Carbohydrate Derivatives of the Antitumour Alkaloid 9-Hydroxyellipticine

Article abstract of DOI:10.1002/1099-0690(200301)2003:1<63::AID-EJOC63>3.0.CO;2-V

The synthesis of L-arabinosyl derivatives of 2-naphthol and the quaternised ellipticines Celiptium (2) and Ellipravin (3) is reported. Naphth-2-yl 2',3',4'-tri-O-acetyl-α-L-arabinopyranoside was prepared under Koenigs-Knorr and Mitsunobu conditions in non

Full text of DOI:10.1002/1099-0690(200301)2003:1<63::AID-EJOC63>3.0.CO;2-V

FULL PAPER
Carbohydrate Derivatives of the Antitumour Alkaloid 9-Hydroxyellipticine
Annaleise R. Grummitt,[a] Margaret M. Harding,*[a] Pia I. Anderberg,[a] and
Alison Rodger[b]
Keywords: Glycosides / Antitumour agents / Heterocycles / DNA / Drug design
The synthesis of
the quaternised ellipticines Celiptium (2) and Ellipravin (3)
is reported. Naphth-2-yl 2Ј,3Ј,4Ј-tri-O-acetyl-α-
L
-arabinosyl derivatives of 2-naphthol and
by incorporation of an ether linker between the sugar and
ellipticine to give derivatives 15 and 16. The glycolate esters
L-arabinopyr- 17 and 18, which were prepared using 2-(2Ј,3Ј,4Ј-tri-O-
anoside was prepared under Königs−Knorr and Mitsunobu
conditions in nonpolar aprotic solvents and using 2,3,4-tri-
O-acetyl-L-arabinopyranosyl fluoride as the glycosyl donor.
acetyl-α-L-arabinopyranosyl)glycolic acid, undergo hydro-
lysis suggesting that these compounds could function as pro-
drugs in vivo. Linear dichroism studies of the interaction of
These conditions were not applicable to the corresponding
glycosidation reactions with the quaternised ellipticines 2
and 3 which are soluble only in polar solvents. Formation of
Celiptium (2) and the stable L-arabinosyl ellipticine derivat-
ives 3, 15 and 16 with calf thymus DNA are consistent with
intercalation of the ellipticine chromophore, positioning the
sugars at the 2- and 9-positions in the major and minor
grooves of DNA.
the 9-(α-
L-arabinopyranosyl)ellipticine derivatives 13 and 14
was achieved by using 2,3,4-tri-O-acetyl-L-arabinopyranosyl
bromide in the presence of sodium methoxide in methanol.
Improved yields were obtained under the same conditions
( Wiley-VCH Verlag GmbH & Co. KGaA, 69451 Weinheim,
Germany, 2003)
Introduction
Ellipticine and 9-methoxyellipticine are plant alkaloids
isolated from the leaves of the plant Ochrosia elliptica, of
the Apocynaceae family.[1] While both compounds exhibit
high in vitro anticancer activity against a number of experi-
mental tumours,[2] poor aqueous solubility has precluded
clinical trials with these compounds. Research to address
this limitation identified hydroxylation at C-9, to give 9-
hydroxyellipticine (1) (Figure 1, part a),[3] and quaternis-
ation at N-2,[4] as key structural modifications that led to
improved aqueous solubility, increased antitumour activity
and altered tissue distribution. As a result of these studies,
preclinical and clinical evaluations of a number of derivat-
ives of 9-hydroxyellipticine have been carried out, with 9-
hydroxy-N-methylellipticinium acetate (2) (Celiptium, Fig-
ure 1, part a) having entered the market for the treatment
of advanced breast cancer.[5,6]
Figure 1. (a) 9-Hydroxyellipticine (1) and important clinical
derivatives Celiptium (2) and Ellipravin (3), where -arab ϭ α--
arabinopyranosyl and (b) schematic drawing showing the relative
orientation of the proposed major and minor groove binding
groups at the 2- and 9-positions of ellipticines
trast to many other established DNA intercalators, there
are no X-ray crystal structures or NMR solution structures
of 9-hydroxyellipticine (1) or derivatives with DNA. Two
recent studies have provided important information about
the nature of the drug/DNA complex.[11,12] Spectroscopic
studies[12] and theoretical calculations[11] have shown that
the chromophore of 1 is oriented perpendicular to the helix
axis, with the 9-OH group protruding into the minor groove
and the pyridine nitrogen atom lying in the major
groove.[11,12] The placement of these two functional groups
in the DNA grooves allows for their use as chemical
handles for the attachment of other DNA-targeting groups,
9-Hydroxyellipticine (1) and derivatives exert their anti-
tumour activity via a multi-faceted mechanism that includes
DNA intercalation, the generation of DNA strand breaks
and interference with the action of topoisomerase II.[7Ϫ10]
DNA intercalation is believed to be a key interaction that
is directly related to anticancer activity. However, in con-
[a]
School of Chemistry, University of Sydney
N.S.W. 2006, Australia
Fax: (internat.) ϩ 61-2/9351-6650
E-mail: harding@chem.usyd.edu.au
[b]
Department of Chemistry, University of Warwick
Coventry CV4 7AL, UK
Eur. J. Org. Chem. 2003, 63Ϫ71  2003 WILEY-VCH Verlag GmbH & Co. KGaA, 69451 Weinheim 1434Ϫ193X/03/0101Ϫ0063 $ 20.00ϩ.50/0
63

A. R. Grummitt, M. M. Harding, P. I. Anderberg, A. Rodger
FULL PAPER
including groove binders, at these positions. Whilst a range the reaction conditions including the use of different solv-
of 2- and 9-substituted derivatives of 1 have been re- ents (toluene, acetone), the number of equivalents of re-
ported,[13] almost all derivatives contain simple alkyl and agents (1.2Ϫ5 equiv. of silver carbonate, 0.4Ϫ1.5 equiv. of
related substituents. One exception is the reported synthesis sugar 4) and the use of both silver carbonate and silver
and in vitro screening of 40 N-glycosides of 1.[14,15] Sugars zeolite as insoluble promoters failed to significantly im-
were chosen for attachment to the ellipticine chromophore prove the yield of 12. Celiptium (2) is insoluble in dichloro-
due to their prevalence in many antitumour drugs, and in methane/diethyl ether, the most successful solvent system
order to improve aqueous solubility. 2-(α--Arabinopyrano- identified for the formation of the naphthyl-O-glycoside 12.
syl)-9-hydroxyellipticinium bromide (3) (Ellipravin, Fig- The reaction was nevertheless attempted with Celiptium (2)
ure 1, part a) was selected from 40 glycosides for preclinical in dimethylformamide (DMF). However, there was no evid-
evaluation. Whilst no DNA-binding studies on 3 have been ence for formation of the corresponding O-aryl glycoside
reported, modelling studies with oligonucleotides suggest using this solvent.
that the sugar is located in the major groove.[16]
The low yields often observed in the synthesis of aryl
The establishment of the binding orientation of 9-hydro- glycosides are generally attributed to the poor nucleophilic-
xyellipticine (1) with DNA provides the basis for the ra- ity of the phenol functional group compared to alcohols in
tional design of DNA threaders in which major and minor glycosidation reactions. This is a major contributing factor
groove binding motifs are tethered to the ellipticine core to the low observed yield in the synthesis of 12 using
(Figure 1, part b).[13] Carbohydrates are an attractive choice KönigsϪKnorr conditions. This limitation has been ad-
for the development of threaders of this type, due to the dressed recently by the application of the Mitsunobu proto-
importance of sugars as groove binders in naturally occur- col to the preparation of O-aryl glycosides of a number of
ring DNA-binding antitumour antibiotics,[17Ϫ19] and due to naturally occurring DNA-binding antitumour antibi-
the significant improvements in aqueous solubility generally otics.[25] Using this procedure, the generation of the phenox-
conferred by carbohydrates.[20] In addition, glycosylated ide ion under conditions that favoured an SN2 pathway pro-
phenols have been widely used in anticancer drug design.[21] vided a highly efficient route for the stereoselective syn-
While the exact mechanism and the mode of DNA binding thesis of aryl 2-deoxy-β--glycosides.
of Ellipravin (3) are unknown, the high clinical activity of
The Mitsunobu conditions[25] were first applied to the
3 has demonstrated that addition of carbohydrates to 1 has synthesis of the naphthyl O-glycoside 12 (Scheme 1). Treat-
the potential to lead to novel antitumour derivatives. In this ment of hemiacetal 5,[26] prepared by hydrolysis of the gly-
paper, we report the synthesis of six new carbohydrate de- cosyl bromide 4, with 2-naphthol (diethyl azodicarboxylate,
rivatives of Celiptium (2) and Ellipravin (3) and preliminary triphenylphosphane, 0 °C, tetrahydrofuran) afforded the de-
DNA binding studies that confirm that these derivatives re- sired aryl glycoside 12 in 39% yield. While this yield was
tain the ability to bind to DNA by intercalation. Of particu- higher than the yield of 12 prepared under KönigsϪKnorr
lar interest is the synthesis of a new ellipticine derivative conditions (9%), the reaction failed in DMF, the solvent
containing two sugars, that is proposed to act as a DNA of choice for the corresponding reaction with ellipticines 2
threading agent.
and 3.
The third approach to form the O-aryl glycoside 12 in-
volved the use of the glycosyl fluoride 6. Glycosyl fluorides
have attracted significant attention as they are readily pre-
pared, stable to chromatography, and are excellent glycosyl
donors in the presence of a range of activating agents.[22]
The required glycosyl fluoride 6 was prepared as a mixture
of anomers (α/β ϭ 3:2) by treatment of hemiacetal 5 with
(diethylamino)sulfur trifluoride at low temperatures
(Scheme 1). Treatment of the fluoride 6 with boron
trifluorideϪdiethyl ether followed by 2-naphthol afforded
the O-aryl glycoside 12 in 10% estimated yield from analysis
of the crude product. Due to the poor yield, this method
was not investigated further.
Results and Discussion
-Arabinose was selected as the carbohydrate for syn-
thetic studies, due to the demonstrated clinical activity of
Ellipravin (3). Attachment of -arabinose to the 9-hydroxy
group of Celiptium (2) and Ellipravin (3)[15] by direct O-
aryl glycosidation, and by the incorporation of both a short
ester and a short ether linker between the sugar and the 9-
position, were investigated.
Preparation of O-Aryl Glycosides
Given the limited availability of ellipticines 2 and 3, the
The use of other glycosyl donors (trifluoroacetimid-
lack of solubility of these quaternised compounds in aprotic ates,[27] thioglycosides[28]) was not investigated as glycosida-
solvents that are generally favoured for glycosidations,[22] tions with these systems are not favoured in DMF or polar
and the susceptibility of 9-hydroxyellipticines to oxidation, protic solvents, in which the quaternised ellipticines 2 and
particularly under basic conditions,[8,10,13,23] glycosidation 3 are soluble. Similarly, initial formation of the O-aryl gly-
was initially investigated using 2-naphthol as a model sys- coside from the free-base 9-hydroxyelliptine (1) using the
tem. The naphthyl O-glycoside 12 was prepared under Schmidt procedure,[27] followed by N-glycosidation, is also
KönigsϪKnorr conditions in low yield (9%) from the glyco- problematic as 1 is also only soluble in DMF and polar
syl bromide 4[24] and 2-naphthol in the presence of silver protic solvents. Whilst the use of nucleophilic solvents is
carbonate (dichloromethane/diethyl ether, 1:1). Variation of highly undesirable in glycosidation reactions, the single ex-
64
Eur. J. Org. Chem. 2003, 63Ϫ71

Carbohydrate Derivatives of the Antitumour Alkaloid 9-Hydroxyellipticine
FULL PAPER
Scheme 2. (i) 4, sodium methoxide, methanol, 0 °C; (ii) 11, sodium
methoxide, methanol, 0 °C, where -arab ϭ α--arabinopyranosyl
Scheme 1. (i) H2O/acetone, Ag2CO3; (ii) (diethylamino)sulfur tri-
tions. In agreement with the literature,[29] a large excess of
the glycosyl bromide 4, short reaction times and low tem-
fluoride, dichloromethane, Ϫ30 to 25 °C; (iii) HOCH2COOBn, Ag
˚
zeolite, toluene, MS (4 A); (iv) H2, Pd/C, ethanol; (v) HOCH2-
˚
CH2OBn, Ag zeolite, toluene, MS (4 A); (vi) H2, Pd/C, ethanol;
(vii) MsCl, NEt3, tetrahydrofuran, 0 °C; (viii) AgCO3, 2-naphthol, peratures (0 °C) were essential in order to obtain the O-aryl
dichloromethane/diethyl ether (1:1); (ix) triphenylphosphane, di-
glycosides 13 and 14, which were identified by analysis of
ethyl azodicarboxylate, tetrahydrofuran, 2-naphthol; (x) boron
the NMR spectra and mass spectra of the crude products.
However, insufficient material was obtained to allow these
products to be fully characterised.
trifluorideϪdiethyl ether, acetonitrile, 2-naphthol
ample in the literature of O-glycosides of 9-hydroxyelliptic-
ine involved a low-yielding (20%) synthesis, which was ac-
complished by treatment of Celiptium (2) with a glycosyl
bromide in sodium methoxide/methanol.[29] These condi-
Preparation of Ester-Linked Glycosides
Given the difficulties in formation of the O-aryl glycos-
tions were originally reported for the O-glycosidation of 5- ides outlined in Scheme 1, and the low overall yields of 13
hydroxyindole[30] and were subsequently applied to the gly- and 14, the synthesis of ester-linked derivatives of 9-hydro-
cosidation of 2 to give O-aryl glycosides. Large excesses of xyellipticine was investigated. The ester linkage was chosen
the glycosyl bromides and short reaction times (Ͻ 5 min) as syntheses of numerous esters of 9-hydroxyellipticine (1)
were reported, presumably due to competing nucleophilic have been reported by reaction of 1 with a range of carb-
attack of the base and solvent on the sugar. Due to this oxylic acids in the presence of N,NЈ-dicyclohexylcarbodiim-
literature precedent and the good solubility of Celiptium (2) ide and 1-hydroxybenzotriazole in DMF.[31] Hence, the
and Ellipravin (3) in alcohols and water, these conditions carboxylic acid functional group was introduced into -ara-
were investigated for the synthesis of arabinosyl derivatives binose as shown in Scheme 1.
of 9-hydroxyellipticine.
Reaction of benzyl glycolate with bromide 4 in the pres-
The desired ellipticine O-aryl glycosides 13 and 14 were ence of silver zeolite afforded the protected glycoside 7
successfully prepared by treatment of Celiptium (2) and elli- which was subjected to hydrogenolysis to give the carb-
pravin (3), respectively, with sodium methoxide/methanol at oxylic acid 8 in reasonable yield (Scheme 1). The acid 8 was
0 °C for 15 min in the presence of 5 equiv. of arabinosyl treated with N,NЈ-dicyclohexylcarbodiimide and 1-hydro-
bromide 4 (Scheme 2). Rigorous exclusion of oxygen was xybenzotriazole in DMF followed by 9-hydroxyellipticine
required in all reactions in order to minimise oxidative de- (1), in the presence of a catalytic amount of (dimethyl-
gradation of ellipticine under these strongly basic condi- amino)pyridine to give the desired ester 17 in 17% isolated
Eur. J. Org. Chem. 2003, 63Ϫ71
65

A. R. Grummitt, M. M. Harding, P. I. Anderberg, A. Rodger
FULL PAPER
of the ester linker suggested that these compounds would
hydrolyse in vivo and hence further chemistry on these com-
pounds was not investigated.
Preparation of Ether-Linked Glycosides
Replacement of the ester linker with the hydrolytically
stable ether functional group was achieved as shown in
Scheme 2. Treatment of the arabinosyl bromide 4 with 2-
(benzyloxy)ethanol afforded the protected arabinose deriv-
ative 9, which was converted into the mesylate 11 under
standard conditions via the alcohol 10 (Scheme 1). Initial
attempts to synthesise 15 and 16 using mild bases (Cs2CO3,
K2CO3) in DMF were unsuccessful. The ellipticine O-gly-
coside 15 and bis(glycoside) 16 were both successfully pre-
pared by treatment of the 9-hydroxyellipticines 2 and 3, re-
spectively, with sodium methoxide in methanol at 0 °C for
15 min in the presence of 5 equiv. of mesylate 11
(Scheme 2).
The reaction conditions were optimised by monitoring
the reaction of Celiptium (2) with mesylate 11 by elec-
trospray mass spectrometry, as the positively charged el-
lipticines in the reaction mixture give strong positive mo-
lecular ions, allowing aliquots of the reaction to be rapidly
analysed at short time intervals. The reaction time of
15 min, as well as the presence of at least 5 equiv. of the
mesylate 4, resulted in maximum formation of the glycoside
15. With less than 5 equiv. of mesylate, unchanged and de-
graded starting materials were recovered, while prolonged
reaction times (30 min to 2 h) resulted in the formation of
a higher molecular weight peak by mass spectral analysis,
corresponding to a bis(sugar) adduct of Celiptium (2) at the
expense of the desired O-glycoside 15; this by-product was
not isolated or characterised. While the isolated yields of
analytically pure 15 and 16 after HPLC were very low (Ͻ
5%), the small scale of the reactions and the purification of
charged salts most likely contribute to this figure; the mass
spectral analysis of the reaction with time suggest that the
glycosides are formed in approximately 20% yield. Hence it
is possible that on a larger scale an improved isolated yield
could be obtained.
Scheme 3. (i) (a) N,NЈ-dicyclohexylcarbodiimide, 1-hydroxybenzo-
triazole, DMF, (b) 1, (dimethylamino)pyridine or (a) thionyl chlor-
ide, (b) 1, CHCl3, NEt3;(ii) 4, CdCO3, nitromethane
yield (Scheme 3). A slightly improved yield (23%) was ob-
tained by initial conversion of the acid 8 to the correspond-
ing acid chloride, which was immediately treated with 9-
hydroxyellipticine in chloroform/triethylamine (3:1) to give
the protected O-glycoside 17. The isolated yield of ester 17
was lower than expected from analysis of the crude product
by NMR spectroscopy, as significant hydrolysis of the ester
to regenerate 9-hydroxyellipticine (1) occurred during puri-
fication by flash chromatography.
The protected O-glycoside 17 was treated with glycosyl
bromide 4 under the literature conditions used for the pre-
paration of 3.[15] The crude product contained the per-
acetylated bis(glycoside) 18, as evidenced by electrospray
mass spectrometry (m/z ϭ 837 [Mϩ]) and analysis of the
DNA-Binding Studies
The interactions of Celiptium (2), Ellipravin (3), O-gly-
product by NMR spectroscopy, but by-products resulting coside 15 and bis(glycoside) 16 with calf thymus DNA (ct-
from cleavage of the glycolate ester were also observed. DNA) were studied by flow linear dichroism (LD) spectro-
Analytically pure material could not be obtained, as at- scopy. In this technique, the DNA is oriented with respect
tempted purification of 18 by HPLC in aqueous ammo- to the incident (linearly polarised) radiation allowing the
nium acetate and methanol resulted in quantitative hydro- interaction of the oriented DNA with a ligand that has ab-
lysis of the ester bond. Attempted deacetylation of the O- sorption bands in the UV/Vis region to be monitored. Any
glycoside 17 and the bis(glycoside) 18 by treatment with ligand that becomes oriented as a result of becoming associ-
ammonia in methanol at low temperatures (Ϫ15 °C) also ated with the DNA will give rise to an LD signal in the
resulted in cleavage of the glycolic ester bond. This suscept- chromophore absorption band of the ligand with an intens-
ibility of the glycolate ester bond to hydrolysis is presumed ity that is usually proportional to the amount of bound li-
to arise from the increased electrophilicity of the carbonyl gand.[32,33]
group due to the inductive, electron-withdrawing effect of
LD has been used previously to study the interaction of
the ether substituent on the α-carbon atom. Thus, while the the parent 9-hydroxyellipticine 1 with DNA.[12] Titration of
protected esters 17 and 18 were prepared, the poor stability a solution of 1 into a solution of ct-DNA at constant DNA
66
Eur. J. Org. Chem. 2003, 63Ϫ71

Carbohydrate Derivatives of the Antitumour Alkaloid 9-Hydroxyellipticine
FULL PAPER
concentration showed that the binding mode is dependent ally exhibit strong variations of the LD signal with wave-
on the ratio of the drug to the DNA base-pairs.[12] Initial length, giving both positive and negative signals.[32,33] Small
titration studies with Celiptium (2) using LD and CD increases in the LD signals of the four ellipticine/DNA
showed similar trends to those reported for 9-hydroxy- complexes at 260 nm, compared with the LD of the DNA
ellipticine (1) (data not shown). Hence, LD studies with el- alone (data not shown), are also consistent with intercala-
lipticines 2, 3, 15 and 16 were carried out at a drug/DNA tion, which occurs due to lengthening or stiffening of the
ratio of 1:10. Under these conditions, all the potential inter- DNA about the intercalation site.
calation sites are not saturated and a single binding mode
is present.
The LDr spectra (the difference between the normal ab-
sorption spectrum and the linear dichroism spectrum),
Figure 2 shows the UV and LD spectra obtained on ad- shown in the lower part of Figure 2 (curve iv) give further
dition of each of the ellipticines 2, 3, 15 and 16 to ct-DNA. information about the binding mode. The relatively flat and
The UV spectra of the ellipticines (curve i, upper spectra) negative LDr spectra confirm the orientation of the el-
overlap with the DNA absorptions below 300 nm (data not lipticine chromophores perpendicular to the long axis of
shown). Upon addition of Celiptium (2) to DNA (Figure 2, the DNA. In addition, the increase in the LDr signal at
part a, curve ii), the Celiptium (2) absorption maxima at wavelengths corresponding to the ellipticines (i.e., Ͼ
326, 371 and 465 nm shift to longer wavelengths, consistent 300 nm in the region where there is no overlap with the
with formation of a Celiptium/DNA complex. Similar DNA absorptions) in comparison to those due to the DNA,
trends are observed in the UV spectra of ellipticines 3, 15 confirm lengthening and stiffening of the DNA, consistent
and 16 (Figure 2, parts bϪd) in the presence of ct-DNA.
with the intercalation of the compound. The strong similar-
The lower curves in Figure 2 show the corresponding LD ity between the spectra in Figure 2, parts aϪd indicates a
spectra of the ellipticine plus ct-DNA solutions (curve iii). similar mode of binding for all compounds.
In each case, the presence of an LD signal at wavelengths
From these data alone, the relative orientation of the el-
corresponding to the maxima in the absorbance spectrum lipticine chromophores with respect to the DNA bases (i.e.,
of the ellipticine plus DNA solution is immediate evidence whether the pyridine nitrogen atom projects into the major
that all four ellipticines interact with DNA and become ori- or the minor groove) cannot be determined. The character-
ented. In addition, the negative LD signal for the aromatic istics of the DNA plus Celiptium (2) spectrum (Figure 2,
transitions of the ellipticines above 300 nm, where there is part a) are almost identical to the published data for DNA
no overlap with the DNA absorptions, necessitates that the plus 9-hydroxyellipticine 1.[12] This is not surprising since
ellipticine chromophore is more perpendicular than parallel the addition of a single methyl group at the 2-position,
to the DNA helical axis, consistent with intercalation be- which converts 1 into 2, would not be expected to alter the
tween the base pairs. For comparison, groove binders typic- binding orientation of the compound with DNA. Further-
Figure 2. UV and LD spectra showing the binding of ellipticines 2 (a) 3 (b) 15 (c) and 16 (d) to calf thymus DNA at pH ϭ 7; the UV
spectra show the ellipticines (40 µ) (curve i) and the ellipticines (50 µ) in the presence of ct-DNA (500 µ) (curve ii); shown below are
the corresponding LD spectra (curve iii) and LDr spectra (curve iv) of the ellipticine plus DNA solutions; the LD spectra are multiplied
by 10 to correct for different cell path lengths
Eur. J. Org. Chem. 2003, 63Ϫ71
67

A. R. Grummitt, M. M. Harding, P. I. Anderberg, A. Rodger
FULL PAPER
at 25 °C with a PerkinϪElmer 241 polarimeter. [α]D values are
given in 10Ϫ1deg·cm2·gϪ1. Microanalyses were performed by the
Microanalytical Unit at the University of Otago, New Zealand.
Ultraviolet spectra were recorded with a Cary 4E UV/Vis/NIR
spectrophotometer at 20 °C. Infrared spectra were recorded as thin
films with a PerkinϪElmer 1600 Fourier transform spectrophoto-
meter. 1H NMR spectra were recorded with a Bruker AC200 NMR
spectrometer or a Bruker WM AMX 400 and referenced to the
residual solvent peak. EI mass spectra were recorded with an AEI
MS902 spectrometer, CI mass spectra were recorded with a Therm-
more, the steric bulk of the additional glycosides in com-
pounds 3, 15 and 16 would be expected to strongly disfa-
vour any orientation that does not allow these groups to
project out into the grooves of the DNA.
The bis(glycoside) 16 is of particular interest as a new
class of DNA-threading agent. Intercalation of the chromo-
phore is only possible if the molecule threads through the
DNA, positioning one sugar in the major groove and one
sugar in the minor groove (see Figure 1, part b), with the
potential to provide multiple sites of interaction between oQuest PolarisQ Ion Trap mass spectrometer and ES mass spectra
were recorded with an LCQ Ion Trap mass spectrometer, using ca-
pillary voltages of ϩ10 V (positive ion) and Ϫ10 V (negative ion).
High-resolution mass spectra were recorded as follows. EI: 70 eV
with a source temperature of 200 °C and a 5.3 kV accelerating volt-
age, using perfluorokerosene as the reference mass; ES: Kratos
Analytical Concept ISQ mass spectrometer to a resolution of 7000
(10% valley definition); LSIMS: recorded using a 10 kV caesium
ion primary beam, with m-nitrobenzoic acid as a liquid matrix and
internal mass reference. 9-Acetoxyellipticine was a gift from Pro-
fessor M. Ishiguro, Suntori Institute for Bioorganic Chemistry,
Osaka and Celiptium (3) was provided by Sanofi Chimie, Sisteron.
the DNA and the threader. Further structural characteris-
ation of this complex, including the measurement of equi-
librium association constants and dissociation rate con-
stants would be informative in establishing whether the two
sugars significantly increase the residence time of 16 on
DNA compared with Celiptium (2).
Conclusion
The O-aryl ester- and ether-linked arabinosyl derivatives
of Celiptium and ellipravin 13Ϫ18 represent a new class of
derivatives of 9-hydroxyellipticine. While the limited
stability of the ester derivatives 17 and 18 suggests that
these compounds could function as pro-drugs in vivo, the
stable bis(glycoside) 16 is of particular interest as a DNA
threading agent; DNA intercalation of the ellipticine chro-
mophore of 16 positions one sugar in the major groove and
one sugar in the minor groove. While the impact on biolo-
gical activity of the sugars at the 9-position in 13Ϫ16 can-
not be predicted, the absence of the phenol functional
group at the 9-position will prevent the formation of react-
ive quinone imines; formation and subsequent alkylation of
these intermediates is a major pathway related to the cyto-
xicity of 9-hydroxyellipticines including 1 and 2.[8,10] This
change in chemical reactivity, as well as the effect of the
new sugars on DNA binding, topoisomerase II poison ac-
tivity and inhibition of p53 phosphorylation, should pro-
vide valuable information on the in vivo mechanism of ac-
tion of the ellipticines and may identify new lead com-
pounds with interesting biological activity profiles.
2,3,4-Tri-O-acetyl-L-arabinose (5): Silver carbonate (54.5 mg,
0.198 mmol) was added to a stirred solution of bromide 4[24]
(44.0 mg, 0.130 mmol) in acetone (0.80 mL) and water (9.0 µL,
0.50 mmol) at 0 °C with minimum exposure to light. After 4 h, the
insoluble silver salts were removed by filtration (0.45 µm Millipore
filter) and the solvent removed to give hemiacetal 5[26] (35.3 mg,
98%) as a cream-coloured oily solid; α/β ϭ 1:2. 1H NMR
(400 MHz, CDCl3, 27 °C): δ ϭ 5.47 (d, 3J1β,2 ϭ 3.4 Hz, 1-Hβ), 4.61
3
(d, J1α,2 ϭ 7.5 Hz,1-Hα) ppm.
2,3,4-Tri-O-acetyl-L-arabinopyranosyl Fluoride (6): Hemiacetal 5
(390 mg, 1.4 mmol) was dissolved in dichloromethane (20 mL) and
chilled to Ϫ30 °C with stirring under nitrogen. (Diethylamino)sul-
fur trifluoride (250 µL, 1.9 mmol) was added rapidly and the solu-
tion allowed to warm to room temperature. After 1 h, the solution
was chilled to Ϫ30 °C and methanol (3 mL) added. The solvent
was removed, the residue taken up in chloroform (60 mL), washed
with water (5 ϫ 40 mL), dried with anhydrous sodium sulfate and
the solvent removed to give the crude product (370 mg). Purifica-
tion by flash chromatography (dichloromethane) afforded sugar 6
(338 mg, 86%) as a clear oil; α/β ϭ 3:2. C11H15FO7 (278.2): calcd.
C 47.48, H 5.43; found C 47.67, H 5.48. IR: ν˜max ϭ 1748 cmϪ1
1
(CO). H NMR (400 MHz, CDCl3, 27 °C): δ ϭ 5.78 (dd, 3J1β,2β
ϭ
2.7, 3J1β,F ϭ 54.5 Hz, 1-Hβ), 5.42 (m, 4-H), 5.37 (dd, 3J3β,4β ϭ 3.4,
3J3β,2β ϭ 10.7 Hz, 3-Hβ), 5.30Ϫ5.22 (m), 5.16 (ddd, J2β,1β ϭ 2.7,
3
3J2β,3β ϭ 10.7, J2β,F ϭ 29.8 Hz, 2-Hβ), 4.12 (m), 3.88 (dd, J5e,4
ϭ
3
3
3
3
3
2.0, J5e,5a ϭ 13.3 Hz, 5-He), 3.77 (ddd, J5a,4 ϭ 1.0, J5a,5e
ϭ
Experimental Section
11.5 Hz, 5-Ha) 2.17Ϫ2.03 (m, 9 H, CH3CO’s) ppm. 19F NMR
(376 MHz, CDCl3, 27 °C): δ ϭ Ϫ133.7 (d, J ϭ 49 Hz), Ϫ149.2 (dd,
J ϭ 23, J ϭ 53 Hz) ppm. MS (EI): m/z (%) ϭ 278 (3) [Mϩ],
259.0816 [M Ϫ F·] {C11H15O7 requires 259.0818}, 115 (100), 216
(5) [M Ϫ F Ϫ Ac].
General Remarks: Reactions requiring anhydrous conditions were
performed under nitrogen in oven-dried glassware using freshly dis-
tilled and purified solvents. Flash chromatography was performed
on Merck silica gel (type 9385, 230Ϫ400 mesh). HPLC was per-
formed using a Waters 600E Multisolvent Delivery System, a
2-(2Ј,3Ј,4Ј-Tri-O-acetyl-α-
L
-arabinopyranosyl)glycolic Acid (8): Sil-
˚
Rheodyne 7725i injector and a Waters 486 tuneable absorbance ver zeolite (2 g) and activated 4-A molecular sieves were added to
detector (254 nm analytical; 350 nm preparative), and Alltech
Alltima C18 columns: analytical 5 micron column (250 mm ϫ
4.6 mm i.d.), flow rate 0.8 mL minϪ1; preparative 10 micron col-
umn (300 mm ϫ 22 mm i.d.), flow rate 7 mL minϪ1. The solvents
used were 50 m ammonium acetate buffer (solvent A) and meth-
anol (solvent B). Melting points were determined using a Reichert
heating stage and are uncorrected. Optical rotations were recorded
a solution of benzyl glycolate (153 mg, 0.92 mmol) in toluene
(12 mL) and the mixture stirred in the dark for 5 min at room tem-
perature. A solution of bromide 4 (462 mg, 1.6 mmol) in toluene
(30 mL) was added and the reaction mixture stirred at room tem-
perature. Further portions of silver zeolite (1 g) were added after
11.5 h and 23 h. After 39 h, the reaction mixture was diluted with
dichloromethane, filtered through silica, and the solvent was re-
68
Eur. J. Org. Chem. 2003, 63Ϫ71

Carbohydrate Derivatives of the Antitumour Alkaloid 9-Hydroxyellipticine
FULL PAPER
moved to give the crude product (530 mg). Purification by flash
chromatography (0.8 Ǟ 3% acetone in dichloromethane) afforded
glycolate 7 as a white solid (342 mg, 87%). [α]2D0 ϭ Ϫ14 (c ϭ 4.2,
2Ј-(Mesyloxy)ethyl 2,3,4-Tri-O-acetyl-α-L-arabinopyranoside (11):
Triethylamine (0.20 mL, 1.4 mmol) was added to a stirred solution
of sugar 10 (41 mg, 0.13 mmol) in tetrahydrofuran (4 mL) at 0 °C
CHCl3). 1H NMR (200 MHz, CDCl3, 27 °C): δ ϭ 7.35 (s, 5 H, under nitrogen, followed by dropwise addition of mesyl chloride
Ph), 5.3Ϫ5.2 (m, 2 H, 2Ј-H and 4Ј-H), 5.18 (s, 2 H, 2 ϫ 2-H), 5.05 (0.10 mL, 1.3 mmol) and the mixture was stirred at 0 °C for 2 d.
3
3
3
(dd, J3Ј,4Ј ϭ 3.5, J3Ј,2Ј ϭ 9.3 Hz, 1 H, 3Ј-H), 4.58 (d, J1Ј,2Ј
ϭ
The solvent was removed and the residue taken up in cold chloro-
3
6.6 Hz, 1 H, 1Ј-Hα), 4.32 (s, 2 H, PhCH2), 4.02 (dd, J5eЈ,5aЈ ϭ 13, form, washed with ice-cold water and dried with anhydrous sodium
3J5eЈ,4Ј ϭ 3.5 Hz, 1 H, 5Ј-He), 3.60 (dd, J5aЈ,5eЈ ϭ 12.8, J5aЈ,4Ј
ϭ
sulfate. Removal of the solvent afforded mesylate 11 as a clear oil
in quantitative yield. 1H NMR (200 MHz, CDCl3, 27 °C): δ ϭ
3
3
1.8 Hz, 1 H, 5Ј-Ha), 2.12 (s, 3 H, CH3CO), 2.06 (s, 3 H, CH3CO),
2.02 (s, 3 H, CH3CO) ppm. The glycolate 7 (328 mg, 770 µmol) 5.30Ϫ5.25 (m, 1 H, 4-H), 5.19 (dd, J2,1 ϭ 6.8, J2,3 ϭ 9.4 Hz, 1
was dissolved in ethanol (35 mL), Pd/C (10%, 320 mg) added and
the mixture stirred under hydrogen for 15.5 h. The reaction mixture
was filtered through Celite and the solvent removed to give the
crude product (220 mg) which was dissolved in diethyl ether and
3
3
3
3
H, 2-H), 5.04 (dd, J3,4 ϭ 3.4, J3,2 ϭ 9.3 Hz, 1 H, 3-H), 4.48 (d,
3J1,2 ϭ 6.8 Hz, 1 H, 1-H), 4.37 (t, 3J ϭ 4.5 Hz, 4 H, 2 ϫ 2Ј-H),
4.13Ϫ4.00 (m, 3 H, 5-He and 2 ϫ 1Ј-H), 3.67Ϫ3.61 (m, 1 H, 5-Ha),
3.04 (s, 1 H, SO2CH3), 2.14 (s, 3 H, CH3CO), 2.08 (s, 3 H, CH3CO),
extracted into saturated NaHCO3. The combined aqueous extracts 2.01 (s, 3 H, CH3CO) ppm.
were acidified to pH ϭ 3 with hydrochloric acid (10 ) and ex-
tracted into dichloromethane. The combined organic extracts were
Naphth-2-yl 2Ј,3Ј,4Ј-Tri-O-acetyl-α-L-arabinopyranoside (12)
dried with anhydrous sodium sulfate and the solvent was removed
Mitsunobu Conditions: Triphenylphosphane (27.9 mg, 106 µmol)
and diethyl azodicarboxylate (16.5 µL, 105 µmol) in tetrahydrofu-
ran (0.15 mL) were added to a stirred solution of sugar 5 (23.8 mg,
86.2 µmol) and naphth-2-ol (10.1 mg, 70.1 µmol) in tetrahydrofu-
ran (0.5 mL) at 0 °C, and the mixture was stirred at 0 °C for 2.75 h.
Removal of the solvent afforded a cream-yellow oily solid which,
upon purification by flash chromatography (dichloromethane), af-
forded glycoside 12 as a white solid (10.8 mg, 39%); m.p. Ͼ 300
1
to give sugar 8 (190 mg, 72%). [α]2D0 ϭ Ϫ7.0 (c ϭ 6.3, CHCl3). H
NMR (200 MHz, CDCl3, 27 °C): δ ϭ 5.30Ϫ5.20 (m, 2 H, 2Ј-H
3
and 4Ј-H), 5.09 (dd, J3Ј,4Ј ϭ 3.5, 3J3Ј,2Ј ϭ 9.3 Hz, 1 H, 3Ј-H), 4.57
3
(d, J1Ј,2Ј ϭ 6.5 Hz, 1 H, 1Ј-H), 4.34 (s, 2 H, 2 ϫ 2-H), 4.06 (dd,
3
3
3J5eЈ,4Ј ϭ 3.3, J5eЈ,5aЈ ϭ 13.1 Hz, 1 H, 5Ј-He), 3.68 (d, J5aЈ,5eЈ
ϭ
13.2 Hz, 1 H, 5Ј-Ha), 2.15 (s, 3 H, CH3CO), 2.11 (s, 3 H, CH3CO)
and 2.04 (s, 3 H, CH3CO) ppm. MS (ESϪ): m/z (%) ϭ 668 (100)
[M2], 334 (72) [M]. MS (CIϪ): m/z (%) ϭ 334 (12) [M], 333 (100)
[M Ϫ H], 291 (40) [M Ϫ Ac]. LSIMS: m/z ϭ 335.0973 [M ϩ Hϩ]
{C13H19O10 requires 335.0978}.
1
3
°C. H NMR (200 MHz, CDCl3, 27 °C): δ ϭ 7.77 (dd, J ϭ 1.5,
3
6.1 Hz, 2 H, aryl), 7.40 (m, 4 H, aryl), 7.20 (dd, J ϭ 2.4, 8.9 Hz,
3
3
1 H, aryl), 5.50 (dd, J2Ј,1Ј ϭ 6.2, J2Ј,3Ј ϭ 8.7 Hz, 1 H, 2Ј-H), 5.37
3
(m, 1 H, 4Ј-H), 5.25 (d, J1Ј,2Ј ϭ 6.1 Hz, 1 H, 1Ј-Hα), 5.20 (dd,
3J3Ј,4Ј ϭ 3.3, J3Ј,2Ј ϭ 8.7 Hz, 1 H, 3Ј-H), 4.17 (dd, J5eЈ,4Ј ϭ 4.1,
3
3
2Ј-(Benzyloxy)ethyl 2,3,4-Tri-O-acetyl-α-
L
-arabinopyranoside (9):
3J5eЈ,5aЈ ϭ 12.8 Hz, 1 H, 5Ј-He), 3.81 (dd, J5aЈ,4Ј ϭ 2.1, J5aЈ,5eЈ
ϭ
3
3
˚
Silver zeolite (5.14 g) and molecular sieves (4 A) were added to a
solution of 2-(benzyloxy)ethanol (0.79 mL, 5.6 mmol) in toluene
(50 mL). A solution of bromide 4 (3.20 g, 9.4 mmol) in distilled
toluene (300 mL) was added and the mixture stirred at room tem-
perature under nitrogen for 5 d, filtered through Celite and the
solvent removed to give the crude product (3.46 g). Purification by
flash chromatography (25 Ǟ 30% EtOAc in hexane) gave sugar 9
(2.05 g, 90%) as a clear oil. [α]2D0 ϭ ϩ5.8 (c ϭ 5.0, CHCl3): IR:
ν˜max ϭ 1742 (CO) cmϪ1. 1H NMR (200 MHz, CDCl3, 27 °C): δ ϭ
7.28Ϫ7.21 (m, 5 H, Ph), 5.19Ϫ5.11 (m, 2 H, 2-H and 4-H), 4.97
(dd, 3J ϭ 3.5, 9.4 Hz, 1 H, 3-H), 4.48 (s, 2 H, PhCH2), 4.46 (d,
3J ϭ 6.9 Hz, 1 H, 1-H), 4.00Ϫ3.86 (m, 3 H, 2 ϫ 2Ј-H and 5-He),
3.73Ϫ3.52 (m, 3 H, 2 ϫ 1Ј-H and 5-Ha), 2.06 (s, 3 H, CH3CO),
1.95 (6 H, 2 ϫ CH3CO) ppm. MS (EI): m/z (%) ϭ 410 (3) [M],
367.1389 [M Ϫ Ac] {C18H23O8 requires 367.1393}, 91 (100).
12.8 Hz, 1 H, 5Ј-Ha), 2.16 (s, 3 H, CH3CO), 2.11 (s, 3 H, CH3CO),
2.10 (s, 3 H, CH3CO) ppm. MS(EI): m/z (%) ϭ 402.1315 [Mϩ]
{C21H22O8 requires 402.1330}, 259 (13.8) [M Ϫ OC10H7], 144 (100)
[C10H7OH], 43 (98.4) [CH3CO].
Königs؊Knorr Conditions: Silver carbonate (24.4 mg, 88 µmol) was
added to a solution of sugar 4 (36.9 mg, 109 µmol) and naphth-2-
ol (10.7 mg, 74 µmol) in diethyl ether/dichloromethane (1:1,
˚
1.6 mL) over 4-A molecular sieves. The mixture was stirred in the
dark at room temperature for 25.5 h, passed through a layer of
silica to remove the silver salts and the solvent removed to give
the crude product (19.5 mg). Purification by flash chromatography
twice (0 Ǟ 2% methanol in dichloromethane; 0 Ǟ 0.5% methanol
in dichloromethane), gave compound 12 (2.7 mg, 9%), which
showed similar spectral properties to that isolated using the Mitsu-
nobu conditions.
2Ј-Hydroxyethyl 2,3,4-Tri-O-acetyl-α-L-arabinopyranoside (10):
Glycosyl Fluoride Route: Boron trifluorideϪdiethyl ether (50 µL,
400 µmol) was added to a solution of fluoride 6 (24 mg, 87 µmol)
and naphth-2-ol (6.8 mg, 47 µmol) in acetonitrile (2 mL), and the
mixture stirred under nitrogen for 7 d. The solvent was removed
and the residue taken up in CHCl3, washed with saturated
NaHCO3 and water, and dried with anhydrous sodium sulfate. Re-
moval of the solvent gave the crude product (12.5 mg), which was
purified by flash chromatography (0 Ǟ 1% methanol in dichloro-
methane) to afford impure 12 (3.1 mg) which was identified on
comparison with authentic material prepared as described above.
Sugar 9 (1.54 g, 3.75 mmol) was dissolved in ethanol (150 mL), Pd/
C (10%, 1.57 g) was added, and the mixture stirred under hydrogen
for 22 h. The mixture was filtered through Celite and the solvent
removed to give the crude product (1.23 g), which was purified by
flash chromatography (0 Ǟ 2% methanol in dichloromethane) to
give sugar 10 (1.15 g, 96%) as a cream-colored, crystalline solid;
m.p. 90.5Ϫ91.5 °C. [α]2D0 ϭ ϩ53 (c ϭ 1.9, CHCl3). C13H20O9
(320.3): calcd. C 48.75, H 6.29; found C 49.03, H 6.44%. IR: ν˜max ϭ
1748 (CO) cmϪ1
5.30Ϫ5.25 (m, 1 H, 4-H), 5.21 (dd, J2,1 ϭ 7.1, J2,3 ϭ 9.6 Hz, 1
.
1H NMR (200 MHz, CDCl3, 27 °C): δ ϭ
3 3
3
3
H, 2-H), 5.05 (dd, J3,4 ϭ 3.5, J3,2 ϭ 9.5 Hz, 1 H, 3-H), 4.46 (d, 9-(α-L-Arabinopyranosyl)-2-methylellipticinium Acetate (13): Celip-
3J1,2 ϭ 6.9 Hz, 1 H, 1-H), 4.05 (dd, J5e,4 ϭ 3.0, J5e,5a ϭ 13.1 Hz,
tium (2) (20.0 mg, 59.5 µmol) was dissolved in distilled methanol
(5.0 mL) and chilled to 0 °C with stirring under nitrogen. A solu-
3
3
3
1 H, 5-He), 3.92Ϫ3.73 (m, 4 H, 2 ϫ CH2), 3.66 (dd, J5a,4 ϭ 1.6,
3J5a,5e ϭ 13.2 Hz, 1 H, 5-Ha), 2.15 (s, 3 H, CH3CO), 2.08 (s, 3 H, tion of sodium methoxide in methanol (0.30 mL, 2.1 , 630 µmol)
CH3CO), 2.03 (s, 3 H, CH3CO) ppm. MS (EI): m/z (%) ϭ 319.1030 was added, followed by glycosyl bromide 4 (105 mg, 310 µmol),
[M Ϫ H] {C13H19O9 requires 319.1029}, 73 (100).
and the reaction mixture stirred at 0 °C for 10 min. The solution
Eur. J. Org. Chem. 2003, 63Ϫ71
69

A. R. Grummitt, M. M. Harding, P. I. Anderberg, A. Rodger
was diluted with diethyl ether (3.0 mL) and the precipitate retained. 60 min) to give a number of impure samples containing 16. Pure
FULL PAPER
Mass spectral analysis of the crude product was consistent with the
compound 16 was isolated from several fractions as an orange solid
(2.0 mg, 3%). UV (H2O): λmax (ε) ϭ 211 (6900), 211 (6900), 249
presence of glycoside 13. 1H NMR (400 MHz, CD3OD, 27 °C):
δ ϭ 9.63 (1-H), 8.15 (3-H), 8.05 (4-H), 7.60 (10-H), 7.28 (7-H), 7.05 (9600), 314 (16500), 454 (1700) nm (dm3molϪ1cmϪ1). 1H NMR
(8-H), 4.1Ϫ3.6 (m, H-sugars) ppm. MS (ESϩ): m/z (%) ϭ 409 (10)
(400 MHz, CD3OD, 27 °C): δ ϭ 9.87 (s, 1 H, 1-H), 8.48 (dd, 3J3,1 ϭ
3 3
[Mϩ], 278 (100).
1.2, J3,4 ϭ 7.5 Hz, 1 H, 3-H), 8.30 (d, J4,3 ϭ 7.5 Hz, 1 H, 4-H),
3
7.74 (s, 1 H, 10-H), 7.44 (d, J7,8 ϭ 8.6 Hz, 1 H, 7-H), 7.22 (dd, 1
2-(α-L-Arabinopyranosyl)-9-(α-L-arabinopyranosyl)ellipticinium
3
3
3
H, J8,10 ϭ 2.0, J8,7 ϭ 8.5 Hz, 8-H), 5.69 (d, J1ЈЈЈ,2ЈЈЈ ϭ 8.7 Hz, 1
Acetate (14): Compound 3[15] (14.5 mg, 24 µmol) was dissolved in
methanol (8 mL) and chilled to 0 °C with stirring under nitrogen.
A solution of sodium methoxide in methanol (0.255 mL, 0.94 ,
240 µmol) was added, followed by glycosyl bromide 4 (41.4 mg,
122 µmol), and the reaction mixture stirred at 0 °C for 15 min, then
allowed to warm to room temperature for a further 10 min. The
solvent was removed and the residue washed with dichloromethane
to removed unchanged sugar starting material, yielding the crude
product. This contained predominantly unchanged starting mat-
erial 3, with only trace amounts of the desired product 14 which
could not be isolated in sufficient amounts to allow full characteris-
3
H, 1ЈЈЈ-H), 4.42 (d, J1ЈЈ,2ЈЈ ϭ 7.0 Hz, 1 H, 1ЈЈ-H), 4.34Ϫ4.24 (m, 4
H, 2 ϫ 2Ј-H and 1Ј-HA and 5ЈЈЈ-He), 4.12Ϫ4.09 (m, 2 H),
4.05Ϫ4.01 (m, 1 H, 1Ј-HB), 3.99Ϫ3.97 (m, 1 H), 3.92 (dd, 3J5eЈЈ,4ЈЈ
3.0, J5eЈЈ,5aЈЈ ϭ 17.7 Hz, 1 H, 5ЈЈ-He), 3.88Ϫ3.86 (m, 2 H, 4ЈЈ-H
ϭ
3
3
3
and 1 H), 3.70 (dd, J2ЈЈ,1ЈЈ ϭ 7.0, J2ЈЈ,3ЈЈ ϭ 8.9 Hz, 1 H, 2ЈЈ-H),
3.64 (dd, 3J5aЈЈ,4ЈЈ ϭ 1.6, 3J5aЈЈ,5eЈЈ ϭ 14.4 Hz, 1 H, 5ЈЈ-Ha), 3.61 (dd,
3J3ЈЈ,4ЈЈ ϭ 3.4, J3ЈЈ,2ЈЈ ϭ 8.9 Hz, 1 H, 3ЈЈ-H), 3.21 (s, 3 H, 11-CH3),
3
2.77 (s, 3 H, 5-CH3), 1.92 (s, 3 H, CH3CO2Ϫ) ppm. MS (ESϩ):
m/z ϭ 571.2271 [Mϩ] {C29H35N2O10 requires 571.2292}.
Ellipticin-9-yl 2Ј-(2ЈЈ,3ЈЈ,4ЈЈ-Tri-O-acetyl-α-L-arabinopyranosyl)gly-
1
ation. H NMR (400 MHz, CD3OD, 27 °C): δ ϭ 9.66 (1-H), 8.22
colate (17): Acid 8 (17.5 mg, 52 µmol), N,NЈ-dicyclohexylcarbodii-
mide (13.1 mg, 63 µmol) and 1-hydroxybenzotriazole (7.4 mg, 55
µmol) were dissolved in DMF (1.5 mL) and stirred at room temper-
ature for 4.5 h. 9-Hydroxyellipticine (1) (11.6 mg, 44 µmol) and (di-
methylamino)pyridine (0.5 mg, 4 µmol) were added and stirring
was continued for 42.5 h. The solvent was removed and the residue
taken up in dichloromethane. The organic soluble portions were
combined and the solvent removed to give the crude product, which
was purified by flash chromatography (0 Ǟ 20% methanol in
dichloromethane) to give compound 17 as a yellow solid (4.3 mg,
(3-H), 7.90 (4-H), 7.90 (10-H), 7.50 (7-H), 7.30 (8-H) ppm. MS
(ESϩ): m/z (%) ϭ 527 (11) [Mϩ], 395 (100) [M Ϫ arabϩ].
9-(2Ј-[α-l-Arabinopyranosyl]ethyloxy)-2-methylellipticinium Acetate
(15): Celiptium (2) (20.0 mg, 59.5 µmol) was dissolved in methanol
(3.5 mL) and chilled to 0 °C with stirring under nitrogen. A solu-
tion of sodium methoxide in methanol (0.65 mL, 0.94 , 610 µmol)
was added, followed by a solution of mesylate 11 (119 mg, 297
µmol) in methanol (3.0 mL), and the solution stirred under nitro-
gen at 0 °C for 20 min. The solvent was removed, the residue
washed with dichloromethane to remove unchanged sugar starting
materials and the resultant crude product was purified by reverse
phase HPLC (40Ϫ85% methanol in 25 m ammonium acetate over
45 min). Integration of the analytical HPLC trace (254 nm) showed
the O-glycoside 15 was the major product present (ca. 25%). Puri-
fication of a portion of the material gave a number of mixed frac-
tions in addition to pure 15 which was isolated as an orange solid
(1.8 mg, 6%). UV (H2O): λmax (ε) ϭ 210 (14300), 247 (17900), 278
(16800), 307 (27300), 379 (3700), 439 (2300) nm (dm3 molϪ1 cmϪ1).
1H NMR (400 MHz, CD3OD, 27 °C): δ ϭ 9.85 (s, 1 H, 1-H), 8.40
1
17%). H NMR (400 MHz, CD3OD, 27 °C): δ ϭ 9.58 (s, 1 H, 1-
H), 8.34 (d, 3J3,4 ϭ 6.2 Hz, 1 H, 3-H), 8.11 (d, 1 H, 3J10,8 ϭ 2.0 Hz,
3
3
10-H), 7.98 (d, J4,3 ϭ 6.1 Hz, 1 H, 4-H), 7.54 (d, J7,8 ϭ 8.7 Hz,
3
3
1 H, 7-H), 7.32 (dd, 1 H, J8,10 ϭ 2.1, J8,7 ϭ 8.7 Hz, 8-H), 5.32
(d, 3J ϭ 1.3 Hz, 1 H, 3ЈЈ-H), 5.24Ϫ5.17 (m, 2 H, 2ЈЈ-H and 4ЈЈ-H),
3
3
4.83 (d, J1ЈЈ,2ЈЈ ϭ 6.6 Hz, 1 H, 1ЈЈ-H), 4.67 (ABq, J ϭ 4.8 Hz, 2
H, 2 ϫ 2Ј-H,), 4.08 (dd, J5eЈЈ,4ЈЈ ϭ 3.0, J5eЈЈ,5aЈЈ ϭ 13.2 Hz, 1 H,
3
3
3
3
5ЈЈ-He), 3.86 (dd, J5aЈЈ,4ЈЈ ϭ 1.4, J5aЈЈ,5eЈЈ ϭ 13.3 Hz, 1 H, 5ЈЈ-Ha),
3.18 (s, 3 H, 11-CH3), 2.76 (s, 3 H, 5-CH3), 2.16 (s, 3 H, CH3CO),
2.06 (s, 3 H, CH3CO), 2.01 (s, 3 H, CH3CO) ppm. MS (EI): m/z
(%) ϭ 578.1893 [Mϩ] {C30H30N2O10 requires 578.1901}, 262 (100)
[1]. MS (ESϩ): m/z (%) ϭ 579 (87) [M ϩ Hϩ], 239 (100).
3
3
3
(d, J3,4 ϭ 7.2 Hz, 1 H, 3-H), 8.26 (dd, J4,1 ϭ 1.2, J4,3 ϭ 7.2 Hz,
3
3
1 H, 4-H), 7.96 (d, J10,8 ϭ 2.3 Hz, 1 H, 10-H), 7.56 (d, J7,8
ϭ
8.8 Hz, 1 H, 7-H), 7.34 (dd, 3J8,10 ϭ 2.4, 3J8,7 ϭ 8.8 Hz, 1 H, 8-H),
DNA-Binding Studies: Water was purified using a Millipore Alpha-
3
4.49 (s, 3 H, 2-CH3), 4.39 (d, J1ЈЈ,2ЈЈ ϭ 6.9 Hz, 1 H, 1ЈЈ-H), 4.36
Q
system. Stock solutions of calf thymus DNA (ct-DNA)
3
(t, J2Ј,1Ј ϭ 4.8 Hz, 2 H, ϪOCH2CH2OAr), 4.23 and 4.01 (2 dt,
(1500 µ) and separate solutions of the ellipticines 2, 3, 15 and 16
(800 µ) were prepared in water. The DNA concentration in bases
3J1Ј,2Ј ϭ 4.8, 3J1AЈ,1BЈ ϭ 11.4 Hz, (2 ϫ 1 H, ϪOCH2CH2OAr), 3.93
3
3
(dd, J5eЈЈ,4ЈЈ ϭ 3.1, J5eЈЈ,5aЈЈ ϭ 12.5 Hz, 1 H, 5ЈЈ-He), 3.86Ϫ3.84
was determined spectroscopically using ε258
ϭ
6600
(m, 1 H, 4ЈЈ-H), 3.67 (dd, 3J2ЈЈ,1ЈЈ ϭ 6.8, 3J2ЈЈ,3ЈЈ ϭ 8.9 Hz, 1 H, 2ЈЈ-
molϪ1 dm3 cmϪ1. UV and LD titrations were performed by prepar-
ing a solution of DNA (500 µ) and adding aliquots of both the
DNA stock and ellipticine stock solutions to give solutions with a
constant DNA concentration and the following ratios of ellipticine/
DNA: 1:50, 1:40, 1:30, 1:20, 1:10 and 1:5. Titration solutions
(pH ϭ 7) contained NaCl (20 m) and Na2HPO4 (1 m).
3
3
H), 3.61 (dd, J5aЈЈ,4ЈЈ ϭ 1.6, J5aЈЈ,4ЈЈ ϭ 12.5 Hz, 1 H, 5ЈЈ-Ha), 3.58
3
3
(dd, J3ЈЈ,4ЈЈ ϭ 3.5, J3ЈЈ,2ЈЈ ϭ 8.9 Hz, 1 H, 3ЈЈ-H), 3.33 (s, 3 H, 11-
CH3), 2.87 (s, 3 H, 5-CH3), 1.90 (s, 3 H, CH3CO2Ϫ) ppm. MS
(ESϩ): m/z ϭ 453.2019 [Mϩ] {C25H29N2O6 requires 453.2026}.
9-[2Ј-(α-L-Arabinopyranosyl)ethyloxy]-2-(α-L-arabinopyranosyl)-
ellipticinium Acetate (16): Compound 3 (70 mg, 0.12 mmol) was
dissolved in methanol (25 mL) and chilled to 0 °C with stirring
under nitrogen. A solution of sodium methoxide in methanol
(1.1 mL, 1.075 , 1.2 mmol) was added and the mixture stirred for
10 min. A solution of mesylate 11 (231 mg, 0.58 mmol) in methanol
(20 mL) was added, and the solution stirred at 0 °C for 30 min.
The solvent was removed and the residue washed with dichlorome-
thane to remove unchanged sugar starting material. The crude
UV Absorbance: UV spectra were recorded with a Cary 4E UV/
Vis/NIR spectrophotometer at 20 °C in a 1-cm quartz cuvette.
Blank spectra of the relevant solvent were collected and subtracted
from the sample spectra. UV spectra of the ellipticines (40 µ) in
water were also acquired.
Linear Dichroism (LD): LD spectra were measured with a Jasco J-
710 spectropolarimeter, adapted for LD measurements. Orientation
product was dissolved in water and purified by reverse phase HPLC of the ellipticine/DNA samples was achieved in a flow Couette
(40Ϫ85% methanol in 25 m ammonium acetate over 45 min fol-
cell[34] with an inner rotating cylinder with a base plate adapted for
lowed by 40Ϫ70% methanol in 25 m ammonium acetate over the smaller compartment. The experimental path length was 1 mm
70
Eur. J. Org. Chem. 2003, 63Ϫ71

Carbohydrate Derivatives of the Antitumour Alkaloid 9-Hydroxyellipticine
FULL PAPER
[14]
T. Honda, M. Inoue, M. Kato, K. Shima, T. Shimamoto,
Chem. Pharm. Bull. 1987, 35, 3975Ϫ3978.
T. Honda, M. Kato, M. Inoue, T. Shimamoto, K. Shima, T.
Nakanishi, T. Yoshida, T. Noguchi, J. Med. Chem. 1988, 31,
1295Ϫ1305.
T. Shimamoto, S. Imajo, T. Honda, S. Yoshimura, M. Ishiguro,
Bioorg. Med. Chem. Lett. 1996, 6, 1331Ϫ1334.
S. Walker, K. G. Valentine, D. Kahne, J. Am. Chem. Soc. 1990,
112, 6428Ϫ6429.
and the rotation speed was ca. 1000 rpm for all samples. A back-
ground spectrum of phosphate buffer, at the same rotation speeds
[15]
as the samples, was collected and subtracted from all spectra. Data
were collected with a response time of 0.5 s, at a speed of
1000 nm minϪ1 and a step resolution of 0.5 nm. Data were aver-
[16]
aged over 16 acquisitions for all spectra. LDr spectra were calcu-
lated using LDr (λ) ϭ LD (λ)/A (λ), after adjustment of absorbance
[17]
values due to different path lengths.
[18]
[19]
D. Kahne, Chem. Biol. 1995, 2, 7Ϫ12.
K. C. Nicolaou, K. Ajito, H. Komatsu, B. M. Smith, P. Bertin-
ato, L. Gomez-Paloma, Chem. Commun. 1996, 1495Ϫ1496.
E. Palomino, Adv. Drug Deliv. Rev. 1994, 13, 311Ϫ323.
L. F. Tietze, T. Feuerstein, A. Fecher, F. Haunert, O. Panknin,
U. Borchers, I. Schuberth, F. Alves, Angew. Chem. Int. Ed.
2002, 41, 759Ϫ761.
R. V. Stick, Carboydrates Ϫ The Sweet Molecules of Life, Aca-
demic Press, Salisbury, 2001.
P. Potier, Chem. Soc. Rev. 1992, 113Ϫ119.
V. P. Miller, D.-y. Yang, T. M. Weigel, O. Han, H.-w. Liu, J.
Org. Chem. 1989, 54, 4175Ϫ4188.
W. R. Roush, X.-F. Lin, J. Am. Chem. Soc. 1995, 117,
2236Ϫ2250.
V. Magnus, D. Vikic-Topic, S. Iskric, S. Kveder, Carbohydr. Res.
1983, 114, 209Ϫ224.
R. R. Schmidt, W. Kinzy, Adv. Carbohydr. Chem. Biochem.
1994, 50, 21Ϫ123.
P. J. Garegg, Adv. Carbohydr. Chem. Biochem. 1997, 52,
179Ϫ205.
B. Dugue, B. Meunier, C. Paoletti, Eur. J. Med. Chem. Chim.
Ther. 1983, 18, 551Ϫ554.
K. Yoshida, N. Iino, I. Koga, Chem. Pharm. Bull. 1975, 32,
1759Ϫ1763.
N. Harada, K. Ozaki, K. Oda, N. Nakanishi, M. Ohashi, T.
Hashiyama, K. Tsujihara, Chem. Pharm. Bull. 1997, 45,
1156Ϫ1162.
Acknowledgments
Financial assistance from the Sydney University Cancer Research
Fund (M. M. H.) and the award of a University of Sydney Post-
graduate Scholarship (A. R. G.) are gratefully acknowledged.
[20]
[21]
[22]
[1]
W. Goodwin, A. F. Smith, E. C. Horning, J. Am. Chem. Soc.
1959, 81, 1903Ϫ1908.
L. K. Dalton, S. Demerac, B. C. Elmes, J. W. Loder, J. M.
[23]
[24]
[2]
Swan, T. Teitei, Aust. J. Chem. 1967, 20, 2715Ϫ2727.
J.-B. Le Pecq, N. Dat-Xuong, C. Gosse, C. Paoletti, Proc. Natl.
[3]
[25]
[26]
[27]
[28]
[29]
[30]
[31]
Acad. Sci. USA 1974, 71, 5078Ϫ5082.
C. Paoletti, J.-B. Le Pecq, N. Dat-Xuong, P. Juret, H. Garnier,
[4]
J.-L. Amiel, J. Rouesse, Recent Res. Cancer Res. 1980, 74,
107Ϫ123.
[5]
´
J. Roue¨sse, M. Spielmann, F. Turpin, T. Le Chevalier, M. Azab,
´
J. M. Mondesir, Eur. J. Cancer 1993, 29A, 856Ϫ859.
[6]
M. Ohashi, T. Oki, Exp. Opin. Ther. Patents 1996, 6,
1285Ϫ1294.
[7]
[8]
C. Auclair, Arch. Biochem. Biophys. 1987, 259, 1Ϫ14.
´
´
P. Fosse, B. Rene, M. Charra, C. Paoletti, J.-M. Saucier, Mol.
Pharmacol. 1992, 42, 590Ϫ595.
[9]
E. Multon, J.-F. Riou, D. LeFevre, J.-C. Ahomadegbe, G. Riou,
Biochem. Pharmacol. 1989, 38, 2077Ϫ2086.
P. Formisyn, F. Pautet, C. Tran Minh, J. Bourgois, J. Pharm.
Biomed. Anal. 1992, 10, 427Ϫ436.
[10]
[11]
[12]
[13]
[32]
[33]
´
´
B. Norden, T. Kurucsev, J. Mol. Recognit. 1994, 7, 141Ϫ156.
A. H. Elcock, A. Rodger, W. G. Richards, Biopolymers 1996,
B. Norden, M. Kubista, T. Kurucsev, Q. Rev. Biophys. 1992,
39, 309Ϫ326.
25, 51Ϫ170.
M. A. Ismail, K. J. Sanders, G. C. Fennell, H. C. Latham, P.
Wormell, A. Rodger, Biopolymers 1998, 46, 127Ϫ143.
M. M. Harding, A. R. Grummitt, Mini-Rev. Med. Chem., in
press.
[34]
A. Rodger, Methods Enzymol. 1993, 226, 232Ϫ258.
Received July 19, 2002
[O02401]
Eur. J. Org. Chem. 2003, 63Ϫ71
71