Several studies on third-strand binding to a homo-purinehomo-pyrimidine duplex have established
the following features:
A third homo-pyrimidine strand binds parallel to the homo-purine strand using Hoogsteen
hydrogen bonds (i.e.the homo-pyrimidine strands are anti-parallel).
A third homo-purine strand binds anti-parallel to the original homo-purine strand using reversed
Hoogsteen hydrogen bonds.
The bases in the third strand have a regular anticonformation of the glycosylic bond.
Synthetic oligodeoxyribonucleotides having an -glycosylic linkage also bind as a third strand, par-
allel for poly(d--T) and anti-parallel for poly(d--TC).Triple helices are less stable than duplexes. Thermodynamic parameters have been obtained from melting
curves, from kinetics, and from the use of differential scanning calorimetry (DSC) (Section 11.4.4). Using
this last technique, values of H° 22 2 kJ mol^1 and S° 70 7 J mol^1 K^1 have been found for
d(CTTCCTCCTCT). The pKavalue of cytidine in isolation is 4.3 but it is higher in oligonucleo-
tides because of their polyanionic phosphate backbone. So it is to be expected that the stability of triple
helices is seen to decrease as the pH rises above 5.
Triple helix stability can be enhanced by the use of modified nucleotides. 5-Methylcytosine increases
stability at neutral pH, probably by a hydrophobic effect, and 5-bromouracil can usefully replace thymine.
Oligoribonucleotides bind more strongly than do deoxyribonucleotides and 2-O-methylribonucleotides
bind even better. Finally, Hélène has shown that at the attachment of an intercalating agent to the 5- (or
3 -end) of the third strand can greatly enhance the stability of the triple-stranded helix.
The major application of triple helices relates to the specificity of the interaction between the single
strand and a much larger DNA duplex. This is because homo-pyrimidines have been identified as poten-
tial vehicles for the sequence-specific delivery of agents that can modify DNA and thereby control genes.
The DNA of the bacterium E. colihas 4.5 Mbp, so the minimum number of base pairs needed to define a
unique sequence in its genome is 11bp (i.e. 411 4,194,304 assuming a statistically random distribution
of the four bases). The corresponding number for the human genome is about 17 bp. Thus, a synthetic 17-mer
could be expected to identify and bind to a unique human DNA target and thus deliver a lethal agent to a
specific sequence of DNA. In practice, the energetics of mismatched base triples is complex and depends
on nearest neighbours, metal ions and other parameters. However, a value of about 1.5 kJ mol^1 per mis-
match seems to fit much of the data and suggests that the specificity of triple helices is at least as good as
that of double-helical complexes.
2.3.6.1 H-DNA. A new polymorph of DNA was discovered in 1985 within a sequence of d(A-G) 16 in
the polypurine strand of a recombinant plasmid pEJ4. Its requirement for protons led to the name H-DNA (half
of its C residues are protonated, so the transition depends on acid pH as well as on a degree of negative
supercoiling). Probes for single-stranded regions of DNA (especially osmium tetroxide:pyridine (Section 8.3)
and nuclease P1 cleavage) were used to identify specific sites and provide experimental support for the
model advanced earlier for a triple helical H-DNA (Figure 2.33d). This has a Watson–Crick duplex which
extends to the centre of the (dT–dC)n(dG–dA)ntract and the second half of the homo-pyrimidine tract
then folds back on itself, anti-parallel to the first half and winding down the major groove of the helix. The
second half of the poly-purine tract also folds back, probably in an unstructured single-stranded form. The
energetics of nucleation of H-DNA suggests that it requires at least 15bp for stability and the consequent
loss of twist makes H-DNA favoured by negative supercoiling.
Although antibodies have been raised to detect triple-stranded structures, no evidence has yet been
found for their natural existence in cells in vivo.
2.3.7 Other Non-Canonical DNA Structures
2.3.7.1 Four-Stranded Motifs. Both G- and C-rich DNA sequences have been found to adopt four-
stranded motifs, also called tetraplexesor quadruplexes(see also Section 9.10.2). Sequences containing
52 Chapter 2