Design, synthesis, and evaluation of a novel pyrrolobenzodiazepine DNA-interactive agent with highly efficient cross-linking ability and potent cytotoxicity

Article abstract of DOI:10.1021/jm001064n

A novel sequence-selective pyrrolobenzodiazepine (PBD) dimer 5 (SJG-136) has been developed that comprises two C2-exo-methylene-substituted DC-81 (3) subunits tethered through their C8 positions via an inert propanedioxy linker. This symmetric molecule is

Full text of DOI:10.1021/jm001064n

J . Med. Chem. 2001, 44, 737-748
737
Design , Syn th esis, a n d Eva lu a tion of a Novel P yr r oloben zod ia zep in e
DNA-In ter a ctive Agen t w ith High ly Efficien t Cr oss-Lin k in g Ability a n d P oten t
Cytotoxicity
Stephen J . Gregson,† Philip W. Howard,† J ohn A. Hartley,‡ Natalie A. Brooks,‡ Lesley J . Adams,§
Terence C. J enkins,§ Lloyd R. Kelland,| and David E. Thurston*,†
CRC Gene Targeted Drug Design Research Group, Cancer Research Laboratories, University of Nottingham,
University Park, Nottingham NG7 2RD, U.K., CRC Drug-DNA Interactions Research Group, Department of Oncology,
University College London Medical School, London W1P 8BT, U.K., Yorkshire Cancer Research Laboratory of Drug Design,
Cancer Research Group, University of Bradford, Bradford, West Yorkshire BD7 1DP, U.K.,
and CRC Centre for Cancer Therapeutics, Institute for Cancer Research, Clifton Avenue, Sutton, Surrey SM2 5PX, U.K.
Received August 28, 2000
A novel sequence-selective pyrrolobenzodiazepine (PBD) dimer 5 (SJ G-136) has been developed
that comprises two C2-exo-methylene-substituted DC-81 (3) subunits tethered through their
C8 positions via an inert propanedioxy linker. This symmetric molecule is a highly efficient
minor groove interstrand DNA cross-linking agent (XL50 ) 0.045 µM) that is 440-fold more
potent than melphalan. Thermal denaturation studies show that, after 18 h incubation with
calf thymus DNA at a 5:1 DNA/ligand ratio, it increases the Tm value by 33.6 °C, the highest
value so far recorded in this assay. The analogous dimer 4 (DSB-120) that lacks substitution/
unsaturation at the C2 position elevates melting by only 15.1 °C under the same conditions,
illustrating the effect of introducing C2-exo-unsaturation which serves to flatten the C-rings
and achieve a superior isohelical fit within the DNA minor groove. This behavior is supported
by molecular modeling studies which indicate that (i) the PBD units are covalently bonded to
guanines on opposite strands to form a cross-link, (ii) 5 has a greater binding energy compared
to 4, and (iii) 4 and 5 have equivalent binding sites that span six base pairs. Dimer 5 is
significantly more cytotoxic than 4 in a number of human ovarian cancer cell lines (e.g., IC50
values of 0.0225 nM vs 7.2 nM, respectively, in A2780 cells). Furthermore, it retains full potency
in the cisplatin-resistant cell line A2780cisR (0.024 nM), whereas 4 loses activity (0.21 µM)
with a resistance factor of 29.2. This may be due to a lower level of inactivation of 5 by
intracellular thiol-containing molecules. A dilactam analogue (21) of 5 that lacks the electrophilic
N10-C11/N10′-C11′ imine moieties has also been synthesized and evaluated. Although unable
to interact covalently with DNA, 21 still stabilizes the helix (∆Tm ) 0.78 °C) and has significant
cytotoxicity in some cell lines (i.e., IC50 ) 0.57 µM in CH1 cells), presumably exerting its effect
through noncovalent interaction with DNA.
In tr od u ction
of particular proteins critical for tumor cell proliferation,
metastasis, or drug resistance. For complete biological
specificity, such agents must be able to recognize duplex
DNA in order to target individual gene sequences.1
DNA is the target for many clinically useful antitumor
agents; however, few of the DNA-interactive drugs in
current clinical use exhibit sequence-specificity extend-
ing beyond two to three base pairs (bp).1 Selectivity is
generally thought to favor the targeting of rapidly
growing tumor cells, rather than discriminating against
any fundamental differences between normal and tumor
cells at the genetic level. Human genes (e.g., oncogenes)
that are responsible for transforming activity in tumor
cells have been identified as potential targets for new
drugs. The ultimate goal is to design and synthesize
agents capable of specifically inhibiting the expression
The pyrrolo[2,1-c][1,4]benzodiazepines (PBDs) are a
family of DNA-interactive antitumor antibiotics derived
from various Streptomyces species.2 Well-known mem-
bers include DC-81 (1) and tomaymycin (2) (Figure 1).
They exert their biological activity through covalent
binding via their N10-C11 imine/carbinolamine moiety
to the C2-amino position of a guanine residue within
the minor groove of DNA. PBD monomers span three
DNA base pairs with a preference for Pu-G-Pu (where
Pu ) purine; G ) guanine) sequences. Extensive studies
have been carried out on both the solution-3 and solid-
phase4 synthesis of PBDs, and a sound understanding
of structure-activity relationships within the family has
been developed.5 The PBDs have been shown to inter-
fere with the interaction of endonuclease enzymes with
DNA6 and to block transcription by inhibiting RNA
polymerase in a sequence-specific manner,7 processes
which are thought to account for the biological activity
* Address correspondence to David E. Thurston, Professor of Cancer
Chemotherapy, Director, CRC Gene Targeted Drug Design Research
Group, Cancer Research Laboratories, University of Nottingham,
University Park, Nottingham NG7 2RD, U.K. Phone (direct): +44 (0)-
115 8466076 (+ voice mail). Phone (secretary): +44 (0)115 951 3404.
Fax (direct): +44 (0)115 951 3114. Fax (secretary): +44 (0)115 951
3412. E-mail: david.thurston@nottingham.ac.uk.
† CRC Gene Targeted Drug Design Research Group.
‡ CRC Drug-DNA Interactions Research Group.
§ Yorkshire Cancer Research Laboratory of Drug Design.
| CRC Centre for Cancer Therapeutics.
10.1021/jm001064n CCC: $20.00 © 2001 American Chemical Society
Published on Web 01/31/2001

738 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 5
Gregson et al.
Resu lts a n d Discu ssion
Ch em istr y. Synthesis of 5 was achieved by employ-
ing the B-ring cyclization strategy first reported by
Fukuyama and co-workers.21 First, commercially avail-
able trans-4-hydroxy-L-proline (6) was N-protected as
the allyl carbamate 7 in 87% yield (Scheme 1).22
Following esterification of 7 in modest yield (43%) using
catalytic H2SO4 in refluxing MeOH, the methyl ester 8
was reduced to the diol 9 with LiBH4 in almost
quantitative yield. Differential silylation of the primary
rather than secondary alcohol was achieved by employ-
ing DBU as a silyl transfer agent. Any disilylated
product and unreacted diol were removed by column
chromatography to provide the TBDMS ether 10 in 52%
yield. Oxidation to the ketone 11 was achieved using
either the Swern reaction or TPAP in the presence of
NMO and 4 Å molecular sieves.23 Both methods pro-
duced the ketone 11 in almost quantitative yield. The
key C4/C4′- (i.e., pro-C2/C2′ for the final PBD structure)
unsaturation was introduced by performing the Wittig
reaction on ketone 11 to provide the olefin 12 in 87%
yield. Initial attempts to deprotect 12 using PPh3/Pd-
(PPh3)4 in the presence of a suitable allyl scavenger (i.e.,
pyrrolidine,24 dimedone,25 or 2-ethylhexanoic acid26
)
F igu r e 1. Structures of the PBD monomers DC-81 (1),
tomaymycin (2), and C2-methylene DC-81 (3), and the PBD
dimers 4 and 5.
were unsuccessful. Eventually, the Alloc group was
cleaved by palladium-catalyzed hydrostannolysis with
tributyltin hydride to provide amine 13 in 77% yield.27
The known PBD dimer core13 14 was converted to the
corresponding acid chloride by treatment with oxalyl
chloride followed by a catalytic amount of DMF. Sub-
sequent coupling to 13 provided the bis-nitro-amide 15
in 74% yield. The TBDMS protecting groups were
removed rapidly under mild conditions using TBAF in
THF to afford the bis-nitro-alcohol 16 in 94% yield.
Reduction of the nitro groups was achieved by employ-
ing SnCl2‚2H2O in refluxing MeOH.28 The resulting bis-
aniline 17 was obtained in 61% yield with the C4/C4′-
unsaturation intact. Following Alloc protection of the
anilino moieties in 50% yield, 18 was subjected to Swern
oxidation in an attempt to synthesize the protected
carbinolamine 19. Unfortunately, 18 was prone to over-
oxidation to the tetralactam 20 under these conditions.
Instead, the spontaneous oxidation/ring closure was
achieved by treating 18 with TPAP/NMO/4 Å molecular
sieves. Although the yield was modest (32%), the forma-
tion of tetralactam 20 was not observed using TPAP.
Finally, treatment of 19 with Pd(PPh3)4/ΡΡh3/pyrroli-
dine24 afforded the novel PBD dimer20 5 in 77% yield.
Cleavage of the N10/N10′-Alloc protecting groups from
20 under identical conditions afforded 21, the first
example of a PBD tetralactam dimer, in 43% yield.
Biop h ysica l Stu d ies. (a ) DNA Bin d in g Stu d ies.
The DNA binding affinity of the novel C2-functionalized
PBD dimers 5 and 21 was examined by thermal
denaturation using calf thymus (CT) DNA.6 Studies on
5 were carried out at [DNA]/[ligand] molar ratios of 100:
1, 50:1, and 5:1. The increase in helix melting temper-
ature (∆Tm) for each ratio was examined after 0, 4, and
18 h incubation at 37 °C (Table 1). Data for 21, 4, and
tomaymycin (2) are included in Table 1 for comparison.
The data clearly show that 5 is a highly efficient
stabilizing agent for double-stranded calf thymus DNA.
For a [ligand]/[DNA] molar ratio of 1:5, 5 elevates the
helix melting temperature of CT DNA by an unprec-
of the compounds. The PBDs have also been used as a
scaffold to attach EDTA8 and epoxide9 moieties, leading
to novel sequence-selective DNA cleaving and cross-
linking agents, respectively. A homologous series of C8-
diether-linked PBD dimers (e.g., 4, Figure 1) has been
synthesized, members of which exhibit potent in vitro
cytotoxicity10 and enhanced DNA-binding affinity and
sequence-specificity compared to the natural product
DC-81 (1, Figure 1).11-13 This improved biological activ-
ity can be attributed to the ability of these compounds
to cross-link DNA irreversibly, as demonstrated by a
gel electrophoresis assay.14 In addition, NMR spectros-
copy15 and molecular modeling studies16 indicate that
4 spans six DNA base pairs, actively recognizing a 5′-
GATC sequence. This represents an effective doubling
of the DNA binding-site size of DC-81, which is known
to be selective for Pu-G-Pu triplets.
Subsequent in vivo studies with 4 in the murine ADJ /
PC6 plasmacytoma model were disappointing,17 and the
low therapeutic index observed was thought to be partly
due to reaction of the molecule with cellular thiol-
containing molecules prior to reaching the tumor site.
PBD monomer natural products with unsaturation at
the C2 position are known to be more biologically potent
than their C2-saturated relatives. One explanation for
this is that C2-exo-unsaturation can lead to lower
electrophilicity at the N10-C11 position.18,19 This may
allow greater availability of the agent at the target DNA
sequence due to a lower level of deactivation by cellular
nucleophiles. For example, tomaymycin (2) is consider-
ably more cytotoxic than DC-81 and binds more ef-
ficiently to DNA, although it is less electrophilic over-
all.6,18,19 Therefore, it was of interest to synthesize and
evaluate a PBD dimer with unsaturation at the C2/C2′
positions. Here, we report the synthesis and biological
evaluation of an exquisitely potent bis(desmethyltomay-
mycin) PBD dimer, 5.20

Pyrrolobenzodiazepine DNA-Interactive Agent Synthesis
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 5 739
Sch em e 1a
a
Reagents and conditions: (i) Alloc-Cl, aq. NaOH, THF, 0 °C, 87%; (ii) MeOH, H2SO4, ∆, 43%; (iii) LiBH4, THF, 0 °C, 99%; (iv) TBDMS-
Cl, TEA, DBU, CH2Cl2, 52%; (v) TPAP, NMO, 4 Å sieves, CH2Cl2-CH3CN, 92%; or (COCl)2, DMSO, TEA, CH2Cl2, -70 °C, 95%; (vi)
Ph3PCH3Br, KOtBu, THF, 0 °C, 87%; (vii) Bu3SnH, Pd(PPh3)2Cl2, H2O, CH2Cl2, 77%; (viii) (COCl)2, DMF, THF then 13, TEA, H2O, 0 °C,
74%; (ix) TBAF, THF, 0 °C, 94%; (x) SnCl2‚2H2O, MeOH, ∆, 61%; (xi) Alloc-Cl, pyridine, CH2Cl2, 0 °C, 50%; (xii) TPAP, NMO, 4 Å sieves,
CH2Cl2, CH3CN, 32%; (xiii) (COCl)2, DMSO, TEA, CH2Cl2, -45 °C 51%; (xiv) Pd(PPh3)4, PPh3, pyrrolidine, CH2Cl2, CH3CN, 0 °C, 77% for
5 and 43% for 21.
Ta ble 1. Thermal Denaturation Data for Tomaymycin (2), 4, 5,
60-80% of its stabilizing effect without prior incubation,
and 21 with Calf Thymus DNA
suggesting a kinetic effect in the PBD reactivity profile.
induced ∆Tm (°C)a after
However, the comparative ∆Tm curves (not shown)
indicate that, on a concentration basis alone, 5 is g10-
fold more efficient than 4. Even at a [DNA]/[ligand]
molar ratio of 100:1, 5 exhibits significantly greater
DNA-binding affinity than that for the corresponding
monomer tomaymycin at a lower 5:1 molar ratio.
Furthermore, after 72 h incubation at a [DNA]/[ligand]
ratio of 5:1, the ∆Tm for 5 increases to a value of 34.4
°C, suggesting a plateau saturation level for covalent
DNA interaction. Although the tetralactam 21 is sig-
nificantly less effective at stabilizing DNA than 5, the
results confirm that it does provide some helix stabiliza-
tion, elevating the DNA melting temperature by 0.78
°C. This result clearly demonstrates that replacement
of the electrophilic N10-C11/N10′-C11′ bis-imine moi-
eties present in 5 with bis-lactam functionalities abol-
ishes covalent DNA-interaction, presumably leaving
only the noncovalent component of interaction. This
phenomenon has been previously observed with PBD
monomers.29
incubation at 37 °C for
[PBD]:[DNA]
molar ratiob
compound
0 h
4 h
18 h
5
5
1:100
1:50
1:5
1:5
1:5
7.1
11.3
25.7
0.78
10.2
0.97
8.0
12.3
31.9
-
9.1
15.0
33.6
-
15.1
2.56
5c
21
4
13.1
tomaymycin (2)
1:5
2.38
a
For CT-DNA alone at pH 7.00 ( 0.01, ∆Tm ) 67.83 ( 0.06 °C
(mean value from 30 separate determinations). All ∆Tm values
b
are (0.1-0.2 °C. For a 1:5 molar ratio of [ligand]/[DNA], calf
thymus DNA concentration ) 100 µM and ligand concentration
) 20 µM in aqueous sodium phosphate buffer [10 mM sodium
phosphate + 1 mM EDTA, pH 7.00 ( 0.01]. c The ∆Tm for 5 at a
[ligand]:[DNAp] molar ratio of 1:5 increased to a value of 34.4 °C
after 72 h incubation.
edented 33.6 °C after incubation at 37 °C for 18 h. In
the same experiment, the DC-81-based dimer 4 gives a
∆Tm of 15.1 °C, illustrating the significant effect of
introducing C2/C2′-unsaturation. As usually observed
for the PBD dimers, 5 exerts most of its effect upon the
GC-rich or high-temperature regions of the DNA melt-
ing curves. In a fashion similar to 4, 5 provides some
(b) DNA Cr oss-Lin k in g Stu d ies. The DNA cross-
linking efficiency of 5 was investigated using an assay
involving linear double-stranded DNA derived from the

740 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 5
Gregson et al.
carried out using an established protocol previously used
to examine the interactions of DNA with both reversible
and irreversible groove-binding agents.9,16,30-32 Helical
analysis shows that upon binding to 5 the DNA main-
tains its B-like conformation with the helical rise for
the adduct between DNA and 5 being 3.55 Å and the
mean helical rotation, 37.0° (9.7 bp/turn). This suggests
that there is little or no disruption of DNA secondary
structure upon adduct formation and cross-linking and
that base-pairing or induced distortion effects are absent
(see Figure 3). This behavior has been reported previ-
ously for the d(CICGATCICG)2-4 complex.16 Both 4 and
5 are well accommodated within the minor groove, such
that very little of either molecule remains exposed
beyond the periphery of the host DNA duplex (Figure
3). This feature may be responsible for the observed
resistance to repair enzymes that are reliant upon
tracing distortion or helical perturbation in DNA in-
duced by most alkylating agents.
(a)
(b)
The calculated binding energies (Table 3) show that
PBDs with unsaturation in the C-ring (i.e., 5 and the
hypothetical exo-CH2 monomer, 3, Figure 1) form more-
favorable adducts than analogues with fully saturated
C-rings (i.e., 4 and DC-81, respectively). Introduction
of the exocyclic methylene group at the C2 position,
where the sp3 f sp2 hybridization change stiffens the
C-ring, leads to a more planar arrangement that enables
this portion of the molecule to fit more snugly in the
minor groove. Importantly, atomic clash with the groove
walls is ameliorated by the superior isohelical matching
of the ligand with the groove contours.
F igu r e 2. (a) Autoradiograph of a neutral agarose gel showing
DNA interstrand cross-linking by 5 in linear 32P-end-labeled
pBR322 DNA following a 2 h incubation at 37 °C. The lanes
are as follows: C, double-stranded DNA control; 0, single-
stranded DNA control; and 0.001, 0.003, 0.01, 0.03, 0.1, 0.3,
1.0, 3.0, 10.0 µM 5. DS and SS are double- and single-stranded
DNA, respectively. (b) Concentration dependence of DNA
cross-linking for 5 in linear 32P-end-labeled pBR322 DNA from
gel in Figure 2a.
Covalent interaction between the DNA and each
dimer is approximately twice as favorable as that for
the corresponding monomers, reflecting their difunc-
tional character and the tolerance afforded to the inert
1,3-propanedioxy linker present between the PBD A-
rings. Both halves of the dimer behave in an essentially
identical fashion without disruption of the host DNA
integrity. The flexible linker between the A-rings of each
subunit makes favorable non-bonding contacts with the
floor and walls of the minor groove which enhances
overall binding.
Ta ble 2. DNA Cross-Linking Data for 4, 5, and Melphalan
with pBR322 DNA
a
compound
XL50 (µM)
4
5
0.055
0.045
20.0
melphalan
a
Both DC-81 (1) and the C2-exo-unsaturated monomer
(3) generally prefer to bind with their aromatic A-rings
oriented toward the 3′-end of the alkylated DNA strand
(i.e., 3S), although this is sensitive to sequence context.
In contrast, the spatially separated PBD units in each
dimer prefer to adopt the alternative 5S orientation.
This difference largely reflects the superior hydrogen
bonding capability in the 5S-5S or “reverse” cross-link
configuration. Interestingly, this 5S-5S/ 3S-3S prefer-
ence is very marked (13 kcal mol-1) for the more
discriminating 4, although the differential is absent for
the more reactive 5 where formation of forward (5′-
AGATCT) and reverse (5′-TCTAGA) cross-links are
essentially isoenergetic. The different behaviors pre-
dicted for cross-linking suggest that the DNA reactivity
profiles are particularly sensitive to the isohelical
characteristics of the dimer molecules.
Concentration of agent required for 50% cross-linking of
pBR322 DNA.
plasmid pBR322 (4632 bp, linearized with Hind III and
then 32P-end-labeled).14 Following denaturation condi-
tions which completely separate the DNA strands, the
presence of an interstrand cross-link results in rena-
turation to double-stranded DNA during electrophoresis
in a neutral agarose gel.
The highly efficient DNA cross-linking ability of 5 is
illustrated in the autoradiograph in Figure 2a which is
quantitated by laser densitometry in Figure 2b. After 2
h incubation at 37 °C, cross-linking is measurable above
background at 0.001 µM, with 100% cross-linking at
doses of >0.3 µM. The concentration required to produce
50% double-stranded (i.e., interstrand cross-linked)
DNA was calculated from Figure 2b and is shown in
Table 2. Compound 5 is a more efficient cross-linking
agent than 4 (XL50 values: 0.045 µM versus 0.055 µM)
and is 440-fold more efficient than the major groove
cross-linking agent melphalan.
In Vitr o Cytotoxicity in Cisp la tin -Resista n t Cell
Lin es. The cytotoxicity of 5 in a panel of human ovarian
cell lines (SKOV-3, A2780/A2780cisR, and CH1/CH1cisR)
was determined using the 96 h continuous exposure
sulforhodamine B (SRB) growth delay assay.33 The cell
Molecu la r Mod elin g of In ter str a n d Cr oss-Lin k
F or m a tion . Studies on the energetics of binding were

Pyrrolobenzodiazepine DNA-Interactive Agent Synthesis
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 5 741
F igu r e 3. Energy-minimized models for the minor groove cross-linked adducts formed with d(CGCAGATCTGCG)2 and either 4
(A1, A2) or the more extended dimer 5 (B1, B2). Only the central eight base pairs are shown for each structure with all hydrogen
atoms removed for clarity. The bound molecules are highlighted in space fill; color scheme: PBD subunits (yellow), inert ligand
C8-C8′ diether linkage (cyan), and alkylated guanine bases (green). Panels A2 and B2 represent views looking down the long
dyad or z-axis of the DNA duplex, showing the effective burial of the dimer molecules within the minor groove conduit. The only
exposed molecular fragments are the 7-methoxy substituents in each PBD A-ring.
Ta ble 3. Interaction Energies for DNA Cross-Linking by the PBD Dimers 4 and 5 and for Monoalkylation by the Corresponding PBD
Monomers 1 and 3 Using Defined Oligonucleotide Duplex Targets
b
b
b
b
PBD
cmpd
∆EvdW
∆Eelec
∆EHB
Ebind
DNA sequencea
kcal mol-1
kcal mol-1
kcal mol-1
kcal mol-1
5′-CGCAGATCTGCG
4
5
-53.0
-38.1
-4.6
-88.5
-58.5
-43.7
-5.7
-104.1
1c
3c
4
-29.1/-25.7
-30.0/-24.2
-62.5
-20.6/-14.0
-22.6/-24.1
-39.7
-0.9/-2.8
-1.3/-2.8
-3.4
-48.0/-40.5
-51.5/-48.7
-101.4
5′-CGCTCTAGAGCG
5′-CGCAGAGCG
5
-60.1
-46.5
-2.7
-104.3
1c
3c
1c
3c
-27.5/-28.5
-25.1/-29.1
-17.5/-26.3
-27.1/-23.0
-17.2/-21.6
-22.7/-23.6
-13.0/-6.3
-33.1/-23.2
-3.0/-2.5
-2.8/-2.0
-2.2/-2.0
-2.1/-2.5
-45.6/-50.2
-48.8/-52.0
-31.7/-34.8
-61.2/-48.0
a
Sequence for one strand of the modeled host DNA duplexes, showing the alkylated guanine and the spanned cross-linking site for the
PBD dimers. The embedded 5′-AGATCT and 5′-TCTAGA tracts present twin optimal 5′-AGA sites for the PBD monomers for “forward”
(3S-3S) and “reverse” (5S-5S) cross-linking, respectively (see text).Where equivalent sites are present in the sequence, only one site was
b
examined for the PBD monomers. ∆EvdW, ∆Eelec, and ∆EHB are the calculated van der Waals, electrostatic, and H-bonded energy
contributions to the overall binding interaction energy (Ebind) as defined by Ebind ) E(adduct) - [E(free DNA) + E(free compound)] where each term
refers to the energy-minimized covalent adduct or native reactant. c Energy values refer to the alternative 3S or 5S alignments of the
PBD monomer with respect to the covalently modified guanine.
lines were chosen on the basis of their inherent sensi-
tivity to the DNA cross-linking agent cisplatin and the
availability of sub-lines with acquired resistance to
cisplatin. IC50 values for human leukemic K562 were
obtained using the microculture tetrazolium (MTT)
assay.34
The IC50 values obtained for 5 (Table 4) show that it
is exquisitely cytotoxic, exhibiting the lowest IC50 values
so far obtained for any synthetic PBD monomer or
dimer. In each cell line, 5 is significantly more potent
than either 4 or the clinically used antitumor agent
cisplatin. In the CH1 cell line, 4 is approximately 6-fold
more active than cisplatin, whereas 5 is 1500-fold more
potent. It is generally assumed that an interstrand
cross-link, if not repaired, will interfere with the process
of DNA replication and lead to cell death. The cisplatin-
resistant sub-line CH1cisR owes its resistance to the
enhanced repair of platinum-DNA adducts. Whereas 4

742 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 5
Gregson et al.
Ta ble 4. In Vitro Cytotoxicity Data for the PBD Dimers 4 and
5, the Tetralactam 21, and Cisplatin in Cisplatin-Resistant Cell
Lines
stabilize double-stranded DNA as evaluated by thermal
denaturation studies. Modeling studies suggest that this
is due to a flattening of the C-rings by the C2/C2′-
methylene groups, leading to a more snug fit in the
minor groove and a lower-energy adduct. This appears
to be reflected in the exquisite cytotoxicity of 5, par-
ticularly toward some cisplatin-resistant cell lines,
where it is possible that the low-energy non-distortive
adducts formed remain invisible to the DNA repair
surveillance systems. Furthermore, C2-exo-unsaturation
is known to reduce the electrophilicity of PBD molecules,
thus decreasing the likelihood of collateral interactions
with thiol-containing proteins and other nucleophilic
molecules such as glutathione.18,19 This suggests that
5 may have greater bioavailablity than 4.
IC50 (µM)a
cell line
5
21
28
2.5
11
4
cisplatin
SKOV-3
A2780
A2780cisR
RFb
0.0091
0.0000225
0.000024
1.1
0.74
0.0072
0.21
10.5
0.265
8.4
4.4
29.2
32
CH1
0.00012
0.0006
5
0.57
2.9
3.5
0.033
0.022
0.7
0.18
1.10
6.1
CH1cisR
RFb
K562
c
0.0425
ND
0.25
ND
a
Dose of PBD required to inhibit cell growth by 50% compared
to PBD-free controls. The cells were incubated with the compounds
b
for 96 h at 37 °C. RF is the resistance factor (IC50 resistant/
c
parent). K562 is a human leukemia cell line in which IC50 values
In addition to the significant in vitro cytotoxicity in
cisplatin-resistant ovarian cell lines (e.g., IC50 ) 0.024
nM in A2780cisR), selective cytotoxicity has been ob-
served for 5 in the NCI’s 60 cell line screen. In addition,
significant antitumor activity has been achieved in both
the Hollow Fiber assay and a number of human tumor
xenografts, and these data will be reported elsewhere.
Compound 5 is currently in preclinical development36
and has potential in a number of human cancers
including cisplatin-resistant ovarian disease.
were measured using an MTT assay following a 1 h exposure to
drug; ND ) not determined.
showed no cross-resistance in the CH1cisR line, 5 and
cisplatin gave resistance factors of 5 and 6.1, respec-
tively. Conversely, the potency of 5 is maintained in the
human chronic myelogenous cell line K562, whereas 4
is approximately 6-fold less effective. The most signifi-
cant results were obtained with the human ovarian
carcinoma cell lines SKOV-3 and A2780/A2780cisR. The
IC50 values for 5 in the A2780/A2780cisR pair of cell
lines are extremely low (∼0.00002 µM) and very little
cross-resistance is observed (RF ) 1.1). In contrast, 4
shows potent cytotoxicity in the parent A2780 line but,
like cisplatin, experiences almost 30-fold resistance in
the A2780cisR cells (RF ) 29.2). The A2780cisR sub-
line is known to possess elevated GSH levels, an
increased level of repair of DNA-cisplatin adducts, and
a decreased ability to uptake cisplatin.33
Similar results were obtained with the intrinsically
cisplatin-resistant SKOV-3 cell line, which possesses
levels of glutathione 4-fold higher than cisplatin-sensi-
tive ovarian carcinoma cells.35 It is thought that 4 failed
in initial in vivo studies due to its interaction with
biological nucleophiles, such as glutathione or thiol-
containing proteins in the blood and tissue.17 Unlike 4,
the behavior of 5 in SKOV-3 and A2780 cis R suggests
that the potency of the drug is unaffected by high
cellular GSH levels as predicted during the design of
the molecule. In summary, these results highlight the
effect that C2-unsaturation has on in vitro cytotoxicity
and suggest that 5 should perform significantly better
than 4 in in vivo studies.17
Exp er im en ta l Section
Syn th etic Ch em istr y. Reaction progress was monitored
by thin-layer chromatography (TLC) using GF254 silica gel,
with fluorescent indicator on glass plates. Visualization was
achieved with UV light and iodine vapor unless otherwise
stated. Flash chromatography was performed using 14 cm of
J .T. Baker 30-60 µm silica gel. The majority of reaction
solvents were purified by distillation under nitrogen from the
indicated drying agent and used fresh: dichloromethane
(calcium hydride), tetrahydrofuran (sodium benzophenone
ketyl), methanol (magnesium methoxide), acetonitrile (calcium
hydride), and toluene (sodium benzophenone ketyl). Extraction
and general chromatography solvents were bought and used
without further purification from J .T. Baker. All organic
chemicals were purchased from Aldrich Chemical Co. Drying
agents and inorganic reagents were purchased from BDH.
IR spectra were recorded with a Perkin-Elmer FT/IR-
Paragon 1000 spectrophotometer. 1H and 13C NMR spectra
were obtained on a J EOL GSX 270 MHz (67.8 MHz for 13C
NMR spectra) FT-NMR operating at 20 °C ( 1 °C. Chemical
shifts are reported in parts per million (ppm) downfield from
tetramethylsilane. Spin multiplicities are described as s (sin-
glet), br s (broad singlet), d (doublet), br d (broad doublet), t
(triplet), q (quartet), quint (quintet), or m (multiplet). NMR
interpretations are quoted for the pro-PBD numbering system
(see Figure 1). Mass spectra were recorded on a J EOL J MS-
DX 303 GC-mass spectrometer. Electron impact (EI) mass
spectra were measured at 70 eV, chemical ionization (CI)
spectra were obtained using isobutane as reagent gas, and fast
atom bombardment (FAB) spectra were recorded using thioglyc-
erol as a matrix. Accurate molecular masses were determined
by peak matching using perfluorokerosene (PFK) as an
internal standard. Optical rotations were measured at ambient
temperature using a Bellingham and Stanley ADP 220 pola-
rimeter.
Surprisingly, the tetralactam 21 exhibits limited
cytotoxicity in most of the ovarian carcinoma cell lines.
The IC50 of 0.57 µM in the CH1 cell line is particularly
striking. This activity is in accord with the thermal
denaturation data; presumably the cytotoxic effects are
due to noncovalent interaction with DNA, a phenom-
enon previously observed with the PBD monomers.29
Furthermore, comparison of the IC50 data for 5 and 21
clearly demonstrates that, for maximum cytotoxicity, an
electrophilic imine or carbinolamine moiety is essential
at the N10-C11 position of the PBD units.
(2S,4R)-N-(Allyloxyca r bon yl)-4-h yd r oxyp yr r olid in e-2-
ca r boxylic Acid (7). A solution of allyl chloroformate (29.2
mL, 33.2 g, 275 mmol) in THF (30 mL) was added dropwise to
a suspension of trans-4-hydroxy-L-proline (6; 30 g, 229 mmol)
in a mixture of THF (150 mL) and H2O (150 mL) at 0 °C (ice/
acetone), while maintaining the pH at 9 with 4 M NaOH. After
being stirred at 0 °C for 1 h, the mixture was saturated with
solid NaCl and extracted with EtOAc (100 mL). The aqueous
phase was separated, washed with further EtOAc (100 mL),
Con clu sion s
Introduction of C2-exo-unsaturation into the C-rings
of PBD dimer 4 to give 5 leads to a significant increase
in both DNA cross-linking efficiency and the ability to

Pyrrolobenzodiazepine DNA-Interactive Agent Synthesis
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 5 743
and then adjusted to pH 2 with 12 M HCl. The resulting milky
emulsion was extracted with EtOAc (2 × 100 mL), washed
with brine (200 mL), dried (MgSO4), filtered, and evaporated
(50% EtOAc/ petroleum ether 40°-60°) revealed the presence
of the major product, unreacted diol, and the presumed
disilylated derivative. Flash chromatography (20-100% EtOAc/
petroleum ether 40°-60°) gave the monosilylated compound
in vacuo to give the allyl carbamate 7 as a viscous oil (42.6 g,
10 as a pale yellow oil (13.85 g, 52%): [R]21 ) -58.6° (c )
1
87%): [R]20 ) -62.1° (c ) 0.69, MeOH); H NMR (CDCl3
+
D
D
1.14, CHCl3); 1H NMR (CDCl3) (rotamers) δ 6.01-5.86 (m, 1H),
5.34-5.18 (m, 2H), 4.59-4.49 (m, 3H, Alloc and H11a), 4.06-
3.50 (m, 5H, H3, H2 and H11), 2.20-2.01 (m, 2H, H1), 0.87
(s, 9H, SiC(CH3)3), 0.00 (s, 6H, Si(CH3)2); 13C NMR (CDCl3)
(rotamers) δ 155.0, 133.1, 117.6/117.1, 70.3/69.7 (C2), 65.9/65.6,
63.9/62.8 (C11), 57.8/57.4 (C11a), 55.7/55.2 (C3), 37.3/36.6 (C1),
25.9 (SiC(CH3)3), 18.2 (SiC(CH3)3), -5.5 (Si(CH3)2); MS (EI)
m/z (rel intensity) 316 ([M + H]+•, 29), 315 ([M]+, 4), 300 (26),
284 (4), 261 (8), 260 (50), 259 (100), 258 (100), 218 (13), 215
(10), 214 (52), 200 (12), 170 (100), 156 (40), 126 (58), 115 (33),
108 (41), 75 (35); IR (neat) 3422 (br), 2954, 2858, 1682, 1467,
DMSO-d6) (rotamers) δ 5.98-5.81 (m, 1H), 5.40-5.14 (m, 2H),
4.64-4.42 (m, 4H, Alloc, H2 and H11a), 3.82-3.51 (m, 2H, H3),
2.34-2.08 (m, 2H, H1); 13C NMR (CDCl3 + DMSO-d6) (rota-
mers) δ 175.0/174.5 (C11), 155.1/154.6 (C5), 132.9/132.8, 117.6/
116.7, 69.5/68.8 (C2), 65.9/65.8, 58.0/57.7 (C11a), 55.0/54.5
(C3), 39.3/38.3 (C1); MS (EI) m/z (rel intensity) 215 ([M]+•, 10)
197 (12), 170 (100), 152 (24), 130 (97), 126 (34), 112 (50), 108
(58), 86 (11), 68 (86), 56 (19); IR (neat) 3500-2100 (br), 1745,
1687, 1435, 1415, 1346, 1262, 1207, 1174, 1133, 1082 cm-1
;
HRMS [M]+• Calcd for C9H13NO5 m/z 215.0794, Obsd m/z
215.0791.
1434, 1412, 1358, 1330, 1255, 1196, 1180, 1120, 1054 cm-1
;
Meth yl (2S,4R)-N-(Allyloxyca r bon yl)-4-h yd r oxyp yr r o-
lid in e-2-ca r boxyla te (8). A solution of 7 (43 g, 200 mmol) in
MeOH (300 mL) was treated with concentrated H2SO4 (4.5 mL)
and heated at reflux for 2 h. After cooling to room temperature,
the reaction mixture was treated with TEA (43 mL) and
concentrated in vacuo. The residue was dissolved in EtOAc
(300 mL), washed with brine (200 mL), dried (MgSO4), filtered,
and evaporated under reduced pressure to give a viscous oil.
Purification by flash chromatography (40% EtOAc/petroleum
ether 40°-60°) gave the pure ester 8 as a yellow oil (19.6 g,
HRMS [M]+• Calcd for C15H29NO4Si m/z 315.1866, Obsd m/z
315.1946.
(2S)-N-(Allyloxyca r b on yl)-2-(ter t-b u t yld im et h ylsily-
loxym eth yl)-4-oxop yr r olid in e (11). Meth od A. A solution
of DMSO (12.9 mL, 14.3 g, 183 mmol) in CH2Cl2 (90 mL) was
added dropwise to a solution of oxalyl chloride (45.1 mL of a
2.0 M solution in CH2Cl2, 90.2 mmol) at -60 °C (dry ice/
acetone) under a N2 atmosphere. After the mixture was stirred
at -70 °C for 30 min, a solution of the alcohol 10 (25.8 g, 81.9
mmol) in CH2Cl2 (215 mL) was added dropwise at -60 °C. The
reaction mixture was allowed to stir for 1.5 h at -70 °C and
was then treated dropwise with TEA (57.2 mL, 41.5 g, 410
mmol) and allowed to warm to 10 °C. The reaction mixture
was then treated with brine (150 mL) and acidified to pH 3
with 12 M HCl. The organic phase was separated and washed
with brine (200 mL), then dried (MgSO4), filtered, and
concentrated in vacuo. Purification of the resulting orange oil
by flash chromatography (40% EtOAc/petroleum ether 40°-
60°) furnished the ketone 11 as a pale yellow oil (24.24 g, 95%).
43%): [R]23 ) -79.0° (c ) 0.35, CHCl3); 1H NMR (CDCl3)
D
(rotamers) δ 5.98-5.78 (m, 1H), 5.35-5.16 (m, 2H), 4.65-4.45
(m, 4H, Alloc, H2 and H11a), 3.75/3.72 (s, 3H, CO2CH3), 3.70-
3.54 (m, 2H, H3), 3.13/3.01 (br s, 1H, OH), 2.39-2.03 (m, 2H,
H1); 13C NMR (CDCl3) (rotamers) δ 173.4/173.2 (C11), 155.0/
154.6, 132.6/132.4, 117.6/117.3, 70.0/69.2 (C2), 66.2, 57.9/57.7
(C11a), 55.2/54.6 (C3), 52.4 (CO2CH3), 39.1/38.4 (C1); MS (EI)
m/z (rel intensity) 229 ([M]+•, 7), 170 (100), 144 (12), 126 (26),
108 (20), 68 (7), 56 (8); IR (neat) 3438 (br), 2954, 1750, 1694,
1435, 1413, 1345, 1278, 1206, 1130, 1086 cm-1; HRMS [M]+•
Calcd for C10H15NO5 m/z 229.0950, Obsd m/z 229.0940.
Meth od B. A solution of 10 (4.5 g, 14.3 mmol) in CH2Cl2
(67.5 mL) was treated with CH3CN (7.5 mL), powdered
molecular sieves (4 Å, 3.54 g), and NMO (2.4 g, 20.5 mmol).
After the mixture was stirred for 15 min at room temperature,
TPAP (0.24 g, 0.7 mmol) was added to the reaction mixture,
and a green to black color change was observed. The reaction
mixture was stirred for a further 2.5 h until TLC (50% EtOAc/
petroleum ether 40°-60°) indicated complete consumption of
starting material. Concentration of the reaction mixture in
vacuo followed by flash chromatography (50% EtOAc/petro-
leum ether) gave the pure ketone 11 as a yellow oil (4.1 g,
(2S,4R)-N-(Allyloxyca r bon yl)-4-h yd r oxy-2-(h yd r oxym -
eth yl)p yr r olid in e (9). A solution of ester 8 (19.5 g, 85 mmol)
in THF (326 mL) was cooled to 0 °C and treated portion-wise
with LiBH4 (2.78 g, 128 mmol). The stirred reaction mixture
was allowed to warm to room temperature over 2.5 h under a
N2 atmosphere, after which TLC (50% EtOAc/petroleum ether
40°-60°) revealed the complete consumption of starting mate-
rial. The mixture was cooled to 0 °C and carefully treated with
water (108 mL) and then 2 M HCl (54 mL). After concentration
in vacuo, the mixture was adjusted to pH 7 with 10 M NaOH,
saturated with solid NaCl, and then extracted with EtOAc (5
× 100 mL). The combined organic phase was washed with
brine (200 mL), dried (MgSO4), filtered, and evaporated in
92%): [R]22 ) +1.25° (c ) 10.0, CHCl3); 1H NMR (CDCl3)
D
(rotamers) δ 6.00-5.90 (m, 1H), 5.35-5.22 (m, 2H), 4.65-4.63
(m, 2H), 4.48-4.40 (m, 1H, H11a), 4.14-3.56 (m, 4H, H3 and
H11), 2.74-2.64 (m, 1H, H1trans), 2.46 (d, 1H, J ) 18.7 Hz,
H1cis), 0.85 (s, 9H, SiC(CH3)3), 0.00 (s, 6H, Si(CH3)2); 13C NMR
(CDCl3) (rotamers) δ 210.1 (C2), 154.1, 132.7, 118.0/117.7, 66.0/
65.8, 65.0 (C11), 55.7 (C11a), 53.6 (C3), 40.8/40.1 (C1), 25.7
(SiC(CH3)3), 18.1 (SiC(CH3)3), -5.7/-5.8 (Si(CH3)2); MS (CI) m/z
(rel intensity) 314 ([M + H]+•, 100), 256 (65); IR (neat) 2930,
2858, 1767, 1709, 1409, 1362, 1316, 1259, 1198, 1169, 1103,
1016 cm-1; HRMS [M]+• Calcd for C15H27NO4Si m/z 313.1710,
Obsd m/z 313.1714.
vacuo to furnish the pure diol 9 as a colorless oil (16.97 g,
1
99%): [R]24 ) -57.0° (c ) 0.61, CHCl3); H NMR (CDCl3) δ
D
6.01-5.87 (m, 1H), 5.36-5.20 (m, 2H), 4.84 (br s, 1H, H11),
4.60 (d, 2H, J ) 5.31 Hz), 4.39 (br s, 1H, H11), 4.18-4.08 (m,
1H, H2), 3.90-3.35 (m, 4H, H3, H11a, and OH), 3.04 (br s,
1H, OH), 2.11-2.03 (m, 1H, H1), 1.78-1.69 (m, 1H, H1); 13C
NMR (CDCl3) δ 157.1, 132.6, 117.7, 69.2 (C2), 66.4/66.2 (Alloc
and C11), 59.2 (C11a), 55.5 (C3), 37.3 (C1); MS (EI) m/z (rel
intensity) 201 ([M]+•, 2), 170 (100), 144 (6), 126 (26), 108 (20),
68 (9); IR (neat) 3394 (br, OH), 2946, 2870, 1679, 1413, 1339,
1194, 1126, 1054 cm-1; HRMS [M]+• Calcd for C9H15NO4 m/z
201.1001, Obsd m/z 201.1028.
(2S)-N-(Allyloxyca r b on yl)-2-(ter t-b u t yld im et h ylsily-
loxym eth yl)-4-m eth ylid en ep yr r olid in e (12). Potassium
tert-butoxide (41.0 mL of a 0.5 M solution in THF, 20.5 mmol)
was added dropwise to a suspension of methyltriphenylphos-
phonium bromide (7.29 g, 20.4 mmol) in THF (20 mL) at 0 °C
under a N2 atmosphere. After the mixture was stirred for 2 h
at 0 °C, a solution of ketone 11 (3.20 g, 10.2 mmol) in THF (10
mL) was added dropwise, and the reaction mixture was
allowed to warm to room temperature. After being stirred for
30 min, the reaction mixture was diluted with EtOAc (150 mL)
and water (150 mL), and the organic layer was separated,
washed with brine, dried (MgSO4), filtered, and evaporated
in vacuo to give a yellow oil. Purification by flash chromatog-
raphy (5% EtOAc/petroleum ether 40°-60°) gave the pure
(2S,4R)-N-(Allyloxyca r bon yl)-2-(ter t-bu tyld im eth ylsi-
lyloxym eth yl)-4-h yd r oxyp yr r olid in e (10). A solution of diol
9 (16.97 g, 84 mmol) and TEA (11.7 mL, 8.5 g, 84 mmol) in
CH2Cl2 (235 mL) was allowed to stir for 15 min at room
temperature and then treated with tert-butyldimethylsilyl
chloride (TBDMSCl, 9.72 g, 64 mmol) and DBU (16.8 mmol,
2.51 mL, 2.56 g). The reaction mixture was allowed to stir for
a further 16 h under a N2 atmosphere. After dilution with
EtOAc (500 mL), the organic phase was washed with saturated
aqueous NH4Cl (160 mL) and brine (160 mL), then dried
(MgSO4), filtered, and evaporated in vacuo to give an oil. TLC
olefin 12 as a colorless oil (2.76 g, 87%): [R]21 ) -22.2° (c )
D

744 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 5
Gregson et al.
0.25, CHCl3); 1H NMR (CDCl3) (rotamers) δ 6.02-5.87 (m, 1H),
5.31 (ddd, 1H, J ) 1.7, 3.1, 17.2 Hz), 5.21 (dd, 1H, J ) 1.5,
10.4 Hz), 4.99-4.61 (m, 2H, H12), 4.60 (d, 2H, J ) 4.9 Hz),
4.19-3.98 (m, 2H, H11), 3.93-3.87 (m, 1H, H11a), 3.66-3.42
(m, 2H, H3), 2.80-2.56 (m, 2H, H1), 0.87 (s, 9H, SiC(CH3)3),
0.03-0.02 (m, 6H, Si(CH3)2); 13C NMR (CDCl3) (rotamers) δ
154.4, 145.1/144.1 (C2), 133.1 (C7), 117.5/117.2, 107.5/106.9
(C12), 65.8/65.6, 63.7/63.1 (C11), 58.7/58.3 (C11a), 51.1 (C3),
34.9/34.2 (C1), 25.8 (SiC(CH3)3), 18.2 (SiC(CH3)3), -5.5 (Si-
(CH3)2); MS (CI) m/z (rel intensity) 312 ([M + H]+•, 82), 296
(9), 279 (5), 255 (20), 254 (100), 168 (8), 122 (14); IR (neat)
3083, 2954, 2847, 1709, 1533, 1467, 1404, 1360, 1310, 1252,
1207, 1174, 1103, 1076, 1006 cm-1; HRMS [M + H]+• Calcd
for C16H30NO3Si m/z 312.1995, Obsd (FAB) m/z 312.1976.
C11′), 60.2 (C11a/C11a′), 58.1/56.6 (OCH3 at C7/C7′), 52.8/50.5
(C3/C3′), 35.0/33.9 (C1/C1′), 30.8/28.6 (C14), 25.8/25.7 (SiC-
(CH3)3), 18.2 (SiC(CH3)3), -5.5/-5.6 (Si(CH3)2); MS (EI) m/z (rel
intensity) 885 ([M]+•, 7), 828 (100), 740 (20), 603 (3), 479 (26),
391 (27), 385 (25), 301 (7), 365 (10), 310 (14), 226 (8), 222 (13),
170 (21), 168 (61), 82 (39), 75 (92); IR (Nujol) 2923, 2853, 2360,
1647, 1587, 1523, 1461, 1429, 1371, 1336, 1277, 1217, 1114,
1061, 1021 cm-1; HRMS [M + H]+• Calcd for C43H65N4O12Si2
m/z 885.4138, Obsd (FAB) m/z 885.4165.
(2S)-1,1′-[[(P r op a n e-1,3-d iyl)d ioxy]bis[(2-n itr o-5-m eth -
oxy-1,4-p h en ylen e)ca r b on yl]]b is[2-(h yd r oxym et h yl)-4-
m eth ylid en ep yr r olid in e] (16). A solution of TBAF (3.98 mL
of a 1 M solution in THF, 3.98 mmol) was added to the bis-
silyl ether 15 (1.41 g, 1.59 mmol) in THF (35 mL) at 0 °C. The
reaction mixture was allowed to warm to room temperature,
and saturated aqueous NH4Cl (120 mL) was added after 30
min. The reaction mixture was extracted with EtOAc (3 × 80
mL), and the combined extracts were washed with brine (80
mL), then dried (MgSO4), and evaporated in vacuo to give a
dark oil. Flash chromatography (97% CHCl3/MeOH) provided
the diol 16 as a pale orange solid (0.98 g, 94%): [R]19D ) -31.9°
(2S )-2-(t er t -Bu t yld im e t h ylsilyloxym e t h yl)-4-m e t h -
ylid en ep yr r olid in e (13). A catalytic amount of PdCl2(PPh3)2
(92 mg, 0.131 mmol) was added to a mixture of the allyl
carbamate 12 (1.0 g, 3.22 mmol) and H2O (0.34 mL, 18.9 mmol)
in CH2Cl2 (30 mL). After the mixture was stirred for 5 min at
room temperature, Bu3SnH (0.96 mL, 1.04 g, 3.57 mmol) was
added rapidly in one portion. A mildly exothermic reaction
occurred immediately with vigorous gas evolution. The mixture
was then stirred for 16 h at room temperature under N2 when
TLC (50% EtOAc/petroleum ether 40°-60°) revealed formation
of the free amine. After dilution with CH2Cl2 (30 mL), the
solution was dried (MgSO4), filtered, and evaporated in vacuo
to give an orange oil. Purification by flash chromatography
(50-100% EtOAc/petroleum ether 40°-60°) afforded the pyr-
rolidine 13 as a pale orange oil (0.56 g, 77%): [R]21 ) -3.90°
(c ) 5.0, CHCl3); 1H NMR (CDCl3) δ 4.93/4.90 (2 × Dt, each 1H,
J ) 2.0 Hz, H12), 3.68-3.46 (m, 4H, H11 and H3), 3.30-3.21
(m, 1H, H11a), 2.53-2.41 (m, 2H, H1 and NH), 2.26-2.17 (m,
1H, H1), 0.90 (s, 9H, SiC(CH3)3), 0.06 (s, 6H, Si(CH3)2); 13C
NMR (CDCl3) δ 150.0 (C2), 104.7 (C12), 64.7 (C11), 60.5 (C11a),
51.3 (C3), 34.9 (C1), 25.9 (SiC(CH3)3), 18.3 (SiC(CH3)3), -5.4
(Si(CH3)2); MS (EI) m/z (rel intensity) 227 ([M]+•, 8), 212 (6),
170 (36), 96 (8), 82 (100), 75 (11); IR (neat) 3550-3100, 3074,
2929, 2857, 1664, 1472, 1424, 1391, 1380, 1361, 1255, 1190,
1101, 1006, cm-1; HRMS [M + H]+• Calcd for C12H26NOSi m/z
228.1784, Obsd (FAB) m/z 228.1777.
1
(c ) 0.09, CHCl3); H NMR (CDCl3) (rotamers) δ 7.75/7.71 (2
× s, 2H, H6/H6′), 6.96/6.84 (2 × s, 2H, H9/H9′), 5.08, 5.02/
4.88 (3 × br s, 4H, H12/H12′), 4.61-4.50 (m, 2H, H11a/H11a′),
4.35-4.33 (m, 4H, H13/H13′), 4.02-3.65 (m, 14H, H11/H11′,
H3/H3′ and 2 × OCH3 at C7/C7′), 2.88-2.43 (m, 6H, H1/H1′
and H14); 13C NMR (CDCl3) (rotamers) δ 167.9/166.9 (C5/C5′),
154.9/154.3 (Cquat), 148.4/148.2 (Cquat), 143.3/142.6 (Cquat), 137.2/
137.0 (Cquat), 127.6/127.3 (Cquat), 109.1 (C6/C6′), 108.4 (C12/
C12′), 108.2 (C9/C9′), 65.6/65.4 (C13/C13′), 64.5/63.3 (C11/
C11′), 60.5/60.0 (C11a/C11a′), 56.8/56.7 (OCH3 at C7/C7′), 52.9
(C3/C3′), 35.0/34.3 (C1/C1′), 29.6/28.6 (C14); MS (FAB) (rel
intensity) 657 ([M + H]+•, 10), 639 (2), 612 (1), 544 (4), 539
(1), 449 (16), 433 (9), 404 (8), 236 (32), 166 (65), 151 (81), 112
(82), 82 (100); IR (Nujol) 3600-3200 (br), 2923, 2853, 2360,
1618, 1582, 1522, 1459, 1408, 1375, 1335, 1278, 1218, 1061
cm-1; HRMS [M + H]+• Calcd for C31H37N4O12 m/z 657.2408,
Obsd (FAB) m/z 657.2388.
(2S)-1,1′-[[(P r opan e-1,3-diyl)dioxy]bis[(2-am in o-5-m eth -
oxy-1,4-p h en ylen e)ca r bon yl]]-bis[2-(h yd r oxym eth yl)-4-
m eth ylid en ep yr r olid in e] (17). A mixture of 16 (0.98 g, 1.49
mmol) and SnCl2‚2H2O (3.36 g, 14.9 mmol) in MeOH (35 mL)
was heated at reflux, and the progress of the reaction was
monitored by TLC (90% CHCl3/MeOH). After 45 min, the
solvent was evaporated in vacuo and the resulting residue
cooled in ice, treated carefully with saturated aqueous NaH-
CO3 (120 mL), and then diluted with EtOAc (120 mL). After
the mixture was stirred for 16 h at room temperature, the
inorganic precipitate was removed by filtration through Celite.
The organic phase was separated, washed with brine (100 mL),
then dried (MgSO4), filtered, and evaporated in vacuo to give
a brown solid. Flash chromatography (95% CHCl3/MeOH)
(2S)-1,1′-[[(P r op a n e-1,3-d iyl)d ioxy]-bis[(2-n itr o-5-m eth -
oxy-1,4-p h en ylen e)ca r bon yl]]-bis[2-(ter t-bu tyld im eth yl-
silyloxym eth yl)-4-m eth ylid en ep yr r olid in e] (15). A cata-
lytic amount of DMF (2 drops) was added to a solution of dimer
acid13 14 (0.66 g, 1.42 mmol) and oxalyl chloride (0.31 mL, 0.45
g, 3.55 mmol) in THF (12 mL), and the mixture was stirred
for 16 h under a N2 atmosphere. Following the removal of
volatiles in vacuo, the resulting bis-acid chloride was redis-
solved in THF (10 mL) and added dropwise to a stirred mixture
of amine 13 (0.65 g, 2.86 mmol), H2O (0.84 mL), and TEA (0.83
mL, 0.60 g, 5.93 mmol) in THF (2 mL) maintained at 0 °C
under N2. The reaction mixture was allowed to warm to room
temperature and then stirred for a further 2 h, at which time
TLC (EtOAc) revealed the complete consumption of starting
material. After removal of the THF by evaporation in vacuo,
the residue was partitioned between H2O (100 mL) and EtOAc
(100 mL), and the separated aqueous layer washed with EtOAc
(3 × 50 mL). The combined organic phase was then washed
with saturated aqueous NH4Cl (100 mL) and brine (100 mL),
and then dried (MgSO4), filtered, and concentrated in vacuo
to afford a dark orange oil. Purification by flash chromatog-
raphy (50% EtOAc/petroleum ether 40°-60°) afforded the pure
afforded the bis-amine 17 as an orange solid (0.54 g, 61%):
1
[R]19 ) -31.8° (c ) 0.30, CHCl3); H NMR (CDCl3) δ 6.74 (s,
D
2H, H6/H6′), 6.32 (s, 2H, H9/H9′), 5.00 and 4.93 (2 × br s, 4H,
H12/H12′), 4.54 (br s, 2H, H11a/H11a′), 4.24-4.14 (m, 4H,
H13/H13′), 3.98-3.50 (m, 14H, H11/H11′, H3/H3′ and 2 ×
OCH3 at C7/C7′), 2.76 (dd, 2H, J ) 8.6, 15.9 Hz, H1/H1′trans),
2.46-2.41 (m, 2H, H1/H1′cis), 2.33-2.28 (m, 2H, H14); 13C
NMR (CDCl3) δ 171.0 (C5/C5′), 151.0 (Cquat), 143.5 (Cquat), 141.3
(Cquat), 140.6 (Cquat), 112.4 (C6/C6′), 111.9 (Cquat), 107.8 (C12/
C12′), 102.4 (C9/C9′), 65.2 (C13/C13′), 65.0 (C11/C11′), 59.8
(C11a/C11a′), 57.1 (2 × OCH3 at C7/C7′), 53.3 (C3/C3′), 34.4
(C1/C1′), 29.0 (C14); MS (FAB) (rel intensity) 597 ([M]+•, 1.5),
596 ([M]+•, 13), 484 (14), 389 (10), 371 (29), 345 (5), 224 (8),
206 (44), 166 (100), 149 (24), 112 (39), 96 (34), 81 (28); IR
(Nujol) 3600-3000 (br), 3349, 2922, 2852, 2363, 1615, 1591,
1514, 1464, 1401, 1359, 1263, 1216, 1187, 1169, 1114, 1043
cm-1; HRMS [M + H]+• Calcd for C31H41N4O8 m/z 597.2924,
Obsd (FAB) m/z 597.2896.
bis-amide 15 as a pale yellow glass (0.93 g, 74%): [R]21
)
D
-51.1° (c ) 0.08, CHCl3); 1H NMR (CDCl3) (rotamers) δ 7.77/
7.74 (2 × s, 2H, H6/H6′), 6.81/6.76 (2 × s, 2H, H9/H9′), 5.09-
4.83 (m, 4H, H12/H12′), 4.60 (m, 2H, H11a/H11a′), 4.35-4.31
(m, 4H, H13/H13′), 4.08-3.74 (m, 14H, H11/H11′, H3/H3′ and
2 × OCH3 at C7/C7′), 2.72-2.45 (m, 6H, H1/H1′ and H14),
0.91/0.79 (2 × s, 18H, SiC(CH3)3), 0.09/-0.09/-0.12 (s × 3, 12H,
Si(CH3)2); 13C NMR (CDCl3) (rotamers) δ 166.2 (C5/C5′), 154.7/
154.5 (Cquat), 148.4/148.2 (Cquat), 144.1/143.2 (Cquat), 137.2
(Cquat), 128.2/127.4 (Cquat), 110.1/108.6 (C6/C6′), 109.1/108.3
(C9/C9′), 107.5 (C12/C12′), 65.7/65.5 (C13/13′), 63.9/62.6 (C11/
(2S)-1,1′-[[(P r op a n e-1,3-d iyl)d ioxy]bis[(2-a llyloxyca r -
b on yla m in o-5-m et h oxy-1,4-p h en ylen e)ca r b on yl]]b is[2-
(h yd r oxym eth yl)-4-m eth ylid en ep yr r olid in e] (18). Pyri-
dine (0.47 mL, 0.46 g, 5.82 mmol) was added to a stirred

Pyrrolobenzodiazepine DNA-Interactive Agent Synthesis
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 5 745
solution of the bis-amine 17 (0.857 g, 1.44 mmol) in CH2Cl2
(30 mL) at 0 °C. This mixture was treated dropwise with a
solution of allyl chloroformate (0.33 mL, 0.38 g, 3.15 mmol) in
CH2Cl2 (10 mL). After being stirred for 2.5 h at room temper-
ature, the reaction mixture was diluted with CH2Cl2 (60 mL),
washed with 1 M HCl (2 × 50 mL), H2O (80 mL), and brine
(80 mL), then dried (MgSO4), filtered, and evaporated in vacuo.
The residue was purified by flash chromatography (70-100%
EtOAc/petroleum ether 40°-60°) to afford the bis-carbamate
18 as a pale orange glass (0.548 g, 50%): 1H NMR (CDCl3) δ
8.58 (br s, 2H, NH), 7.56 (s, 2H, H6/H6′), 6.78 (s, 2H, H9/H9′),
6.03-5.88 (m, 2H, H17/H17′), 5.39-5.21 (m, 4H, H18/H18′),
5.00 and 4.93 (2 × br s, 4H, H12/H12′), 4.70-4.57 (m, 4H, H16/
H16′), 4.30-4.25 (m, 4H, H13/H13′), 4.17-3.90 (m, 8H, H11/
H11′and H3/H3′), 3.81-3.54 (m, 8H, H11a/H11a′ and 2 ×
OCH3 at C7/C7′), 2.76 (dd, 2H, J ) 8.5, 15.9 Hz, H1/H1′trans),
2.49-2.44 (m, 2H, H1/H1′cis), 2.36-2.28 (m, 2H, H14); 13C
NMR (CDCl3) δ 170.3 (C5/C5′), 153.8 (C15/C15′), 150.5 (Cquat),
144.8 (Cquat), 143.1 (Cquat), 132.5 (C17/C17′), 130.7 (Cquat), 118.1
(C18/C18′), 116.8 (Cquat), 110.9 (C6/C6′), 108.1 (C12/C12′) 106.9
(C9/C9′), 65.7 (C16/C16′), 65.4 (C13/C13′), 65.1 (C11/C11′), 59.8
(C11a/C11a′), 56.5 (2 × OCH3 at C7/C7′), 53.9 (C3/C3′), 34.2
(C1/C1′), 29.7/29.2 (C14); MS (FAB) (rel intensity) 765 ([M +
H]+•, 10), 652 (32), 594 (4), 539 (2), 481 (51), 441 (31), 290 (3),
249 (13), 232 (38), 192 (83), 166 (49), 149 (32), 114 (100).
5 as a pale orange glass (78 mg, 77%). This material was
repeatedly evaporated from CHCl3 in vacuo to generate the
bis-imine form: [R]21 ) +357.7° (c ) 0.07, CHCl3); reverse-
D
phase HPLC (C4 stationary phase, 65% MeOH/H2O mobile
phase, 254 nm) Rt ) 6.27 min, % peak area ) 97.5%; 1H NMR
(CDCl3) (imine form) δ 7.68 (d, 2H, J ) 4.4 Hz, H11/H11′),
7.49 (s, 2H, H6/H6′), 6.85 (s, 2H, H9/H9′), 5.20/5.17 (2 × br s,
4H, H12/H12′), 4.46-4.19 (m, 8H, H13/H13′ and H3/H3′), 3.93
(s, 6H, 2 × OCH3 at C7/C7′), 3.89-3.80 (m, 2H, H11a/H11a′),
3.12 (dd, 2H, J ) 8.6, 16.2 Hz, H1/H1′trans), 2.94 (d, 2H, J )
16.3 Hz, H1/H1′cis), 2.45-2.38 (m, 2H, H14); 13C NMR (CDCl3)
(imine form) δ 164.7 (C5/C5′), 162.6 (C11/C11′), 150.7 (Cquat),
147.9 (Cquat), 141.5 (Cquat), 140.6 (Cquat), 119.8 (Cquat), 111.5 (C6/
C6′), 110.7 (C9/C9′), 109.4 (C12/C12′), 65.4 (C13/C13′), 56.1 (2
× OCH3 at C7/C7′), 53.8 (C11a/C11a′), 51.4 (C3/C3′), 35.4 (C1/
C1′), 28.8 (C14); MS (FAB) (rel intensity) (imine form) 773 ([M
+ H + (2 × thioglycerol adduct)]+•, 3), 665 ([M + H +
thioglycerol adduct]+•, 7), 557 ([M + H]+•, 9), 464 (3), 279 (12),
257 (5), 201 (5), 185 (43), 166 (6), 149 (12), 93 (100); IR (Nujol)
3600-3100 (br), 2923, 2849, 1599, 1511, 1458, 1435, 1391,
1277, 1228, 1054, 1011 cm-1; HRMS [M + H]+• Calcd for
C31H33N4O6 m/z 557.2400, Obsd (FAB) m/z 557.2394; Calcd for
C31H32N4O6: C 65.57, H 5.85, N 9.60. Found: C 65.50, H 5.86,
N 9.56.
1,1′-[[(P r op a n e-1,3-d iyl)d ioxy]bis[(11a S)-7-m eth oxy-2-
m eth ylid en e-1,2,3,10,11,11a -h exa h yd r o-5H-p yr r olo[2,1-c]-
[1,4]ben zod ia zep in -5,11-d ion e] (21). A solution of DMSO
(0.17 mL, 0.19 g, 2.43 mmol) in CH2Cl2 (10 mL) was added
dropwise to a solution of oxalyl chloride (0.60 mL of a 2.0 M
1,1′-[[(P r op a n e-1,3-d iyl)d ioxy]bis[(11S,11a S)-10-(a lly-
loxycar bon yl)-11-h ydr oxy-7-m eth oxy-2-m eth yliden e-1,2,3,-
10,11,11a-h exah ydr o-5H-pyr r olo[2,1-c][1,4]ben zodiazepin -
5-on e] (19). A stirred solution of the bis-carbamate 18 (150
mg, 0.196 mmol) in CH2Cl2/CH3CN (12 mL, 3:1) was treated
with powdered molecular sieves (4 Å, 0.2 g) and NMO (70 mg,
0.598 mmol) at room temperature. After the mixture was
stirred at room temperature for 15 min, TPAP (7 mg, 19.9
µmol) was added and stirring continued for a further 2 h, after
which time TLC (95% CHCl3/MeOH) indicated formation of
the cyclized product and the presence of some unreacted
starting material. The mixture was treated with a further
portion of NMO (35 mg, 0.299 mmol) and TPAP (3.5 mg, 9.96
µmol), and it was allowed to stir for 30 min, when TLC showed
complete reaction. Solvent removal in vacuo afforded a black
residue which was purified by flash chromatography (98%
CHCl3/MeOH) to give the pure bis-carbinolamine derivative
19 as a white solid (47 mg, 32%): 1H NMR (CDCl3) δ 7.23 (s,
2H, H6/H6′), 6.74 (s, 2H, H9/H9′), 5.90-5.65 (m, 2H, H17/
H17′), 5.60 (d, 2H, J ) 8.2 Hz, H11/H11′), 5.26-5.07 (m, 8H,
H12/H12′ and H18/H18′), 4.67-4.10 (m, 14H, H16/H16′, H3/
H3′, H13/H13′ and 2 × OH at C11/C11′), 3.89 (s, 6H, 2 × OCH3
at C7/C7′), 3.63 (m, 2H, H11a/H11a′), 2.91 (dd, 2H, J ) 8.8,
15.8 Hz, H1/H1′trans), 2.68 (d, 2H, J ) 16.1 Hz, H1/H1′cis),
2.42-2.24 (m, 2H, H14); 13C NMR (CDCl3) δ 166.7 (C5/C5′),
150.1 (Cquat), 149.0 (Cquat), 141.7 (Cquat), 131.7 (C17/C17′), 130.6
(Cquat), 128.9 (Cquat), 128.8 (Cquat), 118.3 (C18/C18′), 114.7 (C6/
C6′), 110.7 (C9/C9′), 109.8 (C12/C12′), 85.9 (C11/C11′), 66.9
(C16/C16′), 66.0 (C13/C13′), 59.7 (C11a/C11a′), 56.1 (2 × OCH3
at C7/C7′), 50.7 (C3/C3′), 35.0 (C1/C1′), 29.7/29.1 (C14); MS
(FAB) (rel intensity) no molecular ion, 743 (16), 725 (17), 632
(13), 574 (8), 548 (13), 490 (10), 481 (9), 441 (7), 425 (6), 257
(12), 232 (20), 192 (46), 166 (52), 149 (100), 91 (59); IR (Nujol)
3234 (br), 2923, 2853, 2361, 1707, 1604, 1515, 1464, 1410,
solution in CH2Cl2, 1.20 mmol) at -45 °C under
a N2
atmosphere. After the mixture was stirred at -45 °C for 1 h,
a solution of 18 (0.41 g, 0.54 mmol) in CH2Cl2 (10 mL) was
added dropwise at -45 °C. The reaction mixture was allowed
to stir for 2 h at -45 °C before treating dropwise with TEA
(0.76 mL, 0.55 g, 5.40 mmol) in CH2Cl2 (10 mL) and allowing
to warm to 10 °C. TLC (95% CHCl3/MeOH) indicated complete
consumption of starting material. The mixture was then
treated with brine (100 mL) and acidified to pH 3 with 12 M
HCl. The organic phase was separated and washed with brine
(100 mL), then dried (MgSO4), filtered, and concentrated in
vacuo to afford 20 as an orange oil (0.21 g). Without further
purification, the intermediate 20 was dissolved in CH2Cl2 (20
mL), cooled to 0 °C, and treated sequentially with PPh3 (7.6
mg, 29.0 µmol), pyrrolidine (41 mg, 0.576 mmol), and a
catalytic amount of tetrakis(triphenylphosphine)palladium
(16.6 mg, 14.4 µmol). The mixture was allowed to warm to
room temperature and was then stirred for 2 h, at which point
TLC (95% CHCl3/MeOH) revealed that the reaction was
complete. The solvent was evaporated in vacuo, and the
residue was purified by flash chromatography (97% CHCl3/
MeOH) to afford the tetralactam 21 as a yellow glass (70 mg,
43%): [R]22 ) +274.1° (c ) 0.06, CHCl3); 1H NMR (CDCl3) δ
D
8.95 (br s, 2H, D2O exchangeable, NH), 7.40 (s, 2H, H6/H6′),
6.62 (s, 2H, H9/H9′), 5.14/5.08 (2 × br s, 4H, H12/H12′), 4.38
(br d, 2H, J ) 16.3 Hz, 1 × H3/H3′), 4.20-4.10 (m, 8H, 1 ×
H3/H3′, H11a/H11a′ and H13/H13′), 3.85 (s, 6H, 2 × OCH3 at
C7/C7′), 3.43 (br d, 2H, J ) 15.9 Hz, 1 × H1/H1′), 2.84-2.75
(m, 2H, 1 × H1/H1′), 2.34-2.30 (m, 2H, H14); 13C NMR
(CDCl3) δ 170.1 (C11/C11′), 165.2 (C5/C5′), 151.5 (Cquat), 146.8
(Cquat), 141.4 (Cquat), 129.8 (Cquat), 119.0 (Cquat), 112.4 (C6/C6′),
109.0 (C12/C12′), 105.3 (C9/C9′), 65.4 (C13/C13′), 56.8/56.3
(C11a/C11a′ and 2 × OCH3 at C7/C7′), 51.5 (C3/C3′), 31.8 (C1/
C1′), 28.5 (C14); MS (EI) (rel intensity) 588 ([M]+•, 39), 570
(100), 554 (5), 461 (13), 314 (27), 300 (17), 274 (24), 257 (15),
236 (19), 166 (12), 82 (19); IR (Nujol) 3700-3200, 2923, 2854,
1697, 1607, 1518, 1491, 1463, 1435, 1377, 1267, 1227, 1183,
1119, 1020 cm-1; HRMS [M + H]+• Calcd for C31H33N4O8 m/z
589.2298, Obsd (FAB) m/z 589.2282.
1377, 1302, 1267, 1205, 1163, 1120, 1045 cm-1
.
1,1′-[[(P r op a n e-1,3-d iyl)d ioxy]bis[(11a S)-7-m eth oxy-2-
m eth ylid en e-1,2,3,11a -tetr a h yd r o-5H-p yr r olo[2,1-c][1,4]-
ben zod ia zep in -5-on e] (5). A catalytic amount of tetrakis-
(triphenylphosphine)palladium (11 mg, 9.52 µmol) was added
to a stirred solution of bis-carbamate 19 (139 mg, 0.183 mmol),
Ph3P (4.8 mg, 18.3 µmol), and pyrrolidine (27 mg, 0.380 mmol)
in CH2Cl2/CH3CN (13 mL, 10:3 v/v) at 0 °C under a N2
atmosphere. The reaction mixture was allowed to warm to
room temperature, and progress of the reaction was monitored
by TLC (95% CHCl3/MeOH). After 135 min, TLC revealed the
complete consumption of starting material and formation of a
highly fluorescent product. The solvent was evaporated in
vacuo, and the resulting residue was purified by flash chro-
matography (98% CHCl3/MeOH) to give the target PBD dimer
Th er m a l Den a tu r a tion Stu d ies. The compounds were
subjected to DNA thermal denaturation (melting) studies29
using calf thymus DNA (CT-DNA, type-I, highly polymerized
sodium salt; 42% G+C [Sigma]) at a fixed concentration of 100
µM (DNAp), determined using an extinction coefficient of 6600
(M phosphate)-1 cm-1 at 260 nm.37 Aqueous solutions were

746 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 5
Gregson et al.
prepared in pH 7.00 ( 0.01 buffer containing 10 mM NaH2-
PO4/Na2HPO4 and 1 mM EDTA. Working solutions containing
CT-DNA and the test compound (20 µM) were incubated at
37.0 ( 0.1 °C for 0-18 h (or 48 h) using an external water
bath. Samples were monitored at 260 nm using a Varian-Cary
NaOAc/20 mM EDTA/100 µg/mL tRNA). After precipitation
of the DNA by addition of EtOH (3 volumes) followed by
centrifugation and the removal of supernatant, the pellet was
dried by lyophilization.
Electr op h or esis. Samples were dissolved in strand sepa-
ration buffer (10 µL of 30% DMSO/1 mM EDTA/0.04% bro-
mophenol blue/0.04% xylene cylanol), heated at 90 °C for 2
min, and then chilled immediately in an ice-water bath prior
to loading. Control non-denaturated samples were dissolved
in 6% sucrose/0.04% bromophenol blue (10 µL) and loaded
directly. Electrophoresis was performed on a 20 cm 0.8%
submerged horizontal agarose gel at 40 V for 16 h using a 40
mM Tris/20 mM HOAc/2 mM EDTA gel running buffer.
Au tor a d iogr a p h y. Gels were dried at 80 °C on one layer
of Whatman 3 mm and one layer of DE81 filter papers, with
a vacuum-connected Bio-Rad Model 583 gel drier. Autorad-
iography was performed with Hyperfilm MP (Amersham plc)
for 4 h at -70 °C using a DuPont-Cronex Lighting-Plus
intensifying screen, and the percentage double-stranded cross-
linked DNA was calculated.
Molecu la r Mod elin g Stu d ies. The strategy and protocol
used for the modeling studies were adapted from our recent
studies of covalent DNA-PBD adduct formation.16,32 Two host
DNA 12-mer duplexes of sequence 5′-CGCAGATCTGCG and
5′-CGCTCTAGAGCG (where the spanned and G-G cross-
linked sequences and the alkylated guanine bases are high-
lighted) were selected for energy calculations as each of the
embedded tracts offer optimal 5′-AGA sites for alkylation by
a PBD unit. The symmetric 5′AGATCT duplex core tract has
been shown previously to be a favored sequence for cross-
linking by the dimer 4.16 We have recently shown that 4 and
5 can form interstrand DNA cross-links at sites with 5′-GATC
and 5′-CTAG duplex sequences (unpublished results). By
analogy, the “reverse” cross-link 5′-TCTAGA core sequence
retains the preferred 5′-AGA sequence at each site, but the
PBD orientations are reversed with respect to the modified
guanines. Each strand was terminated with an alternating G/C
base stretch of sufficient length to confer duplex stability
during extended dynamics simulations, particularly to prevent
strand separation or “fraying”.
400 Bio spectrophotometer fitted with
a Peltier heating
accessory. Heating was applied at a rate of 1 °C/min in the
45-98 °C temperature range, with optical and temperature
data sampling at 200 ms intervals. A separate experiment was
carried out using buffer alone, and this baseline was sub-
tracted from each DNA melting curve. Optical data were
imported into the Origin 5 computer package (MicroCal Inc.,
Northampton, MA) for analysis. DNA helix f coil transition
temperatures (Tm) were determined at the midpoint of the
normalized melting profiles using a published procedure.29
Results are given as the mean ( standard deviation for at least
three determinations. Ligand-induced alterations in DNA
melting behavior (∆Tm) are given by ∆Tm ) Tm(DNA + ligand)
- Tm(DNA), where the Tm value for the free CT-DNA is 67.83
( 0.06 °C (averaged from >50 runs). All compounds were
dissolved in HPLC-grade MeOH to give working solutions
containing e0.6% v/v MeOH; Tm results were corrected for the
effects of MeOH co-solvent by using a linear correction term.
Other [DNAp]/[ligand] molar ratios (i.e., 50:1 and 100:1) were
examined in the case of 5 to ensure that the fixed 5:1 ratio
used in this assay did not result in saturation of the host DNA
duplex.
DNA Cr oss-Lin k in g Assa y. The agarose gel electrophore-
sis assay was carried out with 5 using pBR322 DNA and T4
polynucleotide kinase (PNK) purchased from Northumbria
Biological Ltd. and used as supplied. Bacterial alkaline phos-
phatase (BAP) and Hind III enzymes were purchased from
BRL and used as supplied.
Lin ea r iza tion of p BR322 DNA. A mixture of circular
pBR322 DNA (10 µg in 20 µL), React 2 aqueous buffer (3 µL,
[500 mM Tris-HCl, pH 8, 100 mM MgCl2, 500 mM NaCl]),
Hind III (3 µL, 30 units), and water (4 µL) was incubated at
37 °C for 1 h. Aqueous 4 M NaOAc was added to give a 0.3 M
solution, then EtOH (95% v/v, 3 volumes) was added, and the
mixture was chilled. After centrifugation and removal of
supernatant, the pellet was lyophilized.
The interaction of dimers 4 and 5 together with their
corresponding monomeric subunits were examined with each
host DNA duplex sequence. The PBD monomers were also
modeled using the 5′-CGCAGAGCG duplex to enable compari-
son with a single optimized and embedded 5′-AGA alkylation
site and to remove the influence of any asymmetric flanking
base sequence. The C11(S) geometry was adopted for all PBD
molecules studied, as this stereochemistry is known to lead
to energetically favored adducts with the exocyclic C2-NH2
group of an embedded guanine in the context of a B-DNA
duplex.32 Thus, the forward and reverse cross-links examined
can be described as 3S-3S and 5S-5S structures due to the
enforced head-to-head tethering through the distal A-rings of
the PBD subunits. Here, the numbering used refers to the
orientation of the aromatic A-ring with respect to the alkylated
strand. However, as both 3S and 5S alignments can be
achieved for a PBD monomer, this allows an estimation of the
energy penalty resulting from their symmetric tethering in a
dimer ligand.
In Vitr o Cytotoxicity Stu d ies. In vitro cytotoxicity was
evaluated using the human ovarian carcinoma cell lines
SKOV-3, A2780, and CH1, together with the cisplatin-resistant
counterparts of the A2780 and CH1 lines (A2780cisR and
CH1cisR, respectively). Viable cells were seeded in growth
medium (160 µL) into 96-well microtiter plates and allowed
to attach overnight. The PBDs 4, 5, and 21, and cisplatin, were
dissolved in DMSO (to give a 20 mM concentration in each
case) immediately prior to adding to the cells in quadruplicate
wells. Final drug concentrations in the wells ranged from 100
µM to 2.5 nM as follows: 100, 25, 10, 2.5, 1 µM, and 250, 100,
25, 10, 2.5 nM. This was achieved by diluting the drugs in
growth medium and then adding 40 µL to the existing well
volume of 160 µL to give the final concentrations stated above.
After 4 days (96 h), the medium was removed and the
Dep h osp h or yla tion of Lin ea r ized DNA. A mixture of
linearized DNA in water (15 µL), BAP buffer (4 µL, [50 mM
Tris-HCl, pH 8, 600 mM NaCl]), and BAP (1 µL, 100 units)
was incubated at 65 °C for 1 h. After extraction with Tris/
EDTA buffer (10 mM Tris-HCl, pH 7.5, 1 mM EDTA)/saturated
phenol (1:1 v/v), the aqueous layer was collected, and the
phenol layer back-extracted with buffer. The combined aque-
ous phase was extracted twice with CHCl3/isoamyl alcohol
(24:1 v/v), then aqueous 4 M NaOAc was added to give a 0.3
M solution, followed by EtOH (3 volumes). After chilling and
centrifugation, the supernatant was removed and the pellet
lyophilized.
5′-En d La belin g. The resultant dephosphorylated DNA
was 5′-labeled with [γ-32P]-ATP (5000 Ci/mmol, Amersham
plc). A mixture of the DNA (10 µg in 30 µL water), forward
reaction buffer (9 µL, [300 mM Tris-HCl, pH 7.8, 75 mM
2-thioethanol, 50 mM MgCl2, 1.65 µM ATP]), [γ-32P]-ATP (3
µL, 30 µCi), and PNK (4 µL, 24 units) was incubated at 37 °C
for 30 min. Aqueous NH4OAc (7.5 M, 0.5 volume) and EtOH
(2 volumes) were added, and the mixture was chilled and
centrifuged (13000 rpm/10 min). The supernatant was removed
and the pellet resuspended in 20 µL Tris/EDTA buffer (10 mM
Tris-HCl, pH 7.5, 1 mM EDTA, 0.3 M NaOAc). After repre-
cipitation with EtOH (60 µL) and centrifugation, the super-
natant was removed and the pellet washed with EtOH (20 µL)
and lyophilized. The DNA was then resuspended in sterile
double-distilled water at 1 µg/µL. Approximately 10 ng of
labeled DNA was used for each experiment.
Rea ction P r otocols. Reactions were performed with 4, 5,
and melphalan at concentrations of 0.001, 0.003, 0.01, 0.03,
0.1, 0.3, 1.0, 3.0, and 10 µM in 25 mM triethanolamine/1 mM
EDTA at pH 7.2 and 37 °C. Reactions were terminated after
2 h by the addition of an equal volume of stop solution (0.6 M

Pyrrolobenzodiazepine DNA-Interactive Agent Synthesis
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 5 747
(13) Thurston, D. E.; Bose, D. S.; Thompson, A. S.; Howard, P. W.;
Leoni, A.; Croker, S. J .; J enkins, T. C.; Neidle, S.; Hartley, J .
A.; Hurley, L. H. Synthesis of Sequence-Selective C8-Linked
Pyrrolo[2,1-c][1,4]benzodiazepine DNA Interstrand Cross-Link-
ing Agents. J . Org. Chem. 1996, 61, 8141-8147.
(14) Hartley, J . A.; Berardini, M. D.; Souhami, R. L. An Agarose-Gel
Method for the Determination of DNA Interstrand Cross-Linking
Applicable to the Measurement of the Rate of Total and 2nd-
Arm Cross-Link Reactions. Anal. Biochem. 1991, 193, 131-134.
(15) Mountzouris, J . A.; Wang, J . J .; Thurston, D.; Hurley, L. H.
Comparison of a DSB-120 DNA Interstrand Cross-Linked Ad-
duct with the Corresponding Bis-Tomaymycin Adduct - an
Example of a Successful Template-Directed Approach to Drug
Design Based Upon the Monoalkylating Compound Tomaymy-
cin. J . Med. Chem. 1994, 37, 3132-3140.
remaining cells were fixed using 10% trichloroacetic acid on
ice for 30 min. Wells were then washed 3-4 times with tap
water and air-dried overnight, and 100 µL of 0.4% sulfor-
hodamine B (dissolved in 1% glacial HOAc) was added to each
well. Staining was allowed to continue for 10-15 min; the wells
were then washed 3-4 times with 1% acetic acid and air-dried,
and Tris base (100 µL of 10 mM) was added to each one. The
plates were then shaken and absorbance readings taken at
540 nm using a plate reader. From plots of concentration
versus percentage absorbance (compared to 8 untreated wells),
IC50 values were calculated using the Quattro-Pro software
package.
(16) J enkins, T. C.; Hurley, L. H.; Neidle, S.; Thurston, D. E.
Structure of a Covalent DNA Minor-Groove Adduct With a
Pyrrolobenzodiazepine Dimer - Evidence For Sequence-Specific
Interstrand Cross-Linking. J . Med. Chem. 1994, 37, 4529-4537.
(17) Walton, M. I.; Goddard, P.; Kelland, L. R.; Thurston, D. E.;
Harrap, K. R. Preclinical Pharmacology and Antitumour Activity
of the Novel Sequence-Selective DNA Minor-Groove Cross-
Linking Agent DSB-120. Cancer Chemother. Pharmacol. 1996,
38, 431-438.
(18) Morris, S. J .; Thurston, D. E.; Nevell, T. G. Evaluation of the
Electrophilicity of DNA-Binding Pyrrolo[2,1-c][1,4]benzodiazepines
by HPLC. J . Antibiot. 1990, 43, 1286-1292.
(19) Morris, S. M. Design, Synthesis and Evaluation of a Sequence-
Selective DNA-Cleaving Agent Based on the Pyrrolo[2,1-c][1,4]-
benzodiazepine Ring System: Investigation of the DNA-Reactive
Species. Ph.D. Thesis, University of Portsmouth, Portsmouth,
U.K., 1992.
(20) Gregson, S. J .; Howard, P. W.; J enkins, T. C.; Kelland, L. R.;
Thurston, D. E. Synthesis of a Novel C2/C2′-Exo Unsaturated
Pyrrolobenzodiazepine Cross-linking Agent with Remarkable
DNA-Binding Affinity and Cytotoxicity. Chem. Commun. 1999,
797-798.
(21) Fukuyama, T.; Liu, G.; Linton, S. D.; Lin, S. C.; Nishino, H. Total
Synthesis of (+)-Porothramycin-B. Tetrahedron Lett. 1993, 34,
2577-2580.
(22) Murata, M.; Chiba, T.; Yamada, A. European Patent Application
89102859.9: J apan, 1989.
(23) Griffith, W. P.; Ley, S. V.; Whitcombe, G. P.; White, A. D.
Preparation and Use of Tetra-N-Butylammonium Per-Ruthenate
(TBAP Reagent) and Tetra-N-Propylammonium Per-Ruthenate
(TPAP Reagent) as New Catalytic Oxidants for Alcohols. Chem.
Commun. 1987, 1625-1627.
(24) Deziel, R. Mild Palladium (0)-Catalyzed Deprotection of Allyl
Esters - A Useful Application in the Synthesis of Carbapenems
and Other Beta-Lactam Derivatives. Tetrahedron Lett. 1987, 28,
4371-4372.
(25) Kunz, H.; Unverzagt, C. The Allyloxycarbonyl (Aloc) Moiety -
Conversion of an Unsuitable into a Valuable Amino Protecting
Group for Peptide Synthesis. Angew. Chem., Int. Ed. Engl. 1984,
23, 436-437.
(26) Martin, S. F.; Campbell, C. L. Total Syntheses of (()-Crinine
and (()-Buphanisine. J . Org. Chem. 1988, 53, 3184-3190.
(27) Dangles, O.; Guibe, F.; Balavoine, G.; Lavielle, S.; Marquet, A.
Selective Cleavage of the Allyl and Allyloxcarbonyl Groups
through Palladium-Catalyzed Hydrostannolysis with Tributyltin
Hydride - Application to the Selective Protection-Deprotection
of Amino-Acid Derivatives and in Peptide-Synthesis. J . Org.
Chem. 1987, 52, 4984-4993.
Ack n ow led gm en t. This work was supported in part
by the Cancer Research Campaign UK (Grant num-
bers: SP1938/0301 to D.E.T., T.C.J., and J.A.H.; SP1938/
0401 to D.E.T.; SP/DC/STU2330/0201 to L.R.K.) and
Yorkshire Cancer Research (to T.C.J .). Dr. Robert
Schultz (NCI) is thanked for helpful discussions and for
the analytical results for 5. Dr. Lesley Canfield, Dr.
Kevin Wellam, and Mr. Nigel Armstrong are thanked
for analytical support.
Su p p or tin g In for m a tion Ava ila ble: Purity table for
target molecules. This material is available free of charge via
the Internet at http://pubs.acs.org.
Refer en ces
(1) Thurston, D. E. Nucleic Acid Targeting: Therapeutic Strategies
for the 21st Century. Br. J . Cancer 1999, 80, 65-85.
(2) Thurston, D. E. Advances in the Study of Pyrrolo[2,1-c][1,4]-
benzodiazepine (PBD) Antitumour Antibiotics. Molecular Aspects
of Anticancer Drug-DNA Interactions; The Macmillan Press
Ltd.: London, U.K., 1993; pp 54-88.
(3) Thurston, D. E.; Bose, D. S. Synthesis of DNA-Interactive
Pyrrolo[2,1-c][1,4]benzodiazepines. Chem. Rev. 1994, 94, 433-
465.
(4) Berry, J . M.; Howard, P. W.; Thurston, D. E. Solid-Phase
Synthesis of DNA-Interactive Pyrrolo[2,1-c][1,4]benzodiazepines.
Tetrahedron Lett. 2000, 41, 6171-6174.
(5) Thurston, D. E.; Bose, D. S.; Howard, P. W.; J enkins, T. C.; Leoni,
A.; Baraldi, P. G.; Guiotto, A.; Cacciari, B.; Kelland, L. R.;
Foloppe, M. P.; Rault, S. Effect of A-Ring Modifications on the
DNA-Binding Behavior and Cytotoxicity of Pyrrolo[2,1-c][1,4]-
benzodiazepines. J . Med. Chem. 1999, 42, 1951-1964.
(6) Puvvada, M. S.; Hartley, J . A.; J enkins, T. C.; Thurston, D. E.
A Quantitative Assay to Measure the Relative DNA-Binding
Affinity of Pyrrolo[2,1-c][1,4]benzodiazepine (PBD) Antitumor
Antibiotics Based on the Inhibition of Restriction-Endonuclease
BamH1. Nucleic Acids Res. 1993, 21, 3671-3675.
(7) Puvvada, M. S.; Forrow, S. A.; Hartley, J . A.; Stephenson, P.;
Gibson, I.; J enkins, T. C.; Thurston, D. E. Inhibition of Bacte-
riophage T7 RNA Polymerase In Vitro Transcription by DNA-
Binding Pyrrolo[2,1-c][1,4]benzodiazepines. Biochemistry 1997,
36, 2478-2484.
(28) Bellamy, F. D.; Ou, K. Selective Reduction of Aromatic Nitro
Compounds with Stannous Chloride in Non-Acidic and Non-
Aqueous Medium. Tetrahedron Lett. 1984, 25, 839-842.
(29) J ones, G. B.; Davey, C. L.; J enkins, T. C.; Kamal, A.; Kneale, G.
G.; Neidle, S.; Webster, G. D.; Thurston, D. E. The Non-Covalent
Interaction of Pyrrolo[2,1-c][1,4]benzodiazepine-5,11-diones with
DNA. Anti-Cancer Drug Des. 1990, 5, 249-264.
(30) Adams, L. J .; Morris, S. J .; Banting, L.; J enkins, T. C.; Thurston,
D. E. Molecular Mechanics Study of the Sterochemistry of
Formation of Covalent Pyrrolobenzodiazepine-DNA Adducts.
Pharm. Sci. 1995, 1, 151-154.
(31) J enkins, T. C.; Lane, A. N. AT Selectivity and DNA Minor Groove
Binding: Modelling, NMR and Structural Studies of the Interac-
tions of Propamidine and Pentamidine with d(CGCGAAT-
TCGCG)2. Biochim. Biophys. Acta Gene Struct. Expression 1997,
1350, 189-204.
(32) Adams, L. J .; J enkins, T. C.; Banting, L.; Thurston, D. E.
Molecular Modelling of a Sequence-Specific DNA-Binding Agent
Based on the Pyrrolo[2,1-c][1,4]benzodiazepines. Pharm. Phar-
macol. Commun. 1999, 5, 555-560.
(8) Thurston, D. E.; Morris, S. J .; Hartley, J . A. Synthesis of a Novel
GC-Specific Covalent-Binding DNA Affinity-Cleavage Agent
Based on Pyrrolobenzodiazepines (PBDs). Chem. Commun. 1996,
563-565.
(9) Wilson, S. C.; Howard, P. W.; Forrow, S. M.; Hartley, J . A.;
Adams, L. J .; J enkins, T. C.; Kelland, L. R.; Thurston, D. E.
Design, Synthesis and Evaluation of a Novel Sequence-Selective
Epoxide-Containing DNA Cross-Linking Agent Based on the
Pyrrolo[2,1-c][1,4]benzodiazepine System. J . Med. Chem. 1999,
42, 4028-4041.
(10) Smellie, M.; Kelland, L. R.; Thurston, D. E.; Souhami, R. L.;
Hartley, J . A. Cellular Pharmacology of Novel C8-Linked
Anthramycin-Based Sequence-Selective DNA Minor-Groove Cross-
Linking Agents. Br. J . Cancer 1994, 70, 48-53.
(11) Bose, D. S.; Thompson, A. S.; Ching, J . S.; Hartley, J . A.;
Berardini, M. D.; J enkins, T. C.; Neidle, S.; Hurley, L. H.;
Thurston, D. E. Rational Design of a Highly Efficient Irreversible
DNA Interstrand Cross-Linking Agent Based on the Pyrroloben-
zodiazepine Ring System. J . Am. Chem. Soc. 1992, 114, 4939-
4941.
(12) Bose, D. S.; Thompson, A. S.; Smellie, M.; Berardini, M. D.;
Hartley, J . A.; J enkins, T. C.; Neidle, S.; Thurston, D. E. Effect
of Linker Length on DNA-Binding Affinity, Cross-Linking
Efficiency and Cytotoxicity of C8-Linked Pyrrolobenzodiazepine
Dimers. Chem. Commun. 1992, 1518-1520.
(33) Kelland, L. R.; Barnard, C. F. J .; Mellish, K. J .; J ones, M.;
Goddard, P. M.; Valenti, M.; Bryant, A.; Murrer, B. A.; Harrap,
K. R. A Novel Trans-Platinum Coordination Complex Possessing
In-Vitro and In-Vivo Antitumor-Activity. Cancer Res. 1994, 54,
5618-5622.

748 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 5
Gregson et al.
(34) Alley, M. C.; Scudiero, D. A.; Monks, A.; Hursey, M. L.;
Czerwinski, M. J .; Fine, D. L.; Abbott, B. J .; Mayo, J . G.;
Shoemaker, R. H.; Boyd, M. R. Feasibility of Drug Screening
with Panels of Human-Tumor Cell Lines Using a Microculture
Tetrazolium Assay. Cancer Res. 1988, 48, 589-601.
(35) Mistry, P.; Kelland, L. R.; Abel, G.; Sidhar, S.; Harrap, K. R.
The Relationships Between Glutathione, Glutathione-S-Trans-
ferase and Cytotoxicity of Platinum Drugs and Melphalan in 8
Human Ovarian Carcinoma Cell Lines. Br. J . Cancer 1991, 64,
215-220.
(36) Thurston, D. E.; Howard, P. W. Novel Anticancer Agents:
Monomeric and Dimeric PBDs. British Patent, No. WO 00/12508,
United Kingdom, 2000.
(37) Manzini, G.; Barcellona, M. L.; Avitabile, M.; Quadrifoglio, F.
Interaction of Diamidino-2-Phenylindole (DAPI) with Natural
and Synthetic Nucleic Acids. Nucleic Acids Res. 1983, 11,
8861-8876.
J M001064N