In vitro fluorine-19 nuclear magnetic resonance study of the liberation of antitumor nitrogen mustard from prodrugs

Article abstract of DOI:10.1039/b111549a

The syntheses of two novel glucuronide prodrugs of nitrogen mustard with a fluorine tag are reported. These prodrugs, which differ only by the presence or not of a spacer, were designed for selective activation in the necrotic area of tumors by β-glucuron

Full text of DOI:10.1039/b111549a

View Article Online / Journal Homepage / Table of Contents for this issue
In vitro fluorine-19 nuclear magnetic resonance study of the
liberation of antitumor nitrogen mustard from prodrugs
Frédéric Schmidt* and Claude Monneret
1
UMR 176 CNRS-Institut Curie, Section Recherche 26, rue d’Ulm, 75248 Paris cedex 05,
France. E-mail: Frederic.Schmidt@curie.fr; Fax: ϩ33-1/4234-6631; Tel: ϩ33-1/4234-6664
Received (in Cambridge, UK) 19th December 2001, Accepted 25th March 2002
First published as an Advance Article on the web 24th April 2002
The syntheses of two novel glucuronide prodrugs of nitrogen mustard with a fluorine tag are reported. These
prodrugs, which differ only by the presence or not of a spacer, were designed for selective activation in the necrotic
area of tumors by β-glucuronidase. Kinetics of enzymatic hydrolysis of these prodrugs were determined by 19F-NMR
and HPLC in vitro studies. The spacer-containing prodrug was found to be the more stable and more quickly cleaved
than the other. Thereby it appeared to be a promising candidate for in vivo studies.
line phosphatase,14 and Springer et al. focused their studies on
Introduction
carboxypeptidase G2.15–19 Work by Roffler20–22 and in our
Lack of selectivity is a major limitation in cancer chemo-
laboratory23–28 has been centered on the use of β-glucuronidase.
therapy. Indeed, most drugs are not selective enough to be used
Noteworthy is the fact that the only prodrugs based on the
in optimal dose without severe drawbacks. A solution to this
ADEPT strategy which have entered clinical trials are prodrugs
problem would be to target the active drugs to the cancer cells.
of nitrogen mustards.15–17,29
In this context, a number of protocols using antibody–drug
To develop ADEPT and PMT studies at the preclinical level
there is a need to find methods to monitor events in vivo in real
time. This is particularly relevant for the study of unstable
nitrogen mustard drugs. In vivo fluorine-19 NMR spectroscopy
conjugates have been conducted.1 However, few conjugates have
reached the level of clinical trials, primarily due to problems of
tumor penetration, internalisation, and antitumor activity of
the conjugate.2
allows non-invasive real-time identification of administered
Alternatively, the ADEPT (Antibody-Directed Enzyme
fluorinated compounds inside human tumors. Furthermore, for
Prodrug Therapy) strategy3 provides the means to selectively
kinetic measurements (prodrug : drug ratios) 19F NMR offers
deliver an active drug to its intended target from an inactive
advantages, such as a spin of ¹, 100% natural abundance, a large
¯
²
prodrug precursor. In this approach, an enzyme is first targetted
to the tumor site by monoclonal antibody(Mab)–antigen
recognition, and in the second step enzymatic cleavage of the
prodrug liberates the active drug at the tumor cell surface. More
convenient is the PMT (Prodrug MonoTherapy) strategy4–6
based on the presence of tumor-associated enzymes7 such as
β-glucuronidase. In normal tissues, this glycosidase is localised
in the lysosome compartment of the cell and found in low con-
centration in the blood. In contrast, it has been shown by
enzyme histochemistry that it is specifically liberated extracel-
lularly in necrotic areas. Monocytes and granulocytes are most
likely the cells responsible for this effect.6 Many of the glucuro-
nide prodrugs, such as the anthracycline prodrug HMR 1826,
which have been developed for ADEPT have also be used in the
PMT approach. From experiments run with human perfused
lungs, it appeared that administration of this prodrug led to
higher concentration of doxorubicin inside the tumor than
those resulting from conventional treatment by doxorubicin
itself.8
Prodrugs of nitrogen mustards have been developed in the
context of the ADEPT strategy.9,10 Nitrogen mustards are cyto-
toxic compounds which alkylate DNA.11,12 Interestingly, these
compounds induce less resistance than other classes of anti-
cancer agents, and are highly efficient against quiescent cancer
cells. An attractive feature of nitrogen mustard prodrugs is
that, whereas the prodrug is stable, the released drug is rapidly
transformed into an inactive species (reaction of the nitrogen
mustard with a nucleophilic species in the medium). This
instability effectively limits the damage to healthy cells in the
zone around the tumor.13
chemical-shift dispersion (200 ppm for organofluorine com-
pounds), high detection sensitivity (83% of that of protons),
and a low background signal. This technique has been used to
study the antitumoral drug 5-FU27,30–34 but, to date, not for the
study of nitrogen mustards.
As the nitrogen mustard prodrugs we have previously syn-
thesised did not contain fluorine,23 we have prepared two new
nitrogen mustards 8 and 18 with a fluorine tag. It must be
reminded that some fluorinated nitrogen mustard prodrugs
have already been reported by Springer et al.,35 but they widely
differ from the prodrugs which are reported in the present
study. Indeed, instead of glutamic-containing prodrugs as
reported by Springer et al.,35 the prodrugs that constitute the
basis of this study are phenol glucuronides. Compound 8 bears
an analogy to Roffler’s prodrug20 in that the drug is directly
linked to glucuronic acid. In compound 18, a self-immolative
spacer is present between the glucuronic acid and the drug.23
An attractive design feature in both systems is that the chemical
shifts for the free drug and the prodrug form are clearly
separated.
Results and discussion
Synthesis
To obtain prodrug 8, 2-fluoro-4-nitrophenol 2 was coupled
(99%) to the protected bromoglucuronate 1.36 The aromatic
nitro compound 3 was then converted to the nitrogen mustard 6
in three steps: catalytic hydrogenation to amine 4 (85%), alkyl-
ation with ethylene oxide to diol 5 (61%), and transformation of
the terminal alcohol functions to the corresponding chlorides
via mesylation and reaction with LiCl (79%). Finally, hydrolysis
Different enzymes have been exploited for the cleavage of
nitrogen mustard prodrugs. Wallace and Senter used an alka-
1302
J. Chem. Soc., Perkin Trans. 1, 2002, 1302–1308
This journal is © The Royal Society of Chemistry 2002
DOI: 10.1039/b111549a

View Article Online
of the ester groups on the glucuronic acid moiety was effec-
ted by treatment with sodium methoxide to give triol 7
(86%), followed by sodium hydroxide to give acid 8 (95%),
(Scheme 1).
Scheme 2 Synthesis of prodrug 18. Reagents and conditions: i) H2, Pd/
C (99%); ii) ethylene oxide, acetic acid (68%); iii) BnBr, KOH, ethanol
(38%); iv) MsCl, LiCl, pyridine (44%); v) H2, HCl, Pd/C (100%); vi)
i
COCl2, NEt3, THF (47%); vii) Pr2NEt, THF (66%); viii) MeONa,
MeOH (80%); ix) NaOH, H2O–acetone (96%).
Scheme 1 Synthesis of prodrug 8. Reagents and conditions: i) Ag2O,
acetonitrile (99%); ii) H2, Pd/C (85%); iii) ethylene oxide, acetic acid
(61%); iv) MsCl, LiCl, pyridine (79%); v) MeONa, MeOH (86%); vi)
NaOH, H2O–acetone (95%).
To obtain prodrug 18, an approach was devised wherein a
phenolic nitrogen mustard is coupled to an intermediate in
which the glucuronic acid motif is already linked to the aro-
matic spacer. 2-Fluoro-4-nitrophenol 2 was reduced to 2-fluoro-
4-aminophenol37 9 by hydrogenation over Pd/C (99%). The
amine function in 9 was then alkylated through reaction with
ethylene oxide to give 10 (68%), and the phenol function was
then protected as the benzyl ether 11 (38%). Conversion of
this intermediate to chloride 12 (44%), followed by reductive
O-debenzylation in acidic media, provided the phenolic nitro-
gen mustard 13 as its hydrochloride salt. This unstable
compound was immediately converted to the corresponding
chloroformate 14 by reaction with phosgene in the presence of
triethylamine (47%). Chloroformate 14 was then treated with
amine 1525 in the presence of diisopropylethylamine to give 16
(66%). Two subsequent deprotection steps (77% overall) gave
prodrug 18, via 17 (Scheme 2).
Liberation of the drug
Scheme 3 Cleavage of prodrug 8.
Both prodrugs 8 and 18 were designed such that the same
fluorinated nitrogen mustard mustard 20 (and glucuronic acid
19) would be liberated in the presence of β--glucuronidase
(Scheme 3 and 4). For prodrug 18, the spacer would also
undergo cyclisation to give 21 as a secondary product. The
released nitrogen mustard 20 was immediately transformed into
a more stable compound 10 by reaction with water as the
nucleophile.
Cytotoxicity
In comparison with the non-fluorinated nitrogen mustard, the
fluorinated analog 20 is slightly less cytotoxic (IC50 15.4 µM
versus 10.5 µM). The behavior of the fluorinated prodrugs 8
and 18 is similar to that of the corresponding non-fluorinated
prodrugs. All these prodrugs are less cytotoxic (IC50 > 400
µM) than the corresponding drugs, but, in the presence of
J. Chem. Soc., Perkin Trans. 1, 2002, 1302–1308
1303

View Article Online
aqueous buffered solution. The chemical shifts were measured
with trifluoroacetic acid as the reference31,38 and all fluorine
spectra were recorded with proton decoupling in order to
simplify the signals.
The prodrug 8 without spacer gave a signal at δ Ϫ55.2 for the
fluorine. For the prodrug 18 with spacer, fluorine appears as
three signals δ Ϫ52.5, Ϫ53.0, and Ϫ53.2. By heating at 90 ЊC,
the signals coalesced to a unique signal at δ Ϫ52.9. The inter-
pretation of that phenomenon is a mixture of rotamers due, in
one part, to the bond cis or trans between the carbamate nitro-
gen and the carbonyl, and in the other part to another restricted
rotation about a different bond. Rotamers could also be
detected in 1H-NMR for the N-methyl signal, for instance in the
protected prodrug 16. For the measurements, these three signals
were integrated together.
The free nitrogen mustard introduced as its chlorhydrate
(hydrochloride) 13 gave a broad signal at δ Ϫ59.1. After
hydrolysis, the drug gave the dialcohol 10, which gave a signal at
δ Ϫ45.7.
All these data indicate that the various signals are different
enough to allow their measurement by integration.
Kinetics (NMR and HPLC) of drug release
Scheme 4 Cleavage of prodrug 18.
The kinetics of drug release by β--glucuronidase (E. coli) from
both prodrugs were measured by 19F-NMR and HPLC (UV
detection). In all cases, rapid hydrolysis of the glucuronyl
moiety occurred with release of the nitrogen mustard 20.
By fluorine NMR, starting from the prodrug, the signals
corresponding to the free nitrogen mustard 20 and the diol 10
resulting from the hydrolysis of the drug appeared and
increased (Figs 2 and 3). At the end, only the signal of the
β--glucuronidase, the cytotoxicity of the free drug 20 was
restored.
Stability
In the absence of β--glucuronidase, we could not detect by
HPLC or 19F-NMR any compound resulting from the cleav-
age of the glycosidic bond. Nevertheless, the two prodrugs 8
and 18 and the free drug 20 were not completely stable in
solution. Each nitrogen mustard moiety was hydrolysed to the
corresponding diol by displacement of the chlorine atoms.
Their stabilities were measured by HPLC measurements in
phosphate buffer at pH 7.2. Interestingly, it was noticed that the
transformation of the drug into prodrug led to a stabilisation.
Thus, the half-life of the free drug 20 was 14 min whereas those
of the prodrugs 8 (without spacer) and 18 (with spacer) were 32,
and 170 min, respectively (Fig. 1). The increasing stability can
Fig. 2 Hydrolysis of prodrug 8 (NMR measurements).
Fig. 1 Stability of prodrugs and free drug.
be related to the decreasing electrophilic character of the nitro-
gen, promoting the formation of the intermediate aziridinium,
as a phenol is more electro-donating than is a glycoside oxygen
and much more than is a carbamate oxygen.
19F NMR
Before starting in vitro kinetic measurements, different
species had to be identified by their 19F-NMR signals in
Fig. 3 Hydrolysis of prodrug 18 (NMR measurements).
1304
J. Chem. Soc., Perkin Trans. 1, 2002, 1302–1308

View Article Online
dialcohol 10 was seen. Comparison between the two prodrugs
was also very clear. Without a spacer, the cleavage was slower
and needed a higher concentration of enzyme. For the same
prodrug concentration, the half-life of 8 with an enzyme con-
centration of 2000 units mLϪ1 (1527 µg mLϪ1) was 23 min, and
for 18 with an enzyme concentration of 500 units mLϪ1 (382 µg
mLϪ1), the half-life was 15 min.
were taken at various times and analysed by HPLC after dilu-
tion with eluent.
Enzymatic cleavage by E. coli ꢀ-D-glucuronidase (HPLC)
Hydrolysis was investigated by incubating a solution of 250 µg
(mL of prodrug)Ϫ1 and variable concentrations of E. coli
β-glucuronidase in 0.02 M phosphate buffer (pH 7.2) at 37 ЊC.
Aliquots (100 µL) were taken at various times and analysed by
HPLC after dilution with 300 µL of eluent.
By HPLC (UV detection), we could clearly compare the con-
centrations of prodrugs, free drug 20, and spacer in its cyclised
form, 21, which accumulated. Once again, the prodrug 18 was
cleaved faster than was the prodrug 8. For a concentration of
250 µg mLϪ1) in prodrug and a concentration of 250 µg mLϪ1
)
In vitro Cytotoxicity
of β--glucuronidase, the half-lives of prodrugs were respect-
ively 7 min for 18, and 15 min for 8. The same 15 min half-life
for the prodrug 18 could be observed under treatment with a
2.5-less-fold concentration in enzyme.
Cytotoxicity was tested against LoVo cells [human colon cancer
cell line obtained from ATCC (Rockville, MD, USA)] using
the Methylene Blue assay. The concentration of prodrug or
drug inducing 50% of inhibition (IC50) was calculated from the
dose–response curve.
Conclusions
Methyl (2-fluoro-4-nitrophenyl 2,3,4-tri-O-acetyl-ꢀ-D-
glucopyranosid)uronate 3
From these results, it clearly appears that these prodrugs can
be used for in vitro studies, and consequently for in vivo
studies. However, among them, the spacer-containing prodrug
undoubtly displayed both better stability and easier enzymatic
cleavage. For both reasons, such a prodrug was selected for
further investigations, in particular by in vivo 19F-NMR for the
detection and the quantification of the drug in tumors.
To a solution of methyl 2,3,4-tri-O-acetyl-bromo-1-deoxy-α-1-
-glucopyranuronate36 1 (16.1 g, 40.5 mmol) in 280 mL of
acetonitrile were added silver oxide (22 g, 94.9 mmol) and
2-fluoro-4-nitrophenol 2 (6.27 g, 39.9 mmol). The mixture was
stirred overnight at room temperature, filtered over Celite, and
concentrated. The product was purified by column chrom-
atography on silica using CH2Cl2–MeOH (97.5 : 2.5) as eluent
to give the glucuronate 3 (18.8 g, 99%) as a white solid, mp 150
ЊC (from toluene); [α]2D0 Ϫ48 (c 0.808 in CHCl3) (Found: C,
48.30; H, 4.26; N, 2.96; F, 4.01. Calc. for C19H20FNO12: C,
48.21; H, 4.26; N, 2.93; F, 3.90%); νmax/cmϪ1 (CDCl3) 1761
(CO), 1533, 1351 (NO2); δH (300 MHz; CDCl3; Me4Si) 8.06–
7.99 (2H, m, H3Ј, H5Ј), 7.33–7.17 (1H, m, H6Ј), 5.40–5.26 (4H,
m, H1, H2, H3, H4), 4.25 (1H, d, J 8.7, H5), 3.74 (3H, s,
CO2CH3), 2.10 (3H, s, Ac), 2.07 (3H, s, Ac), 2.06 (3H, s, Ac);
m/z (CI, NH3) 491 [M ϩ NH4]ϩ.
Experimental
General
Melting points were taken on a Koffler Bench and are un-
corrected. Optical rotations were obtained on a Perkin-Elmer
241 polarimeter. [α]D-Values are given in units of 10Ϫ1 deg cm2
g
Ϫ1. Infrared spectra were measured on a Perkin–Elmer 1710
FT/IR spectrometer. 1H-NMR (300 MHz) spectra were
recorded on a Bruker Avance 300 spectrometer (chemical shifts
δ in ppm and coupling constants J in Hz). For NMR descrip-
tions, unprimed numbers are used for the glucuronic moiety,
primed numbers for the nitrogen mustard, and double-primed
numbers for the nitroaromatics. Chemical ionisation mass
spectra (CI-MS) were recorded on a NERMAG R10–10C
spectrometer. Electrospray ionisation spectra were acquired
using a quadrapole instrument with a mass of charge (m/z)
range of 2000.
Methyl (2-fluoro-4-aminophenyl 2,3,4-tri-O-acetyl-ꢀ-D-gluco-
pyranosid)uronate 4
The nitro derivative (200 mg, 0.423 mmol) 3 was dissolved in
30 mL of ethanol, 10% palladium on carbon (50 mg) was
added, and the mixture was hydrogenated overnight at room
temperature. After filtration over Celite and evaporation, the
product was purified by column chromatography on silica
using CH2Cl2–MeOH (97.5 : 2.5) as eluent to give the amine 4
(160 mg, 85%) as a white solid, mp 106 ЊC (from toluene); [α]2D0
Ϫ39 (c 0.804 in CHCl3); νmax/cmϪ1 (CDCl3) 3694, 3497 (NH),
1758 (CO); δH (300 MHz; CDCl3; Me4Si) 7.02 (1H, dd, JH–F (m)
Column chromatography was carried out on Merck silica
Kieselgel 60 (230–400 Mesh).
Microanalyses were performed on a Perkin–Elmer 2400
CHN microanalyser.
19F-NMR
8.9, JH–H 8.9, H6Ј), 6.40 (1H, dd, JH–F 12.2, JH–H
2.7,
(o)
(o)
(m)
H3Ј), 6.33 (1H, dd, JH–H (o) 8.7, JH–H (m) 1.8, H5Ј), 5.32–5.21 (3H,
m, H2, H3, H4), 4.85 (1H, d, J 7.3 (β), H1), 4.05 (1H, J 9.0,
H5), 3.76 (3H, s, CO2CH3), 3.65 (2H, s, NH2), 2.10 (3H, s, Ac),
2.04 (3H, s, Ac), 2.03 (3H, s, Ac); m/z (CI, NH3) 461 [M ϩ
NH4]ϩ; (CI, CH4) 444.130 [M ϩ H]ϩ. C19H23FNO10 requires
m/z, 444.130.
19F-NMR (282 MHz) signals were measured on a Bruker
Avance 300 at room temperature with an external reference
(CF3CO2H in D2O 5% v/v). Proton CPD (Composite Pulse
Decoupling) coupling was made possible by a QNI (Quad-
rupole Nucleus Inverse) Probe (Bruker). The prodrug (concen-
tration 5 mM) was dissolved in D2O buffered at pH 7.2 by a
phosphate buffer (0.02 M).
Methyl {2-fluoro-4-[bis(2-hydroxyethyl)amino]phenyl 2,3,4-tri-
O-acetyl-ꢀ-D-glucopyranosid}uronate 5
HPLC Analysis
To a solution of the amine 4 (1.81 g, 14.1 mmol) in 35 mL of
acetic acid at 0 ЊC was added ethylene oxide (2.42 mL, 48.4
mmol) and the mixture was stirred for 48 hours at room tem-
perature. The acetic acid was evaporated off under vacuum.
The solution was neutralised with saturated aqueous sodium
bicarbonate. After drying over sodium sulfate and evaporation,
the product was purified by column chromatography on silica
using EtOAc as eluent to give the dialcohol 5 (1.33 g, 61 %) as a
white solid, mp 115 ЊC (from toluene); [α]2D0 Ϫ63 (c 0.99 in
CHCl3) (Found: C, 51.97; H, 5.64; N, 2.65; F, 3.59. Calc. for
Analytical HPLC was carried out using a Gilson HPLC system
with UV detection at 254 nm. The separation was performed on
a reversed-phase column (Zorbax SB-C18, 5 µm, 150 × 4.6 mm)
using the isocratic conditions (0.6 mL minϪ1) of 40% 0.02 M
phosphate buffer (pH 3) and 60% acetonitrile.
Stability of compounds in a buffer solution
A solution of 250 µL of prodrug in 0.02 M phosphate buffer
(pH 7.2) was incubated for various times at 37 ЊC. Aliquots
J. Chem. Soc., Perkin Trans. 1, 2002, 1302–1308
1305

View Article Online
C23H30FNO12: C, 51.98; H, 5.69; N, 2.64; F, 3.57%); νmax/cmϪ1
(CDCl3) 3621 (OH), 1758 (CO); δH (300 MHz; CDCl3; Me4Si)
7.09 (1H, dd, JH–F (m) 9.1, JH–H (o) 9.1, H6Ј), 6.41 (1H, dd, JH–F (o)
CH2N); δF (382 MHz; D2O buffered; TFA) Ϫ55.2; m/z (ES) 450
(2Cl35), 452 (Cl35, Cl37), 454 (2Cl37) [M ϩ Na]ϩ; (ES) 427.061
(2Cl35). C16H20FNO7Cl2 requires M, 427.060.
14.0, JH–H
2.9, H3Ј), 6.34 (1H, dd, JH–H 8.6, JH–H
2.8,
(m)
(o)
(m)
2-Fluoro-4-[bis(2-hydroxyethyl)amino]phenol 10
H5Ј), 5.35–5.21 (3H, m, H2, H3, H4), 4.85 (1H, d, J 7 (β), H1),
4.03 (1H, J 9.3, H5), 3.84 (4H, t, J 4.8, CH2OH), 3.76 (3H, s,
CO2CH3), 3.53 (4H, t, J 4.8, CH2N), 3.09 (2H, s, OH), 2.09 (3H,
s, Ac), 2.04 (3H, s, Ac), 2.02 (3H, s, Ac); m/z (CI, NH3) 532
[M ϩ H]ϩ.
To a solution of the aminophenol 937 (4.0 g, 31.5 mmol) in 85
mL of acetic acid and 85 mL of water at 0 ЊC was added ethyl-
ene oxide (15 mL, 300 mmol) and the mixture was stirred at
room temperature for 72 hours. The acetic acid was evaporated
off under vacuum. The product was purified by two successive
column chromatographic separations on silica using EtOAc for
the first and CH2Cl2–MeOH (9 : 1) for the second eluent, giving
the triol 10 (4.83 g, 68%) as a pink gum which turned brown
within a few minutes. We did not obtain a sample suitable for
microanalysis and used this crude compound in the next step.
νmax/cmϪ1 (KBr) 3392 (OH); δH (300 MHz; DMSO; Me4Si) 6.78
(1H, dd, JH–F (m) 9.2, JH–H (o) 9.2, H6Ј), 6.70–6.10 (2H, H3Ј, H5Ј),
4.72 (2H, OH), 3.67–3.15 (8H, OCH2, NCH2); δF (382 MHz;
D2O buffered; TFA) Ϫ45.7; m/z (CI, CH4) 216.104 [M ϩ H]ϩ.
C10H15FNO3 requires m/z, 216.104.
Methyl {2-fluoro-4-[bis(2-chloroethyl)amino]phenyl 2,3,4-tri-O-
acetyl-ꢀ-D-glucopyranosid}uronate 6
To a solution of the diol 5 (194 mg, 0.365 mmol) in 2 mL
pyridine at 0 ЊC was added methanesulfonyl chloride (0.15 mL,
0.365 mmol) and the mixture was stirred for 5 min at room
temperature, then for 15 min at reflux. Lithium chloride (50 mg,
1.18 mmol) was added and the mixture was kept for 1 hour at
reflux. The cooled mixture was poured into aqueous citric acid,
extracted with EtOAc, and the extract was dried over sodium
sulfate and evaporated. The product was purified by column
chromatography on silica using CH2Cl2–MeOH (97.5 : 2.5) as
eluent to give the dichloride 6 (164 mg, 79%) as a white solid,
mp 175 ЊC (from toluene); [α]2D0 Ϫ31 (c 1.06 in CHCl3) (Found:
C, 48.75; H, 5.00; N, 2.46; F, 3.18. Calc. For C23H28FNO10Cl2:
C, 48.60; H, 4.97; N, 2.46; F, 3.34%); νmax/cmϪ1 (CDCl3) 1758
1-Benzyloxy-2-fluoro-4-[bis(2-hydroxyethyl)amino]benzene 11
To a solution of the triol 10 (3.71 g, 17.2 mmol) in 35 mL of
95% ethanol was added benzyl bromide (2.06 mL, 17.3 mmol),
then commercial 85% potassium hydroxide pellets (1.14 g,
17.3 mmol) were added and the mixture was stirred for 2 hours
at reflux, evaporated, the residue was taken up in EtOAc, and
the solution was washed with water. The organic phase was
dried over sodium sulfate, evaporated, and purified by column
chromatography on silica using EtOAc as eluent to give the
benzyl ether 11 (3.71 g, 38%) as a viscous brown liquid;
νmax/cmϪ1 (CDCl3) 3404 (OH); δH (300 MHz; CDCl3; Me4Si)
(CO); δH (300 MHz; CDCl3; Me4Si) 7.14 (1H, dd, JH–F
9.1,
(m)
JH–H 9.1, H6Ј), 6.41 (1H, dd, JH–F 13.6, JH–H 3.0, H3Ј),
(o)
(o)
(m)
6.34 (1H, dd, JH–H 8.7, JH–H
2.9, H5Ј), 5.33–5.21 (3H, m,
(m)
(o)
H2, H3, H4), 4.87 (1H, d, J 7.5 (β), H1), 4.05 (1H, J 9.4, H5),
3.77 (3H, s, CO2CH3), 3.72–3.66 (4H, m, CH2Cl), 3.61–3.58
(4H, m, CH2N), 2.11 (3H, s, Ac), 2.05 (3H, s, Ac), 2.03 (3H, s,
Ac); m/z (CI, NH3) 568 (2Cl35), 570 (Cl35, Cl37), 572 (2Cl37)
[M ϩ H]ϩ.
7.50–7.20 (5H, br, Ph), 6.89 (1H, dd, JH–F
H6Ј), 6.50 (1H, dd, JH–F 14.3, JH–H
9.2, JH–H 9.2,
3.0, H3Ј), 6.35 (1H,
(m)
(m)
(o)
(o)
Methyl {2-fluoro-4-[bis(2-chloroethyl)amino]phenyl ꢀ-D-gluco-
pyranosid}uronate 7
ddd, JH–H
9.2, JH–H
3.0, JH–F
1.1, H5Ј), 5.04 (2H, s,
(o)
(m)
(p)
CH2Ph), 3.81 (4H, t, J 4.8, OCH2), 3.49 (4H, t, J 4.7, NCH2),
3.11 (2H, s, OH); m/z (CI, NH3) 306 [M ϩ H]ϩ.
To a solution of 6 (574 mg, 10.1 mmol) in 30 mL of CH2Cl2
and 30 mL of MeOH at Ϫ15 ЊC was added sodium methoxide
(30 mg, 0.555 mmol) and the mixture was stirred for 4 hours at
Ϫ15 ЊC then further sodium methoxide (10 mg, 0.185 mmol)
was added and the mixture stirred at Ϫ15 ЊC overnight, neutral-
ised with IRC 50 (Hϩ) Amberlite, and filtered. The solution was
concentrated and the residue was purified by column chrom-
atography on silica using EtOAc as eluent to give 7 (384 mg,
86%) as a white solid, mp 64 ЊC; [α]2D0 Ϫ62 (c 0.925 in CHCl3)
(Found: C, 45.76; H, 5.18; N, 3.12; Calc. for C17H22FNO7Cl2: C,
46.16; H, 5.01; N, 3.17%); νmax/cmϪ1 (KBr) 3349 (OH), 1746
1-Benzyloxy-2-fluoro-4-[bis(2-chloroethyl)amino]benzene 12
To a solution of the diol 11 (1.68 g, 5.5 mmol) in 11 mL of
pyridine at 0 ЊC was added methanesulfonyl chloride (1.66 mL,
21.4 mmol) and the mixture was stirred for 5 min at room tem-
perature, then for 15 min at reflux. Lithium chloride (0.33 g,
7.79 mmol) was added and the mixture was kept 15 min at
reflux. The cooled mixture was poured into aqueous citric acid,
extracted with EtOAc, and the extract was dried over sodium
sulfate and evaporated. The product was purified by column
chromatography on silica using cyclohexane–EtOAc (3 : 1) as
eluent and crystallisation from petroleum ether (80–110 ЊC) to
give the dichloride 12 (824 mg, 44%) as a beige solid, mp 76 ЊC
(Found: C, 59.84; H, 5.44; N, 3.94; F, 5.32. Calc. for C17H18-
FNOCl2: C, 59.66; H, 5.30; N, 4.09; F, 5.55%); νmax/cmϪ1
(CDCl3) 1519; δH (300 MHz; CDCl3; Me4Si) 7.46–7.23 (5H, br,
(CO); δH (300 MHz; CDCl3; Me4Si) 7.12 (1H, dd, JH–F
9.1,
(m)
JH–H 9.1, H6Ј), 6.42 (1H, dd, JH–F 12.5, JH–H 2.6, H3Ј),
(o)
(o)
(m)
6.33 (1H, dd, JH–H (o) 9.0, JH–H (m) 1.5, H5Ј), 4.72 (1H, d, J 6.1 (β),
H1), 4.14 (2H, s, OH), 3.95–3.82 (4H, m, H2, H3, H4, H5), 3.80
(3H, s, CO2CH3), 3.74–3.56 (8H, m, CH2Cl, CH2N), 1.96 (1H,
s,OH); m/z (CI, NH3) 442 (2Cl35), 444 (Cl35, Cl37), 446 (2Cl37)
[M ϩ H]ϩ.
Ph), 6.91 (1H, dd, JH–F
9.2, JH–H 9.2, H6Ј), 6.49 (1H, dd,
(m)
(o)
JH–F (o) 14.0, JH–H (m) 3.0, H3Ј), 6.33 (1H, ddd, JH–H (o) 8.9, JH–H (m)
2.9, JH–F 1.2, H5Ј), 5.06 (2H, s, CH2Ph), 3.70–3.58 (8H, m,
2-Fluoro-4-[bis(2-chloroethyl)amino]phenyl ꢀ-D-glucopyrano-
siduronic acid 8
(p)
NCH2, CH2Cl); m/z (CI, NH3) 342 (2Cl35), 344 (Cl35, Cl37), 346
(2Cl37) [M ϩ H]ϩ.
To a solution of ester 7 (70 mg, 0.158 mmol) in 5 mL of acetone
at Ϫ15 ЊC was added 1 M sodium hydroxide (0.5 mL, 0.5
mmol), the solution was stirred for 5 min at Ϫ15 ЊC, neutralised
with 1 M HCl, and evaporated. The residue was purified by
column chromatography on silica using CH3CN–H2O (80 : 20)
as eluent to give, after lyophilisation, acid 8 (64 mg, 95%) as a
white solid, mp 220 ЊC (dec.); [α]2D0 Ϫ48 (c 0.456 in CH3OH);
νmax/cmϪ1 (KBr) 3355 (OH), 1738 (CO); δH (300 MHz; CD3OD;
Me4Si) 7.19 (1H, dd, JH–F (m) 9.4, JH–H (o) 9.4, H6Ј), 6.53 (1H, dd,
JH–F (o) 14.2, JH–H (m) 2.9, H3Ј), 6.43 (1H, dd, JH–H (o) 8.8, JH–H (m)
2.9, H5Ј), 3.69–3.49 (13H, m, H1, H2, H3, H4, H5, CH2Cl,
2-Fluoro-4-[bis(2-chloroethyl)ammonio]phenol chloride 13†
Dry HCl gas was bubbled into a suspension of the benzyl ether
12 (710 mg, 2.08 mmol) in absolute ethanol until dissolution.
Palladium on carbon (10%; 0.2 g) was added and the com-
pound was hydrogenated for 3 hours. The mixture was filtered
and evaporated under vacuum at room temperature to give the
† Strictly speaking, the center takes precedence for principal group:13 is
bis(2-chloroethyl)(3-fluoro-4-hydroxyphenyl)ammonium chloride
1306
J. Chem. Soc., Perkin Trans. 1, 2002, 1302–1308

View Article Online
salt (600 mg, quantitative) as a white solid. Compound 13 was
utilised as soon as possible without any further purification
as it is rather unstable, especially in solution. νmax/cmϪ1 (CDCl3)
3404 (OH), 1525; δH (300 MHz; DMSO; Me4Si) 6.85 (1H, dd,
JH–F (m) 9.3, JH–H (o) 9.3, H6Ј), 6.63 (1H, dd, JH–F (o) 14.4, JH–H (m)
2.7, H3Ј), 6.44 (1H, dd, JH–H (o) 8.8, JH–H (m) 2.0, H5Ј), 3.71–3.60
(8H, m, NCH2, CH2Cl); δF (382 MHz; D2O buffered; TFA)
Ϫ59.1; m/z (CI, NH3) 252 (2Cl35), 254 (Cl35, Cl37), 256 (2Cl37)
[MH]ϩ.
96%) as a white solid, mp 180 ЊC; [α]2D0 Ϫ43 (c 0.174 in MeOH);
νmax/cmϪ1 (KBr) 3418 (OH), 1718 (CO), 1520, 1350 (NO2); δH
(300 MHz; CD3OD; Me4Si) ) 8.28–8.22 (2H, m, H3Љ, H5Љ), 7.52
(1H, d, J 8.8, H6Љ), 7.20–7.05 (1H, m, H6Ј), 6.56–6.43 (2H, m,
H3Ј, H5Ј), 5.21 (1H, br, H1), 3.87 (1H, d, J 9.1, H5), 3.72–3.26
(14H, m, NCH3, NCH2, CH2Cl, H2, H3, H4); δF (382 MHz;
D2O buffered; TFA) Ϫ52.5, Ϫ53.0, Ϫ53.2 (rotamers); m/z (ES)
644 (2Cl35), 646 (Cl35, Cl37), 648 (2Cl37) [M ϩ Na]ϩ; (ES)
621.093 (2Cl35). C24H26FN3O11Cl2 requires M, 621.093.
Methyl {2-[({2-fluoro-4-[bis(2-chloroethyl)amino]phenoxy}-
carbonyl)methylamino]-4-nitrophenyl 2,3,4-tri-O-acetyl-ꢀ-D-
glucopyranosid}uronate 16
Acknowledgements
We would like to thank Hoechst Marion Roussel for cyto-
toxicity measurements and Mrs Ioana Ungureanu for the
synthesis of some compounds during her one-month training
in our laboratory.
To a suspension of chloride 13 (556 mg, 1.93 mmol) in
anhydrous THF at 0 ЊC was added a 20% solution of phosgene
in toluene (8.2 mL), followed by triethylamine (1 mL, 7.17
mmol). The mixture was stirred for 1 hour at 0 ЊC, then for 2
hours at room temperature. After filtration under argon and
evaporation at room temperature under vacuum, the com-
pound was purified by column flash chromatography on silica
using CH2Cl2 as eluent to give the chloroformate 14 (282 mg,
47%) as a pale pink liquid. This reactive compound was utilised
immediately as it cannot be kept in the absence of excess of
phosgene. Any attempt to characterise this product resulted in
loss of the chloroformate function.
To a solution of freshly prepared chloroformate 14 (282 mg,
0.896 mmol) in 10 mL of THF were added the aniline derivative
1525 (400 mg, 0.826 mmol) and diisopropylethylamine (0.25
mL, 1.44 mmol). The solution was brought to reflux for 2 hours
and was then evaporated. The product was purified by column
chromatography on silica using CH2Cl2–MeOH (95 : 5) to give
16 (415 mg, 66%) as a white solid, mp 108 ЊC; [α]2D0 Ϫ41 (c 0.888
in CHCl3) (Found: C, 48.83; H, 4.59; N, 5.45; Calc. for
C31H34FN3O14Cl2: C, 48.83; H, 4.49; N, 5.51%); νmax/cmϪ1
(CDCl3) 1759 (CO), 1520, 1351 (NO2); δH (300 MHz; CDCl3;
Me4Si) 8.28–8.19 (2H, m, H3Љ, H5Љ), 7.41 (1H, d, J 8.7, H6Љ),
References
1 D. Blakey, Acta Oncol., 1992, 31, 91–97.
2 R. Jain, Cancer Res., 1987, 47, 3039–3051.
3 K. Bagshawe, Br. J. Cancer, 1987, 56, 531–532.
4 K. Bosslet, R. Straub, M. Gerken, H. Petrul, J. Czech, J.-C. Florent,
M. Koch, F. Tillequin, F. Schmidt, C. Monneret and J.-P. Gesson,
in Neue Therapeutische Prinzipien in der Krebsbehandlung. 4.
Salzburger Symposium zur Lebensqualitaet Chronisch Kranker,
Georg Thieme Verlag, Stuttgart and New York, 1996, pp. 123–130.
5 K. Bosslet, J. Czech and D. Hoffmann, Tumor Targeting, 1995, 1,
45–50.
6 K. Bosslet, R. Straub, M. Blumrich, J. Czech, M. Gerken, B.
Sperker, H. K. Kroemer, J.-P. Gesson, M. Koch and C. Monneret,
Cancer Res., 1998, 58, 1195–1201.
7 A. Sinhababu and D. Thakker, Adv. Drug Deliv. Rev., 1996, 19, 241–
273.
8 T. Murdter, B. Sperker, K. Kivistö, M. McClellan, P. Fritz,
G. Friedel, A. Linder, K. Bosslet, H. Toomes, R. Diekesmann and
H. Kroemer, Cancer Res., 1997, 57, 2440–2445.
9 C. Springer and I. Niculescu-Duvaz, Anti-Cancer Drug Des., 1995,
10, 361–372.
10 P. Burke, Adv. Drug Deliv. Rev., 1996, 22, 331–340.
11 S. Rink, M. Salomon, M. Taylor, S. Rajur, L. McLaughlin and
P. Hopkins, J. Am. Chem. Soc., 1993, 115, 2551–2557.
12 G. Bauer and L. Povirk, Nucleic Acids Res., 1997, 25, 1211–1218.
13 D. Blakey, D. Davies, R. Dowell, S. East, P. Burke, S. Sharma,
C. Springer, A. Mauger and R. Melton, Br. J. Cancer, 1995, 72,
1083–1088.
7.00 (1H, dd, JH–F
8.9, JH–H 8.9, H6Ј), 6.42–6.34 (2H, m,
(m)
(o)
H3Ј, H5Ј), 5.45–5.30 (4H, m, H1, H2, H3, H4), 4.29 (1H, d,
J 9.2, H5), 3.77–3.59 (11H, m, OCH3, NCH2, CH2Cl), 3.40,
3.27 (3H, 2s, NCH3 rotamers) 2.14–2.02 (9H, m, OAc); m/z (CI,
NH3) 762 (2Cl35), 764 (Cl35, Cl37), 766 (2Cl37) [M ϩ H]ϩ.
14 P. Wallace and P. Senter, Bioconjugate Chem., 1991, 2, 349–352.
15 C. Springer, G. Poon, S. Sharme and K. Bagshawe, Cell Biophys.,
1993, 22, 9–26.
16 C. Springer, R. Dowell, P. Burke, E. Hadley, D. Davies, D. Blakey,
R. Melton and I. Niculescu-Duvaz, J. Med. Chem., 1995, 38, 5051–
5065.
Methyl{2-[({2-fluoro-4-[bis(2-chloroethyl)amino]phenoxy}-
carbonyl)methylamino]-4-nitrophenyl ꢀ-D-glucopyranosid}-
uronate 17
To a solution of 16 (212 mg, 0.278 mmol) in 10 mL of MeOH at
Ϫ20 ЊC was added sodium methoxide (5 mg, 0.0925 mmol) and
the mixture was stirred for 6 hours at Ϫ20 ЊC, neutralised with
IRC 50 (Hϩ) Amberlite, and filtered. The solution was concen-
trated and the residue was purified by column chromatography
on silica using EtOAc as eluent to give 17 (361 mg, 80%) as a
white solid, mp 96 ЊC; [α]2D0 Ϫ68 (c 1.01 in CHCl3); νmax/cmϪ1
(CDCl3) 3438 (OH), 1724 (CO), 1521, 1350 (NO2); δH (300
MHz; CDCl3; Me4Si) 8.23–8.14 (2H, m, H3Љ, H5Љ), 7.22 (1H, d,
J 8.8, H6Љ), 7.10–6.92 (1H, m, H6Ј), 6.53–6.38 (2H, m, H3Ј,
H5Ј), 5.07 (1H, br, H1), 4.21 (1H, s, OH), 4.14–4.04 (1H, m,
H5), 3.89–3.25 (17H, m, OCH3, NCH3, NCH2, CH2Cl, H2, H3,
H4), 1.84 (2H, s, OH); m/z (CI, NH3) 636 (2Cl35), 638 (Cl35,
Cl37), 640 (2Cl37) [M ϩ H]ϩ.
17 J. Martin, S. Stribbling, G. Poon, R. Begent, M. Napier, S. Sharma
and C. Springer, Cancer Chemother. Pharmacol., 1997, 40, 189–201.
18 S. Stribbling, J. Martin, R. Pedley, J. Boden, S. Sharma and
C. Springer, Cancer Chemother. Pharmacol., 1997, 40, 277–284.
19 D. Niculescu-Duvaz, I. Niculescu-Duvaz, F. Friedlos, J. Martin,
R. Spooner, L. Davies, R. Marais and C. Springer, J. Med. Chem.,
1998, 41, 5297–5309.
20 S. Wang, J. Chern, M. Yeh, J. Ng, E. Tung and S. Roffler, Cancer
Res., 1992, 52, 4484–4491.
21 T. Chen, S. Wei, B. Chen, J. Chern, M. Wu, P. Liu and S. Roffler,
Br. J. Cancer, 1999, 79, 1378–1385.
22 T. Cheng, W. Chou, B. Chen, J. Chern and S. Roffler, Biochem.
Pharmacol., 1999, 58, 325–328.
23 F. Schmidt, J.-C. Florent, C. Monneret, R. Straub, J. Czech,
M. Gerken and K. Bosslet, Bioorg. Med. Chem. Lett., 1997, 7, 1071–
1076.
24 F. Schmidt, J.-C. Florent and C. Monneret, C. R. Acad. Sci., Ser. IIc:
Chim., 1998, 1, 421–424.
25 S. Desbene, H. van Dufat-Trinh, S. Michel, F. Tillequin, M. Koch,
F. Schmidt, J.-C. Florent, C. Monneret, R. Straub, J. Czech,
M. Gerken and K. Bosslet, Anti-Cancer Drug Des., 1999, 14, 93–
106.
26 J.-C. Florent, X. Dong, G. Gaudel, S. Mitaku, C. Monneret,
J.-P. Gesson, J.-C. Jacquesy, M. Mondon, B. Renoux, S. Andriano-
menjanahary, S. Michel, M. Koch, F. Tillequin, M. Gerken,
J. Czech, R. Straub and K. Bosslet, J. Med. Chem., 1998, 41, 3572–
3581.
2-[({2-Fluoro-4-[bis(2-chloroethyl)amino]phenoxy}carbonyl)-
methylamino]-4-nitrophenyl ꢀ-D-glucopyranosiduronic acid 18
To a solution of ester 17 (87 mg, 0.137 mmol) in 10 mL of
acetone at Ϫ15 ЊC was added 1 M sodium hydroxide (0.3 mL,
0.3 mmol), and the solution was stirred for 15 min at Ϫ15 ЊC,
neutralised with 1 M HCl, and evaporated. The residue was
purified by column chromatography on silica using CH3CN–
H2O (80 : 20) as eluent to give, after lyophilisation, 18 (82 mg,
J. Chem. Soc., Perkin Trans. 1, 2002, 1302–1308
1307

View Article Online
27 R. Madec-Lougerstay, J.-C. Florent and C. Monneret, J. Chem.
Soc., Perkin Trans. 1, 1999, 1369–1375.
28 F. Schmidt, I. Ungureanu, R. Duval, A. Pompon and C. Monneret,
Eur. J. Org. Chem., 2001, 2129–2134.
33 C. Presant, W. Wolf, V. Waluch, C. Wiseman, P. Kennedy,
D. Blayney and R. Brechner, Lancet, 1994, 343, ii, 1184–1187.
34 E. Aboagye, D. Artemov, P. Senter and Z. Bhujwalla, Cancer Res.,
1998, 58, 4075–4078.
29 K. Bagshawe, S. Sharma, P. Burke, R. Melton and R. Knox, Curr.
Opin. Immunol., 1999, 11, 579–583.
30 J.-L. Guerquin-Kern, A. Volk, E. Chenu, R. Lougerstay-Madec,
C. Monneret, J.-C. Florent, D. Carrez and A. Croisy, NMR Biomed.,
2000, 13, 306–310.
31 J.-L. Guerquin-Kern, F. Leteurtre, A. Croisy and J.-M. Lhoste,
Cancer Res., 1991, 51, 5770–5773.
32 P. Price and J. Griffiths, Lancet, 1994, 343, ii, 1174–1175.
35 C. Springer, I. Niculescu-Duvaz and R. Pedley, J. Med. Chem., 1994,
37, 2361–2370.
36 G. Bollenback, J. Long, D. Benjamin and J. Lindquist, J. Am. Chem.
Soc., 1955, 77, 3310–3315.
37 D. Aymes and M. Paris, Bull. Soc. Chim. Fr., 1980, 2, 175–
178.
38 P. Hansen, H. Dettman and B. Sykes, J. Magn. Reson., 1985, 62,
487–496.
1308
J. Chem. Soc., Perkin Trans. 1, 2002, 1302–1308