sequence of the type 5 -XGXCX, (again X can be either A or T). An NMR structure of the complex showed
that ImPyPy binds as an antiparallel 2:1 dimerin a similar manner to distamycin binding to A T-tracts in
the minor groove. In this arrangement, an imidazole ring is stacked against a pyrrole ring and thus the
antiparallel pair can distinguish a G C base pair from a C G, i.e. the pair Im/Py targets G C and Py/Im rec-
ognizes C G. Each G C base pair has an imidazole on the G strand and a pyrrole on the C strand. A hydro-
gen bond is formed between the N-3 atom of the imidazole ring and the amino group of guanine as well as a
series of hydrogen bonds between polyamide NH groups and the edge of the G C base pair.
Thermodynamic studies reveal that the sequence selectivity of the Im/Py pair is mostly driven by a large
favourable enthalpy term. Imidazole nitrogen atoms do not hydrogen bond to the cytosine side of the base
pair due to an unfavourable bond angle for this interaction. In addition, a Py/Py pair is degenerate and can
recognize A T and T A base pairs equally well in preference to G C or C G base pairs.
These discoveries suggested a new design strategy: unsymmetrical polyamide ring pairs can be used for
specific recognition of the DNA minor groove where one pair of rings in the ligand targets one base pair
of DNA. An Im is used for a guanine and a Py for a cytosine in G C or C G base pairs, whereas two Py
rings are used to target either A T or T A.
To test these pairing rules, Dervan changed an Im in one of the polyamides of a homodimer to Py to give
the heterodimer ImPyPy/PyPyPy. As predicted, this pair of ligands binds specifically to the sequence
5 -XGXXX. Further, ImPyImPy binds to the minor groove of the sequence 5 -GCGC. By analogy with
distamycin, binding of an alkyl chain is always to an A T site, so that the exact recognition sequence is
XGCGCX. The specificity is extremely high, such that even a single mismatch with a recognition motif, e.g.
G C in place of C G, leads to a 10-fold drop in binding affinity.
Since polyamide dimersdo not distinguish A T from T A base pairs in the minor groove, a third type of
monomer was designed as a thymine-selective moiety, N-methyl-3-hydroxypyrrole (Hp). An Hp/Py pair
recognizes a T A base pair whereas a Py/Hp combination recognizes A T base pairs. Both orientations dis-
favour binding to G C and C G base pairs. X-ray structures of two Hp-containing polyamide–DNA complexes
confirm that the hydroxyl group of Hp fits snugly into the asymmetric cleft between adenine C-2 and
thymine O-2 atoms. In addition, there is a specific hydrogen bond between the hydroxyl group of Hp and
thymine O-2. Hp-containing polyamides bind with lower affinity than the equivalent Py-containing ones.
The reason for this is not clear but may be related to factors such as partial melting of the T A base pair rec-
ognized by Hp/Py or a lengthening of the amide–DNA hydrogen bond on the C-terminal of the Hp residue.
Nevertheless, with the addition of this third type of monomer, it has become possible to create polyamides
containing different combinations of Im, Py and Hp rings. Using the Dervan Pairing Rules, one can dis-
tinguish all four types of Watson–Crick base pairs and thus design polyamide ligands to target any specific
DNA sequence (Figure 9.16).
Linked polyamidesare a further synthetic development where two separate molecules are covalently
joined together to reduce unfavourable entropy loss upon binding. It also results in an increase in specificity
by ensuring that the two molecules bind at the same site. An additional problem is that once the number of
rings in the polyamide ligand increases above six, affinity is decreased. This is because the ligand loses
synchronization in either length or curvature with respect to the DNA groove. The problem is addressed
by introduction of linkers with torsional flexibility, such as a -alanine subunit, which allows the natural
curvature of the polyamide to be relaxed. Molecules have been designed that can bind as overlapping
homodimers to recognize up to 11 base pairs with subnanomolar dissociation constants. The eventual aim
is to target a sequence of 16–18 base pairs, since these would be unique in the human genome.
9.7.5 Hoechst 33258
Hoechst 33258(Figure 9.13) is a fluorescent dye often used as a chromosome stain. Like other minor
groove binders, Hoechst 33258 binds to A T rich sequences preferentially. In fact it has a 50-fold greater
affinity for AATT than for TATA and higher binding to AATT than to AAAA, which is in contrast to other
A T-specific minor groove drugs such as distamycin, netropsin and berenil.^44
366 Chapter 9