Mitomycin C linked to DNA minor groove binding agents: Synthesis, reductive activation, DNA binding and cross-linking properties and in vitro antitumor activity

Article abstract of DOI:10.1016/S0968-0896(99)00223-0

Mitomycin C (MC) is a natural cytotoxic agent used in clinical anticancer chemotherapy. Its antitumor target appears to be DNA. Upon bioreductive activation MC alkylates and cross-links DNA. MC derivatives were synthesized in which MC was linked to DNA minor groove binding agents, analogous to netropsin and distamycin. One, two and three N-methylpyrrole carboxamide units were conjugated with MC by a (CH₂)₅-tether to the 7-amino group of MC (11, 12 and 13, respectively). In contrast to MC 11, 12 and 13 displayed non-covalent affinity to DNA. Their bioreductive activation by NADPH-cytochrome c reductase proceeded as fast as that of MC. Metabolites arising from reductive and low-pH activation were characterized and found to be analogous to those of MC. DNA cross-linking activities were weak and decreased with an increasing number of N-methylpyrrole carboxamide units linked with the mitomycin molecule. No adducts were formed with calf thymus DNA in detectable amounts. In vitro antitumor activities of 11-13 were determined using the NCI in vitro antitumor screen. The conjugates 11-13 are growth inhibitory; however, their activities are 1.5-2 orders of magnitude lower than that of MC. COMPARE analysis indicates that the mechanism of the action of 11 and 12 correlates moderately with MC but negatively with distamycin. Conjugate 13 correlates neither with MC nor with distamycin. The results suggest that the basic cause of the observed low activity of the MC-minor groove binder conjugates is the fast irreversible decay of the activated MC, competing effectively with the slow drug delivery to CpG sites, required for the alkylation. Copyright (C) 1999 Elsevier Science Ltd.

Full text of DOI:10.1016/S0968-0896(99)00223-0

Bioorganic & Medicinal Chemistry 7 (1999) 2713±2726
Mitomycin C Linked to DNA Minor Groove Binding Agents:
Synthesis, Reductive Activation, DNA Binding and Cross-Linking
Properties and In Vitro Antitumor Activity
Manuel M. Paz,y Arunangshu Das and Maria Tomasz*
Department of Chemistry, Hunter College, City University of New York, New York, NY 10021, USA
Received 21 April 1999; accepted 22 June 1999
AbstractÐMitomycin C (MC) is a natural cytotoxic agent used in clinical anticancer chemotherapy. Its antitumor target appears to
be DNA. Upon bioreductive activation MC alkylates and cross-links DNA. MC derivatives were synthesized in which MC was
linked to DNA minor groove binding agents, analogous to netropsin and distamycin. One, two and three N-methylpyrrole car-
boxamide units were conjugated with MC by a (CH2)5-tether to the 7-amino group of MC (11, 12 and 13, respectively). In contrast
to MC 11, 12 and 13 displayed non-covalent anity to DNA. Their bioreductive activation by NADPH-cytochrome c reductase
proceeded as fast as that of MC. Metabolites arising from reductive and low-pH activation were characterized and found to be
analogous to those of MC. DNA cross-linking activities were weak and decreased with an increasing number of N-methylpyrrole
carboxamide units linked with the mitomycin molecule. No adducts were formed with calf thymus DNA in detectable amounts. In
vitro antitumor activities of 11±13 were determined using the NCI in vitro antitumor screen. The conjugates 11±13 are growth
inhibitory; however, their activities are 1.5±2 orders of magnitude lower than that of MC. COMPARE analysis indicates that the
mechanism of the action of 11 and 12 correlates moderately with MC but negatively with distamycin. Conjugate 13 correlates nei-
ther with MC nor with distamycin. The results suggest that the basic cause of the observed low activity of the MC±minor groove
binder conjugates is the fast irreversible decay of the activated MC, competing e€ectively with the slow drug delivery to CpG sites,
required for the alkylation. # 1999 Elsevier Science Ltd. All rights reserved.
lesions occur only after reductive activation of MC,
Introduction
mediated in vivo by various ¯avoreductases. This pro-
cess can be mimicked in cell-free systems by puri®ed
reductase or chemical reducing agents.1 On account of
this property, MC is regarded as the prototype bio-
reductive antitumor agent.2 Members of this class are
hypoxia-selective cytotoxins; i.e. more cytotoxic under
hypoxic than anaerobic conditions of drug exposure.
Bioreductive drugs are also selectively active against
hypoxic regions of solid tumors. In contrast, such
regions are relatively resistant to treatment with radia-
tion or common chemotherapeutic agents since these
treatments require the presence of oxygen for cell kill-
ing.3 Synthetic bioreductive drugs have been developed
for postoperative treatment of solid tumors used in
conjunction with treatment with radiation.4 Further-
more both a combination of MC or por®romycin (N1a-
methyl MC) with radiation are treatment protocols
currently undergoing large-scale clinical evaluation.5,6
Mitomycin C (MC; 1) is a natural antitumor antibiotic,
used in the clinic in anticancer chemotherapy. Sub-
stantial evidence indicates that its direct molecular
target in tumor cells is DNA. Damage to DNA is gen-
erated by mono- and bifunctional alkylation of guanine
residues by MC, leading to MC±guanine monoadducts
(2, 3) and MC±guanine bisadducts; the latter constitute
DNA interstrand and intrastrand cross-links (4, 5)
which are formed speci®cally at CpG and GpG sequen-
ces, respectively (Scheme 1).1 MC is the only natural
antitumor antibiotic known to cross-link DNA in the
cell. Both the monofunctional and bifunctional DNA
Keywords: Antitumor activity; COMPARE; DNA; mitomycin C;
mitomycin-lexitropsin conjugates.
Abbreviations: MC, mitomycin C; dT, 20-deoxythymidine; MC±Py1,
MC±monopyrrole (11); MC±Py2, MC±bispyrrole (12); MC±Py3, MC±
trispyrrole (13).
* Corresponding author. Tel.: +1-212-772-5387; fax: +1-212-650-
3501; e-mail: tomasz@patsy.cuny.edu
Current address: Department Quimica Organica, Facultade de Cen-
cias, Universidade de Santiago de Compostela, Campus de Lugo,
27002 Lugo, Spain.
Recent studies of the mechanism of the reductive acti-
vation of MC indicated that, in addition to its reaction
with DNA, the unstable active form decomposes rapidly
y
0968-0896/99/$ - see front matter # 1999 Elsevier Science Ltd. All rights reserved.
PII: S0968-0896(99)00223-0

2714
M. M. Paz, et al. / Bioorg. Med. Chem. 7 (1999) 2713±2726
Scheme 1. Mitomycin C±deoxyguanosine adducts.
to inactive metabolites, mainly to 2,7-diaminomitosene
(6; Scheme 2).7±9 This reductive inactivating pathway
also prevails in tumor cells,10,11 diminishing the in vivo
potency of MC in its alkylation and cross-linking of
cellular DNA. The partitioning of the active form of
MC into reaction with DNA versus conversion to inac-
tive metabolites depends on the kinetic competition of
the two processes.7,8,11 In contrast to most natural pro-
ducts which interact with DNA, MC has no DNA
binding anity until after its conversion to activated
mitosene forms.12,13 Therefore, it is likely that much
of the MC that enters the cell is activated at intracel-
lular locations which are not in proximity to DNA.
Flavoreductases that activate MC are present in the
cytoplasm, microsomal fraction and mitochondria, as
well as in the nucleus.14±16 Consequently in non-nuclear
locations the active form is expected to decay mainly to
inactive metabolites. As consistent with these notions,
MC is approximately 2±3 orders of magnitude less
potent than CC-1065 or natural enediynes, both of
which bind DNA prior to their activation.17
In order to construct more potent mitomycins a rational
approach may be to append MC with pre-activation
DNA-binding capacity. Accordingly, we designed a
series of hybrid drugs in which MC is linked to the oligo

M. M. Paz, et al. / Bioorg. Med. Chem. 7 (1999) 2713±2726
2715
Scheme 2. Formation of reductive metabolite 6 from activated MC.
N-methylpyrrole carboxamide framework, present in
netropsin and distamycin, two well-characterized minor
groove binders with (A/T)n sequence speci®city.18 We
hypothesized that on account of their anity to DNA
the MC±pyrrole carboxamide hybrids will be rapidly
attracted to the cell nucleus, in proximity to DNA.
Reductive activation would then occur near their DNA
targets, kinetically facilitating the reactions of the acti-
vated form with DNA. This would make MC a more
ecient DNA damaging agent. A successful model for
this scenario in vitro is the cross-linking of DNA by 2,3-
bis(hydroxymethyl)pyrrole (7; Chart 1). The cross-link
(8) is structurally closely analogous to that involving
MC (9; see also 4) and favors the same sequence,
CpG.19 When this agent was conjugated to a distamycin
derivative (10) the yield of DNA cross-linking increased
1000-fold. In addition, the conjugate was quite cyto-
toxic, in contrast to the marginally toxic parent compo-
nents.19 Although the selective binding of the
distamycin moiety to A/T sequences may have reduced
the number of CpG cross-link sites available to the
conjugate relative to free 2,3-bis(hydroxymethyl)-
pyrrole, the increased eciency of the cross-linking
reaction at those sites apparently over-compensated for
the decrease in the number of cross-link target sites.
The similarly enhanced cytotoxicity and cross-linking
Chart 1. 2,3-Bis(hydroxymethyl)pyrrole and analogous geometry of its DNA cross-link to the geometry of the MC±DNA cross-link.

2716
M. M. Paz, et al. / Bioorg. Med. Chem. 7 (1999) 2713±2726
eciency of a psoralen±distamycin conjugate as com-
pared to free psoralen further substantiated the success
of this type of design.20
Results
Synthesis of mitomycin C±N-methylpyrrole carboxamide
conjugates 11, 12 and 13 (Scheme 4)
Several conjugates of mustard-type cross-linking agents
and distamycin analogues have been synthesized and
investigated for activity.21 The most notable of these,
tallimustin, a benzoic acid mustard±distamycin hybrid,
is under clinical trial as an anticancer drug. Structure±
activity relationships of such hybrids, however, are not
well understood. For example, tallimustin, unlike the
parent mustard, does not generate DNA cross-links.22,23
The key step of linking the chemically fragile mitosane
moiety25 to the N-methylpyrrole carboxamides was
based on the well-precedented facile displacement of the
7-methoxy group of mitomycin A (14) by primary
amines (Scheme 3).26 The requisite primary amine deri-
vatives of the mono-, bis- and tris-N-methylpyrrole car-
boxamides (21, 22 and 23) were synthesized starting
from the known27 nitropyrroles 15, 16 and 17, respec-
tively. The nitro group was reduced to the amino group
by catalytic hydrogenation. This was followed by in situ
acylation of the primary amine by the N-t-BOC-pro-
tected 6-aminocaproic acid, activated by carbonyldiimi-
dazole, to give 18, 19 and 20. Subsequent removal of the
t-BOC groups by TFA yielded 21, 22 and 23. Incuba-
tion of each with mitomycin A (14) in methanol for 16 h
at room temperature yielded the desired MC±mono-,
bis- and tris-N-methylpyrrole carboxamide conjugates
11, 12 and 13, respectively, in excellent yields. All new
structures were rigorously veri®ed by HRMS or MS,
An important di€erence of MC from other DNA
cross-linking and alkylating agents is that MC requires
enzymatic reductive activation and the active form
decomposes very rapidly. Due to lack of appropriate
precedent it was not possible to predict whether activa-
tion of MC would occur in its conjugate forms, bound
or unbound to DNA. In an e€ort to determine experi-
mentally the merits of the design of MC-oligo-N-
methylpyrrole carboxamide conjugates as cytotoxic
antitumor agents, we synthesized a series of compounds
consisting of one, two and three N-methylpyrrole car-
boxamide units linked to the MC moiety by a ¯exible
linker (Chart 2; 11±13). Bioreductive activation and
DNA binding and cross-linking properties of the con-
jugates in comparison with free MC were determined.
In addition, the in vitro antitumor activity of each of the
new agents 11±13 was evaluated against a panel of 60
human tumor cell lines of the National Cancer Institute.24
1
and H NMR.
Proof of purity of 11, 12 and 13
HPLC of the samples showed single peaks as detected
by UV absorbance at 254 nm (elution times: 11, 22 min;
1
12, 24 min; 13, 26 min). H NMR spectra indicated no
impurities in any of the compounds. Rigorous proof
Chart 2. Mono-, bis- and tris-N-methylpyrrole carboxamide conjugates of MC.

M. M. Paz, et al. / Bioorg. Med. Chem. 7 (1999) 2713±2726
2717
Scheme 3. Reaction of mitomycin A with primary amines.
Scheme 4. Synthesis of MC±N-methylpyrrole carboxamide conjugates 11, 12 and 13.
was provided by determining the molar extinction coef-
®cients (e) of 11±13 at 374 nm, corresponding to a lmax
of the MC chromophore. At this wavelength the
methylpyrrole carboxamide moiety has negligible
absorption, as seen in the UV spectra of 21, 22 and 23
(data not shown). The e374 values of 11±13 were very
similar to that reported for MC (Table 1). Purity in
Reductive enzymatic activation of 11±13 in the absence
or presence of DNA
In order to assess whether the modi®ed mitomycin
moieties of 11, 12 and 13 were substrates of a MC-acti-
vating enzyme the rate of reduction of 11, 12 and 13 and
MC by NADH-cytochrome c reductase in neutral aqu-
eous bu€er was measured under identical assay condi-
tions. The rate of reduction of MC can be assayed
conveniently by the disappearance of MC from the
reaction mixture followed by HPLC at various times.30
The same method was employed here for 11±13. The
results show that all three conjugates 11±13 were
reduced by NADH-cytochrome c reductase and NADH
at rates similar to that observed for MC itself (Fig. 1a).
1
HPLC, the e374 values and the H NMR spectra toge-
ther constitute reliable proof of purity of 11±13.
Non-covalent binding anity of 11±13 to calf thymus
DNA
Using the EtBr displacement assay28 as in previous
DNA binding studies of drug±lexitropsin conjugates,29
it is shown that the bis- and tris-pyrrole MC conjugates
12 and 13 bound to DNA in the micromolar range;
binding by the mono-pyrrole 11 was approximately 100-
fold weaker. The unconjugated MC showed no binding,
as expected12 (Table 1).
The same method was employed for assaying the
reduction rate when DNA was present in the reaction
mixture. The procedure was more complex, however,
since the DNA had to be enzymatically hydrolyzed

2718
M. M. Paz, et al. / Bioorg. Med. Chem. 7 (1999) 2713±2726
Table 1. Molar extinction coecients at 374 nm (e274) and apparent
binding constants (Kapp) to calf thymus DNA of 11, 12, 13, MC and
distamycin
reduction by Na2S2O4 and NADH-cytochrome c
reductase, respectively. All activation conditions resul-
ted in the disappearance of the starting materials and in
the formation of several metabolites. Characterization
of the metabolites by their UV and mass spectra (data
not shown) indicated a similar chemistry to that of
MC:7,32,33 the acidic treatment or the enzymatic reduc-
tion produced metabolites 24±26, containing the inac-
tive mitosene moiety; enzymatic activation (NADH-
cytochrome c reductase) produced, in addition, the 2,7-
diaminomitosene derivatives 27±29. The chemical
reduction with sodium dithionite gave additional meta-
bolites resulting from bifunctional activation, i.e. acti-
vation of both the C-1 and C-10 functions of the MC
residue (30±35). In addition to the expected metabolites,
the activation of 11, 12 and 13 produced 21, 22 and 23,
respectively, under all activation conditions indicating
dissociation of the pyrrole±mitomycin linkage. These
compounds were identi®ed by their spectroscopic data
and by comparison to authentic samples.
1
1
1
e
374 (M cm
(in H2O)
)
Kapp (mM )
11
12
13
MC
18,900
17,500
18,100
0.036 (‹0.009)
1.29 (‹0.07)
3.87 (‹0.68)
No binding12
0.7744
21,840 (365 nm)
Distamycin
before the HPLC analysis at each time point, in order to
release all remaining starting material (11±13) from its
DNA-bound form. The results indicated that the
reduction proceeds with equal eciency both in the
presence and absence of DNA (Fig. 1b).
Products of acidic and reductive activation of the
conjugates 11±13 (Chart 3)
Cross-linking of DNA by the conjugates 11±13 in a
cell-free system
Diode-array HPLC and LCMS were used to study the
formation of metabolites after acid treatment of con-
jugates 11±13 or after their chemical or enzymatic
pBR322 plasmid DNA linearized by cleavage by the
restriction enzyme EcoRI served as the substrate for
testing the cross-linking activity of the conjugates and
MC. The assay method of Hartley and co-workers31
was employed. According to this method, linearized
DNA is 30-end-labeled by 32P-phosphate, treated with
the cross-linking agent, and subsequently denatured by
heating. On cooling, the cross-linked DNA undergoes
quick renaturation on account of the covalent linkage(s)
between the two strands, while non-cross-linked DNA
remains single stranded. The renatured and single-
stranded fractions are separated by non-denaturing
agarose gel electrophoresis and quantitated by phos-
phorimagery. The method was used for mitomycin and
its derivatives previously.30 DNA cross-linking by MC
requires a reducing agent for activation of the drug;
Na2S2O4 was employed under anaerobic conditions in
the present case. The results (Fig. 2) show that all three
conjugates 11±13 cross-link pBR322 DNA. The mono-
pyrrole conjugate 11 is more active than MC while 12,
and especially 13, are less so. At 100 mM drug con-
centration MC and 11 cause 100% cross-linking, while
12 and 13 cross-link 60 and 18% of the DNA, respec-
tively. At >10 mM concentration the eciency of 12
and especially of 13 decline relative to that of MC.
Lack of formation of DNA adducts
MC±pyrrole conjugates 11±13 were activated with either
sodium dithionite, NADH-cytochrome c reductase/
NADH or catalytic hydrogenation in the presence of
calf thymus DNA. Non-covalently bound drug meta-
bolites were removed by dialysis or successive extrac-
tions with n-butanol,34 until the absorbance at 320 nm
remained constant. The DNA was then enzymatically
digested and analyzed by HPLC. The only signi®cant
peak, in addition to the expected nucleosides, corre-
sponded to the metabolites 21, 22 and 23 (Fig. 3b and
c). In positive control experiments MC was employed
Figure 1. Reduction of MC and MC±Pyn conjugates 11±13 by
NADPH-cytochrome c reductase. (a) Time course of the reduction in
the absence of DNA: circles, MC; squares, MC±Py1 (11); triangles,
MC±Py2 (12); rhombus, MC±Py3 (13). (b) Comparison of the extent of
reductions in the absence (light bars) and presence (dark bars) of
DNA at 8 min [data from (a)].

M. M. Paz, et al. / Bioorg. Med. Chem. 7 (1999) 2713±2726
2719
Chart 3. Metabolites formed after chemical or enzymatic activation of 11±13.
under identical activating conditions. Deoxyguanosine±
MC adducts were readily detectable in the HPLC assay1
(e.g. Fig. 3a). Three oligodeoxynucleotides having
sequences that are methylpyrrole carboxamide binding
sites ¯anked by CpG19 (MC target) were also tested as
substrates. These were the following: [d-(ATATACG-
TATAT)]2, [d-(GATCGAATATTCGATC)]2 and [d-
(GATCGAATTCGATC)]2 (Chart 4). Reactions were
run with MC as positive controls. The reaction mixtures
were analyzed by DPAGE. Monoalkylated and cross-
linked oligonucleotides are known to migrate slower
than unmodi®ed controls under such conditions in gen-
eral. However, only the reactions with MC and, to a
lesser extent, with 11 led to the detection by UV sha-
dowing of slower migrating bands (data not shown).
orders of magnitude. The same conclusion can also be
drawn when data are compared for individual cell lines
(data not shown). The measured cytotoxicity of the
substances (LC50) (data not shown) also leads to the
same conclusion. The activities of 11, 12 and 13 have
the same order of magnitude as those of distamycin
(Table 2).
In order to discern the mechanism of action of the MC±
minor groove binder hybrid agents 11±13 the results of
the tests were evaluated by the COMPARE algorithm.35
This program is used to calculate the pattern of varia-
tion of the inhibitory activity of a drug within a series of
cell lines. Patterns generated by di€erent drugs are then
compared with one another. The extent of similarity is
expressed by a correlation value, the Pearson coecient,
which is the measure of the extent of a correlation with
the ``seed'' compound for which the Pearson coecient,
is taken as 1. Such correlation values have been shown
to be indicators of the degree of similarity of the
mechanism of action between two drugs. In the present
work, correlations were computed between (i) the GI50
of MC and the hybrid agents, (ii) the GI50 of distamycin
and the hybrid agents, and (iii) the GI50 of the hybrid
agents within the series 11±13. The results are shown in
Table 3. They indicate, ®rst of all, that there is no cor-
relation between the mechanism of action of MC and
distamycin, as expected. The hybrid drugs 11 and 12, on
the other hand, show a moderate correlation with MC
(0.490 and 0.589) but no correlation with distamycin.
Hybrid 13 does not correlate with MC (0.163) and
distamycin (0.203) even though the 3-pyrrole DNA
binding unit of 13 is essentially identical structurally to
In vitro antitumor activities of the mitomycin C±N-
methylpyrrole carboxamide conjugates 11, 12 and 13
Each substance was evaluated for in vitro antitumor
activities using a panel of over 60 di€erent human can-
cer cell lines derived from 9 di€erent tissues. This panel
of cell lines, known as the in vitro antitumor screen of
the National Cancer Institute, has been developed and
operated by the NCI and has been used to test over
10,000 substances per year for growth inhibition and
cytotoxicity.24 The mean of the 50% growth inhibitory
activities (GI50) in individual cell lines observed for 11±
13 are given in Table 2, together with those of mitomy-
cin C and distamycin, determined previously in the
same screen. Netropsin has not been tested. The data
show that 11, 12 and 13 are cytotoxic; however, they are
less active overall than MC, by approximately two

2720
M. M. Paz, et al. / Bioorg. Med. Chem. 7 (1999) 2713±2726
Figure 2. Comparison of the cross-linking eciency of pBR322 DNA
by MC and 11±13 under activation by Na2S2O4. (a) Phosphorimaging
of agarose gels showing cross-linking of 30-32P-labeled pBR322 DNA.
Concentration of drug is noted at the bottom of each lane (see
Experimental for details). (b) Percentage cross-link as a function of the
concentration, as measured by phosphorimagery. Results shown are
the average of at least three di€erent experiments. Key: circles, MC;
squares, 11; triangles, 12; rhombus, 13.
distamycin. The mechanisms of the three hybrid agents
moderately correlate with one another (Table 3, last 3
columns).
To augment the results of the NCI testing, agents 11±13
were tested for cytotoxicity in a mouse mammary tumor
cell line, under both hypoxic and aerobic conditions. No
cytotoxicity was observed uponꢀ treatment with up to
5
40 mM (4Â10 M) drug (2 h, 37 C) under either condi-
tion in this cell line.36
Discussion
Figure 3. HPLC of enzymatic digests of calf thymus DNA after reac-
tions with MC, 11, 12 or 13 under activation by Na2S2O4. A mixture
of calf thymus DNA (5 mM) and drug (1 mM) was treated with
Na2S2O4 (1.5 mM) under anaerobic conditions. Following puri®cation
(see text for details) the DNA was digested with P1 nuclease/phos-
phodiesterase I/alkaline phosphatase and analyzed by reverse-phase
HPLC. (a), MC; (b), 11; (c), 12; (d), 13.
It has been a general experience that cytotoxic DNA-
reactive agents become more cytotoxic when conjugated
to minor groove binders. Examples21 are conjugates of
the bifunctional alkylator 2,3-bis(hydroxymethyl)pyrrole
(7)19, CPI-type minor groove-alkylating cytotoxins,
duocarmycin, anthramycin, DNA cleaving enediynes,

M. M. Paz, et al. / Bioorg. Med. Chem. 7 (1999) 2713±2726
2721
veri®ed that even a single pyrrole carboxamide unit (in
11) confers anity to DNA, and the anity increases
with addditional pyrrole units (Table 1), in common
with the binding properties of other alkylator con-
jugates.39 Enzymatic reduction of the MC quinone of
the conjugates, required for the DNA alkylation and
cross-linking activities, proceeded smoothly (Fig. 1).
Apparently, the presence of the large substituents in the
7-amino group of the MC moiety did not inhibit reduc-
tion by NADPH-cytochrome c reductase, a ¯avoenzyme
implicated in the activation of MC in tumor cells in
vivo.15 This has been a general experience, even with
potent N7-substituted MC analogues.40 Therefore a lack
of activation is not likely to account for the low biolo-
gical activity. The products resulting from acidic and
reductive activation (24±35) formed in the absence of
DNA were analogous to those formed from MC under
the same conditions, indicating that the pyrrole carbox-
amide moiety did not in¯uence the metabolic transfor-
mations of MC to stable, inactive mitosene type
substances. The conjugate linkage showed some
instability upon acidic or reductive activation, as evi-
denced by the isolation of 21±23 from the appropriate
reaction mixtures. However, this dissociation occurred
only to the extent of 10±15% under either condition. In
neutral aqueous solution at room temperature in the
absence of activation the conjugates 11±13 were stable
for 16 h. Thus instability of the conjugate linkage can-
not account for the low biological activity of 11±13.
Chart 4. Synthetic oligodeoxyribonucleotides tested as substrates of
alkylation and cross-linking by 11±13.
and even psoralen, which reacts with DNA in the major
groove. Similarly, positive results were obtained in the
benzoic acid N-mustard (BAM) and chlorambucil mus-
tard (CHL) series.34,37 In most cases the cytotoxicity
increased with the increasing number of pyrrole37 or
imidazole units.38 In light of this experience, the low
biological activity of the MC±oligopyrrole carboxamide
conjugates 11±13 in the human tumor cell panel of the
NCI was unexpected. The results showed clearly that
the attachment of minor groove binders to MC failed to
increase the cytotoxicity of the conjugates relative to
MC. On the contrary, the cytotoxicities decreased 50-
to 100-fold and even the most active conjugate (13) was
only 2-fold more cytotoxic than the minor groove bin-
der distamycin itself (Table 2).
We attempted to determine why this popular and gen-
erally successful design aiming at increased cytotoxicity
was apparently ine€ective in the case of MC. First, the
MC conjugates 11±13 were examined for their intended
increased binding anity relative to MC. The results
A conspicuous di€erence in behavior from MC, the
parent cytotoxin, however, is a lack of ability of 11±13
to form DNA adducts. MC generates covalent calf thy-
mus DNA adducts after acidic or reductive activation,
readily detectable by HPLC of the digest of the DNA.
In contrast, the conjugates 11±13 yielded no detectable
adducts of DNA under a variety of conditions. The
digests contained unmodi®ed nucleosides, and only one
additional component, corresponding to 21, 22 or 23,
respectively; that is the activation-induced decomposi-
tion products from 11, 12 and 13. The presence of 21±23
in the digests is probably due to their tenacious non-
covalent binding to DNA, resistant even to extraction
with n-butanol. The lack of covalent adduct formation
with the conjugates was con®rmed by testing a variety
of self-complementary oligonucleotides as substrates,
including those known to be excellent substrates of 10.19
Table 2. Growth inhibitory properties of the MC±mono-, bis- and
tris-N-methylpyrrole carboxamide conjugates 11, 12 and 13 in com-
parison with MC and distamycin as determined in the National Can-
cer Institute tumor cell line panel24
a
Compound (NCI test
reference number)
GI50 (mean graph
midpoint), M
7
Mitomycin C (S26980A)
11 (S702522A)
12 (S702523A
13 (S702524A)
Distamycin (S82150A)
8.32Â10
5
8.71Â10
5
6.71Â10
5
4.27Â10
5
8.91Â10
a
GI50: drug concentration (M) causing 50% inhibition of growth.
``Mean graph midpoint'' was calculated from experimental values of
GI50 of the approximately 60 individual cell lines constituting the test
panel and it corresponds closely to their mean.
These results suggest that the origin of the low cyto-
toxic activity of 11±13 relative to MC lies in their lower
Table 3. Pearson correlation of the pattern of growth inhibitory properties (GI50) of MC±N-methylpyrrole carboxamide conjugates 11±13 with
those of MC and distamycina
Compound (NCI test reference number)
Pearson correlation coecient comparing GI50
MC
Distamycin
11
12
13
Mitomycin C (S26980A)
11 (S702522A)
12 (S702523A)
13 (S702524A)
Distamycin (S82150A)
1.000
0.490
0.589
0.163
0.007
0.007
0.160
0.103
0.203
1.000
0.490
1.000
0.676
0.582
0.160
0.589
0.676
1.000
0.404
0.103
0.163
0.582
0.404
1.000
0.206
a
Calculated by the COMPARE program using the log GI50 patterns.24

2722
M. M. Paz, et al. / Bioorg. Med. Chem. 7 (1999) 2713±2726
eciency of covalent alkylation and cross-linking of
DNA. This, in turn, may be due to a unique di€erence
in the mechanism of MC from that of the other alky-
lating agents discussed above: the extremely short life of
the active alkylating species.41 The non-covalent bind-
ing of the activated MC conjugate to DNA at non-
cross-linkable sites may present a kinetic disadvantage
for the covalent reactions by inhibiting the movement of
this species toward CpG sequences, required for alkyla-
tion and cross-linking. Consequently, the fast irrever-
sible MC inactivation processes, e.g. hydrolysis and
conversion to 2,7-diaminomitosene (Scheme 2),8 will
compete e€ectively with the site-speci®c guanine alkyla-
tion. Furthermore, the inactivated conjugate will inhibit
binding of the active conjugate. No such inhibitory
e€ects are expected in the case of the simpler, successful
model 10, since its alkylating function (7) is stable in
contrast to the MC conjugates. Indeed, 10 has enhanced
alkylation and cytotoxic activity compared to the
unconjugated drug, 7. Enhancement may also be
expected in cases where the covalent reaction is anity
driven rather than base- and sequence-speci®c, as was
found to be the case of enediyne±distamycin conjugates.21
Further indication that the mode of action of the MC±
methylpyrrole carboxamide conjugates 11 and 12 is
related to that of MC rather than distamycin is given by
the results of COMPARE, which gave moderate Pear-
son coecients in relation with MC (0.490 and 0.589
respectively) (Table 3). No correlation was obtained
with 13, however, which may be related to the observed
poor cross-linking activity of 13 relative to 11 and 12.
In analogy to the present ®ndings, recently another
group of bioreductive antitumor agents, the nitrodea-
za¯avins, were reported to show decreased cytotoxicity
upon their conjugation to methylpyrrole carboxamide
minor groove binders.43 Like in the case of the MC
conjugates, their cytotoxicity to tumor cells remained
low or decreased with increasing number of pyrrole
units conjugated to the nitrodeaza¯avin moiety. How-
ever, the DNA cleaving eciency in the in vitro pBR322
plasmid test system increased in the conjugated forms,
in further analogy to the present case. These agents, like
MC, are reductively activated to form DNA-reactive
species, which, alternatively, decay to inactive metabo-
lites irreversibly. Although the mechanism involves
enzymatic reduction of the nitro group details of these
processes are not yet known. It is tempting to speculate
that the negative e€ect of the minor groove binding
moieties on the cytotoxicity of nitrodeaza¯avin is
caused by a similar kinetic inhibition, proposed above
for the MC±minor groove conjugates 11±13.
An alternative cause of the lack of appreciable reactivity
of the MC conjugates with DNA may be considered,
namely that the MC±pyrrole tether of the minor groove-
bound conjugate is not ¯exible enough to allow MC to
orient properly for the reaction with guanine at the
CpG site. Nevertheless, quantitative inspection of the
NMR-derived structure of an oligonucleotide duplex
cross-linked by 10 at the 50-CGAATT sequence42 sug-
gests that the (CH2)5 tethers used in our investigation
should be long enough to allow sucient ¯exibility for
the analogous alignment of the MC cross-link formed
by the conjugates 11±13.
Summary and Conclusion
Conjugation of MC to mono-, bis- and tris-methylpyr-
role carboxamide units resulted in hybrid drugs which
were less cytotoxic than MC. In vitro studies identi®ed a
likely basis for the reduced cytotoxicity of these agents:
a lower eciency of DNA alkylation and cross-linking.
As a cause, we propose that binding of the hybrid
agents at non-alkylatable sites of DNA slows down the
approach of the short-lived reductively activated MC to
its required sequence-speci®c (CpG) target sites. Fur-
thermore, conjugate molecules bearing an irreversibly
inactivated MC residue may themselves compete with
the active conjugates for the target sites. Our results and
analysis of analogous precedents43 suggest that con-
jugation of A/T-speci®c minor groove binders with
prodrugs which are enzymatically activated to an irre-
versibly short-lived site-speci®c reactive agent con-
stitutes an ine€ective design of DNA-targeted cytotoxic
drugs. However, conjugation of MC to non-speci®c
DNA binding groups may lead to enhanced potency as
desired. This possibility is under investigation in our
laboratory.
Employing a sensitive assay of cross-link formation
using pBR322 DNA as the substrate31 we obtained evi-
dence for DNA cross-linking capability of the three
conjugates. This result is reconcilable with the negative
results observed using calf thymus DNA or short oligo-
nucleotides when one considers the much greater sensi-
tivity of the pBR322 DNA assay: an average of one
cross-link per 5000 basepairs scores as 100% cross-
linked DNA, compared to one per only 10 basepairs in
a 10-mer duplex oligonucleotide substrate. The positive
results with pBR322 DNA demonstrate that the MC
moiety of the conjugates is capable of cross-linking
DNA, albeit at low eciency. This suggests that the
cytotoxicity of 11±13 is related to this activity. The
cross-linking activity declines with increasing number of
methylpyrrole units (Fig. 3) supporting the above
hypothesis that the stronger the AT-selective binding is
to DNA, the less likely it is that the activated MC
®nds its CpG target before it gets inactivated. It is
hard to explain why the parent MC was less active in
this assay than the monopyrrole conjugate 11. It is pos-
sible that some of the CpG-containing sequences of
pBR322 DNA are uniquely reactive with the con-
jugated mitomycins, in analogy to the high reactivity
found of a unique short sequence of pBR322 DNA with
tallimustin.22
Experimental
Materials and methods
Mitomycin C and mitomycin A were received from
Kyowa Hokko Kogyo Co., Ltd., Tokyo, Japan. Other
materials and their sources were as follows: synthesis

M. M. Paz, et al. / Bioorg. Med. Chem. 7 (1999) 2713±2726
2723
reagents, Aldrich. EcoRI, pBR322 and Klenow frag-
ment of DNA Polymerase I (exo-), New England Bio-
labs. a-32P-Labeled nucleotide triphosphates, NEN;
agarose (DNA grade) and HPLC solvents, Fisher; calf
thymus DNA and NADPH-cytochrome c reductase,
Sigma; DNase I (code D), phosphodiesterase I, and
an analogous procedure. 18: 1H NMR (Cl3CD) d (ppm)
1.27±1.50 (m, 4H), 1.43 (s, 9H), 1.61±1.76 (m, 4H),
2.21±2.26 (m, 2H), 2.29 (s, 6H), 2.46 (t, 2H, J=6.3 Hz),
3.06 (q, 2H, J=6.4 Hz), 3.40 (q, 2H, J=5.8 Hz), 3.85 (s,
3H), 4.78 (br s, 1H), 6.47 (d, 1H, J=1.7 Hz), 7.11 (d,
1H, J=1.6 Hz), 7.73 (t, 1H, J=4.8 Hz), 8.34 (s, 1H).
+
1
.
HRMS calcd for C22H40N5O4 (18 H ) 438.3080, found
alkaline phosphatase, Worthington. H NMR spectra
1
were recorded in GE 300 MHz or Jeol JNM GX400
spectrometers. UV spectra were recorded in a Cary 3
UV±Visible spectrophotometer. Absorbance readings
were performed in a Gilford 250 spectrophotometer.
LCMS was performed with a Hewlett±Packard Series
1100 diode array HPLC system connected to a Hewlett±
Packard Series 1100 MSD mass spectrometer. HRMS
was provided by Richard Milberg, Chemical Sciences,
University of Illinois, Urbana-Champaign. HPLC was
performed in a Beckman System Gold 125 instrument,
equipped with a diode array detector System Gold 168,
®tted with a Rainin C-18, 5Â250 mm column. A linear
gradient of 6±50% CH3CN in 30 mM NaH2PO4, pH
5.8, was used. Agarose gels were run on a DNA Sub
Cell apparatus (Biorad). Agarose gels were prepared by
dissolving 1 g of agarose in 100 mL of bu€er containing
40 mM Tris±AcOH, 2 mM EDTA, pH 8.2. After elec-
trophoresis, gels were transferred onto sequencing ®lter
paper (Biorad), covered with Saran-Wrap, and dried at
80ꢀC in a Biorad model 583 gel dryer. Phosphorimagery
438.3080. 19: H NMR (Cl3CD) d (ppm) 1.22±1.38 (m,
4H), 1.43 (s, 9H), 1.59±1.75 (m, 4H), 2.25±2.30 (m, 2H),
2.26 (s, 6H), 2.42 (t, 2H, J=6.4 Hz), 3.05 (q, 2H,
J=6.6 Hz), 3.41 (q, 2H, J=5.9 Hz), 3.86 (s, 6H), 4.77
(br s, 1H), 6.51 (d, 1H, J=1.4 Hz), 6.67 (d, 1H,
J=1.2 Hz), 7.03 (d, 1H, J=1.5 Hz), 7.17 (d, 1H,
J=1.5 Hz), 7.66 (t, 1H, J=4.8 Hz), 8.19 (s, 1H), 8.29 (s,
+
.
1H). HRMS calcd for C28H47N7O5 (19 H ) 560.3560,
1
found 560.3569. 20: H NMR (Cl3CD) d (ppm) 1.23±
1.31 (m, 2H), 1.40±1.45 (m, 2H), 1.41 (s, 9H), 1.58±1.73
(m, 4H), 2.23 (t, 2H, J=8.1 Hz), 2.29 (s, 6H), 2.47 (t,
2H, J=6.1 Hz), 3.00 (q, 2H, J=6.0 Hz), 3.31±3.35 (m,
2H), 3.79 (s, 6H), 3.80 (s, 3H), 4.88 (t, 1H, J=5.6 Hz),
6.69 (s, 2H), 6.73 (s, 1H), 7.03 (d, 1H, J=1.5 Hz), 7.20
(s, 1H), 7.22 (s, 1H), 7, 28 (s, 1H), 7.60 (br s, 1H), 8.51
(br s, 1H), 8.77 (br s, 1H), 8.86 (br s, 1H). HRMS calcd
+
.
for C34H52N9O6 (20 H ) 682.4040, found 682.4039.
Conversion of 18±20 to the respective free amines 21±23.
A solution of 18 (58 mg, 0.13 mmol) in CH2Cl2 (1.5 mL)
was treated with TFA (0.5 mL), stirred for 2 h at rt, then
concentrated in vacuo. The residue was dissolved in
CH2Cl2 (25 mL), treated with anhydrous K2CO3, ®l-
tered, and concentrated in vacuo to give crude 21.
Compounds 22 and 23 were prepared from 19 and 20 in
an analogous manner. 21: 1H NMR (D2O) d (ppm)
1.37±1.45 (m, 2H), 1.63±1.70 (m, 4H), 1.95±2.04 (m,
2H), 2.36 (t, 2H, J=7.3 Hz), 2.89 (s, 6H), 2.98 (t, 2H,
J=7.0 Hz), 3.18 (t, 2H, J=7.9 Hz), 3.39 (t, 2H,
J=6.6 Hz), 3.78 (s, 3H), 6.70 (s, 1H), 7.48 (s, 1H).
was performed with
a Storage Phosphor Screen
(Applied Biosystems), using Image Quant software
(Applied Biosystems). Ethanol precipitations were per-
formed by admixing the aqueous solution of DNA with
0.1 vol of 3 M NaOAc, pH 5, and 7 vol of cold ethanol,
followed by centrifugation (12,000 rpm, 4ꢀC, 15 min).
The supernatant was removed and the pellet was dried
in a speed-vac. Calf thymus DNA was sonicated and
dialyzed before use.
.
Molar extinction coecients (el) of 11±13 were deter-
mined by weighing 1±2 mg of the dry sample on a
microbalance in an aluminum boat. The boat with the
sample was transferred into water. The solution was
diluted to volume. e274 was calculated from the mea-
sured absorbance of the solution at 274 nm.
HRMS calcd for C17H32N5O2 (21 H+) 338.2556, found
1
338.2559. 22: H NMR (D2O) d (ppm) 1.35±1.43 (m,
2H), 1.61±1.72 (m, 4H), 1.95±2.01 (m, 2H), 2.29 (t, 2H,
J=7.4 Hz), 2.89 (s, 6H), 2.99 (t, 2H, J=7.4 Hz), 3.17 (t,
2H, J=8.0 Hz), 3.35 (t, 2H, J=6.7 Hz), 3.72 (s, 6H),
6.65 (s, 1H), 6.68 (s, 1H), 7.47 (s, 1H), 7.91 (s, 1H).
.
HRMS calcd for C23H38N7O3 (22 H+) 460.3036, found
1
Synthesis of compounds 18±20. A solution of nitro-
pyrrole 1524 (254 mg, 1.0 mmol) in CH3OH (5 mL) con-
taining 5% Pd/C (50 mg) was hydrogenated (1 atm) for
6 h at rt, ®ltered through silica gel, and concentrated
under vacuum. The residue was dissolved in toluene,
concentrated, and dried under vacuum to give the crude
aminopyrrole that was used in the next step without
further puri®cation: A solution of carbonyldiimidazole
(225 mg, 1.4 mmol) in DMF (1 mL) was added to a
solution of N-Boc-6-aminocaproic acid (350 mg,
1.5 mmol) in DMF (2 mL), stirred for 4 h at rt, then
treated with a solution of crude aminopyrrole in DMF
(1 mL). The mixture was stirred at rt for 16 h, then
concentrated under vacuum. The residue was dissolved
in Cl3CH (50 mL), washed with 0.5 M K2HPO4, dried
over MgSO4 and concentrated under vacuum. The resi-
due was puri®ed by ¯ash chromatography (Cl3CH/
CH3OH 4:1, containing 1% Et3N) to give 18 (370 mg,
84%). Compounds 19 and 20 were synthesized following
460.3035. 23: H NMR (D2O) d (ppm) 1.29±1.34 (m,
2H), 1.51±1.62 (m, 4H), 1.84±1.88 (m, 2H), 2.23 (t, 2H,
J=7.4 Hz), 2.81 (s, 6H), 2.75±2.88 (m, 2H), 3.04 (t, 2H,
J=7.5 Hz), 3.27 (t, 2H, J=6.4 Hz), 3.75 (s, 3H), 3.76 (s,
3H), 3.80 (s, 3H), 6.70 (d, 1H, J=1.7 Hz), 6.75 (d, 1H,
J=1.7 Hz), 6.81 (d, 1H, J=1.6 Hz), 7.03 (d, 1H,
J=1.7 Hz), 7.06 (d, 1H, J=1.6 Hz), 7.43 (s, 1H). HRMS
.
calcd for C29H44N9O4 (23 H+) 582.3516, found
582.3515.
Synthesis of 11±13 by coupling of mitomycin A (14) with
21±23. Crude 21 was dissolved in CH3OH (1 mL), trea-
ted with pyridine (0.2 mL) and mitomycin A (14)
(58 mg, 0.13 mmol). The mixture was stirred for 16 h at
rt protected from light, then concentrated in vacuo
and puri®ed by ¯ash chromatography (CHCl3/CH3OH
5:1 containing 1% Et3N) to give 11 (73 mg, 86%).
Conjugates 12 and 13 were synthesized following an
analogous procedure. 11: 1H NMR (D2O) d (ppm)

2724
M. M. Paz, et al. / Bioorg. Med. Chem. 7 (1999) 2713±2726
1.29±1.40 (m, 2H), 1.59±1.70 (m, 4H), 1.88 (s, 3H),
1.85±1.95 (m, 2H), 2.29 (t, 2H, J=7.5 Hz), 2.63 (s, 6H),
2.80±2.92 (m, 2H), 3.10±3.18 (m, 2H), 3.18 (s, 3H), 3.33
(t, 2H, J=6.5 Hz), 3.49±3.58 (m, 4H), 3.82 (s, 3H), 4.15
(d, 2H, J=10.5 Hz), 4.63 (dd, 1H, J=10.6 Hz, 4.3 Hz),
6.70 (d, 1H, J=1.9 Hz), 7.10 (d, 1H, J=1.9 Hz).
ESIMS: m/e 654 (M+H+). UV (H2O) lmax (e) 220, 374
(18,900). 12: 1H NMR (D2O) d (ppm) 1.40±1.46 (m,
2H), 1.62±1.74 (m, 4H), 1.90±1.95 (m, 2H), 1.98 (s, 3H),
2.33 (t, 2H, J=7.2 Hz), 270 (s, 6H), 2.86±2.99 (m, 4H),
3.21 (s, 3H), 3.31±3.40 (m, 2H), 3.52±3.60 (m, 4H), 3.81
(s, 3H), 3.88 (s, 3H), 4.16±4.25 (m, 2H), 4.66 (dd, 1H,
J=10.6 Hz, 4.3 Hz), 6.82 (d, 1H, J=1.4 Hz), 6.86 (d,
1H, J=1.4 Hz), 7.13 (s 1H), 7.17 (s, 1H). ESIMS: m/e
777.32 (M+H+). UV (H2O) lmax (e) 224, 302, 374
reduction of 11±13 by NADH-cytochrome c reductase
and NADH was performed as described above. (ii)
Acidic activation: 1 mM solutions of 11±13 were incu-
bated in 0.02 M HCl for 2 h, then stored at 20ꢀC for
later analysis. (iii) Sodium dithionite activation: A solu-
tion of the conjugate (11±13) (1 mM, 0.50 mL; 0.5 mmol
total) in 10 mM NaH2PO4±Na2HPO4, pH 7.5, was de-
aerated by argon bubbling. A solution of freshly pre-
pared 15 mM sodium dithionite in the same reaction
bu€er were added in ®ve 10 mL portions at 10 min
intervals. The mixtures were analyzed by diode array
HPLC and LCMS using the following conditions: sol-
vent A, 10 mM NH4OAc, pH 7; solvent B, 60% aceto-
nitrile in A; linear gradient 6±100% B in 45 min.
1
(17,500). 13: H NMR (CD3OD) d (ppm) 1.40±1.44 (m,
pBR322 DNA cross-linking assay.31 A mixture of
pBR322 DNA (5 mg, 5 mL), EcoRI restriction endonu-
clease (5 mL, 50 units), 10 mL EcoRI bu€er (10 mL, sup-
plied with the EcoRI enzyme) and 10 mL sterile H2O was
incubated for 60 min at 37ꢀC. The linear DNA was
labeled at the 30 end by adding a-32P-dATP (2 mL,
20 mCi) and Klenow polymerase (1 mL, 5 units). The
mixture was incubated for 20 min at 37ꢀC and the DNA
was puri®ed by ethanol precipitation. A mixture of
labeled pBR322 DNA (approximately 100 ng) and MC±
pyrrole conjugate (at various concentrations) in 90 mL
10 mM NaH2PO4±Na2HPO4±1 mM EDTA (pH 7.5)
was deaerated by bubbling argon for 1 min, and 10 mL
of a freshly prepared 10 mM solution of sodium dithio-
nite in deaerated bu€er was added. The samples were
incubated for 45 min at room temperature, and the
reactions were stopped by opening the tubes to air and
gently stirring. The DNA was puri®ed by ethanol pre-
cipitation, dissolved in strand separation bu€er (30%
DMSO, 1 mM EDTA, pH 9, 0.25% xylene cyanol,
0.25% bromophenol blue), heated at 90ꢀC for 5 min,
then immediately ice-cooled. Samples were loaded on a
1% agarose gel and run at 40 V for 16 h. The gels were
dried and the bands were visualized and quanti®ed by
phosphorimagery.
2H), 1.63±1.74 (m, 4H), 1.90±1.97 (m, 2H), 1.99 (s, 3H),
2.33 (t, 2H, J=7.2 Hz), 2.70 (s, 6H), 2.82±2.94 (m, 4H),
3.21 (s, 3H), 3.31±3.40 (m, 2H), 3.52±3.74 (m, 4H), 3.88
(s, 3H), 3.89 (s, 3H), 3.90 (s, 3H), 3.95±4.20 (m, 2H),
4.66 (dd, 1H, J=10.5 Hz, 4.3 Hz), 6.84 (d, 1H,
J=1.6 Hz), 6.86 (s, 1H), 6.95 (d, 1H, J=1.4 Hz), 7.13 (d
1H, J=7.5 Hz), 7.18 (d, 1H, J=1.5 Hz), 7.19 (d, 1H,
J=1.5 Hz). ESIMS: m/e 899.4 (M+H+). UV (H2O)
lmax (e) 227, 309, 374 (18,100).
Reduction kinetics of MC±pyrrole conjugates. (a) In the
absence of DNA NADPH-cytochrome c reductase
(®nal concentration 2.5 units/mL) was added to a de-
aerated solution containing 1 mM 11, 2 mM NADPH
and 2 mM dT in 10 mM NaH2PO4±Na2HPO4; 1 mM
EDTA, pH 7.5. Aliquots were removed at several time
intervals, quenched with 10 vol of cold ethanol and
stored at 20ꢀC until analyzed. The consumption of 11,
12 or 13 was monitored by HPLC using the following
conditions: solvent A, 30 mM sodium phosphate, pH
5.8; solvent B, 60% acetonitrile in A; linear gradient 6±
80% B in 45 min. The percentage consumption of start-
ing material was calculated from the percentage
decrease of the HPLC peak area ratio of starting mate-
rial to dT present as a marker. (b) In the presence of
DNA cytochrome c reductase (®nal concentration 1.2
units/mL) was added to a deaerated solution containing
15 mM calf thymus DNA, 0.10 mM 11, 0.5 mM NADH
and 0.1 mM dT in 10 mM NaH2PO4±Na2HPO4; 1 mM
EDTA, pH 7.5. Aliquots were removed at several time
intervals, quenched with 10 vol of 10 mM NaH2PO4±
Na2HPO4; 1 mM EDTA, pH 7.5, previously heated to
90ꢀC. The mixture was heated for 5 min at 90ꢀC to
ensure the inactivation of the enzyme. The solutions
were cooled to room temperature and admixed with
DNase I (25 units), phosphodiesterase I (2 units) and
alkaline phosphatase (5 units). The enzymatic digestion
ensures the release of DNA-bound drug. For compar-
ison, reactions in which DNA was omitted were per-
formed in parallel. The consumption of 11 was
determined by HPLC as above, except that bromophe-
nol blue, instead of dT, was added as marker. The rate
of disappearance of 12 and 13 was determined following
an analogous procedure.
Reaction of calf thymus DNA with reductively activated
11±13. A solution of calf thymus DNA in 0.01 M Tris±
HCl, pH 7.4, containing 1 mmol DNA was admixed with
a solution of 11 in methanol (0.2 mmol total), the mix-
ture was lyophilized, redissolved in 0.1 M Tris±HCl, pH
7.5 (200 mL), and deaerated by bubbling argon for
10 min. Five portions of a freshly prepared 10 mM
solution of sodium dithionite (6 mL each; 0.3 mmol total)
in deaerated bu€er were added at 10 min intervals, and
the reactions were stopped by opening the tubes to air
and gently stirring. The DNA was puri®ed by 5 extrac-
tions with n-BuOH, followed by ethanol precipitation,
then it was dissolved in 10 mM NH4OAc (900 mL),
admixed with 3 units of P1 nuclease and incubated for
3 h at 37ꢀC. After this period, the solution was admixed
with 100 mL of 500 mM Tris±HCl±5 mM MgCl2 (pH
8.3), phosphodiesterase I (5 units) and bacterial alkaline
phosphatase (10 units). The mixture was incubated 5 h
at 37ꢀC, then analyzed by HPLC, using the following
conditions: solvent A, 30 mM sodium phosphate, pH
5.8; B, 60% acetonitrile in A; linear gradient 6±80% B
in 45 min.
Formation of metabolites after activation of 11±13. (i)
Enzymatic activation: The formation of metabolites by

M. M. Paz, et al. / Bioorg. Med. Chem. 7 (1999) 2713±2726
2725
Determination of DNA binding constants of conjugates
11±13 by the ethidium displacement assay.28 Concentra-
tion of ethidium bromide (EtBr) was measured by UV
spectrometry by using the molar extinction coecient
7. Schiltz, P.; Kohn, H. J. Am. Chem. Soc. 1993, 115, 10510.
8. Suresh Kumar, G.; Lipman, R.; Cummings, J.; Tomasz, M.
Biochemistry 1997, 36, 14128.
9. Siegel, D.; Beall, H.; Senekowitsch, C.; Kasai, M.; Arai, H.;
Gibson, N. W.; Ross, D. Biochemistry 1992, 31, 7879.
10. Chirrey, L.; Cummings, J.; Halbert, G. W.; Smyth, J. F.
Cancer Chemother. Pharmacol. 1995, 35, 318.
11. Palom, Y.; Belcourt, M. F.; Suresh Kumar, G.; Arai, H.;
Kasai, M.; Sartorelli, A. C.; Rockwell, S.; Tomasz, M. Oncol.
Res. 1998, 10, 509.
12. Suresh Kumar, G.; He, Q. Y.; Behr-Ventura, D.; Tomasz,
M. Biochemistry 1995, 34, 2662.
13. Peterson, D. M.; Fischer, J.; Beall, H. D.; Ross, D. Cancer
Lett. 1995, 90, 133.
14. Kennedy, K.; Sligar, S. G.; Polomski, L.; Sartorelli, A. C.
Biochem. Pharmacol. 1982, 11, 2011.
15. Keyes, S. R.; Fracasso, P. M.; Heimbrook, D. C.; Rock-
well, S.; Sligar, S. G.; Sartorelli, A. C. Cancer Res. 1984, 44,
5638.
16. Hodnick, W. F.; Sartorelli, A. C. Cancer Res. 1993, 53,
4907.
17. Montzouris, J. A.; Hurley, L. H. In Bioorganic Chemistry:
Nucleic Acids, Hecht, S. M., Ed.; Oxford University Press;
New York, 1996; p 288.
18. Zimmer, C.; Wahnert, U. Prog. Biophys. Mol. Biol. 1986,
47, 31.
19. Sigurdsson, S. Th.; Hopkins, P. B. Tetrahedron 1994, 50,
12065.
20. Lee, M.; Roldan, M. C.; Haskell, M. K.; McAdam, S. R.;
Hartley, J. A. J. Med. Chem. 1994, 9, 513.
21. For a comprehensive review, see Bailly, C.; Chaires, J. B.
Bioconjugate Chem. 1998, 9, 513.
22. Broggini, M.; Erba, E.; Ponti, M.; Ballinardi, D.; Geroni,
C.; Sprearico, F.; D'Incalci, M. Cancer Res. 1991, 51, 199.
23. Broggini, M.; Coley, H. M.; Mongelli, N.; Pesenti, E.;
Wyatt, M. D.; Hartley, J. A.; D'Incalci, M. Nucleic Acids Res.
1995, 23, 81.
1
5850 (M.cm) at 480 nm in water. Calf thymus DNA
mononucleotide concentration was measured by using
the molar extinction coecient 6600 at 260 nm. Con-
centrations of MC analogues were measured by using
the molar extinction coecients at 374 nm, listed in
Table 1. A 3 mL cuvette was used and the titration
mixture was constantly stirred with the help of a mag-
netic stirrer. The bu€er was 10 mM Tris, 1 mM EDTA,
pH 7.1. Ethidium bromide concentration was main-
tained between 1.0 and 3.0 mM. A constant ratio of EtBr
and DNA concentration was maintained as 1:30 for
maximum ¯uorescence (excitation wavelength, 546 nm;
emission wavelength, 595 nm). Aliquots of a stock drug
solution in 10% MeOH (concentration in the range of
2±8 mM) were added to the ¯uorescence solution and
the ¯uorescence measured after each addition until a
50% reduction of ¯uorescence occurred. Fluorescence
was plotted against drug concentration after correcting
for the dilution factor. Drug concentration at 50%
quenching was calculated from the plots. The apparent
binding constant (Kb) of drug to CT DNA was then
calculated from the equation, KEtBr [EtBr]=Kb [drug]50,
where KEtBr is binding constant of EtBr to CT DNA,
[EtBr] is concentration of ethidium bromide and
[drug]50 is concentration of the drug required to reduce
the ¯uorescence of the EtBr±DNA complex to half of its
original value.
Acknowledgements
24. Boyd, M. R. In Anticancer Drug Development Guide: Pre-
clinical Screening, Clinical Trials, and Approval: Teicher, B.,
Ed.; Humana Press Inc.; Totowa, NJ, 1996; pp 23±42.
25. Franck, R. W.; Tomasz, M. In Chemistry of Antitumour
Agents; Wilman, D. E. V., Ed.; Chapman and Hall, New
York, 1990; pp 379±394.
26. Iyengar, B. S.; Lyn, H.-Y.; Cheng, L.; Remers, W. A. J.
Med. Chem. 1981, 24, 975.
27. Nishiwaki, E.; Tanaka, S.; Lee, H.; Shibuya, M. Hetero-
cycles 1988, 27, 1945.
28. Jenkins, T. C. In Methods in Molecular Biology, Vol. 90:
Drug±DNA Interaction Protocols; Fox, K. R., Ed.; Humana
Press, Totowa, NJ, 1997; pp 195±240.
29. Brooks, N.; Hartley, J. A.; Simpson, Jr., J. E.; Wright, S.
R.; Woo, S.; Centioni, S.; Fontaine, M. D.; McIntyre, T. E.;
Lee, M. Bioorg. Med. Chem. 1997, 5, 1497.
This research was supported by a grant (CA28681) from
the National Cancer Institute and a Research Centers in
Minority Institutions award (RR003037) from the
Division of Research Resources, NIH. The generous
gift of mitomycin A from Dr. Katsusige Gomi, Kyowa
Hakko Kogyo Co., Tokyo, Japan is gratefully
acknowledged. We thank Dr. Snorri Th. Sigurdsson,
University of Washington, for the generous donation of
intermediates of the synthesis of the minor groove
binding moieties, and Drs. A. C. Sartorelli and M. F.
Belcourt, Yale University for determination of drug
cytotoxicities in EMT6 mouse mammary tumor cells.
References
30. Gargiulo, D.; Musser, S. M.; Yang, L.; Fukuyama, T.;
Tomasz, M. J. Am. Chem. Soc. 1995, 117, 9388.
31. Hartley, J. A.; Berardini, M. D.; Souhami, R. L. Anal.
Biochem. 1991, 193, 131.
1. For a recent review, see Tomasz, M.; Palom, Y. Pharmacol.
Ther. 1997, 76, 73.
2. Rockwell, S.; Kennedy, K. A.; Sartorelli, A. C. Int. J.
Radiat. Onc. Biol. Phys. 1982, 8, 753.
3. For a review, see Brown, M. J.; Garcia, A. J. Cancer Res.
1998, 58, 1408.
4. Siim, B. G.; Menke, D. R.; Dorie, M. J.; Brown, M. J.
Cancer Res. 1997, 57, 2922.
5. Ha€ty, B. G.; Son, Y. H.; Moini, M.; Papac, R.; Fischer,
D.; Rockwell, S.; Sartorelli, A. C.; Fischer, J. J. Radiat. Oncol.
Invest. 1994, 1, 297.
32. Tomasz, M.; Lipman, R. J. Am. Chem. Soc. 1979, 101, 6063.
33. Pan, S.; Andrews, P. A.; Bachur, N. R. J. Biol. Chem.
1984, 259, 959.
34. Arcamone, F. M.; Animati, F.; Barbieri, B.; Con®gliacci,
E.; D'Alessio, R.; Geroni, C.; Giuliani, F. C.; Lazzari, E.;
Menozzi, M.; Mongelli, N.; Penco, S.; Veroni, M. A. J. Med.
Chem. 1989, 32, 774.
35. Paull, K. D.; Shoemaker, R. H.; Hodes, L.; Monks, A.;
Scudiero, D. A.; Rubinstein, L. J. Natl. Cancer Inst. 1989, 81,
1088.
36. These tests were conducted by Dr. M. F. Belcourt; Yale
University School of Medicine, New Haven, CT.
6. Ha€ty, B. G.; Son, Y. H.; Sasaki, C. T.; Papac, R.; Fischer,
D.; Rockwell, S.; Sartorelli, A. C.; Fischer, J. J. Int. J. Radiat.
Oncol. 1993, 27, 241.

2726
M. M. Paz, et al. / Bioorg. Med. Chem. 7 (1999) 2713±2726
37. Wyatt, M. D.; Garbiras, B. J.; Haskell, M. K.; Lee, M.; Sou-
hami, R. L.; Hartley, J. A. Anti-cancer Drug Design 1994, 9, 511.
38. Lee, M.; Rhodes, A. L.; Wyatt, M. D.; Forrow, S.; Hart-
ley, J. A. Anti-cancer Drug Design 1993, 8, 173.
39. Wyatt, M. D.; Lee, M.; Garbiras, B. J.; Souhami, R. L.;
Hartley, J. A. Biochemistry 1995, 34, 13034.
42. Fagan, P. A.; Spielmann, H. P.; Sigurdsson, S. Th.; Rink,
S. M.; Hopkins, P. B.; Wemmer, D. E. Nucleic Acids Res.
1996, 24, 1566.
43. Kanaoka, Y.; Ikeuchi, Y.; Kawamoto, T.; Bessho, K.;
Akimoto, N.; Mikata, Y.; Nishida, M.; Yano, S.; Sasaki, T.;
Yoneda, F. Bioorg. Med. Chem. 1998, 6, 301.
40. Pan, S.; Gonzalez, H. Mol. Pharmacol. 1990, 37, 966.
41. Hoey, B. M.; Butler, J.; Swallow, A. J. Biochemistry 1988,
27, 2608.
44. Lee, M.; Rhodes, A. L.; Wyatt, M. D.; D'Incalci, M.;
Forrow, S.; Hartley, J. A. J. Med. Chem. 1998, 36, 863.