Alternative heterocycles for DNA recognition: The benzimidazole/imidazole pair

Article abstract of DOI:10.1002/chem.200204689

Boc-protected benzimidazole-pyrrole, benzimidazole-imidazole, and benzimidazole-methoxypyrrole amino acids were synthesized and incorporated into DNA binding polyamides, comprised of N-methyl pyrrole and N-methyl imidazole amino acids, by means of solid-phase synthesis on an oxime resin. These hairpin polyamides were designed to determine the DNA recognition profile of a side-by-side benzimidazole/imidazole pair for the designated six base pair recognition sequence. Equilibrium association constants of the polyamide-DNA complexes were determined at two of the six base pair positions of the recognition sequence by quantitative DNase I foot-printing titrations on DNA fragments each containing matched and single base pair mismatched binding sites. The results indicate that the benzimidazole-heterocycle building blocks can replace pyrrole-pyrrole, pyrrole-imidazole, and pyrrole-hydroxypyrrole constructs while retaining relative site specifities and subnanomolar match site affinities. The benzimidazole-containing hairpin polyamides represent a novel class of DNA binding ligands featuring tunable target recognition sequences combined with the favorable properties of the benzimidazole type DNA minor groove binders.

Full text of DOI:10.1002/chem.200204689

FULL PAPER
Alternative Heterocycles for DNA Recognition:
The Benzimidazole/Imidazole Pair
[a]
Christoph A.Briehn, Philipp Weyermann, and Peter B.Dervan*
Abstract: Boc-protected
zole-pyrrole, benzimidazole-imidazole,
and benzimidazole-methoxypyrrole
benzimida-
sequence. Equilibrium association con-
stants of the polyamide-DNA com-
plexes were determined at two of the
six base pair positions of the recognition
sequence by quantitative DNase I foot-
printing titrations on DNA fragments
each containing matched and single base
pair mismatched binding sites. The re-
sults indicate that the benzimidazole-
heterocycle building blocks can replace
pyrrole-pyrrole, pyrrole-imidazole, and
pyrrole-hydroxypyrrole constructs while
retaining relative site specifities and
subnanomolar match site affinities. The
benzimidazole-containing hairpin poly-
amides represent a novel class of DNA
binding ligands featuring tunable target
recognition sequences combined with
the favorable properties of the benzimi-
dazole type DNA minor groove binders.
amino acids were synthesized and in-
corporated into DNA binding polya-
mides, comprised of N-methyl pyrrole
and N-methyl imidazole amino acids, by
means of solid-phase synthesis on an
oxime resin. These hairpin polyamides
were designed to determine the DNA
recognition profile of a side-by-side
benzimidazole/imidazole pair for the
designated six base pair recognition
Keywords: base pairing
affinity DNA recognition
hydrogen bonds ¥ polyamides
¥ binding
¥
¥
In addition to the development of various polyamide
binding motifs such as hairpins,[3] cycles,[4] H-pins,[5] and
dimers,[6] efforts have been devoted to extend the ensemble of
five-membered aromatic amino acids that are capable of
cooperatively pairing with each other to recognize DNA base
pair sequences in the minor groove.[7] Hitherto, less attention
was focused on the incorporation of six-membered ring
heterocycles and fused heteroaromatic rings into polyamide
sequences and on the manipulation of the heterocycle tether-
ing amide bonds. In earlier studies of polyamide-type minor
groove binders pyridine- and phenol-based monomers (when
paired with Py) were employed to target G ¥ C and A ¥ T base
pairs, respectively.[8, 9] Fused heterocycles such as benzimida-
zoles, imidazopyridines, and indoles as part of small molecule
ligands with extended p-conjugated systems are known to
bind to the minor groove of DNA with high affinities[10] and
some of them were shown to have biological activity.[11]
Hoechst 33258, which comprises of a bis-benzimidazole, a
N-methylpiperazine, and a phenol moiety, is one of the most
prominent examples of these fused heterocycle derivatives
due to its favorable fluorescence characteristics.[12] The
crescent-shaped dye binds to the minor groove of DNA at
A/T tracks as a 1:1 complex.[13]
Introduction
The deciphering of the human genome has spurred efforts to
develop various strategies for gene regulation at the tran-
scriptional and translational level. As promising modulators
of gene transcription, DNA binding polyamides have proved
to be useful tools that recognize predetermined sequences
within the minor groove of DNA at subnanomolar concen-
trations.[1, 2] DNA recognition by polyamides depends on a
code of side-by-side amino acid pairings that are oriented N
C with respect to the 5' 3'-direction of the DNA helix in the
minor groove.[1] An antiparallel pairing of N-methyl imidazole
opposite N-methyl pyrrole (Im/Py pair) distinguishes G ¥ C
from C ¥ G and both of these from A ¥ T/T¥ A base pairs (bp).
An N-methyl pyrrole/N-methyl pyrrole (Py/Py) pair binds
both A ¥ T and T¥ A in preference to G ¥ C/C ¥ G. The discrim-
ination of T¥ A from A ¥ T by means of N-methyl hydrox-
ypyrrole/N-methyl pyrrole (Hp/Py) pairs completes the four
base pair code.[1]
[a] Prof. Dr. P. B. Dervan, Dr. C. A. Briehn, Dr. P. Weyermann
Division of Chemistry and Chemical Engineering
California Institute of Technology
Pasadena, CA 91125 (USA)
Fax : (‡1)6266838753
As shown in Scheme 1, the benzimidazole moiety has the
same atomic connectivity and hydrogen bonding surface
along the recognition site (shown as bold bonds) as the
1-methylpyrrole-aminoacetyl moiety. In addition to van der
Waals interactions of the aromatic ligand and the minor
E-mail: dervan@caltech.edu
Supporting information [ab initio calculations for four-ring polyamide
subunits PyPyPyIm and PyBiPyIm (Figure S1)] for this article is
available on the WWW under http://www.chemeurj.org/ or from the
author.
2110
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
DOI: 10.1002/chem.200204689
Chem. Eur. J. 2003, 9, 2110 2122

2110 2122
groove contacts, the proton
at the N(1) position of the
benzimidazole can form
hydrogen bonds with the
lone pairs of the N(3) of
purines and the O(2) of
pyrimidines like the pro-
tons of the amide bonds
that link the aromatic rings
heterocycles may influence the cell permeation and distribu-
tion behavior as well as degradation pathways.
Here we report the synthesis of benzimidazole-containing
Boc-protected amino acids, their incorporation into hairpin
polyamides by applying the solid-phase synthetic protocols
published previously,[14] and the evaluation of the DNA
recognition properties of the side-by-side benzimidazole/
imidazole (Bi/Im) pair within the hairpin polyamide motif. A
series of benzimidazole-containing hairpin polyamides with the
general sequence ImImPyPy-(R)H Ng-ImXBiPy-CONHMe
2
in
a
polyamide.
In
(Xˆ Py: 1b; X ˆ Hp: 3b), ImImPyPy-(R)H Ng-PyImBiPy-
2
Scheme 1. Structures of benzi-
midazole-heterocycle construct
and pyrrole-heterocycle dimer
as they are incorporated into
hairpin polyamides. Hydrogen
bonding surfaces that are essen-
tial for minor-groove binding
are indicated in bold type.
Hoechst 33258, which pre-
fers a 1:1 ligand/DNA-rec-
ognition-site binding mode,
the N(1) protons of each
CONHMe (2b) and the corresponding parent compounds
ImImPyPy-(R)H Ng-ImXPyPy-CONHMe (Xˆ Py: 1a; X ˆ
2
Hp: 3a) and ImImPyPy-(R)H Ng-PyImPyPy-CONHMe (2a)
2
benzimidazole
interact
were synthesized (Figure 2). A schematic drawing of the
benzimidazole monomer-containing hairpin polyamides bound
to the matched DNA sequences is illustrated in Figure 1. Four
plasmids were designed to probe the recognition behavior of
these polyamides at the six base pair recognition sites 5'-
TGRACA-3' and 5'-TGGXCA-3' (for 1a,b and 3a,b; R, X ˆ
A, T, G, C) as well as 5'-TGRCAA-3' and 5'-TGGXAA-3' (for
2a,b), which differ at a single internal binding position R and
X, respectively, allowing the systematical analysis of the
recognition profile of the benzimidazole/imidazole pair in a
hairpin polyamide at the two individual base pair positions R¥
S and X ¥ Y (Figure 3). The key building block replacements for
dimers PyPy and PyIm, benzimidazole-pyrrole (BiPy) and
benzimidazole-imidazole (BiIm), respectively, have recently
been reported by Singh[15] and Matsuba.[11] To our knowledge,
benzimidazole moieties incorporated into a side-by-side bind-
ing motif such as shown in Figure 1 have not been examined in
the context of the pairing rules.
through bifurcated hydro-
gen bonds with adenine
N(3)- and thymine O(2)-
atoms on the floor of the
minor groove.[13] In contrast, when incorporating the benzi-
midazole ring into a side-by-side 2:1 ligand/DNA-recognition-
site binding mode (e.g. in a hairpin polyamide) the benzimi-
dazole might be expected to recognize A ¥ Tor T¥ A base pairs
when paired with Py or Hp and a C ¥ G base pair when paired
with Im (Figure 1). Despite its formal similarity with the
binding surface of aminoacetyl-pyrrole, the different elec-
tronic structure, larger ring size, restricted flexibility, and
associated difference in curvature of the benzimidazole may
interfere with a cooperative side-by-side pairing and result in
an altered recognition behavior. In view of biological studies
with polyamide-based minor groove binders, one could
imagine that the replacement of amide bonds by fused
Figure 1. Binding models of the 1:1 polyamide-DNA complexes formed between the benzimidazole-containing polyamides ImImPyPy-(R)H Ng-
2
ImXBiPyCONHMe (X ˆ Py: 1b; X ˆ Hp: 3b) and ImImPyPy-(R)H Ng-PyImBiPyCONHMe (2b) and their respective six base pair match sites 5'-
2
TGGACA-3' and 5'-TGGCAA-3', respectively. Circles with two dots represent the lone pairs of N(3) of purines and O(2) of pyrimidines. Circles containing
an ™H∫ represent the N(2) hydrogen of guanine. Putative hydrogen bonds are illustrated by dotted lines. The incorporated benzimidazole-pyrrole,
benzimidazole-hydroxypyrrole, and benzimidazole-imidazole moieties are shown in bold. Ball-and-stick models of polyamides are shown with shaded and
nonshaded circles indicating imidazole and pyrrole carboxamides, respectively. Two touching squares represent benzimidazole-pyrrole, benzimidazole-
imidazole, and benzimidazole-hydroxypyrrole constructs: the nonshaded square represents the benzimidazole moiety, the adjacent pyrrole, imidazole rings
are depicted as nonshaded and shaded circles inside the second square as described above. Hydroxypyrrole rings are annotated with an ™H∫ in the center of
an unshaded circle. (R)-Diaminobutyric acid (DABA) is depicted as a curved line and a plus sign, the methyl amide tail is annotated as a triangle.
Chem. Eur. J. 2003, 9, 2110 2122
www.chemeurj.org
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2111

P. B. Dervan et al.
FULL PAPER
Figure 2. Structures and ball-and-stick models of the parent eight-ring polyamides 1a, 2a, and 3a, and the corresponding benzimidazole-containing
polyamides 1b, 2b, and 3b.
9a), benzimidazole-methoxypyrrole (Boc-OpBi-OH, 9b),
benzimidazole-pyrrole (Boc-BiPy-OH, 9c), and benzimida-
zole-imidazole-pyrrole (Boc-PyImBi-OH, 11) amino acids. As
shown in Scheme 2, the benzimidazole scaffold was obtained
by cyclodehydration of amides 6a,b, which were in turn
prepared from precursors 4a,b by means of an HBTU-
mediated coupling, in acetic acid[16] or by condensation of
III
pyrrole-aldehyde 4c[17] with ortho-diamine 5 and in situFe
/
FeII-catalyzed oxidative cyclodehydrogenation of the Schiff×s
base intermediate.[15] Subsequent reduction of the nitro group
in 7a c and Boc protection of the resulting aminoesters
furnished the protected aromatic amino acid esters 8a c in
moderate to good yield. The desired carboxylic acids 9a
c
were finally obtained by hydrolysis of the esters 8a c with
sodium hydroxide in dioxane. As will be discussed in the
following paragraph, it was found advantageous for the solid-
phase synthesis to use Boc protected amino acid 11 instead of
9a as building block to incorporate the BiIm moiety into the
hairpin polyamide 2b. For the synthesis of trimer 11 nitro
compound 7a was reduced and subsequently reacted with
Boc-Py-OBt to give ester 10 in moderate yield. Saponification
of 10 furnished the protected amino acid 11 in 94% yield.
Figure 3. Schematic ball-and-stick models of hairpin polyamides 1b, 2b,
and 3b bound to recognition sequences 5'-TGRACA-3', 5'-TGRCAA-3',
5'-TGGXCA-3', and 5'-TGGXAA-3' (R, X ˆ A, T, G or C). Ring pairing-
DNA interactions investigated by quantitative DNase I footprint titrations
are boxed.
Solid-phase synthesis: Parent polyamides 1a, 2a, and the
O-protected derivative 12a were synthesized by solid-phase
methods on Kaiser×s oxime resin (0.48 mmolgÀ1) in a stepwise
manner from Boc-N-methyl pyrrole (Boc-Py-OBt, 13a), Boc-
N-methyl imidazole (Boc-Im-OH, 14), Boc-N-methyl me-
Results and Discussion
Synthesis of building blocks: Preparation of benzimidazole-
containing polyamides 1b, 2b, and 3b required the synthesis
of Boc-protected benzimidazole-imidazole (Boc-ImBi-OH,
2112
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2003, 9, 2110 2122

Alternative Heterocycles for DNA Recognition
2110 2122
Scheme 2. Synthesis of Boc-protected benzimidazole amino acids 9a c and 11. a) HBTU, DIEA, DMF; b) AcOH; c) FeCl3 ¥ 6H2O, DMF; d) for 7a,c:
1) H2, Pd/C, DMF; 2) Boc2O, DIEA; for 7b: 1) SnCl2 ¥ 2H2O, DMF; 2) Boc2O, DIEA; e) NaOH, dioxane; f) 1) H2, Pd/C, DMF; 2) Boc-Py-OBt, DIEA.
(3a) and ImImPyPy-(R)H Ng-ImHpBiPy-CONHMe (3b). The
synthetic protocol for the solid-phase synthesis of benzimi-
dazole-containing polyamides 1b, 2b, and 3b is illustrated in
Scheme 3.
2
thoxypyrrole (Boc-Op-OH, 15) monomers, a-Fmoc-g-Boc-
(R)-2,4-diaminobutyric acid (a-Fmoc-g-Boc-(R)-DABA, 16),
and N-methyl imidazole dimer (ImIm-OH, 17) in 16 steps.[14]
Similarly, benzimidazoles 9a c could be successfully used as
building blocks for the stepwise polyamide growth furnishing
resin-bound benzimidazole-containing polyamides 1b, 2b,
and the O-protected derivative 12b in 13 steps (Scheme 3). It
is noteworthy that the same reaction conditions proved to be
suitable for peptide coupling and amine deprotection of
benzimidazole monomers 9a c and the standard building
blocks 13 17: Benzimidazoles 9a c were coupled in DIEA/
NMP after activation with HBTU and (when resin-bound)
deprotected with 20% TFA/CH2Cl2 (9c) or 50% TFA/CH2Cl2
(9a, b). However, due to the moderate efficiency of the Boc-
Py-OBt to imidazole amine coupling,[18] it was found advanta-
geous in terms of the overall polyamide recovery to employ
building block 11 instead of 9a for the synthesis of polyamide
2b. To liberate the resin-bound compounds and concomitant-
ly remove the Fmoc group of the DABA turn, a sample of
each polyamide-loaded polymer support was treated with
methylamine in CH2Cl2/THF (14 h, 378C). The crude poly-
amides were isolated and purified by preparatory reversed-
DNA binding affinities and sequence specificities: When
replacing PyPy, PyIm, and PyHp dimers with BiPy (as in 1b),
BiIm (as in 2b), and BiHp constructs (as in 3b), respectively,
the question arises whether a) the benzimidazole ring is
capable of cooperating with the imidazole ring on the
opposite strand of the hairpin polyamide to specifically
recognize a C ¥ G base pair according to the binding model
and the general pairing rules discussed above, and b) whether
the steric and electronic peculiarities of the benzimidazole
ring disturb the binding ability of the adjacent heterocycle. In
order to address this question for the given polyamide design,
the recognition behavior of the benzimidazole-containing
polyamides was probed in two of the six base pair positions
R ¥ S and X ¥ Y of the recognition sequence (5'-TGRACA-3'
and 5'-TGGXCA-3' for 1b and 3b; 5'-TGRCAA-3' and 5'-
TGGXAA-3' for 2b; Figure 3) and compared with the DNA-
binding ability of the parent compounds 1a, 2a, and 3a at the
same recognition sequences.
Evaluation of the DNA recognition properties of the new
ligands 1b, 2b, and 3b at the imidazole/benzimidazole
position (referred to as R ¥ S position) should reveal the
sequence preference of the benzimidazole ring. The binding
properties at the neighboring ring position (referred to as X ¥
Y position) were expected to disclose whether the recognition
capabilities of the adjacent internal five-membered ring are
disturbed by the benzimidazole moiety. To examine the
phase HPLC to yield ImImPyPy-(R)H Ng-ImPyPyPy-CON-
2
HMe (1a), ImImPyPy-(R)H Ng-ImPyBiPy-CONHMe (1b),
2
ImImPyPy-(R)H Ng-PyImPyPy-CONHMe (2a), ImImPyPy-
2
(R)H Ng-PyImBiPy-CONHMe (2b), ImImPyPy-(R)H Ng-Im-
2
2
OpPyPy-CONHMe (12a), ImImPyPy-(R)H Ng-ImOpBiPy-
2
CONHMe (12b). O-Methyl protected polyamides 12a, b
were O-demethylated with sodium thiophenoxide/DMF
(858C, 2 h)[19] and purified by preparatory reversed-phase
HPLC to yield ImImPyPy-(R)H Ng-ImHpPyPy-CONHMe
2
Chem. Eur. J. 2003, 9, 2110 2122
www.chemeurj.org
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2113

P. B. Dervan et al.
FULL PAPER
Scheme 3. Solid-phase synthesis of hairpin polyamides 1b, 2b, and 3b and amino acid building blocks for the polyamide synthesis. a) 13a, DIEA, NMP;
b) 20% TFA/CH2Cl2; c) for X ˆ N: 9a, HBTU, DIEA, NMP; for X ˆ COCH3: 9b, HBTU, DIEA, NMP; for X ˆ CH: 9c, HBTU, DIEA, NMP; d) for X ˆ
CH: 20% TFA/CH2Cl2; for X ˆ N, COCH3: 50% TFA/CH2Cl2; e) for X ˆ COCH3, CH: 14, HBTU, DIEA, NMP; for X ˆ N: 13b, DCC, DMAP, CH2Cl2;
f) for Yˆ CH: 20% TFA/CH2Cl2; for Yˆ N: 50% TFA/CH2Cl2; g) 16, HBTU, DIEA, NMP; h) 20% TFA/CH2Cl2; i) 13a, DIEA, NMP; j) 20% TFA/
CH2Cl2; k) 13a, DIEA, NMP; l) 20% TFA/CH2Cl2; m) 17, DCC, HOBt, DIEA, NMP; n) 2.0m CH3NH2/THF, CH2Cl2; o) PhSNa, DMF; p) 13a, DIEA,
NMP; q) 20% TFA/CH2Cl2; r) 11, HBTU, DIEA, NMP; s) 20% TFA/CH2Cl2; t) 16, HBTU, DIEA, NMP; u) 20% TFA/CH2Cl2; v) 13a, DIEA, NMP;
w) 20% TFA/CH2Cl2; x) 13a, DIEA, NMP; y) 20% TFA/CH2Cl2; z) 17, DCC, HOBt, DIEA, NMP.
consequences of the ring replacement for DNA binding in
terms of affinity and specificity, the benzimidazole-containing
hairpin polyamides 1b, 2b, and 3b were evaluated by
quantitative DNase I footprint titrations and the results
compared to those from parent compounds 1a, 2a, and 3a,
which were investigated under identical conditions. Quanti-
tative DNase I footprint titration experiments (10 mm Tris-
HCl, 10 mm KCl, 10 mm MgCl2, 5 mm CaCl2, pH 7.0, 228C,
equilibration time: 12 h)[20] were performed on 32P-3'-labeled
EcoRI/PvuII restriction fragments of plasmids pPWF2 and
pCAB1 (to probe the R ¥ S position, Figure 4); and pAU2[21]
and pCAB2 (to probe the X ¥ Y position, Figure 5). Equili-
brium association constants (Ka) for polyamides 1a, b, 2a, b,
and 3a, b at the binding sites of interest were determined as
described previously.[20]
On first inspection, analysis of the footprint data at position
R ¥ S (Table 1) and at position X ¥ Y (Table 2) extracted from
the autoradiograms shown in Figures 6 8 reveals that the
benzimidazole-containing polyamides 1b, 2b, 3b and the
corresponding parent compounds 1a, 2a, 3a exhibit the same
binding site preferences for matched and mismatched sites on
the 3'-labeled restriction fragments derived from pPWF2,
pCAB1, pAU2, and pCAB2. The expected match sites out of
the sequence families 5'-TGRACA-3' or 5'-TGRCAA-3' (R¥ S
position) and 5'-TGGXCA-3' or 5'-TGGXAA-3' (X ¥ Y
position) are bound with comparable equilibrium association
constants by the benzimidazole-containing compounds 1b,
2b, and 3b and the parent hairpin polyamides 1a, 2a, and 3a.
All compounds bind their match sites with high affinities in
the subnanomolar range (Ka values range from 4.9 Â 1010 mÀ1
for R ¥ S position of 1b and X ¥ Y position of 1a down to 3.5 Â
109 mÀ1 for X ¥ Y position of 2a).
A closer inspection of the binding affinities at the two
individual base pair positions elucidates the subtle effects of
the benzimidazole introduction in the molecular recognition:
Molecular recognition at the R ¥ S position: While both
benzimidazole-containing polyamide 1b and parent com-
pound 1a display comparable match-site affinities (Ka ˆ 4.9 Â
1010 mÀ1 and Ka ˆ 2.7  1010 mÀ1, respectively), benzimidazole-
containing polyamides 2b and 3b exhibit a six to seven-fold
higher match site affinity (Ka ˆ 3.1  1010 mÀ1 and Ka ˆ 2.5 Â
1010 mÀ1, respectively) than the parent DNA binders 2a and 3a
(Ka ˆ 5.2  109 mÀ1 and Ka ˆ 3.8  109 mÀ1, respectively). A
significant difference can be seen for the two sets of
compounds when their mismatch-site affinities and the
2114
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2003, 9, 2110 2122

Alternative Heterocycles for DNA Recognition
2110 2122
Figure 4. Illustration of the EcoRI/PvuII restriction fragments derived from plasmids a) pPWF2 and b) pCAB1. The four designed binding sites that were
analyzed in quantitative footprint titrations are indicated with a box surrounding each of the six bp sites. The R ¥ S base pair position within the binding sites is
shown in bold type.
Figure 5. Illustration of the EcoRI/PvuII restriction fragments derived from plasmids a) pAU2 and b) pCAB2. The four designed binding sites that were
analyzed in quantitative footprint titrations are indicated with a box surrounding each of the six bp sites. The X ¥ Y base pair position within the binding sites is
shown in bold type.
associated single site specificities are compared. The replace-
ment of the PyPy, PyIm, and PyHp constructs by the
benzimidazole-Py, -Im, and -Hp analogues is generally
accompanied by an increase in the binding affinities for the
single base pair mismatch sites of benzimidazole-containing
polyamides 1b, 2b, and 3b resulting in overall mildly
diminished binding specificities. More precisely, for polya-
mide 1b single base pair mismatch specificities are lower by
roughly one order of magnitude when compared with the
corresponding parent compounds 1a. By contrast, compound
3b displays a specificity profile that is comparable to that of
the parent compound 3a. For benzimidazole-containing
compound 2b the match site/mismatch site discrimination is
comparable for R ˆ A but reveals at least ten-fold lower
specificities for R ˆ T or C.
mismatch sites and the associated binding specificities reveal
the same overall trend and range in the same order of
magnitude for both the benzimidazole-containing polyamides
1b, 3b and the parent compounds 1a, 3a. By contrast,
benzimidazole-containing polyamide 2b exhibits a six-fold
higher match site affinity than its parent binder 2a (Ka ˆ 1.9 Â
1010 mÀ1 vs Ka ˆ 3.5  109 mÀ1, respectively). Along with the
significantly higher match site affinity of compound 2b
slightly increased binding affinities for single base pair
mismatch sites are displayed, which result in mildly dimin-
ished specificities when compared with the values of the
parent compounds 2a. Obviously, the introduction of the
benzimidazole moiety has only a marginal influence on the
binding properties of the adjacent five-membered internal
ring.
While the relative contributions of various ligand DNA
interactions to the overall recognition profile of a DNA
binder are still subject of discussion,[22, 23] hydrogen bonding,
van der Waals contacts and electrostatic interactions proved
to be the major determinants for DNA recognition. Factors
that have to be taken into account to rationalize the slightly
different DNA recognition properties of the benzimidazole-
Molecular recognition at the X ¥ Y position: As expected, for
the X ¥ Y position of the sequence family 5'-TGGXCA-3'
(compounds 1a, b and 3a, b) and 5'-TGGXAA-3' (com-
pounds 2a, b) the response to the heterocycle replacement in
terms of match site/single base pair mismatch site discrim-
ination is less pronounced. Binding affinities for match and
Chem. Eur. J. 2003, 9, 2110 2122
www.chemeurj.org
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2115

P. B. Dervan et al.
FULL PAPER
Table 1. Equilibrium association constants Ka [mÀ1] and specificities for polyamides 1a, 1b, 2a, 2b, 3a, and 3b on
with the aminoacetyl-pyrrole
system the benzimidazole moi-
ety has with its greater aromatic
surface a higher hydrophobicity
and polarizability that may al-
ter both van der Waals DNA/
ligand interactions and the in-
tra-ligand p-stacking of the two
opposite polyamide strands of
the hairpin.[1a, 10b] b) Gas-phase
calculations of PyBiPyIm and
PyPyPyIm fragments indicate
that the benzimidazole-con-
taining strand PyBiPyIm has a
smaller degree of curvature
than the corresponding PyPy-
PyIm strand (see Figure S1,
Supporting Information).[24] In
general, as hairpin polyamides
comprised of Py, Im, and Hp
are slightly overcurved with
respect to the DNA, the de-
creased curvature of the benzi-
midazole-containing strand of
the hairpin polyamide should
improve shape complementari-
ty between the recognition face
of the ligand and the floor of
the minor groove of DNA and
therefore increase binding af-
finity. However, for the given
benzimidazole-containing hair-
pin polyamides it should be
noticed that the larger curva-
ture of the remaining unmodi-
fied strand might diminish the
cooperative side-by-side ring
pairing and as a consequence
reduce single base pair mis-
match specificity compared to
the parent hairpin polyamides.
c) The replacement of an amide
d]
plasmids pPWF2 and pCAB1 (R ¥ S position).[a
Polyamide on pPWF2
5'-aTGAACAt-3'
5'-aTGTACAt-3'
5'-aTGGACAt-3'
2.7 (Æ0.5) Â 1010
5'-aTGCACAt-3'
4.4 (Æ1.2) Â 108
ꢀ 1 Â 108
[ꢁ270]
ꢀ 1 Â 108
[ꢁ270]
[61]
1a
1b
3a
3.9 (Æ0.2) Â 109
3.3 (Æ0.3) Â 109
4.9 (Æ0.1) Â 1010
3.8 (Æ0.2) Â 109
2.5 (Æ0.4) Â 1010
9.0 (Æ0.5) Â 108
[13]
[15]
[54]
ꢀ 3 Â 107
[ꢁ127]
ꢀ 1 Â 107
[ꢁ380]
ꢀ 1 Â 107
[ꢁ380]
3.3 (Æ0.7) Â 108
2.4 (Æ0.4) Â 108
1.0 (Æ0.3) Â 108
[76]
[104]
[250]
3b
Polyamide on pCAB1
5'-gTGACAAt-3'
5'-gTGTCAAt-3'
5'-gTGGCAAt-3'
5.2 (Æ0.2) Â 109
5'-gTGCCAAt-3'
ꢀ 1 Â 108
[ꢁ52]
ꢀ 1 Â 107
[ꢁ520]
ꢀ 1 Â 107
[ꢁ520]
2a
7.9 (Æ1.4) Â 108
6.2 (Æ1.4) Â 108
3.1 (Æ0.5) Â 1010
4.0 (Æ0.7) Â 108
[39]
[50]
[78]
2b
[a] The reported association constants Ka are the average values obtained from three DNase I footprint titration
experiments, with the standard deviation for each data set indicated in parentheses. [b] The assays were carried
out at 228C at pH 7.0 in the presence of 10 mm Tris-HCl, 10 mm KCl, 10 mm MgCl2, and 5 mm CaCl2 with an
equilibration time of 12 h. [c] Match site equilibration association constants are shown in bold. [d] Specificities
are given in square brackets under the Ka values and calculated as Ka (match)/Ka(mismatch).
Table 2. Equilibrium association constants Ka [mÀ1] and specificities for polyamides 1a, 1b, 2a, 2b, 3a, and 3b on
d]
plasmids pAU2 and pCAB2 (X ¥ Y position).[a
Polyamide on pAU2
5'-aTGGACAt-3'
4.9 (Æ1.8) Â 1010
5'-aTGGTCAt-3'
4.9 (Æ0.2) Â 1010
5'-aTGGGCAt-3'
5'-aTGGCCAt-3'
2.0 (Æ0.5) Â 109
4.5 (Æ0.8) Â 109
[25]
[11]
1a
1b
3a
2.7 (Æ1.0) Â 1010
8.9 (Æ1.3) Â 109
3.6 (Æ0.9) Â 1010
3.4 (Æ0.6) Â 109
5.7 (Æ0.5) Â 109
[11][e]
[6][e]
1.3 (Æ0.1) Â 109
1.8 (Æ0.4) Â 108
1.7 (Æ0.6) Â 108
[7]
[49]
[52]
1.8 (Æ0.7) Â 1010
2.5 (Æ0.3) Â 109
5.8 (Æ1.8) Â 108
7.0 (Æ0.6) Â 108
[7]
[31]
[25]
3b
linkage by
a benzimidazole
Polyamide on pCAB2
5'-gTGGAAAt-3'
5'-gTGGTAAt-3'
5'-gTGGGAAt-3'
5'-gTGGCAAt-3'
3.5 (Æ0.5) Â 109
moiety reduces the degree of
conformational freedom which
is expected to result in an
increased preorganisation of
the ligand leading to higher
binding affinity.
ꢀ 1 Â 107
[ꢁ350]
ꢀ 1 Â 108
[ꢁ35]
ꢀ 1 Â 107
[ꢁ350]
2a
ꢁ 1 Â 108
[ꢁ190]
1.6 (Æ0.1) Â 109
2.8 (Æ0.3) Â 108
1.9 (Æ0.3) Â 1010
[12]
[68]
2b
The results from the dis-
cussed quantitative DNase I
footprint titrations demonstrate
that the benzene ring of the
benzimidazole moiety is able to
pair with an Im residue within
the context of a hairpin poly-
[a] The reported association constants Ka are the average values obtained from three DNase I footprint titration
experiments, with the standard deviation for each data set indicated in parentheses. [b] The assays were carried
out at 228C at pH 7.0 in the presence of 10 mm Tris-HCl, 10 mm KCl, 10 mm MgCl2, and 5 mm CaCl2 with an
equilibration time of 12 h. [c] Match site equilibration association constants are shown in bold type.
[d] Specificities are given in square brackets under the Ka values and calculated as Ka (match)/Ka (mismatch).
[e] Specificity calculated as Ka (5'-aTGGTCAt-3')/Ka (mismatch).
containing hairpin polyamides when compared to their parent
compounds can be categorized into structural and electronic
characteristics of the benzimidazole moiety: a) In comparison
amide and that, as a consequence, this pairing is able to
distinguish G ¥ C from C ¥ G, A ¥ T, and T¥ A. Therefore, the
recognition behavior of the studied benzimidazole/imidazole
2116
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2003, 9, 2110 2122

Alternative Heterocycles for DNA Recognition
2110 2122
Figure 6. Quantitative DNase I footprint titration experiments on the 3'-32P-labeled 283-bp and 263-bp EcoRI/PvuII restriction fragments derived from
plasmids pPWF2 (a and b) and pAU2 (c and d), respectively. a) and c) polyamide 1a: lane 1, intact DNA; lane 2, A-specific reaction; lane 3, G-specific
reaction; lane 4, DNase I standard; lanes 5 13: 3, 10, 30, 100, 300pm and 1, 3, 10, 30nm polyamide, respectively. b) and d) polyamide 1b: lane 1, intact DNA;
lane 2, A-specific reaction; lane 3, G-specific reaction; lane 4, DNase I standard; lanes 5 13: 3, 10, 30, 100, 300pm and 1, 3, 10, 30nm polyamide, respectively.
The analyzed 6-bp binding site locations are designated in brackets along the right side of each autoradiogram with their respective unique base pairs
indicated. Schematic binding models of 1a and 1b with their putative binding sites are shown on the top side of the autoradiograms. Flanking sequences are
designated in lower case while the binding site is given in capitals. The boxed R ¥ S and X ¥ Y base pairs indicate the positions that were examined in the
experiments. [a] Additional match site for 1a,b; sequence 5'-aTGGTCAt-3'.
(Bi/Im) pair is consistent with the well-established binding
model discussed above. Presumably, a combination of elec-
tronic properties, steric bulk, rigidity and the associated
alteration of the overall curvature of the benzimidazole
moiety translate into a mildly diminished single base pair
mismatch specificity in the R ¥ S position, when compared with
the parent compounds. However, the benzimidazole moiety
minimally disturbs the recognition properties of the adjacent
internal five-membered heterocycle resulting in single base
pair mismatch specificities at the X ¥ Y position that are
similar to those of the parent polyamides. In addition, all
benzimidazole-polyamides retain or even exceed the subna-
nomolar association constants of their corresponding parent
compounds at their match sites.
(9a), benzimidazole-methoxypyrrole (9b), benzimidazole-
pyrrole (9c), and benzimidazole-imidazole-pyrrole (11) ami-
no acids as a formal replacement for PyIm, PyHp, PyPy
dimers and a PyImPy trimer, respectively. The new building
blocks were synthesized from diamine 5 and pyrrole carbal-
dehyde 4c, imidazole carboxylic acid 4a, and hydroxypyrrole
carboxylic acid 4b, respectively. The evaluation of the DNA
binding properties of the hairpin polyamides 1b, 2b, and 3b
by means of quantitative DNase I footprint titrations dem-
onstrated that the benzimidazole ring in a Im/Bi side-by-side
pair exhibits the same recognition preferences for a G ¥ C base
pair as a Py ring in an Im/Py side-by-side pair. However, the
pyrrole-to-benzimidazole ring replacement results in a mildly
diminished specificity at the replacement position R ¥ S. The
binding properties of the adjacent five-membered hetero-
cycles at position X ¥ Y are only marginally influenced by the
steric demand and electronic properties of the benzimidazole
ring. The subnanomolar DNA binding affinities for matched
binding sites displayed by the parent polyamides are fully
retained (for 1b) or even exceeded (for 2b, 3b). The BiIm-
Conclusion
Benzimidazole moieties were incorporated into hairpin poly-
amides 1b, 2b and 3b in the form of benzimidazole-imidazole
Chem. Eur. J. 2003, 9, 2110 2122
www.chemeurj.org
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2117

P. B. Dervan et al.
FULL PAPER
Figure 7. Quantitative DNase I footprint titration experiments on the 3'-32P-labeled 284-bp EcoRI/PvuII restriction fragments derived from plasmid pCAB1
(a and b) and pCAB2 (c and d). a) and c) polyamide 2a: lane 1, intact DNA; lane 2, A-specific reaction; lane 3, G-specific reaction; lane 4, DNase I standard;
lanes 5 13: 3, 10, 30, 100, 300pm and 1, 3, 10, 30 nm polyamide, respectively. b) and d) polyamide 2b: lane 1, intact DNA; lane 2, A-specific reaction; lane 3,
G-specific reaction; lane 4, DNase I standard; lanes 5 13: 3, 10, 30, 100, 300pm and 1, 3, 10, 30nm polyamide, respectively. The analyzed 6-bp binding site
locations are designated in brackets along the right side of each autoradiogram with their respective unique base pairs indicated. Schematic binding models of
2a and 2b with their putative binding sites are shown on the top side of the autoradiograms. Flanking sequences are designated in lower case while the
binding site is given in capitals. The boxed R ¥ S and X ¥ Y base pairs indicate the positions that were examined in the experiments. [a] Three overlapping 1-bp
mismatch sites for 2a,b; sequence 5'-gAAGCTTGGCGTa-3'.
from Halocarbon. All other chemicals were obtained reagent-grade from
Aldrich (unless otherwise stated) and used without further purification.
1H NMR and 13C NMR spectra were recorded on a Varian Mercury 300
and BiHp-containing hairpin polyamides 2b and 3b exhibit a
significant increase in binding affinity compared to the parent
compounds. Even if the recognition profile of the benzimida-
instrument. Chemical shifts are reported in ppm with reference to the
zole ring was only studied in a side-by-side pair with Im, it is
expected that in analogy Bi/Py and Bi/Hp side-by-side pairs
will recognize A ¥ T or T¥ A base pairs in accordance with the
general pairing rules.[25] Future studies will address the
influence of the benzimidazole moiety on cellular uptake
and cellular distribution properties.[26]
solvent residual signal. UV spectra were measured on a Beckman Coulter
DU 7400 diode array spectrophotometer. Matrix-assisted, laser desorption/
ionization time-of-flight mass spectrometry (MALDI-TOF-MS), electro-
spray injection (ESI-MS) and high resolution mass spectrometry (FAB, EI)
was carried out at the California Institute of Technology. HPLC analysis
was performed on a Beckman Gold system using a RAININ C18, Microsorb
MV, 5 mm, 300 Â 4.6 mm reversed-phase column in 0.1% (w/v) TFA with
acetonitrile as eluent and a flow rate of 1.0 mLminÀ1, gradient elution
1.25% acetonitrileminÀ1. Preparatory HPLC was carried out on a Beckman
HPLC using a Waters DeltaPak 100 Â 25 mm, 100 mm C18 column, 0.1%
(w/v) TFA, 0.25% acetonitrileminÀ1. 18 MW water was obtained from a
Millipore MilliQ water purification system, and all buffers were 0.2 mm
filtered. DNA oligonucleotides were synthesized by the Biopolymer
Synthesis Center at the California Institute of Technology and used
without further purification. Plasmids were sequenced by the Sequence/
Structure Analysis Facility (SAF) at the California Institute of Technology.
dNTPs (PCR nucleotide mix) and all enzymes (unless otherwise stated)
were purchased from Roche Diagnostics and used with their supplied
buffers. pUC19 was from New England Biolabs. [a-32P]-Deoxyadenosine
triphosphate and [a-32P]-thymine triphosphate was from New England
Nucleotides. RNase-free water (used for all DNA manipulations) was from
Experimental Section
General: N,N'-Dicyclohexylcarbodiimide (DCC), N-hydroxybenzotriazole
(HOBt), and 2-(1H-benzotriazole-1-yl)-1,1,3,3-tetramethyluronium hexa-
fluorophosphate (HBTU) were purchased from Peptides International.
Oxime resin was purchased from Novabiochem (0.48 mmolgÀ1, batch no.
A18763). N,N-Diisopropylethylamine (DIEA) and N,N-dimethylforma-
mide (DMF) were purchased from Applied Biosystems. (R)-2-Fmoc-4-
Boc-diaminobutyric acid (a-Fmoc-g-Boc-(R)-DABA, 16) was from Bach-
em, methyl 3,4-diaminobenzoate (5) from Avocado, dichloromethane
(DCM) was reagent grade from EM, and trifluoroacetic acid (TFA) was
2118
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2003, 9, 2110 2122

Alternative Heterocycles for DNA Recognition
2110 2122
Figure 8. Quantitative DNase I footprint titration experiments on the 3'-32P-labeled 283-bp and 263-bp EcoRI/PvuII restriction fragments derived from
plasmids pPWF2 (a and b) and pAU2 (c and d), respectively. a) and c) polyamide 3a: lane 1, intact DNA; lane 2, A-specific reaction; lane 3, G-specific
reaction; lane 4, DNase I standard; lanes 5 13: 10, 30, 100, 300pm and 1, 3, 10, 30, 100nm polyamide, respectively. b) and d) polyamide 3b: lane 1, intact
DNA; lane 2, A specific reaction; lane 3, G specific reaction; lane 4, DNase I standard; lanes 5 13: 10, 30, 100, 300pm and 1, 3, 10, 30, 100 nm polyamide,
respectively. The analyzed 6-bp binding site locations are designated in brackets along the right side of each autoradiogram with their respective unique base
pairs indicated. Schematic binding models of 3a and 3b with their putative binding sites are shown on the top side of the autoradiograms. Flanking sequences
are designated in lower case while the binding site is given in capitals. The boxed R ¥ S and X ¥ Y base pairs indicate the positions that were examined in the
experiments. [a] Additional 1-bp mismatch site for 3a,b; sequence 5'-aTGGTCAt-3'.
US Biochemicals. Ethanol (100%) was from Equistar, isopropanol from
Mallinckrodt. Bromophenol blue and xylene cyanol FF were from Acros.
DNA manipulations were performed according to standard protocols.[27]
Autoradiography was performed with a Molecular Dynamics Typhoon
HBTU (5.72 g, 15.0 mmol) was added to the solution followed by DIEA
(3 mL). After stirring for 20 h at ambient temperature the reaction mixture
was poured into ice water. The precipitate was filtered, washed with water
and dried in vacuo to yield the title compound 6a as a yellow powder
1
Phosphorimager.
1-Methyl-4-nitro-1H-imidazole-2-carboxylic
acid
(4.22 g, 88%). H NMR (300 MHz, [D6]DMSO): d ˆ 3.74 (s, 3H), 4.01 (s,
(4a),[14a] 1-methyl-4-nitro-1H-pyrrole-2-carbaldehyde (4c),[17] 3-methoxy-
1-methyl-4-nitro-1H-pyrrole-2-carboxylic acid ethyl ester[28] were synthe-
sized according to literature procedures.
3H), 5.89 (s, 2H), 6.74 (d, J ˆ 8.8 Hz, 1H), 7.59 (dd, J1 ˆ 8.8 Hz, J2 ˆ 1.6 Hz,
1H), 7.71 (d, J ˆ 1.6 Hz, 1H), 8.64 (s, 1H), 10.04 (s, 1H); 13C NMR
(75 MHz, [D6]DMSO): d ˆ 36.6, 51.4, 114.5, 115.9, 120.3, 126.5, 128.7, 129.0,
‡
137.7, 144.1, 148.6, 156.7, 165.8; MS (ESI): m/z: 320 [M‡H] ; HR-MS (EI):
Building block synthesis
calcd for C13H13N5O5: 319.0916; found: 319.0929.
3-Methoxy-1-methyl-4-nitro-1H-pyrrole-2-carboxylic acid (4b): 3-Me-
thoxy-1-methyl-4-nitro-1H-pyrrole-2-carboxylic ethyl ester (5.50 g,
24.1 mmol) was dissolved in ethanol (100 mL). NaOH (aqueous, 1m,
100 mL) was added and the solution stirred for 13 h at ambient temper-
ature. The yellow solution was carefully acidified with aqueous HCl (1m) to
pH 2 3. The formed white precipitate was filtered, washed with water and
dried in vacuo to yield the title compound 4b as a white powder (3.85 g,
80%). 1H NMR (300 MHz, [D6]DMSO): d ˆ 3.80 (s, 3H), 3.82 (s, 3H), 8.10
(s, 1H); 13C NMR (75 MHz, [D6]DMSO): d ˆ 38.4, 62.4, 114.6, 126.4, 127.0,
4-Amino-3-[(3-methoxy-1-methyl-4-nitro-1H-pyrrole-2-carbonyl)-amino]-
benzoic methyl ester (6b): Carboxylic acid 4b (2.17 g, 10.8 mmol) and
methyl 3,4-diamino-benzoate (5) (1.81 g, 10.8 mmol) were dissolved in
DMF (26 mL). HBTU (4.11 g, 10.8 mmol) was added to the solution
followed by DIEA (2.6 mL). After stirring for 20 h at ambient temperature
the reaction mixture was poured into ice water. The precipitate was
filtered, washed with water and dried in vacuo to yield the title compound
6b as a yellow powder (3.66 g, 97%). 1H NMR (300 MHz, [D6]DMSO):
d ˆ 3.75 (s, 3H), 3.86 (s, 3H), 3.95 (s, 3H), 5.82 (s, 2H), 6.78 (d, J ˆ 8.4 Hz,
1H), 7.57 (dd, J1 ˆ 8.4 Hz, J2 ˆ 2.0 Hz, 1H), 7.93 (d, J ˆ 2.0 Hz, 1H), 8.13 (s,
1H), 8.95 (s, 1H); 13C NMR (75 MHz, [D6]DMSO): d ˆ 38.0, 51.4, 62.9,
114.7, 116.4, 116.5, 121.3, 125.9, 126.4, 127.3, 128.0, 141.4, 147.3, 157.8, 165.9;
‡
145.1, 160.5; MS (EI): m/z: 201 [M‡H] ; HR-MS (EI): calcd for C7H8N2O5:
200.0433; found: 200.0435.
4-Amino-3-[1-methyl-4-nitro-1H-imidazole-2-carbonyl)-amino]-benzoic
methyl ester (6a): Carboxylic acid 4a (2.57 g, 15.0 mmol) and methyl 3,4-
diamino-benzoate (5) (2.49 g, 15.0 mmol) were dissolved in DMF (30 mL).
Chem. Eur. J. 2003, 9, 2110 2122
www.chemeurj.org
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2119

P. B. Dervan et al.
FULL PAPER
‡
MS (ESI): m/z: 349 [M‡H] ; HR-MS (EI): calcd for C15H16N4O6: 348.1069;
solvent was evaporated, and the residue was extracted several times with
ethyl acetate. The extracts were combined, washed with water, and dried
with MgSO4. After evaporation of the solvent the crude product was
purified by column chromatography on silica gel (CH2Cl2/ethyl acetate 4:1)
to afford the title compound 8b as a white solid (0.56 g, 48%). 1H NMR
(300 MHz, [D6]DMSO): d ˆ 1.44 (s, 9H), 3.79 (s, 3H), 3.85 (s, 3H), 3.96 (s,
3H), 6.96 (s, 1H), 7.56 7.79 (m, 2H), 8.16 (s, 1H), 8.55 (s, 1H), 11.94, 12.01
found: 348.1072.
2-(1-Methyl-4-nitro-1H-imidazol-2-yl)-1H-benzimidazole-5-carboxylic
methyl ester (7a): Amide 6a (3.99 g, 12.5 mmol) was suspended in acetic
acid (50 mL) and heated to 1408C for 6 h. The solvent was removed in
vacuo, diethyl ether was added, the yellow precipitate filtered and dried in
vacuo affording the title compound 7a as an off-white powder (3.57 g,
95%). 1H NMR (300 MHz, [D6]DMSO): d ˆ 3.87 (s, 3H), 4.26 (s, 3H),
7.57 8.34 (m, 3H), 8.71 (s, 1H), 13.78 (s, 1H); 13C NMR (75 MHz,
[D6]DMSO): d ˆ 36.8, 52.1, 112.2, 113.8, 119.1, 120.9, 123.8, 124.5, 126.2,
‡
(2s, 1H); MS (ESI): m/z: 401 [M‡H] ; HR-MS (EI): calcd for C20H24N4O5:
400.1746; found: 400.1742.
2-(4-tert-Butoxycarbonylamino-1-methyl-1H-pyrrol-2-yl)-1H-benzimida-
zole-5-carboxylic acid methyl ester (8c): Nitro-ester 7c (1.50 g, 5.00 mmol)
was dissolved in DMF (250 mL). Pd/C (10 wt.%, 300 mg) was added and
the mixture was hydrogenated at 600 psi overnight at ambient temperature.
The reaction mixture was filtered through Celite to remove the catalyst and
the filtrate was treated with Boc2O (1.63 g, 7.50 mmol) and DIEA (25 mL)
and stirred at 708C for 2 d. After evaporation of the solvent the residue was
dissolved in diethyl ether (150 mL) and the solution was washed with water
and brine. The organic phase was dried with MgSO4, the solvent was
evaporated, and the crude product purified by column chromatography on
silica gel (CH2Cl2/EtOAc 4:1) to give the title compound 8c as a white solid
‡
136.3, 144.2, 145.6, 166.3; MS (ESI): m/z: 302 [M‡H] ; HR-MS (EI): calcd
for C13H11N5O4: 301.0811; found: 301.0805.
2-(3-Methoxy-1-methyl-4-nitro-1H-pyrrol-2-yl)-3H-benzimidazole-5-car-
boxylic methyl ester (7b): Amide 6b (3.56 g, 10.2 mmol) was suspended in
acetic acid (40 mL) and refluxed at 1408C for 12 h. Upon cooling to room
temperature a yellow solid precipitated that was filtered, suspended in
diethyl ether, filtered, and dried in vacuo to yield the title compound 7b as
a yellow solid (3.33, 98%). 1H NMR (300 MHz, [D6]DMSO): d ˆ 3.93 (s,
1.5H), 3.94 (s, 1.5H), 3.97 (s, 3H), 4.15 (s, 3H), 7.72 7.95 (m, 2H), 8.24 (s,
1H), 8.25 8.32 (m, 1H), 12.44 (s, 0.5H), 12.52 (s, 0.5H). At room
temperature the presence of two benzimidazole tautomers (1H/3H) could
be deduced from the 1H NMR spectra. The ratio between the two isomers
was roughly estimated as 1:1 based on the integration of the protons at d ˆ
3.93/3.94 and 12.44/12.52. 13C NMR (75 MHz, [D6]DMSO): d ˆ 38.5, 52.0,
62.7, 111.8, 113.62, 113.75, 118.2, 120.0, 122.72, 123.18, 123.45, 123.65,
126.38, 126.53, 133.5, 137.4, 140.63, 140.87, 142.2, 144.29, 145.16, 145.98,
1
(1.08 g, 58%). H NMR (300 MHz, [D6]DMSO): d ˆ 1.44 (s, 9H), 3.82 (s,
3H), 4.02 (s, 3H), 6.84 (s, 1H), 6.99 (s, 1H), 7.42 8.16 (m, 3H), 9.21 (s, 1H),
‡
12.74, 12.79 (2s, 1H); MS (ESI): m/z: 371 [M‡H] ; HR-MS (EI): calcd for
C19H22N4O4: 370.1641; found: 370.1645.
2-(4-tert-Butoxycarbonylamino-1-methyl-1H-imidazol-2-yl)-1H-benzimi-
dazole-5-carboxylic acid (9a): Methyl ester 8a (700 mg, 1.88 mmol) was
dissolved in dioxane (140 mL). NaOH (aqueous, 1m, 140 mL) was added
and the solution stirred for 18 h at ambient temperature. The yellow
solution was carefully acidified with aqueous HCl (1m) to pH 3 4. The
formed precipitate was filtered, washed with water, and dried in vacuo to
provide the title compound 9a as a white powder (512 mg, 76%). 1H NMR
(300 MHz, [D6]DMSO): d ˆ 1.47 (s, 9H), 4.13 (s, 3H), 7.29 (s, 1H), 7.43
8.30 (m, 3H), 9.49 (s, 1H), 12.71 (s, 1H), 12.95 (s, 1H); 13C NMR (75 MHz,
[D6]DMSO): d ˆ 28.2, 35.2, 78.9, 111.5, 112.5, 113.5, 118.3, 120.6, 123.3,
‡
166.4; MS (ESI): m/z: 331 [M‡H] ; HR-MS (EI): calcd for C15H14N4O5:
330.0964; found: 330.0965.
2-(1-Methyl-4-nitro-1H-pyrrol-2-yl)-1H-benzimidazole-5-carboxylic meth-
yl ester (7c): Methyl 3,4-diamino-benzoate (5) (2.66 g, 16.0 mmol) was
dissolved in DMF (75 mL). A solution of 1-methyl-4-nitro-1H-pyrrole-2-
carbaldehyde (4c, 2.31 g, 15.0 mmol) in DMF (200 mL) was added and the
reaction mixture was heated to 908C for 60 min. Iron-trichloride hexahy-
drate (120 mg) was added and the mixture was heated to 1208C for 6 h
while air was bubbled through the solution. The reaction mixture was
cooled to room temperature. The volume of the solvent was reduced in
vacuo to ca. 50 mL and the brown solution was cooled to À208C for 4 h.
The formed precipitate was filtered, washed with CH2Cl2 until the washing
solution was colorless, and dried in vacuo to yield the title compound 7c as
a yellow powder (3.13 g, 69%). 1H NMR (300 MHz, [D6]DMSO): d ˆ 3.85
(s, 1.5H), 3.86 (s, 1.5H), 4.16 (s, 3H), 7.50 8.31 (m, 5H), 13.15 (s, 0.5H),
13.18 (s, 0.5H). At room temperature the presence of two benzimidazole
tautomers (1H/3H) could be deduced from the 1H NMR spectra. The ratio
between the two isomers was roughly estimated as 1:1 based on the
integration of the protons at d ˆ 3.85/3.86 and 13.15/13.18. 13C NMR
(75 MHz, [D6]DMSO): d ˆ 38.0, 52.0, 106.9, 111.0, 112.5, 118.7, 120.3, 122.9,
‡
124.7, 132.4, 137.9, 145.8, 152.8, 167.6; MS (ESI): m/z: 358 [M‡H] ; HR-MS
‡
(FAB): calcd for C17H20N5O4: 358.1515; found: 358.1504 [M‡H] .
2-(4-tert-Butoxycarbonylamino-3-methoxy-1-methyl-1H-pyrrol-2-yl)-3H-
benzimidazole-5-carboxylic acid (9b): Methyl ester 8b (400 mg, 1.00 mmol)
was dissolved in dioxane (10 mL). NaOH (aqueous, 1m, 10 mL) was added
and the solution stirred for 12 h at ambient temperature. The yellow
solution was carefully acidified with aqueous HCl (1m) to pH 3 4, and
extracted several times with diethyl ether. Evaporation of the solvent
afforded the title compound 9b as a yellow powder (351 mg, 91%).
1H NMR (300 MHz, [D6]DMSO): d ˆ 1.43 (s, 9H), 3.81 (s, 3H), 3.98 (s,
3H), 6.96 (s, 1H), 7.61 (d, J ˆ 8.4 Hz, 1H), 7.81 (d, J ˆ 8.4 Hz, 1H), 8.20 (s,
1H), 8.53 (s, 1H), 11.96 (s, 1H), 12.60 (s, 1H); 13C NMR (75 MHz,
[D6]DMSO): d ˆ 28.3, 37.3, 61.1, 78.6, 109.6, 111.2, 112.9, 113.3, 117.3, 119.8,
‡
123.5, 123.9, 128.0, 134.8, 146.6, 166.4; MS (ESI): m/z: 301 [M‡H] ; HR-
‡
MS (EI): calcd for C14H12N4O4: 300.0858; found: 300.0847.
120.3, 123.0, 123.8, 141.4, 147.1, 154.1, 168.0; MS (ESI): m/z: 387 [MH] ;
‡
HR-MS (FAB): calcd for C19H23N4O5: 387.1668; found: 387.1676 [M‡H] .
2-(4-tert-Butoxycarbonylamino-1-methyl-1H-imidazol-2-yl)-1H-benzimi-
dazole-5-carboxylic methyl ester (8a): Nitro-ester 7a (2.00 g, 6.64 mmol)
was dissolved in DMF (200 mL). Pd/C (10 wt.%, 360 mg) was added and
the mixture was hydrogenated at 600 psi for 2 h at ambient temperature.
The reaction mixture was filtered through Celite to remove the catalyst and
the filtrate was treated with Boc2O (2.17 g, 9.96 mmol) and DIEA (20 mL)
and stirred at 708C for 2 d. After evaporation of the solvent the residue was
dissolved in diethyl ether (150 mL) and the solution was washed with water
and brine. The organic phase was dried with MgSO4, the solvent was
evaporated, and the crude product purified by column chromatography on
silica gel (n-hexane/EtOAc 1:1) to give the title compound 8a as a white
solid (0.74 g, 30%). 1H NMR (300 MHz, [D6]DMSO): d ˆ 1.45 (s, 9H), 3.84
(s, 3H), 4.11 (s, 3H), 7.28 (s, 1H), 7.54 8.22 (m, 4H), 9.48 (s, 1H), 12.98,
2-(4-tert-Butoxycarbonylamino-1-methyl-1H-pyrrol-2-yl)-1H-benzimida-
zole-5-carboxylic acid (9c): Methyl ester 8c (1.04 g, 2.80 mmol) was
dissolved in dioxane (280 mL). NaOH (aqueous, 1m, 280 mL) was added
and the solution heated to 808C for 2 h. The yellow solution was carefully
acidified with aqueous HCl (1m) to pH 3 4. The formed precipitate was
filtered, washed with water, and dried in vacuo to provide the title
compound 9c as a white powder (990 mg, 99%). 1H NMR (300 MHz,
[D6]DMSO): d ˆ 1.46 (s, 9H), 4.04 (s, 3H), 7.04 (s, 1H), 7.24 (s, 1H), 7.73 (d,
J ˆ 8.4 Hz, 1H), 7.96 (dd, J1 ˆ 8.4 Hz, J2 ˆ 1.5 Hz, 1H), 8.20 (d, J ˆ 1.5 Hz,
1H), 9.40 (s, 1H); 13C NMR (75 MHz, [D6]DMSO): d ˆ 28.2, 36.4, 78.7,
107.2, 113.5, 115.0, 120.2, 124.8, 125.6, 126.7, 132.9, 136.2, 144.2, 152.6, 166.7;
‡
MS (ESI): m/z: 357 [M‡H] ; HR-MS (FAB): calcd for C18H21N4O4:
‡
‡
13.01 (2s, 1H); MS (ESI): m/z 372 [M‡H] ; HR-MS (EI): calcd for
357.1562; found: 357.1576 [M‡H] .
C18H21N5O4: 371.1593; found: 371.1584.
2-{4-[(4-tert-Butoxycarbonylamino-1-methyl-1H-pyrrole-2-carbonyl)-ami-
no]-1-methyl-1H-imidazol-2-yl}-1H-benzimidazole-5-carboxylic methyl es-
ter (10): Pd/C (10 wt.%, 150 mg) was added to a solution of nitro-ester 7a
(1.00 g, 3.32 mmol) in DMF (100 mL) and the mixture was hydrogenated at
450 psi for 2 h at ambient temperature. The reaction mixture was filtered
through Celite to remove the catalyst and the filtrate was treated
immediately with Boc-Py-OBt (13a, 1.42 g, 3.98 mmol) and DIEA
(20 mL) and stirred at 608C for 18 h. The resulting yellow solution was
2-(4-tert-Butoxycarbonylamino-3-methoxy-1-methyl-1H-pyrrol-2-yl)-3H-
benzimidazole-5-carboxylic methyl ester (8b): Nitro-ester 7b (1.00 g,
3.03 mmol) was dissolved in 9:1 DMF/acetonitrile (60 mL). Tin-dichloride
dihydrate (5.00 g, 22.2 mmol) was added to the solution and the reaction
mixture was heated to 508C for 14 h. After cooling the yellow solution to
room temperature Boc2O (3.96 g, 18.1 mmol), DMF (5 mL) and DIEA
(7 mL) were added and the reaction mixture heated to 508C for 4 h. The
2120
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2003, 9, 2110 2122

Alternative Heterocycles for DNA Recognition
2110 2122
poured into ice water and extracted with EtOAc. The organic phase was
washed with 10% citric acid, brine, and saturated aqueous sodium
bicarbonate and dried with MgSO4. Evaporation of the solvent yielded
the crude product as a yellow foam which was purified by column
chromatography on silica gel (n-hexane/EtOAc 1:2) to give the title
compound 10 as a beige solid (409 mg, 25%). 1H NMR (300 MHz,
[D6]DMSO): d ˆ 1.45 (s, 9H), 3.82 (s, 3H), 3.87 (s, 3H), 4.16 (s, 3H), 6.88 (s,
1H), 6.99 (s, 1H), 7.60 8.20 (m, 4H), 9.11 (s, 1H), 10.23 (s, 1H), 13.03 (brs,
Plasmid pPWF2: 5'-GATCATGAACATCGATCTCTATGTACATGC-
TATGCGATGGACATATCAGCTATGCACATTCGATCTCTATGC-
TATGCGA-3'.
Plasmid pCAB1: 5'-GATCAGTGACAATCGATCTCTGTGTCAATGA-
TATGCGGTGGCAATATGAGCTGTGCCAATTCGATCTCTATGA-
TATACGA-3'.
Plasmid pCAB2: 5'-GATCAGTGGAAATCGATCTCTGTGGTAAT-
GATATGCGGTGGGAATATGAGCTGTGGCAATTCGATCTCTAT-
GATATACGA-3'.
‡
1H); MS (ESI): m/z: 494 [M‡H] ; HR-MS (FAB): calcd for C24H28N7O5:
‡
494.2151; found: 494.2143 [M‡H] .
The inserts were obtained by annealing complementary synthetic oligonu-
cleotides and were then ligated to the large BamHI/HinDIII restriction
fragment of pUC19 using T4 DNA ligase. The ligated plasmids were then
used to transform E. coli XL-1 Blue Supercompetent cells. Colonies were
selected for a-complementation on 25 mL Luria-Bertani agar plates
containing 50 mgmLÀ1 ampicillin and treated with XGAL and IPTG
solutions and grown overnight at 378C. Well defined white colonies were
transferred into 100 mL Luria-Bertani medium containing 50 mgmLÀ1
ampicillin. Cells were harvested after overnight growth at 378C. Large
scale plasmid purification was performed using Qiagen purification kits.
The presence of the desired insert was determined by dideoxy sequencing.
DNA concentration was determined at 260 nm using the relation 1 OD
unit ˆ 50 mg mLÀ1 duplex DNA.
2-{4-[(4-tert-Butoxycarbonylamino-1-methyl-1H-pyrrole-2-carbonyl)-ami-
no]-1-methyl-1H-imidazol-2-yl}-1H-benzimidazole-5-carboxylic acid (11):
Methyl ester 10 (220 mg, 0.45 mmol) was dissolved in dioxane (20 mL).
NaOH (aqueous, 1m, 20 mL) was added and the solution stirred for 18 h at
ambient temperature. The yellow solution was carefully acidified with
aqueous HCl (1m) to pH 3 4. The formed precipitate was filtered, washed
with water, and dried in vacuo to provide the title compound 11 as a beige
powder (203 mg, 94%). 1H NMR (300 MHz, [D6]DMSO): d ˆ 1.44 (s, 9H),
3.82 (s, 3H), 4.16 (s, 3H), 6.88 (s, 1H), 6.99 (s, 1H), 7.59 8.18 (m, 4H), 9.12
(s, 1H), 10.23 (s, 1H), 12.95 (brs, 1H); 13C NMR (75 MHz, [D6]DMSO):
d ˆ 28.3, 35.2, 36.2, 78.3, 104.9, 114.01, 114.32, 117.8, 121.99, 122.32, 123.4,
124.71, 124.81, 132.5, 137.4, 143.7, 145.76, 145.84, 152.7, 158.6, 167.6; MS
‡
(ESI): m/z: 480 [M‡H] ; HR-MS (ESI): calcd for C23H26N7O5: 480.1995;
‡
Preparation of 3'-end-labeled DNA restriction fragments: Plasmids
pPWF2, pCAB1, pAU2,[21] and pCAB2 were linearized with EcoRI and
PvuII and then radiolabelled by 3'-fill in using [a-32P]-dATP, [a-32P]-TTP,
and the Klenow fragment of DNA polymerase II at 378C for 25 min. The
found: 480.2014 [M‡H] .
Solid-phase synthesis: Polyamides 1a, b, 2a, b, 3a, b were generated by
manual solid-phase synthesis on oxime resin according to recently reported
protocols[14] using Boc-protected amino acid building blocks. The synthesis
product mixture was purified on
a 7% non-denaturing preparatory
of polyamide ImImPyPy-(R)H N -ImPyPyPyCONHMe (1a) has been
previously described.[29] For the coupling of benzimidazole building blocks
Boc protected amino acids 9a c and 11 were activated with HBTU
(1 equiv in DIEA, NMP) for 15 min and subsequently coupled for 4 h at
ambient temperature. Resultant resin-bound benzimidazoles were depro-
tected using 20% TFA in CH2Cl2 for 25 min (for resin-bound 9c and 11) or
50% TFA in CH2Cl2 for 25 min (for resin-bound 9a,b).
2
g
polyacrylamide gel (5% cross-linkage) and the desired fragment was
isolated after visualization by autoradiography. The DNA was precipitated
with isopropanol (1.5 volumes). The pellet was washed with 75% ethanol,
lyophilized to dryness, and then resuspended in RNase-free H2O. Chemical
sequencing reactions were performed according to published protocols.[30]
Quantitative DNase I footprint titrations: Concentrations of benzimida-
zole-containing polyamide and parent polyamide stock solutions were
determined by UV absorption at 331 nm (e ˆ 41500 cmmÀ1) and 310 nm
(e ˆ 69500 cmmÀ1), respectively. All reactions were carried out in a volume
of 400 mL according to published procedures.[20] Quantitation by storage
phosphor autoradiography and determination of equilibrium association
constants were as previously described.[20]
General procedure for MeNH2 cleavage:[14b] A sample of the derivatized
resin (100 mg) was suspended in CH2Cl2 (2 mL) to which was added
methylamine in THF (2 mL, 2.0m). The mixture was agitated at 378C for
14 h. The resin was filtered, rinsed with CH2Cl2 and DMF, the eluant
concentrated in vacuo, and the residue purified by prep. reversed-phase
HPLC to yield the desired polyamides.
ImImPyPy-(R)H Ng-ImPyBiPyCONHMe (1b): The title compound 1b was
2
recovered upon lyophilization of the appropriate fractions as a yellow
powder. MALDI-TOF-MS: calcd for C51H55N21O8: 1090.1; found: 1090.6.
Acknowledgement
ImImPyPy-(R)H Ng PyImPyPyCONHMe (2a): The title compound 2a
2
was recovered upon lyophilization of the appropriate fractions as a white
powder. MALDI-TOF-MS: calcd for C49H56N22O9: 1097.1; found: 1097.6.
We are grateful to the National Institute of Health (GM-27681) for
research support, to the Alexander von Humboldt-Foundation for a
Feodor-Lynen grant to C.A.B. and to the Swiss National Science
Foundation for a postdoctoral fellowship to P.W. We also wish to thank
Michael A. Marques for conducting the ab initio calculations.
ImImPyPy-(R)H Ng PyImBiPyCONHMe (2b): The title compound 2b
2
was recovered upon lyophilization of the appropriate fractions as a yellow
powder. MALDI-TOF-MS: calcd for C50H54N22O8: 1091.8; found: 1091.1.
ImImPyPy-(R)H Ng ImOpPyPyCONHMe (12a): The title compound 12a
2
was recovered upon lyophilization of the appropriate fractions as a beige
powder. MALDI-TOF-MS: calcd for C51H59N21O10: 1126.2; found: 1126.7.
[1] a) C. L. Kielkopf, S. White, J. W. Szewczyk, J. M. Turner, E. E. Baird,
P. B. Dervan, D. C. Rees, Science 1998, 282, 111 115; b) P. B. Dervan,
R. W. B¸rli, Curr. Opin. Chem. Biol. 1999, 3, 688 693; c) D. E.
Wemmer, Biopolymers 1999, 52, 197 211; d) P. B. Dervan, Bioorg.
Med. Chem. 2001, 9, 2215 2235.
[2] a) J. M. Gottesfeld, L. Neely, J. W. Trauger, E. E. Baird, P. B. Dervan,
Nature 1997, 387, 202 205; b) L. A. Dickinson, R. J. Gulizia, J. W.
Trauger, E. E. Baird, D. E. Mosier, J. M. Gottesfeld, P. B. Dervan,
Proc. Natl. Acad. Sci. USA 1998, 95, 12890 12895; c) S. Janssen, T.
Durussel, U. K. Laemmli, Mol. Cell 2000, 6, 999 1011; d) A. Z.
Ansari, A. K. Mapp, D. H. Nguyen, P. B. Dervan, M. Ptashne, Chem.
Biol. 2001, 8, 583 592; e) J. J. Coull, G. He, C. Melander, V. C.
Rucker, P. B. Dervan, D. M. Margolis, J. Virology 2002, 76, 12349
12354.
ImImPyPy-(R)H Ng ImOpBiPyCONHMe (12b): The title compound 12b
2
was recovered upon lyophilization of the appropriate fractions as a yellow
powder. MALDI-TOF-MS: calcd for C52H57N21O9: 1120.2; found: 1120.7.
O-Demethylation of polyamides 12a,b: According to the previously
reported protocol[19] polyamides were treated with sodium thiophenoxide
in DMF at 858C for 2 h and subsequently purified by prep. reversed-phase
HPLC to yield the desired demethylated polyamides.
ImImPyPy-(R)H Ng ImHpPyPyCONHMe (3a): The title compound 3a
2
was recovered upon lyophilization of the appropriate fractions as a beige
powder. MALDI-TOF-MS: calcd for C50H57N21O10: 1112.1; found: 1113.1.
ImImPyPy-(R)H Ng ImHpBiPyCONHMe (3b): The title compound 3b
2
was recovered upon lyophilization of the appropriate fractions as a yellow
powder. MALDI-TOF-MS: calcd for C51H55N21O9: 1106.1; found: 1106.5.
[3] J. W. Trauger, E. E. Baird, P. B. Dervan, Nature 1996, 382, 559 561.
[4] D. M. Herman, J. M. Turner, E. E. Baird, P. B. Dervan, J. Am. Chem.
Soc. 1999, 121, 1121 1129.
[5] W. A. Greenberg, E. E. Baird, P. B. Dervan, Chem. Eur. J. 1998, 4,
796 805.
Construction of plasmid DNA: The restriction fragment pAU2 was
generated as previously described.[21] Similarly, plasmids pPWF2, pCAB1
and pCAB2 were constructed by ligation of the following hybridized inserts
into the BamHI/HinDIII polycloning site of pUC19.
Chem. Eur. J. 2003, 9, 2110 2122
www.chemeurj.org
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2121

P. B. Dervan et al.
FULL PAPER
[6] a) P. Weyermann, P. B. Dervan, J. Am. Chem. Soc. 2002, 124, 6872
6878; b) I. Kers, P. B. Dervan, Bioorg. Med. Chem. 2002, 10, 3339
3349.
[15] M. P. Singh, S. Sasmal, W. Lu, M. N. Chatterjee, Synthesis 2000, 1380
1390.
[16] J. B. Wright, Chem. Rev. 1951, 48, 397 541.
[7] a) Z. Y. J. Zhan, P. B. Dervan, Bioorg. Med. Chem. 2000, 8, 2467
2474; b) D. H. Nguyen, J. W. Szewczyk, E. E. Baird, P. B. Dervan,
Bioorg. Med. Chem. 2001, 9, 7 17; c) C. C. O×Hare, D. Mack, M.
Tandon, S. K. Sharma, J. W. Lown, M. L. Kopka, R. E. Dickerson,
J. A. Hartley, Proc. Natl. Acad. Sci. USA 2002, 99, 72 77; d) M. A.
Marques, R. M. Doss, A. R. Urbach, P. B. Dervan, Helv. Chim. Acta
2002, 85, 4485 4517.
[8] a) W. S. Wade, M. Mrksich, P. B. Dervan, J. Am. Chem. Soc. 1992, 114,
8783 8794; b) M. Mrksich, P. B. Dervan, J. Am. Chem. Soc. 1993, 115,
9892 9899.
[17] Y. Yamamoto, T. Kimachi, Y. Kanaoka, S. Kato, K. Bessho, T.
Matsumoto, T. Kusakabe, Y. Sugiura, Tetrahedron Lett. 1996, 37,
7801 7804.
[18] Moderate to low reaction conversions are often observed for
couplings of Boc-Py-OBt (13a) to resin-bound imidazole amines.[14a]
To increase the coupling efficiency the use of PyIm dimers proved to
be advantageous. Alternatively, Py to Im couplings can be performed
by using Boc-Py-OH (13b) in the presence of DCC and DMAP.[14a]
[19] C. Melander, D. M. Herman, P. B. Dervan, Chem. Eur. J. 2000, 6,
4487 4497.
[9] U. Ellervik, C. C. C. Wang, P. B. Dervan, J. Am. Chem. Soc. 2000, 122,
9354 9360.
[20] J. W. Trauger, P. B. Dervan, Methods Enzymol. 2001, 340, 450 466.
[21] A. R. Urbach, J. W. Szewczyk, S. White, J. M. Turner, E. E. Baird, P. B.
Dervan, J. Am. Chem. Soc. 1999, 121, 11621 11629.
[22] R. Kakkar, R. Garg, Suruchi, J. Mol. Struct. 2002, 584, 37 44; and
references therein.
[23] D. E. Wemmer, Annu. Rev. Biophys. Biomol. Struct. 2000, 29, 439
461.
[24] Gas-phase ab initio HF/6-31G* energy minimization; Spartan Essen-
tial Software, Wavefunction Inc., 2001.
[25] D. Renneberg, P. B. Dervan, unpublished observations.
[26] J. M. Belitsky, S. J. Leslie, P. S. Arora, T. A. Beerman, P. B. Dervan,
Bioorg. Med. Chem. 2002, 10, 3313 3318.
[27] J. Sambrook, E. F. Fritsch, T. Maniatis, Molecular Cloning, Cold
Spring Harbor Laboratory, Cold Spring Harbor, NY, 1989.
[28] D. P. Phillion, M. Singh (Monsanto Technology LLC), WO 02/04417
A1, 2002 [Chem. Abstr. 2002, 136, 102283].
[29] N. R. Wurtz, J. L. Pomerantz, D. Baltimore, P. B. Dervan, Biochem-
istry 2002, 41, 7604 7609.
[30] a) B. L. Iverson, P. B. Dervan, Nucleic Acids Res. 1987, 15, 7823 7830;
b) A. M. Maxam, W. S. Gilbert, Methods Enzymol. 1980, 65, 499 560.
[10] a) R. L. Lombardy, F. A. Tanious, K. Ramachandran, R. R. Tidwell,
W. D. Wilson, J. Med. Chem. 1996, 39, 1452 1462; b) T. G. Minehan,
K. Gottwald, P. B. Dervan, Helv. Chim. Acta 2000, 83, 2197 2213;
c) L. Wang, C. Carrasco, A. Kumar, C. E. Stephens, C. Bailly, D. W.
Boykin, W. D. Wilson, Biochemistry 2001, 40, 2511 2521; d) W.
Zhang, Y. Dai, U. Schmitz, T. W. Bruice, FEBS Lett. 2001, 509, 85 89;
e) Y.-H. Ji, D. Bur, W. Haesler, V. R. Schmitt, A. Dorn, C. Bailly, M. J.
Waring, R. Hochstrasser, W. Leupin, Bioorg. Med. Chem. 2001, 9,
2905 2919; f) A. L. Satz, T. C. Bruice, J. Am. Chem. Soc. 2001, 123,
2469 2477; g) C. Behrens, N. Harrit, P. E. Nielsen, Bioconjugate
Chem. 2001, 12, 1021 1027.
[11] Y. Matsuba, H. Edatsugi, I. Mita, A. Matsunaga, O. Nakanishi, Cancer
Chemother. Pharmacol. 2000, 46, 1 9.
[12] T. Stokke, H. B. Steen, J. Histochem. Cytochem. 1985, 33, 333 338.
[13] a) P. E. Pjura, K. Grzeskowiak, R. E. Dickerson, J. Mol. Biol. 1987,
197, 257 271; b) M. Teng, N. Usman, C. A. Frederick, A. Wang,
Nucleic Acids Res. 1988, 16, 2671 2690; c) S. Kumar, B. Yadagiri, J.
Zimmermann, R. T. Pon, J. W. Lown, J. Biomol. Struct. Dyn. 1990, 8,
331 357.
[14] a) E. E. Baird, P. B. Dervan, J. Am. Chem. Soc. 1996, 118, 6141 6146;
b) J. M. Belitsky, D. H. Nguyen, N. R. Wurtz, P. B. Dervan, Bioorg.
Med. Chem. 2002, 10, 2767 2774.
Received: December 19, 2002 [F4689]
2122
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2003, 9, 2110 2122