Hypoxia-selective antitumor agents. 16. Nitroarylmethyl quaternary salts as bioreductive prodrugs of the alkylating agent mechlorethamine

Article abstract of DOI:10.1021/jm010202l

Nitrobenzyl quaternary salts of nitrogen mustards have been previously reported as hypoxia-selective cytotoxins. In this paper we describe the synthesis and evaluation of a series of heterocyclic analogues, including pyrrole, imidazole, thiophene, and pyrazole examples, chosen to cover a range of one-electron reduction potentials (from -277 to -511 mV) and substitution patterns. All quaternary salt compounds were less toxic in vitro than mechlorethamine, and all were more toxic under hypoxic than aerobic conditions, although the differentials were highly variable within the series. The most promising analogue, imidazole 2, demonstrated DNA crosslinking selectively in hypoxic RIF-1 cells, and was active in vivo in combination with radiation or cisplatin. However, 2 also produced unpredictable toxicity in vivo, suggestive of nonspecific nitrogen mustard release, and this has restricted further development of these compounds as hypoxia-selective cytotoxins.

Full text of DOI:10.1021/jm010202l

J . Med. Chem. 2001, 44, 3511-3522
3511
Hyp oxia -Selective An titu m or Agen ts. 16. Nitr oa r ylm eth yl Qu a ter n a r y Sa lts a s
Bior ed u ctive P r od r u gs of th e Alk yla tin g Agen t Mech lor eth a m in e
Moana Tercel,* Alan E. Lee, Alison Hogg, Robert F. Anderson, Ho H. Lee, Bronwyn G. Siim,
William A. Denny, and William R. Wilson*
Auckland Cancer Society Research Centre, Faculty of Medical and Health Sciences, University of Auckland,
Private Bag 92019, Auckland, New Zealand
Received May 7, 2001
Nitrobenzyl quaternary salts of nitrogen mustards have been previously reported as hypoxia-
selective cytotoxins. In this paper we describe the synthesis and evaluation of a series of
heterocyclic analogues, including pyrrole, imidazole, thiophene, and pyrazole examples, chosen
to cover a range of one-electron reduction potentials (from -277 to -511 mV) and substitution
patterns. All quaternary salt compounds were less toxic in vitro than mechlorethamine, and
all were more toxic under hypoxic than aerobic conditions, although the differentials were highly
variable within the series. The most promising analogue, imidazole 2, demonstrated DNA cross-
linking selectively in hypoxic RIF-1 cells, and was active in vivo in combination with radiation
or cisplatin. However, 2 also produced unpredictable toxicity in vivo, suggestive of nonspecific
nitrogen mustard release, and this has restricted further development of these compounds as
hypoxia-selective cytotoxins.
In tr od u ction
Hypoxia-selective cytotoxins are designed to be pref-
erentially activated to toxic species in the low-oxygen
microenvironments present in many solid tumors.1 This
hypoxic cell subpopulation has been directly linked to
tumor resistance to radiotherapy,2 and these slower
cycling cells at a greater distance from blood vessels are
also considered more likely than well-oxygenated cells
to be resistant to various forms of chemotherapy.
Further, there is increasing evidence that hypoxia is
involved in several aspects of tumor progression, as a
consequence of upregulation of the expression of angio-
genic factors,3 selection for apoptosis-resistant cells,4
and stimulation of metastasis.5 Hence, a successful
hypoxia-selective cytotoxin should offer a considerable
therapeutic benefit if combined with other treatments
that eliminate the major aerobic portions of a tumor.
um chloride (1) combines a number of attractive fea-
Several classes of compounds have been considered
tures: very strong deactivation of the mustard and high
for development as hypoxia-selective cytotoxins, such
water solubility by virtue of the permanent positive
as the quinone porfiromycin (under clinical evaluation
charge, and a reduction potential (-358 mV) within the
range expected to be suitable for metabolic activation.
1 was found to display extremely high hypoxic selectivi-
in combination with radiotherapy6) and the nitroimid-
azole CI-1010 (highly effective against hypoxic cells in
murine tumors,7 but withdrawn from clinical evaluation
ties in vitro (up to several thousand fold), with the
due to retinal toxicity8). Two very promising and more
mechanism of action being established as reductive
recently investigated analogues are the amine N-oxides
fragmentation to release mechlorethamine.14 Release of
AQ4N9 and tirapazamine,10 the latter of which has
a diffusible cytotoxin was also inferred from the much
progressed to phase III evaluation and appears particu-
greater cytotoxicity of 1 against intact versus dissociated
larly effective in combination with cisplatin11 or as an
EMT6 spheroids.14 A number of benzyl-substituted,
adjunct to cisplatin/radiotherapy treatment.12
naphthyl, and mustard analogues of 1 were prepared
We have previously reported a further class of com-
and investigated.15 Although the various substituents
pounds that have potential as hypoxia-selective cyto-
affected the reduction potential in a reasonably predict-
toxins, namely, nitrobenzyl quaternary salts of nitrogen
able manner, and there was general evidence for DNA
mustards.13 The prototype member of this class, N,N-
alkylation as the common cytotoxic event, the observed
bis(2-chloroethyl)-N-methyl-N-(2-nitrobenzyl)ammoni-
hypoxic selectivities were highly variable, ranging from
no selectivity up to several thousand fold, i.e., up to the
levels observed for the parent compound 1. However,
* To whom correspondence should be addressed. Phone: 649 3737
599. Fax: 649 3737 502. E-mail: m.tercel@auckland.ac.nz, wr.wilson@
auckland.ac.nz.
when investigated in vivo in combination with radiation,
10.1021/jm010202l CCC: $20.00 © 2001 American Chemical Society
Published on Web 09/13/2001

3512 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 21
Tercel et al.
Ta ble 1. One-Electron Reduction Potentials, E(1), Determined by Pulse Radiolysis, and in Vitro Cytotoxicity of Nitroarylmethyl
Quaternary Nitrogen Mustards under Aerobic and Hypoxic Conditions
IC50 (µM), aerobic 4 h exposurea
C10 (µM), 3 h exposure, EMT6b
compd
E(1) (mV)
AA8
UV4
EMT6
Skov3
air
N2
HCRc
4
5
6
1
7
2
8
9
-277 ( 10
-287 ( 8
-344 ( 8
-358 ( 10
-361 (10
-397 ( 8
-433 ( 8
-500 ( 8
-511 ( 10
47 ( 18
4260 ( 760
32 ( 7
0.19 ( 0.02
6.5 ( 2.6
2.7 ( 0.8
200 ( 120
3.0 ( 0.4
300 ( 70
5.1 ( 1.0
14 ( 2
29 ( 11
2400 ( 800
32 ( 4
22 ( 5
3450 ( 450
30 ( 2
1.3 ( 0.2
2.2 ( 0.4
4.5 ( 2.5
66 ( 30
6 ( 2
16 ( 2
2500 ( 500
14 ( 9
0.36 ( 0.08
61 ( 14
24600 ( 4400
140 ( 90
10000 ( 3600
43 ( 8
12300 ( 540
30 ( 19
400 ( 290
6 ( 4
0.38 ( 0.12
1.9 ( 0.2
150 ( 90
330 ( 60
550 ( 170
1000 ( 400
2.5 ( 0.3
140 ( 25
250 ( 130
215 ( 30
120 ( 40
1.09 ( 0.04
57 ( 22
85 ( 20
245 ( 40
320 ( 140
1.0 ( 0.1
1.2 ( 0.1
3.5 ( 1.0
18 ( 10
27 ( 9
46 ( 15
17 ( 7
1.4 ( 0.2
15 ( 6
1.6 ( 0.1
17 ( 2
9.9 ( 1.3
0.15 ( 0.02
16 ( 6
3
1.8 ( 0.6
12 ( 5
HN2d
0.029 ( 0.004
0.85 ( 0.07
1.17 ( 0.02
a
b
Drug concentration to reduce cell density to 50% of that of the controls. Values are mean ( SEM for 2-7 experiments. Drug
concentration to reduce surviving fraction to 10% of that of the controls. Values are mean ( SEM for 2-3 experiments. c Hypoxic cytotoxicity
ratio ) C10(air)/C10(N2). Mechlorethamine.
d
Sch em e 1. Reductive Fragmentation of Compound 2
1 was found to exhibit only marginal activity against
hypoxic cells in the KHT tumor.15
The detailed mechanism of the reductive activation
step was initially assumed to involve fragmentation at
the one-electron (nitro radical anion) stage, as previ-
ously documented for nitrobenzyl halides.16 However,
investigation by steady-state17 and pulse18 radiolysis has
conclusively shown that multielectron reduction of 1 is
required before fragmentation occurs, and is accompa-
nied by the production of many aromatic byproducts in
low yield. This is true for all nitrobenzyl and nitronaph-
thyl derivatives investigated, but is in sharp contrast
to the results obtained with two nitroheterocyclic
Given the distinctive difference between the reductive
chemistry of 2 and 3 compared to previous benzyl
examples, and the much higher yielding release of
analoguessthe nitroimidazole 2 and the nitropyrrole
3.17 For these compounds clean one-electron reductive
mechlorethamine from these compounds, we have un-
dertaken a detailed study of their hypoxia-selective
activity. We also report the synthesis and activity of a
series of further heterocyclic analogues with a variety
of substitution patterns and reduction potentials. This
study seeks to optimize the nitroheterocyclic system in
relation to masking of mustard toxicity in the prodrug
form, enzymatic activation by hypoxic cells in tissue
culture, and activity against hypoxic cells in tumors.
Resu lts a n d Discu ssion
Ch em istr y. The reference compounds 1-3 and newly
synthesized quaternary salts are presented in Table 1.
Heterocyclic systems investigated include imidazole,
fragmentation occurs, with high-yielding release of
mechlorethamine, along with the product derived from
arylmethyl radical dimerization (Scheme 1). As previ-
ously noted, 2 and 3 also display significant hypoxic
selectivity in vitro, although not to the same extreme
levels as observed for 1.19 The observation that these
compounds fragment efficiently on radiolytic reduction
also suggests the possibility, ultimately, of activating
nitroarylmethyl quaternary salts under hypoxic condi-
tions using radiotherapy rather than enzymes to effect
reduction, although this would probably require release
of more potent cytotoxic amines than mechlorethamine
to be therapeutically useful.19
pyrrole, thiophene, and pyrazole, in which the nitro
group and methyl quaternary substituent are in conju-
gated positions. The general synthetic plan follows from
the route previously described,15 i.e., displacement of the
leaving group from a nitro(halomethyl)heterocycle with
N-methyldiethanolamine, and conversion of the result-
ing quaternary salt diol to the mustard using SOCl2.
In several cases the required (chloromethyl)- or (bromo-
methyl)heterocycles were known compounds: 5-(chloro-
methyl)-1-methyl-4-nitroimidazole20 for the synthesis of
2, 2-(chloromethyl)-5-nitrothiophene21 for the synthesis
of 4, and 2-(bromomethyl)-3-nitrothiophene22 for the
synthesis of 7. In two other cases the (chloromethyl)-

Hypoxia-Selective Antitumor Agents
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 21 3513
Sch em e 2a
Sch em e 5a
a
Conditions: (a) Cl2CHCO2-t-Bu, t-BuOK; (b) AcOH, 110 °C.
Sch em e 3a
a
Conditions: (a) Me2SO4, K2CO3; (b) NaBH4; (c) NaOH and then
Cu, quinoline, 180 °C; (d) Ph3PCl2; (e) MeN(CH2CH2OH)2; (f)
SOCl2.
a
Conditions: (a) Me2SO4, K2CO3; (b) NaBH4; (c) NaOH and then
Cu, quinoline, 180 °C; (d) Ph3PCl2.
described above and is illustrated in Scheme 5. With
this isomer pattern, although the chloromethyl com-
pound 27 could be isolated in a pure state, it was found
to be very sensitive to nucleophilic displacement and
hydrolysis. The quaternary salt analogues of 27, con-
taining an even better leaving group, were anticipated
to be more unstable than 27, and their synthesis was
not pursued. However, with 28, the addition of the extra
electron-withdrawing ester substituent was sufficiently
deactivating so that both 28 and the quaternary salt
diol 29 could be isolated without difficulty. A crude
sample of the mustard 30 was also prepared, but was
found to decompose on attempted purification.
Sch em e 4
heterocycles were readily obtained from the correspond-
ing aldehyde or acid: 1-methyl-5-nitropyrrole-2-carbox-
aldehyde23 and 1-methyl-5-nitropyrazole-4-carboxylic
acid24 were each reduced to the hydroxymethyl ana-
logues (NaBH4 or BH3‚DMS) and then converted to
the chlorides with MsCl/Et3N, thus leading to the
Red u ction P oten tia ls. One-electron reduction po-
tentials, E(1), were determined in aqueous solution by
measuring the equilibrium constant for electron transfer
between the radical anions of the compounds and
viologen reference standards.28 Although reduction po-
tentials for compounds 1-3 have been recorded previ-
ously,18 the reported value for 3 (-561 mV) was found
to require revision as a consequence of the very fast
breakdown of its radical anion, and the relatively slow
electron transfer from the reducing 2-propanyl radical
to 3. Using fast reduction by the aquated electron, and
high substrate concentrations to establish equilibrium
before any significant breakdown, a revised E(1) value
of -511 mV was determined. The reduction potentials
of all the compounds are collected in Table 1, with the
compounds arranged in order of decreasing electron
affinity. The compounds span an E(1) range from -277
to -511 mV, with the previously reported examples (1-
3) lying in the middle and at the bottom of the range.
Although there are only a limited number of structur-
ally diverse heterocycles, some SAR is apparent: for a
given substitution pattern, nitropyrroles are consider-
ably harder to reduce than either nitrothiophenes or
nitropyrazoles (3 vs 4, 9 vs 5), while the addition of an
electron-withdrawing ester substituent raises the re-
duction potential as anticipated (6 vs 9, where the ester
is conjugated to the nitro group). We have previously
noted that the quaternary methyl substituent is itself
strongly electron withdrawing (the Hammett σp value
for the CH2N+Me3 substituent has been variously
reported as +0.44 and +0.6729) and were therefore
interested in comparing the reduction potentials in
Table 1 with those of analogues bearing a more neutral
“benzyl” substituent. Where the hydroxymethyl ana-
logues were available for comparison (i.e., analogues of
2-4, 6, and 9),30 the change from CH2OH (σp 0.0029) to
25
preparation of 3 and 5, respectively.
The route to a further compound (8) made use of the
vicarious nucleophilic substitution (VNS) strategy re-
ported by Makosza and Bialecki20 (Scheme 2). Reaction
of 1-methyl-4-nitropyrazole (10) with the anion derived
from tert-butyl dichloroacetate gave, as anticipated,26
the product of VNS reaction at the 5-position (11). The
tert-butyl ester was cleaved with AcOH and the result-
ing acid decarboxylated to give the desired chloromethyl
intermediate 12.
A common intermediate for the final two analogues,
6 and 9, was ethyl 4-formyl-5-nitropyrrole-2-carboxyl-
ate27 (13) (Scheme 3). This pyrrole was N-methylated
and the aldehyde 14 reduced with NaBH4. Hydrolysis
of the ester 15 and decarboxylation using Cu/quinoline
at elevated temperature gave the unsubstituted ana-
logue 16. Both 15 and 16 were converted to their
chloromethyl analogues, and these were carried through
to the quaternary mustard compounds 6 and 9 as
described above.
Unsuccessful attempts were also made to prepare a
number of other analogues. Two isomers of the imid-
azole 2, with the quaternary group in the 2-position,
were successfully carried through to the diol stage
(Scheme 4), but all attempts to convert 19 and 20 to
the mustard forms led to isolation of the products of
intramolecular quaternization. Presumably other imid-
azole isomers (2-nitro-4-substituted and 5-nitro-4-
substituted) would show the same self-quaternization
behavior. Isomers of the pyrroles 6 and 9 in which the
substituents at the 2- and 3-positions were reversed
were also investigated. The chemistry follows that

3514 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 21
Tercel et al.
methyl quaternary salt was found to raise the reduction
potential, as expected. However, ∆E(1) was quite vari-
able (+94 mV for 3, +180 mV for 2), suggesting that
there is a complex electronic perturbation on introduc-
ing a methyl quaternary substituent, which likely varies
with both the heterocycle and substitution pattern.
Aer obic Toxicity. Aerobic cytotoxicities were deter-
mined as IC50 values after 4 h of drug exposure in four
cell lines: the Chinese hamster lines AA8 and UV4, the
murine mammary carcinoma EMT6, and the human
ovarian carcinoma Skov3. The UV4 cell line is an
ERCC1 mutant defective in nucleotide excision repair31
and is hypersensitive to DNA cross-linking agents such
as nitrogen mustards.32 Striking variations in toxicity
were observed among the different cell lines and com-
pounds (Table 1). Although all the quaternary salts were
less toxic than mechlorethamine, the ratio varied from
7 (4 in UV4) to nearly 10000 (1 in AA8). Despite these
large variations some patterns do emerge: the rela-
tive cell line sensitivity is in all cases UV4 > EMT6 >
Skov3 > AA8, the same order as observed for mechlor-
ethamine itself. Further, it can be seen that 1 is the
least toxic compound to all cell lines, while 4 (with a
single exception) is the most toxic. The extent to which
toxicity is masked in the prodrugs is clearly not directly
related to reduction potential, as, among the hetero-
cycles, 4 and 5 are at the extremes of toxicity but their
E(1) values are separated by only 10 mV. These obser-
vations suggest that aerobic toxicity is unlikely to be a
consequence of reductive activation (unless there is a
great variation in the stability of the reduced intermedi-
ates); an alternative possibility is that there is a slow
release of mechlorethamine by direct nucleophilic dis-
placement (either chemical or enzymatic) during the
drug exposure phase.33
F igu r e 1. Clonogenic survival curves for EMT6 cells (106 cells/
mL) exposed to mechlorethamine (circles), 2 (triangles), or
5-(chloromethyl)-1-methyl-4-nitroimidazole (squares) for 3 h
under aerobic (open symbols) or hypoxic (filled symbols)
conditions. Representative data from a single experiment for
each compound. The control plating efficiencies were 80-92%
under aerobic conditions and 69-82% under hypoxia.
quaternary salts was investigated by determining dose-
response relationships for clonogenic killing of EMT6
cells, obtained by dissociation of multicellular spheroids,
under aerobic and anoxic conditions. In these experi-
ments, stirred and continuously gassed cell suspensions
(106 cells/mL) were exposed to drugs for 3 h and washed
by centrifugation, and the plating efficiency determined.
Typical results are illustrated for mechlorethamine, for
2, and for the corresponding chloromethyl compound
[5-(chloromethyl)-1-methyl-4-nitroimidazole] in Figure
1. This shows that 2 is 60-fold less toxic than mechlor-
ethamine under aerobic conditions (similar to the 90-
fold differential against EMT6 cells in the IC50 assay),
but is almost as potent as mechlorethamine under
hypoxic conditions. This suggests almost quantitative
release of mechlorethamine from 2 under hypoxia. The
chloromethyl analogue was slightly less potent than 2
under aerobic conditions, and showed no significant
hypoxic selectivity. This is consistent with the proposed
mechanism for the hypoxic toxicity of 2, in which
cytotoxicity is due not to reactive species derived from
the reduced nitro group (e.g., nitroso or hydroxylamine),
but to intramolecular electron transfer leading to frag-
mentation and release of mechlorethamine.
The same assay was used to compare the hypoxic
selectivity (quantified as the hypoxic cytotoxicity ratio,
HCR, as defined in Table 1) of the series of heterocyclic
analogues. As noted previously with 1,15 cytotoxicity was
quite variable in repeat experiments, as indicated by
the large standard errors for cytotoxic potency (C10) and
HCR in Table 1. Despite considerable investigation of
possible sources of variability (differences in cell density,
growth stage, medium age, ascorbate concentration,
exposure to light, etc.), we have not been able to
circumvent this problem. Nevertheless, it is clear that
there are substantial differences in hypoxic selectivity
between the different heterocycles, with the pyrazole 5
(HCR 2500) being the most selective and the thiophene
7 (HCR 6) the least selective. Since many of the
analogues are almost as toxic as mechlorethamine
under hypoxic conditions, these differences in HCR
largely reflect variations in the aerobic toxicity of the
Release of mechlorethamine, whether via reduction
or nucleophilic displacement, is implicated in the aerobic
toxicity by the large hypersensitivity factor [HF ) IC50-
(AA8)/IC50(UV4)] between the AA8 and UV4 cell lines.
Typical HF values for cross-linking agents fall in the
range 10-7032 (in Table 1 mechlorethamine gives a
value of 86), while the results for the quaternary salts
lie between 50 and 660. Such abnormally high HF
values were also previously observed with several
nitrobenzyl quaternary mustards, and in those cases it
was suggested that saturable uptake of the quaternary
salts might magnify the toxicity differential.15 A further
potential contributor to aerobic toxicity is the generation
of reactive oxygen species through redox cycling, which
would be expected to be more pronounced with the more
easily reduced (higher reduction potential) compounds,
and was in fact implicated in the aerobic toxicity of
nitrobenzyl analogues.15
Clearly, there are a number of factors potentially
involved in the aerobic toxicity of mustard quaternary
salts, and quaternizing a mustard does not always
eliminate its toxicity. However, in most of the examples
in Table 1, the toxicity differential compared to mechlor-
ethamine is large enough that substantial hypoxia-
selective cytotoxicity should still be achievable if these
prodrugs are mainly reduced by oxygen-sensitive re-
ductases.
Hyp oxia -Selective Cytotoxicity. The ability of
endogenous reductases to activate the nitroarylmethyl

Hypoxia-Selective Antitumor Agents
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 21 3515
Ta ble 2. In Vivo Toxicity (Maximum Tolerated Dose in C3H/HeN Mice) and Activity against Aerobic and Hypoxic Cells in RIF-1
Tumors
activity against RIF-1 tumor
MTD,
single ip dose
(µmol/k g)
log(cell kill), drug + radiation (15 Gy)
dose
(µmol/k g)
log(cell kill),
drug onlyb
compda
pc
ns
ns
<0.001
ns
<0.001
ns
ns
ns
ns
<0.001
ns
ns
ns
drug 30 min beforeb
pc
drug 5 min afterb
pc
TPZ
CI-1010
PORF
AQ4N
300
940
75
225
705
0.1 ( 0 (2)
0.53 ( 0.15 (4)
0.72 ( 0.13 (4)
1.21 ( 0.22 (6)
-0.17 ( 0.09 (3)
-0.16 ( 0.24 (5)
-0.16 ( 0.16 (6)
-0.55 ( 0.07 (5)
-0.01 ( 0.08 (5)
0.16 ( 0.13 (5)
0.39 ( 0.09 (6)
-0.17 ( 0.21 (6)
-0.38 ( 0.03 (5)
0.48 ( 0.09 (5)
0.04 ( 0.14 (8)
ns
0.0170
<0.001
ns
ns
ns
ns
ns
ns
ns
ns
ns
ns
ns
0.52 ( 0.07 (22)
1.08 ( 0.11d (38)
0.86 ( 0.23 (6)
0.09 ( 0.19 (5)
-0.55 ( 0.22 (5)
0.17 ( 0.15 (6)
-0.73 ( 0.09 (5)
0.92 ( 0.11 (5)
0.11 ( 0.19 (5)
1.01 ( 0.13 (41)
0.10 ( 0.13 (6)
-0.47 ( 0.17 (5)
0.23 ( 0.21 (5)
-0.03 ( 0.13 (8)
0.014
<0.001
<0.001
ns
0.07 ( 0.02 (2)
0.63 ( 0.15 (4)
0.08 ( 0.13 (3)
0.54 ( 0.10 (2)
-0.04 ( 0.05 (4)
0.03 ( 0.00 (2)
0.24 ( 0.05 (2)
0.23 ( 0.03 (2)
0.38 ( 0.12 (4)
-0.12 ( 0.02 (4)
0.24 ( 0.06 (2)
0.25 ( 0.02 (2)
-0.18 ( 0.06 (2)
56.2
450
56.2
31.6
31.6
100
56.2
100
23.7
56.2
42.1
17.8
450
56.2
31.6
31.6
100
56.2
100
23.7
56.2
42.1
17.8
4
ns
5
ns
6
ns
1
7
<0.001
ns
2
8
<0.001
ns
9
ns
3
ns
HN2
ns
ns
a
b
TPZ, tirapazamine; PORF, porfiromycin; HN2, mechlorethamine. Numbers in parentheses refer to the number of mice per group.
c Significance of drug-induced cell killing, assessed as the decrease in clonogens relative to the appropriate control (untreated, or radiation
only). Note that because of differences in the number of animals per group the magnitude of the effect provides a more useful basis for
comparing compounds than does the statistical significance. One out of thirty-eight tumors below the limit of detection (103 clonogens/
d
g) was converted to 103 clonogens/g for this analysis.
compounds. There is also, however, a trend to lesser
hypoxic potency at lower E(1) values (particularly if the
single nonheterocyclic analogue 1 is excluded from the
comparison). Presumably this reflects less facile meta-
bolic reduction of these analogues, but even so, the
lowest potential compound (pyrrole 3, E(1) ) -511 mV)
still shows appreciable hypoxic selectivity (HCR 12) in
this assay.
DNA Cr oss-Lin k in g in Hyp oxic RIF -1 Cells. The
hypoxia-selective cytotoxicity of 2 was investigated
further in RIF-1 tumor cells, under the same conditions
as used above for EMT6 cells, but with 1 h of drug
exposure. This demonstrated an HCR of 59, with C10
values of 200 µM under aerobic conditions and 3.4 µM
under hypoxia (data not shown). The ability of 2 to form
DNA cross-links in these cells was tested using the
comet assay (single-cell gel electrophoresis) following 1
h of drug exposure at 3.5 µM. In the comet assay,34 DNA
breakage in individual cells is detected by alkaline lysis
F igu r e 2. Comet analysis of RIF-1 cell suspensions (106 cells/
in gels on microscope slides, followed by electrophoresis,
mL) following exposure to 2 at 3.5 µM for 1 h under aerobic
and hypoxic conditions. Cells were washed, and then half were
which draws DNA fragments out of nuclei to form
comet-like tails. Comets are stained with propidium,
and the tail moment is determined using an image
analysis system. Exposure of RIF-1 cells to 2 under
either aerobic or hypoxic conditions did not cause DNA
breakage (Figure 2A-C), whereas γ irradiation (10 Gy)
under aerobic conditions gave extensive breakage (Fig-
ure 2D). Pretreatment with 2 under aerobic conditions
had no effect on subsequent DNA breakage by radiation
(Figure 2E), whereas breakage was strongly suppressed
by pretreatment with 2 under hypoxia before irradiation
(Figure 2F). These results are similar to the effect of
known DNA cross-linking agents, including mechlor-
ethamine, in preventing radiation-induced changes in
DNA breakage in this assay,35 and demonstrate hy-
poxia-selective induction of DNA interstrand cross-links
in RIF-1 cells.
irradiated on ice with 10 Gy γ irradiation. Cells were analyzed
for DNA breakage or cross-linking, with the histograms
showing the distibution of tail moments from 100 comets per
treatment condition.
days later at the next two dose levels. The prodrugs all
showed lower host toxicity than mechlorethamine,
although the change was only modest, with MTD values
ranging from 23.7 µmol/kg for 8 to 100 µmol/kg for 1
and 2. The dose-response curve was shallow, with only
moderate toxicity (1-2 deaths/group of six, and weight
loss <10% at 2-3 dose levels above the MTD). This
makes the MTD an imprecise measure of toxicity in this
series. Nonetheless, it is obvious that the prodrugs
protect against the toxicity of the mustard effector much
less well in vivo than in aerobic cell cultures (Table 1),
and that the MTD does not correlate with the aerobic
IC50 or C10 values. At doses above the MTD, the median
day of death was 5 days after treatment, and signs of
gut toxicity (diarrhea and/or bloody stools) were some-
times seen. These findings are consistent with nitrogen
In Vivo Toxicity. The maximum tolerated doses
(MTDs) of the quaternary nitrogen mustard derivatives
in mice were evaluated using single ip doses at 1.33-
fold dose increments (Table 2). The MTD of mechlor-
ethamine was 17.8 µmol/kg with deaths occurring 2-8

3516 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 21
Tercel et al.
F igu r e 4. Dose-response activity of 2 either alone (open
symbols) or in combination with radiation (closed symbols)
against RIF-1 tumors. In the combination treatment 2 was
administered 5 min after a 15 Gy radiation dose. Values are
the mean ( SEM for 3-9 mice. Groups showing a statistically
significant drug effect (p < 0.05) relative to the appropriate
controls (untreated, or radiation only) are marked with an
asterisk.
F igu r e 3. Activity of the reference hypoxia-selective cytotoxin
CI-1010 (A) and the nitroarylmethyl quaternary nitrogen
mustards 2 and 5 (B) against aerobic and hypoxic cells in RIF-1
tumors. The radiation dose was 15 Gy, and the drug doses were
705 µmol/kg for CI-1010, 100 µmol/kg for 2, and 31.6 µmol/kg
for 5. Data points are individual tumors. The downward arrow
indicates that no colonies were detected. Groups showing a
statistically significant drug effect (p < 0.05) relative to the
appropriate controls (untreated, or radiation only) are marked
with an asterisk.
The above assay was used to evaluate the nitroaryl-
methyl quaternary nitrogen mustards, using the ex-
perimental design illustrated in Figure 3B. The latter
experiment, for example, demonstrates the lack of
activity of the pyrazole 5 in comparison with the
imidazole 2, which was used as an internal standard in
all experiments. The calculated log(cell kill) for each of
the prodrugs alone or in combination with radiation is
summarized in Table 2. Mechlorethamine itself showed
no significant activity either alone or in combination
with radiation in this assay. The most active of the
nitroarylmethyl quaternary mustards was compound 2,
which provided an additional log(cell kill) when admin-
istered 5 min after radiation, while analogue 1 showed
slightly less activity. Using this schedule of administra-
tion, 2 was at least as active as CI-1010 and porfiro-
mycin (at 75% of their MTD), and significantly better
than both tirapazamine (75% MTD) and AQ4N (100%
MTD). Both 1 and 2 were less active in the absence of
radiation, thus demonstrating some selectivity for hy-
poxic cells in tumors. They also appeared to be less
active when administered 30 min before irradiation,
suggesting possible interference with tumor blood flow
(and consequent radioresistance via induction of hy-
poxia). In contrast, none of the other nitroarylmethyl
quaternary mustards showed any significant activity
either pre- or postirradiation. There was also no obvious
relationship between the in vitro and in vivo results,
with the compound having the highest hypoxic selectiv-
ity in culture (5) lacking activity in tumors.
mustard toxicity. Taken together, these results suggest
that the prodrugs are activated extensively, either by
nitroreduction or by nucleophilic displacement of mechlor-
ethamine, in non-tumor-bearing mice.
An titu m or Activity. Antitumor activity was as-
sessed by excising RIF-1 tumors 18 h after treatment
of mice and measuring cell survival by assaying clono-
genic cells. Each quaternary salt compound was admin-
istered at the MTD, either alone or in combination with
a dose of ionizing radiation (15 Gy) sufficient to sterilize
all except the most hypoxic cells in the tumors. In this
assay, selective activity against hypoxic cells manifests
as a greater increase in killing due to the drug in
combination with radiation than for the drug alone.
Typical results are shown for the reference bioreduc-
tive drug CI-1010 at 75% of its MTD in Figure 3A, which
provided statistically significant additional cell killing
when administered 30 min before or 5 min after radia-
tion, but lacked activity as a single agent. The mean
log(cell kill) due to CI-1010 under each of these condi-
tions is shown in Table 2, where its activity is compared
to three other hypoxic cytotoxins: tirapazamine, por-
firomycin, and AQ4N. Tirapazamine at 75% of its MTD
was not quite as active as CI-1010, but also showed
selective toxicity in combination with radiation. Por-
firomycin at 75% of its MTD showed activity comparable
to that of CI-1010, but this was not restricted to its
combination with radiation, indicating that it is not
highly selective for hypoxic cells in this tumor. AQ4N
was inactive at its MTD against aerobic or hypoxic cells
in RIF-1 tumors, although Patterson et al. have reported
this compound to enhance the radiation response of
RIF-1 tumors using a growth delay end point.36
Given the encouraging results with 2, further studies
of its antitumor activity were undertaken. Investigation
of the time course of interaction with radiation showed
statistically significant activity at most postirradiation
times investigated (up to 90 min), with no clear opti-
mum (data not shown). The dose-response relationship
was also investigated (Figure 4), with the drug admin-
istered 5 min after radiation, using drug doses at and
above the MTD. Cytotoxicity due to the drug in combi-

Hypoxia-Selective Antitumor Agents
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 21 3517
the 35 animals treated with the combination. Evalua-
tion of the bioreductive drug dose response (Figure 5C)
confirmed these results; both drugs gave marked re-
sponses when administered 1 h before cisplatin, but 2
was more toxic than tirapazamine in the combination.
In this experiment, 2 was toxic by itself (four deaths in
seven mice) at only 75% of its nominal MTD, so it is
not clear that the toxicity in the combination is due to
an interaction with cisplatin.
The variable toxicity of 2 in tumor-bearing mice
confounded further attempts to assess its therapeutic
activity using tumor growth delay assays. A protracted
series of experiments (data not shown) demonstrated
statistically significant activity of 2 at 75 µmol/kg in
combination with radiation against RIF-1 and MDAH-
MCa-4-tumors in some groups, but severe toxicity (up
to fives deaths in seven mice) in other groups under
identical treatment conditions. Reduction of the dose of
2 to 56 µmol/kg did not completely eliminate this
toxicity, and generally provided nonsignificant activity.
Thus, 2 appears to have activity as a hypoxia-selective
cytotoxin in tumors, but its variable toxicity is a serious
problem for further development of this series of com-
pounds as bioreductive drugs.
Con clu sion s
On the basis of the observed clean one-electron
reduction chemistry of a nitroimidazole and a nitropyr-
role quaternary salt of mechlorethamine, we have
prepared and analyzed a series of nitroheterocyclic
analogues as hypoxia-selective cytotoxins. Although all
quaternary salt compounds were less toxic than me-
chlorethamine in vitro, and all were more toxic under
hypoxic than aerobic conditions (with HCR ratios as
much as several thousand fold), the differentials were
quite variable and there was no obvious correlation with
reduction potential or structure. The 4-nitroimidazole
2 was found to become as toxic as mechlorethamine
under hypoxic conditions in EMT6 cell suspensions
(Figure 1), while the corresponding chloromethyl ana-
logue showed no significant differential. Compound 2
also demonstrated hypoxia-selective DNA cross-linking
in RIF-1 cells (Figure 2). Both of these observations
strongly support the proposed mechanism of activation,
involving reductive release of mechlorethamine. In vivo,
in a RIF-1 excision assay, 2 showed comparable or
superior activity to hypoxia-selective cytotoxins of cur-
rent clinical interest, when combined with radiation or
cisplatin treatment. Unfortunately, this activity was
accompanied by unpredictable host toxicity, and the
accumulated evidence suggests that these nitroaryl-
methyl quaternary salts are too unstable with regard
to nonspecific release of mechlorethamine to be of use
as bioreductive agents.
F igu r e 5. Activity of 2 and tirapazamine in combination with
cisplatin (27 µmol/kg) against RIF-1 tumors, determined by
measurement of tumor growth delay. All drugs were admin-
istered as single ip doses. (A) Time course of interaction
between tirapazamine (270 µmol/kg) and cisplatin. (B) Time
course of interaction between 2 (75 µmol/kg) and cisplatin. (C)
Bioreductive drug dose response, with tirapazamine or 2
administered 1 h before cisplatin. Cisplatin (triangles), tira-
pazamine (circles), and 2 (squares), either alone (open symbols)
or in combination with cisplatin (closed symbols). Each point
is the mean ( SEM for a group of seven mice, with the number
of deaths from drug toxicity shown in parentheses. An asterisk
indicates that the time to the end point (1.2 g of tumor) is
significantly different from that of cisplatin alone.
nation with radiation was greater than for the drug
alone, with, for example, a log(cell kill) at 178 µmol/kg
of 2.4 log units following radiation versus an interpo-
lated value of 1.8 log units for drug alone at this dose.
This confirms that 2 is a hypoxia-selective cytotoxin in
RIF-1 tumors, at least at high doses, although the
notable cell killing in the absence of radiation also
implicates significant toxicity to aerobic cells in vivo.
Tirapazamine has been reported to be highly active
against the RIF-1 tumor when combined with cis-
platin,10,37 probably because it inhibits the repair of
cisplatin-induced DNA cross-links.38 We compared ti-
rapazamine and 2 in this model, using tumor growth
delay as the end point (Figure 5). Varying the time
between administration of the two agents confirmed the
large enhancement of cisplatin activity when tira-
pazamine was given before (but not after) cisplatin
(Figure 5A). Statistically significant increases in tumor
growth delay relative to that of cisplatin alone were also
observed when 2 was administered either before or after
cisplatin, at each time tested, with no obvious time
dependence (Figure 5B). However, unlike tirapazamine,
enhancement of cisplatin antitumor activity by 2 was
accompanied by substantial toxicity, with 13 deaths in
Exp er im en ta l Section
Ch em istr y. Analyses were carried out in the Microchemical
Laboratory, University of Otago, NZ. Melting points were
determined using an Electrothermal Model 9200 digital melt-
ing point apparatus and are as read. NMR spectra were
obtained on a Bruker AM-400 spectrometer at 400 MHz (1H)
or 100 MHz (13C). Mass spectra were obtained on a Varian
VG 7070 mass spectrometer at nominal 5000 resolution.

3518 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 21
Tercel et al.
Gen er a l Meth od for th e Con ver sion of Nitr o(h a lo-
m eth yl)h eter ocycles to th eir Qu a ter n a r y Sa lt Diol a n d
Mu sta r d An a logu es. P r ep a r a tion of N,N-Bis(2-ch lor oeth -
yl)-N-m eth yl-N-[(1-m eth yl-4-n itr o-5-im id a zolyl)m eth yl]-
a m m on iu m Ch lor id e (2). A solution of 5-(chloromethyl)-1-
methyl-4-nitroimidazole20 (1.22 g, 6.9 mmol) and N-methyl-
diethanolamine (1.24 g, 10.4 mmol) in CH3CN (40 mL) was
stirred at reflux for 18 h and then cooled to 20 °C and the
supernatant decanted. The oily residue was crystallized from
EtOH/Et2O to give N,N-bis(2-hydroxyethyl)-N-methyl-N-[(1-
methyl-4-nitro-5-imidazolyl)methyl]ammonium chloride as a
This diol in SOCl2 for 4 days gave 5 as a white solid (88%):
mp 155-156 °C (i-PrOH); H NMR [(CD3)2SO] δ 7.98 (s, 1 H,
H-3), 4.91 (s, 2 H, ArCH2N), 4.22-4.14 (m, 4 H, NCH2CH2Cl),
4.18 (s, 3 H, NCH3), 3.91-3.77 (m, 4 H, NCH2CH2Cl), 3.12 (s,
3 H, +NCH3). Anal. (C10H17N4Cl3O2) C, H, N, Cl.
1
N,N-Bis(2-ch lor oeth yl)-N-m eth yl-N-[(2-(eth oxyca r bon -
yl)-1-m eth yl-5-n itr o-4-p yr r olyl)m eth yl]a m m on iu m Ch lo-
r id e (6). Ethyl 4-(chloromethyl)-1-methyl-5-nitropyrrole-2-
carboxylate (17) and N-methyldiethanolamine (1.0 equiv) at
reflux for 13 h gave on cooling and dilution with EtOH N,N-
bis(2-hydroxyethyl)-N-methyl-N-[(2-(ethoxycarbonyl)-1-methyl-
5-nitro-4-pyrrolyl)methyl]ammonium chloride as yellow prisms
(61%): mp 78-82 °C (MeOH/Et2O); 1H NMR [(CD3)2SO] δ 7.31
(s, 1 H, H-3), 5.47 (t, J ) 4.9 Hz, 2 H, OH), 4.88 (s, 2 H,
ArCH2N), 4.35 (q, J ) 7.1 Hz, 2 H, CH2CH3), 4.18 (s, 3 H,
NCH3), 3.94-3.84 (m, 4 H, NCH2CH2O), 3.65-3.57 (m, 2 H,
NCH2CH2O), 3.49-3.42 (m, 2 H, NCH2CH2O), 3.05 (s, 3 H,
+NCH3), 1.33 (t, J ) 7.0 Hz, 3 H, CH2CH3). Anal. (C14H24N3-
ClO6‚MeOH) C, H, N, Cl.
1
cream solid (1.51 g, 74%): mp 177-178 °C; H NMR [(CD3)2-
SO] δ 8.13 (s, 1 H, H-2), 5.59 (t, J ) 4.6 Hz, 2 H, 2 × OH),
5.25 (br s, 2 H, ArCH2N), 3.98-3.86 (m, 4 H, NCH2CH2O), 3.90
(s, 3 H, NCH3), 3.85-3.75 (m, 2 H, NCH2CH2O), 3.58-3.48
(m, 2 H, NCH2CH2O), 3.12 (s, 3 H, +NCH3); 13C NMR δ 147.5,
139.8, 120.6, 63.5, 55.4, 55.0, 47.6, 33.8. Anal. (C10H19N4ClO4)
C, H, N, Cl.
This diol (1.04 g, 3.5 mmol) was added in portions to SOCl2
(5 mL) and the mixture stirred at 20 °C for 3 days and then
evaporated at 20 °C. The residue was crystallized from MeOH
to give 2 as a white solid (0.77 g, 66%): mp 173-176 °C dec;
1H NMR [(CD3)2SO] δ 8.15 (s, 1 H, H-2), 5.22 (s, 2 H, ArCH2N),
4.26-4.12 (m, 4 H, NCH2CH2Cl), 4.06-3.97 (m, 2 H, NCH2-
CH2Cl), 3.95-3.86 (m, 2 H, NCH2CH2Cl), 3.89 (s, 3 H, NCH3),
3.18 (s, 3 H, +NCH3); 13C NMR δ 147.8, 140.2, 119.4, 61.4, 54.9,
47.3, 36.0, 33.8. Anal. (C10H17N4Cl3O2) C, H, N, Cl.
This diol in SOCl2 for 5 days gave 6 as a pale green powder
1
(85%): mp 125-127 °C (EtOH/Et2O); H NMR [(CD3)2SO] δ
7.27 (s, 1 H, H-3), 4.90 (s, 2 H, ArCH2N), 4.35 (q, J ) 7.1 Hz,
2 H, CH2CH3), 4.22-4.11 (m, 4 H, NCH2CH2Cl), 4.17 (s, 3 H,
NCH3), 3.92-3.76 (m, 4 H, NCH2CH2Cl), 3.12 (s, 3 H, +NCH3),
1.32 (t, J ) 7.1 Hz, 3 H, CH2CH3). Anal. (C14H22N3Cl3O4‚H2O)
C, H, N, Cl.
N,N-Bis(2-ch lor oet h yl)-N-m et h yl-N-[(3-n it r o-2-t h ien -
yl)m et h yl]a m m on iu m Ch lor id e (7). 2-(Bromomethyl)-3-
nitrothiophene22 and N-methyldiethanolamine (1.0 equiv) at
reflux for 2 h gave on cooling and dilution with EtOAc the
crude quaternary bromide salt. This was dissolved in MeOH/
H2O (1:1) and stirred with a large excess of Biorad AG 1-X4
resin (HCl form) at 20 °C for 1 h. The resin was filtered and
washed with MeOH/H2O, the filtrate was evaporated, and the
residue was recrystallized to give N,N-bis(2-hydroxyethyl)-N-
methyl-N-[(3-nitro-2-thienyl)methyl]ammonium chloride as a
The following compounds were prepared by the same
general method.
N,N-Bis(2-ch lor oeth yl)-N-m eth yl-N-[(1-m eth yl-5-n itr o-
2-p yr r olyl)m eth yl]a m m on iu m Ch lor id e (3). 2-(Chloro-
methyl)-1-methyl-5-nitropyrrole and N-methyldiethanolamine
(1.0 equiv) at reflux for 3 h gave on cooling N,N-bis(2-hy-
droxyethyl)-N-methyl-N-[(1-methyl-5-nitro-2-pyrrolyl)methyl]-
ammonium chloride as a cream solid (76%); mp 180-181 °C;
1H NMR [(CD3)2SO] δ 7.34 (d, J ) 4.5 Hz, 1 H, H-3 or H-4),
6.73 (d, J ) 4.5 Hz, 1 H, H-3 or H-4), 5.59 (t, J ) 5.0 Hz, 2 H,
OH), 4.93 (s, 2 H, ArCH2N), 3.99 (s, 3 H, NCH3), 3.93-3.88
(m, 4 H, NCH2CH2O), 3.67-3.60 (m, 2 H, NCH2CH2O), 3.49-
3.42 (m, 2 H, NCH2CH2O), 3.06 (s, 3 H, +NCH3); 13C NMR δ
139.6, 127.8, 115.8, 113.0, 62.7, 57.5, 54.8, 47.8, 34.8. Anal.
(C11H20N3ClO4) C, H, N.
1
pale yellow solid (70%): mp 149-151 °C (MeOH/EtOAc); H
NMR [CD3)2SO] δ 8.40 (d, J ) 5.6 Hz, 1 H, H-5), 7.84 (d, J )
5.6 Hz, 1 H, H-4), 5.47 (t, J ) 4.8 Hz, 2 H, OH), 5.34 (s, 2 H,
ArCH2), 3.98-3.86 (m, 4 H, NCH2CH2O), 3.71-3.63 (m, 2 H,
NCH2CH2O), 3.58-3.50 (m, 2 H, NCH2CH2O), 3.13 (s, 3 H,
CH3). Anal. (C10H17N2ClO4S) C, H, N.
This diol in SOCl2 for 3 days gave 3 as a very pale yellow
powder (0.84 g, 92%): mp 140-141 °C (MeOH/Et2O); 1H NMR
[(CD3)2SO] δ 7.36 (d, J ) 4.5 Hz, 1 H, H-3 or H-4), 6.66 (d,
J ) 4.5 Hz, 1 H, H-3 or H-4), 4.98 (s, 2 H, ArCH2N), 4.24-
4.12 (m, 4 H, NCH2CH2Cl), 3.99 (s, 3 H, NCH3), 3.97-3.89
(m, 2 H, NCH2CH2Cl), 3.84-3.76 (m, 2 H, NCH2CH2Cl), 3.13
(s, 3 H, +NCH3); 13C NMR δ 140.0, 126.7, 115.9, 113.1, 61.2,
57.5, 47.7, 36.4, 35.3. MS (35Cl, FAB) m/z 294 (100, M+ for
ammonium ion); HRMS calcd for C11H18N3Cl2O2 294.0776,
found 294.0766.
N,N-Bis(2-ch lor oeth yl)-N-m eth yl-N-[(5-n itr o-2-th ien yl)-
m eth yl]a m m on iu m Ch lor id e (4). 2-(Chloromethyl)-5-ni-
trothiophene21 and N-methyldiethanolamine (1.0 equiv) at
reflux for 5 h gave crude N,N-bis(2-hydroxyethyl)-N-methyl-
N-[(5-nitro-2-thienyl)methyl]ammonium chloride as a dark oil
that was not purified but was treated directly with SOCl2 for
17 h to give 4 (27% for two steps): mp (EtOAc/MeOH) 161-
163 °C dec; 1H NMR [CD3)2SO] δ 8.23 (d, J ) 4.2 Hz, 1 H,
H-4), 7.58 (d, J ) 4.2 Hz, 1 H, H-3), 5.13 (s, 2 H, ArCH2), 4.24-
4.13 (m, 4 H, NCH2CH2Cl), 3.94-3.76 (m, 4 H, NCH2CH2Cl),
3.21 (s, 3 H, CH3). Anal. (C10H15N2Cl3O2S) C, H, N.
This diol in SOCl2 for 20 h gave 7 as a cream solid (87%):
mp 145-147 °C (MeOH/EtOAc); 1H NMR [(CD3)2SO] δ 8.90
(d, J ) 5.7 Hz, 1 H, H-5), 7.86 (d, J ) 5.7 Hz, 1 H, H-4), 5.39
(s, 2 H, ArCH2), 4.24-4.12 (m, 4 H, NCH2CH2Cl), 3.99-3.85
(m, 4 H, NCH2CH2Cl), 3.22 (s, 3 H, CH3). Anal. (C10H15N2-
Cl3O2S) C, H, N, Cl.
N,N-Bis(2-ch lor oeth yl)-N-m eth yl-N-[(1-m eth yl-4-n itr o-
5-p yr a zolyl)m et h yl]a m m on iu m Ch lor id e (8). 5-(Chloro-
methyl)-1-methyl-4-nitropyrazole (12) and N-methyldiethanol-
amine (1.0 equiv) at reflux for 16 h gave on cooling and dilution
with Et2O N,N-bis(2-hydroxyethyl)-N-methyl-N-[(1-methyl-4-
nitro-5-pyrazolyl)methyl]ammonium chloride as a white solid
(80%): mp 135-137 °C; 1H NMR [(CD3)2SO] δ 8.54 (s, 1 H,
H-3), 5.52 (t, J ) 4.7 Hz, 2 H, OH), 5.28 (s, 2 H, ArCH2N),
4.10 (s, 3 H, NCH3), 3.95-3.87 (m, 4 H, NCH2CH2O), 3.86-
3.71 (m, 2 H, NCH2CH2O), 3.56-3.48 (m, 2 H, NCH2CH2O),
3.13 (s, 3 H, +NCH3). Anal. (C10H19N4ClO4‚1/4H2O) C, H, N,
Cl.
This diol in SOCl2 for 3 days gave 8 as a white solid (90%):
1
mp 147-150 °C (MeOH/Et2O); H NMR [(CD3)2SO] δ 8.58 (s,
1 H, H-3), 5.27 (s, 2 H, ArCH2N), 4.26-4.14 (m, 4 H, NCH2-
CH2Cl), 4.12 (s, 3 H, NCH3), 4.09-4.01 (m, 2 H, NCH2CH2Cl),
3.97-3.88 (m, 2 H, NCH2CH2Cl), 3.20 (s, 3 H, +NCH3). Anal.
(C10H17N4Cl3O2) C, H, N, Cl.
N,N-Bis(2-ch lor oeth yl)-N-m eth yl-N-[(1-m eth yl-5-n itr o-
4-p yr a zolyl)m et h yl]a m m on iu m Ch lor id e (5). 4-(Chloro-
methyl)-1-methyl-5-nitropyrazole and N-methyldiethanol-
amine (1.0 equiv) at reflux for 26 h gave on cooling N,N-
bis(2-hydroxyethyl)-N-methyl-N-[(1-methyl-5-nitro-4-pyra-
zolyl)methyl]ammonium chloride as a cream solid (76%): mp
151-153 °C; 1H NMR [(CD3)2SO] δ 8.02 (s, 1 H, H-3), 5.52 (t,
J ) 4.9 Hz, 2 H, OH), 4.87 (s, 2 H, ArCH2N), 4.18 (s, 3 H,
NCH3), 3.95-3.84 (m, 4 H, NCH2CH2O), 3.63-3.54 (m, 2 H,
NCH2CH2O), 3.50-3.42 (m, 2 H, NCH2CH2O), 3.06 (s, 3 H,
+NCH3). Anal. (C10H19N4ClO4) C, H, N, Cl.
N,N-Bis(2-ch lor oeth yl)-N-m eth yl-N-[(1-m eth yl-2-n itr o-
3-p yr r olyl)m eth yl]a m m on iu m Ch lor id e (9). 1-Methyl-3-
(chloromethyl)-2-nitropyrrole (18) and N-methyldiethanol-
amine (1.0 equiv) at 20 °C for 3 days gave on cooling N,N-
bis(2-hydroxyethyl)-N-methyl-N-[(1-methyl-2-nitro-3-pyr-
rolyl)methyl]ammonium chloride as a cream solid (73%): mp
1
147-150 °C; H NMR [(CD3)2SO] δ 7.45 (d, J ) 2.8 Hz, 1 H,

Hypoxia-Selective Antitumor Agents
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 21 3519
H-4 or H-5), 6.62 (d, J ) 2.9 Hz, 1 H, H-4 or H-5), 5.42 (t, J )
4.9 Hz, 2 H, OH), 4.89 (s, 2 H, ArCH2N), 3.98 (s, 3 H, NCH3),
3.92-3.86 (m, 4 H, NCH2CH2O), 3.63-3.55 (m, 2 H, NCH2-
CH2O), 3.48-3.41 (m, 2 H, NCH2CH2O), 3.03 (s, 3 H, +NCH3).
Anal. (C11H20N3ClO4) C, H, N.
and evaporated, and the residue was purified by chromatog-
raphy (1:3 EtOAc/petroleum ether) to give recovered 1-methyl-
4-nitropyrazole (0.18 g, 17%) and tert-butyl chloro(1-methyl-
4-nitro-5-pyrazolyl)acetate (11) as a pale yellow solid (1.39 g,
60%): 1H NMR (CDCl3) δ 8.11 (s, 1 H, H-3), 6.54 (s, 1 H,
CHCO2-t-Bu), 3.98 (s, 3 H, NCH3), 1.48 (s, 9 H, t-Bu).
This diol in SOCl2 for 3 days gave 9 as a pale yellow powder
1
This crude ester (1.38 g, 5.0 mmol) was stirred at reflux in
AcOH (15 mL) for 21 h, and the solvent was evaporated. TLC
analysis showed a mixture of the acid and decarboxylated
product. The residue was dissolved in EtOAc, washed with
water, and extracted with aqueous NaHCO3. The organic layer
was dried (Na2SO4) and evaporated to give a brown oil (0.62
g). The aqueous layer was acidified and extracted with EtOAc,
and the extracts were dried (Na2SO4) and evaporated. The
residue (0.22 g) was dissolved in DMF (5 mL) and the solution
heated for 10 min to 110 °C to complete the decarboxylation.
The solvent was evaporated, the residue was combined with
the above brown oil, and the product was purified by chroma-
tography (1:3 EtOAc/petroleum ether) to give 12 as a very pale
yellow liquid (0.71 g, 81%). A sample was crystallized from
PhH/petroleum ether: mp 67-69 °C; 1H NMR (CDCl3) δ 8.09
(s, 1 H, H-3), 5.01 (s, 2 H, CH2Cl), 4.00 (s, 3 H, NCH3). MS
(35Cl, EI) m/z 175 (25, M+), 140 (100, M - Cl); HRMS calcd for
C5H6ClN3O2 175.01485, found 175.01482. Anal. (C5H6N3ClO2)
C, H, N.
Eth yl 4-F or m yl-1-m eth yl-5-n itr op yr r ole-2-ca r boxyla te
(14). Dimethyl sulfate (0.31 mL, 3.2 mmol) was added to a
mixture of ethyl 4-formyl-5-nitropyrrole-2-carboxylate (13)
(0.57 g, 2.7 mmol) and K2CO3 (0.56 g, 4.0 mmol) in DMSO (4
mL), and the brown suspension was stirred at room temper-
ature for 1 h. The mixture was diluted with water, acidified
with HCl (2 N), and extracted with EtOAc (2×). The extracts
were dried (Na2SO4), evaporated, and purified by chromatog-
raphy (1:9 EtOAc/petroleum ether) to give 14 as a pale green
solid (0.53 g, 86%). A sample was recrystallized from PhH/
petroleum ether: mp 59-60.5 °C; 1H NMR (CDCl3) δ 10.32
(s, 1 H, CHO), 7.42 (s, 1 H, H-3), 4.37 (q, J ) 7.1 Hz, 2 H,
CH2CH3), 4.33 (s, 3 H, NCH3), 1.39 (t, J ) 7.1 Hz, 3 H,
CH2CH3). Anal. (C9H10N2O5) C, H, N.
Eth yl 4-(Hyd r oxym eth yl)-1-m eth yl-5-n itr op yr r ole-2-
ca r boxyla te (15). NaBH4 (0.33 g, 8.7 mmol) was added in
portions to a solution of 14 (3.96 g, 17.5 mmol) in EtOH (100
mL), and the mixture was stirred at 20 °C for 20 min. Water
was slowly added, the EtOH was evaporated, and the residue
was diluted with aqueous NaCl and extracted with EtOAc
(2×). The extracts were washed with aqueous NaCl, dried (Na2-
SO4), and evaporated, and the resulting solid was recystallized
from PhH to give 15 (1.40 g, 35%) as white needles: mp 95.5-
96.5 °C. The mother liquor was evaporated and purified by
chromatography (1:9 EtOAc/petroleum ether) to give more 15
(1.46 g, 37%): 1H NMR (CDCl3) δ 7.01 (s, 1 H, H-3), 4.80 (br
s, 2 H, CH2OH), 4.36 (q, J ) 7.1 Hz, 2 H, CH2CH3), 4.31 (s, 3
H, NCH3), 2.49 (br s, 1 H, OH), 1.38 (t, J ) 7.1 Hz, 3 H,
CH2CH3). Anal. (C9H12N2O5) C, H, N.
(90%): mp 145-146 °C (MeOH/Et2O); H NMR [(CD3)2SO] δ
7.48 (d, J ) 2.9 Hz, 1 H, H-4 or H-5), 6.57 (d, J ) 2.8 Hz, 1 H,
H-4 or H-5), 4.92 (s, 2 H, ArCH2N), 4.22-4.12 (m, 4 H, NCH2-
CH2Cl), 3.98 (s, 3 H, NCH3), 3.90-3.75 (m, 4 H, NCH2CH2Cl),
3.10 (s, 3 H, +NCH3). Anal. (C11H18N3Cl3O2) C, H, N, Cl.
2-(Hydr oxym eth yl)-1-m eth yl-5-n itr opyr r ole. NaBH4 (0.19
g, 5.03 mmol) was added in portions to a stirred solution of
1-methyl-5-nitropyrrole-2-carboxaldehyde23 (0.78 g, 5.07 mmol)
in MeOH (40 mL) at 20 °C under N2. After addition was
complete, the reaction mixture was stirred for a further 20
min, then water (40 mL) was added, and the mixture was
saturated with solid K2CO3. Extraction with EtOAc gave
2-(hydroxymethyl)-1-methyl-5-nitropyrrole (0.77 g, 97%): mp
76-77 °C (EtOAc/petroleum ether) (lit.25 mp 83-86 °C); 1H
NMR (CDCl3) δ 7.16 (d, J ) 4.3 Hz, 1 H, H-3 or H-4), 6.17 (d,
J ) 4.3 Hz, 1 H, H-3 or H-4), 4.68 (s, 2 H, CH2), 4.02 (s, 3 H,
CH3). Anal. (C6H8N2O3) C, H, N.
2-(Ch lor om eth yl)-1-m eth yl-5-n itr op yr r ole. A stirred so-
lution of 2-(hydroxymethyl)-1-methyl-5-nitropyrrole (0.13 g,
0.83 mmol) in dry CH2Cl2 (5 mL) at 0 °C was treated with
MsCl (0.36 mL, 1.25 mmol), followed by Et3N (0.50 mL). After
being stirred at 0 °C for 15 min, the mixture was evaporated
to dryness under reduced pressure at 20 °C, and the resulting
brown solid was partitioned between water and EtOAc. The
organic layer was dried (Na2SO4), evaporated, and purified by
chromatography (1:1 CH2Cl2/petroleum ether) to give 2-(chlo-
romethyl)-1-methyl-5-nitropyrrole (0.13 g, 88%): mp 82.5-83.5
1
°C (EtOAc/petroleum ether); H NMR (CDCl3) δ 7.16 (d, J )
4.4 Hz, 1 H, H-3 or H-4), 6.26 (d, J ) 4.4 Hz, 1 H, H-3 or H-4),
4.60 (s, 2 H, CH2Cl), 4.02 (s, 3 H, CH3). Anal. (C6H7N2ClO2) C,
H, N.
4-(Hyd r oxym et h yl)-1-m et h yl-5-n it r op yr a zole. Borane
dimethyl sulfide (2 M solution in THF, 4.2 mL, 8.4 mmol) was
added to a solution of 1-methyl-5-nitropyrazole-4-carboxylic
acid24 (1.11 g, 6.5 mmol) in dry THF (50 mL) under nitrogen,
and the mixture was stirred at reflux for 80 min and then
cooled. MeOH, then water, and then 2 N HCl were added, and
the THF was evaporated. The residue was diluted with water
and extracted with EtOAc. The extracts were dried (Na2SO4),
evaporated, and purified by chromatography (1:1 EtOAc/
petroleum ether) to give 4-(hydroxymethyl)-1-methyl-5-nitro-
pyrazole as a white solid (0.52 g, 51%): mp 78-80 °C (PhH);
1H NMR (CDCl3) δ 7.58 (s, 1 H, H-3), 4.82 (d, J ) 3.4 Hz, 2 H,
CH2), 4.25 (s, 3 H, NCH3), 2.39 (br s, 1 H, OH). Anal.
(C5H7N3O3) C, H, N.
4-(Ch lor om eth yl)-1-m eth yl-5-n itr op yr a zole. MsCl (0.46
mL, 6.0 mmol) was added to a solution of 4-(hydroxymethyl)-
1-methyl-5-nitropyrazole (468 mg, 3.0 mmol) and Et3N (1.25
mL, 9.0 mmol) in CH2Cl2 (10 mL), and the mixture was stirred
at reflux for 1 h. The mixture was cooled, diluted with water,
and extracted with CH2Cl2. The extracts were dried (Na2SO4)
and evaporated, and the residue was purified by chromatog-
raphy (1:3 EtOAc/petroleum ether) to give 4-(chloromethyl)-
1-methyl-5-nitropyrazole as a very pale yellow liquid (374 mg,
71%): 1H NMR (CDCl3) δ 7.64 (s, 1 H, H-3), 4.77 (s, 2 H, CH2-
Cl), 4.25 (s, 3 H, NCH3). MS (35Cl, EI) m/z 175 (23, M+), 140
(100, M - Cl); HRMS calcd for C5H6N3ClO2 175.01485, found
175.01421.
Eth yl 4-(Ch lor om eth yl)-1-m eth yl-5-n itr op yr r ole-2-ca r -
boxyla te (17). Dichlorotriphenylphosphorane (1.66 g, 5.14
mmol) was added to a solution of 15 (390 mg, 1.71 mmol) in
CH3CN (20 mL), and the mixture was stirred at 20 °C for 2 h.
The mixture was evaporated, the residue purified by chroma-
tography (1:9 EtOAc/petroleum ether), and the crude product
crystallized from PhH/petroleum ether to give 17 as colorless
1
needles (267 mg, 63%): mp 82-83.5 °C; H NMR (CDCl3) δ
7.07 (s, 1 H, H-3), 4.81 (br s, 2 H, CH2Cl), 4.36 (q, J ) 7.2 Hz,
2 H, CH2CH3), 4.29 (s, 3 H, NCH3), 1.39 (t, J ) 7.2 Hz, 3 H,
CH2CH3). Anal. (C9H11N2ClO4) C, H, N.
5-(Ch lor om eth yl)-1-m eth yl-4-n itr op yr a zole (12). A so-
lution of 1-methyl-4-nitropyrazole (10)39 (1.07 g, 8.45 mmol)
and tert-butyl dichloroacetate (2.35 g, 12.7 mmol) in dry DMF
(6 mL) was added dropwise over 30 min to a suspension of
t-BuOK (3.32 g, 30 mmol) in dry DMF (20 mL) under nitrogen,
keeping the internal temperature in the range -25 to -35 °C.
The deep purple solution was stirred at -25 °C for a further
20 min, then poured into cold HCl (0.5 N, 80 mL), and
extracted with EtOAc (3×). The extracts were dried (Na2SO4)
3-(Hyd r oxym eth yl)-1-m eth yl-2-n itr op yr r ole (16). A so-
lution of NaOH (2.7 g, 68 mmol) in water (15 mL) was added
to a solution of 15 (1.02 g, 4.47 mmol) in EtOH (30 mL), and
the mixture was stirred at 20 °C for 1 h. The EtOH was
evaporated, and the aqueous phase was washed with EtOAc
and then acidified (HCl). The aqueous mixture was extracted
with EtOAc (3×), and the extracts were dried (Na2SO4) and
evaporated to give crude 4-(hydroxymethyl)-1-methyl-5-nitro-
pyrrole-2-carboxylic acid (0.84 g, 96%) as a red-brown solid.

3520 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 21
Tercel et al.
This acid was suspended in quinoline (6 mL) with Cu
powder (0.44 g), and the mixture was heated at 170-180 °C
for 50 min. The cooled mixture was diluted with HCl (2 N)
and extracted with EtOAc (3×), and the extracts were dried
(Na2SO4) and evaporated. The residue was purified by chro-
matography (2:3 EtOAc/petroleum ether) to give 16 as a pink
solid (0.45 g, 64% from 15). A sample was recrystallized from
PhH: mp 79-80.5 °C; 1H NMR (CDCl3) δ 6.78 (d, J ) 2.8 Hz,
1 H, H-4 or H-5), 6.27 (d, J ) 2.5 Hz, 1 H, H-4 or H-5), 4.80 (d,
J ) 6.8 Hz, 2 H, CH2OH), 4.00 (s, 3 H, NCH3), 2.75 (t, J ) 6.8
Hz, 1 H, OH). Anal. (C6H8N2O3) C, H, N.
(PhH/petroleum ether); 1H NMR (CDCl3) δ 10.55 (s, 1 H, CHO),
7.49 (s, 1 H, H-3), 4.38 (q, J ) 7.2 Hz, 2 H, CH2CH3), 4.33 (s,
3 H, NCH3), 1.40 (t, J ) 7.2 Hz, 3 H, CH2CH3). Anal.
(C9H10N2O5) C, H, N.
Da ta for Eth yl 5-(Hyd r oxym eth yl)-1-m eth yl-4-n itr o-
p yr r ole-2-ca r boxyla te (25): cream solid (81%); mp 119-
120.5 °C (PhH); 1H NMR (CDCl3) δ 7.48 (s, 1 H, H-3), 4.97 (d,
J ) 6.8 Hz, 2 H, CH2OH), 4.32 (q, J ) 7.1 Hz, 2 H, CH2CH3),
4.06 (s, 3 H, NCH3), 2.72 (t, J ) 7.1 Hz, 1 H, OH), 1.37 (t, J )
7.1 Hz, 3 H, CH2CH3). Anal. (C9H12N2O5) C, H, N.
Da ta for 2-(Hyd r oxym eth yl)-1-m eth yl-3-n itr op yr r ole
1
3-(Ch lor om eth yl)-1-m eth yl-2-n itr op yr r ole (18). Dichlo-
rotriphenylphosphorane (3.9 g, 12.1 mmol) was added to a
solution of 16 (0.63 g, 4.0 mmol) in CH3CN (50 mL), and the
mixture was stirred at 20 °C for 30 min. The mixture was
evaporated and the residue purified by chromatography (1:4
EtOAc/petroleum ether) to give 18 as a pale yellow solid (0.60
g, 85%). A sample was recrystallized from PhH/petroleum
(26): pale yellow solid (54%); mp 63-64 °C (PhH); H NMR
(CDCl3) δ 6.73 (d, J ) 3.4 Hz, 1 H, H-4 or H-5), 6.50 (d, J )
3.4 Hz, 1 H, H-4 or H-5), 4.89 (d, J ) 7.2 Hz, 2 H, CH2OH),
3.73 (s, 3 H, NCH3), 2.83 (t, J ) 7.2 Hz, 1 H, OH). Anal.
(C6H8N2O3) C, H, N.
Da t a for 2-(Ch lor om et h yl)-1-m et h yl-3-n it r op yr r ole
(27): white flakes (91%); mp 70-71 °C (PhH/petroleum ether);
1H NMR (CDCl3) δ 6.76 (d, J ) 3.2 Hz, 1 H, H-4 or H-5), 6.57
(d, J ) 3.2 Hz, 1 H, H-4 or H-5), 5.07 (s, 2 H, CH2Cl), 3.74 (s,
3 H, NCH3). Anal. (C6H7N2ClO2) C, H, N.
1
ether: mp 80-81 °C; H NMR (CDCl3) δ 6.75 (d, J ) 2.8 Hz,
1 H, H-4 or H-5), 6.37 (d, J ) 2.7 Hz, 1 H, H-4 or H-5), 4.86 (s,
2 H, CH2Cl), 4.00 (s, 3 H, NCH3). Anal. (C6H7N2ClO2) C, H, N.
N,N-Bis(2-h yd r oxyeth yl)-N-m eth yl-N-[(1-m eth yl-4-n i-
tr o-2-im id a zolyl)m eth yl]a m m on iu m Ch lor id e (19). Reac-
tion of 2-(chloromethyl)-1-methyl-4-nitroimidazole40 and N-
methyldiethanolamine according to the general method above
gave 19 as a light brown foam (90%): 1H NMR [(CD3)2SO] δ
8.58 (s, 1 H, H-5), 5.58 (t, J ) 4.9 Hz, 2 H, OH), 4.95 (s, 2 H,
ArCH2N), 3.98-3.85 (m, 4 H, NCH2CH2O), 3.88 (s, 3 H, NCH3),
3.75-3.65 (m, 4 H, NCH2CH2O), 3.24 (s, 3 H, +NCH3); 13C
NMR δ 145.7, 137.2, 124.9, 63.6, 56.4, 54.9, 48.9, 34.5. MS
(FAB) m/z 259 (100, M+ for ammonium ion); HRMS (FAB)
calcd for C10H19N4O4 259.1406, found 259.1399.
Da ta for Eth yl 5-(Ch lor om eth yl)-1-m eth yl-4-n itr op yr -
r ole-2-ca r boxyla te (28): cream crystalline solid (83%); mp
1
92-83 °C (PhH/petroleum ether); H NMR (CDCl3) δ 7.50 (s,
1 H, H-3), 5.08 (s, 2 H, CH2Cl), 4.33 (q, J ) 7.1 Hz, 2 H, CH2-
CH3), 4.06 (s, 3 H, NCH3), 1.38 (t, J ) 7.1 Hz, 3 H, CH2CH3).
Anal. (C9H11N2ClO4) C, H, N.
N,N-Bis(2-h yd r oxyet h yl)-N-m et h yl-N-[(2-(et h oxyca r -
b on yl)-1-m et h yl-4-n it r o-5-im id a zolyl)m et h yl]a m m on i-
u m Ch lor id e (29). Reaction of 28 and N-methyldiethanol-
amine according to the general method above gave 29 as a
cream solid (21%): mp 124-126 °C (MeOH/Et2O); 1H NMR
[(CD3)2SO] δ 7.57 (s, 1 H, H-3), 5.7-4.9 (v br m, 4 H, ArCH2N
and 2 × OH), 4.31 (q, J ) 7.0 Hz, 2 H, CH2CH3), 4.05 (s, 3 H,
NCH3), 3.95-3.75 (m, 6 H, NCH2CH2O), 3.55-3.46 (m, 2 H,
NCH2CH2O), 3.08 (s, 3 H, +NCH3), 1.32 (t, J ) 7.1 Hz, 3 H,
CH2CH3). Anal. (C14H24N3ClO6) C, H, N, Cl.
N,N-Bis(2-ch lor oeth yl)-N-m eth yl-N-[(2-(eth oxyca r bon -
yl)-1-m eth yl-4-n itr o-5-im idazolyl)m eth yl]am m on iu m Ch lo-
r id e (30). Reaction of 29 with SOCl2 according to the general
method above gave 30, which was isolated in an impure state
as a white solid (EtOH/Et2O): 1H NMR [(CD3)2SO] δ 7.59 (s,
1 H, H-3), 5.6-5.0 (v br m, 2 H, ArCH2N), 4.32 (q, J ) 7.1 Hz,
2 H, CH2CH3), 4.28-3.87 (m, 8 H, NCH2CH2Cl), 4.06 (s, 3 H,
NCH3), 3.16 (s, 3 H, +NCH3), 1.33 (t, J ) 7.0 Hz, 3 H, CH2CH3).
On e-Electr on Red u ction P oten tia ls. The pulse radiolysis
equipment used in these studies has been described.18 Ther-
modynamic reversible equilibria were established between the
one-electron-reduced compounds and reference viologen stan-
dards at pH 7 in deaerated aqueous solutions containing
2-propanol or 2-methyl-2-propanol (0.2 M) and phosphate
buffer (10 mM). The equilibrium constants were used to
calculate ∆E values and hence E(1) values for the compounds.28
The following are the reference compounds used and their
reduction potentials E(1) (mV): for 4, duroquinone, -260 (
10;28 for 1, 5, and 7, benzyl viologen, -375 ( 10;28,42 for 2, 8,
and 6, methyl viologen, -447 ( 7;43 for 3 and 9, triquat,
-548 ( 7.43
In Vitr o Cytotoxicity. Inhibition of proliferation of log-
phase cell monolayers was assessed in 96-well plates as
described previously,44 using 4 h exposure to compounds under
aerobic conditions followed by sulforhodamine B staining 3
days later. The IC50 was determined by interpolation as the
drug concentration required to inhibit cell density to 50% of
that of thecontrols on the same plate. Cytotoxic potency was
compared under aerobic (20% O2) and hypoxic (<20 ppm O2)
conditions using continuously gassed and stirred supensions
of EMT6 cells.45 Cells for this assay were obtained by enzy-
matic dissociation of EMT6 spheroids (ca. 800 µm diameter),
and cells were plated to determine clonogenicity after exposure
to drugs for 3 h. The C10 was determined as the interpolated
drug concentration required to give 10% cell survival relative
to the controls.
7-(2-Ch lor oeth yl)-1,7-d im eth yl-3-n itr o-5,6,7,8-tetr a h y-
d r o-1H-im id a zo[1,2-a ]p yr a zin iu m Dih yd r och lor id e (21).
Reaction of 19 with SOCl2 gave 21 as white prisms (70%): mp
1
150-155 °C dec (MeOH/Et2O); H NMR [(CD3)2SO] δ 9.35 (s,
1 H, H-2), 5.83 (AB coupling, J ) 17.1 Hz, 2 H, ArCHAHBN),
5.11 (dt, J ) 15.6, 4.7 Hz, 1 H, NCH2CH2N), 4.98-4.90 (m, 1
H, NCH2CH2N), 4.39-4.30 (m, 1 H, NCH2CH2N), 4.31 (t, J )
6.9 Hz, 2 H, NCH2CH2Cl), 4.27-4.18 (m, 1 H, NCH2CH2N),
4.18 (t, J ) 6.9 Hz, 2 H, NCH2CH2Cl), 3.96 (s, 3 H, NCH3),
3.48 (s, 3 H, +NCH3); 13C NMR [(CD3)2SO] δ 138.4, 137.1, 126.3,
64.4, 55.6, 54.3, 42.1, 49.2, 36.0, 35.4. MS (FAB) m/z 259 [100,
M2+ (bisammonium ion) - H+], 196 (80%, M2+ - H+ - CH2-
CH2Cl). Anal. (C10H17N4Cl3O2‚H2O) C, H, N, Cl.
N,N-Bis(2-h yd r oxyeth yl)-N-m eth yl-N-[(1-m eth yl-5-n i-
tr o-2-im id a zolyl)m eth yl]a m m on iu m Ch lor id e (20). Reac-
tion of 2-(chloromethyl)-1-methyl-5-nitroimidazole41 and N-
methyldiethanolamine according to the general method above
1
gave 20 as cream plates (74%): mp 162-163 °C (EtOH); H
NMR [(CD3)2SO] δ 8.22 (s, 1 H, H-4), 5.61 (t, J ) 5.6 Hz, 2 H,
OH), 5.04 (s, 2 H, ArCH2N), 4.03 (s, 3 H, NCH3), 3.98-3.90
(m, 4 H, NCH2CH2O), 3.77-3.64 (m, 4 H, NCH2CH2O), 3.26
(s, 3 H, +NCH3); 13C NMR δ 141.9, 140.0, 132.0, 63.7, 56.6,
54.9, 49.2, 34.2. Anal. (C10H19N4ClO4) C, H, N, Cl.
7-(2-Ch lor oeth yl)-1,7-d im eth yl-2-n itr o-5,6,7,8-tetr a h y-
d r o-1H-im id a zo[1,2-a ]p yr a zin iu m Dih yd r och lor id e (22).
Reaction of 20 with SOCl2 according to the general method
above gave 22 (6%): mp 135 °C dec; 1H NMR [(CD3)2SO] δ
9.42 (s, 1 H, H-3), 5.92 (AB coupling, J ) 17.0 Hz, 2 H,
ArCHAHBN), 4.89 (dt, J ) 15.3, 4.6 Hz, 1 H, NCH2CH2N),
4.83-4.73 (m, 1 H, NCH2CH2N), 4.46-4.31 (m, 2 H, NCH2-
CH2N), 4.34 (t, J ) 6.9 Hz, 2 H, NCH2CH2Cl), 4.23 (t, J ) 6.9
Hz, 2 H, NCH2CH2Cl), 4.05 (s, 3 H, NCH3), 3.52 (s, 3 H,
+NCH3); 13C NMR δ 138.8, 138.4, 124.6, 64.8, 55.8, 54.6, 41.5,
49.4, 35.5, 35.3; MS (FAB) m/z 259 [30, M2+ (bisammonium
ion) - H+], 196 (100, M2+ - H+ - CH2CH2Cl). Anal. (C10H17N4-
Cl3O2‚2H2O) C, H, N, Cl.
By the methods described above (for the synthesis of 14-
18), the following compounds were prepared starting from
ethyl 5-formyl-4-nitropyrrole-2-carboxylate27 (23).
Da ta for Eth yl 5-F or m yl-1-m eth yl-4-n itr op yr r ole-2-
ca r boxyla te (24): pale green solid (89%); mp 70.5-71.5 °C

Hypoxia-Selective Antitumor Agents
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 21 3521
Com et Assa y for DNA Cr oss-Lin k in g. A single cell
suspension was obtained from two RIF-1 tumors by enzymatic
dissociation at 37 °C for 30 min using Pronase (0.5 mg/mL),
collagenase (0.2 mg/mL), and DNAase I (0.2 mg/mL). The cells
and drug solution were equilibrated under aerobic or hypoxic
conditions as for the clonogenicity experiments above. After 1
h of exposure to 3.5 µM 2, 2 × 105 cells were washed in cold
phosphate-buffered saline (PBS) by centrifugation and resus-
pended in fresh cold PBS at the same density. Aliquots were
irradiated on ice (10 Gy, cobalt-60, 1.5 Gy/min), and 0.5 mL
samples of irradiated or unirradiated cell suspensions (104
cells) were added to 1.5 mL of a 1% solution of low-gelling-
temperature agarose (Sigma Type VII), mixed, and pipetted
onto a microscope slide. Slides were placed in lysis solution
(30 mM sodium hydroxide, 1 M sodium chloride, 0.1% N-
lauroylsarcosine) for 1 h and then washed for 3 × 20 min in
rinse solution (30 mM sodium hydroxide, 2 mM EDTA),
followed by electrophoresis at 0.5 V/cm for 25 min in a fresh
solution of 30 mM sodium hydroxide and 2 mM EDTA. Slides
were rinsed with distilled water for 15 min and then stained
with propidium iodide (2.5 µg/mL) for 15 min. The DNA from
individual cells was examined using a Nikon epifluorescence
Refer en ces
(1) (a) Brown, J . M. The Hypoxic Cell: A Target for Selective Cancer
Therapy - Eighteenth Bruce F. Cain Memorial Award Lecture.
Cancer Res. 1999, 59, 5863-5870. (b) Brown, J . M. Exploiting
the Hypoxic Cancer Cell: Mechanisms and Therapeutic Strate-
gies. Mol. Med. Today 2000, 6, 157-162.
(2) (a) Brizel, D. M.; Sibley, G. S.; Prosnitz, L. R.; Scher, R. L.;
Dewhirst, M. W. Tumor Hypoxia Adversely Affects the Prognosis
of Carcinoma of the Head and Neck. Int. J . Radiat. Oncol., Biol.,
Phys. 1997, 38, 285-289. (b) Nordsmark, M.; Overgaard, M.;
Overgaard, J . Pretreatment Oxygenation Predicts Radiation
Response in Advanced Squamous Cell Carcinoma of the Head
and Neck. Radiother. Oncol. 1996, 41, 31-39.
(3) (a) Koong, A. C.; Dennko, N. C.; Hudson, K. M.; Schindler, C.;
Swiersz, L.; Koch, C.; Evans, S.; Ibrahim, H.; Le, Q. T.; Terris,
D. J .; Giaccia, A. J . Candidate Genes for the Hypoxic Tumor
Phenotype. Cancer Res. 2000, 60, 883-887. (b) Dachs, G. U.;
Tozer, G. M. Hypoxia Modulated Gene Expression: Angiogen-
esis, Metastasis annd Therapeutic Exploitation. Eur. J . Cancer
2000, 36, 1649-1660.
(4) (a) Graeber, T. G.; Osmanian, C.; J acks, T.; Housman, D. E.;
Koch, C. J .; Lowe, S. W.; Giaccia, A. J . Hypoxia-Mediated
Selection of Cells with Diminished Apoptotic Potential in Solid
Tumours. Nature 1996, 379, 88-91. (b) Ho¨ckel, M.; Schlenger,
K.; Ho¨ckel, S.; Vaupel, P. Hypoxic Cervical Cancers with Low
Apoptotic Index are Highly Aggressive. Cancer Res. 1999, 59,
4525-4528.
(5) (a) Sundfor, K.; Lyng, H.; Rofstad, E. K. Tumour Hypoxia and
Vascular Density as Predictors of Metastasis in Squamous Cell
Carcinoma of the Uterine Cervix. Br. J . Cancer 1998, 78, 822-
827. (b) Rofstad, E. K.; Danielsen, T. Hypoxia-Induced Metasta-
sis of Human Melanoma Cells: Involvement of Vascular En-
dothelial Growth Factor-Mediated Angiogenesis. Br. J . Cancer
1999, 80, 1697-1707.
(6) Haffty, B. G.; Son, Y. H.; Wilson, L. D.; Papac, R.; Fischer, D.;
Rockwell, S.; Sartorelli, A. C.; Ross, D.; Sasaki, C. T.; Fischer,
J . J . Bioreductive Alkylating Agent Porfiromycinn in Combina-
tion with Radiation Therapy for the Management of Squamous
Cell Carcinoma of the Head and Neck. Radiat. Oncol. Invest.
1997, 5, 235-245.
(7) (a) J enkins, T. C.; Naylor, M. A.; O’Neill, P.; Threadgill, M. D.;
Cole, S.; Stratford, I. J .; Adams, G. E.; Fielden, E. M.; Suto, M.
J .; Stier, M. A. Synthesis and Evaluation of R-[[(2-Haloethyl)-
amino]methyl]-2-nitro-1H-imidazole-1-ethanols as Prodrugs of
R-[(1-Aziridinyl)methyl]-2-nitro-1H-imidazole-1-ethanol (RSU-
1069) and its Analogues which are Radiosensitisers and Biore-
ductively Activated Cytotoxins. J . Med. Chem. 1990, 33, 2603-
2610. (b) Cole, S.; Stratford, I. J .; Fielden, E. M.; Adams, G. E.;
Leopold, W.; Elliot, W.; Suto, M.; Sebolt-Leopold, J . Dual
Function Nitroimidazoles Less Toxic Than RSU 1069: Selection
of Candidate Drugs for Clinical Trial (RB 6145 and/or PD
130908). Int. J . Radiat. Oncol., Biol., Phys. 1992, 22, 545-548.
(8) Lee, A. E.; Wilson, W. R. Hypoxia-Dependent Retinal Toxicity
of Bioreductive Anticancer Prodrugs in Mice. Toxicol. Appl.
Pharmacol. 2000, 163, 50-59.
(9) Patterson, L. H.; McKeown, S. R. AQ4N: A New Approach to
Hypoxia-Activated Cancer Chemotherapy. Br. J . Cancer 2000,
83, 1589-1593.
(10) Denny, W. A.; Wilson, W. R. Tirapazamine: A Bioreductive
Anticancer Drug That Exploits Tumour Hypoxia. Expert Opin.
Invest. Drugs 2000, 9, 2889-2901.
(11) Von Pawel, J .; von Roemeling, R.; Gatzemeier, U.; Boyer, M.;
Elisson, L. O.; Clark, P.; Talbot, D.; Rey, A.; Butler, T. W.;
Hirsch, V.; Olver, I.; Bergman, B.; Ayoub, J .; Richardson, G.;
Dunlop, D.; Arcenas, A.; Vescio, R.; Viallet, J .; Treat, J . Tira-
pazamine Plus Cisplatin Versus Cisplatin in Advanced Non-
Small-Cell Lung Cancer: A Report of the International CATA-
PULT I Study Group. J . Clin. Oncol. 2000, 18, 1351-1359.
(12) Rischin, D.; Peters, L.; Hicks, R.; Hughes, P.; Fisher, R.; Hart,
R.; Sexton, M.; D’Costa, I.; von Roemeling, R. Phase I Trial of
Concurrent Tirapazamine, Cisplatin, and Radiotherapy in Pa-
tients with Advanced Head and Neck Cancer. J . Clin. Oncol.
2001, 19, 535-542.
(13) Tercel, M.; Wilson, W. R.; Denny, W. A. Nitrobenzyl Mustard
Quaternary Salts: A New Class of Hypoxia-Selective Cytotoxins
Showing Very High in Vitro Selectivity. J . Med. Chem. 1993,
36, 2578-2579.
(14) Denny, W. A.; Wilson, W. R.; Tercel, M.; Van Zijl, P.; Pullen, S.
M. Nitrobenzyl Mustard Quaternary Salts: A New Class of
Hypoxia-Selective Cytotoxins Capable of Releasing Diffusible
Cytotoxins on Bioreduction. Int. J . Radiat. Oncol., Biol., Phys.
1994, 29, 317-321.
(15) Tercel, M.; Wilson, W. R.; Anderson, R. F.; Denny, W. A.
Hypoxia-Selective Antitumor Agents. 12. Nitrobenzyl Quater-
nary Salts as Bioreductive Prodrugs of the Alkylating Agent
Mechlorethamine. J . Med. Chem. 1996, 39, 1084-1094.
microscope with a 10× objective, attached to a J VC TK-C1360
1
digital
/ in. CCD color video camera and image analysis
2
system using VComet software (Digital Science Consulting,
Auckland). DNA damage was quantitated as the tail moment,
the product of the percentage of DNA in the tail and the
distance between the means of the head and tail distributions.
DNA cross-links were detected as inhibition of radiation-
induced single-strand breaks.
Toxicity a n d An titu m or Activity in Mice. Nitroaryl-
methyl quaternary nitrogen mustards, mechlorethamine, por-
firomycin, AQ4N, and cisplatin were formulated in 0.9% saline,
tirapazamine was formulated in 10% DMSO/saline, and CI-
1010 was formulated in 0.05 M lactate buffer, pH 4. The
compounds were administered as single ip doses at 0.01 mL/g
of body weight. The toxicity in nontumor-bearing mice was
determined using dose increments of 1.33-fold on a fixed scale
with a 28 day observation time. The MTD was defined as the
highest dose causing no deaths in 6 mice and <15% mean body
weight loss on day 5.
RIF-1 cells were maintained by alternating in vivo and in
vitro passage as described.46 Tumors were grown by injecting
105 cells im in 20 µL into the gastrocnemius muscle. Mice were
individually ear-tagged and randomized to treatment groups
when the tumor size reached 0.5 g (leg plus tumor diameter
10 mm). For clonogenic assays, mice received drugs with or
without whole body cobalt-60 radiation (2.5 Gy/min) using the
experimental design illustrated in Figure 3B. Each of the
compounds 1-9 was tested either in one experiment with five
mice per group or in two experiments with three mice per
group, which were then pooled. Tumors were removed by
dissection 18 h after treatment, and dissociated enzymatically
to assess the number of clonogenic cells as described previ-
ously.47 For tumor growth delay assays, tumors were irradiated
by restraining unanesthetized mice in jigs allowing local
irradiation of the leg (cobalt-60, 1.7 Gy/min). Leg diameters
were measured three times per week until the end point (13
mm leg diameter corresponding to ca. 1.2 g of tumor). The
growth delay was calculated as the difference in time to the
end point for treated versus control groups. The statistical
significance of drug effects was evaluated using the log of
clonogens per gram of tumor for the clonogenic assay, and the
time to the end point for the growth delay assay, by ANOVA
followed by Dunnett’s test to determine p values for individual
groups.
Ack n ow led gm en t. This work was carried out with
the support of the Auckland Division of the Cancer
Society of New Zealand, the Health Research Council
of New Zealand, and the Marsden Fund administered
by The Royal Society of New Zealand.

3522 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 21
Tercel et al.
(16) (a) Neta, P.; Behar, D. Intramolecular Electron Transfer in the
Anion Radicals of Nitrobenzyl Halides. J . Am. Chem. Soc. 1980,
102, 4798-4802. (b) Bays, J . P.; Blumer, S. T.; Baral-Tosh, S.;
Neta, P. Intramolecular Electron Transfer and Dehalogenation
of Nitroaromatic Radical Anions. J . Am. Chem. Soc. 1983, 105,
320-324.
(17) Wilson, W. R.; Ferry, D. M.; Tercel, M.; Anderson, R. F.; Dennny,
W. A. Reduction of Nitroarylmethyl Quaternary Ammonium
Prodrugs of Mechlorethamine by Radiation. Radiat. Res. 1998,
149, 237-245.
(18) Anderson, R. F.; Denny, W. A.; Li, W.; Packer, J . E.; Tercel, M.;
Wilson, W. R. Pulse Radiolysis Studies on the Fragmentation
of Arylmethyl Quaternary Nitrogen Mustards by One-Electron
Reduction in Aqueous Solution. J . Phys. Chem. A 1997, 101,
9704-9709.
(19) Wilson, W. R.; Tercel, M.; Anderson, R. F.; Denny, W. A.
Radiation-Activated Prodrugs as Hypoxia-Selective Cytotoxins:
Model Studies with Nitroarylmethyl Quaternary Salts. Anti-
Cancer Drug Des. 1998, 13, 663-685.
(20) Makosza, M.; Bialecki, M. Synthesis of (Nitroaryl)chloromethanes
via Vicarious Nucleophilic Substitution of Hydrogen. Synlett
1991, 181-182.
(21) Newcombe, P. J .; Norris, R. K. Substitution Reactions of ni-
trothiophenes. III. Radical and Ionic Reactions of 4- and 5-Nitro-
2-thienylmethyl and -ethyl Chlorides and Acetates. Aust. J .
Chem. 1979, 32, 2647-2658.
(22) Snyder, H. R.; Carpino, L. A.; Zack, J . F., J r.; Mills, J . F.
Synthesis of the Thieno[3,2-b]pyrrole System. J . Am. Chem. Soc.
1957, 79, 2556-2559.
(33) Displacement by thiol nucleophiles in aqueous buffer has been
investigated in a related series of nitroarylmethyl quaternary
salts in which the leaving group was a cytotoxic amine other
than mechlorethamine. In this series there was no obvious
correlation between the ease of nucleophilic displacement and
toxicity of the prodrug (unpublished results).
(34) Fairbairn, D. W.; Olive, P. L.; O’Neill, K. L. The Comet Assay:
a Comprehensive Review. Mutat. Res. 1995, 339, 37-59.
(35) Olive, P. L.; Banath, J . P. Sizing Highly Fragmented DNA in
Individual Apoptotic Cells using the Comet Assay and a DNA
Cross-linking Agent. Exp. Cell Res. 1995, 221, 19-26.
(36) Patterson, L. H.; McKeown, S. R.; Ruparelia, K.; Double, J . A.;
Bibby, M. C.; Cole, S.; Stratford, I. J . Enhancement of Chemo-
therapy and Radiotherapy of Murine Tumours by AQ4N, a
Bioreductively Activated Anti-Tumour Agent. Br. J . Cancer
2000, 82, 1984-1990.
(37) Dorie, M. J .; Brown, J . M. Tumor-Specific, Schedule-Dependent
Interaction Between Tirapazamine (SR 4233) and Cisplatin.
Cancer Res. 1993, 53, 4633-4636.
(38) Kovacs, M. S.; Hocking, D. J .; Evans, J . W.; Siim, B. G.; Wouters,
B. G.; Brown, J . M. Cisplatin Antitumor Potentiation by Tira-
pazamine Results From a Hypoxia-Dependent Cellular Sensi-
tization to Cisplatin. Br. J . Cancer 1999, 80, 1245-1251.
(39) Wright, D. E. Pyrazole derivatives. U.S. Patent 3,102,890; Chem.
Abstr. 1963, 60, 1762a.
(40) J entzer, O.; Vanelle, P.; Crozet, M. P.; Maldonado, J .; Barreau,
M. New Highly Active 5-Nitroimidazoles Bearing Lactam Nu-
clei: Synthesis and Antibacterial Properties. Eur. J . Med. Chem.
1991, 26, 687-697.
(41) Adebayo, A. T. O. M.; Bowman, W. R.; Salt, W. G. Radical-
Nucleophilic Substitution (SRN1) Reactions. Part 5. Anions of
Nitroimidazoles in SRN1 and Oxidative Addition Reactions. J .
Chem. Soc., Perkin Trans. 1 1987, 2819-2827.
(23) Fournari, P. Research on Heterocyclic Series. VI. Synthesis of
the Nitro Derivatives of N-Methylpyrrole. Application of Ana-
lytical Polarography for the Separation of Nitro Isomers. Bull.
Soc. Chim. Fr. 1963, 488-491.
(24) Cheng, C. C. Pyrazoles. II. Reactions of 1-Methyl-5-amino-4-
pyrazolecarboxamide and Nitrous Acid. Introduction of a Nitro
Group at Position 5 in the Pyrazole Ring. J . Heterocycl. Chem.
1968, 5, 195-197.
(42) van Leeuwen, J . W.; van Duk, C.; Veeger, C. A Pulse Radiolysis
Study of the Reduction of Flavodoxin by Viologen Radicals: A
Conformational Change as a Possible Regulating Mechanism.
Eur. J . Biochem. 1983, 135, 601-607.
(25) Borch, R. F.; Liu, J .; Schmidt, J . P.; Marakovits, J . T.; J oswig,
C.; Gipp, J . J .; Mulcahy, R. T. Synthesis and Evaluation of
Nitroheterocyclic Phosphoramidates as Hypoxia-Selective Alky-
lating Agents. J . Med. Chem. 2000, 43, 2258-2265.
(26) Bernard, M. K.; Makosza, M.; Szafran, B.; Wrzeciono, U. Azoles,
26. Vicarious Nucleophilic Substitution of Hydrogen in Nitro-
pyrazole Derivatives. Liebigs Ann. Chem. 1989, 545-549.
(27) Fournari, P.; Farnier, M.; Fournier, C. Research on Heterocyclic
Series. XIX. Synthesis of Nitropyrrolealdehydes. Bull. Soc. Chim.
Fr. 1972, 283-291.
(43) Steckhan, E.; Kuwana, T. Spectroelectrochemical Study of
Mediators. I. Bipyridylium Salts and their Electron Transfer to
Cytochrome c. Ber. Bunsen-Ges. Phys. Chem. 1974, 78, 253-
259.
(44) Wilson, W. R.; Denny, W. A.; Pullen, S. M.; Thompson, K. M.;
Li, A. E.; Patterson, L. H.; Lee, H. H. Tertiary Amine N-Oxides
as Bioreductive Drugs: DACA N-Oxide, Nitracrine N-Oxide and
AQ4N. Br J . Cancer 1996, 74 (Suppl. XXVII), S43-S47.
(45) Siim, B. G.; Atwell, G. J .; Wilson, W. R. Oxygen Dependence of
the Cytotoxicity and Metabolic Activation of 4-Alkylamino-5-
nitroquinoline Bioreductive Drugs. Br. J . Cancer 1994, 70, 596-
603.
(46) Twentyman, P. R.; Brown, J . M.; Gray, A. W.; Franko, A. J .;
Scoles, M. A.; Kallman, R. F. A New Mouse Tumor Model System
(RIF-1) for Comparison of Endpoint Studies. J . Natl. Cancer Inst.
1980, 64, 595-604.
(47) Siim, B. G.; Laux, W. T.; Rutland, M. D.; Palmer, B. D.; Wilson,
W. R. Scintigraphic Imaging of the Hypoxia Marker 99mTc-HL-
91 (PrognoxTM): Noninvasive Detection of Tumor Response to
the Anti-vascular Agent DMXAA. Cancer Res. 2000, 60,
4582-4588.
(28) Wardman, P. Reduction Potentials of One-Electron Couples
Involving Free Radicals in Aqueous Solution. J . Phys. Chem.
Ref. Data 1989, 18, 1637-1755.
(29) Hansch, C.; Leo, A.; Hoekman, D. Exploring QSAR. Vol 2.
Hydrophobic, Electronic, and Steric Constants; American Chemi-
cal Society: Washington, DC, 1995.
(30) Anderson, R. F.; Tercel, M.; Hay, M. P. H. Unpublished results.
(31) Hoeijmakers, J . H. J .; Bootsma, D. Molecular Genetics of
Eukaryotic DNA Excision Repair. Cancer Cells 1990, 2, 311-
320.
(32) Hoy, C. A.; Salazar, E. P.; Thompson, L. H. Rapid Detection of
DNA-Damaging Agents Using Repair-Deficient CHO Cells.
Mutat. Res. 1984, 130, 321-332.
J M010202L