Unified mechanistic framework for the Fe(II)-induced cleavage of Qinghaosu and derivatives/analogues. The first spin-trapping evidence for the previously postulated secondary C-4 radical

Article abstract of DOI:10.1021/ja973080o

Qinghaosu and derivatives were easily reduced by ferrous sulfate in aqueous acetonitrile to give results different from those reported for other reducing systems. The unstable epoxide 7, a compound that was postulated earlier as a species responsible for

Full text of DOI:10.1021/ja973080o

3316
J. Am. Chem. Soc. 1998, 120, 3316-3325
Unified Mechanistic Framework for the Fe(II)-Induced Cleavage of
Qinghaosu and Derivatives/Analogues. The First Spin-Trapping
Evidence for the Previously Postulated Secondary C-4 Radical
†
†
,†
†
†
Wen-Min Wu, Yikang Wu, Yu-Lin Wu,* Zhu-Jun Yao, Cheng-Ming Zhou,
,‡
‡
Ying Li,* and Feng Shan
Contribution from the State Key Laboratory of Bio-organic and Natural Products Chemistry,
Shanghai Institute of Organic Chemistry, Chinese Academy of Sciences, Shanghai 200032, China, and
Shanghai Institute of Materia Medica, Chinese Academy of Sciences, Shanghai 200031, China
ReceiVed September 2, 1997
Abstract: Qinghaosu and derivatives were easily reduced by ferrous sulfate in aqueous acetonitrile to give
results different from those reported for other reducing systems. The unstable epoxide 7, a compound that
was postulated earlier as a species responsible for the antimalarial activity, now has been isolated and
characterized. The earlier speculative secondary C-4 radical has also been trapped with 2-methyl-2-
nitrosopropane and thus provides the very first direct evidence for the involvement of radicals in the in vitro
cleavage of QHS-type compounds. A unified mechanism featuring interchangeable radical anions and reversible
intramolecular radical reactions is proposed for the ferrous ion induced cleavage of the 1,2,4-trioxanes (i.e.,
QHS and the like). On the basis of this framework, together with consideration of counterion and solvent
effects, a large body of divergent experimental outcomes can be satisfactorily rationalized, not only the formation
of the main products but also the product ratios as well as their deviation from those obtained under other
reaction conditions.
Introduction
Qinghaosu (artemisinin, 1, QHS) is a novel sesquiterpene
1
endoperoxide isolated in 1971 from Chinese medicinal herb
qinghao (Artemisia annua L.). Due to their potent antimalarial
activity, fast action, and low toxicity, QHS and its derivatives
have distinguished themselves as a new generation of antima-
2
larial drugs, especially in the treatment of multi-drug-resistant
cases. QHS has a unique carbon framework different from that
of all previous antimalarials. Structure-activity studies have
shown that the peroxy group is essential for the antimalarial
activity. The absence of this peroxy bridge, e.g. in deoxyqing-
3
haosu (deoxyQHS, 2), leads to complete loss of potency.
4
Deoxoqinghaosu (deoxoQHS, 3) and dihydroqinghaosu (dihy-
3
droQHS, 4) as well as its derivatives that retain an intact peroxy
group, on the other hand, show even higher antimalarial
activities than that of QHS itself. These discoveries have
stimulated a number of studies on the role of the peroxy group
in the biological activity of QHS.
in the antimalarial mechanism. As models for the biochemical
The mode of action of QHS type drugs against Plasmodium
has been suggested to be at the intraerythrocytic stage. Since
the parasitized erythrocytes (red blood cells) are rich in iron, it
is natural to conjecture that Fe(II) and Fe(III) might be involved
course of reaction of 1 in vivo, several groups around the world
5
5-10
have conducted investigations
cleavage of the peroxy bond since 1990. Some
on the Fe(II)-mediated in vitro
6
-10
used 1 and
5
closely related compounds, while others worked mainly on
†
Shanghai Institute of Organic Chemistry.
Shanghai Institute of Materia Medica.
(5) (a) 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., Jr.; Meshnick, S. R. PerspectiVes in
Medicinal Chemistry; Testa, B., Kyburz, E., Fuhrer, W., Giger, R., Eds.;
VCH Publishers, Inc.: New York, 1993; pp 459-472. (b) Jefford, C. W.;
Favarger, F.; Vicente, M.da G. H.; Jacruier, Y. HelV. Chim. Acta 1995, 78,
452-458. (c) 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, 1475-1487. (d) Bloodworth, A. J.; Shah, A. Tetrahedron
Lett. 1995, 36, 7551-7554.
‡
(
1) Lin, J.-M.; Ni, M.-Y.; Fan, J. F.; Tu, Y.-Y.; Wu, Z.-H.; Wu, Y.-L.;
Chou, W.-S. Acta Chim. Sin. 1979, 37, 129-141.
2) China Cooperative Research Group. J. Trad. Chin. Med. 1982, 2 (1),
7-24.
3) Gu, H.-M.; Lu, B.-F.; Qu, Z.-X. Acta Pharmacol. Sin. 1980, 1 , 48-
(
1
5
2
(
0.
(
4) Ye, B.; Wu, Y.-L.; Li, G.-F.; Jiao, X.-Q. Acta Pharm. Sin. 1991, 26,
28-230.
S0002-7863(97)03080-1 CCC: $15.00 © 1998 American Chemical Society
Published on Web 03/26/1998

Fe(II)-Induced CleaVage of Qinghaosu and DeriVatiVes
J. Am. Chem. Soc., Vol. 120, No. 14, 1998 3317
simpler 1,2,4-trioxane systems. The reports on the cleavage
have been accumulating at an increasing rate. Mechanistic
understanding, however, has not kept pace, partially because
the results of different groups are divergent (at least superfi-
cially) and partially because the interpretations of the experi-
mental data are not unequivocal (better alternatives exist) and
the mechanisms proposed contain unexplained or even question-
able steps. Since the knowledge of the in vitro cleavage may
play a critical role in elucidation of the parasiticidal mechanism
of this new class of drugs, further experimentation as well as
scrutiny of previously reported experiments and mechanistic
interpretations in the literature is warranted.
Scheme 1
Results and Discussion
We have been engaged in a long-term project on QHS
chemistry dealing with the structure, reactions, and synthesis
10
of QHS and its derivatives, as well as analogues. In the late
11
1
980s, the publications on DNA damage caused by the Fenton
reagent (Fe(II)/H2O2) caught our attention. The chemical
similarity between the Fenton reagent and Fe(II)/QHS and the
high concentration of iron in the parasitized red blood cells
inspired us to explore the reaction of 1 (and its derivatives)
with Fe(II). In this paper, we detail12 our findings and present
the current understanding of the cleavage mechanism as well
as its relevance to the parasiticidal activity of QHS-type drugs
at the molecular level.
Our cleavage reactions were carried out in aqueous acetoni-
8
trile (H2O/CH3CN 1:1, pH 4) containing 1 equiv of FeSO4,
significant changes in the product composition or yields, either.
If the FeSO was replaced by FeCl , however, the reactions
simply did not occur at all (cf. ref 8).
conditions closer (compared with THF) to the biological ones.
In this system at room temperature under a nitrogen atmosphere,
could be easily cleaved. Addition of a phosphate buffer (to
adjust the medium acidity to pH 6-7) slowed the reaction
significantly (probably by affecting the acid-catalyzed rear-
rangements, vide infra) but did not result in any other changes.
Introduction of EDTA or varying the amount of FeSO4 (from
catalytic amounts to more than 1 equiv) did not lead to any
4
3
1
The cleavage of 1 with FeSO4 (Scheme 1) gave two major
products, 5 and 6 (in 25% and 67% yields, respectively), which
13,14
had also been detected in the extract from Artemisia annua
1
5
16
and in the products of pyrolysis and metabolism of QHS.
Repeated chromatography of the fractions containing minor
components afforded another crystalline product (accounting for
ca. 1∼2% yield), which was later identified as epoxide 7, a
(
6) (a) Posner, G. H.; Cumming, J. N.; Ploypradith, P.; Oh, C. H. J. Am.
6a
Chem. Soc. 1995, 117, 5885-5886. (b) Posner, G. H.; Wang, D.-S.;
Cumming, J. N.; Oh, C. H.; French, A. N.; Bodley, A. L.; Shapiro, T. A.
J. Med. Chem. 1995, 38, 2273-2275. (c) Posner, G. H.; Park, S. B.;
Gonzalez, L.; Wang, D.-S.; Cumming, J. N.; Klinedinst, D.; Shapiro, T.
A.; Balci, M. D. J. Am. Chem. Soc. 1996, 118, 3537-3538. (d) Posner, G.
H.; Wang, D.-S.; Gonzalez, L.; Tao, X.-L.; Cumming, J. N.; Klinedinst,
D.; Shapiro, T. A. Tetrahedron Lett. 1996, 37, 815-818. (e) Posner, G.
H.; Gonzalez, L.; Cumming, J. N.; Klinedinst, D.; Shapiro, T. A.
Tetrahedron 1997, 53, 37-50. (f) Posner, G. H.; Oh, C. H. J. Am. Chem.
Soc. 1992, 114, 8328-8329. (g) Posner, G. H.; Oh, C. H.; Wang, D.; Gerena,
L.; Milhous, W. K.; Meshnick, S.; Asawamahasadka, W. J. Med. Chem.
compound Posner has speculated as an intermediate that might
be responsible for the parasiticidal activity due to its potent
alkylating property. The existence of 7, however, was assumed7a
to be “difficult to prove” because of its intrinsic instability.
Together with 7 were isolated some other components of
higher polarity as a low-melting mass (8). Acetylation of this
mixture with acetic anhydride led to a single crystalline acetate
9
(83% yield), whose configuration at C-4 was established
according to the NOESY spectroscopy (a cross-peak correlating
the H’s at C-4 and C-5a was observed). The configuration at
C-4 immediately after the formation of the THF ring should be
as shown in structure 10, but under the reaction conditions, the
less stable isomer (10, E ) 32.26 kcal/mol, as calculated by
1
994, 37, 1256-1258.
(
7) (a) Avery, M. A.; Fan, P.-C.; Karle, J. M.; Bonk, J. D.; Miller, R.;
Goins, D. K. J. Med. Chem., 1996, 39, 1885-1897. (b) Avery, M. A.;
Mehrotra, S.; Bonk, J. D.; Vroman, J. A.; Goins, D. K.; Miller, R., J. Med.
Chem. 1996, 39, 9, 2900-2906. (c) Avery, M. A.; Mehrotra, S.; Johnson,
T. L.; Bonk, J. D.; Vroman, J. A.; Miller, R. J. Med. Chem. 1996, 39,
4
149-4155.
+
MM method using Hyperchem 4.5 program) is converted into
(8) (a) Haynes, R. K.; Vonwiller, S. C. Tetrahedron Lett. 1996, 37, 253-
the more stable one (9, E ) 30.19 kcal/mol), although so far
we do not know whether this inversion occurred before or after
the acetylation. It should be emphasized here that no deoxyQHS
2
2
56. (b) Haynes, R. K.; Vonwiller, S. C. Tetrahedron Lett. 1996, 37, 257-
60.
(9) O’Neill, 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-
1
7
(
2) was obtained, although this compound is often found as
4
6
266.
(
10) Wu, Y.-L.; Li, Y. Med. Chem. Res. 1995, 5 (8), 569-586.
(
11) (a) Imlay, J. A.; Chin, S. M.; Linn, S. Science 1988, 240, 640-
(13) Wei, Z.-X.; Pan, J.-P.; Li, Y. Planta Med. 1992, 58, 229-230.
(14) Tu, Y.-Y.; Ni, M.-Y.; Zhong, Y.-R.; Li, L.-N.; Cui, S.-L.; Zhong,
42. (b) Imlay, J. A.; Linn, S. Science 1988, 240, 1302. For recent reports
in this respect, see: (c) Luo, Y.; Henle, E. S.; Linn, S. J. Biol. Chem. 1996,
71, 21167-21176. (d) Henle, E. S.; Luo, Y.; Gassmann, W.; Linn, S. J.
Biol. Chem. 1996, 271, 21177-21186.
12) For the preliminary communication, see: Wu, W.-M.; Yao, Z.-J.;
Wu, Y.-L.; Jiang, K.; Wang, Y.-F.; Chen, H.-B.; Shan, F.; Li, Y. J. Chem.
Soc., Chem. Commun. 1996, 2213-2214. The results on the DNA damage
caused by this system will be published elsewhere.
M.-Q.; Wang, X.-Z.; Liang, X.-T.Planta Med 1982, 44, 143.
(15) Lee, I.-S.; Hufford, C. D. Pharmacol. Ther. 1990, 48, 345-355.
(16) (a) Luo, X.-D.; Yeh, H. J. C.; Brossi, A. Heterocycles 1985, 23,
881-887. (b) Lin, A.-J.; Klayman, D. L.; Hoch, J. M.; Silverton, J. V.;
George, C. F. J. Org. Chem., 1985, 50, 4504-4508.
2
(
(17) Li, Y.; Yu, P.-L.; Chen, Y.-X.; Zhang, J.-L.; Wu, Y.-L. Kexue
Tongbao 1986, 31, 1038.

3318 J. Am. Chem. Soc., Vol. 120, No. 14, 1998
Wu et al.
In our preliminary communication,12 we presented a sketch
of the possible mechanism. While it was compatible with the
main features of the experimental outcome, many details were
unexplained. Having noticed this and similar problems in the
work of others, we have reconsidered all of this, taking into
account as much relevant information in the literature as
possible. We have now worked out a unified mechanistic
picture as outlined in Scheme 4, which seems to cover all known
facts. For the convenience of discussion, in the following, we
shall refer to this picture as “the main mechanism” and the
products obtained in the cleavage of QHS with FeSO4 as “ the
normal products”.
Table 1. Results of Cleavage of 1 and Derivatives with FeSO
Aqueous CH CN
4
in
3
compd
products (yields, %)
1
2
3
4
5
6
1
11
4a
4b
4c
4d
5 (25)
12 (37)
14a (45)
14b (39)
14c (46)
14d (59)
6 (67)
13 (45)
15a (23)
15b (56)
15c (25)
15d (25)
7 and 8 (<10)
15b (4, R + â)
15b (25)
a
a
Hydrolysis of 14b led to 16.
It has been agreed by most workers in this field that the Fe-
(II)-mediated cleavage of 1,2,4-trioxanes begins with a single
electron transfer (SET) from a Fe(II) ion to the peroxy bond,
resulting in two radical anions (Scheme 4, exemplified by QHS),
1A (O-1 radical) and 1B (O-2 radical), each having its own
routes to evolve further to give the final products. It has been
assumed by others that the two radical anions were irreversibly
generated and were not interchangeable as established from the
corresponding end product ratio. This presumption leads to a
serious problem: Why does modification at an atom (e.g., C-10)
remote from the peroxy bond so strongly affect the ratio at which
the two radical anions are generated? Up to now, all mecha-
nistic work has only tried to rationalize the formation of the
main products obtained in a given run without addressing the
deviation in the ratios and the types of the products observed
under different conditions or/and with different substrates.
Scheme 2
5c
Jefford et al. have mentioned such a concept of interchange-
able radical anions purely for heuristic purposes, neither giving
the arguments for its existence nor employing the concept in
their interpretation. In rationalizing the divergent product ratios
reported by different authors, we find that, by assuming that
radical anions 1A,B are rapidly interchangeable, either through
reversible cleavage of the peroxy bond or, more likely, through
direct SET between the two oxygen atoms (or an oxy radical
substitution at the Fe atom), many “conflicting” results can be
easily interpreted.
Scheme 3
We have also assumed that the steps after formation of 1A
or 1B (before the unpaired electron is eliminated from the
substrate molecule through either substitution or â-scission) are
reversible. This is not often encountered in radical chemistry,
presumably because in most cases the reactions take place
intermolecularly; the resultant radical is no longer in the same
molecule as the starting radical.
Owing to the much higher reactivity of free radicals than oxy
anions, the subsequent transformations after generation of the
radical anions should be predominated by the radical reactions.
Thus, 1A rearranges by an intramolecular 1,5-hydrogen shift
to produce a secondary carbon-centered radical (17). This is
one of the earliest proposed transformations in the cleavage
chemistry of QHS and has been widely accepted by the workers
the predominant product when using many other reducing
agents. This is a major difference between the reaction
conducted in aqueous medium and that in THF.
Artemether (11), a derivative of QHS already on the market
as an antimalarial drug, reacted with FeSO4 similarly, giving
5
c
in this field. Recently, however, some authors question this
,5-H transfer process on the basis of calculated distances
1
between relevant atoms. The main argument used to reject
1,5-H transfer is that the distance between O-1 and H-4 (i.e.,
the H at C-4) in the conformation of lowest energy exceeds the
critical distance of 2.1 Å. Conformational adjustments, though,
could narrow the gap but would never realize the desirable O-1,
H-4, and C-4 collinear arrangement. Houk’s calculation is also
cited as a supporting argument: The 1,5-H transfer requires a
boatlike transition state (of high energy) and therefore the radical
17 would be unattainable.
1
2 and 13 in 37% and 45% yields, respectively (Scheme 2). In
addition to these two major products, a small amount of the
demethyl product (15b) was also obtained (ca. 4%). Artesunate
(4a), another marketed antimalarial drug, and some other
derivatives of QHS were also cleaved by FeSO4 to give
compounds 14 and 15 (Scheme 3). Again, neither deoxy
derivatives of 11 or 4, nor 4â-hydroxy isomers of 13 or 15 were
detected. The results are summarized in Table 1.

Fe(II)-Induced CleaVage of Qinghaosu and DeriVatiVes
J. Am. Chem. Soc., Vol. 120, No. 14, 1998 3319
Scheme 4
secondary C-4 radical (17). However, there is so far no direct
evidence that the trapped radical contains iron. It deserves to
be emphasized here that, although such a species6g was
postulated several years ago and supporting evidence for its
involvement in the reaction has been reported, this is very first
direct evidence for the existence of a secondary carbon-centered
radical in the cleavage reaction.
During the preparation of this manuscript, Robert and
1
8
Meunier’s work appeared in the literature. They isolated and
characterized an adduct that contained a moiety of QHS structure
corresponding to the primary C-4 radical 18. Although strictly
speaking their results do not provide any direct evidence for
the involvement of a radical, it was possible that the adduct
was formed via radical addition under the given circumstances.
The secondary C-4 radical (17) may have several pathways
for further reaction. A â-scission between C-3 and O-2 would
give enol 19, which could be easily converted to deoxyQHS
Figure 1. ESR signal recorded in a run in aqueous CH
3
CN with 1 as
the substrate in the presence of 1 equiv of FeSO
agent (cf. text and the Experimental Section).
4
with MNP as trapping
(
2) by intramolecular addition of the C-12a OH to the enol
2+
double bond. During this process, OdFe(IV) (i.e., OdFe ),
a species that has been attracting a great deal of attention as
We find these arguments are not appropriate for rejecting
the 1,5-H transfer. The collinear arrangement has never been
a prerequisite for the 1,5-H transfer. Probably no 1,5-H
transfers, including those well-established cases of Norrish Type
II process, ever occurred via a collinear transition state. Houk’s
theory (boatlike TS) was derived for open-chain systems. In
QHS, the boatlike conformation already exists in the starting
entity and only slight twisting of the ring would meet the
requirement for the 1,5-H transfer. The energy barrier would
not be as high as in an open chain starting species.
We have also obtained firm experimental evidence for the
proposed secondary C-4 radical. In a run with QHS (1) as the
substrate under the same conditions as aforementioned, we
added the spin-trapping agent 2-methyl-2-nitrosopropane (MNP)
to the reaction mixture and recorded the ESR spectrum (Figure
6
possible basis of parasiticidal action, is expected to form.
2+
Alternatively, radical substitution at O-2 with Fe as the leaving
group would lead to epoxide 7. Due to its intrinsic instability,
7
may be converted to several secondary products by acid (either
protonic or Lewis acids) catalyzed rearrangements. Note that
2+
the Fe is regenerated here, which accounts for the observation
2
+
that a catalytic amount of Fe added in the beginning could
work equally well. Finally, the C-4 radical may also abstract
a hydrogen intermolecularly (if there exists a good hydrogen
donor, such as 1,4-cyclohexadiene, in the reaction mixture) to
afford intermediate 20, which could be easily transformed into
+
2
via 21, either via O-13 assisted elimination of OdFe or by
hydrolysis of the Fe-O bond followed by elimination of water
the process of forming ketal from hemiketal). The observation
(
1). The detected signal showed additional splitting besides that
6
by Posner and co-workers that the presence of an excess of
,4-cyclohexadiene led to significantly increased amounts of 2
caused by the coupling to the nitrogen; each line of the triplet
was further split into two, indicating that the observed species
1
is most likely due to facilitated intermolecular hydrogen
•
was t-Bu(R1R2CH)NO . Thus, the trapped radical was a
secondary one (bearing only one proton), which fit best the
(18) Robert, A.; Meunier, B. J. Am. Chem. Soc. 1997, 119, 5968-5969.

3
320 J. Am. Chem. Soc., Vol. 120, No. 14, 1998
Wu et al.
Table 2. Published Product Ratios of Iron-Induced Cleavage of
Scheme 5
QHSa
product ratios
reagent
FeBr2
FeBr2
FeCl2‚4H2O
FeSO4
solvent
THF
THF/CHD 17
2
6
5
remarks sources
6
6
6
a
a
a
1
2
3
4
6
1
1
1
3
4
6
Posner
Posner
Posner
CH3CN
aq CH3CN
0
0
>67
25 6 plus
this work
7
& 8
8
5
6
7
8
FeCl2/imid
FeCl2‚4H2O
CH3CN
CH3CN
6
(0)
(0)
16
17
78
78
Haynes
Jefford
Jefford
5
5
c
c
FeCl2‚4H2O/CyH CH3CN
1∼8 84
6
a
hemin/BnSH
THF
0.2
1
11
Posner
a
abstraction (and of course the intermolecular hydrogen abstrac-
tion by 1A,B), a process the authors6a confusingly termed a
CHD ) 1,4-cyclohexadiene; CyH ) cyclohexene; imid ) imida-
zole. The product ratios of Posner were determined by NMR, while
others were isolated yields.
“
radical trapping experiment”. This explanation is much more
5c
convincing than the alternative that the increased amount of
was due to transfer hydrogenation with 1,4-cyclohexadiene
components (8) are more difficult to conclusively identify
because they were obtained as a mixture. However, since they
all can be converted to 9 (whose structure has been undoubtedly
established) by acetylation, all of these compounds must be
derived from 7 with an oxygen attached to C-4 in such a way
that the oxygen can leave easily under the acetylation conditions
while the OH at C-12a cuts in to form the THF ring.
With the knowledge of the above main mechanism, we are
in a position to see if the mechanism described above can
rationalize the divergent experimental outcomes reported in the
literature. Let’s start with QHS since most groups have
examined its cleavage. For the convenience of discussion, we
have gathered relevant literature data in Table 2.
2
as the H2 donorsthe absence of a hydrogenation catalyst was
overlooked!
Bromide ion from FeBr2 in THF has also been proposed5c as
a reducing agent, which might be responsible for the formation
of 2. However, the reduction potentials19 for Br2/2Br and QHS
-
8
b
do not seem to support this proposal. Haynes and Vonwiller
have also reported that ZnBr2 in diethyl ether could not induce
any reaction.
The O-2 radical (1B), unlike the O-1 radical which has an
R-hydrogen atom (at C-4) available for abstraction, reacts
predominantly via â-scission. Since elimination of a methyl
radical is disfavored, the possible â-scissions available to 1B
are those leading to 22 and 18. The scission of C-3/O-13 bond
affords 22, a species with at least two alternatives at the
subsequent step. In the case of QHS (1), scission of C-12 and
O-11 wins out, as no products derived from 23 were detected.
To decipher these data, at least two factors must be
considered. One is the effect of the counterion, which we
believe plays a role in reaction selectivity. FeCl2‚4H2O seems
to be the most active one. The reaction with this catalyst
usually takes only a few minutes. FeBr2 is probably a little
6a
(Note that conversion of 22 into 23 can be regarded as an
6a
less reactive, with which it would take some 10-30 min to
irreversible process because the radical is eliminated from the
molecule). This is an interesting crossroads; which way the
reaction goes seems to be dictated by the nature of the
substituents at C-10. When a carbonyl is present as in 1, the
cleavage leads to 24, giving a carboxyl radical, a better leaving
group than alkoxide radical due to delocalization of the unpaired
electron. The two successive scissions (1B to 22 then to 24)
might well occur in a concerted manner. When C-10 is a methyl
or substituted methyl group, the scission corresponding to that
leading to 23 prevails (vide infra). Another possible route for
finish the reaction. The activity of FeSO4 is obviously lower;
several hours (often overnight) is normally required to consume
all of the starting material. Fe(ClO4)2 has been reported to
8b
be completely inactive. This trend might simply be a conse-
quence of the increasing number of oxygen atoms in the anion,
2+
which reduces the probability for Fe to deliver one electron
to the peroxy oxygen (breaking the peroxy bond) and that for
3+
Fe to receive one electron from the formal alkoxide (breaking
the Fe-OR bond) and thus slow the reactions.
Solvent may also contribute to the divergence of the results.
In THF (less polar than CH3CN), the Fe-OR bond is probably
1
B to take is the scission of the C-3/C-4 bond. This is also
supposed to be a rapid reversible process. Radical substitution
at the O-1 (leading to 5, a step that has been part of Jefford’s
mechanism5 although it was put in a different way without
indicating that this is a radical substitution at oxygen), however,
is irreversible and likely to be rate-limiting.
“
stronger/tighter” than in CH3CN (especially the aqueous (aq)
one). Breaking of the Fe-OR bond (e.g., in the formation of
and 7) may thus become more difficult and the â-scission
breaking the FeO-R bond) becomes more significant. In aq
c
5
(
CH3CN, the oxygen in the Fe-OR bond is probably more like
a solvated alkoxide anion, which renders the â-scission (formally
elimination an O atom radical) very difficult. This explains
why the major product in THF (entry 1, Table 2), 2, is not
The unstable epoxide 7 may have a variety of ways to
rearrange in the presence of protonic or Lewis acid catalysts.
The formation of 6 (Scheme 5) is probably the simplest.
However, it deserves to be pointed out that 6 is most likely
formed by O-13 assisted cleavage of the O-2/C-3 bond, followed
by attack of O-1 as shown in Scheme 5 (This process is similar
to the exchanging of alkoxy groups of a ketal). The other minor
formed at all in CH CN (entries 3, 4, 6, and 7).
3
In our case (FeSO /aq CH CN, entry 4), the less active form
4
3
2
+
of Fe (compared with entry 1) would slow the steps of 17 to
7 and 18 to 5, while the solvent effect of aq CH CN works in
3
the opposite way. Hence the net effects at these two steps are
(
19) Reduction potantials: qinghaosu, 0.87 V; dihydroqinghaosu, 1.05
almost null. But, at the step 17 to 19, the solvent effect is not
counterbalanced by the reagent effect; in aq CH3CN, this path
is shut off (all 17 formed from A is converted to 7, and in turn
to 6, etc.). For these reasons, the ratio of 5/(2 + 6 + 5) is
more or less the same in both cases, but the ratios of 2/6 differ
greatly. With FeCl2‚4H2O/CH3CN (entry 6), the solvent effect
V; arteether, 1.33 V; artemether, 1.50 V. See: Jiang, H.-L.; Chen, K.-X.;
Tang, Y.; Cheng, J.-Z.; Li, Y.; Wang, Q.-M.; Ji, R.-Y.; Zhuang, Q.-K. To
be published. Cf. also the reduction potentials (taken from CRC Handbook
of Chemistry and Physics, 73rd ed.; Chemical Rubber Co.: Boca Raton,
-
3+
2+
FL, 1992) for the following: Br2/2Br , 1.087 V; Fe /Fe , 0.771 V; I2/
-
3+
2+
2
I , 0.536 V; Sm /Sm , -1.55 V; the latter three are able to reduce the
peroxy bond in QHS.

Fe(II)-Induced CleaVage of Qinghaosu and DeriVatiVes
acceleration of the steps 17 to 7 and 18 to 5) is probably weaker
J. Am. Chem. Soc., Vol. 120, No. 14, 1998 3321
(
Scheme 6
than that with aq CH3CN. However, this time, it is not canceled
by the reagent (anion) effect; the product distribution pattern
changed. The “O-2 product” 5 now has become the major
product. Why would the “O-2 route” predominate over the “O-1
route” in this case? A likely reason is that 18 to 5 is the rate-
limiting step of the “O-2 route”, whereas the 1,5-H shift (which
is not affected by the solvent effect) in “O-1 route” may be not
as fast as the C-3/C-4 bond breaking/reforming and thus
becomes part of the rate-limiting steps.
Anhydrous FeCl2 has been found5b to be less active than its
tetrahydrate. However, Haynes’ reaction (entry 5) took only 5
min to complete. This may reflect the effect of imidazole. It
seems to be likely that imidazole may form a complex with the
iron ion (resembling the situation in heme), at least a small
portion of it, and stabilizes both imid-Fe (this formally makes
2+
Fe a better leaving group) and imid-Fe-O species (this
2+
formally makes imid-Fe dO a better leaving group in the
â-scission) by delocalization of the charge or radical. Again,
due to the difference in the rate-limiting steps, the “O-2 route”
would be more sensitive to the acceleration than the “O-1 route”
and 5 would be the major product. Since both epoxidation and
â-scission are accelerated, some 2 would still form even in CH3-
CN.
The outcomes8a of the reaction catalyzed by hemin seem to
be very anomalous. It appears to us that the effects of
complexation of the iron ion with porphyrin in hemin are similar
to those described above for imidazole. There are, however,
some additional complications when using hemin as reagent.
3+
For instance, hemin (Fe ) must be reduced (by, for example,
2+
a thiol) to heme (Fe ) before triggering the main mechanism.
The reactions may go rather fast (finished in a few minutes).
Therefore, the way of mixing the reactants (i.e., starting with
how much heme?) might affect the outcome of the reaction;
unlike simple Fe3 species (e.g., FeCl3 alone), hemin may react
7
a
in the literature ) as discussed for QHS. The difference at C-10
changes the relative rates in the “O-2 route” and consequently
the main products (compared with QHS). The “O-1 route” is
more or less the same as that of QHS. Since the radical
substitution at oxygen is relatively slow in THF, the â-scission
manifests itself, giving the enol 26 and then 27 (cf. conversion
of 19 to 2) in 8% yield.
+
8
to some extent on its own as suggested by Haynes before it is
reduced to heme and turns on “the main mechanism”. The
presence of a thiol (good hydrogen donor) may also activate
the intermolecular hydrogen abstraction pathway and thus
change the product ratio. For instance, owing to the presence
of thiol, the radicals involved in the main mechanism (1A, 1B,
and 17) may abstract a hydrogen from the thiol and eventually
lead to increased amounts of 2. Finally, the additional func-
tionalities present in hemin and thiol may contribute to the low
recovery (forming, e.g., heme adduct) as well.
The “O-2 route” is entirely different now (different sub-
branches are now in operation). Unlike in QHS case (Scheme
) where 22 is mainly converted to 24 (or, conversion of 1B to
2 then to 24 occurs in a concerted manner due to the relative
ease of breaking C-12/O-11 bond), in 3B, the scission of the
C-12/O-11 bond becomes a disfavored process (because alkoxy
groups are much poorer radical leaving group compared with
carboxy groups, or it can be interpreted as that the reverse, i.e.,
the addition of the alkoxy radical to the aldehyde, is rather fast)
4
2
The QHS derivatives (Table 1) we have examined fall into
two groups, one includes 11 and 4b, both having a relatively
small substituent at C-10. These two compounds gave more
so that the reaction takes another path: breaking the C-12/C-
2a bond followed by leaving of Fe ion. Again, these three
“
6-type” products than “5-type” products in the cleavage
2+
1
reaction (similar to QHS). The situation with the other three
compounds (4a,c,d), all carrying a larger group at C-10, is
reversed, with the “5-type” product predominating over the “6-
type” one. We currently attribute the decreased amount of “6-
type” products to a slower 1,5-H shift that may result from the
conformational changes caused by the large substituents.
â-scissions may well occur concertedly at an overall rate not
only much higher than that for 3B to 32 (the primary C-4 radical
needs time to move to the right position to attack O-1, whereas
the three â-scissions may occur without any movements of the
atoms) but also remarkably higher than the overall rate for the
“
O-1 route”, so that 34 becomes the predominant product.
7
7a
Avery and co-workers have conducted a large amount of
The two carba-analogues (35 and 36) of QHS made by
elegant work in the structure modification of QHS. However,
their mechanistic studies did not lead to convincing interpreta-
tions. Their mechanisms rely heavily on a less common 1,3-H
shift, a process the authors themselves have questioned. We
have found that all their results7 can be easily explained by
our main mechanism. Thus, 3 is cleaved by FeBr2 to afford
corresponding “O-1 radical” 3A and “O-2 radical” 3B (Scheme
Avery et al. gave interesting results in the FeBr2-induced
cleavage reactions. The first one (35) afforded alcohol 37 as
the predominant product (in 79% yield, Scheme 7). After
establishment of the structure for 37 by spectroscopic means,
they concluded that the radical of 38 did not produce epoxide
39 and attributed the formation of 37 to an uncommon 1,3-H
shift. As a matter of fact, this product could well be derived
from 39 as proposed in Scheme 7. Thus, the epoxide ring
a
6
, note that the configurations at C-12a were incorrectly drawn

3322 J. Am. Chem. Soc., Vol. 120, No. 14, 1998
Wu et al.
Scheme 7
Scheme 8
opening by the bromide results in 40a. Although most of 40a
is expected to be converted back to 39, some of the 40a may
undergo Fe2 transposition (ligand exchange) to afford 40b,
which would soon lead to 37 due to the favorable position of
the C-12a OH for both the ligand exchange and the nucleophilic
substitution. The reaction ran for 8 h, which provided enough
time for the overall transformation mediated by trace amounts
of 40a,b.
+
When position 13 is an oxygen, the epoxide opening is
dominated by the powerful ketal exchange mechanism as
discussed for the 7 to 6 transformation (Scheme 5). Replace-
ment of oxygen with a carbon shuts this path off, providing a
chance for the slower reaction caused by the bromide to
manifest. This is why two epoxides (7 vs 39) gave different
products under the same conditions. The other carba-analogue
Scheme 9
(36), differing only at the configurations at C-3, C-12, and
C-12a, produced entirely different products (Scheme 8). At first
sight this seems to be very strange because these two compounds
are so similar to each other. We found the clue by inspection
of the models of these two molecules: The inversion of the
configurations makes the 1,5-H transfer impossible, unless the
seven-membered ring flips to turn the â-hydrogen at C-4 from
the quasi-equatorial position to the quasi-axial position. The
ring flip certainly takes time, and therefore, the rate for 36A to
known aspects of free radical chemistry. Perhaps the elimination
reaction barrier for forming an alkene (41) is higher than that
for corresponding enol (26); when the â-scission is slow, most
of 38 would of course take another way out to form epoxide
39. Similarly, the ketone 50 is probably more reactive than
ester 18, so that the generation of the C-4 radical in the case
of, e.g., 35B becomes a disfavored process (the reverse is too
fast). This explains why no 51 was formed.
4
2 is remarkably lower than that for 35A to 38. Consequently,
6f
the yield of epoxide 44 is much lower than that of 39 (37). It
deserves to be mentioned here that the back of the epoxy ring
in 44 is blocked by the H atoms at C-12 and C-5a. This makes
the bromide attack impossible, and hence, no 45 is formed.
Without competition of a rather fast “O-1 route”, the “O-2 route”
now has the chance to operate. Once again, since the scission
of the C-12/O-11 bond is a disfavored process (cf. the discussion
for conversion of 30 to 31), the reaction is directed to the
elimination of Fe2 from O-1 via three â-scissions (that of C-3/
C-13, C-12/C-12a, and Fe/O-1) and finally yields 49.
Posner and co-workers have prepared a simplified analogue,
52, and examined its cleavage under various conditions (Scheme
10). One of the pathways (the non-iron one-electron reduction
path) proposed by them obviously does not fit in with our
mechanistic picture. This led us to look into the work. And
after closer scrutiny we have found that the discord stems from
a difference in interpretation. The main problem with the
previously proposed mechanism is the requirement for a highly
reactive radical to persist while the more stable anion performs
a series of transforms. This appears to be irrational to us. The
reason for which the authors chose the anion mechanism instead
of the normal radical one was that 53 was not formed during
the cleavage reaction. However, the failure of the intramolecular
alkylation may have other causes than the absence of alkoxide
54. The counterion (cation), solvent, and temperature may all
+
It is noted that, in the cleavage of the carba-analogues, there
are no products from the â-scission of the secondary C-4 radical
(e.g., 38 did not give 41, Scheme 7) and substitution of the
primary C-4 radical at O-1 (e.g., 35B did not produce 50,
Scheme 9). These phenomena may reveal some not very well

Fe(II)-Induced CleaVage of Qinghaosu and DeriVatiVes
J. Am. Chem. Soc., Vol. 120, No. 14, 1998 3323
Scheme 10
of the structure of the substrate, the anion, the cation, the solvent,
and the temperature. Now we shall turn to its relevance to the
molecular-level antimalarial mechanism of QHS-type drugs. It
has been generally accepted that, in hemin-induced cleavage
of QHS and related compounds, some cytotoxic species is (are)
produced that is (are) directly responsible for the antiparasitistic
activity. The question is what is it (or what are they)?
6
One of the suggestions is carbon-centered radicals. Later,
OdFe(IV) and epoxide 7 (as a potential alkylating agent) were
6
also thought to be possible killers. The possible existence of
the high-valence iron species OdFe(IV) initially was probably
deduced by logic from the â-scission of secondary C-4 radicals
such as 17 and has never been proven beyond all doubt. The
evidence available so far includes the rearrangement of hex-
amethyl Dewar benzene (HDB) to hexamethylbenzene (HB),
oxidation of a sulfide to sulfoxide, and oxidation of tetralin to
hydroxytetralin; none of them is necessarily bound to a OdFe-
(IV) participated reaction. For instance, rearrangement of HDB
to HD may be initiated by losing one electron. The receiver of
this electron could be, e.g., 1A (or 1B); this would lead to
increased relative amount of 20 and eventually 2 (product ratio6a
for 6:2:5 changed from 1:6:3 to 1:11:3). As for the latter two
cases, there is no reason to say the oxidant must be OdFe(IV).
It ought also to be noted that there exists no knowledge about
the oxidation power of this so far speculative species. We only
know about its counterpart in hemin, which is not necessarily
the same. The results from nonhemin model systems have
revealed a trend that in aqueous media the â-scission that is
expected to generate OdFe(IV) may not occur at all (since no
2 was formed)! After all, in the parasite, the true species would
be oxyheme, not simple OdFe(IV). Digging very hard in non-
hemin model systems for the existence of OdFe(IV) would not
help in understanding the antimalarial mechanism.
+
+
contribute. A “wrong” cation (other than, e.g., Li or Na ), an
improper solvent, or too low temperature may all lead to such
failures. It should be noted that some of their reducing agents
5
c
(
e.g., Zn metal ) may only cause two-electron reduction.
The only thing requiring explanation here is that, if the
cleavage is initiated by a single-electron reduction, why 54
would form as the predominant product (this requires excising
a methoxy radical, a process that normally does not occur) while
other common “O-1 and O-2 products” were not obtained. A
possible answer is shown in Scheme 10. We must note that
the formation of the common products relies on the facile
interchange between the two low-lying energy states of iron
Fe /Fe ). If the Fe is replaced by other metal ions (M ),
both the â-scission (elimination of OdM from 55) and radical
substitution at oxygen atom (forming epoxide or THF ring
The speculative alkylating power of 7 is another thing to be
questioned. Posner et al. got the idea that 7 might be an
21
alkylating agent from the easy formation of C-O or C-N
bonds by reaction of ketal epoxides (at the carbonyl carbon of
the ketals) with alcohols or amines. Those ketal epoxides,
however, do not have a favorably positioned built-in OH group
to compete with the “outcomers” for C-3. Considering the ease
with which it rearranges into other products, we believe that
epoxide 7 would make a poor alkylating agent. On the basis
of their biological evaluation of 36 and 44 (2/3 of QHS′ activity
and null, respectively), Avery et al. have also concluded that
the epoxide ring is not an essential part for the antimalarial
activity. Their results rule out the possibility for epoxides as
alkylating agents by reacting at C-4: in the carba-analogues,
the C-3 is far not so reactive as that (carbonyl carbon) in, e.g.,
3
+
2+
3+
n+
(
n+
(n-1)+
accompanied by elimination of M
) may be drastically
retarded.20 Thus, radicals 56 and 55 would go nowhere except
for returning to 52B and 52A, respectively. Under these
circumstances, the ordinarily very slow, insignificant process
of excising alkoxy radical has the opportunity to manifest itself
via scission of C-3/O-13 and C-12/OMe (for convenience, we
still use QHS numbering here). The intramolecular rearrange-
ment of 57 leads to 54 (formally by the attack of the O-1
alkoxide at ketone carbonyl and subsequent attack of O-2 at
the aldehyde). If the ion pairs in 54 are not well solvated
1
and thus does not need to be considered. The C-4 in 39 ought
to be much more reactive than in 44. However, 36 shows higher
antimalarial activity than 35 (44 is inert). Taking all these as
a whole, it seems that epoxide 7 is unlikely to be the “killer”.
The carbon-centered radicals generated in the cleavage remain
the most likely species that are responsible for the parasiticidal
activity. As shown in the schemes, there are two types of carbon
radicals at C-4: one (primary) is formed from â-scission of O-2
radical (e.g., 18) and the other (secondary) is produced by 1,5-H
transfer (e.g., 17). Our spin-trapping experiment in aqueous
CH3CN only recorded the latter. This is compatible with the
fact that the relative amount of 17 produced in our system is
much more than that of 26 (since the combined yields of 6, 7,
and 9 are much higher than the yield of 5). Robert and
Meunier’s results point to the more reactive primary radical 18,
(
“wrong” cation or poor solvent) or the medium temperature is
too low, the reaction would stop at 54 (tosylates are far not as
reactive as carbonyls).
Up to this point, we have explained almost all of the known
results of the iron-induced cleavage of QHS and derivatives/
analogues and shown that behind the diverse outcomes of
experiments under different conditions, there exists a relatively
simple mechanism framework. We have also shown how the
particular mechanisms (i.e., particular sub-branches within the
mechanism framework) for a given case are rationalized in terms
(
20) A stronger reducing agent may reduce the peroxy bond more easily
than Fe , but later, it may be much more difficult to get back one electron
2
+
2+
from the oxygen. For example, radical substitution at O-Sm requires Sm
to leave. This is an Sm3+ f Sm reduction with a reduction potential of
2+
3
+
2+
-
1.55 V, much more difficult than Fe f Fe (0.771 V).
(21) References 13 and 23 in ref 6a.

3324 J. Am. Chem. Soc., Vol. 120, No. 14, 1998
Wu et al.
but their reaction was done with the manganese counterpart of
hemin in CH2Cl2, where the situation probably differs very much
from that in our system. It is possible that the adduct they
obtained was formed from an intramolecular radical addition
to the porphyrin. The primary C-4 radical has the unpaired
electron at the end of the two-carbon side chain that can stretch
out to reach the right target carbon in the porphyrin. The
secondary one does not have this possibility because of its rigid
ring system. This might explain why they did not get the other
adduct.
The radical trapping experiments have confirmed the pos-
sibility for the highly reactive radicals generated in the cleavage
reaction to interact with other molecules. This possibility is a
prerequisite for the radicals to be qualified for the lethal agent
in the antimalarial mechanism. Considerable amounts of work
have already been done to reveal the relationship between the
C-4 radicals’ stability and the antimalarial activity. The
conclusion is not very clear yet because antimalarial activity
may be affected by a number of factors other than the reactivity/
lifetime of the radical. However, the establishment of a unified
mechanism framework for the iron-induced cleavages will
certainly facilitate the process of understanding the mode of
action of QHS-type drugs at the molecular level and provide a
basis for rational drug design efforts.
from the Sigma Chemical Co. Microanalyses were carried out at the
microanalysis laboratories of these Institutes. Flash chromatography
was performed on silica gel H (400 mesh).
Typical Procedure for the Reaction of Qinghaosu and Its
Derivatives with Ferrous Sulfate: Qinghaosu or its derivative (2
mmol) and FeSO
4
(2 mmol) were dissolved in aqueous acetonitrile
(50%, v/v, 20 mL), and the mixture was stirred at 37 °C under a nitrogen
atmosphere, until the starting material was completely consumed (often
overnight) as shown by TLC (petroleum ether/ethyl acetate, 4:1∼2:1
depending on the substrate, visualized with ammonium molybdate).
The brown insoluble substance was filtered off, and the acetonitrile
was removed under reduced pressure (rotary evaporator). The residue
was extracted with ethyl acetate (2 × 10 mL). The combined extracts
4
were washed with brine, dried (MgSO ), and concentrated. Flash
column chromatography of the residue (eluting with petroleum ether/
ethyl acetate, 4:1∼2:1 depending on the substrate) gave the products.
Spin Trapping Experiment: To the solution of qinghaosu (1 mmol)
and ferrous sulfate (1 mmol) in acetonitrile and water (1:1, 10 mL) at
3
.
7 °C was added 1 mg of MNP (t-BuNO) After 1 h its ESR spectrum
was taken. The spectrum (Figure 1) is shown with A
AH ) 0.301 mT, and g ) 2.0059.
N
) 1.481 mT,
Compound 5: colorless crystal (25% yield), mp 92∼94 °C (hex-
anes-ethyl acetate) (91-92 °C,1 91-92 °C ), [R]
+109.2 (c 1.0,
15.5 +93.5 (c 1.0
6b
13
6b
20
D
1
CHCl
CHCl
CDCl
3
3
3
) {[R]
D
20 +109.2 (c 0.109, CHCl
3
), [R]
D
) }; IR (KBr) νmax 1760, 1746 cm-1; 1H NMR (300 MHz,
13
) δ 6.65 (1H, s), 4.21 (1H, t, J ) 7.6 Hz), 3.94 (1H, dd, J ) 7.4,
1
5.8 Hz), 3.16 (1H, dq, J ) 2.7, 7.2 Hz), 2.16 (3H, s), 1.8-2.1 (4H,
m),1.73 (1H, m), 1.60 (1H, m), 1,47 (1H, m), 1.21 (3H, d, J ) 7.3
Hz), 1.08 (1H, m), 0.99 (3H, d, J ) 6.3 Hz); MS(m/z) 283 (M + 1),
Summary
+
1
65, 151, 137 (100), 109, 43.
In the preceding paragraphs, we have detailed our cleavage
reactions of QHS and derivatives catalyzed by FeSO4 in aqueous
acetonitrile, including the isolation and characterization of
epoxide 7, whose existence has been proposed earlier but
thought to be very difficult to prove. These results play a very
important part in analyzing the anion and solvent effects, which
have never been addressed in previous studies, on the product
distribution pattern. We have also presented the very first direct
evidence for the involvement of the earlier proposed secondary
C-4 radical. A unified mechanism framework featuring inter-
changeable radical anions and reversible intramolecular radical
reactions has been suggested for the iron-induced cleavage of
Compound 6: colorless crystal (67% yield), mp 199-201 °C
16b
20
(
(
hexanes-ethyl acetate) (190-192 °C ), [R]
D
3
-131.1 (c 1.0, CHCl )-
16b
-1
1
-131.1, (c 0.119, CHCl
) δ 5.64 (1H, s), 3.63 (1H, d, J ) 7.2 Hz),
.21 (1H, dq, J ) 2.5,7.2 Hz), 2.07 (1H, dt, J ) 4.4, 13.3 Hz), 1.99
1H, dd, J ) 1.67, 9.1 Hz), 1.95 (1H, dq, J ) 3.3, 13.6 Hz), 1.87 (1H,
3
) ); IR (KBr) νmax 3050, 1730 cm ; H
NMR (300 MHz, CDCl
3
3
(
brs), 1.83 (1H, dq, J ) 3.2, 13.6 Hz), 1.58 (3H,s), 1.54 (2H, m), 1.28
(1H, m), 1.21 (3H, d, J ) 7.6 Hz), 1.12 (1H, dq, J ) 3.5, 13.3 Hz),
1.00 (1H,dq, J ) 3.2, 13.3 Hz), 0.94 (3H, d, J ) 6.4 Hz); MS (m/z)
+
283 (M + 1), 265, 247, 222, 150, 137.
Compound 7: colorless crystal (1% yield), mp 148-152 °C
(
(
1
petroleum ether-ethyl acetate); [R]19
D
+78.5 (c 0.14, CHCl
3
); IR-
-
1
+
KBr) νmax 3500, 1728, 1635, 1144, 1022 cm ; MS (m/z) 283 (M +
1
,2,4-trioxanes. It ought to be emphasized that, although some
1
), 222, 204, 151, 150, 149, 137, 122, 121, 107, 93; H NMR (400
pieces (only the forward steps) of the present mechanism
framework have been part of the mechanistic paths proposed
by others, their correctness is rather vulnerable to doubt due to
the lack of explanation for the disappearance of these paths in
some cases and the introduction of uncommon or unlikely
mechanisms in rationalization of the unexpected main products.
The mechanism described above allows rationalization of a
large body of unexplained experimental observations regarding
the formation of individual product as well as the predominance
of particular pathways. Thus, the fuzzy picture of the cleavage
mechanism caused by the lack of an overall elucidation of the
diverse experiments now becomes much better focused. This
would help to save further unnecessary efforts on clarifying the
in vitro cleavage mechanism itself and direct attention to the
relevance of the findings in the model systems to the molecular-
level antimalarial mechanism, the rationale of all the work with
model systems.
MHz, C
6
D
6
) δ 0.91 (3H, d, J ) 6.4 Hz), 1.18 (3H, d, J ) 7.4 Hz),
13
1
.55 (3H, s), 3.17 (1H, m), 3.60 (1H, br), 5.61 (1H, s); C NMR (100
MHz, C
6
D
6
) δ 13.26, 18.93, 21.32, 23.99, 31.12, 33.60, 34.03, 35.22,
4
0.99, 42.74, 69.90, 83.36, 99.53, 109.54, 170.89. Anal. Calcd for
C H O : C, 63.81; H, 7.85. Found: C, 63.73; H, 7.79.
15
22
5
Compounds 8: colorless solid; mp 85-86 °C; [R]2
0
-49.9 (c 1.05,
1 1
D
-
CHCl
MHz, CDCl
3
); IR (KBr) νmax 3320, 2990, 1750, 1720 cm ; H NMR (300
) δ 9.98 (1H, s), 5.98 (1H, d, J ) 13.3 Hz), 5.60 (1H, d,
3
J ) 13.3 Hz), 5.54 (1H, s), 4.74 (1H, dd, J ) 4.2, 10.5 Hz), 4.21 (1H,
dd, J ) 2.14, 7.6 Hz), 4.12 (1H, dd, J ) 3.1, 6.4 Hz), 3.08 (1H, m),
1
.20 (3H, dd, J ) 7.2 Hz), 1.02 (3H, dd, J ) 6.3 Hz); MS (m/z) 283
+
+
+
(
M + 1), 265 (M - H
O) 265.1449, C14
65.1440, 253.1440.
Compound 9 via acetylation of compounds 8: To a solution of 8
2
O), 237, 207; HRMS found C15
22 4
H O (M -
+
H
2
2
H
21
O
4
(M - H
2
O - HCdO) 253.1437, calcd
(25 mg, 0.089 mmol) in dry CH Cl2 (0.5 mL) were added acetic
2
anhydride (31 mg, 0.3 mmol), pyridine (40 mg, 0.5 mmol), and a
catalytic amount of DMAP. The solution was stirred at 25 °C for
several hours and evaporated under reduced pressure. The residue was
dissolved with ethyl acetate, washed, and concentrated to give crude
product, which was recrystallized from ethyl acetate-petroleum ether
Experimental Section
20
ESR spectra were recorded at room temperature on a Varian E-112
spectrometer with X-band, field modulation frequency 100 kHz,
microwave power 20 mw, response time 0.25 s, center field set 3240
to give compound 9 (27 mg, 83% yield): mp 162-164 °C; [R]
D
+55.3
); IR (KBr) νmax 1760, 1720 cm ; H NMR (600 MHz,
) δ 6.66 (1H, s), 4.41 (1H, t, J ) 7.8 Hz), 3.23 (1H, dq, J ) 5.4,
-
1 1
(c 0.95, CHCl
CDCl
3
3
-
1
-2
×
10 T, and sweep width 10 T. The magnetic field was determined
7.4 Hz), 2.38 (3H, s), 2.27 (1H, dt, J ) 6.9 12.2 Hz), 2.02 (1H, m),
2.00 (3H, s), 1.96 (2H, m), 1.75 (1H, dt, J ) 8.4, 13.6 Hz), 1.69 (1H,
m), 1.66 (1H, m), 1.26 (1H, m), 1.24 (3H, d, J ) 7.3 Hz), 1.08 (1H,
1
by an H NMR fieldmeter and the microwave frequency by a superhigh
frequencymeter. 2-Methyl-2-nitrosopropane (MNP) was purchased

Fe(II)-Induced CleaVage of Qinghaosu and DeriVatiVes
J. Am. Chem. Soc., Vol. 120, No. 14, 1998 3325
dq, J ) 4.8, 13.4 Hz), 0.97 (3H, d, J ) 6.0 Hz); 13C NMR (150 MHz,
Compound 14c: colorless crystal (47% yield), mp 86-88 °C
(petroleum ether-ethyl acetate); [R]19
+99.1 (c 0.19, CHCl
CDCl
3
) δ 210.37, 171.50, 168.16, 92.05, 85.28, 80.31, 55.11, 46.41,
D
3
); IR (KBr)
νmax 3400, 1760, 1450, 1360, 1200, 1080, 1000 cm-1; MS (m/z) 373
3
4.90, 34.50, 30.97, 30.29, 27.25, 24.53, 20.92, 20.29, 12.44; MS (m/
+
+
+
1
z) 281 (M - Ac), 265 (M - AcO), 239, 221, 208, 193, 180, 165,
1
(M - 1), 315, 165, 139, 138, 137, 96, 91, 55; H NMR (400 MHz,
+
47, 121, 43; HRMS found C17
H
24
O
6
(M ) 324.1552, calcd 324.1572.
CDCl ): δ 0.90 (6H, d, J ) 6.8 Hz), 2.14 (3H, s), 3.90 (1H, q, J ) 7.9
3
Compound 12: colorless crystal (37% yield), mp 96-97 °C (hexane);
Hz), 4.26 (1H, t, J ) 7.9 Hz), 4.44 (1H, d, J ) 12.1 Hz), 4.79 (1H, d,
J ) 3.9 Hz), 4.93 (1H, d, J ) 12.1 Hz), 6.32 (1H, s), 7.33 (5H, m).
R]2
0
+117.4 (c 0.87, CHCl
); IR (KBr) νmax 1760, 1365, 1230, 1090,
020, 930, 500 cm ; H NMR (400 MHz, CDCl ) δ 0.87 (3H, d, J )
.5 Hz), 0.91 (3H, d, J ) 6.5 Hz), 2.11 (3H, s), 3.40 (3H, s), 3.90 (1H,
[
1
7
D
3
-
1 1
3
Anal. Calcd for C22
.11.
Compound 15c: colorless crystal (25% yield), mp 102-104 °C
30 5
H O : C, 70.56; H, 8.08. Found: C, 70.51; H,
8
m), 4.25 (1H, m), 4.61 (1H, d, J ) 4.2 Hz), 6.22 (1H, s). Anal. Calcd
for C16 : C, 64.40; H, 8.78. Found: C, 64.64; H, 8.95.
Compound 13: colorless needle crystal (44% yield), mp 65-67 °C
26 5
H O
1
9
(
ν
petroleum ether-ethyl acetate) [R]
D
-53.6 (c 0.19, CHCl
3
); IR (KBr)
-
1 1
max 3500, 1660, 1060 cm ; H NMR (400 MHz, CDCl ) δ 0.85 (3H,
3
2
0
(
1
hexane); [R]
D
+18.1 (c 1.30, CHCl ); IR (KBr) νmax 3500, 2920,
3
1
-
1
d, J ) 6.3 Hz), 0.95 (3H, d, J ) 7.4 Hz), 1.51 (3H, s), 2.46 (1H, m),
.55 (1H, d, J ) 2.2 Hz), 4.49 (1H, d, J ) 12.4 Hz), 4.83 (2H, m),
3
465, 1386, 1228, 1026, 930, 870 cm ; H NMR (400 MHz, CDCl )
3
5
8
δ 0.86 (3H, d, J ) 6.4 Hz), 0.91 (3H, d, J ) 7.5 Hz), 1.55 (3H, s),
.44 (1H, m), 3.38 (3H, s), 3.56 (1H, m, J ) 3.0 Hz), 4.63 (1H, d, J
4.2 Hz), 5.25 (1H, s). Anal. Calcd for C16 : C, 64.40; H, 8.78.
Found: C, 64.77; H, 8.94.
Compound 15b (from 11): colorless crystal (4% yield), mp 141-
.26 (1H, s), 7.30 (5H, m). Anal. Calcd for C22
30 5
H O : C, 70.56; H,
2
.08. Found: C, 71.30; H, 8.06.
)
26 5
H O
Compound 14d: colorless crystal (59% yield), mp 160-163 °C
1
9
(petroleum ether-ethyl acetate) [R]
D
+20.4 (c 0.74, CHCl
3
); IR (KBr)
20
-1 1
1
43 °C (petroleum ether-ethyl acetate); [R]
D
-68.7 (c 1.32, CHCl
IR (KBr) νmax 3470, 1454, 1383 cm ; H NMR (400 MHz, CDCl ) δ
.88, 0.89 (3H, dd, J ) 6.2, 6.2 Hz), 0.96, 1.00 (3H, dd, J ) 7.5, 7.2
Hz), 1.54 and 1.57 (3H altogether, 2 s), 2.31,2.44 (1H, m), 3.53, 3.55
3
);
νmax 1750, 1720, 1592 cm ; H NMR (100 MHz, CDCl ) δ 0.91 (3H,
3
-
1 1
3
d, J ) 7.3 Hz), 0.92 (3H, d, J ) 4.2 Hz), 2.08 (3H, s), 3.93 (1H, q, J
) 8.1 Hz), 4.30 (1H, t, J ) 7.5 Hz), 6.08 (1H, d, J ) 9.5 Hz), 6.30
(1H, s), 7.45 (3H, m), 8.08 (2H, m). Anal. Calcd for C H O : C,
0
2
2
28
6
(
1H, m), 4.80, 5.26 (1H, dd, J ) 6.8, 4.0 Hz), 5.28 and 5.32 (1H
68.02; H, 7.27. Found: C, 67.82; H, 7.30.
+
altogether, 2 s); MS (m/z) 267 (M + 1 - H
1
6
2
O), 249, 224, 206, 191,
Compound 15d: amorphous solid (25% yield), mp 97-98 °C
37, 121. Anal. Calcd for C15
H
24
O
5
: C, 63.38; H, 8.51. Found: C,
19
(
petroleum ether-ethyl acetate) [R]
D
-56.9 (c 0.28, CHCl
3
); IR (KBr)
3.47; H, 8.65.
-1
1
νmax 3527, 1713, 1281, 1013, 818, 714 cm ; H NMR (300 MHz,
2
0
Compound 14a: amorphous solid (45% yield), mp 69-71 °C; [R]
23.9 (c 1.34, CHCl
D
CDCl ) δ 0.91 (3H, d, J ) 6.3 Hz), 1.04 (3H, d, J ) 7.1 Hz), 1.55
3
-1 1
+
3
); IR (KBr) νmax 1761, 1734, 1714 cm ; H NMR
(3H, s) 2.75 (1H, m), 3.52 (1H, m) 5.33 (1H, s), 5.98 (1H, d, J ) 6.9
(300 MHz, CDCl
3
) δ 0.78 (3H, d, J ) 7.0 Hz), 0.86 (3H, d, J ) 6.1
Hz), 7.43 (2H, t, J ) 7.7 Hz), 7.56 (1H, t, J ) 7.4 Hz), 8.08 (2H, d,
Hz), 2.10 (3H, s), 2.64 (4H, m), 3.89 (1H, q, J ) 8.2 Hz), 4.21 (1H,
t, J ) 8.2), 5.82 (1H d, J ) 9.7 Hz), 6.14 (1H, s); ESI-MS (m/z) 430
J ) 7.1 Hz). Anal. Calcd for C22
28 6
H O : C, 68.02; H, 7.27. Found:
C, 67.74; H, 7.30.
+
+
(M + 2Na), 407 (M + Na).
Compound 15a: colorless crystal (23% yield); mp 139-141 °C;
R]2
0
-1
Acknowledgment. We thank the Chinese Academy of
Sciences (KJ95-A1-504), the State Committee of Science and
Technology of China, and the National Science Foundation for
financial support. We also express our appreciation to Dr. A.
R. Butler and Prof. Yao-Quan Chen for their helpful discussions.
[
D
-139.1 (c 1.16, CHCl
3
); IR (KBr) νmax 3290, 1740, 1703 cm
;
1
H NMR (600 MHz, CDCl
3
) δ 0.91 (3H, d, J ) 6.4 Hz), 0.99 (3H, d,
J ) 7.4 Hz), 1.59 (3H, s), 2.55 (1H, m), 2.68 (4H, m), 3.54 (1H, m),
5
COH)
Compound 15b (from 4): colorless crystal (56% yield), mp 149-
+
.27 (1H, s), 5.79 (1H d, J ) 6.4 Hz); MS (m/z) 266 (M - (CH
2
-
2
), 206, 162, 147.
2
0
1
50 °C, [R]
compound 15b from 11.
Compound 16: colorless crystal (39% yield), mp 107-109 °C; [R]
D
3
-73.4 (c 1.07, CHCl ); identified by comparison with
Supporting Information Available: Reproductions of MS
1
20
for 7, H NMR for 5, 6, 7, 8, 9, 12, 13, 15b, 14a, 14c, 14d,
D
-1 1
+
36.5 (c 0.60, CHCl
600 MHz, CDCl ) δ 0.95 (3H, d, J ) 6.3 Hz), 1.17 (3H, d, J ) 7.1
Hz), 2.41 (1H, m), 3.97 (1H, q, J ) 8.5 Hz), 4.12 (1H, dt, J ) 8.5, 2.0
3
); IR (KBr) νmax 3440, 1734, 1724 cm ; H NMR
15a, 15c, 15d, and 16, NOESY and COSY spectra of compound
1
3
(
3
9, C NMR of compounds 7, 9, and 16, and the ESR spectrum
22 pages, print/PDF). See any current masthead page for
(
1
3
Hz), 9.56 (1H, d, J ) 2.4 Hz), 9.93 (1H, s); C NMR (150 MHz,
ordering information and Web access instructions.
CDCl
3
) δ 13.13, 20.40, 27.00, 27.38, 30.50, 35.20, 46.96, 48.92, 56.73,
7.22, 88.53, 203.84, 205.54; MS (m/z) 225 (M + 1), 207, 195, 137.
+
6
JA973080O