doi:10.1006/jmbi.2000.3894 available online at http://www.idealibrary.com on
J. Mol. Biol. (2000) 300, 1113±1131
Water and Ion Binding Around RNA and DNA
(C,G) Oligomers
Pascal Auffinger and Eric Westhof*
Institut de Biologie MoleÂculaire
et Cellulaire du CNRS
ModeÂlisations et Simulations
des Acides NucleÂiques, UPR
9002, 15 rue Rene Descartes
67084, Strasbourg Cedex
France
The dynamics, hydration, and ion-binding features of two duplexes, the
A(r(CG)12) and the B(d(CG)12), in a neutralizing aqueous environment
with 0.25 M added KCl have been investigated by molecular dynamics
(MD) simulations. The regular repeats of the same CˆG base-pair motif
have been exploited as a statistical alternative to long MD simulations in
order to extend the sampling of the conformational space. The trajectories
demonstrate the larger ¯exibility of DNA compared to RNA helices. This
¯exibility results in less well de®ned hydration patterns around the DNA
than around the RNA backbone atoms. Yet, 22 hydration sites are clearly
characterized around both nucleic acid structures. With additional results
from MD simulations, the following hydration scale for CˆG pairs can
be deduced: A-DNA < RNA (‡3 H2O) and B-DNA < RNA (‡2 H2O). The
calculated residence times of water molecules in the ®rst hydration shell
of the helices range from 0.5 to 1 ns, in good agreement with available
experimental data. Such water molecules are essentially found in the
vicinity of the phosphate groups and in the DNA minor groove. The calculated number of ions that break into the ®rst hydration shell of the
nucleic acids is close to 0.5 per base-pair for both RNA and DNA. These
ions form contacts essentially with the oxygen atoms of the phosphate
groups and with the guanine N7 and O6 atoms; they display residence
times in the deep/major groove approaching 500 ps. Further, a signi®cant sequence-dependent effect on ion binding has been noted. Despite
slight structural differences, K‡ binds essentially to GpC and not to CpG
steps. These results may be of importance for understanding various
sequence-dependent binding af®nities. Additionally, the data help to
rationalize the experimentally observed differences in gel electrophoretic
mobility between RNA and DNA as due to the difference in hydration
(two water molecules in favor of RNA) rather than to strong ion-binding
features, which are largely similar for both nucleic acid structures.
# 2000 Academic Press
*Corresponding author
Keywords: molecular dynamics; RNA; DNA; hydration shell;
gel electrophoresis
Introduction
The aqueous solvent plays a central role in the
maintenance of the three-dimensional architecture
of biopolymers. It has been shown by experimental
and theoretical methods that water molecules,
even located at long distances from the solute
Ê ), are structurally important (Auf®nger et al.,
(>10 A
1996; Israelachvili & WennerstroÈm, 1996). These
apparently unstructured hydration shells are
Abbreviations used: MD, molecular dynamics; MMD,
multiple MD; HB%, hydrogen bonding percentage.
E-mail address of the corresponding author:
[email protected]
0022-2836/00/051113±19 $35.00/0
involved in the formation of hydration forces at
play in folding and recognition events and should
be studied thoroughly (Sorenson et al., 1999). On
the other side, a precise knowledge of the shape of
the primary hydration shell at the nucleotide and
base-pair level is required. Indeed, ligands (nucleotides, amino acids, drugs, . . .) interact with nucleic
acids either by direct contacts, thereby replacing
some water molecules from their locations in the
®rst hydration shell, or by forming water-mediated
interactions. Most commonly, ligands establish
both types of contacts with nucleic acids.
Crystallography is the method of choice for
studying the primary hydration shell of nucleic
acids and several reviews have surveyed the ®eld
# 2000 Academic Press
1114
(Westhof, 1988, 1993; Westhof et al., 1988; Westhof
& Beveridge, 1990; Berman, 1991, 1994; Feig &
Pettitt, 1998a; Auf®nger & Westhof, 1999). The
shapes of the hydration shells around DNA bases
(Schneider & Berman, 1995), DNA phosphate
groups (Schneider et al., 1998) and RNA CˆG,
A-U, GU base-pairs (Auf®nger & Westhof, 1998c)
have been investigated on the basis of statistical
analysis of high-resolution crystallographic structures. Yet, the main characteristics of the hydration
shells obtained by crystallographic methods are
not always completely determined, and X-ray techniques provide only limited information on the
dynamics of the bound water molecules. In
addition, the hydrogen bond patterns formed
between the solvent and the solute molecules have
often to be guessed, since hydrogen atoms are not
observable. Molecular dynamical (MD) simulations
allow the completion of the structural and dynamical information gained by experimental methods.
Several recent MD studies have been devoted to
the characterization of the properties of the
hydration shells around RNA and DNA structures
(Auf®nger & Westhof, 1997a,b; Cheatham &
Kollman, 1997; Duan et al., 1997; Young et al.,
1997b; Cheatham et al., 1998; Feig & Pettitt, 1999a).
Hydration patterns were often described at the
level of the global nucleic acid structure and many
heterogeneous sequences have been investigated
with the principal aim of characterizing the minor
groove ``spine of hydration'' ®rst observed for the
d(CGCGAATTCGCG)2 dodecamer (Kopka et al.,
1983). Yet, the understanding of the physicochemical basis for changes in hydration induced by the
addition or removal of a single 20 -OH group and
by the U $ T modi®cation in the structural and
functional properties of nucleic acids is of great
importance (Wang & Kool, 1995; Norberg &
Nilsson, 1996; Cheatham & Kollman, 1997). Such
an understanding should provide useful hints to
recognition and folding problems as well as explanations to the fact that, for a similar sequence,
RNA duplexes are thermodynamically more stable
than DNA duplexes (Lesnik & Freier, 1995; Conte
et al., 1997; Kankia & Marky, 1999).
In an earlier study, we described the preferred
orientations and the hydration properties of the 20 OH group (Auf®nger & Westhof, 1997b). The present study focuses on the characterization of the
similarities and differences in the hydration and
ion-binding properties of regular RNA and DNA
duplexes. To this end, two 2.4 ns MD simulations
of two regular helices in an aqueous environment
with 0.25 M KCl added, derived from the Arnott
®ber models (Arnott et al., 1976), the A-form RNA
r(CG)12 duplex and the B-form DNA d(CG)12
duplex, were generated (Figure 1). The choice of
regular sequences containing CˆG pairs may
appear trivial compared to ``more biologically relevant'' structures. Yet, Watson-Crick base-pairs are
the single most important secondary structural
motif in nucleic acids. Short strands of alternating
CˆG pairs can be found in numerous RNA and
Nucleic Acid Water and Ion-binding Features
DNA sequences, and the grooves of CˆG pairs are
involved in a great variety of intermolecular contacts in regular triple helices, base triples, and complexes formed between protein or drugs and
nucleic acids. From a computational point of view,
the aims of selecting sequences containing a regular repeat of CˆG base-pairs are multiple. First, the
comparison of such sequences involves only
chemical changes associated with the presence or
absence of the 20 -OH group and do not consider
the modi®cations associated with the presence of U
(RNA) or T (DNA) bases. Second, among nucleic
acid duplexes, those containing only CˆG basepairs are thermodynamically the most stable
(SantaLucia, 1998; Xia et al., 1998) and, consequently, are supposed to be structurally less ¯exible than A-U or A-T-containing sequences, a fact
that therefore, should limit problems associated
with the sampling of the conformational space.
Third, for a regular sequence, it is possible to calculate average values for various properties of the
CˆG base-pairs in a more precise way than by
analyzing sequences where CˆG pairs are located
in a heterogeneous environment. This is crucial for
investigating dynamical features that need long
equilibration times, such as those related to the
ionic atmosphere. The use of regular or semi-regular (Feig & Pettitt, 1998b, 1999a) sequences is similar in spirit to applications of the multiple
molecular dynamics (MMD) method that have
been designed to extend the sampling of the conformational space of speci®c structural motifs, a
recurrent problem in MD simulations (Auf®nger
et al., 1995; Auf®nger & Westhof, 1997a; Caves
et al., 1998; Worth et al., 1998).
Besides the structural aspects related to the
nucleic acid hydration shells, the question of
whether and where ions may bind to nucleic acids
is of great importance, as it is well appreciated that
ion coordination sites are potential binding sites
for charged groups of proteins or drugs (Draper &
Misra, 1998; Hermann & Westhof, 1998). It has
long been stated that the primary hydration shell
of nucleic acids is impermeable to ions and,
indeed, only a limited number of divalent or
monovalent ions in direct contact with the nucleic
acids have been experimentally observed. MD
simulations have challenged this view, since it
has been reported that direct contacts between
monovalent ions and nucleic acids could occur in
RNA and DNA grooves (Mohan et al., 1993;
Weerasinghe et al., 1995b; Auf®nger & Westhof,
1997a; Cheatham & Kollman, 1997; Young et al.,
1997a,b; Lyubartsev & Laaksonen, 1998; Feig &
Pettitt, 1999b). These observations have been
rationalized and several binding sites have been
proposed in both DNA grooves depending on the
base-pair sequences (Young et al., 1997a). From the
experimental side, recent crystallographic data
have con®rmed that monovalent ions may partially
intrude into the nucleic acid grooves (Shui et al.,
1998; Tereshko et al., 1999), while NMR evidence
for NH‡
4 (Hud et al., 1999) and other monovalent
Nucleic Acid Water and Ion-binding Features
1115
Figure 1. Canonical A-form structure of the r(CG)12 helix (left) and canonical B-form structure of the d(CG)12 helix
(right) in their simulation box. Middle, sequence of the RNA and DNA duplexes analyzed in this study. Only the
central 12 base-pairs (in yellow) were used for the determination of the hydration and ion-coordination sites. Each
box is ®lled with 72 K‡ (in blue) and 26 Clÿ (in green). For clarity, the water molecules ®lling the simulation boxes
are not displayed.
ion groove binding (Denisov & Halle, 2000) has
been reported. Further, well-characterized K‡
(Basu et al., 1998) and Na‡ (Su et al., 1999) binding
pockets have been described in RNA crystal structures. Thus, despite some discussions concerning
the signi®cance of effects related to ion-binding
events (Chiu et al., 1999; McFail-Isom et al., 1999;
Beveridge & McConnell, 2000), there is little doubt
that some hydration sites localized in the ®rst
coordination sphere of nucleic acids are partially
occupied by monovalent ions. In the present MD
study, potassium ions were considered instead of
the more common sodium ions. The main reasons
for choosing potassium ions are, ®rst, that they
represent one of the dominant ionic species in the
intracellular medium and, secondly, that it has
been demonstrated that potassium ions are
required in some RNA molecules for structural
and functional integrity (Wang et al., 1993; Gluick
et al., 1997; Basu et al., 1998; Draper & Misra, 1998).
Another aim of this study is to provide a structural basis for the observed difference in gel electrophoretic mobility between RNA and DNA
duplexes (Ratmeyer et al., 1994; Lesnik & Frejer,
1995; Bonifacio et al., 1997; Hagerman, 1997).
Despite the fact that both types of molecules display the same net charge, RNA duplexes migrate
more slowly than DNA duplexes under various
ionic conditions. This behavior has been attributed
to different ion distributions around A-form and Bform duplexes (Bonifacio et al., 1997). However, the
fact that RNA molecules bear an additional 20 -OH
group compared to DNA molecules and are, therefore, supposed to be more hydrated (Egli et al.,
1996) could explain as well the observed differences in mobility. The data on the hydration and
ionic atmosphere of the RNA and DNA duplexes
provided by the MD simulations should help in
unraveling the origins of this behavior.
1116
Results
Global structural features
The r(CG)12 and the d(CG)12 alternating oligomers retain their overall structural characteristics
during the 2.4 ns trajectories. No single base-pair
opening event was observed during the simulations of either oligomer. For the RNA duplex, the
RMS deviation pro®les and the snapshots extracted
from the MD trajectories show that the structures
remain close to the initial A-form shape, while they
Ê from
stay constantly at the high RMS value of 6 A
Nucleic Acid Water and Ion-binding Features
a hypothetical B-form RNA structure (Figure 2).
The superposition, at the base-pair level, of the
initial and the average structure results in a low
Ê , despite a 1.0 A
Ê
RMS deviation value of 0.5 A
Ê interphosphate disincrease of the initial 18.0 A
tance (Figure 3). The sugar puckers stay in the C30 endo domain with low angle and amplitude variations. The average value of the pseudorotation
phase angle P (15 ) is close to the canonical C30 endo value (18 ).
On the contrary, although no clear central hole
characteristic of an A-form DNA can be seen in the
Figure 2. RMS deviations and snapshots extracted from the MD simulations of the r(CG)12 and d(CG)12 duplexes.
Top, RMS deviation pro®les from the starting structures. The red lines correspond to the complete 24 base-pair structures and the black lines to the central 12 base-pair segments. Additionally, for the RNA and the DNA simulations,
RMS deviations with respect to a B-form and an A-form structure are given, respectively (the RNA B-form structure
is ``hypothetical''). The horizontal line marks the RMS deviations between the starting A and B-form structures. The
time during which positional constraints have been applied is emphasized by a red gradation. The end of the equilibration period is marked by a broken line. Bottom, superposition of 20 structures covering 400 ps of MD extracted
from the beginning and the end of the RNA (left) and DNA (right) MD simulations. While the A-RNA structures
remain essentially of the A-type, the B-DNA moves towards a form intermediate between A and B.
Nucleic Acid Water and Ion-binding Features
1117
Figure 3. Structural and dynamical characteristics of the r(CG)12 and d(CG)12 structures extracted from the last 1.5
ns of the 2.4 ns MD trajectories. Top, interphosphate distance pro®les. The base-pair steps are identi®ed on the
abscissa scale. The bold lines give the average values; the vertical lines indicate one standard deviation around the
average values; the thin lines outline the minimum and maximum values for each interphosphate distance; the broken arrows mark the interphosphate distance from the starting structure derived from the ®ber model (Arnott et al.,
1976). Middle, superposition of the starting structure of an RNA (left) and a DNA (right) base-pair (in black) with the
average structure calculated for the central 12 base-pairs over the last 1.5 ns of the MD trajectories (in red). Bottom, a
histogram of the pseudorotation phase angle P calculated for the 48 sugar molecules of the RNA and DNA duplexes
over the last 1.5 ns of the trajectories. The arrows indicate the theoretical values of P for the main conformational
domains. The inset gives the tm values. The mean tm values are marked by a broken line.
overlay of MD structures (Figure 2), it appears that
the ®nal structures are intermediate between an
A-form and a B-form DNA, as further indicated by
the convergence of the RMS deviation pro®les
(MacKerell & Banavali, 2000) calculated with
respect either to the initial B-form structure or to a
standard A-form helix (Figure 2). The calculated
RMS deviation at the base-pair level is larger for
Ê ) compared to the RNA
the DNA structure (0.7 A
Ê ) and the interphosphate distance
structure (0.5 A
Ê (Figure 3). The
increases on the average by 1.3 A
DNA sugar puckers remain essentially in the C20 endo domain with some incursions in the O40 -endo
domain and very few switches to the C30 -endo
domain. The mean value of the pseudorotation
phase angle P for the sugar population centered on
the C20 -endo domain is close to 145 , which is less
than the canonical value (162 ), and is characteristic of the Cornell et al. force-®eld (Cheatham et al.,
1999). The histogram representing the tm values is
broader for DNA than for RNA, re¯ecting the larger mobility of the sugars. The mean value is 2 1118
lower (41.5 for RNA compared to 39.5 for
DNA). Close to 180 backbone dihedral angle transitions (2 transitions residueÿ1 nsÿ1) involving 35
residues were observed for DNA, while for RNA,
only 12 backbone dihedral angle transitions (0.2
transitions residueÿ1 nsÿ1) involving ®ve residues
among the 48 of the duplex were noted during the
last 1.5 ns of simulation. The interphosphate
distance displays a larger range of variability
for DNA than for RNA (Figure 3) but, similarly
Ê than the initial interto RNA, is larger by 1 A
phosphate distance derived from the ®ber model
structure.
The well-behaved structural characteristics displayed by the two duplexes along the MD trajectories allow for an in-depth analysis of their ionbinding and hydration properties. In the following,
we focus our analysis on the central 12 base-pairs
of each duplex (Figure 1) in order to avoid the
inclusion of possible end-effects in the statistics.
Ion-binding sites
Direct ion-coordination sites
At the lowest level of ionic densities, several ion
contacts are observed around the CˆG pairs and
the backbone atoms (Figure 4). However, all the
contacts do not lead to ordered coordination sites.
At a high contour level, densities are visible only
Nucleic Acid Water and Ion-binding Features
around the O6 atom for RNA and around the O6
and N7 atoms for DNA. Thus, while potassium
ions contact, at least episodically, each accessible
electronegative atom, coordination sites are
observed only in the deep/major grooves. On the
average, 0.2 K‡ (RNA) and 0.3 K‡ (DNA) form
contacts with the O6 atoms while 0.1 K‡ are
located close to the N7 atoms (Table 1). This leads
to approximately 0.2 (RNA) to 0.3 K‡ (DNA) in
contact with the electronegative base atoms.
Further, 0.3 K‡ are located close to the phosphate
groups. Thus, 0.5 (RNA) to 0.6 K‡ (DNA) are in
direct contact with one CˆG pair or, on average,
about one K‡ per base-paired GpC dinucleotide
step, whatever the nucleic acid type. The RMS
deviations of these values are between 10 % (RNA)
and 20 % (DNA), which are indicative of large ¯uctuations (Table 1) resulting from the fact that only
certain binding sites are partially occupied during
the MD simulation. However, the data indicate
unambiguously that the deep/major groove of
these duplexes must be considered as an important
ion-binding region.
The K‡ in contact with one O6 atom is generally
also in contact with the O6 atom of an adjacent
base-pair forming an ``ion bridge''. Very interestingly, ion bridges are essentially observed for
27 % (RNA) and 40 % (DNA) of the GpC steps,
while almost no occurrence is observed for CpG
Figure 4. Ion densities calculated
around RNA (top) and DNA (bottom) CˆG steps. Low-level densities (contact sites) are shown in
purple and high-level densities
(strong binding sites) are shown in
red. On average, there is respectively 0.55 and 0.59 K‡ (Table 1) per
RNA and DNA base-pair (purple
density), essentially located in the
deep/major grooves and around
the anionic phosphate oxygen
atoms.
1119
Nucleic Acid Water and Ion-binding Features
Figure 5. Snapshots extracted
from the MD simulations showing
the GpC versus CpG ion-binding
features. Left, on the average, for
GPC steps, 27 % (RNA) and 40 %
(DNA) of K‡ (in yellow) were
involved in bridges between the
(G)O6 atoms (in red). Right, for
CpG steps, almost no K‡ has been
observed to form bridges between
the (G)O6 atoms during the MD
simulations. This may be due to
the electropositive cytosine amino
groups, which protrude into the
major groove of CpG compared to
GpC steps.
steps (Figure 5). This behavior may be due to the
different geometrical arrangement of electronegative atoms lining the deep/major grooves.
Although, in all four cases shown in Figure 5, the
distance between two adjacent G(O6) atoms seems
ideal to bind a K‡, the protruding of the two electropositive cytosine amino groups for CpG steps
prevents the binding of K‡ into the deep/major
groves. Thus, the MD simulations allow characterization of an important sequence effect related to
ion binding.
Ion-coordination dynamics
Given the reduced number of ionic contacts to
the duplexes, the residence times of the ions were
not calculated with respect to a particular coordination site. Rather, the residence times of the ions
in the ®rst coordination shell of the central 12 basepairs of the RNA and DNA duplexes were evaluated. The highest calculated residence times of K‡
in the ®rst coordination shell of the central 12 basepairs approach or even exceed the nanosecond
range (Figure 6) and for each sequence of 12 basepairs, at least three ions are present with residence
times exceeding 500 ps. The remaining ions display
much shorter residence times. It has to be noted
that these data are of qualitative character,
especially with regard to the ions that display the
longest residence times. However, they indicate
that residence times in the nanosecond range can
occur in the deep/major groove of these duplexes,
especially for GpC steps. The bridging position of
the ions between two adjacent O6 atoms is surely
an important sequence-dependent effect (Figure 5).
No chloride ion is observed in the ®rst coordination shells of the duplexes, as was reported for
simulations conducted at a 1.0 M NaCl concentration (Feig & Pettitt, 1999b). However, several
chloride ions do penetrate the second coordination
shell of the double helices by forming ion pairs
with positive ions.
Hydration
Water-coordination sites
The RNA and DNA base-pairs are each surrounded by 22 hydration sites (Figure 7) occupied
by an average of 20.8 and 19.2 water molecules (or
21.4 and 19.8 water molecules and K‡), respectively (Table 2). Thus, approximately two additional
water molecules are found in the RNA primary
hydration shell. At a low density level that represents all the possible contact sites (in Figure 7,
the densities due to both the K‡ and the water
molecules are presented), it is apparent that the
CˆG base-pairs are completely surrounded by
Table 1. Average number of K‡/RNA and K‡/DNA contacts calculated over the last 1.5 ns of MD for the 12 central
base-pairs
Deep/major groove
Shallow/minor groove
(G)N7 (G)O6 (C)H42 (G)N3 (G)H22 (C)O2
K‡ (RNA)
‡
K (DNA)
0.06
0.18
0.11
(0.09) (0.06) (>0.04)
0.14
0.27
0.13
(0.20) (0.20) (0.10)
0
0
0
0
0
0
O20
OR
Backbone and sugar
OS
O30
0.03
0.07
0.09
0
(0.02) (>0.02) (>0.05)
/
0.04
0.09
0.05
(0.03) (>0.07) (0.06)
O40
O50
0
0
0
0
Basepair
All
atoms
0.23
0.55
(>0.04) (0.05)
0.31
0.59
(0.25) (>0.13)
Numbers in parentheses correspond to standard deviations from average values, calculated over four non-overlapping blocks of
three consecutive base-pairs.
1120
Nucleic Acid Water and Ion-binding Features
Figure 6. Number of K‡ present in the ®rst coordination shell of the RNA and DNA duplexes with respect
to their residence times (obviously, the statistics are the
poorest for the ions displaying the longest residence
times).
water molecules, except evidently on the surfaces
where stacking contacts occur and on the 50 and 30 side of the backbone. At a higher density level, the
hydration sites appear much less ordered around
DNA than around RNA steps, particularly around
the more mobile backbone regions. The hydration
sites around the base-pairs are, in each case, relatively well de®ned. For both helices, three sites are
found in the deep/major groove in the vicinity of
N7, O6, and N4, and three sites are found in the
shallow/minor grooves in the vicinity of N3, N2
and O2 atoms. Additionally, for RNA but not for
DNA, a hydration site in the proximity of the C5H group (RNA site 10, Figure 7) and shared with
the OR atom of the attached phosphate group can
be recognized (Auf®nger & Westhof, 1998c). Thus,
close to ®ve water molecules occupying six
hydration sites are found in the primary hydration
shell of a CˆG base-pair (Table 2). The hydration
cones, formed by three water molecules around
each anionic oxygen atom of the phosphate
groups, are easily recognizable in the high-level
densities calculated for the RNA molecule.
Although present in the DNA oligomers, these
hydration cones are less well de®ned (see the medium-level densities, Figure 7). Hydration cones are
also characteristic of the 20 -OH groups of RNA
(Figure 7) and account for the presence of an
additional hydration site in the shallow groove
compared to the DNA minor groove. For DNA,
the absence of these hydration sites is balanced by
the presence of a hydration site in regard to the
O40 atoms (DNA sites 16 and 22, Figure 7).
The hydrogen bonding percentages (HB%) give
a ®ner picture of the contacts established between
the solute and the hydration shell (Table 2). They
are de®ned as the total number of hydrogen bonds
established during a single trajectory between a
particular solute atom and the surrounding water
molecules divided by the total number of con®gurations analyzed (Auf®nger & Westhof, 1997a). It
appears that in the deep/major grooves (G)N7
atoms establish good hydrogen bond contacts
(90 %), followed by the (C)N4 amino groups
(80 %). The HB% value close to 60 for the DNA
(G)O6 atoms, associated with a high standard deviation value, results from the fact that the O6 atoms
are also a principal ion-binding site. In the shallow/minor grooves, the (G)N2 amino group is
poorly hydrated compared to its neighboring
atoms, as re¯ected by the size of its associated density (see Figure 7). Interestingly, the RNA (G)N3
atom is less well hydrated than the corresponding
DNA atom, despite the presence of the 20 -OH
group. The 20 -OH appears to be a relatively poor
donor and acceptor group, probably due to its
rotational dynamics (Auf®nger & Westhof, 1997b).
Table 2. Average number of H2O/RNA and H2O/DNA contacts as well as hydrogen bonding percentages (HB%)
calculated over the last 1.5 ns of MD and the 12 central base-pairs (note that, except in the last column of the Table,
potassium ion shave not been taken into account)
Deep/major groove
Shallow/minor groove
(G)N7 (G)O6 (C)H42 (G)N3 (G)H22 (C)O2
A. Average contacts
H2O (RNA)
1.1
1.0
(>0.1)a (0.1)
H2O (DNA)
1.0
0.9
(0.1) (0.2)
1.3
(>0.1)
1.2
(>0.1)
B. Hydrogen bonding percentages (HB%)
91
71
75
H2O (RNA)
(15)b
(21)
(3)
86
58
79
H2O (DNA)
(23)
(37)
(5)
Backbone and sugar
O20
OR
OS
O30
O40
O50
Basepair
All
atoms
5.3
(0.1)
5.2
(0.2)
20.9c
(21.3)d
19.2c
(19.8)d
0.8
(>0.1)
0.9
(>0.1)
1.5
(>0.1)
1.4
(>0.1)
0.9
(>0.1)
0.9
(0.1)
2.6
(>0.1)
/
/
2.9
(>0.1)
3.4
(>0.1)
3.5
(>0.1)
3.4
(>0.1)
1.3
(>0.1)
1.1
(0.1)
0.9
(>0.1)
0.9
(>0.1)
0.7
(>0.1)
0.9
(>0.1)
56
(5)
81
(3)
79
(3)
58
(11)
83
(3)
94
(5)
88/46
(6)/(5)
/
/
279
(6)
286
(7)
287
(12)
280
(10)
36
(4)
52
(13)
14
(2)
36
(19)
18
(3)
30
(6)
a
Numbers in parentheses correspond to standard deviations from average values calculated over four non-overlapping blocks of
three consecutive base-pairs.
b
Numbers in parentheses correspond to standard deviations from average values.
c
These numbers correspond to the exact number of water molecules surrounding a CˆG base-pair (potassium ions have not been
counted).
d
Numbers in parentheses correspond to the exact number of water molecules and K‡ surrounding a CˆG base-pair.
Nucleic Acid Water and Ion-binding Features
1121
Figure 7. Water and ion densities showing the 22 hydration sites surrounding the RNA (top) and DNA (middle
and bottom) CˆG base-pairs (note that the K‡ are considered as water molecules, see also Figure 9). Low-level densities (contact sites) are shown in blue (left) and high-level densities are shown in yellow (left and right). For DNA,
the high-density patterns (middle) are comparable to those drawn for RNA (top). The medium-density patterns for
DNA (bottom) have been drawn in order to distinguish the 22 hydration sites surrounding the CˆG base-pairs.
Potential hydrogen bonds between hydrophilic groups and potential C-H. . .O hydrogen bonds are represented by
broken lines (right).
The anionic oxygen atoms of the phosphate groups
satisfy each of their three hydrogen bond acceptor
sites, while the O30 , O40 , and O50 atoms are poorly
hydrated, especially in RNA, where they are
involved in C6/8-H. . .O50 and C20 -H20 . . .O40 contacts (Auf®nger & Westhof, 1997b, 1998c; Wahl &
Sundaralingam, 1997).
Water-coordination dynamics
Water molecules establish long-lived hydrogen
bond contacts with the RNA and DNA anionic
oxygen OR and OS atoms. For both molecules, the
characteristic pro®les are slightly different. The
longest residence times are observed for the
(RNA)OR atoms (700 ps) followed by the
(RNA)OS (500 ps) atoms (Figure 8), while the
residence times for the (DNA)OR and (DNA)OS
atoms are comparable (400 ps). Long residence
times are observed for the (DNA)O40 atoms, while
signi®cantly shorter residence times are observed
for the O30 and O50 backbone atoms, and the RNA
20 -OH groups. No marked difference is seen for the
RNA and DNA atoms located in the deep/major
grooves. Surprisingly, in the shallow/minor
grooves, the residence times are much longer in
the vicinity of the (DNA)N3 (700 ps, site 20,
Figure 7) and (DNA)O2 (up to 1500 ps, site 18,
Figure 7) atoms. These long-lived water molecules
are not associated with ions, as none of them is
observed in the shallow/deep grooves.
1122
Nucleic Acid Water and Ion-binding Features
Figure 8. Number of water molecules bound to speci®c hydrophilic atoms of the 12 central CˆG
base-pairs of the RNA and DNA
duplexes as a function of their
hydrogen bonding contact times
(HB%, see Table 2).
Long-lived water bridges
It is useful to characterize the water molecules
involved in long-lived water bridges, since those
water molecules are susceptible to having the largest structural impact (Westhof, 1993). In the RNA
duplex, 17 water molecules are observed to bridge
OR anionic oxygen atoms of adjacent phosphate
groups with 300 to 400 ps lifetimes (sites 6 and 10,
Figure 7). Thus, such long-lived hydration patterns
are observed during 15 to 20 % of the simulation.
A large number of water bridges with signi®cantly
shorter lifetimes is also observed. For the B-form
DNA duplex, no long-lived water bridge is
found between the major groove phosphate
groups because the increased phosphate-phosphate
interatomic distance does not allow the formation
of such hydration patterns (Saenger et al., 1986).
However, the exceptionally long residence
times of the water molecules associated with the
minor groove O2 and, to a lesser extent, N3
atoms (Figure 8) are associated with the
formation of (C)O2(n). . .W. . .O40 (n ‡ 1) and
(G)N3(n). . .W. . .O40 (n ‡ 1) bridges. Five (C)O2(n)
. . .W. . .O40 (n ‡ 1) water bridges with lifetimes ran-
Nucleic Acid Water and Ion-binding Features
ging between 300 and 650 ps and one
(G)N3(n). . .W. . .O40 (n ‡ 1) water bridge with a
600 ps lifetime have been observed.
Calculated ion densities
It has been noted above that the deep/major
groove O6 and N7 atoms are the main K‡ coordination sites. Obviously, water molecules also
occupy these hydration sites and, therefore, the
densities shown in Figure 7 result from the cumulative densities due to the presence of water molecules and K‡ in the vicinity of O6 and N7. In
order to visualize the contribution of the potassium
ions to the patterns shown in Figure 7, two types
of density pro®les have been calculated. The ®rst
(Figure 9, in yellow) considers all the atoms in contact with the CˆG pairs as oxygen atoms of water
molecules (Z ˆ 8), the second (Figure 9, in red)
takes into account the presence of K‡ (Z ˆ 18) in
the vicinity of O6 and N7. It appears that the calculated densities increase signi®cantly in the vicinity
of O6 for the second model.
Discussion
RNA duplexes are more rigid than
DNA duplexes
A large number of MD simulations have been
carried out on DNA duplexes using the particle
Figure 9. Partial occupancy of hydration sites by K‡
calculated for RNA (top) and DNA (bottom) base-pairs.
The yellow densities have been calculated by considering all the atoms as water molecules with Z ˆ 8 (see
also Figure 7). The red densities take into account the
calculated partial occupancy of the hydration sites by
K‡ (Z ˆ 18). The O6 sites for which this difference is the
most signi®cant are marked by an arrow.
1123
mesh Ewald (PME) summation method (Zichi,
1995; Cheatham & Kollman, 1997; Duan et al.,
1997; Young et al., 1997b; Cheatham et al., 1998;
Feig & Pettitt, 1998b; Flatters & Lavery, 1998;
Lyubartsev & Laaksonen, 1998; Winger et al., 1998;
Young & Beveridge, 1998; Sprous et al., 1999;
Bevan et al., 2000; MacKerell & Banavali, 2000;
Trantirek et al., 2000), while only two studies concerned RNA duplexes (Cheatham & Kollman,
1997; MacKerell & Banavali, 2000). In agreement
with 31P NMR experiments (Shindo et al., 1985),
Cheatham & Kollman (1997) emphasized the greater rigidity of RNA over DNA duplexes, as re¯ected
by lower RMS deviations from the starting structure, little sugar repuckering and no correlated BI
to BII backbone transitions. This tendency is con®rmed by our simulations, which show that the
r(CG)12 duplex remains very close to standard
A-RNA structures.
For DNA duplexes, force-®eld dependency
effects have been reported. For simulations using
an older version of the CHARMM force-®eld
(MacKerell et al., 1995), a transition from a B-DNA
to an A-DNA form was observed (Yang & Pettitt,
1996), while for simulations performed with the
AMBER force-®eld (Cornell et al., 1995), A to B
transitions were described (Cheatham & Kollman,
1996; Cieplak et al., 1997), leading to the conclusion
that there is a force-®eld-dependent DNA polymorphism (Auf®nger & Westhof, 1998d). In
addition, sequence effects, as discussed by Feig &
Pettitt (1998b), may play a signi®cant role, and a
certain level of structural heterogeneity is expected.
Our simulation of the d(CG)12 sequence reveals a
drift toward a form intermediate between A and B,
a rather unexpected observation for simulations
performed with the AMBER force-®eld, although
experimentally it is known that CˆG-containing
sequences are rather A-philic compared to A-Tcontaining sequences (Pilet & Brahms, 1972; Arnott
& Selsing, 1974; Saenger, 1984; Minchenkova et al.,
1986; Ng et al., 2000; Trantirek et al., 2000). From
the theoretical perspective, poly(dG)-poly(dC)
sequences were observed to be A-philic, while
poly(dA)-poly(dT) sequences were observed to be
A-phobic (Cheatham et al., 1998). Thus, the present
MD simulations of the r(CG)12 and d(CG)12 structures display a reasonable agreement with our current knowledge of the structure and dynamics of
these A and B-form duplexes, and exemplify the
experimentally well-documented greater rigidity of
RNA compared to DNA duplexes (Hagerman,
1998, 1997). Thus, the AMBER force-®eld appears,
at least partially, to be successful in reproducing
several of the sequence-dependent dynamical
properties of these duplexes, and the results are in
good agreement with MD simulations performed
with the CHARMM27 force-®eld (Foloppe &
MacKerell, 2000; MacKerell & Banavali, 2000).
1124
Hydration sites: comparison between
experimental and simulation data
The results of the present simulations are in
good agreement with studies based on statistical
analysis of crystallographic structures at the basepair level. Schneider & Berman (1995) characterized several well-de®ned hydration sites in the
vicinity of the four A, G, C, and T bases for DNA
duplexes in the B-form. These sites (three in the
major groove and three in the minor groove associated with a small density in the vicinity of the
(C)C6 atom) are all well de®ned in this study
(Figure 7). In agreement with the present data, the
smallest experimentally characterized and most
labile hydration site is located in the middle of the
minor groove in front of the (G)N2 amino group.
The backbone O30 and O50 ester oxygen atoms are
not well hydrated, while O40 shares a hydration
site with the (G)N3 or (C)O2 minor groove atoms
(Schneider et al., 1998). The cones of hydration
around each anionic oxygen atom of the phosphate
groups (Schneider et al., 1998) are visible only at
low-density contour levels indicating that, as a
result of the high mobility of the DNA backbone,
these sites are not especially well ordered. Only
crystallographic studies conducted at a low temperature, which leads to a signi®cant reduction of
the atomic mobility, allow detection of precise
water peaks around the DNA phosphate groups
(Kopka et al., 1983). Given the overall lower mobility of the RNA structure, the hydration sites
located around the bases and the backbone atoms
are much better de®ned than those around DNA
(Figure 7). The eight hydration sites surrounding
the CˆG base-pair are clearly identi®able and
match very well with those derived from an analysis of high-resolution crystallographic structures
(see Figure 1, Auf®nger & Westhof, 1998c). The
slight difference observed for the O6 site in the
deep groove is probably due to the large number
of heterogeneous base-pair steps present in the
experimental sample re¯ecting possible sequencespeci®c dependence. The hydration cones around
the 20 -OH groups and observed in earlier studies
(Egli et al., 1996; Auf®nger & Westhof, 1997b) are
also well de®ned.
In contrast to crystallography, NMR can provide
information about the residence times of speci®c
water molecules (Karplus & Faerman, 1994;
Kochoyan & Leroy, 1995; Otting & Liepinsh, 1995).
Water molecules have essentially been detected in
the minor groove of DNA duplexes containing
ApT tracts (Kubinec & Wemmer, 1992; Liepinsh
et al., 1992, 1994; Jacobson et al., 1996; Denissov
et al., 1997; JoÂhannesson & Halle, 1998; Phan et al.,
1999). Their residence time, involving signi®cant
sequence-dependent effects (Liepinsh et al., 1994;
Phan et al., 1999), was estimated to exceed the
nanosecond at 4 C, reach 0.6 ns at 10 C, and 0.2
ns at 27 C (Denissov et al., 1997; Phan et al., 1999).
In the major groove, the observed residence times,
in the 200 to 500 ps range at 10 C, are associated
Nucleic Acid Water and Ion-binding Features
with water molecules close to the methyl group of
thymine bases not present in the investigated
d(CG)12 sequence (Denissov et al., 1997; Lane et al.,
1997; Phan et al., 1999), while water molecules contacting the phosphate groups are relatively labile.
Given the differences in the investigated sequences,
the present MD results are in qualitative agreement
with these NMR data. Further, the RNA shallow
groove, despite its increased width, was found by
NMR as well as in the present simulations, to be
more hydrated than the DNA minor groove of an
analogous DNA sequence (Conte et al., 1996; Gyi
et al., 1998a). This was attributed to the presence of
the RNA 20 -OH groups.
From the theoretical perspective, several MD studies using the Ewald summation method for the
treatment of the long-range electrostatic interactions focussed on the structure of the hydration
shell of nucleic acids (for reviews on MD studies,
see Westhof & Beveridge, 1990; Auf®nger &
Westhof, 1998d; Beveridge & McConnell, 2000).
For a DNA duplex, Duan et al. (1997) distinguished
three hydration sites in the major groove and three
hydration sites in the minor groove of CˆG basepairs, in agreement with those shown in Figure 7.
The highest density of water molecules was located
in the minor groove, where a spine of hydration is
observed. A lower level of hydration was observed
in the major groove, revealing a larger mobility
and, consequently, reduced residence times for
these water molecules (Cheatham & Kollman,
1997; Young et al., 1997b). The most detailed study
on the hydration of a DNA duplex has been performed by Feig & Pettitt (1999a). These authors
reported an overall good agreement between their
simulation data and the hydration sites characterized by Schneider & Berman (1995). However, as
in the present study, some discrepancies were
noted, especially in the vicinity of the (G)N2
group, and these were attributed to limitations in
the resolution of water sites determined by crystallography (Feig & Pettitt, 1998a). In previous work,
water molecules with residence times exceeding
500 ps in the vicinity of a tRNA hairpin were
described (Auf®nger & Westhof, 1997a). Some of
these stable hydration patterns were associated
with the presence of modi®ed bases (Auf®nger &
Westhof, 1998a). Yet, most of these water molecules were found in the vicinity of the anionic
oxygen atoms of the phosphate groups, in agreement with the present data. Noteworthy, the longer data collection time of the present study (1.5 ns),
compared to the preceding 500 ps time-scale,
allows for a more precise estimate of the water
molecule residence times.
Is RNA more hydrated than DNA?
From Figure 7, it is apparent that the hydration
patterns around the A-form RNA duplex are better
de®ned than those around the B-form DNA
duplex. In all, 22 hydration sites surround the
RNA and DNA CˆG base-pairs accommodating
1125
Nucleic Acid Water and Ion-binding Features
20.8 (RNA) and 19.2 (DNA) water molecules as
well as 0.5 and 0.6 K‡, respectively (Tables 1 and
2). These values are in good agreement with experimental results giving values in the range of 15-20
water molecules per base-pair (Saenger, 1984).
However, derived from MD investigations, Feig &
Pettitt (1999a) estimated the number of water molecules in the ®rst hydration shell of CˆG pairs to
be close to 19.3 (A-DNA) and 20.6 (B-DNA), exemplifying the fact that the evaluations of the number
of water molecules present in a hydration shell
is slightly dependent upon the algorithm used to
calculate them (MacKerell & Banavali, 2000).
Yet, despite an identical number of hydration
sites, an excess of close to two water molecules is
calculated in favor of the RNA structure, allowing
us to state that RNA duplexes are more hydrated
than DNA duplexes. This value results from a
combination of various factors. Among them, the
RNA and DNA duplexes adopt different forms (A
and B), which modulate their ability to interact
with the solvent. The principle of ``economy of
hydration'' (Saenger et al., 1986), which indicates
that the more compact A-DNA form needs less
hydration than the B-DNA form, cannot be applied
in the present case because of the presence of the
RNA 20 -OH group. Interestingly, by using the
results presented by Feig & Pettitt (1999a), it is
possible to propose the rough hydration scale for
CˆG base-pairs shown in Table 3. Thus, an
estimated difference of three water molecules per
CˆG base-pair between the two RNA and DNA
A-form duplexes can be attributed to the presence
of the 20 -hydroxyl group. This scale may be useful
in the interpretation of experimental data
for which precise structural information is not
available.
Short-lived and long-lived ion-binding sites
The main ion-binding sites for the selected RNA
and DNA sequences are located on the Hoogsteen
edges, close to the (G)O6 and (G)N7 electronegative sites. The occurrence of such ion-binding sites
in electronegative pockets in the nucleic acid
grooves was proposed earlier on the basis of the
observation of the intrusion of a sodium ion into a
DNA minor groove (Young et al., 1997a) and is
con®rmed by the present simulations and those
from other groups for RNA (Auf®nger & Westhof,
1997a; Cheatham & Kollman, 1997), for DNA
Table 3. Estimated difference of water molecules in the
®rst hydration shell of A-DNA, B-DNA and A-RNA
nucleic acid (CG)12 duplexes
A-DNA
B-DNA
A-RNA
A-DNA
B-DNA
A-RNA
/
/
/
‡1 H2O
/
/
‡3 H2O
‡2 H2O
/
(Cheatham & Kollman, 1997; Young et al., 1997b;
Lyubartsev & Laaksonen, 1998; Feig & Pettitt,
1999b), and for DNA triple helices (Mohan et al.,
1993; Weerasinghe et al., 1995a). Our observations,
which indicate the occurrence of an ion bridge
between the two deep/major groove (G)O6 atoms
of adjacent base-pairs, are consistent with the
model proposed by Young et al. (1997a). In
addition, the present data demonstrate a signi®cant
deep/major groove ion-binding dependence, as
almost all observed ion bridges involve GpC and
not CpG steps (Figure 5). This ®nding may be of
importance for binding studies trying to understand the af®nities of speci®c amino acids or other
chemical groups to particular base-pair steps.
From a dynamical point of view, unlike the ions
located in the grooves, which display residence
times in the 500 ps range, the ions coming into
contact with the phosphate anionic oxygen atoms
are not well ordered and do not display residence
times exceeding 50 to 100 ps (Figure 6). Interestingly, no ion-binding seems possible in the shallow/minor groove of CˆG steps, while binding
has been observed by crystallography in the minor
groove of ApT steps (Tereshko et al., 1999) exemplifying, along with other MD simulations
(Cheatham & Kollman, 1997; Young et al., 1997a;
Feig & Pettitt, 1999b), the occurrence of sequencespeci®c ion-binding patterns. Thus, the present
results are consistent with previous propositions
(Young et al., 1997a; McFail-Isom et al., 1999) of a
hybrid ion/solvent model, indicating that the
monovalent ions, present in the surrounding of the
nucleic acid, share hydration sites with water molecules. It is noteworthy that the different density
maps shown in Figure 9 are indicative of some
errors that may be introduced into the interpretation of electronic densities resulting from medium
resolution crystallographic studies when one considers that only water molecules are present in the
®rst hydration shell of nucleic acids.
Differences in gel electrophoretic mobility
result from hydration properties
It has been established that RNA duplexes display reduced electrophoretic mobility in polyacrylamide gels compared to DNA duplexes of the
same sequence and length, despite possessing the
same formal negative charge (Ratmeyer et al., 1994;
Lesnik & Freier, 1995; Bonifacio et al., 1997;
Hagerman, 1997; Gyi et al., 1998b). This behavior
was attributed to either a difference in hydration
between the two nucleic acids or to a difference in
their ion-binding abilities. Indeed, the ionic atmosphere should be more concentrated around A-form
than around B-form structures, as the axial charge
density is higher for the former molecules, leading
to a higher degree of neutralization of the duplex
and a resulting reduced electrophoretic force.
The present results (Tables 1 and 2) indicate that,
although there is a signi®cant difference in the
hydration of the two duplexes (approximately two
1126
water molecules per base-pair in favor of RNA),
there is no noticeable variation in the number of
counterions in direct contact with the atoms of the
nucleic acids, especially in the deep/major groove
(Table 1). These ions, which are directly bound to
the helices and should therefore migrate with the
duplexes, should have the strongest impact on
their respective mobility. Thus, it appears that in
order to understand the experimentally observed
difference in mobility, the two additional water
molecules found in the RNA ®rst hydration shell
associated with the lower mobility of the solvent
molecules surrounding the RNA compared to the
DNA molecules should be seriously considered.
About reasonable simulation lengths
Several nucleic acid MD simulations overcoming
the present 2 ns range have been published (5 ns
Young et al., 1997b; 10 ns Feig & Pettitt, 1998b;
Young & Beveridge, 1998; 25 ns Bevan et al., 2000),
and this raises the relevant question of the relation
between the simulation length and its signi®cance.
Although long simulations necessary in order to
extensively sample the conformational space of the
studied system (Caspar, 1995; Clarage et al., 1995),
a few of these nanosecond simulations of nucleic
acids sampled infrequent or improbable events
(Cieplak et al., 1997; Feig & Pettitt, 1998b). Also, in
some instances, it would be desirable to extend the
simulation times; indeed the results obtained by
several groups indicate that long simulations
would sample only protocol and force-®eld-dependent conformational regions of the ``theoretical''
phase space, which may not be congruent with the
``real'' phase space of the studied system
(Auf®nger & Westhof, 1998d). For example, in
some MD simulations, severe and undesirable
structural distorsion appeared only after running
several nanoseconds of simulations (Ravishanker
et al., 1997; Cheatham et al., 1999). Such effects
have been documented for simulations using several protocols of increased accuracy for the determination of the electrostatic interactions (Auf®nger
& Westhof, 1998b). It was made clear that
sampling the conformational space over long timescales while using inaccurate protocols or inappropriate representations of the environment
(Auf®nger et al., 1999) is of little use. Beyond
doubt, even the best force-®elds and MD methods
available present some de®ciencies that place
boundaries to their applicability, despite continual
and praiseworthy efforts to improve them (Cornell
et al., 1995; Cheatham et al., 1999; Foloppe &
MacKerell, 2000; MacKerell & Banavali, 2000).
Thus, the reliability of MD simulations has ®rst to
be assessed on the shorter time-scales (which
extend presently into the nanosecond regime)
before exploring the longer time-scales.
Therefore, as an alternative to long MD simulations (Young et al., 1997b; Feig & Pettitt, 1998b;
Bevan et al., 2000; Daggett, 2000) and multiple MD
simulations (Auf®nger et al., 1995; Auf®nger &
Nucleic Acid Water and Ion-binding Features
Westhof, 1997a; Caves et al., 1998; Worth et al.,
1998), the present strategy of investigating the
dynamical behavior of a structure containing
highly repetitive motifs (e.g. CˆG base-pairs)
appears successful in getting interesting and
valuable statistical data. This is demonstrated,
for example, by the ability of the simulations to
discriminate between the ion-binding features of
GpC and CpG steps.
Summary and Perspectives
The present MD simulations give, in good agreement with a broad range of experimental data, a
precise view of the intrinsic dynamics of an RNA
and a DNA duplex, as well as of the hydration
shell surrounding the important CˆG WatsonCrick base-pairs. The RNA helix is calculated to be
less ¯exible than the DNA duplex, and to remain
closer to its starting structure (A-form) than the
DNA duplex (B-form). It has been shown that
CˆG base-pairs are surrounded by 22 hydration
sites occupied by 21.4 (RNA) and 19.8 (DNA)
water molecules and K‡. This difference of two
water molecules can be ascribed to the presence of
two hydroxyl groups per RNA base-pair. From the
present and previous (Feig & Pettitt, 1999a) results,
it is possible to propose the following scale of
hydration for CˆG base-pairs: A-DNA < RNA
(‡3 H2O) and B-DNA < RNA (‡2 H2O), indicating
that RNA is more hydrated than the two A and BDNA forms. From a dynamical point of view, the
hydration shell is less well de®ned around the
DNA structure than around the RNA structure,
due to the higher mobility of the DNA backbone.
Water molecules with long residence times are
found mainly around the RNA and DNA phosphate groups (0.5 ns range) and some very longlived water molecules are observed in the DNA
minor groove (1 ns range). As expected, some
water molecules form long-lived bridges between
anionic oxygen atoms of RNA phosphate groups
(0.4 ns range) and some form bridges between the
O2/N3 and O40 atoms in the DNA minor groove
(up to 0.6 ns). Again, water is recognized as an
integral part of nucleic acid structures (Westhof,
1988) and, in the present case, the complex interplay between the hydration shell and the presence
(RNA) or absence (DNA) of a ribose 20 -hydroxyl
group leads to nucleic acid structures with different shapes and dynamical features.
The potassium ion-binding properties are similar
for both nucleic acid structures. Approximately,
0.5 K‡ form direct contact with each CˆG pair.
The ions form short-lived contacts with the phosphate group and longer-lived contacts with the
major groove O6 and N7 atoms. Most of the ions
found in the deep/major groove form bridges
between two adjacent (G)O6 atoms. However, for
both B-DNA and RNA duplexes, potassium ions
prefer to interact with GpC instead of CpG steps.
Ion residence times in the deep/major groove
1127
Nucleic Acid Water and Ion-binding Features
approach 0.5 ns. Thus, MD simulations provide a
theoretical opportunity to propose reasonable
binding sites for the experimentally unobservable
or ``lost'' cations present in the surroundings of the
nucleic acids in crystals and in solution. The present results may help to rationalize several experimental observations. The variance in hydration
calculated in this study should be de®nitively considered in order to understand the experimentally
observed differences in gel electrophoretic mobility
of RNA and DNA duplexes of similar sequences,
besides the ion-binding properties, which appear
to be roughly identical for both nucleic acid structures. Thus, MD simulations provide a good basis
for interpreting data originating from NMR or
other experimental techniques for which a detailed
three-dimensional structure including the environment is often lacking.
The publication of very high-resolution crystallographic structures reaching and even overcoming
Ê resolution boundary (Masquida et al.,
the 1.0 A
1999; Soler-LoÂpez et al., 1999) will allow an indepth comparison between experimental and
theoretical results of the hydration patterns around
canonical and non-canonical base-pairs. The use of
hydration patterns has already been shown to be
useful in the re®nement of medium or low-resolution crystal structures (Klosterman et al., 1999).
Further, as it has been proposed that protein atoms
involved in protein/DNA interactions can overlap
with DNA hydration sites (Seeman et al., 1976), the
characterization of hydration sites should help
de®ning possible protein/nucleic acid or drug/
nucleic acid interactions, as proposed earlier for
DNA (Woda et al., 1998).
Computational Methods
System setup
Two 2.4 ns MD trajectories of the r(CG)12 and
the d(CG)12 helices at constant temperature (298 K)
and pressure (1 atm ˆ 101,325 Pa) have been generated with the AMBER 5.0 simulation package
(Pearlman et al., 1995; Case et al., 1997). The
duplexes were placed in a box containing 72 K‡
and 26 Clÿ as well as 5697 (RNA) and 5589 (DNA)
SPC/E water molecules (Berendsen et al., 1987),
corresponding to a concentration close to 0.25 M
KCl (Figure 1). The starting A-RNA and B-DNA
structures were generated from ®ber models by
using the NUCGEN module of AMBER. The calculations employed the all-atom force-®eld described
by Cornell et al. (1995). The van der Waals parameters for the ions were extracted from the work
of Dang (Clÿ, Dang, 1992; K‡, Dang, 1995), which
calibrated them according to the SPC/E water
potential. It has to be noted that the ions displayed
a tendency to form clusters with the present water
potential, a trend that has been observed in simulations of 1.0 M NaCl aqueous solutions (DegreÁve
& da Silva, 1999). The box dimensions were chosen
Ê solvation shell around
in order to ensure a 12 A
the duplexes. The ions were placed around the
nucleic acids on the basis of the electrostatic potential of the solvated system such that no ion was
Ê to any solute atom or closer than
closer than 5 A
Ê to any other ion. Thus, no ion was initially pre4A
sent in the nucleic acid grooves.
MD setup and equilibration procedure
For the treatment of the electrostatic interactions,
the particle mesh Ewald (PME) summation method
(Darden et al., 1999; Sagui & Darden, 1999) has
been used in order to avoid severe artifacts generated by the use of truncation methods (Auf®nger
& Westhof, 1998b). The charge grid spacing was
Ê , and a cubic interpchosen to be close to 1.0 A
Ê truncation
olation scheme was used. A 9.0 A
distance was applied to the Lennard-Jones interactions. The trajectory was run with a time-step of
2 fs and shake bond constraints. The non-bonded
pair list was updated every ten steps. The equilibration procedure consisted of 100 steps of steepest
descent minimization without positional constraints, followed by two 25 ps of MD simulations,
the ®rst with ®xed solute atoms and ions, the latter
with ®xed solute atoms and mobile ions. Then,
seven 50 ps MD runs were performed with positional constraints on the nucleic acid atoms of 10,
Ê ÿ2, respect5, 2, 1, 0.5, 0.1 and 0.01 kcal molÿ1 A
ively, yielding a complete equilibration phase of
400 ps (Figure 2). Such an equilibration procedure
is akin to procedures where the temperature is
slowly raised to the target temperature (Auf®nger
& Westhof, 1998b). However, with this procedure,
water molecules and ions may explore a larger
conformational space, as they are always maintained at the target temperature. This is especially
important for the ions, which explore their conformational space more slowly than the other components of the system. After the 400 ps
equilibration phase, a 2.0 ns production run was
performed from which only the last 1.5 ns were
used to calculate average properties.
Determination of water and ion-binding sites
The direct ion-coordination and hydration sites
were determined by using a procedure ®rst developed by Schneider & Berman for the analysis of
the hydration of DNA bases (Schneider & Berman,
1995) and later applied to the hydration of DNA
phosphate groups (Schneider et al., 1998) and RNA
base-pairs (Auf®nger & Westhof, 1998c). Brie¯y,
for the base-pairs, the water molecules or ions that
Ê from any heavy atom
are located at less than 3.5 A
Ê above or
of the CˆG pairs and less than 2.5 A
below the base-pair plane were used to construct
Ê
the water and ion densities (the out-of-plane 2.5 A
criterion has not been used for the backbone
atoms). An occupancy factor of 0.5 was ascribed to
the ions and water molecules belonging to the
coordination shell of two base-pair steps (especially
for those close to the (G)O6 atoms in order to
1128
avoid counting them twice during the averaging
procedure). The distributions of the water molecules proximate to the base-pairs were transformed into ``pseudoelectron'' densities through a
Fourier transformation by using the SFALL program of the CCP4 library (CCP4, 1994) and the O
program (Jones et al., 1991) was used to display the
calculated pseudoelectron densities.
Ê boundary was set in
The out-of-plane 2.5 A
order to exclude water molecules belonging to the
®rst hydration shell of neighboring base-pairs (the
Ê ). The
average interbase distance is close to 3.4 A
Ê water-to-base and ion-to-base distance
3.5 A
criteria correspond to the ®rst minimum in the corresponding calculated radial distribution functions.
Interestingly, this criterion is valid for the ion as
well as for the water molecules.
The average structures for the CˆG nucleotides
shown in Figure 2 were obtained by superposing
the heavy base-pair atoms of the 12 central CˆG
steps for 300 con®gurations (one con®guration
every 3 ps) over the last 1.5 ns of the trajectories.
For the calculation of the hydrogen bonding percentages, HB %, the following standard hydrogenÊ and
bonding criteria were used: d(H. . .A) < 2.5 A
y(D-H. . .A) < 135 , where A stands for a hydrogen
acceptor atom and D for a hydrogen donor atom.
Acknowledgments
The authors thank Claude Sauter for his help with the
crystallographic software as well as Georges Wipff and
Etienne Engler for the use of their MD Draw display program (Engler & Wipff, 1994).
References
Arnott, S. & Selsing, E. (1974). The structure of polydeoxyguanylic acid polydeoxycytidylic acid. J. Mol.
Biol. 88, 551-553.
Arnott, S., Campbell-Smith, P. J. & Chandrasekaran, R.
(1976). Atomic coordinates and molecular conformations for DNA-DNA, RNA-RNA, and DNARNA helices. In CRC Handbook of Biochemistry and
Molecular Biology (Fassman, G., ed.), vol. 2, pp. 411422, CRC Press, Boca Raton, FL.
Auf®nger, P. & Westhof, F. (1997a). RNA hydration:
three nanoseconds of multiple molecular dynamics
simulations of the solvated tRNAASP anticodon hairpin. J. Mol. Biol. 269, 326-341.
Auf®nger, P. & Westhof, E. (1997b). Rules governing the
orientation of the 20 -hydroxyl group in RNA. J. Mol.
Biol. 274, 54-63.
Auf®nger, P. & Westhof, E. (1998a). Effects of pseudouridylation on tRNA hydration and dynamics: a
theoretical approach. In Modi®cation and Editing of
RNA (Grosjean, H. & Benne, R., eds), pp. 103-112,
American Society for Microbiology, Washington,
DC.
Auf®nger, P. & Westhof, E. (1998b). Molecular dynamics
of nucleic acids. In Encyclopedia of Computational
Chemistry (Schleyer, v. P. R., Allinger, N. L., Clark,
T., Gasteiger, J., Kollman, P. A., Schaefer, H. F., III
Nucleic Acid Water and Ion-binding Features
& Schreiner, P. R., eds), vol. 5, pp. 1629-1640, Wiley
& Sons, Chichester.
Auf®nger, P. & Westhof, E. (1998c). RNA base pair
hydration. J. Biomol. Struct. Dynam. 16, 693-707.
Auf®nger, P. & Westhof, E. (1998d). Simulation of the
molecular dynamics of nucleic acids. Curr. Struct.
Biol. 8, 227-236.
Auf®nger, P. & Westhof, E. (1999). Roles of hydration
on the structure and dynamics of nucleic acids. In
Water Management in the Design and Distribution of
Quality Food (Ross, Y. H., Leslie, R. B. & Liliford,
P. J., eds), pp. 165-198, Technonic Publishing Co.
Inc., Basel.
Auf®nger, P., Louise-May, S. & Westhof, E. (1995). Multiple molecular dynamics simulations of the anticodon loop of tRNAASP in aqueous solution with
counterions. J. Am. Chem. Soc. 117, 6720-6726.
Auf®nger, P., Louise-May, S. & Westhof, E. (1996).
Molecular dynamics simulations of the anticodon
hairpin of tRNAASP: structuring effects of C-H. . .O
hydrogen bends and of long-range hydration forces.
J. Am. Chem. Soc. 118, 1181-1189.
Auf®nger, P., Louise-May, S. & Westhof, E. (1999). Molecular dynamics simulations of the solvated yeast
tRNAASP. Biophys. J. 76, 50-64.
Basu, S., Rambo, R. P., Strauss-Soukup, J., Cate, J. H.,
FerreÂ-D'AmareÂ, A. R., Strobel, S. A. & Doudna, J. A.
(1998). A speci®c monovalent metal ion integral to
the AA platform of the RNA tetraloop receptor.
Nature Struct. Biol. 5, 986-992.
Berendsen, H. J. C., Grigera, J. R. & Straatsma, T. P.
(1987). The missing term in effective pair potential.
J. Phys. Chem. 97, 6269-6271.
Berman, H. M. (1991). Hydration of DNA. Curr. Opin.
Struct. Biol. 1, 423-427.
Berman, H. M. (1994). Hydration of DNA: take 2. Curr.
Opin. Struct. Biol. 4, 345-350.
Bevan, D. R., Li, L., Pedersen, L. O. & Darden, T. A.
(2000). Molecular dynamics simulations of the
d(CCAACGTTGG)2 decamer: in¯uence of the crystal environment. Biophys. J. 78, 668-682.
Beveridge, D. L. & McConnell, K. J. (2000). Nucleic
acids: theory and computer simulation, Y2 K. Curr.
Opin. Struct. Biol. 10, 182-196.
Bonifacio, G. F., Brown, T., Conn, G. L. & Lane, A. N.
(1997). Comparison of the electrophoretic and
hydrodynamic properties of DNA and RNA oligonucleotide duplexes. Biophys. J. 73, 1532-1538.
Case, D. A., Pearlman, D. A., Caldwell, J. W.,
Cheatham, T. E., Ross, W. S., Simmerling, C. L.,
Darden, T. A., Merz, K. M., Stanton, R. V., Cheng,
A. L., Vincent, J. J., Crowley, M., Fergusson, D. M.,
Radmer, R. J., Seibel, G. L., Singh, U. C., Weiner,
P. K. & Kollman, P. A. (1997). AMBER 5, University
of California, San Francisco.
Caspar, D. L. D. (1995). Problems in simulating macromolecular movements. Structure, 3, 327-329.
Caves, L. S. D., Evanseck, J. D. & Karplus, M. (1998).
Locally accessible conformations of proteins: multiple molecular dynamics simulations of crambin.
Protein Sci. 7, 649-666.
Cheatham, T. E. & Kollman, P. A. (1996). Observation
of the A-DNA to B-DNA transition during
unrestrained molecular dynamics in aqueous solution. J. Mol. Biol. 259, 434-444.
Cheatham, T. E. & Kollman, P. A. (1997). Molecular
dynamics simulations highlights the structural
differences among DNA:DNA, RNA:RNA, and
Nucleic Acid Water and Ion-binding Features
DNA:RNA hybrid duplexes. J. Am. Chem. Soc. 119,
4805-4825.
Cheatham, T. E., Srinivasan, J., Case, D. A. & Kollman,
P. A. (1998). Molecular dynamics and continuum
solvent studies of the stability of polyG-polyC and
polyA-polyT DNA duplexes in solution. J. Biomol.
Struct. Dynam. 16, 265-280.
Cheatham, T. E., Cieplak, P. & Kollman, P. A. (1999). A
modi®ed version of the Cornell et al. force ®eld
with improved sugar pucker phases and helical
repeat. J. Biomol. Struct. Dynam. 16, 845-862.
Chiu, T. K., Kaczor-Greskowiak, M. & Dickerson, R. E.
(1999). Absence of minor groove monovalent cations in the crosslinked dodecamer C-G-C-G-A-A-TT-C-G-C-G. J. Mol. Biol. 292, 589-608.
Cieplak, P., Cheatham, T. E. & Kollman, P. A. (1997).
Molecular
dynamics
simulations
®nd
that
30 phosphoramidate
modi®ed
DNA
duplexes
undergo a B to A transition and normal DNA
duplexes an A to B transition. J. Am. Chem. Soc. 119,
6722-6730.
Clarage, J. B., Romo, T., Andrews, B. K., Pettitt, B. M. &
Phillips, G. N. (1995). A sampling problem in molecular dynamics simulations of macromolecules.
Proc. Natl Acad. Sci. USA, 92, 3288-3292.
Collaborative Computing Project Number 4 (1994). The
CCP4 suite: programs for protein crystallography.
Acta Crystallog. Sect. D, 50, 760-763.
Conte, M. R., Conn, G. L., Brown, T. & Lane, A. N.
(1996).
Hydration
of
the
RNA
duplex
r(CGCAAAUUUGCG)2 determined by NMR. Nucl.
Acids Res. 24, 3693-3699.
Conte, M. R., Conn, G. L., Brown, T. & Lane, A. N.
(1997). Conformational properties and thermodynamics of the RNA duplex r(CGCAAAUUUGCG)2:
comparison
with
the
DNA
analogue
d(CGCAAATTTGCG)2. Nucl. Acids Res. 25, 26272634.
Cornell, W. D., Cieplak, P., Bayly, C. I., Gould, I. R.,
Merz, K. M., Ferguson, D. M., Spellmeyer, D. C.,
Fox, T., Caldwel, J. W. & Kollman, P. A. (1995). A
second generation force ®eld for the simulation of
proteins, nucleic acids, and organic molecules. J. Am.
Chem. Soc. 117, 5179-5197.
Daggett, V. (2000). Long time scale simulations. Curr.
Opin. Struct. Biol. 10, 160-164.
Dang, L. X. (1992). The nonadditive intermolecular
potential for water revised. J. Chem. Phys. 97, 26592660.
Dang, L. X. (1995). Mechanism and thermodynamics of
ion selectivity in aqueous solutions of 18-crown-6
ether: a molecular dynamics study. J. Am. Chem.
Soc. 117, 6954-6960.
Darden, T., Perera, L., Li, P. & Pedersen, L. (1999). New
tricks for modelers from the crystallography toolkit:
the particle mesh Ewald algorithm and its use in
nucleic acid simulations. Structure, 7, R55-R60.
DegreÁve, L. & da Silva, F. L. B. (1999). Large ionic clusters in concentrated aqueous NaCl solution. J. Chem.
Phys. 111, 5150-5156.
Denisov, V. P. & Halle, B. (2000). Sequence-speci®c
binding of counterions to B-DNA. Proc. Natl Acad.
Sci. USA, 97, 629-633.
Denissov, V. P., CarlstroÈm, G., Venu, K. & Halle, B.
(1997). Kinetics of DNA hydration. J. Mol. Biol. 268,
118-136.
Draper, D. E. & Misra, V. K. (1998). RNA shows its
metal. Nature Struct. Biol. 5, 927-930.
1129
Duan, Y., Wilkosz, P., Crowley, M. & Rosenberg, J. M.
(1997). Molecular dynamics simulation study
of DNA dodecamer d(CGCGAATTCGCG)2 in
solution: conformation and hydration. J. Mol. Biol.
272, 553-572.
Egli, M., Portmann, S. & Usman, N. (1996). RNA
hydration: a detailed look. Biochemistry, 35, 84898494.
Engler, E. & Wipif, G. (1994). MD Draw: A Program of
Graphical Representation of Molecular Dynamics Trajectories, Universite Louis Pasteur de Strasbourg,
France.
Feig, M. & Pettitt, B. M. (1998a). Crystallographic water
sites from a theoretical perspective. Structure, 6,
1351-1354.
Feig, M. & Pettitt, B. M. (1998b). Structural equilibrium
of DNA represented with different force-®elds. Biophys. J. 75, 134-149.
Feig, M. & Pettitt, B. M. (1999a). Modeling high-resolution hydration patterns in correlation with DNA
sequence and conformation. J. Mol. Biol. 286, 10751095.
Feig, M. & Pettitt, B. M. (1999b). Sodium and chlorine
ions as part of the DNA solvation shell. Biophys. J.
77, 1769-1781.
Flatters, D. & Lavery, R. (1998). Sequence-dependent
dynamics of TATA-box binding sites. Biophys. J. 75,
372-381.
Foloppe, N. & MacKerell, A. D. (2000). All-atom empirical force-®eld for nucleic acids. I. Parameter optimization based on small molecule and condensed
phase macromolecular target data. J. Comput. Chem.
21, 88-104.
Gluick, T. C., Wills, N. M., Gesteland, R. F. & Draper,
D. E. (1997). Folding of an mRNA pseudoknot for
stop codon readthrough: effects of mono- and divalent ions on stability. Biochemistry, 36, 16173-16186.
Gyi, J. I., Lane, A. N., Conn, G. L. & Brown, T. (1998a).
The orientation and dynamics of the C20 -OH and
hydration of RNA and DNA.RNA hybrids. Nucl.
Acids Res. 26, 3104-3110.
Gyi, J. I., Lane, A. N., Conn, G. L. & Brown, T. (1998b).
Solution structure of DNA.RNA hybrids with purine-rich and pyrimidine-rich strands: comparison
with the homologous DNA and RNA duplexes. Biochemistry, 37, 73-80.
Hagerman, P. J. (1988). Flexibility of DNA. Annu. Rev.
Biophys. Biophys. Chem. 17, 265-286.
Hagerman, P. J. (1997). Flexibility of RNA. Annu. Rev.
Biophys. Biomol. Struct. 26, 139-156.
Hermann, T. & Westhof, E. (1998). Aminoglycoside
binding to the hammerhead ribozyme: a general
model for the interaction of cationic antibiotics with
RNA. J. Mol. Biol. 276, 903-912.
Hud, N. V., Sklenar, V. & Feigon, J. (1999). Localization
of ammonium ions in the minor groove of DNA
duplexes in solution and the origin of DNA A-tract
bending. J. Mol. Biol. 286, 651-660.
Israelachvili, J. & WennerstroÈm, H. (1996). Role of
hydration and water in biological and colloidal
interactions. Nature, 79, 219-225.
Jacobson, A., Leupin, W., Liepinsh, E. & Otting, G.
(1996). Minor groove hydration of DNA in aqueous
solution: sequence-dependent next neighbor effect
of the hydration lifetimes in d(TTAA)2 segments
measured by NMR spectroscopy. Nucl. Acids Res.
24, 2911-2918.
JoÂhannesson, H. & Halle, B. (1998). Minor groove
hydration of DNA in solution: dependence on base
1130
composition and sequence. J. Am. Chem. Soc. 120,
6859-6870.
Jones, T. A., Zou, J. Y., Cowan, S. W. & Kejldgaard, M.
(1991). Improved methods for building models in
electron density maps and the location of errors in
these models. Acta Crystallog. sect. A, 47, 110-119.
Kankia, B. I. & Marky, L. A. (1999). DNA, RNA, and
DNA/RNA oligomer duplexes: a comparative
study of their stability, heat, hydration, and Mg2‡
binding properties. J. Phys. Chem. Ser. B, 103, 87598767.
Karplus, P. A. & Faerman, C. (1994). Ordered water in
macromolecular structure. Curr. Opin. Struct. Biol. 4,
770-776.
Klosterman, P. S., Shah, S. A. & Steitz, T. A. (1999).
Crystal structures of two plasmid copy control
related RNA duplexes: an 18 base pair duplex at
Ê resolution and a 19 base pair duplex at
1.20 A
Ê resolution. Biochemistry, 38, 14784-14792.
1.55 A
Kochoyan, M. & Leroy, J. L. (1995). Hydration and solution structure of nucleic acids. Curr. Opin. Struct.
Biol. 5, 329-333.
Kopka, M. L., Fratini, A. V., Drew, H. R. & Dickerson,
R. E. (1983). Ordered water structure around a BDNA dodecamer. A quantitative study. J. Mol. Biol.
163, 129-146.
Kubinec, M. G. & Wemmer, D. E. (1992). NMR evidence
for DNA bound water in solution. J. Am. Chem. Soc.
114, 8739-8740.
Lane, A. N., Jenkins, T. C. & Frenkiel, T. A. (1997).
Hydration and solution structure of d(CGCAA
ATTTGCG)2 and its complex with propamidine
from NMR and molecular modelling. Biochim. Biophys. Acta, 1350, 205-220.
Lesnik, E. A. & Freier, S. M. (1995). Relative thermodynamic stability of DNA, RNA, and DNA:RNA
hybrid duplexes: relationship with base composition
and structure. Biochemistry, 34, 10807-10815.
Liepinsh, E., Otting, G. & WuÈthrich, K. (1992). NMR
observation of individual molecules of hydration
water bound to DNA duplexes: direct evidence for
a spine of hydration water present in aqueous solution. Nucl. Acids Res. 20, 6549-6553.
Liepinsh, E., Leupin, W. & Otting, G. (1994). Hydration
of DNA in aqueous solution: NMR evidence for a
kinetic destabilization of the minor groove
hydration of d-(TTAA)2 versus d-(AATT)2 segments. Nucl. Acids Res. 22, 2249-2254.
Lyubartsev, A. P. & Laaksonen, A. (1998). Molecular
dynamics simulations of DNA in solutions with
different counter-ions. J. Biomol. Struct. Dynam. 16,
579-592.
MacKerell, A. D. & Banavali, N. (2000). All-atom empirical force-®eld for nucleic acids: II. Application to
molecular dynamics simulations of DNA and RNA
in solutions. J. Comput. Chem. 21, 105-120.
MacKerell, A. D., WioÂrkiewicz-Kuczera, J. & Karplus,
M. (1995). An all-atom empirical energy function
for the simulation of nucleic acids. J. Am. Chem. Soc.
117, 11946-11975.
Masquida, B., Sauter, C. & Westhof, E. (1999). A sulfate
Ê
pocket formed by three G U pairs in the 0.97 A
resolution X-ray structure of a nonameric RNA.
RNA, 5, 1384-1395.
McFail-Isom, L., Sines, C. C. & Williams, L. D. (1999).
DNA structure: cations in charge? Curr. Opin.
Struct. Biol. 9, 298-304.
Minchenkova, L. E., Schyolkina, A. K., Chernov, B. K. &
Ivanov, V. I. (1986). CC/GG contacts facilitate the B
Nucleic Acid Water and Ion-binding Features
to A transition of DNA in solution. J. Biomol. Struct.
Dynam. 4, 463-476.
Mohan, V., Smith, P. E. & Pettitt, B. M. (1993). Molecular dynamics simulation of ions and water around
triplex DNA. J. Phys. Chem. 97, 12984-12990.
Ng, H. L., Kopka, M. L. & Dickerson, R. E. (2000). The
structure of a stable intermediate in the A B DNA
helix transition. Proc. Natl Acad. Sci. USA, 97, 20352039.
Norberg, J. & Nilsson, L. (1996). Constant pressure molecular dynamics simulations of the dodecamers
d(GCGCGCGCGCGC)2 and r(GCGCGCGCGCGC)2.
J. Chem. Phys. 104, 6052-6057.
Otting, G. & Liepinsh, E. (1995). Protein hydration
viewed by high-resolution NMR spectroscopy:
implications for magnetic resonance image contrast.
Acc. Chem. Res. 28, 171-177.
Pearlman, D. A., Case, D. A., Caldwell, J. W., Ross,
W. S., Cheatham, T. E., DeBolt, S., Ferguson, D.,
Seibel, G. & Kollman, P. (1995). AMBER, a package
of computer programs for applying molecular
mechanics, normal mode analysis, molecular
dynamics and free energy calculations to simulate
the structural and energetic properties of molecules.
Comput. Phys. Commun. 91, 1-41.
Phan, A. T., Leroy, J. L. & GueÂron, M. (1999). Determination of the residence time for water molecules
hydrating B0 -DNA and B-DNA, by the one-dimensional zero-enhancement nuclear Overhauser effect.
J. Mol. Biol. 286, 505-519.
Pilet, J. & Brahms, J. (1972). Dependence of B-A conformational change in DNA on base composition.
Nature New. Biol. 236, 99-100.
Ratmeyer, L., Vinayak, R., Zhong, Y. Y., Zon, G. &
Wilson, W. D. (1994). Sequence speci®c thermodynamic and structural properties for DNA.RNA
duplexes. Biochemistry, 33, 5298-5304.
Ravishanker, G., Auf®nger, P., Langley, D. R., Jayaram,
B., Young, M. A. & Beveridge, D. L. (1997). Treatment of counterions in computer simulations of
DNA. In Reviews in Computational Chemistry
(Lipkowitz, K. B. & Boyd, D. B., eds), vol. 11, pp.
317-372, VCH Publishers, New York.
Saenger, W. (1984). Principles of Nucleic Acid Structure.
Springer Advanced Text in Chemistry, SpringerVerlag, New-York.
Saenger, W., Hunter, W. N. & Kennard, O. (1986). DNA
conformation is determined by economics in the
hydration of phosphate groups. Nature, 24, 385-388.
Sagui, C. & Darden, T. A. (1999). Molecular dynamics
simulations of biomolecules: long-range electrostatic
effects. Annu. Rev. Biophys. Struct. 28, 155-179.
SantaLucia, J., Jr (1998). A uni®ed view of polymer,
dumbbell, and oligonucleotide DNA nearest-neighbor thermodynamics. Proc. Natl Acad. Sci. USA, 95,
1460-1465.
Schneider, B. & Berman, H. M. (1995). Hydration of the
DNA bases is local. Biophys. J. 69, 2661-2669.
Schneider, B., Patel, K. & Berman, H. M. (1998).
Hydration of the phosphate group in double-helical
DNA. Biophys. J. 75, 2422-2434.
Seeman, N. C., Rosenberg, J. M. & Rich, A. (1976).
Sequence-speci®c recognition of double helical
nucleic acids by proteins. Proc. Natl Acad. Sci. USA,
73, 804-808.
Shindo, H., Fujiwara, T., Akutsu, H., Matsumoto, U. &
Kyogoku, Y. (1985). Phosphorus-31 nuclear magnetic resonance of highly oriented DNA ®bers. 1.
Nucleic Acid Water and Ion-binding Features
Static geometry of DNA double helices. Biochemistry, 24, 887-895.
Shui, X., Sines, C. C., McFail-Isom, L., VanDerveer, D. &
Williams, L. D. (1998). Structure of the potassium
form of CGCGAATTCGCG: DNA deformation by
electrostatic collapse around inorganic cations. Biochemistry, 37, 16877-16887.
Soler-LoÂpez Malinina, L., Liu, J., Huynh-Dinh, T. &
Subirana, J. A. (1999). Water and ions in a high resolution structure of B-DNA. J. Biol. Chem. 274,
23683-23686.
Sorenson, J. M., Hura, G., Soper, A. K., Pertsemiidis, A.
& Head-Gordon, T. (1999). Determining the role of
hydration forces in protein folding. J. Phys. Chem.
ser. B, 103, 5413-5426.
Sprous, D., Young, M. A. & Beveridge, D. L. (1999).
Molecular dynamics? studies of axis bending in
d(G5-(GA4T4C)2-C5) and d(G5-(GT4A4C)2-C5): effects
of sequence polarity on DNA curvature. J. Mol. Biol.
285, 1623-1632.
Su, L., Chen, L., Egli, M., Berger, J. M. & Rich, A. (1999).
Minor groove RNA triplex in the crystal structure
of a ribosomal frameshifting viral pseudoknot.
Nature Struct. Biol. 6, 285-292.
Tereshko, V., Minasov, G. & Egli, M. (1999). A ``hydration'' spine in a B-DNA minor groove. J. Am. Chem.
Soc. 121, 3590-3595.
Trantirek, L., Ste¯, R., Vorlickova, M., Koca, J., Sklenar,
V. & Kypr, J. (2000). An A-type double helix of
DNA having B-type puckering of the deoxyribose
rings. J. Mol. Biol. 297, 907-922.
Wahl, C. M. & Sundaralingam, M. (1997). C-H. . .O
hydrogen bonding in biology. Trends Biochem. Sci.
22, 97-102.
Wang, S. & Kool, T. E. (1995). Origins of the large difference in stability of DNA and RNA helices C-5
methyl and 20 -hydroxyl effects. Biochemistry, 34,
4125-4132.
Wang, Y. X., Lu, M. & Draper, D. E. (1993). Speci®c
ammonium ion requirement for functional ribosomal RNA tertiary structure. Biochemistry, 32, 1227912282.
Weerasinghe, S., Smith, P. E., Mohan, V., Cheng, Y.-K.
& Pettitt, B. M. (1995a). Nanosecond dynamics and
structure of a model DNA triplex helix in saltwater
solution. J. Am. Chem. Soc. 117, 2147-2158.
Weerasinghe, S., Smith, P. E. & Pettitt, M. (1995b). Structure and stability of a model pyrimidine-purinepurine DNA triple helix with a GC.T mismatch by
simulation. Biochemistry, 34, 16269-16278.
Westhof, E. (1988). Water: an integral part of nucleic
acid structure. Annu. Rev. Biophys. Biophys. Chem.
17, 125-144.
1131
Westhof, E. (1993). Structural water bridges in nucleic
acids. In Water and Biological Macromolecules
(Westhof, E., ed.), vol. 17, pp. 227-243, The
Macmillan Press Ltd, London.
Westhof, E. & Beveridge, D. L. (1990). Hydration of
nucleic acids. In Water Science Reviews (Franks, F.,
ed.), vol. 5, pp. 24-123, Cambridge University Press,
Cambridge, UK.
Westhof, E., Dumas, P. & Moras, D. (1988). Hydration
of transfer RNA molecules: a crystallographic
study. Biochimie, 70, 145-165.
Winger, R. H., Liedl, K. R., RuÈdisser, S., Pichler, A.,
Hallbrucker, A. & Mayer, E. (1998). B-DNA's
BI ! BII conformer substate dynamics is coupled
with water migration. J. Phys. Chem. B, 102, 86348940.
Woda, J., Schneider, B., Patel, K., Mistry, K. & Berman,
H. M. (1998). An analysis of the relationship
between hydration and protein-DNA interactions.
Biophys. J. 75, 2170-2177.
Worth, G. A., Nardi, F. & Wade, R. C. (1998). Use of
multiple molecular dynamics trajectories to study
biomolecules in solution: the YTGP peptide. J. Phys.
Chem. ser. B, 102, 6260-6272.
Xia, T., SantaLucia, J., Jr, Burkard, M. E., Kierzek, R.,
Schroeder, S. J., Jiao, X., Cox, C. & Turner, D. H.
(1998). Thermodynamic parameters for an
expanded nearest-neighbor model for formation of
RNA duplexes with Watson-Crick base pairs. Biochemistry, 37, 14719-14735.
Yang, L. & Pettitt, B. M. (1996). B to A transition of
DNA on the nanosecond time scale. J. Phys. Chem.
100, 2564-2566.
Young, M. A. & Beveridge, D. L. (1998). Molecular
dynamics simulations of an oligonucleotide duplex
with adenine tracts phased by a full helix turn.
J. Mol. Biol. 281, 675-687.
Young, M. A., Jayaram, B. & Beveridge, D. L. (1997a).
Intrusion of counterions into the spine of hydration
in the minor groove of B-DNA: fractional occupancy of electronegative pockets. J. Am. Chem. Soc.
119, 59-69.
Young, M. A., Ravishanker, G. & Beveridge, D. L.
(1997b). A 5-nanosecond molecular dynamics trajectory for B-DNA: analysis of structure, motions, and
solvation. Biophys. J. 73, 2313-2336.
Zichi, D. A. (1995). Molecular dynamics of RNA with
the OPLS force ®eld. Aqueous simulation of a hairpin containing a tetranucleotide loop. J. Am. Chem.
Soc. 117, 2957-2969.
Edited by I. Tinoco
(Received 28 February 2000; received in revised form 20 May 2000; accepted 22 May 2000)