Molecular recognition of DNA by Hoechst benzimidazoles: Exploring beyond the pyrrole-imidazole-hydroxypyrrole polyamide-pairing code

Article abstract of DOI:10.1002/1522-2675(20000906)83:9<2197::AID-HLCA2197>3.0.CO;2-N

A series of three-ring analogs of the minor-groove-binding molecule Hoechst 33258 (1), consisting of benzimidazole (B), imidazopyridine (P), and hydroxybenzimidazole (H) monomers, have been synthesized in order to investigate both their sequence specifici

Full text of DOI:10.1002/1522-2675(20000906)83:9<2197::AID-HLCA2197>3.0.CO;2-N

Helvetica Chimica Acta ± Vol. 83 (2000)
2197
Molecular Recognition of DNA by Hoechst Benzimidazoles: Exploring
Beyond the Pyrrole-Imidazole-Hydroxypyrrole Polyamide-Pairing Code
by Thomas G. Minehan, Konstanze Gottwald, and Peter B. Dervan*
Division of Chemistry and Chemical Engineering, California Institute of Technology, Pasadena, California
91125, USA
Dedicated to Professor Albert Eschenmoser on the occasion of his 75th birthday
A series of three-ring analogs of the minor-groove-binding molecule Hoechst 33258 (1), consisting of
benzimidazole (B), imidazopyridine (P), and hydroxybenzimidazole (H) monomers, have been synthesized in
order to investigate both their sequence specificity and binding modes. MPE ´ FeII Footprinting has revealed the
preference of both PBB and BBB ligands for 5'-WGWWW-3' and 5'-WCWWW-3' tracts, as well as A ´ T-rich
sequences. Affinity-cleavage titrations show no evidence for a 2 :1 binding mode of these Hoechst analogs.
Importantly, all derivatives are oriented in one direction at each of their binding sites. The implications of these
results for the design of minor-groove-binding small molecules is discussed.
1. Introduction. ± Small molecules that bind DNA sequence specifically are
important tools in molecular biology and may form the basis for novel classes of gene-
targeted drugs directed toward cancer and infectious diseases [1][2]. The natural
products netropsin (N) and distamycin A (D) are two- and three-ring N-methyl-
pyrrolecarboxamides that bind in the minor groove of double-helical DNA at sites rich
in A ´ T base pairs [3] (Fig. 1). X-Ray diffraction and NMR studies reveal that
specificity is accomplished by these molecules as their 1:1 and 2 :1 complexes with
DNA depending on A ´ T sequence composition [4 ± 10]. The synthetic dye Hoechst
33258 (1) also binds double-helical DNA in the minor groove at sites rich in A ´ T base
pairs [11]. The size and crescent shape of the Hoechst molecule 1 is similar to that of
netropsin (N) and distamycin (D), and, indeed, these molecules all share common
binding sites consisting of four to five consecutive A ´ T base-pairs as revealed by MPE ´
FeII footprinting [12]. In addition, X-ray and NMR studies of 1 bound to DNA as a 1:1
complex [13 ± 15] indicate how the molecule accomplishes recognition of DNA: similar
to netropsin (N) and distamycin (D), specific H-bonds are formed between the
benzimidazole NHs and adjacent adenine N(3)- and thymine O(2)-atoms on the floor
of the minor groove. The binding preference for A ´ T base pairs is a result of the close
contact of the aromatic H-atoms of the Hoechst molecule 1 and the floor of the minor
groove, thus precluding the binding of the drug at G ´ C base pairs due to the presence
of the exocyclic N(2)-amino group of guanine. In addition, the binding of 1 is stabilized
by electrostatic interactions and extensive van der Waals contacts with the walls of the
minor groove. Unlike distamycin (D), only 1:1 complexes of 1 and DNA have been
described.
Efforts to alter the sequence specificity of the natural product distamycin (D) to
allow recognition of G ´ C base pairs came to fruition with the discovery that the simple

2198
Helvetica Chimica Acta ± Vol. 83 (2000)
Fig. 1. Structures and binding modes of the minor-groove-binding molecules netropsin (N), distamycin (D), and
Hoechst 33258 (1)
three-ring analog imidazole-pyrrole-pyrrole (ImPyPy; 2) binds to a 5'-TGTCA-3' site
as an antiparallel side-by-side dimer in the minor groove [16][17]. The 2 :1 model
allowed specific contacts of the ligands with each strand on the floor of the minor
groove, and it is the model from which the pairing rules for polyamide recognition of
DNA arose (Fig. 2). Polyamides consisting of aromatic rings which differ only at the 3
position ± imidazole (Im), pyrrole (Py), and hydroxypyrrole (Hp) ± are now versatile
ligands that can distinguish all four base pairs by a binary recognition code based on
combinations of four building blocks: a Py/Im pair targets C ´ G, an Im/Py pair targets
G ´ C, a Py/Hp pair targets A ´ T, and an Hp/Py pair targets T´ A base pairs [18].
Analogs of Hoechst 33258 (1) have been described in the literature [19], some of which
are capable of recognizing G ´ C-containing sequences. However, although there has
been evidence of cooperativity in the binding of 1 to DNA [20][21], there has been
little investigation of the binding orientation of the derivatives of 1. The goals of the
present study were to rationally develop analogs of 1 that might allow recognition of
DNA sequences of mixed A ´ T and G ´ C composition.
Our design of tris-benzimidazole derivatives of 1 was influenced by the current
paradigm for polyamide recognition of DNA and, in particular, Lownꢀs contributions to

Helvetica Chimica Acta ± Vol. 83 (2000)
2199
Fig. 2. Structures and 2 :1 binding modes of synthetic molecules which recognize DNA in the minor groove.
ImPyPy and ImHpPyPy bind 5'-TGACT-3' and 5'-TGTACA-3', respectively, as antiparallel homodimers. The
pairing rules for polyamide recognition of DNA arose from study of the sequence specificity of these and similar
molecules.
altered bis(benzimidazole) minor-groove binders [22] (Fig. 3). The benzimidazole (B),
imidazopyridine (P), and hydroxybenzimidazole (H) monomers, which differ only at
the 7-position, were envisioned to present an array of functionalities and H-bonding
surfaces to the floor of the minor groove similar to pyrrole, imidazole, and
hydroxypyrrole. It was anticipated that the imidazopyridine unit, containing a N-atom
instead of the 7-CH of the benzimidazole nucleus, would bind opposite the G of a G ´ C
base pair, and that the hydroxybenzimidazole unit, containing a C(7)ÀOH instead of
the 7-CH, would bind opposite the Tof a T´ A base pair. An unmodified benzimidazole
moiety would be degenerate for A ´ T and T´ A base pairs. The trimers 3-Dp (Dp: 3-
(dimethylamino)propanamine; R ˆ 4-MeOC6H; R' ˆ Me) and 4-Dp (R ˆ 4-HOC6H4;
R' ˆ Me), as Hoechst 33258 (1) analogs of distamycin (D) and ImPyPy (2), were thus
designed to investigate the sequence selectivity of these ligands relative to the parent
molecules (Fig. 4). Resolution of the binding-site location and size can be accom-
plished by the method of MPE ´ FeII footprinting [23]. The affinity-cleaving analogs
of 3-ED and 4-ED allow the binding mode and orientation of the ligands at their
binding sites to be determined by affinity-cleavage titrations. An HBB trimer 5 was
also designed to investigate the possibility of A ´ T/T´ A discrimination by the
derivatives of 1.
In addition, although there have been numerous studies characterizing the exact
binding sites of the parent compound 1 [11 ± 15], there have been no reports on the
orientation preferences of 1 at its binding sites. This information can be gained from

2200
Helvetica Chimica Acta ± Vol. 83 (2000)
Fig. 3. Rationale for the design of benzimidazole-based molecules for DNA recognition. Similar H-bonding
surfaces are presented to the floor of the minor groove for both the pyrrole-imidazole-hydroxypyrrole
monomers and the benzimidazole-imidazopyridine-hydroxybenzimidazole monomers.
affinity cleavage of DNA with a simple derivative of 1 modified at its N-terminus with
Hoechst EDTA (1-ED). We report here the synthesis of 1-ED and characterization of
its binding mode by affinity cleavage of pBr322 DNA.
2. Synthesis of Hoechst-EDTA (1-ED). ± Tethering of the EDTA (ethylendiamine-
N,N,N',N'-tetraacetic acid) moiety from the Hoechst core was most easily accomplished
by modifying the terminal piperazine unit (Scheme 1). Thus, 5-chloro-2-nitroaniline (6)
was treated with piperazine in DMF in the presence of K2CO3 at elevated temperature
to afford the corresponding piperazine-substituted nitroaniline. Reaction of this
compound with ethyl bromoacetate and EtN(i-Pr)2 in DMF overnight, followed
directly by hydrogenation over Pd-C gave ethyl 4-(3,4-diaminophenyl)piperazine-1-
acetate (7). Compound 7 was then coupled with carbaldehyde 9 under Lown conditions
(vide infra) to afford ethyl ester 8. Transformation to the EDTA derivative 1-ED took
place in three steps: hydrolysis of the ethyl ester and methyl ether of 8 with refluxing
48% HBr, amide-bond formation by N,N'-dicyclohexylcarbodiimide (DCC)/1-hydrox-
ybenzotriazole (HOBt)-mediated coupling of the acid and N-methylpropane-1,3-
diamine, and reaction of the terminal amine with EDTA-dianhydride as described
below for the preparation of 3-ED and 4-ED. This sequence afforded 1-ED in 5%
overall yield, and the compound was characterized by UV and 1H-NMR spectroscopy,
and high-resolution mass spectrometry.

Helvetica Chimica Acta ± Vol. 83 (2000)
2201
Fig. 4. The BBB and PBB trimers, 3 and 4, along with their affinity-cleaving analogs designed to investigate the
binding mode and sequence selectivity of Hoechst derivatives. The HBB trimer 5 was synthesized to investigate
the possibility of A ´ T discrimination by Hoechst compounds.
3. Synthesis of the BBB and PBB Trimers, 3 and 4, Respectively. ± Because of
difficulties encountered at the monomer stage in the preparation of the imidazopyr-
idine trimer with R ˆ H (Fig. 4), it was decided, for synthetic ease, to construct the PBB
and BBB trimers containing a 4-methoxyphenyl cap (R ˆ 4-MeOC6H4; Fig. 4). As
described by Lee et al., 2-(4-methoxyphenyl)-1H-benzimidazole-6-carbaldehyde (9)
was prepared from methyl 3,4-dinitrobenzoate [24]; 2-(4-methoxyphenyl)-1H-imida-
zo[5,4-b]pyridine-6-carbaldehyde (10) was prepared from 2-amino-6-methylpyridine
according to the procedure of Lown and co-workers [25] (Scheme 2). Coupling of 10
with diamino compound 11 [26] was accomplished by refluxing the mixture in PhNO2 at
1508 for 19 ± 24 h under Ar. Evaporation of the solvent, trituration with Et2O/CH2Cl2,
and filtration conveniently afforded the trimer 13 as a brown solid in 51% yield. This
method was also used to construct trimer 12 in 62% yield from 11 and 9. A variety of
different methods were attempted in order to hydrolyze the CN group of 12 and 13.
Basic conditions at elevated temperatures resulted in extensive decomposition of the

2202
Helvetica Chimica Acta ± Vol. 83 (2000)
Scheme 1. Synthesis of Hoechst-EDTA (1-ED)
Scheme 2. Synthesis of PBB and BBB Trimers and Their Affinity-Cleaving Analogs

Helvetica Chimica Acta ± Vol. 83 (2000)
2203
trimers. Treatment of the carbonitriles with 75% H2SO4 for 1 h at 1508 resulted in
hydrolysis of the carbonitrile to carboxylic acid accompanied by partial hydrolysis of
the aromatic methyl ether. Extended treatment with 75% H2SO4 resulted in
decomposition. However, it was found that hydrolysis with 48% HBr at 1308 for 4 h
resulted in clean conversion to the intermediate hydroxyphenyl carboxylic acids. The
crude acids were immediately coupled under DCC/HOBt conditions at room temper-
ature with either N,N-dimethylpropane-1,3-diamine or MeNPr2 to afford amides 3-
Dp,4-Dp, and 3-NH2,4-NH2, respectively. The EDTA derivatives for PBB and BBB
were prepared from 3-NH2 and 4-NH2 by treatment with EDTA dianhydride in DMSO/
N-methylphthalimide (NMP) solution at 558 for 5 min, followed by hydrolysis with
0.1n NaOH and 558 for 10 min and subsequent HPLC purification [27]. The identity of
1
all compounds was confirmed by UV and H-NMR spectroscopy, and high-resolution
mass spectrometry.
4. Synthesis of the HBB Trimer 5. ± In the initial approach to the HBB trimer, it was
found that the most convenient starting material, which contains the aromatic
substitution pattern required for synthesis of the hydroxybenzimidazole subunit, was
the methyl carboxylate 14 (Scheme 3). Furthermore, for synthetic ease, a 2-Me-
substituted terminal benzimidazole unit was selected as the capping moiety of our
target molecule. A convenient route to carbaldehyde 15 was developed, in which a one-
pot nitro reduction/acid-catalyzed cyclization/dehydration of 14 produced the inter-
mediate 2-methylbenzimidazole-carboxylate, which was subsequently reduced and
oxidized according to the procedure described in [28]. Coupling of 15 and 11 in refluxing
Scheme 3. Synthesis of the HBB Trimer 5

2204
Helvetica Chimica Acta ± Vol. 83 (2000)
PhNO2 at 1508 took place smoothly to afford the benzimidazole-carbonitrile 16 in 78%
yield. Nitrile and methyl-ether hydrolysis followed by direct coupling of the crude acid
under DCC/HOBt conditions with N,N-dimethylpropane-1,3-diamine afforded trimer
amide 5 in 20% yield after HPLC purification. The identity of all compounds was
confirmed by UV and 13C-NMR spectroscopy, and high-resolution mass spectrometry.
5. Analysis of Binding-Site Location and Size by MPE ´ Fe(II) Footprinting. ± The
519 base-pair EcoRI/RsaI restriction fragment of plasmid pBr322 has been used
extensively for characterization of the sequence selectivity of the parent molecules
distamycin (D), ImPyPy-Dp (2-Dp) [16], and Hoechst 33258 (1) [12], and thus was also
selected the present study. MPE ´ Fe(II) footprinting of 1, 2-Dp, 3-Dp, and 4-Dp on 3'-
32P-end-labeled 519-bp restriction fragment reveals similarities in the binding-site
locations of all the benzimidazole-based ligands (Fig. 5,a and c). Both PBB (lanes 11 ±
13) and BBB (lanes 14 ± 16) bind the 5 ± 6 base-pair sequences 5'-TAAGA-3', 5'-
TTATTA-3', 5'-TTAACC-3', and 5'-TATCA-3', sites which they also have in common
with 1. In contrast, a single footprint for 2-Dp at the site 5'-TGACA-3' is observed, and
neither 3-Dp nor 4-Dp bind at this location. Although quantitative evaluation of the
binding affinities for these ligands for their match sites by DNase footprinting was not
performed, comparison of the intensity of the protection patterns for 20 mm 2-Dp at the
site 5'-TGACA-3' with that for 10 mm 4-Dp and 10 mm 3-Dp at the site 5'-AGAAA-3'
suggests that, regardless of binding mode, 4-Dp and 3-Dp have higher affinities for their
respective binding sites than 2-Dp. Footprints for these compounds were also compared
on the 5'-32P PCR-labeled 3991 ± 4143 segment of pBr322 (Fig. 6,a and c). Again, both
PBB and BBB bind the same sites, which can be summarized as 5'-WWWWCC-3', 5'-
WWWWC(W)-3', 5'-WWWWG-3', 5'-WWWCW-3', and 5'-WWWGW-3'. Remarkably,
the sites 5'-WWWGW-3' and 5'-WWWCW-3' (from the previous experiment) are
bound equally well by 3-Dp and 4-Dp. Several conclusions can be drawn from these
data. First, although the PBB trimer differs from the BBB trimer in the terminal
monomer unit by the substitution of the N-atom for an aromatic CH, the two ligands do
not seem to have different DNA-sequence preferences. This may be due to free
rotation about the CÀC bond connecting the terminal and penultimate benzimidazole
residues, allowing surfaces from both the convex and concave faces of the ligand to be
presented to the floor of the minor groove. Second, since neither PBB nor BBB bind 5'-
TGACA-3', and since they occupy virtually identical binding sites to 1, this suggests a
likely 1:1 binding mode for these tris[benzimidazole] ligands. Affinity-cleavage
experiments verified this hypothesis.
MPE Footprinting of the HBB trimer was performed on a 286-bp fragment of the
plasmid pJT3 [29] (data not shown), which contains cassettes of five consecutive As or
Ts. If HBB can discriminate a T base pair from an A base pair, fewer footprints should
be seen than for the parent Hoechst compound 1. HBB showed very weak footprints,
even at 100 mm; comparison with 1 showed that both molecules bind to identical
sequences. Since the HBB trimer has a low affinity for DNA and apparently does not
discriminate T from A, no further experiments were performed with this compound;
other trimer derivatives containing simple substituted hydroxyphenyl rings in the
terminal position are currently being investigated, and footprinting results will be
reported in due course.

Helvetica Chimica Acta ± Vol. 83 (2000)
2205
6. Analysis of Binding Mode and Orientation by Affinity-Cleavage Titration. ± The
affinity-cleavage gel and histograms for ImPyPy-ED (2-ED) (Lanes 6 and 7, Fig. 5,b
and d) show a symmetrical pattern of cleavage on both sides of the single site 5'-
TGACA-3', indicative of a 2 :1 binding mode for 2-ED at this site. In contrast, Hoechst-
ED (1-ED, Lanes 4 and 5), BBB-ED (3-ED, Lanes 8 and 9), and PBB-ED (4-ED,
Lanes 10 and 11) show strong cleavage patterns on only one side of each binding site.
Thus, both tris[benzimidazole] compounds, as well as the parent molecule 1, appear to
bind DNAwith 1:1 stoichiometry; surprisingly, however, these data also indicate that all
three ligands are oriented in one particular direction when bound. Because of the
proximity of some binding sites, especially for Hoechst-ED (1-ED), cleavage patterns
at adjacent sites often overlap so that assignment of ligand orientation in some cases is
inferred from the absence of a strong cleavage pattern on the other side of the binding
site (Fig. 7,a). Interestingly, comparison of the loci of cleavage for 3-ED and 4-ED at
their common binding sites indicates that at some sites, the orientation of PBB and BBB
are opposite (see the upper portion of the gel in Fig. 5,b). This is even more evident in
the affinity cleavage of the 5'-32P PCR-labeled 3991 ± 4143 segment of pBr322 with 3-
ED and 4-ED (Fig. 6,b and d): PBB-ED shows a cleavage pattern to the 3' side of 5'-
ATTTCC-3', whereas BBB-ED shows a cleavage pattern to the 5' side of the same site.
There are several other examples of sequences where this is also the case; a summary of
the binding sites and orientations of 1-ED, 3-ED, and 4-ED is shown in Fig. 7. Although
there are no obvious alignment trends for each ligand, it seems that in many instances,
the terminal benzimidazole/pyridimidazole-phenol portion of the molecule prefers to
lie next to a G ´ C base-pair. Perhaps the freely rotating phenol moiety prefers the wider
minor groove associated with G ´ C base-pairs. These data indicate that steric and/or
van der Waals interactions may be more important than H-bond formation in the
molecular recognition of DNA by Hoechst 33258 (1) and its derivatives.
7. Discussion and Conclusion: The Different DNA-Binding Properties of Poly-
amides and Bis- and Tris[benzimidazoles]. ± CPK Models of distamycin (D) and BBB
(3) show a difference in the curvature of the two crescent-shaped molecules. As with
models of ImPyPy (2) and PBB (4), there is consistently a greater degree of curvature
for the polyamide than for the tris[benzimidazole] in the same series. Gas-phase
minimizations of molecular models also corroborate this observation [30]. However,
the curvature of PBB (4) and distamycin (D) are quite similar, thus making it difficult
to draw any conclusions concerning modes of binding for an entire class of ligands from
these observations. Experimentally, MPE ´ FeII footprinting data indicate the similar
binding site preferences of Hoechst compound 1 and the PBB and BBB trimers, and
also distinguish these three molecules from ImPyPy, both in affinity and sequence
specificity. The higher affinity of benzimidazoles for the DNA sequences studied may
indicate the existence of strong, favorable associations (van der Waals interactions)
between the ligand and the walls of the minor groove at the binding site. Indeed, the
benzimidazole-monomer unit has a greater aromatic surface area for such contacts than
either pyrrole or imidazole. These interactions may further suggest an explanation for
the affinity cleavage data, in which Hoechst-ED (1-ED), PBB-ED (4-ED), and BBB-
ED (3-ED) are clearly distinguished from ImPyPy-ED (2-ED) in binding mode. The
strong associations with the minor groove formed upon binding of the first molecule of

2206
Helvetica Chimica Acta ± Vol. 83 (2000)

Helvetica Chimica Acta ± Vol. 83 (2000)
2207
bis- or tris[benzimidazole] may discourage the binding of a second molecule at the
same site (in either a stacked antiparallel, parallel, or slipped mode). In addition,
tris[benzimidazole] ligands may prefer A ´ T tracts, 5'-WGWWW-3', and 5'-WCWWW-
3' sites, because the narrow minor-groove walls at these locations can hug the ligand,
maximizing the favorable van der Waals interactions. The sequence-dependent
variation in the width of the minor groove must also be considered in the discussion
of ligand orientational preferences at individual binding sites. Virtually all of the sites
bound by the Hoechst compound, PBB, and BBB have a G ´ C base pair at the end,
which serves to slightly widen the otherwise narrow A ´ T tract. As discussed above, all
three molecules adopt an orientation in which the N-terminus is positioned close to the
G ´ C base pair, likely because of the greater steric space required by a freely rotating
phenol moiety. It is difficult, however, to rationalize the observation that, at certain
sequences, PBB and BBB, which differ only by the single atomic substitution of N for
CH, adopt different orientations. Perhaps local dipole moments in the minor groove
may exert differential orienting effects on the different ligands.
These studies demonstrate that much is still unknown about the factors responsible
for sequence selectivity, orientation, binding mode and affinity of Hoechst derivatives
which bind in the minor groove of DNA.
We are grateful for financial support from the National Institutes of Health, the Beckman Institute, BASF,
and Studienstiftung des deutschen Volkes for a postdoctoral fellowship to K. G., and the American Cancer Society
for a postdoctoral fellowship to T. G. M. We also wish to thank Victor Rucker for performing gas-phase
minimizations on distamycin, ImPyPy, PBB, and BBB.
Experimental Part
1. General. ± Abbreviations: DCC: 1,3-Dicyclohexylcarbodiimide, DIEA: EtN(i-Pr)2, Dp: 3-(Dimeth-
ylamino)propylamine, HOBt: 1-Hydroxy-1H-benzotriazole, FC: flash chromatography. Reagents and Equip-
ment: Solvents and reagents were purchased from Aldrich of the highest quality available and were used without
further purification. TLC: Merck silica gel 60-F254 anal. plates. FC: Baker silica gel (40 mm). Anal. reversed-
phase (RP) HPLC: Hewlett-Packard 1090 Liquid Chromatograph or Beckman System Gold (detector 168,
solvent module 125), flow 1 ml/min, UV detector at l 280 ± 340 nm, Varian Microsorb-MV 100 Š C18 column
4.6 Â 250 mm, linear gradient: 0.1% CF3COOH/H2O (A), MeCN (B), 0 min 0% B, 30 min 60% B, 35 min 100%
B; tR values refer to anal. HPLC. Prep. RP-HPLC: Beckman System Gold (detector 166P, solvent module 127P),
flow 8 ml/min, UV detector at l 280 ± 340 nm, Waters PrepLC 25-mm module C18 column, linear gradient: 0.1%
CF3COOH/H2O (A), MeCN (B), 0 min 0% B, 5 min 0% B, 45 min 20% B, 85 min 30% B, 95 min 35% B, 97 min
50% B, 100 min 95% B. UV: Hewlett-Packard 8452 diode-array spectrophotometer, lmax in nm, e in mÀ1 cmÀ1
.
b
Fig. 5. a) MPE-FeII Footprinting of Hoechst 33258 (1), ImPyPy-Dp (2-Dp), BBB-Dp (3-Dp) and PBB-Dp (4-
Dp) trimers on the 3'-end-labelled 519-bp EcoRI/RsaI restriction fragment of PBr322: Lanes 1 and 20: intact 3'-
labeled DNA; Lanes 2 ± 3 and 17 ± 18: A and G reactions; Lanes 4 and 19: MPE-FeII digestion of DNA in the
absence of ligand; Lanes 5 ± 7, 1, 20: 40 mm Hoechst; Lanes 8 ± 10, 1, 10: 20 mm ImPyPy-Dp; Lanes 11 ± 13: 0.5, 10,
40 mm BBB-Dp; Lanes 14 ± 16: 0.5, 10, 40 mm PBB-Dp. b) Affinity cleavage of the 519-bp EcoRI/RsaI restriction
fragment of pBr-322 with Hoechst-EDTA (1-ED), ImPyPy-ED (2-ED), BBB-ED (3-ED), and PBB-ED (4-
ED). Lanes 1 and 12: intact 3'-labeled DNA; Lanes 2 ± 3: A and G reactions; Lanes 4 and 5: 1 and 5 mm Hoechst-
ED; Lanes 6 and 7: 20 and 40 mm ImPyPy-ED; Lanes 8 and 9: 1 and 5 mm BBB-ED; Lanes 10 and 11: 1 and 5 mm
PBB-ED. c) Binding sites (in bold) as determined by MPE-FeII footprinting, shown on the 5' and 3'-end labelled
4281 ± 4350 segment of PBr322. Bar heights are proportional to the relative protection from cleavage at each
band. Shown are the protection patterns for 10 mm Hoechst, 20 mm ImPyPy-Dp, 10 mm BBB-Dp, and 10 mm PBB-
Dp on the 4281 ± 4350 segment of PBr322. d) Affinity-cleavage patterns for 1 mm Hoechst-ED, 20 mm ImPyPy-
ED, 1 mm BBB-EDTA, and 1 mm BBB-ED on the 4281 ± 4350 segment of PBr322. Line heights are proportional
to the relative cleavage at each band; binding sites (as determined above) are in bold.

2208
Helvetica Chimica Acta ± Vol. 83 (2000)

Helvetica Chimica Acta ± Vol. 83 (2000)
2209
Fig. 7. Ligand binding site and orientation summary for Hoechst-ED, BBB-ED, and PBB-ED
b
Fig. 6. a) MPE-FeII Footprinting of BBB-Dp (3-Dp) and PBB-Dp (4-Dp) on the 5'-PCR-labeled 3991 ± 4143
segment of PBr322. Lanes 1 and 18: intact 5'-labeled DNA; Lanes 2 ± 3 and 16 ± 17: A and G reactions; Lanes 4
and 15: MPE-FeII digestion of DNA in the absence of ligand; Lanes 5 ± 9: 0.5, 1, 10, 20, and 40 mm BBB-Dp;
Lanes 10 ± 14: 0.5, 1, 10, 20, and 40 mm PBB-Dp. b) Affinity cleavage of the 5'-PCR-labeled 3991 ± 4143 segment of
PBr322 with BBB-ED (3-ED) and PBB-ED (4-ED). Lanes 1 and 13: intact 5'-labeled DNA; Lanes 2 ± 3 and
11 ± 12: A and G reactions; Lanes 4 and 11: MPE-FeII digestion of DNA in the absence of ligand; Lanes 5 ± 7: 0.5,
1, and 5 mm BBB-ED; Lanes 8 ± 10: 0.5, 1, and 10 mm PBB-ED. c) Binding sites (in bold) as determined by MPE-
FeII footprinting, shown on the 5'- and 3'-end-labeled 3991 ± 4143 segment of pBr-322. Bar heights are
proportional to the relative protection from cleavage at each bond. Shown are the protection patterns for 10 mm
BBB-Dp (3-Dp) and 10 mm PBB-Dp (4-Dp). d) Affinity-cleavage patterns for 5 mm BBB-ED (3-ED) and 5 mm
PBB-ED (4-ED) on the 3991 ± 4143 segment of PBr322. Line heights are proportional to the relative cleavage at
each band.

2210
Helvetica Chimica Acta ± Vol. 83 (2000)
NMR Spectra were recorded with TMS (d ˆ 0 ppm) as internal standard on a Varian Mercury 300 (1H:
300 MHz, 13C: 75 MHz), or on a Varian Mercury 500 (13C: 125 MHz) spectrometer; chemical shifts d are given in
ppm. High-resolution mass spectra (HR-MS) were performed by Mass Consortium Inc., San Diego, CA.
2. Synthesis of Hoechst-ED (1-ED). ± Ethyl 2-(4-{2-[2-(4-methoxyphenyl)-1H-benzimidazol-6-yl]-1H-
benzimidazol-6-yl}piperazin-1-yl)acetate (8): 5-chloro-2-nitroaniline (6; 2 g, 12 mmol) and piperazine (2.6 g,
30 mmol, 2.5 equiv.) were combined in DMF (23 ml), and K2CO3 (1.4 g) was added. The mixture was heated to
1308 for 8 h. Upon cooling, the mixture was poured into H2O and extracted 3 times with AcOEt (100 ml). The
org. extracts were dried (Na2SO4) and evaporated to give 2-nitro-5-(piperazin-1-yl)aniline. The crude
piperazine (155 mg, 690 mmol) was dissolved in DMF (1 ml) and DIEA (0.5 ml), and ethyl bromoacetate (85 ml,
760 mmol, 1.1 equiv.) was added dropwise. The mixture was stirred overnight at r.t. The mixture was poured into
H2O and extracted 3 times with AcOEt (20 ml). The org. extracts were dried (Na2SO4) and evaporated to give
ethyl 4-(3-amino-4-nitrophenyl)piperazinecarboxylate. The crude ethyl ester (20 mg, 68 mmol) was dissolved in
AcOEt (3 ml), and excess 10% Pd-C (100 mg) was added. The mixture was stirred under an atmosphere of H2
for 1 h, and then filtered over Celite and concentrated in vacuo to give the crude ethyl 4-(3,4-diaminophe-
nyl)piperazinecarboxylate (7; 17 mg, 63 mmol), which was immediately coupled with aldehyde 9 (16 mg,
63mmol) by the same procedure for the preparation of 12 from 9 and 11 (vide infra). FC (SiO2; 10%MeOH/
AcOEt) gave 8 (15 mg, 29 mmol, 4% overall). UV (H2O): 260, 344 (42,000). 1H-NMR ((D6)DMSO, 300 MHz):
1.21 (t, J ˆ 7.5, 3 H); 2.48 (d, J ˆ 1.5, 2 H); 2.67 (m, 4 H); 2.95 (q, J ˆ 6.9, 2 H); 3.27 (m, 2 H); 3.83 (s, 3 H); 4.07
(q, J ˆ 7.5, 2 H); 6.89 (d, J ˆ 8.4, 2 H); 7.10 (d, J ˆ 9.0, 4 H); 7.20 ± 7.80 (m, 2 H); 8.14 (d, J ˆ 8.1, 4 H). HR-MS
‡
‡
(MALDI-FTMS): 511.2472. (C29H31N6O3 , [M ‡ H] ; calc. 511.2452).
2-((Carboxymethyl){2-[(carboxymethyl)({N-[3-({3-[2-(4-{2-[2-(4-hydroxyphenyl)-1H-benzimidazol-6-yl]-
1H-benzimidazol-6-yl}-piperazin-1-yl)acetylamino]propyl}methylamino)propyl]carbamoyl}methyl)amino]ethyl}-
amino)acetic Acid (1-ED). Compound 8 (15 mg, 29 mmol) was hydrolyzed with 48% HBr (2 ml) for 1 h at 1308.
Evaporation gave a crude acid, which was carried forward without further purification. The acid was coupled
with bis(3-aminopropyl)methylamine by the same procedure described for the preparation of 3-NH2 to produce
1-NH2 (7 mg, 41%). Excess ethylenediaminetetraacetic acid (EDTA) dianhydride (50 mg, 195 mmol) was
dissolved in DMSO/NMP (1 ml) and DIEA (1 ml) by heating at 558 for 5 min. The dianhydride soln. was added
to 1-NH2 (7 mg, 12 mmol) dissolved in DMSO (50 ml). The mixture was heated (558, 25 min), and the remaining
EDTA anhydride was hydrolyzed (0.1m NaOH, 3 ml, 558, 10 min). Aq. CF3COOH (TFA) (0.1 wt-%/v) was
added to adjust the total volume to 8 ml, and the soln. was purified directly by RP-HPLC (20% MeCN/0.1% aq.
TFA) to provide 1-ED (0.5 mg, 5%). UV (H2O): 261, 346 (42,000). 1H-NMR ((D6)DMSO, 300 MHz): 1.83 (m,
4 H); 2.49 (m, 15 H); 2.75 (m, 4 H); 2.91 (m, 4 H); 3.15 (m, 10 H); 6.95 (d, J ˆ 8.4, 2 H); 7.14 (m, 2 H); 7.60 (m,
2 H); 7.75 (m, 2 H); 8.06 (d, J ˆ 7.5, 2 H); 8.35 (br. s, 4 H); 10.15 (br. s, 3 H). HR-MS(MALDI-FTMS):
‡
‡
892.4052 (C43H55N11O9Na , [M ‡ Na] ; calc. 892.4076).
3. Synthesis of the BBB (3) and PBB (4) Trimers. ± 2-{2-[2-(4-Methoxyphenyl)-1H-benzimidazol-6-yl]-1H-
benzimidazol-6-yl}-1H-benzimidazole-6-carbonitrile (12). A soln. of 2-(4-methoxyphenyl)-1H-benzimidazole-
6-carbaldehyde (9; 44 mg, 170 mmol) and 11 (44 mg, 170 mmol) in PhNO2 (1 ml) was stirred at 1508 under Ar for
19 h. Evaporation of the solvent and trituration with CH2Cl2 and Et2O gave a brown solid after filtration.
Purification by RP (C18) silica-gel chromatography (35% MeCN/0.1% aq. TFA) gave 12 (51 mg, 62%). UV
(MeOH): 250, 341 (30,000). 1H-NMR ((D6)DMSO, 300 MHz): 3.82 (s, 3 H); 7.13 (m, 3 H); 7.67 (m, 1 H); 7.80
‡
‡
(m, 2 H); 8.18 (m, 6 H); 8.49 (m, 1 H). HR-ES-MS: 482.1799. (C29H20N7O , [M ‡ H] ; calc. 482.1729).
N-[3-(Dimethylamino)propyl]-2-{2-[2-(4-methoxyphenyl)-1H-benzimidazol-6-yl]-1H-benzimidazol-6-yl}-
1H-benzimidazole-6-carboxamide (3-Dp). A soln. of 12 (20 mg, 40 mmol) in 48% HBr (1.5 ml) was heated to
1508 for 4 h. The soln. was cooled to r.t. and evaporated to dryness. To a soln. of crude acid (20 mg, 39 mmol) in
DMF (1 ml) was added HOBt (10 mg, 84 mmol, 2.1 equiv.) and DCC (17 mg, 84 mmol, 2.1 equiv.), and the
mixture was allowed to stir for 30 min. The white precipitate was filtered off and DIEA (0.5 ml) and 3-
(dimethylamino)propylamine (0.25 ml) were added dropwise. The mixture was stirred at r.t. for 5 h and then
concentrated in vacuo. Purification by RP-HPLC (30% MeCN/0.1% aq. CF3COOH) gave 3-Dp (8 mg, 36%
overall). UV (H2O): 250, 346(30,000). 1H-NMR ((D6)DMSO, 300 MHz): 1.56 ± 1.81 (m, 2 H); 2.76 (s, 3 H); 2.78
(s, 3 H); 3.00 ± 3.25 (m, 4 H); 3.82 (s, 3 H); 7.13 (m, 3 H); 7.67 (m, 1 H); 7.80 (m, 2 H); 8.16 (m, 5 H); 8.46 (m,
‡
‡
2 H). HR-MS(MALDI-TOF): 585.2731 (C 34H33N8O2 , [M ‡ H] ; calc. 585.2721).
N-{3-[(3-Aminopropyl)methylamino]propyl}-2-{2-[2-(4-methoxyphenyl)-1H-benzimidazol-6-yl]-1H-benz-
imidazol-6-yl}-1H-benzimidazole-6-carboxamide (3-NH2). To a soln. of crude acid derived from hydrolysis of
12 in the previous step (20 mg, 39 mmol) in DMF (1 ml) was added HOBt (10 mg, 84 mmol, 2.1 equiv.) and DCC
(17 mg, 84 mmol, 2.1 equiv.), and the mixture was allowed to stir for 30 min. The white precipitate was filtered
off and DIEA (0.5 ml) and (3-aminopropyl)methylamine (0.25 ml) were added dropwise. The mixture was

Helvetica Chimica Acta ± Vol. 83 (2000)
2211
stirred at r.t. for 5 h and then concentrated in vacuo. Purification by RP-HPLC (30% MeCN/0.1% aq.
CF3COOH) gave 3-NH2 (11 mg, 33% overall). UV (H2O) 255, 348 (30,000). 1H-NMR ((D6)DMSO;
300 MHz): 1.50 ± 1.87 (m, 4 H); 2.52 ± 2.81 (m, 7 H); 3.84 (s, 3 H); 5.54 (m, 2 H); 7.15 (d, J ˆ 8.1, 2 H); 7.84 (m,
‡
‡
5 H); 8.16 (m, 4 H); 8.45 (s, 2 H). HR-MS(MALDI-TOF): 628.3162 (C 36H38N9O2 , [M ‡ H] ; calc. 628.3148).
2-((Carboxymethyl){2-[(carboxymethyl)({N-[3-({3-[(2-{2-[2-(4-methoxyphenyl)-1H-benzimidazol-6-yl]-
1H-benzimid-azol-6-yl}-1H-benzimidazol-6-yl)carbonylamino]propyl}methylamino)propyl]carbamoyl}methyl)-
amino]ethyl}amino)acetic Acid (3-ED). Excess EDTA dianhydride (50 mg, 195 mmol) was dissolved in DMSO/
N-methylphthalimide (NMP) (1 ml) and DIEA (1 ml) by heating at 558 for 5 min. The dianhydride soln. was
added to 3-NH2 (2 mg, 2 mmol) dissolved in DMSO (50 ml). The mixture was heated (558, 25 min) and the
remaining EDTA-anhydride hydrolyzed (0.1m NaOH, 3 ml, 558, 10 min). Aq. CF3COOH (0.1 wt-%/v) was
added to adjust the total volume to 8 ml, and the soln. was purified directly by RP-HPLC (20% MeCN/0.1% aq.
CF3COOH) to provide 3-ED (1.0 mg, 38%). UV (H2O): 253, 346 (30,000). 1H-NMR ((D6)DMSO, 300 MHz):
1.80 (m, 2 H); 1.93 (m, 2 H); 2.49 (d, J ˆ 1.5, 3 H); 2.77 (m, 8 H); 3.11 (m, 12 H); 3.98 (s, 3 H); 7.13 (d, J ˆ 8.0,
‡
2 H); 7.78 (m, 5 H); 8.20 (m, 4 H); 8.45 (s, 2 H). HR-MS(MALDI-TOF): 924.3738 (C 46H51N11O9Na , [M ‡
‡
Na] ; calc. 924.3763).
2-{2-[2-(4-Methoxyphenyl)-1H-imidazo[4,5-b]pyridin-5-yl]-1H-benzimidazol-6-yl}-1H-benzimidazole-6-
carbonitrile (13). A soln. of aldehyde 10 (40 mg, 160 mmol) and 11 (41 mg, 16 mmol) in PhNO2 (1 ml) was stirred
at 1508 under Ar for 19 h. Evaporation of the solvent and trituration with CH2Cl2 and Et2O gave a brown solid
after filtration. Purification by RP (C18) silica-gel chromatography (35% MeCN/0.1% aq. CF3COOH) gave 13
(40 mg, 51%). UV (MeOH): 251, 360 (45,000). 1H-NMR ((D6)DMSO, 300 MHz): 3.83 (s, 3 H); 7.15 (d, J ˆ 10.0,
2 H); 7.65 (d, J ˆ 8.2, 1 H); 7.80 (d, J ˆ 8.1, 1 H); 7.88 (d, J ˆ 8.1, 1 H); 7.80 (d, J ˆ 8.0, 1 H); 8.20 ± 8.40 (m, 6 H);
‡
‡
8.55 (m, 1 H). HR-MS(MALDI-TOF): 483.1656 (C 28H19N8O , [M ‡ H] ; calc. 483.1682).
N-[3-(Dimethylamino)propyl]-2-{2-[2-(4-hydroxyphenyl)-1H-imidazo[4,5-b]pyridin-5-yl]-1H-benzimida-
zol-6-yl}-1H-benzimidazole-6-carboxamide (4-Dp). A soln. of 13 (20 mg, 40 mmol) in 48% HBr was heated to
1508 for 1 h. The soln. was cooled to r.t. and evaporated to dryness. To a soln. of crude acid (20 mg, 39 mmol) in
DMF (1 ml) was added HOBt (10 mg, 84 mmol, 2.1 equiv.) and DCC (17 mg, 84 mmol, 2.1 equiv.) and the
mixture was allowed to stir for 30 min. The white precipitate was filtered off, and DIEA (0.50 ml) and 3-
(dimethylamino)propyl amine (0.25 ml) were added dropwise. The mixture was stirred at r.t. for 5 h and then
concentrated in vacuo. Purification by RP-HPLC (30% MeCN/aq. CF3COOH (0.1 wt-%/v)) gave 4-Dp (5 mg,
25% overall). UV (H2O): 251, 362 (45,000). 1H-NMR ((D6)DMSO, 300 MHz): 1.56 ± 1.81 (m, 2 H); 2.76 (s,
3 H); 2.78 (s, 3 H); 3.00 ± 3.25 (m, 4 H); 6.98 (d, J ˆ 8.0, 5 H); 8.14 (m, 7 H). HR-MS(MALDI-TOF): 572.2505.
‡
‡
(C32H31N9O2 , [M ‡ H] ; calc. 572.2522).
N-{3-[(3-Aminopropyl)methylamino]propyl}-2-{2-[2-(4-hydroxyphenyl)-1H-imidazo[4,5-b]pyridin-5-yl]-
1H-benzimidazol-6-yl}-1H-benzimidazole-6-carboxamide (4-NH2). To a soln. of crude acid derived from
hydrolysis of 13 in the previous step (20 mg, 39 mmol) in DMF (1 ml) was added HOBt (10 mg, 84 mmol,
2.1 equiv.) and DCC (17 mg, 84 mmol, 2.1 equiv.), and the mixture was allowed to stir for 30 min. The white
precipitate was filtered off, and DIEA (0.50 ml) and (3-aminopropyl)methylamine (0.25 ml) were added
dropwise. The mixture was stirred at r.t. for 5 h and then concentrated in vacuo. Purification by RP-HPLC (30%
MeCN/0.1% aq. CF3COOH) gave 4-NH2 (10 mg, 30% overall). UV (H2O): 250, 362 (45,000). 1H-NMR
(CD3CN, 300 MHz): 1.50 ± 2.05 (m, 4 H); 2.81 (s, 3 H); 3.15 ± 3.50 (m, 8 H); 6.14 (d, J ˆ 8.5, 5 H); 7.14 (m, 7 H).
‡
‡
HR-MS(MALDI-TOF): 615.2962. (C 34H35N10O2 , [M ‡ H] ; calc. 615.2944).
2-((Carboxymethyl){2-[(carboxymethyl)({N-[3-({3-[(2-{2-[2-(4-hydroxyphenyl)-1H-imidazo[4,5-b]pyri-
din-5-yl]benzimidazol-6-yl}-1H-benzimidazol-6-yl)carbonylamino]propyl}methylamino)propyl]carbamoyl}-
methyl)amino]ethyl}amino)acetic Acid (4-ED). Excess EDTA dianhydride (50 mg, 195 mmol) was dissolved in
DMSO/NMP (1 ml) and DIEA (1 ml) by heating at 558 for 5 min. The dianhydride soln. was added to 4-NH2
(2 mg, 2 mmol) dissolved in DMSO (50 ml). The mixture was heated (558, 25 min), and the remaining EDTA
anhydride was hydrolyzed (0.1m NaOH, 3 ml, 558, 10 min). Aq. CF3COOH (0.1 wt-%/v) was added to adjust the
total volume to 8 ml, and the soln. was purified directly by RP-HPLC (20% MeCN/0.1% aq. CF3COOH) to
provide 4-ED (0.5 mg, 28%). UV (H2O): 250, 358 (45,000). 1H-NMR ((D6)DMSO, 300 MHz): 1.82 (m, 2 H);
1.91 (m, 2 H); 2.47 (d, J ˆ 1.5, 3 H); 2.80 ± 2.88 (m, 8 H); 3.40 (m, 12 H); 3.80 (s, 3 H); 6.23 (d, J ˆ 8.0, 5 H); 7.15
‡
‡
(m, 7 H). HR-MS(MALDI-TOF): 889.3707. (C 44H49N12O9 , [M ‡ H] ; calc. 889.3740).
4. Synthesis of the HBB Trimer 5. ± 2-[2-(4-Chloro-7-methoxy-2-methyl-1H-benzimidazol-6-yl)-1H-
benzimidazol-6-yl]-1H-benzimidazole-6-carbonitrile (16). A soln. of 15 (117 mg, 520 mmol) and 11 (130 mg,
520 mmol) in PhNO2 (5.2 ml) was stirred at 1508 for 18 h. After evaporation, the residue was triturated with
AcOEt, dissolved in MeOH, and filtered. The filtrate was evaporated to yield 16 (184 mg, 78%). UV (MeOH):

2212
Helvetica Chimica Acta ± Vol. 83 (2000)
245, 336 (37445). 13C-NMR (CD3OD, 125 MHz): 156.29; 151.71; 126.22; 124.29; 123.05; 122.96; 122.03; 120.82;
‡
‡
105.06; 61.79; 14.43. HR-ESI-MS: 454.1167 (C24H17ClN7O , [M ‡ H] ; calc. 454.1183).
N-[3-(Dimethylamino)propyl]-2-[2-(4-chloro-7-hydroxy-2-methyl-1H-benzimidazol-6-yl)-1H-benzimid-
azol-6-yl]-1H-benzimidazole-6-carboxamide (5). A soln. of 16 (33 mg, 73 mmol) in 48% HBr (1.5 ml) was
refluxed for 30 min. Evaporation and azeotropic destillation with benzene (3Â) afforded the crude hydroxy
acid. To a soln. of this acid in DMF (2 ml) and DIEA (0.5 ml) were added DCC (31.8 mg, 153 mmol, 2.1 equiv.)
and HOBt (18.7 mg, 153 mmol, 2.1 equiv.), and the mixture was stirred at r.t. for 20 min. 3-(dimethylamino)-
propylamine (0.25 ml) was added, and the mixture was stirred at r.t. for 3 h. After addition of 0.1% CF3COOH/
H2O (2 ml), the solution was filtered and purified by HPLC (25% MeCN/0.1% TFA) to give 5 (3.9 mg, 20%).
‡
‡
HR-ESI-MS: 543.2042 (C28H28ClN8O2 , [M ‡ H] ; calc. 543.2024).
5. Preparation and Labeling of DNA Restriction Fragment. ± The plasmid pBR322 was linearized with
EcoRI and RsaI and then treated with Sequenase v. 2.0, deoxyadenosine 5'-[a-32P]triphosphate, and thymidine
5'-[a-32P]triphosphate. The 3'-end labeled fragment was loaded onto a 7% nondenaturing polyacrylamide gel.
The desired 519 base-pair band was visualized by autoradiography and isolated. Chemical sequencing reaction
performed according to published methods.
Sequences of the PCR primers for plasmid pBR322: Primer P1: 5'-GTACTCAACCAAGTCATTCTG-3',
Primer P2: 5'-GGAAGAGTATGAGTATTCAAC-3'. Primer 2 was 5'-radiolabeled with g-32P-[ATP] using T4-
polynucleotide kinase. PCR amplification in the presence of P1, P2 (labeled), plasmid pBR322, and Taq
polymerase gave a 298 bp DNA fragment that was purified on a 6% nondenaturating polyacrylamide gel.
Sequences of the PCR primers for plasmid pJT3: Primer P1: 5'-GCAACTGTTGGGAAGGGCGAT-3',
Primer P2: 5'-TTAAGACGCTACGACGCTAGCT-3'. Primer 2 was 5'-radiolabeled with g-32P-[ATP] using T4-
polynucleotide kinase. PCR amplification in the presence of P1, P2 (labeled), plasmid pJT3, and Taq
polymerase gave a 286 bp DNA fragment that was purified on a 6% nondenaturating polyacrylamide gel.
6. MPE ´ FeII Footprint Titrations. ± All MPE footprinting reactions were performed in a total volume of
40 ml containing 5'- or 3'-32P-radiolabeled DNA fragments (17,000 cpm) and final concentrations of 25 mm Tris
acetate, 10 mm NaCl, 100 mm/bp calf thymus DNA, pH 7.0, and either 0.1 to 100 mm polyamide or
polybenzimidazole ligand. The solns. were allowed to equilibrate for 1 ± 3 h at 228. After addition of 4 ml of
50 mm MPE-FeII (freshly prepared from 100 ml of a 100 mm MPE soln. and 100 ml of 100 mm Fe(NH4)2(SO4)2 ´
6H2O soln., the mixtures were incubated for 5 min. Cleavage was initiated by addition of 4 ml of a 50 mm DTT
soln. and allowed to proceed for 14 min at r.t. Reactions were stopped by EtOH precipitation. The precipitates
were resuspended in 1 Â TBE/80% formamide loading buffer, denaturated by heating at 858 for 10 min, and
cooled on ice. The reaction products were separated by electrophoresis on a 6% polyacrylamide gel in 1 Â TBE
at 2000 V for 1 h. Gels were dried on a slab dryer and exposed to a photostimulatable storage phosphor imaging
plate (Kodak Storage Phosphor Screen SO230 obtained from Molecular Dynamics) in the dark at 228 for 12 to
24 h. The data from the storage screens were obtained using a Molecular Dynamics 400SPhosphorImager and
analyzed by volume integration of the target sites and reference blocks using the ImageQuant version 3.3.
software.
7. Affinity-Cleavage Titrations. ± All affinity-cleavage-titration reactions were performed in a total volume
of 40 ml containing 5'- or 3'-32P-radiolabeled DNA fragments (17,000 cpm) and final concentrations of 25 mm
Tris acetate, 20 mm NaCl, 100 mm/bp calf thymus DNA, pH 7.0, and either 0.1 to 10 mm polyamide or
polybenzimidazole ligand. The solns. were allowed to equilibrate for 1 ± 3 h at 228. After addition of 4ml of
100 mm Fe(NH4)2(SO4) ´ 6H2O soln., the mixtures were incubated for 5 min. Cleavage was initiated by addition
of 4ml of a 50 mm DTT soln. and allowed to proceed for 30 min at r.t. Reactions were stopped by EtOH
precipitation. The precipitates were resuspended in 1 Â TBE/80% formamide loading buffer, denaturated by
heating at 858 for 10 min, and cooled on ice. The reaction products were separated by electrophoresis on a 6%
polyacrylamide gel in 1 Â TBE at 2000 V for 1 h. Gels were dried on a slab dryer and exposed to a
photostimulatable storage phosphor imaging plate (Kodak Storage Phosphor Screen SO230 obtained from
Molecular Dynamics) in the dark at 228 for 12 to 24 h. The data from the storage screens were obtained using a
Molecular Dynamics 400SPhosphorImager and analyzed by volume integration of the target sites and reference
blocks using the ImageQuant version 3.3. software.
REFERENCES
[1] P. B. Dervan, Science 1986, 232, 464.
[2] D. E. Wemmer, P. B. Dervan, Curr. Opin. Struct. Biol. 1997, 7, 355.

Helvetica Chimica Acta ± Vol. 83 (2000)
2213
[3] C. Zimmer, U. Wahnert, Prog. Biophys. Mol. Biol. 1986, 47, 31.
[4] M. L. Kopka, C. Yoon, D. Goodsell, P. Pjura, R. E. Dickerson, Proc. Natl. Acad. Sci. U.S.A. 1985a, 82, 1376.
[5] M. L. Kopka, C. Yoon, D. Goodsell, P. Pjura, R. E. Dickerson, J. Mol. Biol. 1985b, 183, 553.
[6] M. Coll, C. A. Fredrick, A. H. Wang, A. Rich, Proc. Natl. Acad. Sci. U.S.A. 1987, 84, 8385.
[7] R. E. Klevit, D. E. Wemmer, R. R. Reid, Biochemistry, 1986, 25, 3296.
[8] J. G. Pelton, D. E. Wemmer, Biochemistry 1988, 27, 8088.
[9] J. G. Pelton, D. E. Wemmer, Proc. Natl. Acad. Sci. U.S.A. 1989, 86, 5723.
[10] J. G. Pelton, D. E. Wemmer, J. Am. Chem. Soc 1990, 112, 1393.
[11] R. F. Martin, N. Holmes, Nature, 1983, 302, 452.
[12] K. D. Harshman, P. B. Dervan, Nucl. Acids Res. 1985, 13, 4825.
[13] M. Teng, N. Usman, C. A. Frederick, A. Wang, Nucl. Acids Res. 1988, 16, 2671.
[14] P. E. Pjura, G. Kazimierz, R. E. Dickerson, J. Mol. Biol. 1987, 197, 257.
[15] S. Kumar, B. Yadagiri, J. Zimmermann, R. T. Pon, J. W. Lown, J. Biomol. Struct. Dyn. 1990, 8, 331.
[16] W. S. Wade, M. Mrksich, P. B. Dervan, J. Am. Chem. Soc 1992, 114, 8783.
[17] M. Mrksich, W. S. Wade, T. J. Dwyer, B. H. Geierstanger, D. E. Wemmer, P. B. Dervan, Proc. Natl. Acad.
Sci. U.S.A. 1992, 89, 7568.
[18] P. B. Dervan, R. W. Burli, Curr. Opin. Chem. Biol. 1999, 3, 688.
[19] S. Frau, J. Bernadou, B. Meunier, Bull. Soc. Chim. Fr. 1996, 133, 1053.
[20] M. Kubista, B. Akerman, B. Norden, Biochemistry 1987, 26, 4545.
[21] L. Wang, C. Bailly, A. Kumar, D. Ding, M. Bajic, D. W. Boykin, W. D. Wilson, Proc. Natl. Acad. Sci. U.S.A.
2000, 97, 12.
[22] J. W. Lown, in ꢀAdvances in DNA Sequence Specific Agentsꢁ, 1998, 3, 67.
[23] M. W. Van Dyke, P. B. Dervan, Nucl. Acids Res. 1983, 11, 5555.
[24] M. Lee, P. H. Spotts, J. Eckert, C. Walker, J. A. Nobles, Heterocycles, 1991, 32, 2093.
[25] M. P. Singh, T. Joseph, S. Kumar, Y. Bathini, L. J. Lown, Chem. Res. Toxicol., 1992, 5, 597.
[26] A. Czarny, W. D. Wilson, J. Boykin, J. Heterocycl. Chem. 1996, 33, 1393.
[27] D. H. Herman, E. E Baird, P. B. Dervan, J. Am. Chem. Soc 1998, 120, 1382.
[28] A. Tanaka, K. Ito, S. Nishino, Y. Motoyama, H. Takasugi, Chem. Pharm. Bull. 1994, 42, 560.
[29] J. W. Trauger, E. E. Baird, M. Mrksich, P. B. Dervan, J. Am. Chem. Soc. 1996, 118, 6160.
[30] Gas-phase ab initio HF/6-31G** energy minimization; Jaguar software package, Schrödinger, Inc.
Received May 22, 2000