Institut für Molekulare Biotechnologie, Postfach 100813, D-07708 Jena, Germany
In studies on the AT tract of DNA dodecamers Nelson et al. [1] and
Coll
et al. [2] have identified for the first time so-called cross-strand
diagonal
hydrogen bonds. In addition to H-bonds in Watson-Crick base-pairs and in
mismatches they represent a further type of hydrogen bonds in DNA
because
they connect different base-pairs. These inter-base-pair hydrogen bonds
have been claimed to contribute to a particular stability and rigidity
of
AT sequence regions in DNA which could possibly explain what
differentiates
AT tracts from other sequence parts. A few further inter-base-pair
H-bonds
have been described in other DNA structures including AT- and non-AT
regions
and in a DNA-protein complex as well.
In order to gain deeper insight into the importance of this type of
H-bonds
for DNA structures it would be interesting to know if they are found in
other structures as well and to learn more about their geometrical
properties.
We have used our new program HBexplore [3] to perform a systematic
search
for non-intra-base-pair H-bonds in a set of 202 DNA structures from the
Protein Data Bank (PDB) including A-, B- and Z-DNA, other unusual DNA
structures,
DNA-drug and DNA-protein complexes, adopting the geometrical criteria
dHA
< 2.5 Å, dDA < 3.9 Å, aDHA
> 90 deg, aDAA1 > 90 deg, aHAA1 > 90 deg,
where d and a stand for distance and angle, D and A are the H-bond donor
and acceptor atoms and A1 is the bonded neighbor of A.
HBexplore identifies in the 202 DNA structures analyzed a total number
of
5776 H-bonds and among them 82 non-intra-base-pair H-bonds in 44
structures.
We report distributions of geometrical H-bond parameters both for the
complete
set and for the non-intra-base-pair H-bonds alone and discuss other
statistical properties and individual examples.
The systematic search for non-intra-base-pair H-bonds in DNA structures
has identified much more bonds of this type than known so far. Even
though
the percentage of non-intra-base-pair H-bonds in DNA structures is very
small they may have important biological implications. The results
obtained
can also be useful for a separation of real structure effects from
effects
of crystal packing and even for the identification of possible errors in
structure determination.