Current Problems in Computer Simulation of Variability …
239
can be expected from MP2 computation of the minimal fragments of DNA. The
method correctly reproduces the conformational regularities mentioned in the section
of Introduction, but predicts shortened contacts between the stacked bases, and that
renders this level of theory unacceptable. This conclusion about the limitations of the
MP2 level of theory in application to DNA fragments had further been confirmed by
systematic study of pairwise interactions of DNA bases [21] at MP2/6-31G** level
of theory. Those computations conclude that MP2 systematically overestimates the
base stacking energy for various base positions and shortens the stacking C–C and
C–N distances up to 3.0 Å for the most of pairwise combinations.
The main QM method that we use for dDMP and cdDMP optimization is based on
the Density Functional Theory (DFT). This method provides reasonable compromise
between the accuracy and computational cost. We use Gaussian09 [22] and ADF
(Amsterdam Density Functional) [23, 24] programs for geometry optimization of
the studied systems. Employed functionals are PW91 [25] and PBE [26], as well
as M05-2X [27] and M06-2X [28], which are designed for computation of stacking
configurations of organic molecules. We also use DFT-D method, which mixes the
quantum mechanical level of theory with classic corrections, on selected dDMPs
and cdDMPs. In addition to popular functionals B97-D3 [29] and B3LYP-D3 [30]
implemented in GAUSSIAN09 program, and PBE-D [31] functional implemented
in ADF package [24], we also use PW91-ulg functional [32] recently added to the
ADF program. Computation of conformational characteristics of DNA fragments
including torsion angles, deoxyribose ring puckering, and mutual base arrangements
employs 3DNA software [33].
3 Sources of Variability in the DNA 3D-Structure
Analysis of the molecular structure of DNA suggests the presence of two main sources
of variability in its 3D structure. These are (1) flexibility of SPB due to rotation around
single bonds (including variability of sugar puckering), and (2) ability of each of the
bases to participate in various base-base complexes. Molecular structure of bases
enables the formation of various pairwise complexes corresponding to the energy
minima of intermolecular H-bonding and stacking interactions. In this work, we limit
our analysis of mutual position of subunits corresponding to the formation of double
helices with Watson-Crick nucleoside pairs, i.e. with A:T and G:C complementary
base pairs and anti-orientation of base and deoxyribose in both nucleosides. Some
of other experimentally observed polynucleotide structures, such as left-handed Zconformation families with purine nucleosides in syn-conformation, Hoogsteen or
parallel duplexes with base pairing different from Watson-Crick one, and duplexes
with mispairs arising as a result of errors of biosynthesis have been considered in
our earlier studies [4–8].
239
can be expected from MP2 computation of the minimal fragments of DNA. The
method correctly reproduces the conformational regularities mentioned in the section
of Introduction, but predicts shortened contacts between the stacked bases, and that
renders this level of theory unacceptable. This conclusion about the limitations of the
MP2 level of theory in application to DNA fragments had further been confirmed by
systematic study of pairwise interactions of DNA bases [21] at MP2/6-31G** level
of theory. Those computations conclude that MP2 systematically overestimates the
base stacking energy for various base positions and shortens the stacking C–C and
C–N distances up to 3.0 Å for the most of pairwise combinations.
The main QM method that we use for dDMP and cdDMP optimization is based on
the Density Functional Theory (DFT). This method provides reasonable compromise
between the accuracy and computational cost. We use Gaussian09 [22] and ADF
(Amsterdam Density Functional) [23, 24] programs for geometry optimization of
the studied systems. Employed functionals are PW91 [25] and PBE [26], as well
as M05-2X [27] and M06-2X [28], which are designed for computation of stacking
configurations of organic molecules. We also use DFT-D method, which mixes the
quantum mechanical level of theory with classic corrections, on selected dDMPs
and cdDMPs. In addition to popular functionals B97-D3 [29] and B3LYP-D3 [30]
implemented in GAUSSIAN09 program, and PBE-D [31] functional implemented
in ADF package [24], we also use PW91-ulg functional [32] recently added to the
ADF program. Computation of conformational characteristics of DNA fragments
including torsion angles, deoxyribose ring puckering, and mutual base arrangements
employs 3DNA software [33].
3 Sources of Variability in the DNA 3D-Structure
Analysis of the molecular structure of DNA suggests the presence of two main sources
of variability in its 3D structure. These are (1) flexibility of SPB due to rotation around
single bonds (including variability of sugar puckering), and (2) ability of each of the
bases to participate in various base-base complexes. Molecular structure of bases
enables the formation of various pairwise complexes corresponding to the energy
minima of intermolecular H-bonding and stacking interactions. In this work, we limit
our analysis of mutual position of subunits corresponding to the formation of double
helices with Watson-Crick nucleoside pairs, i.e. with A:T and G:C complementary
base pairs and anti-orientation of base and deoxyribose in both nucleosides. Some
of other experimentally observed polynucleotide structures, such as left-handed Zconformation families with purine nucleosides in syn-conformation, Hoogsteen or
parallel duplexes with base pairing different from Watson-Crick one, and duplexes
with mispairs arising as a result of errors of biosynthesis have been considered in
our earlier studies [4–8].