Preparation of 10-Trifluoromethyl Artemether and Artesunate. Influence of Hexafluoropropan-2-ol on Substitution Reaction

Article abstract of DOI:10.1021/jo035174o

To prepare 10-trifluoromethyl analogues of important antimalarials such as artemether and artesunate, the substitution reaction of the 10-trifluoromethyl hemiketal 6 and bromide 4 derived from artemisinin was investigated. While 6 appeared to be unreactiv

Full text of DOI:10.1021/jo035174o

P r ep a r a tion of 10-Tr iflu or om eth yl Ar tem eth er a n d Ar tesu n a te.
In flu en ce of Hexa flu or op r op a n -2-ol on Su bstitu tion Rea ction
Guillaume Magueur, Benoit Crousse, Miche`le Oure´vitch, J ean-Pierre Be´gue´, and
Danie`le Bonnet-Delpon*
BIOCIS UPRES A 8076 CNRS, Faculte´ de Pharmacie, Universite´ Paris-Sud, Rue J . B. Cle´ment,
92296 Chaˆtenay-Malabry, France
daniele.bonnet-delpon@cep.u-psud.fr
Received August 11, 2003
To prepare 10-trifluoromethyl analogues of important antimalarials such as artemether and
artesunate, the substitution reaction of the 10-trifluoromethyl hemiketal 6 and bromide 4 derived
from artemisinin was investigated. While 6 appeared to be unreactive under various conditions,
bromide 4 could easily undergo substitution with methanol under electrophilic assistance or
noncatalyzed conditions. Optimization of the reaction revealed the role of CH2Cl2 as solvent to
avoid the competitive elimination process and the crucial influence of hexafluoro-2-propanol (HFIP)
in increasing the rate and the stereoselectivity of the substitution reaction (de >98%). The efficiency
of this reaction was exemplified with various alcohols and carboxylates (yield up to 89%).
In tr od u ction
from artemisinin and trifluoromethyltrimethysilane
(TMSCF3),2 and was the direct etherification or acylation
of the hydroxyl moiety. However, all our attempts failed,
probably because of the very weak nucleophilicity of the
corresponding alkoxide, due to the trifluoromethyl group.
An alternative route could be a substitution reaction of
the hydroxyl of 6 (or the corresponding mesylate) as
already described for trifluoromethyl furanose deriva-
tives.3 However, it is well-known that substitution of
trifluoromethyl alcohols is difficult by either SN1 or SN2
process.4 The electron-withdrawing character of the CF3
group strengthens the C-O bond. Furthermore, combi-
nation of steric and electronic repulsion of the incoming
nucleophile by fluorine atoms decreases the reaction rate
of SN2 process and most examples of SN2 reaction concern
primary and secondary trifluoromethyl alcohol deriva-
tives.5 In addition, the high electron-withdrawing effect
of the CF3 group destabilizes the intermediary carboca-
tion thus disfavoring an SN1 process. Such SN1 substitu-
tion is highly favored when an adjacent stabilizing
substituent counteracts the attractive effect of the CF3
group. This is the case for trifluoromethyl alcohols
substituted in R or â with aryl groups, heteroatoms or
metals with vacant d orbitals.6
Artemisinin 1 is a natural efficient antimalarial drug
for the treatment of multidrug-resistant forms of Plas-
modium falciparum.1 However, pharmacological prob-
lems associated with artemisinin 1, especially a short
plasma half-life, with, as consequence, a limited biodis-
ponibility, prompted scientists to search for more meta-
bolically stable derivatives. In this context, we investi-
gated the chemistry of 10-trifluoromethyl artemisinin
derivatives.2 In this paper, we report our studies aimed
at preparing trifluoromethyl analogues of ethers and
esters 3 of dihydroartemisinin 2 (DHA), which were
expected to be more stable toward metabolism process.
Our strategy is based on substitution reaction at the CF3-
substituted C-10 carbon of bromide 4.
(3) Munier, P.; Giudicelli, M. B.; Picq, D.; Anker, D. J . Carbohydr.
Chem. 1994, 13, 1225-1230.
Resu lts a n d Discu ssion
(4) (a) Allen, A. D.; Tidwell, T. T. In Advances in carbocation
chemistry; Creary, X., Ed.; J AI Press Inc.: Greenwich, CT, 1989; pp
1-44. (b) Creary, X. Chem. Rev. 1991, 91, 1625-1676.
In a first approach, an obvious route to 10-CF3
ketals 5 was to start from hemiketal 6, easily prepared
(5) (a) Matsutani, H.; Poras, H.; Kusumoto, T.; Hiyama, T. Synlett
1998, 1353-1354. (b) Katagiri, T.; Minoru, I.; Uneyama, K. Tetrahe-
dron: Asymmetry 1999, 10, 2583-2589. (c) Hagiwara, T.; Tanaka, K.;
Fuchikami, T. Tetrahedron Lett. 1996, 37, 8187-8190.
(6) (a) Uneyama, K.; Momota, M.; Hayashida, K.; Itoh, T. J . Org.
Chem. 1990, 55, 5364-5368. (b) Nesi, M.; Brasca, M. G.; Longo, A.;
Moretti, W.; Panzeri, A. Tetrahedron Lett. 1997, 4881-4884. (c) Tongco,
E. C.; Prakash, G. K. S.; Olah, G. A. Synlett 1997, 1193-1195. (d)
Kondratenko, M. A.; Male´zieux, B.; Gruselle, M.; Bonnet-Delpon, D.;
Be´gue´, J . P. J . Organomet. Chem. 1994, 487, C15-C17.
* To whom correspondence should be addressed. Fax: +33-1/
46835740
(1) Klayman, D. L. Science 1985, 228, 1049-1055.
(2) (a) Abouabdellah, A.; Be´gue´, J . P.; Bonnet-Delpon, D.; Gantier,
J . C.; Thanh Nga, T. T.; Truong Dinh, T. Bioorg. Biomed. Chem. Lett.
1996, 2717-2720. (b) Than Nga, T. T.; Me´nage, C.; Be´gue´, J . P.;
Bonnet-Delpon, D.; Gantier, J . C.; Pradines, B.; Doury, J . C.; T. D.
Thac. J . Med. Chem. 1978, 21, 4101-4108.
10.1021/jo035174o CCC: $25.00 © 2003 American Chemical Society
Published on Web 11/19/2003
J . Org. Chem. 2003, 68, 9763-9766
9763

Magueur et al.
SCHEME 2. Assign m en t of th e C-10 Con figu r a tion
in 5a
As for acylation, the mesylate of 6 could not be
obtained, and the hemiketal 6 was then directly submit-
ted to SN1-type conditions. However, when placed with
BF3‚Et2O in the presence of MeOH, hemiketal 6 re-
mained unchanged. Substitution of 6 with benzoic acid
under the Mitsunobu SN2 conditions also failed.
However, hemiketal 6 was found to be reactive in
bromination reaction. Reaction of 6 with thionyl bromide
provided the trifluoromethylated bromide 4 in very high
yield.7 The reaction proceeded with retention of config-
uration at C-10 (SNi reaction) as usually observed with
SOCl2 and SOBr2. Bromine atom being more labile than
a hydroxyl group, substitution reactions were then
investigated from 4.
SCHEME 3. Both Activa tion P a th w a y of
Su bstitu tion Rea ction of Br om id e 4
and H-9 in the major isomer, whereas it correlates with
both methyl-16 and H-12 in the minor one. On these
bases, the major isomer with a 19F signal at -75.9 ppm
was assigned to be the isomer where the methoxy group
is â (axial), whereas the minor one (δ 19F ) -75.5 ppm)
was assigned to be the R isomer. Non ambiguous NOE
effect indicated that methoxy in 5a â is axial joint to X-ray
data from a suitable crystal of 6, clearly indicate that a
chair conformation of D-ring is largely predominant.8
In these substitution reactions performed in the pres-
ence of a silver salt, since the reaction is not stereose-
lective it is reasonable to consider that reaction proceeds,
at least in part, in a SN1-type process providing an
oxonium salt with both accessible faces (route a, Scheme
3). As usually observed in the DHA chemistry with Lewis
acids, substitution and elimination are competitive path-
ways.9
SCHEME 1. Su bstitu tion of Br om id e 4 a n d
Ch lor id e 8 w ith MeOHa
a
12% of starting material was recovered.
Before extending the substitution to various nucleo-
philes, conditions had to be optimized in order to avoid
the use of the nucleophile as solvent, which represents a
real drawback in the case of expensive or high boiling
point compounds. For this purpose, the choice of a good
solvent could be crucial since we have previously dem-
onstrated the great influence of solvent on the ratio
elimination/substitution from DHA.2b,10 Bromide 4 was
thus placed in different solvents at room temperature and
Bromide 4 was first placed under silver salt assisted
solvolysis conditions, since silver salts are well-known
to favor SN1 substitution reactions or push-pull pro-
cesses. When bromide 4 was placed in methanol, in the
presence of 1 equiv of silver salt, 5a was obtained as a
mixture of 2 diastereoisomers and was accompanied by
the pseudo glycal 7 (Scheme 1). With AgOTf (entry 1),
bromide 4 provided after 1 h a 71:29 ratio of 5a and 7,
and 5a was obtained as a 39:61 mixture of 5a R (in which
OMe is R) and 5a â. When bromide 4 was replaced with
its chloride analogue 8 (prepared from 4 with SOCl2),7
the reaction rate strongly decreased, and chemoselectivity
was not improved (entry 2). Similar results were obtained
with other silver salts.
Configurations at C-10 were assigned with NOESY
experiments, performed on the mixture of both diastereo-
isomers resulting from the reaction with AgOTf. The
homo-NOE indicated, in the major isomer, a correlation
between protons of the methoxy group and H-12, whereas
in the minor isomer these protons correlate with H-9
(Scheme 2). Moreover, H, F hetero-NOE experiments
revealed a correlation between the trifluoromethyl group
aliquots were taken at t ) 1 and 8 h. The conversion into
19
the glycal 7 was followed in 19F NMR (δ
F
)
bromide 4
19
-78.1 ppm whereas δ
F
) - 65.4 ppm) and
glycal 7
compared to the conversion of 4 into 7 observed in MeOH
as solvent (Table 1). From this study it appeared that
bromide 4 was unstable in aprotic dipolar solvents such
as DMSO or DMF, in which 4 was quickly converted into
(8) Chorki, F.; Grellepois, F.; Crousse, B.; Oure´vitch, M.; Bonnet-
Delpon, D.; Be´gue´, J . P. J . Org. Chem. 2001, 66, 7858-7863.
(9) (a) Lin, A. J .; Lee, M.; Klayman, D. L. J . Med. Chem. 1989, 32,
1249-1252. (b) Hindley, S.; Ward, S. E.; Storr, R. C.; Searle, N. L.;
Bray, P. G.; Kevin Park, B.; Davies, J .; O’Neill, P. M. J . Med. Chem.
2002, 45, 1052-1063.
(10) Hoang, V. D.; Bonnet-Delpon, D.; Be´gue´, J . P. Unpublished
results. The substitution of DHA with fluoro alcohols is dependent on
the solvent used. A mixture of substitution/elimination compounds was
obtained in Et2O, whereas no elimination was observed when hexane
was used as solvent.
(7) Chorki, F.; Grellepois, F.; Crousse, B.; Hoang, V. D.; Hung, N.
V.; Bonnet-Delpon, D.; Be´gue´, J . P. Org. Lett. 2002, 4, 757-759.
9764 J . Org. Chem., Vol. 68, No. 25, 2003

Preparation of 10-Trifluoromethyl Artemether and Artesunate
TABLE 1. Sta bility of 4 in Differ en t Solven ts a t Room
Tem p er a tu r e
TABLE 2. Su bstitu tion of Br om id e 4 w ith Va r iou s
Nu cleop h iles
DMSO
DMF
CH3CN
hexane
CH2Cl2
MeOHb
solvent 1 h 8 h 1 h 8 h 1 h 8 h 1 h 8 h 1 h 8 h 1 h 8 h
4 (%)a
7 (%)a
41
4
68 45 91 77 >95 >95 >95 >95 14
0
59 96 32 55
9
23
<5 <5 <5 <5 16 23
a
b
19F NMR ratio. 5a R and 5a â are the other products obtained.
The 5a R/5a â ratio was the same, 17/83, at 1 and 8 h.
nucleophile
compd
yielda (%)
olefin 7b (%)
glycal 7. Conversely, it was stable in low or nonpolar ones
such as CH2Cl2 or hexane (no elimination was observed
in 19F NMR after 8 h).
H2O
6
87
79
76
76b
73
89
46
83
68
12
5
6
6
5
2
12
0
MeOH
5a
5b
5c
5d
5e
5f
5g
5h
EtOH
Surprisingly, in MeOH the reaction was complete in
less than 2 h, and substitution compound 5a was the
major product accompanied with only 23% of glycal 7.
5a was obtained in a 17/83 ratio of R and â isomers.
The conclusion of these experiments on solvent effects
is that the nucleophilic substitution of bromide 4 can
occur without silver electrophilic assistance when using
methanol as solvent. The diastereoisomer â was also
predominantly formed (de ) 66%) and a non negligible
amount of elimination product was obtained. Addition
of triethylamine or dimethylaminonaphthalin as proton
sponge did not change the substitution/elimination ratio.
Other alcohols such as ethanol or allylic alcohol used as
solvents also allowed the substitution of 4. It is postulated
that in these cases the solvent is able to activate bromine
through a hydrogen bond (RMeOH ) 0.98).11 Such a process,
more or less concerted, could explain the low ratio of
elimination and the better diastereoselection. However,
unlike an SN2 process, the substitution of 4 with MeOH
occurred mainly with retention of configuration, similarly
to an SNi process usually observed when the nucleophiles
also play a role of activator (route b, Scheme 3).12,13 The
second point that this study highlighted, is that CH2Cl2
appears to be a solvent of choice for studying the
substitution of 4 because no elimination was observed
after 8 h and it solubilizes starting material better than
hexane.
Bromide 4 was thus placed in CH2Cl2 with 10 equiv of
MeOH at room temperature. The substitution reaction
was slowed (72% conversion after 15 h). Nevertheless, it
is interesting to note that the diastereoselection was
improved (5a R/5a â ) 8:92) compared to the reaction
performed with MeOH as solvent (5a R/5a â ) 17:83).
Moreover, under these conditions the elimination process
into glycal 7 could be reduced to 3%.
With the hypothesis that the slower reaction rate was
due to the weaker activation process of bromide 4 by
MeOH as represented in Scheme 3, we thought to
reinforce this activation using a stronger hydrogen bond
donor but less electrophilic than a silver salt. A good
candidate was 1,1,1,3,3,3-hexafluoro-2-propanol (HFIP),
which possesses a low nucleophilicity, a high ionizing
power, and a high hydrogen bond donor ability (R )
1.96).11 All these properties contribute to modify the
4-(MeO2C)C6H4CH2OH
CH2dCHCH2OH
HOCH2CH2OH
c
CF3CH2OH
H2O2 (UHP)d
HO2C(CH2)2CO2Hd,e
5
Isolated yield. b NMR ratio of the crude. c Used as solvent, and
a
with addition of Et3N. In a mixture 1:1 CH2Cl2/HFIP. e In the
d
presence of 5 equiv of Et3N.
course and the rate of solvolysis reactions when HFIP is
used as solvent.14 They have also been exploited in
various other reactions, such as hydrogen peroxide
activation, oxirane ring opening,15 Nazarov cyclization,16
or Michael reaction.17
Experiments under various conditions using different
ratios of HFIP/MeOH or HFIP/CH2Cl2/MeOH allowed
selecting the following best conditions for the reaction of
4 with MeOH: 10 equiv of MeOH and 5 equiv of HFIP
in a dichloromethane solution at room temperature.
Under these conditions, the reaction was complete in less
than 5 h, instead of more than 1 day without HFIP.
Moreover, the reaction was still chemoselective, with only
5% of glycal 7 formed, and stereoselective: the R stere-
oisomer was not detected in NMR. However, an excess
of nucleophile was still required for the reaction. When
the nucleophile was used in stoichiometric amount,
degradation of the starting material was observed.
These conditions have been applied to other nucleo-
philes in order to obtain artemisinin derivatives with
better solubility in either oil or water. Treatment of 4
with various alcohols in the presence of HFIP gave 5a -e
in good isolated yields ranging from 73 to 89% (Table 2).
They were accompanied with a small amount of glycal 7
(2-6%). When products could not be separated from
glycal 7 by chromatography because of their close polarity
(5a and 5b), 7 was oxidized into a polar diol with RuO4
generated in situ from RuCl3 and NaIO4. After substitu-
tion of 4 with methyl 4-(hydroxymethyl)benzoate, a fast
purification on silica gel allowed the recycling of the
excess of reagent (90% were recovered). The product 5c
was then hydrolyzed under basic conditions to afford the
corresponding carboxylic acid 5c′ in 53% yield from 4.
(14) See, for example: (a) Allard, B.; Casadevall, A.; Casadevall, E.;
Largeau, C. Nouv. J . Chim. 1980, 4, 539-545. (b) Schadt, F. L.;
Bentley, D. W.; Schleyer, P. V. R. J . Am. Chem. Soc. 1976, 98, 7667.
(15) Iskra, J .; Bonnet-Delpon, D., Be´gue´, J . P. Eur. J . Org. Chem.
2002, 3402-3410.
(11) (a) Reichadt, C. Solvents and solvent effects in organic chemistry,
3rd ed.; Wiley-VCH: Weinheim, 2003. (b) Kamlet, M. J .; Abboud, J .-
L. M.; Abraham, M. H.; Taft, R. W J . Org. Chem. 1983, 48, 2877.
(12) Haynes, R. K.; Chan, H. W.; Cheug, M. K.; Chung, S. T.; Lam,
W. L.; Tsang, H. W.; Voerste, A.; Williams, I. D. Eur. J . Org. Chem.
2003, 2098-2114.
(16) Ichikawa, J .; Miyazaki, S.; Fujiwara, M.; Minami, T. J . Org.
Chem. 1995, 60, 2320-2321.
(13) Be´gue´, J . P.; Benayoud, F.; Bonnet-Delpon, D. J . Org. Chem.
1995, 60, 5029-5036.
(17) Takita, R.; Oshima, T.; Shibasaki, M. Tetrahedron Lett. 2002,
43, 4661-4665.
J . Org. Chem, Vol. 68, No. 25, 2003 9765

Magueur et al.
SCHEME 4. Con for m a tion of Ar tem isin in Rin gs in
Differ en t Activa tion Meth od s
methanol in the presence of silver salts. The predominant
â attack could be ascribed to a substitution reaction
involving an intermediate pyramidal alkoxy carbenium
ion (B, Scheme 4). The R approach of the nucleophile is
electronically and sterically disfavored, because of the
presence of the bulky trifluoromethyl moiety (usually
compared to an isopropyl group) and the oxygen lone
pairs, while the â face is only hindered by the C8a-C8
axial bond. It is assumed that due to the non polar
medium (CH2Cl2 is the major solvent) and to a different
counterion, the alkoxy carbenium ion, after ionization of
the C-Br bond, does not undergo the isomerization of B
into the planar oxonium salt A.
In all cases, the reaction was stereoselective. Configura-
tion of C-10 was assigned by NMR for compound 5a (â
OMe). Interestingly, substitution of 4 with water also
occurred and gave back the â isomer of the ketal 6 in
87% yield, thus confirming the stereochemical assign-
ments.18 All other products were hence supposed to be
the â isomers.
Con clu sion
This study allowed us to highlight the importance of
two cosolvents in the substitution reactions of R trifluo-
romethyl bromide 4 derived from artemisinin. The first
one, dichloromethane, allowed a decrease in the forma-
tion of the elimination product 7 (less than 10%) and thus
an increase in the chemoselectivity of the reaction. The
second, the hexafluoro-2-propanol, could both activate the
reaction and allow a complete diastereoselectivity (up to
95%).
The substitution of bromide 4 with various nucleo-
philes under mild conditions gave an access to new C-10
fluorinated analogues of artemisinin such as trifluoro-
methyl analogues of artemether, arteether, and artesunic
acid.
The reaction was also performed with the poor nucleo-
phile alcohol 2,2,2-trifluoroethanol (TFE), but the sub-
stitution did not occur and degradation of starting
bromide was observed. When TFE was used as solvent,
5f (â isomer as determined by NOE experiments) was
obtained in a poor yield (23%), which could be improved
to 46% when Et3N was added to the mixture. Substitu-
tion of 4 with hydrogen peroxide was also investigated
in view to obtain compound with two peroxide function-
alities. To avoid the competition between water and
hydrogen peroxide substitution, H2O2 was delivered
through urea hydrogen peroxide (UHP) which is a safe
source of anhydrous hydrogen peroxide.19 UHP is com-
mercially available, stable, and can be handled without
the need of specific precautions. Moreover, we have
recently reported that this stable complex could easily
release hydrogen peroxide when HFIP was used as
solvent or as cosolvent.20 Reaction of 4 with UHP in a
1:1 v/v mixture of CH2Cl2/HFIP was thus investigated,
and it was revealed that substitution quickly occurred
in less than 2.5 h and was selective. No traces of glycal
7 were detected. The resulting adduct was stable enough
to be purified on silica gel. 5g (â) was isolated in 83%
yield.
Carboxylic acids were also engaged in substitution
reaction of 4. Reaction with carboxylate generated in situ
from succinic acid and triethylamine (5 equiv), in the
presence of HFIP as a cosolvent (1:1 v/v CH2Cl2/HFIP),
provided 5h which could be isolated after purification in
68% yield (Table 2).
The high selectivity of the substitution reaction with
retention of configuration is very striking. The compari-
son of the reaction performed with MeOH/CH2Cl2 and
MeOH/CH2Cl2/HFIP clearly shows that HFIP favors the
cleavage of the C-Br bond. However, in this case the
hypothesis of an SNi-type reaction is ruled out since the
activator, HFIP, and the nucleophile are two different
molecules. Moreover, if a planar oxonium salt (intermedi-
ate A, Scheme 4) was formed, a mixture of R and â
isomers would be expected as obtained in reactions in
This new process constitutes an easy substitution
reaction on a center bearing a trifluoromethyl group
which could be extended to other substrates. Biological
properties against malaria of these new compounds are
under investigation.
Exp er im en ta l Section
Typ ica l P r oced u r e for th e Su bstitu tion of 4. Bromide
4 (353 mg, 0.85 mmol) was dissolved in CH2Cl2 (5 mL). The
solution was stirred under Ar stream at room temperature,
and ethylene glycol (475 µL, 8.5 mmol) and HFIP (440 µL, 4.2
mmol) were successively added. After being stirred for 5 h,
the mixture was diluted with ethyl acetate, washed with a
saturated aqueous solution of NaHCO3, and dried over MgSO4.
Evaporation of the solvent afforded the crude product which
was purified on a SiO2 column (7:3 petroleum ether/AcOEt).
Compound 5e was obtained (299 mg, 89%) as white crystals:
mp 92°C (petroleum ether/AcOEt).
Ack n ow led gm en t. We thank MENRT for financial
support (PAL+ program) and a fellowship (G.M.) and
CNRS (GDR Etude des protozoaires pathoge`nes: cibles
the´rapeutiques et vaccinales) for financial support. We
are grateful to Dr. V. D. Hoang, P. G. Dien, and N. V.
Hung at the Institute of Natural Products (CNST,
Hanoi, Vietnam) for supplying of artemisinin and
Central glass for supplying HFIP. We thank Sophie
Mairesse-Lebrun at the Microanalysis service of BIOCIS
for elementary analysis.
(18) Grellepois, F.; Chorki, F.; Crousse, B.; Oure´vitch, M.; Bonnet-
Delpon, D.; Be´gue´, J . P. J . Org. Chem. 2002, 67, 1253-1260.
(19) (a) Cooper, M. S.; Heaney, H.; Newbold, A. J .; Sanderson, W.
R. Synlett 1990, 533-535. (b) Heaney, H. Aldrichim. Acta 1993, 26,
35-45.
(20) Legros, J .; Crousse, B.; Bonnet-Delpon, D.; Be´gue´, J . P. Eur.
J . Org. Chem 2002, 3290-3293.
Su p p or tin g In for m a tion Ava ila ble: Experimental pro-
cedures and compound characterization data. This material
is available free of charge via the Internet at http://pubs.acs.org.
J O035174O
9766 J . Org. Chem., Vol. 68, No. 25, 2003