Nitrobenzocyclophosphamides as potential prodrugs for bioreductive activation: Synthesis, stability, enzymatic reduction, and antiproliferative activity in cell culture

Article abstract of DOI:10.1016/S0968-0896(03)00459-0

In efforts to obtain potential anticancer prodrugs for gene-directed enzyme prodrug therapy using Eschericia coli nitroreductase, a series of four benzocyclophosphamide analogues were designed and synthesized incorporating a strategically placed nitro group in a position para to the benzylic carbon for reductive activation. All four analogues were found to be stable in phosphate buffer at pH 7.4 and 37 °C and were good substrates of E. coli nitroreductase with half lives between 7 and 24 min at pH 7.0 and 37 °C. However, only two analogues 6a and 6c, both with a benzylic oxygen in the phosphorinane ring para to the nitro group, showed a modest 33-36-fold enhanced cytotoxicity in E. coli nitroreductase-expressing cells. These results suggest that good substrate activity and the para benzylic oxygen are required for activation by E. coli nitroreductase. Compounds 6a and 6c represent a new structure type for reductive activation and a lead for further modification in the development of better analogues with improved selective toxicity to be used in gene-directed enzyme prodrug therapy.

Full text of DOI:10.1016/S0968-0896(03)00459-0

Bioorganic & Medicinal Chemistry 11 (2003) 4171–4178
Nitrobenzocyclophosphamides as Potential Prodrugs
for Bioreductive Activation: Synthesis, Stability, Enzymatic
Reduction, and Antiproliferative Activity in Cell Culture
Zhuorong Li,a,y Jiye Han,a Yongying Jiang,a Patrick Browne,b
Richard J. Knoxb and Longqin Hua,*
aDepartment of Pharmaceutical Chemistry, Ernest Mario School of Pharmacy, Rutgers, the State University of New Jersey,
160 Frelinghuysen Road, Piscataway, NJ 08854, USA
bEnact Pharma PLC, Porton Down Science Park, Salisbury, Wiltshire SP4 0JQ, UK
Received 25 April 2003; revised 26 June 2003; accepted 11 July 2003
Abstract—In efforts to obtain potential anticancer prodrugs for gene-directed enzyme prodrug therapy using Eschericia coli
nitroreductase, a series of four benzocyclophosphamide analogues were designed and synthesized incorporating a strategically
placed nitro group in a position para to the benzylic carbon for reductive activation. All four analogues were found to be stable in
phosphate buffer at pH 7.4 and 37 ꢀC and were good substrates of E. coli nitroreductase with half lives between 7 and 24 min at pH
7.0 and 37 ꢀC. However, only two analogues 6a and 6c, both with a benzylic oxygen in the phosphorinane ring para to the nitro
group, showed a modest 33–36-fold enhanced cytotoxicity in E. coli nitroreductase-expressing cells. These results suggest that good
substrate activity and the para benzylic oxygen are required for activation by E. coli nitroreductase. Compounds 6a and 6c repre-
sent a new structure type for reductive activation and a lead for further modification in the development of better analogues with
improved selective toxicity to be used in gene-directed enzyme prodrug therapy.
# 2003 Elsevier Ltd. All rights reserved.
Introduction
malignant tumor cells seem to have very little of this
enzyme. Therefore, it is believed that the detoxication
by aldehyde dehydrogenase might be responsible for its
tumor selectivity as well as drug-resistance in resistant
tumor cells.4 The a,b-unsaturated aldehyde acrolein (4)
is a potent electrophile and the causative agent of the
bladder toxicity associated with cyclophosphamide.5
Cyclophosphamide (1) is an anticancer prodrug, which
has to be activated by cytochrome P-450 enzyme in the
liver.1ꢁ3 As shown in Scheme 1, hepatic cytochrome
P-450 oxidation converts cyclophosphamide to
4-hydroxycyclophosphamide (2). Ultimate conversion
to the cytotoxic alkylating species, phosphoramide
mustard (5), is initiated by ring opening of 2 to produce
aldophosphamide (3). The formation of 5 from 3 pro-
ceeds by general base-catalyzed b-elimination. Enzymes
are not required for conversions following the initial
hydroxylation in the liver.3 The aldehyde moiety in 3
can serve as a substrate for aldehyde dehydrogenase and
the corresponding carboxylic acid product is less prone
to b-elimination. Aldehyde dehydrogenase is widely
distributed in normal human tissues and has been found
in cyclophosphamide-resistant tumor cells. But, most
In the last four decades, modifications of cyclophos-
phamide led to the design and synthesis of many cyclic
and acyclic phosphoramidate alkylating agents.1,6,7
However, these extensive structure–activity relationship
studies failed to produce better drugs than cyclophos-
phamide.
Prodrug design is an important strategy that has been
proven to work for many drugs in improving their
undesirable physico-chemical and biological proper-
ties.8ꢁ10 Recently, prodrug strategies have also been
used in targeted drug delivery including antibody-direc-
ted enzyme prodrug therapy (ADEPT) and gene-direc-
ted enzyme prodrug therapy (GDEPT). In these
approaches, an enzyme is delivered site-specifically by
*Corresponding author. Tel.: +1-732-445-5291; fax: +1-732-445-
6312; e-mail: longhu@rci.rutgers.edu
yCurrent address: Institute of Medicinal Biotechnology, Chinese
Academy of Medical Sciences, Tiantan, Beijing 100050, China.
0968-0896/$ - see front matter # 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/S0968-0896(03)00459-0

4172
Z. Li et al. / Bioorg. Med. Chem. 11 (2003) 4171–4178
Scheme 1. Mechanism of activation of cyclophosphamide (1) in the liver.
chemical conjugation or genetic fusion to a tumor spe-
cific antibody or by an enzyme gene delivery system into
tumor cells. This is then followed by the administration
of a prodrug, which is selectively activated by the deliv-
ered enzyme at the tumor cells. A number of these sys-
tems are in development and have been reviewed.11ꢁ15
Among the enzymes under evaluation is a bacterial
nitroreductase from Escherichia coli. This FMN-con-
taining flavoprotein is capable of reducing certain aro-
matic nitro groups to the corresponding amines or
hydroxylamines in the presence of a cofactor NADH or
NADPH.16ꢁ18
this strategically placed nitro group as the trigger. After
reduction by an enzyme such as the bacterial nitror-
eductase, the resulting hydroxyamines or amines 7a–d
will relay their electrons to the para-position and facil-
itate the cleavage of the benzylic C-O/NH bond, pro-
ducing the cytotoxic intermediates (8a–d). These
intermediates 8a–d resemble the phosphoramide mus-
tard (5) produced in the activation process of cyclo-
phosphamide 1 and could be the ultimate cytotoxic
alkylating agent. In addition, 8a–d also possess addi-
tional electrophilic centers that potentially could form
cross-links with functionally important macromolecules,
providing additional mechanism for cytotoxicity.
To increase tumor selectivity and overcome tumor-
resistance, we hope to move the site of activation from
the liver to tumor tissues through structural modifi-
cation of cyclophosphamide and incorporation of a
trigger-activation mechanism that could be activated by
a reductive enzyme such as E. coli nitroreductase used in
the enzyme prodrug therapies mentioned above. In this
paper, we report the synthesis, stability, enzymatic
reduction, and cellular antiproliferative activity of four
nitrobenzocyclophosphamide analogues 6a–d incorpor-
ating a strategically placed nitro group as the trigger
(Fig. 1).
Synthesis
The dioxaphosphorinane analogue 6a was synthesized
as shown in Scheme 3 starting from 2-methyl-5-nitro-
phenol (10). Acetylation with acetic anhydride followed
by bromination with N-bromosuccinimide afforded 2-
acetoxy-4-nitrobenzyl bromide (11) in 76% yield for the
two steps. Complete hydrolysis of both the ester and the
bromide in 11 using CaCO3 in H2O–dioxane (1:1) gave
2-hydroxy-4-nitrobenzyl alcohol (12) in 82% yield.
Subsequent triethylamine-mediated cyclization with
bis(2-chloroethyl)phosphoramidic dichloride gave the
desired 7-nitro-2-[bis(2-chloroethyl)amino]-1,3,2-benzo-
dioxaphosphorinane-2-oxide (6a) in 55% yield. The over-
all yield for the synthesis of 6a before optimization is 34%.
The benzo[e]cyclophosphamide analogue 6b was syn-
thesized using the Gabriel synthesis of primary amines
by converting the bromide 11 via intermediate 13 to 2-
hydroxy-4-nitrobenzylamine (14) in 32% yield as shown
in Scheme 4. Subsequent triethylamine-mediated cycli-
zation with bis(2-chloroethyl)phosphoramidic dichlor-
ide gave the desired 7-nitro-2-[bis(2-chloroethyl)amino]-
1,3,2-benzoxazaphosphorinane-2-oxide 6b in 62% yield.
The overall yield before optimization for the synthesis
of 6b is 15%.
Fig. 1. Nitrobenzocyclophosphamide analogues designed.
The benzo[e]cyclophosphamide analogue, 7-nitro-2-
[bis(2-chloroethyl)amino]-3,1,2-benzoxazaphosphorinane-
2-oxide (6c), and the diaza analogue, 7-nitro-2-[bis(2-
chloroethyl)amino]-1,3,2-benzodiazaphosphorinane-2-
oxide (6d), were synthesized starting from 2-methyl-5-
nitroaniline (15) using the similar series of reactions
discussed above for the synthesis of 6a and 6b and are
shown in Schemes 5 and 6. The overall yields for the
synthesis of 6c and 6d before optimization were 4.5 and
6.8%, respectively. The overall yields of these syntheses
are limited by formation of the phosphorinane ring sys-
tem. The yields reported in literature for the cyclization
and formation of similar systems varies from 15% to
Results and Discussion
Design and proposed mechanism of activation
Compounds 6a–d are cyclophosphamide analogues with
the cyclophosphamide ring fused with a benzene ring,
where a nitro group is placed in the position para to the
benzylic carbon. The nitro group here serves as an elec-
tronic trigger. It is a strong electron-withdrawing group
and is converted to an electron-donating amino or
hydroxyamino group upon reduction. Scheme 2 illus-
trates the potential mechanism of activation for nitro-
benzocyclophosphamide 6a–d designed incorporating

Z. Li et al. / Bioorg. Med. Chem. 11 (2003) 4171–4178
4173
around 50%.19ꢁ23 Structures of the synthetic inter-
mediates and products were confirmed by IR, NMR,
and mass spectrometry.
the benzylic carbon is attached to a phosphoramide nitro-
gen in the case of 6b and 6d, we were able to isolate the
corresponding reduced aminobenzocyclophosphamides
20b and 20d in 97% and 52% yield, respectively
(Scheme 7). In addition, both 20b and 20d were found
to be similarly stable as compared to their precursors
under the same stability testing conditions used above.
Stability
All four nitrobenzocyclophosphamide analogues were
incubated in pH 7.4 phosphate buffer at 37 ꢀC. HPLC
analysis of the incubation mixtures showed no sig-
nificant changes of the analogues over a period of 4
days (<10%, data not shown), suggesting that the
compounds are very stable under these conditions.
Substrate activity for E. coli nitroreductase
The four nitrobenzocyclophosphamide analogues were
evaluated as substrates of E. coli nitroreductase by
incubation of each compound (0.2mM) in 10 mM
phosphate buffer, pH 7.0 at 37 ꢀC in the presence of 1
mM NADH as the cofactor. The reaction was initiated
by the addition 1.8 mg E. coli nitroreductase. Aliquots
were withdrawn at various time intervals, quenched
with acetonitrile and stored frozen prior to HPLC
analysis. The half lives calculated based on the
disappearance of the substrate are tabulated in Table 1.
All four compounds were found to be substrates of E.
coli nitroreductase with half lives between 7 and 24 min,
though not as good a substrate as CB1954, which has a
half life of 5 min under the same assay conditions.
Chemical reduction of nitrobenzocyclophosphamides
We used catalytic hydrogenation or NaBH4 in the pre-
sence 10% Pd/C in methanol to selectively reduce the
nitro group and then characterized the reduced product
with NMR and high resolution MS.24 In the case of
compounds 6a and 6c, where the benzylic carbon is
attached to an ester oxygen, the reduction gave a com-
plex product mixture, suggesting that the corresponding
reduced products were not stable and may undergo the
cleavage reactions proposed in Scheme 2. However, when
Scheme 2. Proposed mechanism of activation of nitrobenzocyclophosphamides 6a–d by bioreduction.
Scheme 3. Synthesis of the dioxaphosphorinane analogue 6a. Reagents and conditions: (a) Ac2O (10 equiv), pyridine (1.2equiv), 0 ꢀC to rt, 6 h,
91%; (b) NBS/CCl4 (1.0 equiv), hn, rt, 14 h, 83%; (c) CaCO3 (5.2equiv), dioxane–H 2O (1:1), reflux, 3 h, 82%; (d) bis(2-chloroethyl)phosphoramidic
dichloride (1.0 equiv), Et3N (2equiv), EtOAc, rt, 18 h, 55%.
Scheme 4. Synthesis of the benzo[e]cyclophosphamide analogue 6b. Reagents and conditions: (a) potassium phthalimide (1.2equiv), 18-crown-6 (0.1
equiv), rt, 20 h, 52%; (b) NH2NH2 (2.4 equiv), CH2Cl2–CH3OH (1:1), rt, 14 h, 61%; (c) bis(2-chloroethyl)phosphoramidic dichloride (1.0 equiv),
Et3N (2equiv), EtOAc, rt, 14 h, 62%.

4174
Z. Li et al. / Bioorg. Med. Chem. 11 (2003) 4171–4178
CB1954, an excellent substrate of E. coli nitroreductase,
is currently in clinical trials and was used as a control in
our experiments.18,25 It should be noted that compound
6d only reached an end point of 58% while all other
compounds reached end points of less than 10% (Fig. 2).
This behavior of compound 6d in reduction by nitror-
eductase is not understood. Since each of our four
compounds contains a chiral phosphorus center, and
because we used racemic mixtures in our assays, one
possibility is that one enantiomer might not be as good
of a substrate of E. coli nitroreductase as the other. It is
also possible that the nitroreductase enzyme was being
inhibited by the product formed upon reduction by the
enzyme. No loss of substrate was observed when the E.
coli nitroreductase was replaced with 50 mg/mL of the
human enzyme NQO1 and it was concluded that these
compounds are not substrates for this enzyme.
Antiproliferative activity in cell culture
Compounds 6a–d were assayed for their cytotoxicity
against cells expressing either E. coli nitroreductase
(T116) or the human quinone oxidoreductase enzyme
NQO1 (hDT7). Cells were Chinese hamster V79 cells
that had been transfected with a bicistronic vector
encoding for the E. coli nitroreductase or the human
quinone oxidoreductase protein and puromycin resis-
tance protein as the selective marker. F179 cells were
transfected with vector only and were used as the con-
trols. The cells were exposed for 72h to each test com-
pound and the maximum concentration used was 100
mM. With the exception of compound 6b, which has an
IC50 of 61 mM in the control cells, most compounds
were not cytotoxic at 100 mM in the control cells. The
IC50 and the ratios of IC50 (F179/T116) of our test
compounds are tabulated in Table 1. In calculating the
ratio of IC50, the value of 100 mM was used for those
compounds with an undetermined IC50 >100 mM so the
ratio is an underestimate. As shown in Table 1, com-
pounds 6a, 6c, and 6d are not very cytotoxic in their
own right and were not activated by endogenous mam-
malian enzymes (at least not those found in V79 cells).
The results with the E. coli nitroreductase-expressing
cells is much more interesting. Generally, the T116 cells
are more cytotoxically effected by our test compounds
than the control cells. All compounds except 6d tested
show ratios >1 indicating activation by E. coli nitro-
reductase. Compounds 6b and 6d were found to have
similar IC50 values in cells expressing or not expressing
Fig. 2. The disappearance of 6a (*), 6b (*), 6c (&), 6d (&), and
CB1154 (~) during reduction by E. coli nitroreductase as monitored
by HPLC. Substrate (0.2mM) was incubated with 1.8 mg of E. coli
nitrorectase in 10 mM phosphate buffer, pH 7.0 in the presence of
1 mM of NADH at 37 ꢀC in a total volume of 250 mL.
Scheme 5. Synthesis of the benzo[e]cyclophosphamide analogue 6c. Reagents and conditions: (a) Ac2O (10 equiv), pyridine (1.1 equiv), rt, 6 h, 84%;
(b) NBS/CCl4 (1.2equiv), h n, rt, 20 h, 56% based on recovered starting material; (c) CaCO3 (6 equiv), dioxane–H2O (1:1), reflux, 3 h, 42%; (d) 6 N
HCl, rt, 16 h, 100%; (e) bis(2-chloroethyl)phosphoramidic dichloride (1.0 equiv), Et3N (2equiv), EtOAc, rt, 48 h, 23%.
Scheme 6. Synthesis of the diazaphosphorinane analogue 6d. Reagents and conditions: (a) potassium phthalimide (1.5 equiv), 18-crown-6 (0.1
equiv), rt, 24 h, 65%; (b) 6 N HCl, 50 ꢀC, 5 h, 64%; (c) bis(2-chloroethyl)phosphoramidic dichloride (1.0 equiv), Et3N (2equiv), EtOAc, rt, 3 h, 35%.
Scheme 7. Hydrogenation of analogues 6b and 6d. Reagents and conditions: (a) H2, 10% Pd/C, 97% for 20b, 52% for 20d.

Z. Li et al. / Bioorg. Med. Chem. 11 (2003) 4171–4178
4175
E. coli nitroreductase even though both were reduced by
E. coli nitroreductase as shown in our enzyme assays.
Both of these compounds contain a benzylic nitrogen,
instead of a benzylic oxygen, para to the nitro group.
Chemical reduction of 6b and 6d produced stable amine
products that are not expected to be alkylating agents.
On the other hand, 6a and 6c with benzylic oxygen at
the para position to nitro group gave no clearly identifi-
able products upon chemical reduction. 6a and 6c were
found to be over 30-fold more cytotoxic in E. coli
nitroreductase-expressing cells with IC50 values of
around 3 mM. These results indicate that E. coli nitror-
eductase-reduction was an important first step but not
sufficient for enhanced cytotoxicty in E. coli nitror-
eductase-expressing cells. Although we have not deter-
mined experimentally the reactive species responsible
for the enhanced cytotoxicity in E. coli nitroreductase-
expressing cells, we believe that nitroreductase converts
6a and 6c to their corresponding amino or hydro-
xylamine analogue 7a and 7c, which would then follow
the electron ‘push and pull’ mechanism outlined in
Scheme 2 to produce the observed cytotoxicity. It should
be pointed out that the 33–36-fold activation shown by 6a
and 6c in E. coli nitroreductase-expressing cells is about
100 fold less than that shown by CB1954. However, our
benzocyclophosphamides represent a new chemo type for
reductive activation by nitroreductases and further
structure modification of 6a and 6c might lead to the
development of a new generation of cyclophosphamide
analogues that are activated by E. coli nitroreductase and
shown to have enhanced cytotoxicity in E. coli nitror-
eductase-expressing cells. These new analogues might
then be used in conjunction with nitroreductase in
enzyme prodrug therapy in the treatment of cancer.
the benzylic carbon for reductive activation. All four
analogues were found to be stable in phosphate buffer
at pH 7.4 and 37 ꢀC. They are all good substrates of E.
coli nitroreductase with half lives between 7 and 24 min
at pH 7.0 and 37 ꢀC. However, only two compounds 6a
and 6c, both with a benzylic oxygen in the phosphor-
inane ring para to the nitro group, showed a modest 33–
36-fold enhanced cytotoxicity in E. coli nitroreductase-
expressing cells. The other two analogues 6b and 6d,
each with a benzylic nitrogen in the same position,
showed no selective toxicity in cells expressing the E.
coli nitroreductase enzyme. Chemical reduction of 6b
and 6d resulted in the isolation of stable amine pro-
ducts. These results suggest that good substrate activity
is not sufficient and a benzylic oxygen in the phosphor-
inane ring para to the nitro group is required for
reductive activation of nitrobenzocyclophosphamide
analogues by E. coli nitroreductase. Compounds 6a and
6c represent a new structure type for reductive activ-
ation and a new lead for further modification in the
development of better analogues with much improved
selective toxicity to be used in gene-directed enzyme
prodrug therapy.
Experimental
General methods
Solvents were either ACS reagent grade or HPLC grade.
Unless otherwise stated, all reactions were magnetically
stirred and monitored by thin-layer chromatography
(TLC) using 0.25 mm Whatman precoated silica gel
plates. TLC plates were visualized using either 7% (w/
w) ethanolic phosphomolybdic acid or 1% (w/w) aqu-
eous potassium permagnate containing 1% (w/w)
NaHCO3. Flash column chromatography was per-
formed using silica gel (Merck 230–400 mesh). Yields
refer to chromatographically and spectroscopically (1H
NMR) homogeneous materials, unless otherwise noted.
All reagents were purchased at the highest commercial
quality and used without further purification.
In summary, a series of four benzocyclophosphamide
analogues were designed and synthesized incorporating
a strategically placed nitro group in a position para to
Table 1. E. coli nitroreductase (NR)-activation of nitrobenzocyclo-
phosphamides 6a–d
Compd
NR assay
t1/2 (min)a
IC50 (mM)b
Ratioc
(F179/T116)
Infrared spectra were recorded with a Perkin–Elmer
model 1600 series FTIR spectrometer using polystyrene
as an external standard. Infrared absorbance is reported
F179
hDT7
T116
2.7
6a
6b
24
11
>100
61
>100
48
>36
1.3
1
in reciprocal centimeters (cmꢁ1). All H and 13C NMR
48
6c
6d
CB1954
13
>100
>100
>100
>100
>100
1.7
3.0
>100
0.036
>33
ꢂ1
spectra were recorded on a Varian Gemini 300 MHz
spectrometer at ambient temperature and calibrated
using residual undeuterated solvents as the internal
reference. Chemical shifts (300 MHz for 1H and 75 MHz
for 13C) are reported in parts per million (d) relative to
7.8d
5.0
>2777
aHalf lives of reduction by E. coli nitroreductase were determined
using 0.2mM of substrate in 10 mM phosphate buffer (pH 7.0) in the
presence of 1 mM of NADH at 37 ꢀC in a total volume of 250 mL. The
reaction was initiated by the addition of 1.8 mg of E. coli nitror-
eductase. Aliquots were withdrawn and analyzed by HPLC.
bIC50 values are the concentration required to reduce cell number to
50% of control. Cytotoxicity was assayed against cells expressing
either E. coli nitroreductase (T116) or human quinone oxidoreductase
NQO1 (hDT7). Cells were Chinese hamster V79 cells that had been
transfected with a bicistronic vector encoding for the E. coli nitro-
reductase or the human quinone oxidoreductase protein and pur-
omycin resistance protein as the selective marker. F179 cells were
transfected with vector only and were used as the controls.
cRatio of IC50 values (F179/T116) as an indication of activation by E.
coli nitroreductase.
1
CD3OD (d 3.3 for H and 49.0 for 13C). Coupling con-
stants (J values) are given in hertz (Hz). The following
abbreviations were used to explain the multiplicities:
s=singlet; d=doublet; t=triplet; q=quartet; p=quin-
tet; m=multiplet; br=broad. Mass spectral data were
obtained from the University of Kansas Mass Spectro-
metry Laboratory (Lawrence, KS).
Acetic acid, 2-bromomethyl-5-nitrophenyl ester (11). 2-
Methyl-5-nitrophenol 10 (2.5 g, 13 mmol) was dissolved
in 50 mL of acetic anhydride (10 eq) and immersed in an
dThe catalysis seemed to reach an end point of 58%.

4176
Z. Li et al. / Bioorg. Med. Chem. 11 (2003) 4171–4178
ice water bath. After the addition of pyridine (2mL, 1.2
equiv), the reaction mixture was stirred at room tem-
perature for 6 h. Excess acetic anhydride was removed
under reduced pressure and the residue was dissolved in
100 mL of CH2Cl2, washed with satd NaHCO3, water,
dried over Na2SO4. 2-Methyl-5-nitrophenyl acetate was
obtained as a white solid (2.9 g, 91%). Mp 68–72 ꢀC, 1H
NMR (300 MHz, CDCl3) d 8.04 (d, 1H, J=8.4 Hz),
7.93 (s, 1H), 7.40 (d, 1H, J=8.4 Hz), 2.37 (s, 3H), 2.29
(s, 3H); MS (FAB+) m/z (relative intensity) 196 (MH+,
12.9), 195 (50.8), 152 (54.1), 135 (70.5), 119 (100).
toluene and mixed with potassium phthalimide (2.63 g,
1.2equiv) and 18-crown-6 (375 mg, 0.1 equiv). The
suspension was stirred at room temperature for 20 h.
The reaction mixture was then diluted with 50 mL of
water and extracted with methylene dichloride. The
CH2Cl2 extract was washed with 5% citric acid, satu-
rated NaHCO3, and H2O. After drying over anhydrous
Na2SO4 and removal of solvent, the residue was purified
through flash column chromatography to giꢀve the
1
desired product 13 (2.5 g. 52%). Mp 175–178 C, H
NMR (300 MHz, CDCl3) d 8.17–7.73 (m, 7H,), 4.89 (s,
2H), 2.47 (s, 3H). MS (FAB+) m/z (relative intensive)
341 (MH+, 5), 299 (7), 195 (33), 152 (39), 135 (100).
HRMS (FAB+) m/z calcd for C17H13N2O6: 341.0773,
found: 341.0773.
2-Methyl-5-nitrophenyl acetate (2.9 g, 14.9 mmol) and
N-bromosuccinimide (2.65 g, 14.9 mmol) were sus-
pended in 50 mL of carbon tetrachloride, and photo-
lyzed with a 300 watt lamp under N2 for 14 h. The
reaction mixture was then diluted with 50 mL of meth-
ylene chloride, washed with water and brine, dried over
anhydrous Na2SO4. The residue after removal of sol-
vents was purified through flash column chromato-
graphy to afford the desired product 11 (3.27 g, 83%).
2-Hydroxy-4-nitrobenzylamine (14). To a solution of
compound 13 (2.5 g, 7.35 mmol) in 50 mL of 1:1 mix-
ture of CH2Cl2 and CH3OH was added 2.4 equiv of
hydrazine. The reaction mixture was stirred at room
temperature for 14 h. After removal of solvent under
reduced pressure, the residue was treated with 6 N HCl
(50 mL) and stirred at room temperature for 1 h. The
filtrate was neutralized to pH=7 with aqueous NaOH
solution and extracted with EtOAc. The combined
EtOAc extract was dried over anhydrous Na2SO4 and
concentrated to dryness to afford the desired product 14
Mp 76.5–78 ꢀC. H NMR (300 MHz, CDCl3) d 8.09–
1
8.04 (m, 2H), 7.60 (d, 1H, J=8.4 Hz), 4.44(s, 2H), 2.43
(s, 3H). MS (FAB+) m/z (relative intensity) 196 (MH+-
Br, 7.9), 195 (82.3).
2-Hydroxy-4-Nitrobenzyl alcohol (12). Compound 11
(200 mg, 0.7 mmol) dissolved in 2 mL of dioxane, was
mixed with 5.2equiv of CaCO 3 in 2mL of H 2O and the
reaction mixture was heated to reflux for 3 h. After the
disappearance of starting material as shown by TLC,
dioxane was removed by evaporation and the residue
was treated with 5 mL of 2N HCl and extracted with 30
mL of EtOAc. The combined extract was washed with
brine (3ꢃ30 mL) and dried over anhydrous Na2SO4.
Final separation through flash column chromatography
afforded the desired product 12 (101 mg, 81.9%). Mp
145–149 ꢀC. 1H NMR (300 MHz, CDCl3), d 7.70 (s,
1H), 7.69 (d, 1H, J=9.0 Hz), 7.34 (d, 1H, J=9.0 Hz),
4.81 (s, 2H), 4.53 (s, 1H), 2.20 (s, 1H). MS (EI) m/z
(relative intensity) 169 (41.6, M+), 151 (100), 105 (54.4),
77 (78.4).
(0.752 g, 60.6%). Mp 210–215 ꢀC. H NMR (300 MHz,
1
CD3OD) d 7.45 (s, 1H), 7.42(d, 1H, J=8.1 Hz), 7.26 (d,
1H, J=8.1 Hz), 4.00 (s, 2H). 1H NMR (300 MHz,
DMSO) d 7.55 (d, 1H, J=8.4 Hz), 7.43 (s, 1H), 7.36 (d,
1H, J=8.4 Hz), 3.91 (s, 2H). MS (FAB+) m/z (relative
intensity) 169 (MH+, 7), 154 (100), 136 (69). HRMS
(FAB+) m/z calcd for C7H9N2O3: 169.0613, found:
169.0613.
7-Nitro-2-[bis(2-chloroethyl)amino]-1,3,2-benzoxazapho-
sphorinane-2-oxide (6b). To a solution of compound 14
(752mg, 4.47 mmol) and 2.0 equiv of Et 3N in 2 0 mL of
EtOAc was added dropwise with stirring a solution of
1.0 equiv of bis(2-chloroethyl)phosphoramidic dichlor-
ide (1.16 g, 4.47 mmol) in 20 mL of EtOAc. After strir-
ring was continued for 14 h, the precipitate was
removed by suction filtration and the filtrate was con-
centrated under reduced pressure. The residue was pur-
ified by flash column chromatography to afford the
desired product 6b (974 mg, 61.9%). Mp 123–126 ꢀC.
1H NMR (300 MHz, CDCl3) d 7.97 (d, 1H, J=8.1 Hz),
7.90 (s, 1H), 7.29 (d, 1H, J=8.1 Hz), 4.61–4.31 (m, 2H),
3.80 (s, 1H), 3.72–3.59 (m, 4H), 3.57–3.47 (m, 4H). IR
7-Nitro-2-[bis(2-chloroethyl)amino]-1,3,2-benzodioxapho-
sphorinane-2-oxide (6a). Compound 12 (100 mg, 0.59
mmol) was dissolved in 1 mL of EtOAc and mixed with
2.0 equiv of Et3N and a solution of bis(2-chloro-
ethyl)phosphamidic dichloride (153 mg, 1.0 equiv) in 1
mL of EtOAc. The mixture was stirred at room tem-
perature for 18 h. After removal of the precipitate
through filtration, the filtrate was purified by flash col-
umn chromatography to give 1,3-dioxa analogue 6a as
(neat) 3100, 1480, 1304, 1175, 1045, 925 and 804 cmꢁ1
;
MS (FAB+) m/z (relative intensity) 354 (MH+, 3.3),
309 (6.5), 195 (28), 152 (68), 135 (90), 119 (100). HRMS
(FAB+) m/z calcd for C11H15N3O4Cl2P: 354.0177,
found: 354.0181.
1
an yellow oil (114.7 mg, 54.6%). H NMR (300 MHz,
CDCl3) d 8.02(d, 1H, J=8.4 Hz), 7.92(s, 1H). 7.32(d,
1H, J=8.4 Hz), 5.71–5.24 (m, 2H), 3.67 (t, 4H, J=6.6
Hz), 3.55–3.46 (m, 4H); IR (neat) 2960–2820, 1520,
1420, 1340, 1260, 970, 840 and 726 cmꢁ1; MS (FAB+)
m/z (relative intensity) 355(MH+, 12.6), 307(16.2),
289(8.9), 154(100); HRMS (FAB+) m/z calcd for
C11H14Cl2N2O5P: 355.0017, found: 354.9992.
2-Acetamido-4-nitrobenzyl bromide (16). To a solution
of 2-methyl-5-nitroaniline 15 (3.04 g, 2mmol) in 50 mL
of CHCl3 were added Ac2O (10 equiv) and pyridine
(1.78 mL, 1.1 equiv). The reaction mixture was stirred at
room temperature overnight. After concentration under
reduced pressure, the residue was dissolved in 100 mL of
CH2Cl2, washed with water, satd NaHCO3 and water,
2-Acetoxy-4-nitro-ꢀ-phthalimido toluene (13). Com-
pound 11 (3.9 g, 14.2mmol) was dissolved in 50 mL of

Z. Li et al. / Bioorg. Med. Chem. 11 (2003) 4171–4178
4177
and dried over anhydrous Na2SO4. After removal of
solvent, the residue was triturated with CCl4 to give the
desired product 2-acetamido-4-nitrotoluene as a solid
(3.36 g, 84%). Mp 154–155 ꢀC. 1H NMR (300 MHz,
CDCl3) d 8.76 (s, 1H), 7.94 (d, 1H, J=8.1 Hz), 7.34 (d,
1H, J=8.1 Hz), 7.09 (br, 1H), 2.37 (s, 3H), 2.26 (s, 3H).
2H), 3.69–3.62 (m, 4H), 3.48–3.39 (m, 4H). IR (neat)
3600–3000 (broad), 2930, 2860, 1600, 1520, 1450, 1340,
1220, 970, 880, 820, and 735 cmꢁ1. MS (FAB+) m/z
(relative intensity) 354 (MH+, 4.9), 307 (20.0), 289
(12.6), 154 (100), 136 (98.8). HRMS (FAB+) m/z calcd
for C11H15Cl2N3O4P: 354.0177, found: 354.0162.
2-Acetamido-4-nitrotoluene (1.0 g, 3.66 mmol) and N-
bromosuccinimide (0.78 g, 1.2equiv) were suspended in
100 mL of CCl4 and photolized with a 300 watt lamp
under N2 for 20 h. After removal of solvent under
reduced pressure, the residue was subjected to flash col-
umn chromatography to afford the desired product 16
(0.46 g, 55.6% after recovery of 0.2g of starting mate-
2-Acetamido-4-nitro-ꢀ-phthalimido toluene (18). A solu-
tion of compound 16 (45.9 mg, 0.168 mmol) in 2mL of
THF was mixed with 1.5 equiv of potassium phthali-
mide (146.6 mg) and a catalytic amount of 18-Crown-6
(4.4 mg, 0.1 equiv). The reaction mixture was stirred at
room temperature for 24 h. After removal of solvent,
the residue was taken up in 20 mL of CH2Cl2, washed
with 5% citric acid, satd NaHCO3, and water, and dried
over Na2SO4. Purification through flash column chro-
matography afforded the desired product 18 (37.2mg,
73.3% after recovery of 5 mg of starting material). Mp
rial). Mp 187.5–189 ꢀC. H NMR (300 MHz, CDCl3) d
1
8.84 (s, 1H), 8.00 (d, 1H, J=8.4 Hz), 7.54 (br, 1H), 7.50
(d, 1H, J=8.4 Hz), 4.52 (s, 2H), 2.32 (s, 3H); MS
(FAB+) m/z (relative intensity) 273 (MH+, 5.6), 195
(25.7), 153 (33.1), 135 (100).
ꢀ
1
221.3–224 C. H NMR (300 MHz, CDCl3) d 8.97 (s,
1H), 7.96–7.76 (m, 6H), 4.88 (s, 2H), 2.39 (s, 3H). MS
(FAB+) m/z (relative intensity) 340 (MH+, 6.2), 307
(16.9), 289 (9.9), 273 (4.0), 154 (100), 136 (67.2).
2-Amino-4-nitrobenzyl alcohol (17). Compound 16 (163
mg, 0.6 mmol) dissolved in 2mL dioxane was mixed
with a suspension of CaCO3 (358.5 mg, 3.6 mmol) in 2
mL of water. The mixture was then heated up to reflux
for 3 h until all starting material disappeared as mon-
itored by TLC. After removal of solvent under reduced
pressure, the residue was treated with 2mL of 2N HCl
and extracted with CH2Cl2. The organic extract was
dried over Na2SO4 and subjected to flash column chro-
matography to give 2-acetamido-4-nitrobenzyl alcohol
2-Amino-4-nitrobenzylamine (19). Compound 18 (50 mg,
0.15 mmol) was suspended in 2mL of 6 N HCl and
stirred at 50 ꢀC for 5 h. After filtration to remove
the solid, the filtrate was neutralized to pH 10 and
extracted with EtOAc. The EtOAc extract was dried
over anhydrous Na2SO4. Removal of EtOAc afforded
1
the desired compound 19 (15.7 mg, 63.8%). H NMR
1
(53.2 mg, 42.2%). H NMR (300 MHz, CDCl3) d 9.01
(300 MHz, CDCl3) d 7.51 (dd, 1H, J1=2.4 Hz, J2=8.1
Hz), 7.49 (d, 1H, J=2.4 Hz), 7.15 (d, 1H, J=8.1 Hz),
3.97 (s, 2H).
(d, 1H, J=2.1 Hz), 8.87 (br, 1H), 7.91 (dd, 1H, J1=2.1
Hz, J2=8.1 Hz), 7.32(d, 1H, J=8.1 Hz), 4.82(d, 2H,
J=5.7 Hz), 2.53 (t, 1H, J=5.7 Hz), 2.24 (s, 3H). MS
(FAB+) m/z (relative intensity) 211 (MH+, 7.5), 195
(34.0), 152(42.0), 135 (100).
7-Nitro-2-[bis(2-chloroethyl)amino]-1,3,2-benzodiazaphos-
phorinane-2-oxide (6d). To a solution of 19 (358 mg,
2.14 mmol) in 8 mL of EtOAc were added with stirring
Et3N (433 mg, 4.28 mmol) and bis(2-chloroethyl)-phos-
phoramidic dichloride (554 mg, 2.14 mmol) in 2 mL of
EtOAc. After the reaction mixture was stirred for an
additional 3 h, the precipitate was removed by suction
filtration and the filtrate was concentrated under
reduced pressure. The residue was purified through flash
column chromatography to give the desired product 6d
as a yellow solid (263 mg, 34.6%). Mp 168–169.5 ꢀC. 1H
NMR (300 MHz, CDCl3) d 7.74 (dd, 1H, J1=2.4 Hz,
J2=8.4 Hz), 7.65 (d, 1H, J=2.4 Hz), 7.16 (d, 1H, J=8.4
Hz), 6.23 (br s, 1H), 4.46–4.12 (m, 2H), 3.66 (t, 4H,
J=5.7 Hz), 3.48–3.37 (m, 4H), 3.24 (br s, 1H). MS
(FAB+) m/z (relative intensity) 324 (MH+, 4.2), 307
(17.9), 289 (10.4), 273 (4.6), 154 (100), 147(58.2), 136
(68.7). HRMS(FAB+) m/z calcd for C11H17Cl2N3O2P:
324.0435, found: 324.0435.
2-Acetamido-4-nitrobenzyl alcohol (53.2 mg, 0.316
mmol) was treated with 1 mL of 6 N HCl and the reac-
tion mixture was stirred at room temperature overnight.
After neutralization with 6 N aqueous NaOH solution
to pH 10, the reaction mixture was extracted with
EtOAc, dried over Na2SO4, purified through flash col-
umn chromatography to give desired product 17 (46
mg, 100%). Mp 178–180 ꢀC. 1H NMR (300 MHz,
CDCl3) d 7.56–7.51 (m, 2H, aromatic), 7.20 (d, 1H,
J=8.1 Hz, aromatic), 4.74 (d, 2H, J=4.5 Hz), 4.52(br
s, 2H), 1.72 (t, 1H, J=4.5 Hz). MS (EI) m/z (relative
intensity) 168(M+, 100), 150(60.8).
7-Nitro-2-[bis(2-chloroethyl)amino]-3,1,2-benzoxazapho-
sphorinane-2-oxide (6c). To a solution of 17 (46 mg,
0.27 mmol) in 0.5 mL of EtOAc were added with stir-
ring Et3N (54.6 mg, 0.54 mmol) and bis(2-chloro-
ethyl)phosphoramidic dichloride (70.8 mg, 0.27 mmol)
in 0.5 mL EtOAc. After 48 h, the precipitate was
removed by suction filtration and the filtrate was con-
centrated under reduced pressure. The residue was pur-
ified through flash column chromatography to give the
desired product 6c as a yellow solid (21.6 mg, 22.5%).
Mp 138–142 ꢀC. 1H NMR (300 MHz, CDCl3) d 7.78
(dd, 1H, J1=2.4 Hz, J2=8.1 Hz), 7.69 (d, 1H, J=2.4
Hz), 7.22 (d, 1H, J=8.1 Hz), 6.57 (d, 1H), 5.56–5.07 (m,
Stability test of benzocyclophosphamides (6a, 6b, 6c, 6d)
in aqueous buffer. A 2mg sample of a benzocyclophos-
phamide (6a, 6b, 6c or 6d) was dissolved in 2mL of 50
mM sodium phosphate buffer (pH=7.40) containing
10% DMSO and incubated at 37 ꢀC. At different time
intervals, aliquots were withdrawn and subjected to
reversed phase HPLC analysis (C18 analytical column,
gradient elution from 5–80% acetonitrile containing
0.1% TFA at a flow rate of 1 mL/min).

4178
Z. Li et al. / Bioorg. Med. Chem. 11 (2003) 4171–4178
Enzyme assays
cells incubated with medium only and evaluated by
interpolation from the dose response curve.
Substrate (0.2mM) was incubated with 1 mM of
NADH at 37 ꢀC in 10 mM phosphate buffer (pH 7.0) in
a total volume of 250 mL. The reaction was initiated by
the addition of 1.8 mg of E. coli nitroreductase. Aliquots
were withdrawn and analyzed by HPLC. The half life of
reduction by E. coli nitroreductase was calculated based
on the disappearance of the substrate.
Acknowledgements
We gratefully acknowledge the financial support of
grant SNJ-CCR 700-009 from the State of New Jersey
Commission on Cancer Research. We thank Dr. Roger
Melton (Enact Pharma Plc, Salisbury, UK) for provid-
ing the E. coli nitroreductase enzyme and Dr. Steve
Hobbs (Institute of Cancer Research, Sutton, Surrey,
UK) for providing the vector constructs.
Plasmid vector construction
Bicistronic eukaryotic expression vectors containing the
coding regions for either human NQO1 or E. coli
nitroreductase together with puromycin acetyl transfer-
ase (conferring puromycin resistance) driven from a
single CMV promoter was constructed by cloning into
the XhoI site of the vector pIRES-P (EMBL:Z75185)
using conventional techniques. Insert orientation and
identity were confirmed by diagnostic restriction
digests and dideoxy sequencing using a Sequenase II
kit (Amersham Pharmacia Biotech, St Albans, Herts,
UK).
References and Notes
1. Zon, G. Prog. Med. Chem. 1982, 19, 205.
2. Stec, W. J. J. Organophosphorous Chem. 1982, 13, 145.
3. Borch, R. F.; Millard, J. A. J. Med. Chem. 1987, 30, 427.
4. Hilton, J. Cancer Res. 1984, 44, 5156.
5. Cox, P. J. Biochem. Pharmacol. 1979, 28, 2045.
6. Wright, J. E. In Cancer Therapeutics: Experimental and
Clinical Agents, Teicher, B. A., Ed.; Humana Press Inc:
Totowa, NJ, 1997; pp 23–79.
7. Arnold, H.; Bourseaux, F.; Brock, N. Nature 1958, 181, 931.
8. Pochopin, N. L.; Charman, W. N.; Stella, V. J. Int. J.
Pharmaceut. 1995, 121, 157.
9. Oliyai, R.; Stella, V. J. Annu. Rev. Pharmacol. Toxicol.
1993, 32, 521.
10. Bundgaard, H. Design of Prodrugs; Elsevier: Amsterdam,
1985.
11. McNeish, I. A.; Searle, P. F.; Young, L. S.; Kerr, D. J.
Adv. Drug Deliv. Rev. 1997, 26, 173.
12. Niculescu-Duvaz, I.; Springer, C. J. Adv. Drug Deliv. Rev.
1997, 26, 151.
13. Senter, P. D.; Svensson, H. P. Adv. Drug Deliv. Rev. 1996,
22, 341.
14. Sherwood, R. F. Adv. Drug Deliv. Rev. 1996, 22, 269.
15. Springer, C. J.; Niculescu-Duvaz, I. Adv. Drug Deliv. Rev.
1996, 22, 351.
16. Bridgewater, J.; Minton, N. P.; Michael, N. P.; Knox,
R.; Springer, C.; Collins, M. Eur. J. Cancer 1995, 31,
2362.
17. Anlezark, G. M.; Melton, R. G.; Sherwood, R. F.; Coles,
B.; Friedlos, F.; Knox, R. J. Biochem. Pharmacol. 1992, 44,
2289.
18. Knox, R. J.; Friedlos, F.; Sherwood, R. F.; Melton, R. G.;
Anlezark, G. M. Biochem. Pharmacol. 1992, 44, 2297.
19. Ludeman, S. M.; Zon, G. J. Med. Chem. 1975, 18, 1251.
20. Takamizawa, A.; Matsumoto, S. Chem. Pharm. Bull.
1978, 26, 790.
21. Shih, Y.-E.; Wang, J.-S.; Chen, C.-T. Hetercycles 1978, 9,
1277.
22. Borch, R. F.; Canute, G. W. J. Med. Chem. 1991, 34,
3044.
23. Viljanen, T.; Tahtinen, P.; Pihlaja, K.; Fulop, F. J. Org.
Chem. 1998, 63, 618.
24. Hu, L.; Liu, B.; Hacking, D. R. Bioorg. Med. Chem. Lett.
2000, 10, 797.
25. Melton, R. G., Knox, R. J., Eds. Enzyme-prodrug Strate-
gies for Cancer Therapy. Kluwer Academic/Plenum: New
York, 1999.
26. Skehan, P.; Storeng, R.; Scudiero, D.; Monks, A.;
McMahon, J.; Warren, J. T.; Bokesch, H.; Kenney, S.; Boyd,
M. R. J. Natl. Cancer Inst. 1990, 82, 1107.
Cell culture and transfection
V79 Chinese hamster lung fibroblasts were grown in
monolayer culture in Dulbecco’s modified Eagle med-
ium (DMEM) containing 10% fetal calf serum and 4
mM glutamine (all from GibcoBRL, Life Technologies
Ltd, Paisley, Scotland, UK). Cells were maintained in a
humidified atmosphere at 37 ꢀC with 5% CO2 and sub-
cultured twice weekly by trypsinization. All cells were
determined to be free of mycoplasma. Plasmid vectors
were transfected into cells by calcium phosphate co-
precipitation (Profection, Promega, Southampton,
Hampshire, UK) and positive clones were selected in
growth medium containing 10mg/mL puromycin and
maintained under selective pressure. Individual clones
were screened for either NQO1 or NR activity and a
suitable nitroreductase (designated T116) or NQO1
(designated hDT7) expressing clone selected for
further use. Transfection of the empty vector supplied
a suitable puromycin-resistant control cell line (desig-
nated F179).
Growth inhibition was measured by the sulforhodamine
B method.26 Cells in exponential phase of growth were
trypsinized, seeded in 96-well plates at a density of 500
cells per well (100 mL) and permitted to recover for 24 h.
For testing, the compounds were dissolved in dimethyl
sulfoxide to give 100 mM stock solutions. These were
diluted into medium to 300 mM, that were then serially
diluted in situ (8- of 3-fold) giving final concentrations
of 100–0.046 mM. Cells were then incubated with the
drug for 3 days at 37 ꢀC. The plates were fixed and
stained with sulforhodamine-B (SRB), before reading
optical absorption at 570 nm; results were expressed as
percentage of control growth. The cytotoxicity of each
compound was expressed as that concentration produ-
cing 50% inhibition of cell growth (IC50) compared with