Novel C8-linked pyrrolobenzodiazepine (PBD)-heterocycle conjugates that recognize DNA sequences containing an inverted CCAAT box

Article abstract of DOI:10.1016/j.bmcl.2011.04.054

A series of novel DNA-interactive C8-linked pyrrolobenzodiazepine (PBD)-heterocycle polyamide conjugates has been synthesised to explore structure/sequence-selectivity relationships. One conjugate (2d) has a greater selectivity and DNA binding affinity for inverted CCAAT sequences within the Topoisomerase IIα promoter than the known C8-bis-pyrrole PBD conjugate GWL-78 (1b).

Full text of DOI:10.1016/j.bmcl.2011.04.054

Bioorganic & Medicinal Chemistry Letters 21 (2011) 3780–3783
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Novel C8-linked pyrrolobenzodiazepine (PBD)–heterocycle conjugates
that recognize DNA sequences containing an inverted CCAAT box
Federico Brucoli a, Rachel M. Hawkins a, Colin H. James a, Geoff Wells a, Terence C. Jenkins b, Tom Ellis c,d
,
John A. Hartley c,d, Philip W. Howard d, David E. Thurston a,d,
⇑
a The School of Pharmacy, University of London, London WC1N 1AX, UK
b University of Greenwich, Chatham, Kent ME4 4TB, UK
c University College London, London WC1E 6BT, UK
d Spirogen Ltd, 29-39 Brunswick Square, London WC1N 1AX, UK
a r t i c l e i n f o
a b s t r a c t
Article history:
A series of novel DNA-interactive C8-linked pyrrolobenzodiazepine (PBD)–heterocycle polyamide conju-
gates has been synthesised to explore structure/sequence-selectivity relationships. One conjugate (2d)
has a greater selectivity and DNA binding affinity for inverted CCAAT sequences within the Topoisomer-
Received 4 February 2011
Revised 11 April 2011
Accepted 12 April 2011
Available online 20 April 2011
ase IIa promoter than the known C8-bis-pyrrole PBD conjugate GWL-78 (1b).
Ó 2011 Elsevier Ltd. All rights reserved.
Keywords:
Pyrrolobenzodiazepine (PBD)-conjugate
Inverted CCAAT box
Nuclear Factor Y
Topoisomerase II
a
Anticancer agents
The CCAAT box is a common DNA sequence motif present in di-
rect or reverse orientation in the promoters of many eukaryotic
genes. It is usually found approximately 80 base pairs upstream
of the transcription start site and may be present in one or more
copies. It provides a recognition site for a number of general tran-
scription factors (known as CCAAT box binding proteins or factors)
including the ubiquitously expressed Nuclear Factor Y (NF-Y), in
which case protein-specific binding is necessary for CCAAT box
activation.1,2 Studies have shown that CCAAT box binding proteins
such as NF-Y3 are associated with a number of important cellular
processes including cell cycle progression,4 differentiation,5 and
cellular aging.5,6 NF-Y has been shown to exert both stimulatory
and inhibitory effects on gene expression, and an important link
between NF-Y and p53 has emerged in which activation of p53
by DNA damage induces repression of G2/M genes through NF-Y
association leading to cell cycle arrest. Conversely, P53 mutants
commonly found in human tumour cells have an increased affinity
for NF-Y, resulting in activation of growth-promoting genes under
NF-Y control.7,8 Therefore, targeting the interaction of NF-Y with
DNA could be a potential strategy for blocking the proliferation
of cancer cells.
a potential means to develop novel therapeutic agents.9,10 We have
previously reported the synthesis of a series of C8-linked PBD–
poly(N-methylpyrrole) conjugates (1a–f, Fig. 1A), members of
which have significant DNA recognition properties and the poten-
tial to modulate the activity of specific transcription factors.11
These conjugates bind covalently through their C11-position to
the C2-NH2 functionality of guanine residues in the DNA minor
groove with a preference for 50-XGXWZ motifs (X = any base, but
mainly a purine; W = A or T; z = 3 1).11 Hochhauser and co-work-
ers demonstrated that one member of this family, GWL-78 (1b),
can inhibit the binding of NF-Y to inverted CCAAT box sequences
1–3 (ICBs 1–3) within the human DNA Topoisomerase II
a (Topo-
IIa
) promoter.12 1b was demonstrated to interact with a variety
of CCAAT-containing promoters leading to p53-independent cell
cycle arrest.12
In an effort to further investigate the interactions of 1b with
ICBs sites, and to improve on the potency and selectivity of the par-
ent ligand, we prepared a library of six analogs of 1b (2a–f, Fig. 1B)
in which the pyrrole residues in the C8-polyamide chain were
changed to various combinations of pyrrole (Py), imidazole (Im)
or thiazole (Th) rings. These heterocycles were selected as they
have been previously used by Dervan to enhance the sequence rec-
ognition of non-covalent minor-groove binding DNA-interactive
ligands.13 Thermal denaturation and DNA footprinting assays were
employed to assess the DNA-binding affinity and sequence selec-
tivity of these molecules, and their cytotoxicities were determined
There have been previous attempts to use small molecules to
disrupt NF-Y/DNA binding both for experimental purposes and as
⇑
Corresponding author.
E-mail address: david.thurston@pharmacy.ac.uk (D.E. Thurston).
0960-894X/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2011.04.054

F. Brucoli et al. / Bioorg. Med. Chem. Lett. 21 (2011) 3780–3783
3781
Table 1
A
Thermal denaturation and in vitro cytotoxicity data for the PBD–heterocyclic
conjugates 2a–f, 1b (GWL-78) and the linker PBD 1g.
Cytotoxicity (NCI 60-cell-line panel)b
a
Compound
D
Tm (°C)
0 h
4 h
18 h
GI50 (nM)
TGI (nM)
LC50 (lM)
2d
1b
2c
2b
2e
2f
14.5
13.7
10.3
10.0
9.3
7.7
2.8
0.2
15.2
14.5
10.9
10.7
9.6
8.0
3.0
0.3
15.8
15.1
11.4
11.2
10.2
8.3
3.1
11.0
35.5
100.0
14.8
70.8
96.0
120.0
933.0
1445.0
363.0
813.0
9.1
10.0
13.5
32.4
19.1
12.3
19.5
46.8
2a
1g
3.1
0.4
691.8
1740.0
4898.0
11,800.0
a
Induced
control CT-DNA alone,
values are 0.1–0.2 °C. For a fixed 1:10 molar ratio of [ligand]/[DNA], DNA con-
centration = 50 (base pairs) and ligand concentration = 5 M, in aqueous
D
Tm following incubation at 37 0.1 °C for the time shown.19 For
Tm = 67.82 0.07 °C (mean from >90 experiments). All Tm
D
D
lM
l
B
sodium phosphate buffer [10 mM sodium phosphate + 1 mM EDTA, pH 7.00 0.01].
b
GI50, TGI and LC50 values (the concentrations causing 50% growth inhibition,
total growth inhibition, and 50% lethality, respectively) are mean across the 60-cell-
line-panel.
Table 2
Minimum concentration of each PBD-conjugate required to provide a measurable
footprint at the ICB1 and ICB2 sites
Compound
Identity
DNA footprinting potencya
(lM)
Figure 1. (A) Structures of the C8-linked pyrrolobenzodiazepine–poly(N-methyl-
pyrrole) conjugates11 including GWL-78 (1b), and the C8-coupled PBD-linker
fragment 1g upon which 1a–f and 2a–f are based. (B) The library (2a–f) of
analogues of 1b containing combinations of imidazole (Im), thiazole (Th) and
pyrrole (Py) heterocycles in the C8-polyamide chain.
ICB1
ICB2
2d
1b
2b
2c
2a
2e
2f
Py-Th-PBD
Py-Py-PBD
Im-Py-PBD
Im-Im-PBD
Py-Im-PBD
Th-Py-PBD
Th-Th-PBD
0.2
1
0.2
1
5
5
25
>25
>25
>25
5
>25
1
in the National Cancer Institute (NCI) 60-cell-line panel. The thia-
zole-containing conjugate 2d emerged as superior to 1b in terms
of overall DNA-binding affinity, selectivity for specific ICB se-
quences and growth inhibitory potency.
>25
a
concentration (lM) at which DNA footprints were observed.
The synthetic approach used to prepare PBD-conjugates 2a–f is
shown in Scheme 1. The heterocyclic amino esters 3a–c and the
Boc-protected heterocyclic amino acids 4a–c were prepared
according to literature methods11,14 and then coupled together
(using EDCI/DMAP) to produce the N-Boc-protected imidazole
(5a–c) and thiazole (5d–f) dimers which, without purification,
were deprotected to provide 6a–f. These dimeric units were then
coupled to the previously described11 PBD capping unit 7 to obtain
PBD conjugates 2a–f in moderate to good yields.15
The relative overall binding affinities of 2a–f were evaluated
from their induced effects upon the melting behaviour (Tm) of dou-
ble-stranded calf thymus (CT) DNA.11,16–19 The results (Table 1)
show that all the conjugated ligands increased the thermal stabil-
ity of the DNA duplex but to different extents. In particular, the Py–
Th–PBD conjugate (2d) had a higher DNA binding affinity than 1b
and all other library members, with a
DTm value of 14.5 °C without
incubation (i.e., 0 h)19 compared to 13.7 °C for 1b. The Im-contain-
3a: R1 = Me; R2 = H
5a/6a
: Py-Im
X, Y = C; Z = N-CH3
X = N; Y = C; Z = N-CH3
X = N; Y = S; Z = C
NHR2
5b/6b: Im-Py
5c/6c: Im-Im
4a: R1 = H; R2 = Boc
3b: R1 = Et; R2 = H
X
R1O
5d/6d
5e/6e: Th-Py
5f/6f: Th-Th
: Py-Th
Y
: R1 = H; R2 = Boc
: R1 = Et; R2 = H
4b
Z
O
3c
4c
1
: R = H; R2 = Boc
O
O
N
O
O
H
O
N
O
O
NHR2
HO
MeO
H
N
Het
R1O
a
7
3a-c
4a-c
2a-f
Het
+
O
a, c
O
5a-f: R1 = Me or Et; R2 = Boc
b
: R1 = Me or Et; R2 = H
6a-f
Scheme 1. Reagents and conditions: (a) EDCI, DMAP, CH2Cl2, 18–48 h, 35–99%; (b) CH2Cl2/TFA/H2O (50:47.5:2.5 v/v), 30 min (or 4 M HCl in dioxane, 30 min, for the
imidazole-containing dimers); (c) Pd[PPh3]4, pyrrolidine, CH2Cl2, 3 h, 30–55% (Het = generic heterocycle).

3782
F. Brucoli et al. / Bioorg. Med. Chem. Lett. 21 (2011) 3780–3783
ing PBD conjugates 2b–c and the Th-containing 2e–f stabilised
DNA less effectively than 2d or 1b, and the Im-containing 2a had
1b (1 lM) (Fig. 2A). In contrast, three imidazole- (2a–c) and two
thiazole-containing (2e–f) PBD-conjugates had a lower affinity
for the ICB sites. Interestingly, compounds 2c and 2e were able
to discriminate between the two ICB sites, binding at a lower con-
centration to ICB2 (see Table 2).
the lowest
DTm compared to all other compounds. For comparative
purposes, the C8-coupled PBD-linker fragment 1g (Fig. 1A) was
evaluated in both the DNA thermal denaturation assay and in the
NCI 60-cell-line panel (Table 1). It had only weak duplex DNA sta-
The conjugates were also screened in the NCI 60-cell-line panel
(Table 1). Once again, 2d was found to be the most potent member
of the series, with a greater cytotoxicity compared to 1b and all
other library members. Furthermore, it was found to be more ac-
tive against breast, colon and renal cancer cells compared to other
tumour types in the NCI panel. The mean GI50 value across all cell
lines was approximately 3-fold lower than for 1b (i.e., 3.1 vs
11.0 nM, respectively). These results correlate with the thermal
denaturation and footprinting data, with 2d exhibiting the highest
potency in all assays. Furthermore, the GI50 value for 2d (3.1 nM) is
bilizing properties (i.e.,
DTm = 0.4 °C after 18 h) and was not appre-
ciably cytotoxic compared to the PBD-conjugates 2a–f, thus
confirming the contribution of the C8-heterocyclic rings to the
activity of 1b and 2a–f.
DNase I footprinting studies were carried out to evaluate the se-
quence selectivity of the PBD–polyamide conjugates using a new
methodology described by Hartley and co-workers.20 This involved
the use of IR dyes rather than a radiolabel, with DNA fragments
visualized on the gel through detection of the dyes with a diode la-
ser. PBD conjugates 2a–f and 1b were footprinted against a 170 bp
nearly 3000-times lower than its LC50 value (9.1 lM), consistent
fragment of the human Topo-II
ICB2 sites. All PBD–heterocycles were incubated with the DNA
fragment at different concentrations (0 ? 25 M) overnight prior
a
promoter containing the ICB1 and
with its proposed mechanism of action as a transcription factor
inhibitor.
l
Molecular modelling studies were performed in an effort to
rationalize the superior DNA binding affinity and sequence selec-
tivity of conjugate 2d compared to 2a, which had the lowest po-
tency in all assays. In each case, a 50-CTACGATTGGTTCTT-30
duplex containing an ICB sequence (underlined) flanked by a 50G
to DNase I cleavage. According to the literature,11 incubating PBD
C8-conjugates such as 1a–f with DNA for 17 h should lead to com-
plete covalent binding. The results are summarized in Table 2, and
a typical gel resulting from the study is shown in Figure 2A.
The results demonstrate that substitution of an imidazole or
thiazole unit for a pyrrole in the C8-polyamide chain of 1b signif-
icantly affects the binding affinity and selectivity of the com-
pounds for the CCAAT motifs. Conjugate 2d showed the greatest
binding affinity, generating measurable footprints at both ICB1
(bold) as found in ICBs 1 and 2 in the Topo-IIa promoter was used.
B-DNA was constructed with the AMBER package,21 and the xleap
program was used to make an initial graphical alignment of the li-
gand in the minor groove prior to forming a covalent attachment
between the C11-position of the PBD and the C2-NH2 of guanine-
5 in the minor groove of the 15-mer DNA model. The PBD A-ring
and ICB2 sites at 0.2 lM, a concentration 5-fold lower than for
Figure 2. (A) DNase I footprinting gel showing the binding of PBD conjugates 1b (GWL-78) and 2d to a 170 bp fragment of the human Topo-II
seen at both the ICB1 and ICB2 sequences as marked on the gel. Lanes a–h correspond to concentrations of 0, 0.0016, 0.008, 0.04, 0.2, 1, 5 and 25
lane (m) allowed sequence identification. (B) Model of complex of 2d covalently bound to a 15-mer DNA duplex containing an embedded 50-GATTGG-30 tract.
a
promoter. Footprints can be
lM, respectively. The marker

F. Brucoli et al. / Bioorg. Med. Chem. Lett. 21 (2011) 3780–3783
3783
was oriented towards the 30-end of the covalently-modified strand
with a S-configuration at the C11-position of the PBD, consistent
with observations reported in the literature.11 Subsequent minimi-
zation was performed using AMBER, initially applying a large posi-
tional restraint to the DNA only. This was gradually reduced to zero
in further rounds of minimization. A long range non-bonded cut-
off was applied along with use of the GBSA implicit solvent model
and monovalent ion screening (0.2 M). After the energy minimiza-
tion procedure, both bound conjugates appeared to be well-accom-
modated within the DNA minor groove with virtually no induced
distortion of the overall host duplex structure. Figure 2B shows
the excellent fit achieved by the most potent ligand 2d in the min-
or groove of the ICB1 sequence when covalently bound to the C2-
NH2 of G5. Surprisingly, there were no obvious differences be-
tween the two models in terms of distortions or steric interactions
that might account for the large difference in experimental binding
affinities observed for 2a and 2d.
ship for R.M.H. (Earmarked QUOTA Award 01300211), and Cancer
Research UK for funding to D.E.T. (C180/A1060, SP1938/0402,
SP1938/0201 and SP1938/0301) and J.A.H. (C2259/A9994).
References and notes
1. Mantovani, R. Gene 1999, 239, 15.
2. Higuchi, T.; Anzai, K.; Kobayashi, S. Biochim. Biophys. Acta-Gen. Subj. 2008, 1780,
274.
3. Ronchi, A.; Bellorini, M.; Mongelli, N.; Mantovani, R. Nucleic Acids Res. 1995, 23,
4565.
4. Sandri, M. I.; Isaacs, R. J.; Ongkeko, W. M.; Harris, A. L.; Hickson, I. D.; Broggini,
M.; Vikhanskaya, F. Nucleic Acids Res. 1996, 24, 4464.
5. Marziali, G.; Perrotti, E.; Ilari, R.; Coccia, E. M.; Mantovani, R.; Testa, U.;
Battistini, A. Blood 1999, 93, 519.
6. Matuoka, K.; Chen, K. Y. Ageing Res. Rev. 2002, 1, 639.
7. Di Agostino, S.; Strano, S.; Emiliozzi, V.; Zerbini, V.; Mottolese, M.; Sacchi, A.;
Blandino, G.; Piaggio, G. Cancer Cell 2006, 10, 191.
8. Imbriano, C.; Gurtner, A.; Cocchiarella, F.; Di Agostino, S.; Basile, V.; Gostissa,
M.; Dobbelstein, M.; Del Sal, G.; Piaggio, G.; Mantovani, R. Mol. Cell. Biol. 2005,
25, 3737.
A further investigation centred on evaluating the free energy of
binding by constructing the two ligands in their most appropriate
conformations prior to covalent interaction. This was followed by
graphical alignment in the minor groove by placing the N10 of each
PBD moiety with its partial negative charge in close proximity to
the partial positive charge of a C2-NH2 hydrogen of guanine-5. A
similar stepwise minimization procedure as used for the cova-
lently-bound ligands was then applied, and free energy calcula-
tions undertaken using the AMBER MM-PBSA approach. This
involved extracting 200 frames from a 2ns molecular dynamics
simulation followed by calculation of the average free energy of
binding in each case. However, this methodology also failed to pro-
vide an explanation for the observed difference in binding affinity
between 2a and 2d. Perhaps this reflects the limitations of model-
ing approaches of this type to discriminate between ligands with
relatively small structural differences when binding to a complex
target such as DNA.
9. Tolner, B.; Hartley, J. A.; Hochhauser, D. Mol. Pharmacol. 2001, 59, 699.
10. Mackay, H.; Brown, T.; Sexton, J. S.; Kotecha, M.; Nguyen, B.; Wilson, W. D.;
Kluza, J.; Savic, B.; O’Hare, C.; Hochhauser, D.; Lee, M.; Hartley, J. A. Bioorg. Med.
Chem. 2008, 16, 2093.
11. Wells, G.; Martin, C. R. H.; Howard, P. W.; Sands, Z. A.; Laughton, C. A.;
Tiberghien, A.; Woo, C. K.; Masterson, L. A.; Stephenson, M. J.; Hartley, J. A.;
Jenkins, T. C.; Shnyder, S. D.; Loadman, P. M.; Waring, M. J.; Thurston, D. E. J.
Med. Chem. 2006, 49, 5442.
12. Kotecha, M.; Kluza, J.; Wells, G.; O’Hare, C. C.; Forni, C.; Mantovani, R.; Howard,
P. W.; Morris, P.; Thurston, D. E.; Hartley, J. A.; Hochhauser, D. Mol. Cancer Ther.
2008, 7, 1319.
13. Dervan, P. B. Bioorg. Med. Chem. 2001, 9, 2215.
14. Boger, D. L.; Fink, B. E.; Hedrick, M. P. J. Am. Chem. Soc. 2000, 122, 6382.
15. Data for 2d: A pale yellow solid (30% yield). 1H NMR (acetone-d6) (400 MHz):
d 11.29 (s, 1H, N–H), 9.28 (s, 1H, N-H), 7.81 (s, 1H, Th-H), 7.74 (d, 1H,
J = 4.4 Hz, H-11), 7.53 (d, 1H, J = 1.9 Hz, Py-H), 7.40 (s, 1H, H-6), 6.96 (d, 1H,
J = 2.0 Hz, Py-H), 6.81 (s, 1H, H-9), 4.19 (m, 2H, side chain H-1), 3.91 (s, 3H,
O/N-CH3), 3.87 (s, 3H, O/N–CH3), 3.77 (s, 3H, O/N–CH3), 3.69 (m, 1H, H-11a),
3.46 (m, 2H, H-3), 2.53 (s, 2H, side chain H-3), 2.36 (m, 2H, H-1), 2.26 (m,
2H, side chain H-2), 2.09 (s, 2H, H-2); LC–MS m/z (ES+): 595 ([M+H]+); Acc.
Mass C28H30N6O7S: calcd 595.1070, found 595.1957 ([M+H]+).
16. McConnaughie, A. W.; Jenkins, T. C. J. Med. Chem. 1995, 38, 3488.
17. Bose, D. S.; Thompson, A. S.; Smellie, M.; Berardini, M. D.; Hartley, J. A.; Jenkins,
T. C.; Neidle, S.; Thurston, D. E. Chem. Commun. 1992, 1518.
18. Thurston, D. E.; Bose, D. S.; Howard, P. W.; Jenkins, T. C.; Leoni, A.; Baraldi, P. G.;
Guiotto, A.; Cacciari, B.; Kelland, L. R.; Foloppe, M. P.; Rault, S. J. Med. Chem.
1999, 42, 1951.
In summary, these results demonstrate that it is possible to
modify the C8-bis-pyrrole side chain of GWL-78 (1b) to improve
DNA binding affinity and selectivity for specific ICB sites within
the human Topo-IIa promoter, although it is not yet possible to
19. The ‘0 h’ values shown actually reflect ꢀ30 min equilibration heating within
the instrument used for the Tm assay. This is unavoidable and cannot be
circumvented in temperature scan-based assays. Heating was applied at a rate
of 1 °C/min in the 50–99 °C temperature range, with optical and temperature
data sampling at 100 ms intervals. A separate experiment was carried out
using buffer alone, and this baseline was subtracted from each DNA melting
curve before data treatment. Experiments were conducted in triplicate.
20. Ellis, T.; Evans, D. A.; Martin, C. R. H.; Hartley, J. A. Nucleic Acids Res. 2007, 35.
Article No.: e89.
21. Case, D. A.; Darden, T. A.; Cheatham III, T. E.; Simmerling, C. L.; Wang, J.; Duke,
R. E.; Luo, R.; Merz, K. M.; Pearlman, D. A.; Crowley, M.; Walker, R. C.; Zhang,
W.; Wang, B.; Hayik, S.; Roitberg, A.; Seabra, G.; Wong, K. F.; Paesani, F.; Wu, X.;
Brozell, S.; Tsui, V.; Gohlke, H.; Yang, L.; Tan, C.; Mongan, J.; Hornak, V.; Cui, G.;
Beroza, P.; Mathews, D. H.; Schafmeister, C.; Ross, W. S.; Kollman, P. A.
University of California, San Francisco, 2006.
rationalize how the improved activity of 2d relates to the struc-
tural change. Further studies of 2d are underway to evaluate its
ability to stabilize shorter fragments of DNA containing the ICB
binding sequence, and to determine whether it can inhibit the
binding of NF-Y to DNA in cells using methods such as chromatin
immunoprecipitation (ChIP). In addition, the effect of extending
the length of 2d with further heterocycles is being investigated.
Acknowledgments
Dr. Emma Sharp is gratefully acknowledged for her help with
preparing the manuscript. The authors thank EPSRC for a student-