C1O-modified artemisinin derivatives: Efficient heme-alkylating agents

Article abstract of DOI:10.1002/anie.200462556

(Chemical Equation Presented) Discussion is reopened regarding C10-substituted derivatives of the peroxide-based antimalarial agent artemisinin (see picture); they are shown to react readily with iron(II)-heme to provide heme-drug adducts in high yields.

Full text of DOI:10.1002/anie.200462556

Correspondence
Artemisinin Derivatives
C10-Modified Artemisinin Derivatives: Efficient Heme-
Alkylating Agents**
Sophie A.-L. Laurent, Anne Robert,* and Bernard Meunier*
Keywords:
alkylation · antiviral agents · heme proteins ·
iron · porphyrinoids
[
1–4]
A
rtemisinin (1, Figure 1) is a sesqui- des.
Among the different possible some of these efficient antimalarial C10-
terpene extracted from the leaves of mechanisms, one suggestion is that re- modified derivatives were unable to
Artemisia annua which contains a 1,2,4- ductive activation of its endoperoxide alkylate free heme. For example, the
highly active 10-deoxoartemisinin deriv-
ative 2 (labeled 23 in Ref. [8]; see
Figure 1) with a methylene group at
II
C10 was totally inert to Fe -heme. On
the other hand, the 10-(4-benzyl)-
piperazinyl derivative 3 (labeled 18 in
Ref. [8]) reacted with heme, however,
the major reaction products, presumably
heme–drug adducts, were not character-
ized.
Herein we report that both com-
pounds 2 and 3 are indeed potent heme-
Figure 1. Structures of artemisinin (1), 10-deoxoartemisinin (2), and 10a-(4-benzyl)piperazinyl-
alkylating agents; both heme–drug ad-
1
0-deoxoartemisinin (3).
ducts have been obtained in high yields.
In fact, heme is poorly soluble in the
mixture of acetonitrile and phosphate
trioxane structure that is essential for its function by the iron(ii) center of heme buffer (pH 7.4) used by Haynes et al.
activity against malaria-infected eryth- generates a C4-centered radical that The saturation concentration of heme
rocytes. The mechanism of action of alkylates the meso positions of proto- under these conditions at room temper-
[
5]
artemisinin has been the center of porphyrin-ix (PPIX). We found that ature is not higher than 100 mm, and such
intense debate over the last two deca- many antimalarial trioxanes related to a low solubility probably does not allow
artemisinin are also able to interact with heme to react with artemisinin. For this
heme or heme-like models and conse- reason of solubility, we used a homoge-
[
*] S. A.-L. Laurent, Dr. A. Robert,
Dr. B. Meunier
[
4c,6]
quently alkylate the porphyrin ring.
neous solution in dimethylsulfoxide
Laboratoire de Chimie de Coordination du We were working on the identification (DMSO) containing 2–4 vol% of water
CNRS
of heme–artesunate and heme–trioxa- in which heme was significantly more
2
3
05 route de Narbonne
1077 Toulouse cedex 4 (France)
[7]
quine adducts
when Haynes et al. soluble (6 mm). Artemisinin itself and
reported the preparation and antimalar- its derivatives 2 and 3 are not very
Fax: (+33)5-6155-3003
E-mail: arobert@lcc-toulouse.fr
bmeunier@lcc-toulouse.fr
ial activities of several C10-substituted soluble in aqueous solution. However,
[
8]
artemisinin derivatives. As all of these many drugs are insoluble in water, yet
compounds are based on different sub- they are highly active, as hydrophobicity
[
**] Comments on “Highly Antimalaria-Active
Artemisinin Derivatives: Biological Activity stituents at the C10 position, no struc- is required for them to cross cell mem-
Does Not Correlate with Chemical Reac-
tivity”. This work was supported by the
CNRS and Palumed, and S.A.-L.L. is
indebted to both for a joint fellowship.
Dr. Y. Coppel (CNRS, Toulouse) is
acknowledged for the measurement of 2D
NMR spectra.
tural modification in terms of steric branes (e.g. the anticancer drug taxol is
effects or reactivity is expected for the insoluble in water). Furthermore, solu-
untouched trioxane entity of these de- tions that contain some volume of
rivatives. Consequently, one can expect DMSO are usually used in biologically
that the heme-alkylating capacities of relevant studies to evaluate the activity
these molecules should be very similar of artemisinin or related compounds in
[
9,10]
to that of artemisinin, artemether, or vitro and in vivo.
artesunate. In Table 2 of the article by Compound 3 was obtained from the
Haynes et al., the authors claimed that corresponding 10-trimethylsilyl ether
Supporting information for this article is
available on the WWW under http://
www.angewandte.org or from the author.
[
8]
2
060 ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
DOI: 10.1002/anie.200462556
Angew. Chem. Int. Ed. 2005, 44, 2060 –2063

Angewandte
Chemie
[
11,12]
4
after modification of the proce- agents other than dithionite can be used. the loss of a molecule of acetic acid at
[
8]
dure published by Haynes et al. In our The use of glutathione (which is a C12. A small amount of unconverted
hands, the original procedure provided “biological” reductant present at ap- heme (m/z = 616.3) was also present. A
essentially the elimination product 5 proximately 5-mm concentration in red ratio of 89:11 of 6-Fe to heme in the
with a double bond between C9 and blood cells) to form heme– or hemoglo- crude reaction mixture was indicated by
C10 (Scheme 1). However, when bin–artemisinin adducts has already mass spectrometry.
The mixture of heme and 6-Fe was
demetalated to provide the free-base
covalent adduct 6, which was character-
+
ized by MS (m/z = 831.3 [MH] ). How-
ever, owing to the strongly acidic con-
ditions required for demetalation, the
parent peak detected at m/z = 789.5
corresponded to compound 7, which
has a hemiacetal structure. The ratio of
6/7 was 3:10. As expected, a small
amount of H PPIX was detected at
2
Scheme 1. Synthesis of 10a-(4-benzyl)piperazinyl-10-deoxoartemisinin (3): a) TMSBr, 25 min,
m/z = 563.3
1:19).
Owing to the low solubility of heme
(ratio
6 + 7/H PPIX =
2
0
8C; b) 1-benzylpiperazine, HFIP, 45 min, 08C. TMS=trimethylsilyl.
8
derivatives, the purification and charac-
This confirms that terization by NMR spectroscopy of
[
5b,c]
1
,1,1,3,3,3-hexafluoropropan-2-ol
been reported.
[
10]
(
HFIP) was used as an additive in the this alkylation process is not dependent these heme–drug adducts was not pos-
amination step of the 10-brominated on the reductant used, provided that its sible. We therefore repeated these reac-
intermediate, the expected compound 3 redox potential is consistent with the tions using heme-dimethylester (heme-
was obtained in 50% yield (see Sup- reduction of iron(iii)-heme to iron(ii)- DME) instead of heme. In this case, the
porting Information). Compound 2 was heme. As previously described for arte- alkylation reaction was immediately
[
2b,5]
prepared by using a modification of the misinin,
described procedure,
the reductive activation of followed by demetalation to afford, as
and then the the peroxide bond of 10-deoxoartemisi- expected, a mixture of adducts 8 and 9,
two compounds 2 and 3 were tested for nin leads to an alkoxy radical on O2 which are the dimethylester analogues
[
13]
their alkylation of heme.
which, after b scission, gives a C4-cen- of compounds 6 and 7. Adducts 8 and 9
tered radical. This species efficiently were detected at m/z = 859.5 and 817.7,
alkylates the meso carbon atoms of respectively (MH ), with a ratio of 8/9 of
+
Alkylation of Heme by 2
heme and leads to the four regioisomer- 34:100, along with H PPIX-DME (m/z
ic covalent adducts 6-Fe. All of the = 591.3). The yield of alkylation calcu-
2
Iron(ii) protoporphyrin-ix, generat- porphyrinic material was precipitated lated from this analysis was 69% (for a
ed in situ from its iron(iii) analogue and from the reaction mixture and analyzed 1:1 molar ratio of heme-DME/drug).
sodium dithionite, readily reacted with 2 by mass spectrometry. The major com-
Column chromatography of this
at room temperature to provide the pound was the adduct 6-Fe, character- mixture yielded pure adduct 9, which
covalent heme–drug adduct 6-Fe (1:1:1 ized by its molecular mass peak (m/z = was characterized by 1D and 2D NMR
III
+
1
molar ratio of [Fe (PPIX)Cl]/2/dithion- 884.4 [M ]) with the parent peak (m/z = spectroscopy. The H NMR spectrum
+
ite; Scheme 2). In fact, several reducing 824.4 [MÀCH COOH] ) resulting from exhibited three signals in the range d =
3
Scheme 2. Alkylation of heme and heme-dimethylester by 10-deoxoartemisinin (2), and demetalation of the obtained adducts. The configuration
at C12 of adducts 7 and 9 has not been verified.
Angew. Chem. Int. Ed. 2005, 44, 2060 –2063
www.angewandte.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2061

Correspondence
1
0.63–9.76 ppm for the protons in the piperazinyl-substituted deoxoartemisi- low). Minor peaks were also detected at
meso positions. The resonance of the nin derivative. Mass spectrometry of m/z = 858.7 (hydration of the aldehyde
methylene group at C4 was detected as a the crude porphyrinic residue revealed function: + 18 amu) and 872.5 (hydra-
broad pattern at d = 5.42–5.05 ppm, a the presence of the adduct 10-Fe, which tion of the aldehyde and methylation of
proof of the covalent coupling between corresponds to alkylation of heme at the the hydroxyl function: + 32 amu). The
the porphyrin and drug moieties. The meso positions by the drug 3 (m/z = ratio 10-Fe/11-Fe was 43:57, as deter-
III
+
hemiacetal proton H12 gave a signal at 998.8, [M ÀCH COOH] ), along with mined from mass spectrometry, while
3
d = 4.63 ppm. The signals for the two a minor amount of unconverted heme the ratio of 10-Fe + 11-Fe/[Fe(PPIX)]
methylene protons at C10 at d = 3.68 (7% measured by HPLC; Scheme 3). was 80:20.
and 3.27 ppm overlapped with those of The structure of 10-Fe was confirmed by
Demetalation of the mixture of 10-
the protons of the b-pyrrolic CH and the presence of peaks at m/z = 999.7 and Fe and 11-Fe was performed and gave
3
CH COOCH3 groups, respectively. 499.9 which could be assigned to rise to the respective free-base adducts
2
II
À
+
These assignments were attested by [M ÀCH COO ]
and 10 and 11. Adduct 11 was characterized
3
1
13
III
À
2+
H– C correlation spectroscopy. Both [M ÀCH COO ] , respectively (see by the presence of its molecular peak at
3
+
protons H10 correlated with C10, which Supporting Information). The reduction m/z = 787.7 (MH ) and a small signal at
was detected at d = 61 ppm, and proton of iron(iii) porphyrins and the loss of m/z = 805.8 owing to hydration of the
H12 correlated with the resonance acetate at C12 in artemisinin-derived aldehyde function. Adduct 10 was char-
found at d = 93–95 ppm for C12. These adducts under the analytical conditions acterized by the presence of a peak at
NMR data allow to rule out the possible of mass spectrometry are well-known m/z = 829.7 owing to the loss of the
[
15]
hydrolysis of the cycle C8a-C10-C12 phenomena.
with formation of an alcohol at C10 Covalent adduct 11-Fe (m/z = 840.3, porting Information).
and an aldehyde at C12; in fact, such a MH ) was also produced during the
benzylpiperazinyl substituent (see Sup-
+
The identification of the covalent
product is expected to exhibit the same alkylation of heme by 3 by hydrolysis of adducts 10-Fe and 11-Fe was confirmed
mass spectrum as 9, but with different the C8a-C10-C12 cycle of adduct 10-Fe, by using heme-dimethylester instead of
NMR characteristics.
followed by nucleophilic attack of HO- heme as the starting material. Under
C12a at C10, and loss of 1-benzylpiper- these circumstances, the analogous ad-
+
azine (Scheme 4). This structure, which ducts 12-Fe (m/z = 1124.9 [MÀH+K] ,
+
Alkylation of Heme by 3
contains an exocyclic aldehyde, was 1026.5 [MÀCH COOH] ) and 13-Fe
3
1
13
+
confirmed by H– C NMR correlation (m/z = 868.5 [M ]) were observed
Iron(ii) protoporphyrin-ix was also spectroscopy after demetalation of the (Scheme 4; yield = 77% from HPLC
efficiently alkylated by the C10-benzyl- heme-dimethylester derivative (see be- analysis). Demetalation of the mixture
Scheme 3. Alkylation of heme and heme-dimethylester by 10a-(4-benzyl)piperazinylartemisinin (3). Bn=benzyl, pip=piperazinyl.
Scheme 4. Modification of the covalent adducts 10-Fe and 12-Fe to give 11-Fe and 13-Fe, respectively. The configuration at C10 is more probable,
but has not been verified.
2
062
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
Angew. Chem. Int. Ed. 2005, 44, 2060 –2063

Angewandte
Chemie
of 12-Fe and 13-Fe yielded 13 as the heme derivatives are supposed to have
main adduct (m/z = 815.6 [MH ]), while similar volatilities under given condi-
a significant peak at m/z = 857.8 was tions. Furthermore, the consistency be-
2001, 17, 122 – 126; b) A. Robert, J.
+
Cazelles, O. Dechy-Cabaret, B. Meuni-
er, Acc. Chem. Res. 2002, 35, 167 – 174.
3] a) U. Eckstein-Ludwig, R. J. Webb,
I. D. A. van Goethem, J. M. East, A. G.
Lee, M. Kimura, P. M. OꢀNeill, P. G.
Bray, S. A. Ward, S. Krishna, Nature
[
also detected that probably corresponds tween the results of HPLC and MS
+
to the molecular ion [M ] of compound analyses confirms that the yields of the
2003, 424, 957 – 961.
[
[
4] P. M. OꢀNeill, G. H. Posner, J. Med.
Chem. 2004, 47, 2945 – 2964.
5] a) A. Robert, J. Cazelles, B. Meunier,
Angew. Chem. Int. Ed. 2001, 40, 1954 –
1
957; b) A. Robert, Y. Coppel, B. Meu-
nier, Inorg. Chim. Acta 2002, 339, 488 –
96; c) K. Selmeczi, A. Robert, C.
Claparols, B. Meunier, FEBS Lett.
004, 556, 245 – 248.
4
2
Scheme 5. Structure of adduct 14.
[
[
6] J. Cazelles, A. Robert, B. Meunier, J.
Org. Chem. 2002, 67, 609 – 612.
7] S. A.-L. Laurent, C. Loup, S. Mourgues,
A. Robert, B. Meunier, ChemBioChem,
in press.
1
4 (see Scheme 5). The presence of this alkylation reactions of heme by 2 or 3
adduct, formed by the elimination of ranged from 70 to 90% under the
,10-piperazine from 12, indicates that reported conditions.
the C8a-C10-C12 cycle may withstand to In conclusion, it is clear that com-
[
8] R. K. Haynes, W.-Y. Ho, H.-W. Chan, B.
Fugmann, J. Stetter, S. L. Croft, L. Vivas,
W. Peters, B. L. Robinson, Angew.
Chem. 2004, 116, 1405 – 1409; Angew.
Chem. Int. Ed. 2004, 43, 1381 – 1385.
9
some extent the harsh conditions re- pounds 2 and 3 are very efficient alky-
quired for the demetalation step. It also lating agents toward iron(ii)-heme to
confirms the high propensity of the provide covalent heme–drug adducts in
artemisinin derivatives that are substi- high yields. The bulky amine substituent
tuted at C10 by an amine to produce the in the a position at C10 of compound 3
C9-C10 unsaturated derivative, as ob- (on the same side as the peroxide bond)
[9] a) P. M. OꢀNeill, A. Miller, L. P. D. Bish-
op, S. Hindley, J. L. Maggs, S. A. Ward,
S. M. Roberts, F. Scheinmann, A. V.
Stachulski, G. H. Posner, B. K. Park, J.
Med. Chem. 2001, 44, 58 – 68; b) S.
Hindley, S. A. Ward, R. C. Storr, N. L.
Searle, P. G. Bray, B. K. Park, J. Davies,
P. M. OꢀNeill, J. Med. Chem. 2002, 45,
served during the synthesis of 3.
did not prevent the interaction with the
The structure of 13 was unambigu- iron center of heme that mediates the
1
13
ously determined by H– C NMR cor- reductive activation. This is consistent
relation spectroscopy. The protons of with previous results obtained with
the meso positions and the aldehyde model trioxanes which suggest that
1052 – 1063; c) M. Jung, K. Lee, H.
Kendrick, B. L. Robinson, S. L. Croft,
(
H12) appeared at d = 9.82–10.24 ppm hindrance at C10 has no influence on
J. Med. Chem. 2002, 45, 4940 – 4944.
(
4H) and correlated with the meso the reactivity of the peroxide toward [10] G. Magueur, B. Crousse, S. Charneau, P.
Grellier, J.-P. Bꢁguꢁ, D. Bonnet-Delpon,
carbon atoms and the carbonyl center heme (active artesunate has an a confi-
C12 at d = 95–99 and 209 ppm, respec- guration at C10). In fact, reduction of
J. Med. Chem. 2004, 47, 2694 – 2699.
11] A. J. Lin, D. L. Klayman, W. K. Milhous,
J. Med. Chem. 1987, 30, 2147 – 2150.
12] R. K. Haynes, H.-W. Chan, M.-K.
Cheung, W.-L. Lam, M.-K. Soo, H.-W.
Tsang, A. Voerste, I. D. Williams, Eur. J.
Org. Chem. 2002, 113 – 132.
[
[
tively. A broad pattern at d = 5.10– the carbonyl function at C10 to give an
3
5
.54 ppm accounted for three protons, sp -hybridized carbon center renders
namely from H C4 and H10, which the C8a-C10-C12 cycle relatively labile,
2
[
16]
correlated with C4 and C10 that were as previously reported for artemether.
detected at d = 31–33 ppm and 107 ppm, This feature somewhat complicates the
respectively. These data confirm the characterization of the heme–drug ad- [13] M. Jung, X. Li, D. A. Bustos, H. N.
ElSohly, J. D. McChesney, W. K. Mil-
hous, J. Med. Chem. 1990, 33, 1516 –
presence of both the exocyclic aldehyde ducts, but does not change the reactivity
and the 5-membered cyclic hemiacetal. of the peroxide bond and, thus, does not
The NMR spectra were complex owing alter the alkylating ability of these
to the presence of four regioisomeric powerful antimalarial drugs.
adducts. The confirmation of the stereo-
1518.
[
[
14] A. J. Lin, M. Lee, D. L. Klayman, J.
Med. Chem. 1989, 32, 1249 – 1252.
15] A. Robert, Y. Coppel, B. Meunier,
Chem. Commun. 2002, 414 – 416.
[16] A. Robert, B. Meunier, Chem. Eur. J.
1998, 4, 1287 – 1296.
chemistry at C10 could therefore not be
obtained. However, the intramolecular
reaction makes the S configuration
probable.
[
1] S. R. Meshnick, T. E. Taylor, S. Kam-
chonwongpaisan, Microbiol. Rev. 1996,
6
0, 301 – 315.
Mass spectrometry is not, in princi-
ple, a quantitative method. However,
[2] a) P. L. Olliaro, R. K. Haynes, B. Meu-
nier, Y. Yuthavong, Trends Parasitol.
Angew. Chem. Int. Ed. 2005, 44, 2060 –2063
www.angewandte.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2063