.9 and 19.9 kcal/mol. Thus, both the 1,5 shift
and C-C scission are strongly exothermic processes. By
extrapolation, the rearrangements 78 80 and 79 81

H
O

H H

foregoing hypothetical reaction is the ready cleavage of 1methylcyclopentyl hydroperoxide by ferrous ion to hexa-5one-1-yl radical [81].

H O

H
H

H
91
Fig. (18). Cleavage of the methoxycyclopentyloxyl radical.

H H HH
H

.
H

O
H H O
92

H
HH

1814

Current Medicinal Chemistry, 2001, Vol. 8, No. 15

O
Me
Me

Charles W. Jefford

OH O

O O

Me

Me

O
Me
Me 94

93

OH

Me
Me

HO
OH

Me

95

Fig. (19). Cleavage and rearrangement of the hypothetical oxy radical from artemisinin.

should be exothermic to the same degree and therefore


irreversible. These conclusions, as it will be seen later, have
special relevance to the mode of action of peroxidic
antimalarials.
Is O=Fe 2+ an Intermediate?
No direct evidence has been reported for the existence of
82, while that in support of O=Fe2+ is circumstantial. The
experiments, the rationalization of which led to the proposal
of such exotic species, were conducted by subjecting
artemisinin (14) to ferrous bromide in THF alone or in the
presence of various easily oxidizable or aromatizable addends
as possible mechanistic witnesses [27, 76]. Under these
conditions, 68 and 69 were formed as before in a similar
ratio, but in diminished yields of 29 and 10%, the balance of
material being deoxyartemisinin (23) (59%). The proportion
of 23 went up on adding progressively more 1,4cyclohexadiene. Hexamethyl Dewar benzene (HMDB) also
boosted production of 23, becoming partially aromatized to
hexamethylbenzene in the process. Consequently, it was
inferred that 23 arose by protonation and intramolecular
closure of the putative cyclic enol ether 82, increasingly
furnished by loss of O=Fe2+ from 81 Fig. (14). A
supplementary avenue to 23 was assumed to issue from 81
by abstraction of hydrogen from 1,4-cyclohexadiene.
However, 23 could have arisen by a less exotic and
mechanistically more likely route. The reduction of
artemisinin (14) or perhaps, more appropriately, the ferric
oxy radicals 78 and 79 by abstraction of hydrogen from THF
or added 1,4-cyclohexadiene without a hydrogenation
catalyst (none is needed) to give the dihydroxy derivative 96
followed by dehydration to 23 offers a reasonable explanation
Fig. (20). Ferrous ion or a transient radical, rather than
O=Fe2+ , may well have catalyzed the aromatization of
HMDB. Other addends such as methyl phenyl sulfide and
tetralin, the oxygenation of which was presented as proof for
O=Fe2+ , may have abstracted oxygen directly from 14 or
could have been oxidized adventitiously. The recent

O
14 or 78

Me
Me

O
HO
OH

H2 O
Me
23

96

Fig. (20). Reduction of artemisinin or its oxy radical 78 to the


diol 96 and deoxyartemisinin 23 .

observation that large amounts of diol are formed when


simple bicyclic endoperoxides are treated with FeBr2/THF
attests to its reductive nature [82]. Finally, it is difficult to
explain why O=Fe2+ , a non-selective oxidant, if really
present, which has no trouble in epoxidizing the enol ether
82, fails to react with admixed 1,4-cyclohexadiene to give its
epoxide.
Lastly, it should be mentioned that the simple extrusion
and transfer of an oxygen atom from a peroxide to a ferrous
cation might be mechanistically difficult. In order to be an
effective oxidant, the high valent iron-oxene species may
need to be perferryl, the species formed in the P450 enzyme
catalytic cycle, rather than ferryl [83]. If this be the case, the
second step after coordination with ferrous ion, the rupture of
the FeO-R bond to produce the perferryl species O=Fe3+
requires the departure of R as an anion, not a radical, which
is more costly in energy and less likely.
In view of the inertness of cyclohexene in the FeCl2induced reactions with 53 and 14 together with the nonformation of deoxyartemisinin (23) or the dioxolane (62), the
original proposal that only C-centered radicals are involved
is undoubtedly correct and does not need to be modified to
fit a P450 mono-oxygenase mechanism in spite of its
attractiveness [84]. The FeCl2-induced reaction of artemether (19) with and without cyclohexene follows the
same course and can be similarly interpreted Fig. (15).
Deoxy--artemether is not formed. The furan-acetate 70 and
hydroxy-deoxyartemisinin 71 arise in the standard way via
cleavage and rearrangement of the pair of ferric oxy radicals
97 and 98. The third product the mixture of diketones 72
must have arisen by the formal loss of a molecule of methyl
formate. How this happens is explicable in terms of the
elimination of ferrous ion in a counterclockwise or clockwise
radical fragmentation of 97 or 98. The resulting
methoxymethyl formate 99 then disintegrates by loss of a
proton, carbon monoxide, and methoxide ion, giving the
penultimate diketo-aldehyde 100, which finally isomerizes to
72.
The final piece in the mechanistic jigsaw puzzle was the
isolation of the unstable epoxide 83, albeit in a tiny amount
(1%), from the reaction mixture obtained by treating 14 with
one equivalent of ferrous sulfate in aqueous acetonitrile. This
valuable piece of information confirms the sequence 79
81 83 69 [85, 86]. However, it must be noted that the
key experiment of actually isomerizing 83 to 69 appears not
to have been done or at least reported. Proof for a secondary
carbon radical, presumably 81, was secured by its capture
with 2-methyl-2-nitrosopropane and the identification of the
resulting nitroso radical by its ESR spectrum. Why the

Implications for the Mode of Action

Current Medicinal Chemistry, 2001, Vol. 8, No. 15

O O

Ph

Me

Ph
O

103
Me

O
O
OH

Me

Me Ph
Ph

O
OH

Me

Ph

Ph

Me

+
N Mn
N
N
N

Ph Ph

102

O
OH

Me

Ph
101

Me

Me Ph

N Mn N
N
N

Ph Ph

Ph

Me

N MnN
N
N

Ph

Me

Me

Me

OO

Me

1815

Me Ph
+

N MnN
N
N

Ph

Ph

Ph

NH N
HN

Ph

Ph
104

105

Fig. (21). Alkylation of Mn(II)TPP by artemisinin.

more active primary radical 80 was not trapped as well or


instead of 81 was not commented on.
By recourse to a hydrophobic model which parallels the
biological event, primary carbon radicals formed from
artemisinin (14) could be trapped [87]. Instead of heme,
which
is
notoriously
unstable,
managanous
tetraphenylporhyrin (Mn(II)TPP), generated in situ by the
interaction of Mn(III)TPPPOAc with tetrabutylammonium
borohydride, was allowed to react with 14 Fig. (21). Single
electron transfer occurred in the usual way by breaking the OO bond. The resulting manganese(III) derivative 101 cleaved
to the primary radical 102. Thereafter, intramolecular
addition to one of the nearby pyrrole rings affords the pyrrole
radical 103, then the cation 104 by internal transfer of an
electron. Reduction by borohydride to the dihydropyrrole
and demetallation afforded the covalent adduct 105. By
applying the same procedure, the primary radicals from artemether (19) and the cyclopentene-trioxane 53 were
similarly produced and captured by the pyrrole ring [88, 89].
It is reasonable to assume that in the biological context
too the decomposition of artemisinin (14) and -artemether

MeO
Me

O-O

Me

Zn
Me

MeO
Me

OO
Zn
O O

Fig. (22). Deoxygenation of -artemether with zinc.

MeO
Me
Me
Me

Me
19

(19) proceeds via primary and secondary carbon radicals.


Metabolic studies with 14 and -arteether confirm the
formation of products like 70 and 71 as well as
deoxyartemisin (23) and its ethoxy derivative [90]. The
origin of the deoxy compounds could be due to enzymatic
deoxygenation which has nothing to do with the
parasiticidal event. For example, an in vitro study of the
metabolism of -arteether in rat liver cytosol revealed that
deoxy--artether was formed directly under the influence of an
NADH-dependent cytosolic enzyme [91]. As a chemical
equivalent of reduction by a Zn-containing NADH
dehydrogenase zinc dissolving in acetic acid was taken as an
appropriate reagent [92]. Adding an equivalent of Zn powder
to a solution of 14 in AcOH with stirring at room
temperature gave after a few hours a quantitative yield of
deoxyartemisinin (23) [68]. Under the same conditions 19
gave deoxy--artemether (108) as the sole product in 68%
yield Fig. (22). These results indicate that deoxygenation
involves a two-electron reduction, which is a characteristic of
zinc, but not of ferrous ion. First, Zn donates two electrons
to the O-O bond of 19 to form the bidentate intermediate
106. The latter thanks to the oxophilic nature of zinc
rearranges to 107. Finally, deoxyartemether (108) is formed

106

Zn=O

Me

O
_

Zn

MeO

Me
Me

O
Me

107

108

1816

Current Medicinal Chemistry, 2001, Vol. 8, No. 15

Table 3.

Charles W. Jefford

Product Yields (%) Obtained by Exposure of Artemisinin (14) to Fe(II) Reagents

Entry

Fe(II) reagent

Reaction speed

Furan 68

Pyran 69

Deoxo 23

Other 83, 109

Ref.

FeBr2 , THF

15 min

29

10

59

[76]

FeBr2 , THF, CHD 40%

15 min

17

71

[76]

FeCl2 .4H2 O, MeCN

5 min

77

11

[76]

FeCl2 .4H2 O, MeCN

5 min

78

17

[68]

FeCl2 .4H2 O, MeCN, CH

5 min

84

[68]

FeCl2 , imidazole, MeCN

5 min

78

16

[109]

Hemin, PhCH2 SH, THF

15-40 min

63

5.5

1.0

[76]

FeSO4 , H2 O, MeCN

hours

25

78

1-2

[86]

by extrusion of ZnO which dissolves in AcOH giving


Zn(OAc)2.
Comments on the Proposed
Framework

Unified

likely that 23, arising by reduction, stems from a reaction


which is separate from that of the catalyzed isomerization to
the furan-acetate 68 and the hydroxypyran 69. Consequently,
a doubt exists about the validity of any mechanism that is
partly based on 23. A scheme to be valid, and indeed to be
unified, can only take into account those products that share
a common mechanism.

Mechanistic

Having in hand sufficient proof for the intermediacy of the


epoxide 83, the primary and secondary radicals 80 and 81
together with their formation as kinetic products, and an
absence of proof for the cyclic enol ether 82 and the ferryl
species O=Fe 2+ , all the pieces are in place for defining the
ultimate mechanism for the ferrous ion-cleavage of
artemisinin and its congeners. A recent attempt, while
ambitious and somewhat audacious, falls short of this goal
[86]. A so-called unified mechanistic framework was
formulated on the basis of an imperfect understanding of the
data and some of the principles governing organic reactions.
The data taken for consideration, although disparate, were
quite limited (Table 3). With such a small sample a reliable
interpretation is already compromised. It is immediately
seen that deoxyartemisinin (23) is only produced in THF
and also to a very minor degree when imidazole was present
(entries 1, 2, 6, and 7). As mentioned earlier, it seems very
O

O O

The reaction speeds (they cannot be called rates), namely,


the time required to complete the reaction, are very rough
estimates (Table 3). All that can be said is that FeSO4 (entry
8) is much slower than the rest, which are more or less
equally rapid in their action. Its slowness differentiates it
from the other reagents by permitting other products to be
identified, notably the epoxide 83 and one of its
rearrangement products (109). As the formation of 109 was
only alluded to, a word on how it arises is appropriate. The
isomerization of 83 to hydroxypyran 69 was correctly
detailed Fig. (23). Protonation of 83 to 110 followed by
opening of the epoxide transfers positive charge to the acetal
function as indicated by the species 111. Annihilation of
charge on the oxonium ion by internal attack of the hydroxyl
group then gives 69. An alternative course is to open the
protonated epoxide 110 to form a five-membered ring. Attack

O O

Me
Me

OH

Me

OH

Me

Me

O O
H

O
Me
Me
O
H
+

110

Me

111

O H

Me

Me

69

OH

110

Me

OH

Me
O H

83

Me

Me
Me

O +
O

O
Me

OR O
O

Me

Me
112

Fig. (23). Opening of the epoxide 83 to give either a six or five-membered ring product.

109 R = H
113 R = Ac

Implications for the Mode of Action

Current Medicinal Chemistry, 2001, Vol. 8, No. 15

by the hydroxyl group on the least substituted terminus of


the epoxide entity leads to the furan 112. Deprotonation and
cleavage gives 109, one of the observed decomposition
products which was subsequently identified as its acetate. In
fact, this dichotomy of epoxide opening is a characteristic
feature in the rearrangement of artemisinins and depends on
the nature of the adjacent ring atoms (v. infra).
The main mechanism proposed incorporates as its core
the pair of equilibrating oxy radicals and 78 and 79
(produced from 14) which are assumed to be in equilibrium
with their derived carbon radicals 80 and 81 Fig. (24). It
should be remarked at the outset that the subsidiary
equilibrating oxy radicals 114 and 115 shown as arising
from 78 and the conversion of 114 to 116 are purely
hypothetical in the case of artemisinin as no products formed
from them were observed. They are meant to serve as
reference structures for reactions of derivatives. The core
equilibrium is supposed to be displaced to favor different
amounts of the products 68, 69, and 23 depending on the
kind of ferrous reagent and solvent employed. Two factors
were invoked to account for the product composition. The
better described is the solvent factor. The basic idea is that
the preferential breakage of one over the other of the two
bonds in Fe-O-R as in 80 and 81 is susceptible to the nature
of the solvent. It is suggested that the Fe-O bond is stronger
in THF than in MeCN, therefore favoring homolysis of the
O-R bond, whereas the reverse is true in aqueous MeCN.
This particular solvent effect was used to explain why 23 is
formed as the major product in THF, but not observed in
MeCN (Table 3, entry 1). Such an explanation is difficult to
accept since radical reactions are generally insensitive to
solvent polarity [93].
The second factor is the nature of the counter ion in the
ferrous reagent which is assumed to affect the ability of
ferrous ion to deliver an electron to the * orbital of the O-O
O

OO

Me
Me
O

O
Me
Me

O Fe2+

C-C

O
O

Me

Me
Fe2+

O
O

116

68

Me

O O
2+

Fe
O O

O
Me

The trouble with the preceding interpretations is that


they are based on incorrect premises and fuzzy factors.
Intermediates 80 and 81 are stable carbon radicals evincing
little tendency to revert to the oxy radicals 78 and 79.
Deoxyartemisinin (23) is the product of a separate reaction. It
should be pointed out too that trying to rationalize product
ratios like 2.7:1 and 1:3 (cf. Table 3, entries 4 and 8), which
reflect transition states not differing greatly in energy, is not
a reliable undertaking.
The above mechanism was also deemed to be applicable
to certain artemisinin derivatives. To simplify discussion,
just two are considered here. -Artemether (19) and its benzyloxy analogue on treatment with FeSO4 in aqueous
MeCN gave derivatives analogous to 68 and 69 in ratios of 1
to 1.2 and 1.8 to 1 respectively. The switch in ratios was
attributed to slower 1,5 shifts due to conformational changes
caused by the "larger" benzyloxy substituent. In reality the

Me

O
Me

Me
Me

O OO
2+
Fe O
O
Me

115

2+

Fe
OO

O
Me
1,5 shift Me

79
O=Fe

114

Me

Me

2+

Me

Me

.
O

Fe O
O

bond. It is argued, without substantiation, that FeSO4,


being less active than FeBr 2 (Table 3, entries 8 and 1) slows
down "the steps of" (the rate of conversion of?) 81 to 69 and
of 80 to 68, but aq. MeCN "works the opposite way"
(speeds them up?). Put another way, it can be supposed that
this means that the ratio of the equilibrating species 81 to 80
is about the same (3:1) when produced by either FeSO4 or
FeBr2 in their respective solvents. Decomposition of 80 just
gives 68 while 81 bifurcates to 23 and 69 in THF, but goes
completely to 69 in MeCN. When FeCl2.4H2O in MeCN is
used as reagent (entry 4), no 23 is formed as the solvent does
not facilitate Fe-OR bond cleavage. The ratio of 68 to 69
now changes to 4.6:1. Apparently in this case the effects of
solvent and reagent do not cancel out which means that the
equilibrium shifts away from 81 towards 80 because as "the
1,5 shift is not so fast" 79 gives 81 less quickly. The major
and minor radicals 80 and 81 then react giving 68 and 69 in
the aforementioned ratio.

78

Me

Fe2+
O

scission

80

Me
Me

Me

Me

1817

Me

Me

OH
O

82
O

Me

Me

81
2+
Fe

OH

2+

Fe
O
OH

Me

2+

2+
Fe

83

Me
O

Me

O
Me

Me

Me

Me

23

Me

Fig. (24). The main mechanism for the Fe(II)-induced reaction of artemisinin and its congeners (ref. [86]).

O
OH
69

1818

Current Medicinal Chemistry, 2001, Vol. 8, No. 15

Charles W. Jefford

difference in steric compression on the rigid tetracyclic


skeleton between such remote substituents must be
negligible.

from 127 which undergoes 1,5 H shift to 128. Even though


it is in THF, a solvent alleged to favor excision of O=Fe2+ ,
128 fails inexplicably, if the unified mechanism is to be
believed, to react in this sense and loses Fe2+ instead
forming the epoxide 129. The next step proposed is farfetched as it entails opening of the epoxide with FeBr2 to
furnish 130. In order to get the groups in the right positions,
the two ligands are now obliged to exchange with each other
giving 131. The goal, the hydroxy furan 125, is finally
reached by nucleophilic displacement of bromide ion by the
oxide substituent. Of course, this tortuous route is not
followed at all, because there is a rational shortcut.
Protonation of the epoxide 129 will engender backside attack
by the hydroxyl group to form the furan ring 125 directly.
This is the expected closure. Closure to the six-membered
ring only occurs when positive charge is formally located at
the methylated terminus by an acetal function, e.g. 83
110 111 69 Fig. (23).

Arguments have been mustered to provide purportedly


correct mechanisms for the reactions of certain carba
artemisinins [94]. The action of FeBr2/THF on
deoxoartemisinin (24) produced the completely unraveled
formyl diketone (117) and the deoxodeoxyartemisinin (118)
in yields of 79 and 8% Fig. (25). The first step is the setting
up of the pair of equilibrating oxy radicals 119 and 120.
Thereafter, their evolution was ascribed to the differential
effect of THF. Cyclization 119 122 was retarded, allowing
cleavage 120 123, which is favored, to take over. Closure
of the hydroxyl group onto the double bond in 123 would
afford 118 in the postulated way. Fragmentation of 119 by
circular movement of four electrons brings about cleavage to
117. It was not made clear why so little of the deoxygenated
product was formed compared with the case of artemisinin. A
simpler explanation is that 120 fragments just as well as 119
which necessarily bypasses the potential excision of O=Fe2+ ,
while deoxygenation to 118 is probably the consequence of
an independent reduction process arising from 119 and 120
through the diol 121.

It is to be noted that the hypothetical equilibrium


between 78, 114 and 115 is unobserved. The path from 78
to 114 and then to 116 only becomes a reality when the
carbonyl is absent, for example in the case of -artemether
(19) and deoxoartemisinin 24. Rather than halt at the 114
stage, a likelier course, as suggested above, is fully concerted
cyclic fragmentation to the analogues of 116.

The dideoxoartemisinin (124) seemed to present


difficulties of interpretation even though the result was
simple enough Fig. (26). With the same reagents as before
just a single product was obtained, the hydroxy furan 125.
The mechanism proposed [86] starts by adjunction of ferrous
ion giving the pair of ferric oxy radicals 126 and 127. This
time there is no avenue of cleavage open to 126 because
thermodynamic stabilization cannot accrue by creation of a
carbon-carbon double bond. The reaction proceeds entirely
O

Me

O-O

Me

24
Fe

In conclusion, it appears on reviewing the selection of


results (Table 3), that, apart from the presence of the odd
product 23, there is not much to choose between them.
Thus, the expansion of the mechanistic scheme to embrace
the controversial formation of 23 is superfluous. To get a
better notion of the scope of the mechanism more results are
needed. New experiments need to be done and old

Me
FeBr 2, THF
25o , 5 min

Me

117

2+
Fe

O O
Me

Me

Me

Me

122

118
H2 O

Me

O O
HO
OH

2+

Fe

Me

Me
121

Me
O

Me

Me

120

119
O

Me

2+
Fe

.
OO

Me

Fe 2+

Me

2+

O
Me

Me

Me

.
OO

O
O

O
Me

OH

Me

123

Fig. (25). Ferrous ion-induced rearrangement of deoxoartemisinin 24 .

Me

Me

Implications for the Mode of Action

Current Medicinal Chemistry, 2001, Vol. 8, No. 15

O
Me
FeBr 2, THF

O-O

Me

1819

Me
Me

Me

125

OH

25 , 5 min
124

Me
2+

Fe

.O

Me

.
OO

Me

Me

2+

Me

Me

Me 127

Me

Fe

Me

OFeBr

Br

Me

OH

O H
129

FeBr2
125

Br

Me

130

Me
Me

FeBr
O

Me
+

128
O

OH

Me

Fe

Me

Me

H
O

2+

O
FeBr2
Me

Me

OH
O

2+

Fe
126

129

131

Fig. (26). Rearrangement of dideoxoartemisinin by FeBr2 . Mechanistic avenues.

experiments carefully repeated with a variety of pure solvents


and pure redox couples, not only with Fe(II), but also with
Mn(II), Ru(II), and Ti(II) reagents, to secure incontrovertible
results before proposing a unified mechanistic framework.

It is now clear that artemisinin gives with ferrous


reagents and presumably with heme, two types of radical,
primary and secondary. An obvious question to ask is what
is the relative parasiticidal power of the two? Is one more
active than the other? Are other types of carbon radical
equally as active? Another question concerns their
susceptibility to steric hindrance.

artemisinin in vitro, having an IC50 value of 4.5 ng/ml


against the W2 clone. The -methyl and dimethyl
derivatives 133 and 134 were inactive. Two conclusions can
be drawn. The first is that 133 and 134 are unable to access
heme failing to elicit electron transfer, whereas 132 sits
nicely on top of heme, generating one or both of the tertiary
and secondary radicals 135 and 136 with perhaps the latter
being the more effective at killing the parasite. The second is
that coordination with heme takes place with all three
trioxanes, but as 133 cannot undergo 1,5 shift only the
secondary radical 136 forms. Similarly, the only option open
to 134 is cleavage to the unreactive tertiary radical 137.
Therefore, it appears that hindrance to coordination with
heme is the determining factor for antimalarial activity in
this instance. Nevertheless, the electronic nature of the alkyl
radical is important too.

A pertinent first answer to such questions was provided


by the methylated tricyclic trioxanes 132-134 [95] Fig. (27).
The -methyl derivative 132 was about as twice as active as

Which of the alkyl radicals, secondary or tertiary, is the


most effective is answered by in vitro tests performed on
derivatives of the tricyclic trioxane 26 [44] Fig. (28). The -

Peroxides Generating Hindered and Substituted


Carbon Radicals have Poor Antimalarial Activity

MeO

OH
O

Me
O O R2

MeO
heme

R1
132 R1 = H, R2 = Me
133 R1 = Me, R2 = H heme
134 R1 = Me, R2 = Me

OH
O
OH
O

Me
Me
or

MeO

MeO

Me

O O Me

heme
135

OH
O

heme
136

OH
O

Me

O O Me
Me
heme
137

Fig. (27). Formation of tertiary and secondary radicals by reacting heme with trioxane 132.

1820

Current Medicinal Chemistry, 2001, Vol. 8, No. 15

Charles W. Jefford

benzyl derivative 138 was about as twice as active as the


parent trioxane 26. However, the -phenyl derivative 139
was devoid of activity confirming that a carbon radical
stabilized by conjugation, or hyperconjugation like the
aforementioned tertiary radical, is not aggressive enough to
kill the parasite.
MeO

Me
OO R

26 R = H
138 R = CH2Ph
139 R = Ph

MeO

Me

O O
Me
140

Fig. (28). Active and inactive tricyclic trioxanes.

Obviously the two factors, electronic and steric, go hand


in hand. Yet the ability of the peroxide bond to get into
intimate contact with heme is of paramount importance. The
methyl derivative 140, in which the underside of the
peroxide bond is sterically blocked provides an apt
illustration Fig. (28). It is totally inactive [96, 97].
The effect of steric hindrance on the availability or
propensity of a carbon radical to accomplish its deadly task
is patently seen on inspecting the in vitro activities of
substituted cyclohexane-1,2,4-trioxanes and 1,2,4,5tetroxanes both of which have the same modus operandi
[98]. The spirocyclic trioxane 141 has an IC50 against the
W2 clone in vitro of 3.9, comparing well with artemisinin
(14) and its value of 1.2 ng/ml Fig. (29). Simply placing
four methyl substituents on the cyclohexane rings sharply
diminishes the activity of 142, its IC50 rising to 94.0 ng/ml.
1,2,4,5-Tetroxanes have long been known to possess
antimalarial activity, but much less so than their trioxane
counterparts [99-101]. The cyclohexane derivative 143 is 4-5
times less active than 141 showing an IC50 of 20 ng/ml Fig.
(29). Loading the tetroxane with two pairs of geminal
Me
O
O

Substitutions on certain parts of the artemisinin skeleton


can also profoundly perturb its activity. As previously
mentioned, substitutions in the lactone ring, either as ester
or ether derivatives of the epimeric lactols, are sufficiently
remote from the peroxide linkage as to have little effect on
docking with heme. In fact, such derivatives, like the lactols
themselves, are often more active than the lactonic parent,
simply for electronic reasons. However, substituents at the
-position to the lactone lie closer to the peroxide linkage.
Changing them alters the activity in a systematic and, in
certain cases, a dramatic way [102]. Replacing the -methyl
group in artemisinin (14) by ethyl or propyl leads to IC50
values against the W2 clone for 147 and 148 that are roughly
12 times smaller than that of 14, in other words, the
compounds are more active Fig. (30). However, substitution
by a much longer aliphatic chain, as evidenced by 149,
annihilates activity completely. Presumably, because the
chain is long enough and sufficiently mobile to reach and
cover the peroxide bond preventing approach to heme.
Surprisingly, in vitro tests carried out on epiartemisinin
151 revealed IC 50 values only 1.5 -2.0 times bigger than
that of 14 [102, 103] Fig. (30). In contrast, the IC50 of the
bromo derivative 150 is 8 times that of than 14. The isostere
of 150, the gem-dimethyl derivative 152, is completely inert.
The expectation is that all three -substituted derivatives
would erect the same kind of steric barrier to an incoming
molecule of heme. A reason for the low IC50 values of 151

Me

Me
Ph

methyl groups on the cyclohexane ring raises the IC 50 of 144


to 255 ng/ml, wiping out activity. It can be concluded that
steric encumbrance at the C3 position on the cyclohexane
ring relative to the spirocyclic carbon atom somehow
interferes with parasiticidal action. As in the case of the
methylated tricyclic trioxanes, two hypotheses are possible.
First, the bulky gem-dimethyl groups could simply prevent
the peroxide bond from getting close to heme. Second, if
heme manages to get near enough to effect electronic transfer,
the resulting neopentyl type radicals, e.g. 145 and 146,
would be unreactive and therefore useless for killing
parasites.

Me
O

Me

Ph

141

O
Ph

Ph

142

Me

Me

O
Me

Me
Me

Me

144

143

Me

heme
Me

heme
Me

Me

Me

.
Me

Me

O
O

heme

Ph
O

Me

Ph

145
Fig. (29). Sterically encumbered trioxanes and tetroxanes.

.
Me

O
heme Me
146

Me
Me
Me

Implications for the Mode of Action

Current Medicinal Chemistry, 2001, Vol. 8, No. 15

might be due to isomerization. Under the conditions of


incubating P. falciparum during the test, 151, the less stable
epimer, or at least some of it, could have reverted to the
more active -epimer 14.

of 14. Against P. yoelii ssp. NS, the resistant line, the ED50
and ED90 values are 4-6 times bigger. Clearly, the
epimer 151 is 4-7 times less effective than its -epimer 14.
The persistence of some activity may be due to some
epimerization to 14 under the test conditions. Comparison of
the X-ray data of 14 and 151 reveals that the two molecules
are superimposable, except for the lactone ring in 151 which
is slightly distorted to alleviate congestion between the methyl group and the peroxide bond. Thus the difference in
activity arises solely from the orientation of the methyl
substituent on the lactone ring.

O
R2
O O

R1

Me

Me
14
147
148
149
150
151
152

R1 = Me, R2 = H
R1 = Et, R 2 = H
R1 = Pr, R2 = H
R1 = (CH 2)13Me, R2 = H
R1 = Me, R2 =Br
R1 = H, R 2 = Me
R1 = R 2 = Me

The diminished activity of 151 and its absence by 152


could be ascribed to poor docking on heme Fig. (31). Thus,
the complexes 154 and 155 owing to obstruction by the methyl substituent either do not form or are not intimate
enough for an electron to jump from the 3d orbital of iron to
the anti-bonding orbital of the O-O sigma bond.
Consequently, neither oxy nor successive carbon radicals
arise from 154 and 155 like they do from 153. An alternative
is that docking in 154 and 155 is close, but skewed obliging
the iron atom of heme to selectively bind to the distal
oxygen atom of the peroxide bond. As a result, electron
transfer occurs leading solely to the secondary carbon radicals
158 and 159. The tight complex formed with artemisinin
(153) would be expected to form mostly, if not all, the
primary radical 156. The relative parasiticidal power of the
primary and secondary radicals (e.g. 156 vs. 157) is not
known, but the selective capture of the primary radical by
Mn(II)TPP (v. supra) argues in favor of the former.

Fig. (30). Some derivatives of artemisinin.


A recent X-ray study of 151 reveals that the -methyl
substituent lies in van der Waals contact with the proximal
oxygen atom of the peroxide bond [104]. Evidently, the
peroxide is hindered. The consequences of this hindrance are
shown up by the in vivo activity. Against Plasmodium
berghei N, the chloroquine-sensitive line, the ED50 and
ED90 values of 151 are about seven times greater than those
O

O
R

O
2
R
O O

O
Me

Me

Fe

N
N

Me

Me

N Fe N

HO2C
HO2 C

Me

Me

Me

Me

Me
Me

Me

153
154
155

Me

Me

N Fe N

HO2C

O
2
R
OH O

R1

Me

HO2C

Me

Me

Me
N

HO2C
HO2C

Me

Me

1821

R1 = Me, R2 = H
R1 = H, R2 = Me
1
2
R = R = Me

156

157
158
159

PP

R1 = Me, R2 = H
1
2
R = H, R = Me
1
2
R = R = Me

Me
O
Me

Me

O
OH PP

Me

N + N
Fe
N
N

HO2C
HO2C
Me

160
Fig. (31). Formation of radicals from artemisinin, epiartemisinin and congeners by reaction with heme.

161

Me

Me

hemozoin

1822

Current Medicinal Chemistry, 2001, Vol. 8, No. 15

Charles W. Jefford

Me
O

Me

Me

heme
Me

Me
OH

Me
O

Me

O
hemin

H OH

13
Me

Me
excision

OH
H OH

163

Me

Me

OH

.
O

Me

H OH

PP + hemin
OH

Me

hemin

164

hemozoin

166

165

Fig. (32). Mode of action of yingzhaosu A.


How Artemisinin and Peroxidic Antimalarials Kill the
Parasite
How the parasite is actually killed is also not known. It
must be emphasized that FeCl2.4H2O in MeCN is not the
same as heme inside the red blood cell. Both entities display
the same redox properties towards artemisinin. The first acts
as a catalyst as already mentioned, but intraerythrocytic
heme acts as a reagent when a parasite is present. Alkylation
by a reactive hemin-radical e.g. 156 undoubtedly takes place

Me

Me
Me heme
OH

O O
Tol-p
162

The cis-fused cyclopentene-trioxanes, exemplified by the


most potent candidate 53, like cis-decalin easily undergo
conformational inversion. They therefore are able to make a
close fit with the surface of heme. The resulting complex
from 53 then unravels to the hemin derivative of the primary
radical 74, which then proceeds to alkylate parasite protein.
The motor for antimalarial activity is the thermodynamic
stability which comes from the formation of a carbonyl group
at its simplest, or better one that is conjugated. Such a
structural feature is found in yingzhaosu A (13) and C (162)

Me
Me

OH
O O
Me
Tol-p hemin
167

Me

O
Tol-p

Me
Me

Me
Me

.
+

168

OH

PP

OH

hemin

hemin

169

170

Fig. (33). Mode of action of yingzhaosu C.

inside the food vacuole of the parasite Fig. (31). Addition to


an unsaturated center in the parasite protein (PP) would
afford a new radical, which could then eject an electron
yielding a cation followed by its deprotonation. Lastly,
protonation of the resulting PP-artemisinin-hemin adduct
156 releases the alkylated protein 160 from its hemin
appendage (161) which subsequently polymerizes to
hemozoin. Experimentally, it was found that on treatment
with artemisinin trophozoites of P. falciparum still produce
hemozoin, while with chloroquine less is produced [105]. In
any event, heme(FeII) is first oxidized to hemin(FeIII),
which then polymerizes. How this happens is controversial,
but it may well be a purely chemical process [106]. In
summary, in a sort of quid pro quo, a toxic C-centered
radical replaces toxic heme to kill the parasite.

O
O

Fe2+
16

Me

Confirmation of the validity of the foregoing schemes has


been obtained by treating arteflene (16) with FeCl2.4H2O in
MeCN [107]. Evidence for cleavage of the ferric oxy radical
171 to the cyclohexyl radical 172 and the enone 173 was
obtained by capture of the former with 5,5-dimethyl-1pyrroline N-oxide and the identification of its adduct by EPR

Me C6H 3(CF3 )2-2,4

Me C 6H3(CF3)2 -2,4
Me

and may contribute to their potency [68] Fig. (32).


Complexation of heme with 13 gives the hemin oxy radical
163. Excision of the ,-unsaturated ketone 164 releases the
cyclohexyl radical 165 which then alkylates parasite protein
finally producing 166 and eventually hemozoin. Yingzhaosu
C (162) behaves the same way Fig. (33). The hemin-oxy
radical 167 cleaves to the acetophenone 168 and the ethyl
radical 169. Alkylation gives the hemin-PP adduct 170
which finally ejects hemin as before.

O
Fe2 +

171

Fig. (34). Ferrous ion-induced cleavage of arteflene.

Me
O

C6 H3(CF3)2-2,4

O +
Me
Fe 2+
172

173

Implications for the Mode of Action

spectroscopy [108] Fig. (34). The potency of arteflene (16) is


proof that secondary C-centered radicals are reactive enough
after all to be effective schizonticides. Nevertheless, the fact
that the p-fluorophenyl-trioxane 53 is even more potent
points to the efficacy of a primary radical (Table 1).
Evidence that artemisinin on treatment with a Lewis acid
or benzylamine opens to a hydroperoxide has been suggested
as a basis of parasiticidal action [109, 110]. It will be
interesting to see if such reactions occur under physiological
conditions and how biomolecules react with the purported
electrophilic oxygenating species derived from the
hydroperoxide. In general, hydroperoxides that are unable to
cleave to a carbon radical only manifest weak antimalarial
activity [29].
Molecular Considerations in Designing New Peroxidic
Antimalarials
The preceding mechanistic schemes clearly depend on the
optimal concurrence of several molecular properties in order
to produce the ultimate antimalarial. In other words, the
parasiticidal action of potent peroxides like 14 and 53
depends on the efficient operation of a sequence of chemical
events, namely, docking with heme, electron transfer,
formation of an oxy radical, scission to an unencumbered Ccentered radical powered by the formation of a carbonyl
group or something similar, and finally, the death of the
parasite by alkylation. Many trioxanes and peroxides are
going to fulfill the above criteria. In designing new peroxidic
antimalarials, any one of the above desiderata needs to be
examined to see if the candidate will fit the bill.
Many spatial arrangements are possible between heme
and artemisinin (14) and it does not necessarily follow that
the one usually drawn is the right one. However, computerassisted molecular modeling of various docking
arrangements of hemin(FeIII) (serving as a model for heme)
and 14 using the Sybil program showed that in the most
stable configuration the peroxide bond and the carbonyl
oxygen atom interact closely with hemin iron [111]. The face of artemisinin is close to heme putting the interatomic
distance between 2.6 and 2.8 from the peroxide oxygen
atoms to the iron. A similar procedure with
deoxyartemisinin (23) and hemin favored close binding on
the opposite or -face.
A QSAR study using Comparative Molecular Field
Analysis of 11-alkyl derivatives of artemisinin as well as
some tricyclic trioxanes similar to D-seco-artemisinin gave a
good correlation between calculated and observed activities
[102]. From the relative contributions of steric and
electrostatic terms the former had the greatest bearing on
activity. The latter finding accords well with docking on
heme as the crucial first step for effective drug action.
In a 3D-QSAR study using a pharmacophore search
method (CATALYST) two hydrophobic features and
hydrogen bonding were identified as the hypothesis or
pharmacophore responsible for activity [112]. By taking a
training set of trioxanes having known in vitro and in vivo
activities the best arrangement of the hypothesis consonant

Current Medicinal Chemistry, 2001, Vol. 8, No. 15

1823

with the highest activity was found. Typically, the highest


antimalarial activity is shown by those cis-fused trioxanes
which map well with the hypothesis (41-44 and 53) as they
are able to adjust their conformations at little cost in energy.
An interesting result was the predicted inertness of the cisfused cyclohexene trioxanes (37-40), the reason being their
inability to adopt a low enough energy conformation that
overlaps with the computed ideal pharmacophore. It was also
found that 53 docks closely with heme putting the iron atom
close to the peroxide bond.
Attempts have been made to correlate computed
molecular electrostatic potentials (MEPS) with antimalarial
activity with little success [113]. A problem is the difficulty
of comparing the MEPS of molecules that are quite different
in shape. A technique, which paradoxically facilitates
comparison, is to reduce the dimensionality of the MEPs by
recourse to Kohonen Neural Network transforms [114]. To
visually improve the characteristic features of a given
transform, each is surround by eight replicas. The first
indications are that such 3 x 3 tilings of active trioxanes,
such as artemisinin, give continuous strips of charge,
whereas inactive peroxides give broken strips. More
calculations on a bigger sample base need to be undertaken
to confirm the aforementioned early conclusions.

CONCLUSION AND PROSPECTS


It is seen from the preceding discussion that the evidence
for the involvement of carbon radicals produced from a
peroxidic precursor by heme is completely established.
Subsequently, the killing of the parasite within its food
vacuole by alkylation is a logical and convincing sequel.
However, the identity and characterization of the alkylated
parasite protein needs to be elucidated to fully confirm the
reality of this crucial step. At the same time, suggestions
that the death of the parasite may occur by oxygen atom
transfer or by the action of an oxy electrophilic species
should be clarified. As alkylation and oxygenation are
chemically different they should be easily distinguishable.
The finding that artemether and arteether are neurotoxic at
high dose in animal models is probably due to alkylation by
radicals, namely, by the same mechanism that works for
killing the parasite [115]. Similarly, the iron-triggered
alkylation mechanism which characterizes antimalarial
trioxanes and peroxides indicates that they would find use as
antitumor agents as well [116].
There is no doubt that the conventional antimalarial
remedies based on quinoline derivatives have outlived their
usefulness. In view of the increase of multi-drug-resistant
falciparum malaria, new, effective drugs are urgently needed.
Experience has shown that reformulations of old drugs are
not going to provide an adequate response to the growing
menace. A new antimalarial must be as simple in structure
as aspirin or paracetamol, taken as a tablet by mouth, and as
cheap and safe to manufacture on the ton scale. The synthetic
trioxane 53 is a prime candidate, fitting the aforementioned
criteria. Nonetheless, the design of other effective simple
peroxides, bearing neutral water soluble groups such as
hydroxyl substituents, is now entirely feasible in the light of

1824

Current Medicinal Chemistry, 2001, Vol. 8, No. 15

the knowledge accumulated from mechanistic studies and the


systematic preparation and testing of many trioxanes and
peroxides of varied structure. For example, easily accessible
alkyl and hydroxyalkyl analogues of 1-methylcyclopropyl
hydroperoxide, as well as other variations on the basic
structure, offer promise and should be prepared and tested.

Charles W. Jefford

[23]

Woerdenbag, H.J.; Pras, N.; van Uden, W.; Wallart, T.E.;


Beekman, A.C.; Lugt, C.B. Pharm. World Sci. 1994, 16 ,
169.

[24]

Jung, M. Curr. Med. Chem. 1994, 1, 35.

[25]

Zaman, S.S.; Sharma, R.P. Heterocycles 1991, 32 , 1593.

[26]

Ziffer, H.; Highet, R.J.; Klayman, D. in Progress in the


Chemistry of Natural Products; Herz, W.; Kirby, G.W.;
Moore, R.E.; Steglich, W.; Tamm, C., Eds.; Springer:
Vienna, New York, 1997, 72 , pp 121-214.

[27]

Cumming, J.N.; Ploypraith, P.; Posner, G.H. In Advances


in Pharmacology; August, J.T.; Anders, M.W.; Murad,
F.; Coyle, J.T., Eds.; Academic Press: San Diego, 1997,
Vol. 37 , pp 253- 297.

REFERENCES
[1]

Weekly Epidemiological Record, October 21, 28, 1994.

[2]

Knell, A.J. Malaria; Oxford University Press: Oxford,


1991.

[3]

Malaria, Obstacles and Opportunities; Oakes, S.C.;


Mitchell, V.S.; Pearson, G.W.; Carpenter, C.J. Eds.;
National Academy Press: Washington, D.C. USA, 1991.

[28]

Meshnick, S.R.; Taylor, T.E.; Kamchonwongpaisan, S.


Microbiol. Rev. 1996, 60 , 301.

[4]

Gramiccia, G. Acta Leidensia 1987, 55 , 5.

[29]

[5]

Sweeney, T.R.; Strube, R.E. In Burgers


Medicinal
Chemistry; Wolff, M.E., Ed.; John Wiley and Sons: New
York, 1979, Part II, pp. 333-413.

Jefford, C.W. In Advances in Drug Research; Testa B.;


Meyer, A. U., Eds.; Academic Press: New York, 1997,
Vol. 29 , Chap. 7.

[30]

Curd, F.H.S.; Davey, D.G.; Rose, F.L. Ann. Trop. Med.


Parasit. 1945, 39 , 139, 157, 208.

Xu, X.-X.; Zhu, J.; Huang, D-Z.; Zhou, W.-S. Tetrahedron


Lett . 1991, 32 , 5785.

[31]

Falco, E.A.; Goodwin, L.G.; Hitchings, G.H.; Rollo,


I.M.; Russell, P.B. Brit. J. Pharm. Chemotherap. 1951, 6,
185.

Hofheinz, H.; Burgin, H.; Gocke, E.; Jaquet, C.;


Masciadri, R.; Schmid, G.; Stohler, H.; Urwyler, H. Trop.
Med. Parasitol. 1994, 45 , 261.

[32]

Russell, P.F. Mans Mastery of


University Press: London, 1955.

Li, Y.; Yu, P.; Chen, Y.; Li, L.; Gai, Y., Wang, D.; Zheng,
Y. Acta Pharm. Sinica 1981, 16 , 429.

[33]

China Cooperative Research Group on Qinghaosu and


its derivatives as Antimalarials, J. Trad. Chin. Med.
1982, 2, 17.

[34]

Brossi, A.; Venugopalan, B.; Gerpe, L.D.; Yeh, H.J.C.;


Flippen-Anderson, J.L.; Buchs, P.; Luo, X.-D.; Milhous,
W.; Peters, W. J. Med. Chem. 1988, 31 , 645.

[35]

Jung, M.; Li, X.; Bustos, D.A.; ElSohly, H.N.;


McChesney, J.D.; Milhous, W.K. J. Med. Chem. 1990, 33 ,
1516.

[36]

Jefford, C.W. In Comprehensive Heterocyclic Chemistry;


Katritzky, A.R.; Rees, C.W.; Scriven, E.F.V. Eds.;
Pergamon: Oxford, 1996, 6, Chap. 20.

[37]

Desjardins, R.E.; Canfield, C.J.; Haynes, D.E.; Chulay,


J.D. Antimicrob. Agents Chemother. 1979, 16 , 710.

[38]

Milhous, W.K.; Weatherly, N.F.; Bowdre, J.H.;


Desjardins, R.E. Antimicrob. Agents Chemother. 1985,
27 , 525.

[6]
[7]

[8]

Malaria; Oxford

[9]

Travis, J. Nature 1997, 152, 340.

[10]

White, N.J. Lancet 1987, March 21.

[11]

Miller, K.D.; Lobel, H.O.; Satriale, R.F.; Kuritsky, J.N.;


Stern, R.; Campbell, C.C. Ann. J. Trop. Med. Hyg. 1986, 5,
451.

[12]

Gutteridge, W.E. J. Roy. Soc. Med. 1989, 82 , 63 (suppl.


17).

[13]

LEstrange Orme, L. Acta Leidensia 1987, 55 , 77.

[14]

Peters, W. Brit. Med. Bull. 1982, 38 , 187.

[15]

Howells, R.E. Brit. Med. Bull. 1982, 38 , 193.

[16]

Shen, C.C.; Zhuang, L.G. Med. Res. Rev. 1984, 4, 47.

[17]

Liang, X.T.; Yu, D.Q.; Wu, W.L.; Deng, H.C. Acta Chim.
Sinica 1979, 37 , 215.

[18]

Liu, J.-M.; Ni, M.-Y.; Fan, J.-F.; Tu, Y.-Y.; Wu, Z.H.;
Zhou, W.-S. Acta Chim. Sinica 1979, 37 , 129.

[39]

Jefford, C.W.; Velarde, J.A.; Bernardinelli, G. Tetrahedron


Lett . 1988, 30 , 4485.

[19]

Klayman, D.L. Science 1985, 228, 1049.

[40]

[20]

Butler, A.R.; Wu, Y.-L. Chem. Soc. Rev. 1992, 85.

Jefford, C.W.; Velarde, J.A.; Bernardinelli, G.; Bray, D.H.;


Warhurst, D.C.; Milhous, W.K. Helv. Chim. Acta 1993,
76 , 2775.

[21]

Luo, X.-D.; Shen, C.- C. Med. Res. Rev. 1987, 7, 29.

[41]

[22]

China Cooperative Research Group on Qinghaosu and


its derivatives as Antimalarials, J. Trad. Chinese Med.
1982, 2, 9.

Jefford, C.W.; Misra, D.; Rossier, J.C.; Kamalaprija, P.;


Burger, U.; Mareda, J.; Bernardinelli, G.; Peters, W.;
Robinson, B L.; Milhous, W.K.; Zhang, F.; Gosser, D.K.;
Meshnick, S.R. In Perspectives in Medicinal Chemistry;
Testa, B.; Kyburz, E.; Fuhrer, W.; Giger R., Eds.; Verlag
Helvetica Chimica Acta: Basel, 1993, Chapter 29.

Implications for the Mode of Action

Current Medicinal Chemistry, 2001, Vol. 8, No. 15

[42]

Posner, G.H.; Oh, C.H.; Milhous, W.K. Tetrahedron Lett.


1991, 32 , 4235.

[43]

Posner, G.H.; Oh, C.H.; Gerena, L.; Milhous, W.K. J.


Med. Chem. 1992, 35 , 2459.

[44]

[45]

Posner, G.H.; Wang, D.; Cumming, J.N.; Oh, C.H.;


French, A.N.; Bodley, A.L.; Shapiro, T.A. J. Med. Chem.
1995, 38 , 2273.
Jefford, C.W.; McGoran, E.C.; Boukouvalas, J.;
Richardson, G.; Robinson, B.L.; Peters, W. Helv. Chim.
Acta 1988, 71 , 1805.

1825

[64]

Asawamahasakda, W.; Ittarat, I.; Pu, Y.-M.; Ziffer, H.,;


Meshnick, S.R. Antimirob. Agents Chemother. 1994, 38 ,
1854.

[65]

Walsh, C. Enzymatic
Reaction Mechanisms; W.H.
Freeman and Company: San Francisco, 1979, Chapter 15.

[66]

Ortiz de Montellano, P.R. Acc. Chem. Res. 1987, 20 , 289.

[67]

Jefford, C.W.; Favarger, F.; Vicente, M.G.H.; Jacquier, Y.


Helv. Chim. Acta 1995, 78 , 452.

[68]

Jefford, C.W.; Vicente, M.G.H.; Jacquier, Y.; Favarger, F.;


Mareda, J.; Millasson-Schmidt, P.; Brunner, G.; Burger,
U. Helv. Chim. Acta 1996, 79 , 1475.

[69]

Herz, W.; Ligon, R.C.; Turner, J.A.; Blount, J.F.J. Org.


Chem. 1977, 42 , 1885.

[46]

Peters, W.; Robinson, B.L.; Rossier, J.C; Jefford, C.W.


Ann. Trop. Med. Hyg. 1993, 87 , 1.

[47]

Jefford, C.W.; Godoy, J. Unpublished work.

[48]

Jefford, C.W.; Jaggi, D.; Kohmoto, S.; Timri, G.;


Bernardinelli, G.; Canfield, C.J.; Milhous, W.K.
Heterocycles 1998, 49 , 375.

[70]

Turner, J.A.; Herz, W. J. Org. Chem. 1977, 42 , 1895.

[71]

Turner, J.A.; Herz, W. J. Org. Chem. 1977, 42 , 1900.

[49]

Jefford, C.W.; Rossier, J.-C.; Canfield, C.J. Unpublished


work.

[72]

Takahashi, K.; Kishi, M. Tetrahedron 1988, 44 , 4737.

[73]

Jefford, C.W. Jpn. J. Trop. Hyg. 1996, 24 Suppl. 1, 77.

[50]

Jaquet, C.; Stohler, H.R.; Chollet, J.; Peters, W. Trop.


Med. Parasitol. 1994, 45 , 266.

[74]

Jefford, C.W.; Brunner, G.; Millasson-Schmidt, P.; Burger,


U. unpublished work.

[51]

Jefford, C.W.; Rossier, J.-C.; Milhous, W.K.; Bray, D.;


Warhurst, D.C. Unpublished.

[75]

Posner, G.H.; Oh, C.H. J. Am Chem. Soc. 1992, 114,


8328.

[52]

Jefford, C.W.; Kohmoto, S.; Jaggi, D.; Timri, G.; Rossier,


J.-C.; Rudaz, M.; Barbuzzi, O.; Grard, D.; Burger, U.;
Kamalaprija, P.; Mareda, J.;
Bernardinelli,
G.;
Manzanares, I.; Canfield, C.J.; Fleck, S.L.; Robinson, B.;
Peters, W. Helv. Chim. Acta 1995, 78 , 647.

[76]

Posner, G.H.; Cumming, J.N.; Ploypradith, P.; Oh, C.H. J.


Am Chem. Soc. 1995, 117, 5885.

[77]

Posner, G.H.; Oh, C.H.; Gerena, L.; Milhous, W.K. J.


Med. Chem. 1992, 35 , 2459.

[53]

Peters, W.; Portus, J.H.; Robinson, B.L. Ann. Trop. Med.


Parasitol. 1975, 69 , 155.

[78]

Wilt J.W. In Free Radicals; J.K. Kochi, Ed.; WileyInterscience: New York, N.Y., 1973, Vol. I, Chap. 3.

[54]

Peters, W.; Robinson, B.L.; Tovey, G.; Rossier, J.-C.;


Jefford, C.W. Ann. Trop. Med. Parasitol. 1993, 87, 111.

[79]

[55]

Slater, A.F.G.; Swiggard, W.J.; Orton, B.R.; Flitter, W.D.;


Goldberg, D.E.; Cerami, A.; Henderson, G.B. Proc. Nat.
Acad. Sci. U.S.A. 1991, 88 , 325.

Brun, P.; Waegell, B. In Reactive Intermediates; Plenum


Press: New York, 1983, Vol. 3, p. 367 and references
cited therein.

[80]

Gu, J.; Chen, K.; Jiang, H.; Leszczynski, J. J. Phys. Chem.


A 1999, 103, 9364.

[56]

Warhurst, D.C. Biochem. Pharmacol. 1981, 30 , 3323.

[81]

[57]

Slater, A.F.G.; Cerami, A. Nature 1992, 355, 167.

Lawesson, S.O.; Sosnovsky, G. Svensk. Kem. Tidskr.


1963, 75 , 343.

[58]

Vippagunta, S.R.; Dorn, A.; Matile, H.; Bhattacharjee,


A.K.; Karle, J.M.; Ellis, W.Y.; Ridley, R.G.; Vennerstrom,
J.L. J. Med. Chem. 1999, 42 , 4630.

[82]

Posner, G.H.; Wang, D.; Gonzlez, L.; Tao, X.; Cumming,


J.D.; Klinedinst, D.; Shapiro, T.A. Tetrahedron Lett.
1996, 37 , 815.

[59]

Meshnick, S.R.; Thomas, A.; Ranz, A.;. Xu, C.; Pan, H.


Mol. Biochem. Parasitol. 1991, 49 , 181.

[83]

Guengerich, F.P.; MacDonald, T.L. Acc. Chem. Res. 1984,


17 , 9.

[60]

Hong, Y.; Zang, Y.; Meshnick, S.R. Mol.


Parasitol. 1994, 63 , 121.

Biochem.

[84]

Meshnick, S.R.; Jefford, C.W.; Posner, G.H.; Avery, M.A.;


Peters, W. Parasitology Today 1996, 12 , 79.

[61]

Zhang, F.; Gosser, D. Jr.; Meshnick, S.R. Biochem.


Pharmacol. 1992, 43 , 1805.

[85]

Wu, W.-M.; Yao, Z.-J; Wu, Y.-L.; Jiang, K.; Wang, Y.-F.;
Chen, H.-B.; Shan, F.; Li, Y. Chem. Commun. 1996, 2213.

[62]

Meshnick, S.R.; Yang, Y.-Z.; Lima V.; Kuypers, F.;


Kamchonwongpaisan, S.; Yuthavong, Y. Antimicrob.
Agents Chemother. 1993, 37, 1108.

[86]

Wu, W.-M.; Wu, Y.; Wu, Y.-L; Yao, Z.-J.; ; Zhou, C.-M.;
Li, Y.; Shan, F. J. Am. Chem. Soc. 1998, 120, 3316.

[87]

Robert, A.; Meunier, B. J. Am. Chem. Soc. 1997, 119,


5968.

[88]

Robert, A.; Meunier, B. Eur. J. Chem. 1998, 4, 1287.

[89]

Robert, A.; Meunier, B. Chem. Soc. Rev. 1998, 27, 273.

[63]

Hong, Y.-L.; Yang, Y.-Z.; Meshnick, S.R. Mol. Biochem.


Parasitol 1994, 63 , 121.

1826

Current Medicinal Chemistry, 2001, Vol. 8, No. 15

[90]

Lee, I.-K.; Hufford, C.D. Pharmac. Ther. 1990, 48 , 345.

[91]

Leskovac, V.; Theoharides,


Physiol.1991, 99C , 391.

[92]

Adams M.J. In Enzyme


Mechanisms;. Page M.I.;
Williams, A., Eds.; Royal Society of Chemistry: London:
1987, Chapter 24.

[93]

A.D.

Reichardt, C. Solvent Effects in


Verlag Chemie: Weinhem, 1979.

Charles W. Jefford

Comp.

Organic

Biochem.

Chemistry;

[94]

Avery, M.A.; Fan, P.; Karle, J.M.; Bonk, J.D.; Miller, R.;
Goins, D.K. J. Med. Chem. 1996, 39 , 1885.

[95]

Posner, G.H.; Oh, C.H.; Wang, D.; Gerena, L.; Milhous,


W.K.; Meshnick, S.R.; Asawamahasadka, W. J. Med.
Chem. 1994, 37 , 1256.

[96]

Zouhiri,F.; Desmale, D.; DAngelo, J.; Riche, C.; Gay F.;


Cicron, L. Tetrahedron Lett. 1998, 39, 2969.

[97]

Zouhiri, F.; Desmale, D.; DAngelo, J.; Mahuteau, J.;


Riche, C.; Gay, F.; Cicron, L. Eur. J. Org. Chem. 1998,
2897.

[98]
[99]

Jefford, C.W.; Rossier, J.-C.; Milhous, W.K. Heterocycles


2000, 52, 1345.
Doorenbos H.E.; Decker, D.L. Annual Technical Report
No. AM-1A-73, The Dow Chemical Company: 1973.

[100] Vennerstrom, J.L.; Fu, H.; Ellis, W.Y. Ager, A.L Jr.;
Wood, J.K.; Andersen, S.L.; Gerena, L.; Milhous, W.K. J.
Med. Chem. 1992, 35 , 3023.
[101] Kim, H.-S.; Shibata,Y.; Wataya, Y.; Tsuchiya, K.,
Masuyama, A.; Nojima, M. J. Med. Chem. 1999, 42 , 2604.
[102] Avery, M.A.; Gao, F.; Chong, W.K.M.; Mehrotra, S.;
Milhous, W.K. J. Med. Chem. 1993, 36 , 4264.
[103] Acton, N.; Klayman, D. Planta Med. 1987, 266.

[104] Jefford, C.W.; Burger, U.; Millasson-Schmidt, P.;


Bernardinell, G.; Robinson, B.L.; Peters, W. Helv. Chim.
Acta 2000, 83 , 1239.
[105] Asawamahasakda, W.; Ittarat, I.; Chang, C.; McElroy, P.;
Meshnick, S.R. Mol. Biochem. Parasitol. 1994, 67, 183.
[106] Dorn, A.; Stoffel, R.; Matile, H.; Bubendorf, A.; Ridley, R.
Nature 1995, 374, 269.
[107] ONeill, P.M.; Bishop, L.P.; Searle, N.L.; Maggs, J.L.;
Ward, S.A.; Bray, P.G.; Storr, R.C.; Park, B.K.
Tetrahedron Lett. 1997, 38 , 4263.
[108] ONeill, P.M.; Bishop, L.P.D.; Searle, N.L.; Maggs, J.L.;
Storr, R.C.; Ward, S.A.; Park, B.K.; Mabbs, F. J. Org.
Chem. 2000, 65 , 1578.
[109] Haynes, R.K.; Vonwiller, S.C. Tetrahedron Lett. 1996,
37 , 257.
[110] Haynes, R.K.; Pai, H.H.; Voerste, A. Tetrahedron Lett.
1999, 40 , 4715.
[111] Shukla, K.L.; Gund, T.M.; Meshnick, S.R. J. Mol. Graph.
1995, 13 , 215.
[112] Grigorov, M.; Weber, J.; Tronchet, J.M.J.; Jefford, C.W.;
Milhous, W.K.; Maric D. J. Chem. Inf. Comput. Sci. 1997,
37 , 124.
[113] Bernardinelli, G;. Maric, D.; Jefford, C.W.; Thomson, C.;
Weber, J. Int. J. Quantum Chem. Biol. Symp. 1994, 21,
117.
[114] Jefford, C.W.; Grigorov, M.; Weber, J.; Lthi, H. P.;
Tronchet J.M.J. J. Chem. Inf. Comput. Sci. 2000, 40 , 354.
[115] Brewer, T.G.; Grate, S.J.; Peggins, J.O.; Weina, P.J.;
Petras, J.M.; Levine, B.S.; Heiffer, M.H.; Schuster, B.G.
Am. J. Trop. Med. Hyg. 1994, 51 , 251.
[116] Moore, J.C.; Lai, H.; Li, J.R.; Ren, R.L.; McDougall, J.A.;
Singh, N.P.; Chou. C.K. Cancer Lett. 1995, 98 , 83.

Das könnte Ihnen auch gefallen