| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Biophys J, September 2000, p. 1213-1227, Vol. 79, No. 3
and
*Center for Structural Biochemistry, Department of Bioscience at
NOVUM, Karolinska Institutet, SE-141 57 Huddinge, Sweden; and
Institute of Bioorganic Chemistry, Polish Academy of
Sciences, 61 704 Poznan, Poland
 |
ABSTRACT |
|---|
 TOP
 ABSTRACT
 INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
|
|---|
Molecular modeling and molecular dynamics have been employed to study the conformation and flexibility of a 15-nucleotide fragment of the plant 5S rRNA containing loop D and a single uridine bulge. Two different model built initial structures were used: one with the bulge localized inside the helical stem and another with the bulge pointing out from the helix. Several independent 700-ps-long trajectories in aqueous solution with Na+ conterions were produced for each starting structure. The bulge nucleotide inside the helix stayed in two main conformations, both of which affected the geometry of the stem part opposite the bulge. When the bulge nucleotide was located outside the helix, we found high base mobility and local backbone flexibility. The dynamics of the hydrogen bond network and conformational changes from a direct to a water mediated hydrogen bond in the sheared G-A basepair in the tetraloop was described. Our results correlate with lead ion induced cleavage patterns in 5S rRNA. Sites resistant to nonspecific lead cleavage appeared in all our simulations as the most rigid fragments independent of the localization of the bulge nucleotide.
| |
INTRODUCTION |
|---|
|
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
|
|---|
5S Ribosomal RNA is an integral component of the
large subunit of both prokaryotic and eukaryotic ribosomes. This rRNA,
about 120 nucleotides long, has been the subject of various studies concerning its structure and biological function. A general secondary structure of prokaryotic and eukaryotic 5S rRNA was proposed on the
basis of sequence alignments and available biochemical data (Specht et
al., 1990
). A large amount of sequence data for molecules of different
origin has been collected and is deposited in the 5S Ribosomal RNA Data
Bank (Szymanski et al., 1999). The consensus secondary structure of all
5S rRNAs possesses a three-way junction with three arms containing
helical regions (I-V) and internal loops (B, E). Hairpins with loops C
and D terminate two of those arms. (In the literature we have found
different notations for the regions of 5S rRNA. In this paper, helices
are designated by roman numerals I-V and loops by capital letters A-E;
Szymanski et al., 1999). The tertiary structure of 5S rRNA is not fully known, and details of the secondary structures for particular molecules
still remain a matter of discussion. Significant contributions toward
understanding 5S rRNA structure in solution come from biochemical methods such as enzymatic digestion, chemical modification,
oligonucleotide binding, or metal ions hydrolysis.
Several attempts have been made to crystallize the whole 5S rRNA
molecule, but the obtained crystals are not suitable for structure
determination at atomic resolution (Lorentz et al., 1991). At present
the tertiary structures are known for parts of the molecule,
corresponding to the two arms of the prokaryotic 5S rRNA. Solution
structures have been obtained for an Escherichia coli 5S
rRNA fragment containing an internal loop E terminated by loop D
(Dallas and Moore, 1997) and for the loop E region in complex with the
ribosomal protein L25 (Stoldt et al., 1999). Crystal structures
encompassing helix I and loop E-helix IV of E. coli 5S rRNA
(Correll et al., 1997), helix I and helix IV terminated by a tetraloop
of Thermus flavus 5S rRNA (Betzel et al., 1994; Perbant et
al., 1998) are also available. For eukaryotic 5S rRNA the only known
structure is that in solution containing the internal loop E (Wimberly
et al., 1993). The recent successful determination of the crystal
structure of the large and small ribosomal subunits at 5 Å and 5.5 Å resolution, respectively, brought again an interest in determining the
structural details and interactions of the components of large
structures (Ban et al., 1999; Clemons et al., 1999).
The subject of this study is a fragment of plant 5S rRNA from wheat germ and lupin seeds corresponding to the natural sequence of nucleotides 81 to 95 of helix IV - loop D (Fig. 1).
|
Helix IV of the eukaryotic 5S rRNAs, in contrast to the bacterial ones,
contains a single nucleotide bulge in the 5' strand. For plant 5S rRNA
an alternative model was proposed where an enlarged loop D encompasses
nucleotides 84-92 and is involved in tertiary interactions (Joachimiak
et al., 1990). Analysis of the lead ion induced cleavage patterns
suggested that the presence of the bulge nucleotide U84 destabilizes
two potential basepairs, G85-C91 and G86-C92, at the bottom of helix IV
and causes an enlargement of loop D (Ciesiolka and Krzyzosiak, 1996),
but tertiary interactions were not confirmed. Additionally, the
cleavage pattern for this fragment of RNA from plants is clearly
different from that for other studied tetraloops in the GNRA family (N
is any nucleotide, R is purine; Ciesiolka et al., 1998). However, this
fragment itself exhibits high thermodynamic stability. A model
consistent with structural transitions observed calorimetrically,
postulates the existence of a basepaired helical region,
G85-G86:C92-C91, and a structured tetraloop, GUGA (Kulinski et al.,
1997). Therefore, knowledge of the potential conformational dynamics of
this RNA fragment may bring us closer to the understanding of RNA
folding and its propensity for intra- and intermolecular interactions.
The aim of this study is to analyze the structural details of plant 5S rRNA fragment (nucleotides 81-95) and its conformational equilibrium in aqueous solution by molecular modeling and molecular dynamics (MD) simulation. The structural diversity obtained from several 0.7-ns-long MD trajectories provides an estimate of the range and stability of possible internal interactions and stabilizing role of water molecules within this RNA structural motif.
| |
METHODS |
|---|
|
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
|
|---|
Two different initial structures were used for the MD
simulations. For each structure two independent, 700-ps-long
simulations that differ in the initial velocities were performed (see
Table 1). Simulations were run with
periodic boundary conditions (PBC). To reduce the number of water
molecules in the system and to compare the influence of the boundary
condition on the results, for one initial structure we have also
performed simulations with stochastic deformable boundary conditions
(SDB; Brooks and Karplus, 1983). MD simulations and all
calculations were performed with the CHARMM program version 25a2 and
26a2 (Brooks et al., 1983) using the force field version 22 (MacKerell
et al., 1995).
|
Simulation protocol
Initial structures
For the investigation of the GGAUGGGUGACCUCC RNA oligomer forming a hairpin with a GUGA tetraloop and a single uridine bulge (underlined; Fig. 1), two initial structures were built: one with the bulge located inside the helix (bulge_in) and another with the bulge outside the helix (bulge_out). In both structures the loop part was derived from the NMR structure for the GGGCGAGAGCCU RNA hairpin loop containing a GAGA tetraloop (PDB code 1zig; Jucker et al., 1996). To get a GUGA tetraloop, in the GAGA tetraloop the adenine base (A) was changed manually to a uracil base (U).
Initial structure bulge_in
First a 16-nucleotide hairpin was built. The geometry of the 4-bp stem fragment (nucleotides 1-4 and 12-15) was derived from the coordinates of canonical A-RNA (Arnott et al., 1976). The stem-loop
part (nucleotides 5-12) came from the nuclear magnetic resonance (NMR)
structure for the hairpin containing GAGA tetraloop, and in this part,
two bases adjacent to the tetraloop were manually changed (C to G and G
to C). Then 100 steps of steepest descent (SD) energy minimization were
applied. To create a 15-mer with the U bulge in position 4, the
base-pairing nucleotide A from the opposite strand was removed and 2500 steps of SD minimization were applied.
Initial structure bulge_out
The 2-bp stem part (nucleotides 1-2 and 14-15) came from the standard A-RNA form. The bulge region was derived from the x-ray structure for an RNA bacteriophage coat protein-operator complex (PDB code 1zdh; Valegård et al., 1997). Coordinates for nucleotides 3-6
and 11-13 were taken from the 5'GAGG/5'CCC stem part
containing a single bulged nucleotide (underlined) of the RNA hairpin.
In this part, three bases were manually changed (the bulge base A to U
and the 5' adjacent basepair G-C to A-U). The loop part (nucleotides 7-10) came from the NMR GAGA tetraloop structure. Finally, the structure was minimized with 100 steps of SD.
All energy minimizations during the initial structure building were
done using harmonic restraints with a 10.0 kcal·mol
1·Å2
force constant on all atoms.
For both structures, 14 Na+ counterions were
placed on the bisector of the phosphate oxygen atoms, 6.0 Å from the
phosphorus atoms. The whole system was solvated in a 45 × 35 × 35 Å size water box (bulge_in,
bulge_out) or in a water sphere (R = 23 Å; bulge_in) containing TIP3P water (Jorgensen et
al., 1983). Water molecules <2.0 Å from RNA or
Na+ were removed. The resulting system contained
a total of 1606 water molecules in the box and 1398 water molecules in
the sphere, corresponding to RNA concentrations of 30 and 32 mM and ion
concentrations of 0.42 and 0.46 M Na+ in the box
and the sphere, respectively. The center of the water sphere was set on
the atom N4(C11). Energy minimization by 100 steps of SD without any
restraints was applied to the solvated system.
Simulations were run with time step 0.002 ps and the SHAKE algorithm
was used to constrain all bonds to hydrogens (Ryckaert et al., 1977).
Coordinates were saved every 100 steps. The nonbonded interaction
energies and forces were smoothly shifted to zero at a cutoff of 12.0 Å by the atom-based truncation method with a force-shifting function
(Steinbach and Brooks, 1994). The nonbonded pair list was regenerated
for atoms within a 13.0-Å cutoff on average every 5 steps, but every
10 steps for simulations Sb2 and Sb7, and every 20 steps for
simulations simulation Sb1. (For a notation of the simulations, see
Table 1.) The relative dielectric constant was set to 1.0. Extensive
simulations of DNA duplexes have shown this protocol to yield
trajectories that are very similar to trajectories obtained using Ewald
summation, in terms of both structural stability and flexibility
(Norberg and Nilsson, 2000).
For MD simulations with stochastic boundary conditions (Brooks and
Karplus, 1983) the Langevin dynamics algorithm with a friction coefficient 50.0 ps1 for
the water oxygen atoms in the buffer region between R = 21 Å and
R = 23 Å was applied.
Simulations were performed in the microcanonical ensemble and the
temperature was periodically checked and kept within the window
300 ± 10 K.
In simulations with PBC, the entire system, with no restraints, was
heated from 50 K to 300 K over 6 ps. The temperature was progressively
increased by velocity scaling in steps of 50 K with 1 ps of MD at each
step. The simulations were continued to 700 ps and the trajectories
were saved for analysis. The initial 100 ps of the trajectories was
considered an equilibration period and the analyses of simulations were
performed on the next 600 ps of the production trajectory. Only data
concerning RMSD (Table 1 and Fig. 2) and
hydrogen bonds in the tetraloop (Table 5 and Fig. 7) come from the
analysis of the full 700-ps-long trajectory. Illustrations of
structures and structural motifs were made with the program MolScript
(Kraulis, 1991).
|
| |
RESULTS |
|---|
|
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
|
|---|
We have performed several molecular dynamics simulations of a 15-nucleotide RNA structure containing a tetraloop and a single nucleotide bulge in the stem part, in aqueous solution with sodium counterions. Two different models of the bulge location were used. Initial structures were built using both NMR and x-ray derived coordinates for the RNA motifs. The loop part in both structures was similar, but the structures differed in the bulge region. In one initial structure the bulge nucleotide was inside (bulge_in), and in the other it was outside the helix (bulge_out).
For the initial structure bulge_in, two independent simulations in a water box were performed (Sb1, Sb2). Simulations in the water box were compared with simulations in a water sphere (Ss3, Ss4, Ss5, Ss6). Two independent simulations, starting from the initial structure bulge_out, were run in the water box (Sb7, Sb8). All simulations are 700 ps long (Table 1).
Structural stability
We obtained stable trajectories both in the water box and in the water sphere without any systematic drift of the root mean square deviation (RMSD) from the initial structure (Fig. 2). The average global RMSD over the last 200 ps of MD was in the range 1.9-2.3 Å with one exception, in the simulation Ss6, where the final value was around 3.0 Å (Table 1). In all simulations the tetraloop part remained close to the initial structure, with an RMSD in the range 1.0-1.6 Å. Trajectories Sb2 and Ss6 showed the lowest and most stable RMSD for the tetraloop. As we will see later, in these cases we did not observe formation of a water-mediated hydrogen bond between bases G7 and A10.
Structural diversity
As a result of starting simulations from two initial structures and running several independent molecular dynamics trajectories for each of them, we obtained different possible conformations of the system. Details of the structures in the bulge region will be analyzed separately for simulations starting from initial structure bulge_in and bulge_out, whereas the tetraloop conformation which was similar for both initial structures, will be discussed for all simulations together.
Bulge region
To describe structures obtained from simulations starting from the initial structure bulge_in in the bulge region, particularly the orientation of the bulge base, we calculated distances between the N1 atom of the bulge uracil (U4) base and the N9 atom of adjacent bases: adenine (A3) and guanine (G5; Fig. 3 A).
|
Structures in the bulge region can be divided into two groups according to the distances between those nitrogens. One group (bul_in_1) represents structures where U4 is closer to G5 with the distance between N9(A3) and N1(U4) in the range 6.4-6.6 Å and the distance between N1(U4) and N9(G5) in the range 4.8-5.0 Å. The second group (bul_in_2) contains structures where U4 is closer to A3 with the distance between N9(A3) and N1(U4) in the range 4.5-4.9 Å and the distance between N1(U4) and N9(G5) in the range 5.8-6.0 Å.
In the structures belonging to the group bul_in_1 (simulations Sb1, Ss3, Ss4), the bulge nucleotide U4 is somewhat outside the helix in the minor groove side and the base is not exactly parallel to the neighboring adenine and guanine bases. The average angles between normal vectors of the bulge uracil base and adjacent adenine or guanine bases are similar and in the range of 25-33° (Fig. 3 B). The uracil base remains in close contact with the flanking adenosine base: (U4)O4...H-C2(A3) (Table 2 and Fig. 4 A). For this group of structures there is also a hydrogen bond (A3)O2'-H...O5'(U4) (Table 4).
|
|
In the structures from the group bulge_in_2 (simulations Sb2, Ss5, Ss6), the bulge nucleotide U4 stays stacked with A3 and a bend between U4 and G5 is observed (Fig. 4 B). The angle between the normal vectors of U4 and G5 is in the range 68-78° (Fig. 3 B). The bulge uracil base is hydrogen bonded with O2(U13) or O6(G5). Hydrogen bonds (A3)O2'-H...O4'(U4) or (C12)O2'-H...O5'(U13) along the backbone are found for two simulations, Sb2 and Ss5 (Table 4).
Structures obtained from the simulations starting from initial structure bulge_out (simulations Sb7, Sb8) do not represent a single well-defined conformation in the bulge region. The position of the bulge nucleotide both differs between simulations and changes along them. With the bulge base on the outside of the helix, stacking of the flanking bases A3 and G5 on each other is possible, and also occurs most of the simulation time. The bulge uracil base is not involved in any direct hydrogen bonds or stacking interaction. Differences in the position of the bulge phosphate group and the bulge base relative to the helix can be seen from the superposition of the two final structures (averaged over the last 20 ps of MD) obtained from simulations Sb7 and Sb8 (Fig. 5 C). Hydrogen bonds involving the 2'OH group of nucleosides G2, A3, U4 are found over the whole trajectories or only during part of simulations (Table 4).
|
The Watson-Crick basepairs flanking the bulge were maintained for most
of the simulation time (Table 2). A few reversible breakings of the
hydrogen bonds were observed in the minor grove side of these basepairs
for the simulations bulge_in. Two reversible breaking events took place for the hydrogen bond N1...H-N3 of
basepair A3-U13 (Ss4, Ss6). The base-pairing distortion observed in the simulation Ss6 was connected with rotation of the 
13 glycosyl torsion angle from anti toward the high anti (-sc) conformational region. For the simulations from the group
bulge_in_1, the relative orientation
of the A3 and U13 bases is similar to the U4 and U13 (range
23°-30°; Table 3). For the G5-C12
basepair, breaking of the N2-H...O2 and
partially of the N1-H...N3 hydrogen bond was observed during
the simulation Ss6. For this simulation the G5-C12 basepair has a less
optimal geometry due to the (U4)N3-H...O6(G6) hydrogen bond between
the bulge base and the subsequent base in the strand (Table 2). The
G5-C12 basepair is not planar and the angle between normal vectors of
the G5-C12 oscillates around 41°; the N4-H...O4 (G5-C12) hydrogen
bond also displays high fluctuations over the whole trajectory.
|
Hydrogen bond system in the loop
The tetraloop contains a sheared G-A basepair. There are up to four hydrogen bonds between G7-A10 and additional hydrogen bonds between O'2-H of G7 and N7 of G9 (Fig. 5 A). For most of the simulations (6 out of 8) the average distance between N3(G7) and H6-H (A10) is in the range 4.6-4.8 Å, which is too long for a standard hydrogen bond (Table 5). In these cases we found a water bridge between N3(G7), N6-H(A10), and N7(G9) created by a single water molecule (Fig. 6 A). This bridging water molecule was exchanged relatively slowly. For five simulations, we found the same water molecule in the bridge position for more than 400 ps. Dynamically stable tetraloop configurations with a direct hydrogen bond N3(G7) to N6 (A10), without water bridges in the loop, were also observed in the course of two 700-ps-long simulations, Sb2 and Ss6. (Fig. 6 B). During one simulation, Ss5, we observed a change of the loop arrangement and creation of water bridges (Fig. 7). The rearrangement to a direct hydrogen bond (G7)N3...H- N6(A10) from a water-mediated hydrogen bond was not observed.
|
|
For the structures with water bridges between bases in the loop (Sb1, Ss3, Ss4, Ss5b, Sb7, Sb8), average distances and angles of the (G7)N2-H...N7(A10) hydrogen bond are in the range 2.33 Å to 2.63 Å and 139° to 144°, respectively (Table 5). We found a stronger hydrogen bond between the N2 amino proton of G7 and the phosphate oxygen of A10: (G7)N2-H...O1P(A10). There is also a hydrogen bond between the N1 imino proton of G7 and the phosphate oxygen of A10: (G7)N1-H...O1P(A10), although its geometry in some cases is less optimal. The hydrogen bond (G7)O2'-H...N7(G9) in some simulations displays breaking events. For the structures without the water bridges between bases in the loop (Sb2, Ss5a, Ss6) there is no (G7)N1-H...O1P(A10) hydrogen bond but the remaining three hydrogen bonds (G7)N2-H...N7(A10), (G7)N2-H...O1P(A10) and (G7)O2'-H...N7(G9) are stronger.
The angle between normal vectors of the bases G7 and A10, which indicates the relative orientation of the bases, is in the 20-30° range. The G7-A10 bases with a water bridge between them are closer to planarity than in the conformation without the water bridge.
We did not observe hydrogen bonds (G7)O2'-H...O6(G9) or
(G7)O2'...H-N6(A10), identified in the NMR study of GNRA tetraloops
(Jucker et al., 1996).
During all simulations, stacking between the G9 and A10 bases in the loop is well maintained. In most of the simulations we also found that U8, the second base in the loop, stacks with G9. In only one case, simulation Sb8, did we observe a transformation to an unstacked conformation. This transformation was connected with a change of the sugar pucker of nucleotide U8 and occurred after 640 ps of simulation.
The A10 and C11 bases do not stack except in the simulation Ss6. Hydrogen bonds between the 2'OH group of A10 and a phosphate oxygen of C11 were noticed during two simulations, Sb2 and Sb8 (Table 4).
|
|
Backbone torsion angle and sugar conformation transformations
In Table 6 the regions for the
backbone torsion angles, which were different from A-form RNA, and
their main transition events are shown. We do not discuss here
transitions that took place during the initial 100 ps of simulation and
were mostly of an equilibration nature connected with adopting new
conformations different from the model built initial structure. Average
torsion angles for the backbone fragments 1_2, 14_15 in the 5' and
3'ends and for fragments 5_6_7, 9_10 have an A-form geometry in all
structures independent of the bulge arrangement (n-1_n denotes backbone
fragment C3'n-1-O3'n-1-Pn-O5'n-C5'n-C4'n
defined by torsion angles 
n-1, 
n-1, 
n,
n, 
n). We want to
point out that the fragment 5_6_7, consisting of three subsequent
guanosines, is very similar in all structures and shows the smallest
fluctuations of the torsion angles during simulations. The 9_10
fragment belonging to the loop part also displays the A-form geometry.
|
Deviations from the A-form and differences between structures are found
in the bulge region, in the loop region, and, for the
bulge_in simulations, for the fragment opposite
the bulge. In the bulge region for the bulge_out
simulations the flexible fragment encompasses the backbone fragment
2_3_4_5 where several transitions took place (Table 2). For simulations
bulge_in deviations from the A-form appear for
the fragments 4_5 and 5_6, and during 600 ps of the production run no
transitions were observed. In the loop part for the fragment 7_8, the
backbone changes direction as a result of 8
being in the +ap region and of a slight rotation of
8 from the +ap region toward the border of the
+ac/+ap region. During one simulation (Sb8), a transition of nucleotide
U8 from stacking to unstacking conformation was observed. This
transition involved a rotation of 8 from +ap
to +ac, a change of the sugar pucker from C3'-endo to C2'-endo and some
changes in the 8, 
8, and 9 torsion angles. An unstacked nucleotide
in this position was observed in the NMR structure of a GCAA loop
(Jucker et al., 1996). At step 8_9 only one structure, Ss4, differs
from the A-form with 9,
9, and 9 in the +ac,
ac, and +ap region, respectively. Variability in the conformation of
the backbone is observed at the A10:C11 step connecting the loop and
the stem part. We observed three kinds of arrangement of the backbone
torsion angles in this region. Conformations of the backbone with
torsion angles different than in A-form result in a lack of stacking
interactions between A10 and C11. Only during the Ss6 simulation do
torsion angles for the backbone fragment 10_11 have A-form geometry,
except for a temporary transition of the angles
10, 10,
11, and stacking between A10 and C11 is
maintained. For all bulge_in structures the stem
part opposite to the bulge (11_12_13_14) differs from the A-RNA
structure. The torsion angle is found in the ap conformation for at
least two of these nucleotides. For the bulge_out
simulations the torsion angles for the strand 11_12_13_14 opposite the
bulge appear in A-RNA geometry.
All structures display short distances between adjacent phosphate atoms P7-P8 (5.3-5.5 Å) and P6-P7 (5.4-5.7 Å), and there are water bridges between adjacent phosphate oxygen atoms in this fragment (J. Sarzynska, T. Kulinski, and L. Nilsson, manuscript in preparation). The remaining Pn-Pn+1 distances differ between positions and simulations. This leads to differences in the hydration pattern.
In general, sugar units are in the C3'-endo puckering mode, as in a
regular A-RNA helix. Most of the average values of the pseudorotation
phase angle P are in the range of 0° to 18°. There are three
regions where a transition from the C3'-endo to the C2'-endo puckering
was observed. In the bulge region we saw changing of the sugar
pucker for the bulge nucleotide U4 and the preceding nucleotide A3
during the simulation bulge_out. In the 3' strand opposite the bulge, sugar transformations took place for nucleotide C12
during the simulations bulge_in: one, Ss6, was
not reversible to the C2'-endo, and one, Ss4, was reversible to
C4'-exo. In the loop region only one transformation to the C2'-endo
puckering was noticed for nucleotide U8 in the loop. Sugar puckering
shifted to C2'-exo for nucleotides: G5 for
bulge_in_1 simulations, C11 for
bulge_out simulations, and C12 for simulation
Ss5. The glycosidic torsion angle about the C1'-N1 bond, , generally
was in the range 180° to 144°, corresponding to the anti
region. Extreme values in this region were observed for 4 (rotated
toward 144°) and 5 (rotated toward 180°). During a high
fraction of the time of the bulge_out
simulations, the 4 angle was higher than 120° and the 3 angle
was above 130°. In one case, for 13 during the Ss6
bulge_in simulation, we saw a temporary rotation
to the high anti (sc) region (to about 105°).
Sugar conformation changes from C3'-endo to C2'-endo, sometimes reversible, were observed at the 3' end in all simulations. For the G1-C15 terminal basepair, hydrogen bond breaking was observed only in one case, Ss3, after about 400 ps.
C-H...O contacts with C...O distances between 3.0 and 4.0 Å and C-H...O angles >90° are considered as contributing to the
stability of RNA (Wahl and Sundaralingam, 1997; Auffinger and Westhof,
1996). Base-backbone C(6/8)-H...O5' interactions are frequently
detected in the examined RNA. The geometry suitable for a C-H...O5'
interaction requires the torsion angle around C4'-C5' bond in the
region +sc, positioning the O5' oxygen atom under the sugar ring and the base in the anti conformation, which is typical for A-RNA. In our
simulations the C-H...O5' contact is well correlated with the
torsion angle . In all bulge_in simulations
this contact is broken very often for nucleotides 11-14, nucleotide 5 next to the bulge and in one case for nucleotide 9. In
bulge_out simulations it is not maintained for
nucleotides 3, 4, and 5 in the bulge region and for nucleotide 11.
Another C-H...O short contact often found in RNA is from C2'-H to O4' of the subsequent nucleotide. Generally this contact is weaker than C-H...O5'. In our simulations only the geometry in the turning point in the loop between nucleotides 7 and 8 is never suitable for the C2'-H...O4' interaction between sugars. The other regions display more variability between structures, but this contact is often detected. The C2'-H...O4' interaction is also destroyed due to the transformation of the sugar conformation from C3'-endo to the C2'-endo (simulation Ss6). The contact (U8)C5-H...O2'(G7) in the loop meets the criteria for a C-H...O contact for over 50% of each trajectory.
Structural flexibility
To identify regions of high and low mobility in the structures, we calculated RMS fluctuations around the average structure for the last 200 ps of the trajectories. RMS fluctuations were averaged per nucleotide separately for the backbone and base atoms (Fig. 8). Of special interest for us was how the bulge nucleotide arrangement influences the dynamics of the whole structure. We found that nucleotides G6 and G7 show the lowest fluctuations both for the backbone and base atoms in all simulations. Nucleotide G5 in the bulge_in simulations is also characterized by low mobility. In fact, nucleotide G7 is the first tetraloop nucleotide. The following three nucleotides from the tetraloop (nucleotides 8-10) display relatively large fluctuations, and the base atoms of the nucleotide U8 fluctuate more than the backbone atoms. The stem part opposite to the bulge (nucleotides 11-14) displays lower fluctuations in comparison to the mobile part of the tetraloop, but some structures from the bulge_in group exhibit quite high backbone fluctuations. For the bulge nucleotide U4 we can clearly see a difference in the mobility between structures bulge_in and bulge_out. Backbone fluctuations for nucleotide 4 in structures from the bulge_in group are similar to that for the adjacent nucleotide G5, and fluctuations of the base atoms are slightly higher for structures from the bulge_in_1 group than from the bulge_out_2 group. However, backbone fluctuations for the bulge nucleotide U4 in the structures from the group bulge_out are higher than for flanking nucleotides with much higher base fluctuations. The largest motion amplitude was found for the terminal nucleotides, with higher fluctuations at the 5' end.
|
Another way to provide insight into the atomic motions is to calculate the RMS fluctuations of the atom-atom distances. These calculations give us information about atom motions relative to the other atoms in the molecule. Of interest to us was the relative mobility of the backbone atoms. Fig. 9 displays examples of RMS fluctuations of the selected backbone atoms (P-O5'-C5'-C4'-C3'-O3') from the last 200 ps of the trajectories. The graphs represent three groups of structures. In the graphs only small fluctuations are displayed (0.0-0.5 Å). Small fluctuations (darker areas in Fig. 9) between atoms from neighboring nucleotides indicate local rigidity of the backbone. The backbone in the bulge region in the simulations Sb1 and Sb2 is more rigid than during the simulation Sb7. The stem fragment opposite the bulge (nucleotides 11-15) is clearly more rigid for structures from the bulge_out than for the bulge_in group. In the loop part for nucleotides, which show relatively high fluctuations around average positions (U8, G9, A10), fluctuation of atomic distances, are not relatively higher. This means that the backbone itself in the loop region is relatively rigid with large fluctuations as a unit around an average position in space. The highest distance fluctuations took place between atoms from nucleotides 2 and 9 and are connected with the motion of the 5' end relative to the tetraloop (data not shown).
|
In addition, the backbone flexibility has been analyzed by examination
of the presence of conformational changes leading to structures with
the O'2(i)...P(i+1)-O5'(i+1) angle
close to 180° and a short
O'2(i)...P(i+1) distance. Such an
in-line conformation is believed to be necessary for ion-induced
hydrolysis of the phosphodiester linkage (Brown et al., 1985). We did
not observe spontaneous fluctuations toward the in-line conformation
during the analysis of all phosphodiester linkages over 8 simulations. In our simulations the
O'2i...P'i+1-O5i+1
distance and angle are centered on 4.0 Å and 75° respectively. The
O'2i...
P'i+1-O5i+1 angles were
never larger than 155° and the O2' never came
closer than 3.05 Å to the cleavable phosphate. A small fraction of the conformations is characterized by angles in the region 100-155° (7%) and distances in the region 3.05-3.55 Å (3.6%). However, when
analyzing particular linkages, we can see that some of them are better
oriented for the cleavage reaction than others. Some angles and
distances display a wide range of values due to conformational variability. One example of this is shown in Fig.
10. Linkage 4_5 in the flexible
region is compared with linkage 6_7 in the rigid region during the two
bulge_out simulations (Sb7, Sb8). The
O'2i...P'i+1-O5i+1 angles for the 4_5 linkage are shifted to the highest values and the
O'2i...Pi+1
distances display a wide range of accessible values.
|
Solvent accessible surface area
The solvent accessible surface area (ASA; Lee and Richards, 1971)
was calculated using a spherical probe of radius 1.4 Å corresponding to the radius of a water molecule. Average ASAs over trajectories for
the RNA fragment from nucleotide 2 to 14 (without terminal nucleotides)
are between 2442-2562 Å2 with slightly higher
values for the bulge_out simulations (Sb7, Sb8).
The ASA of the studied RNA fragment is relatively stable throughout the
MD simulations, with a standard deviation of 30-38 Å2. The average ASA per nucleotide is in the
range 188-197 Å2. The contribution to the ASA
from the backbone and bases is about 75% and 25%, respectively.
To see the differences between regions of the 5S rRNA fragment, the
average ASAs for individual atoms were calculated. Fig. 11 shows ASAs for phosphate oxygens O1P
and O2P in each nucleotide averaged separately over
bulge_in and bulge_out
trajectories. The ASA for the O1P and O2P atoms is slightly larger
(from 5 to 15 Å2) in the 3' side for nucleotides
11 to 15 than in the 5' side for nucleotides 2 to 8. The O1P and O2P
atoms for nucleotides 9 and 10 in the loop part show relatively low
accessibility to water. In the region 12-14 in addition, we observed
an increase of the ASA for the O5' atoms. The ASA for the O5' atoms is
correlated with the backbone torsion angle around the C4'-O5' bond.
With in the +sc region, typical for A-RNA, the ASA of the O5' atom is close to 0. When rotates to the ap region, the ASA for the O5'
atom increases by about 5 Å2. This happens for
most of the nucleotides 11-14 for bulge_in simulations, for nucleotide C11 only for
bulge_out simulations, and for nucleotide G5 for
all simulations.
|
| |
DISCUSSION |
|---|
|
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
|
|---|
The object of this study is a terminal part of one of the arms of
5S rRNA from wheat germ corresponding to the nucleotide sequence from
81 to 95 of helix IV - loop D (Fig. 1). The consensus secondary
structure of this region consists of of a regular helix closed by a
tetraloop. This region is characterized by the lowest free energy of
the structure formation and was proposed as a center of the nucleation
for 5S RNA folding (Kulinski et al., 1997).
Initial structures used for MD simulations were constructed manually
according to a base-pairing scheme as proposed by (Kulinski et al.,
1997) with structural motifs taken from structural databases. We
assumed that the tetraloop sequence GCGA of loop D in the plant 5S rRNA
forms the structure typical for a GNRA loop. The GNRA loops are a
common motif in ribosomal RNA and their structures in solution were
characterized by NMR (Jucker et al., 1996). A similar, characteristic
structure was found in RNA molecules of different origin: hammerhead
ribozyme (Pley et al., 1994), group I ribozyme (Cate et al., 1996;
Kieft and Tinoco, 1997), leadzyme (Hoogstraten et al., 1998),
sarcin/ricin loop from 28S rRNA (Szewczak and Moore, 1995; Correll et
al., 1998, 1999). The unique known crystal structure of the D loop from
bacterial 5S rRNA exhibits some differences in the tetraloop GCGA
conformation in respect to the others previously described (Perbant et
al., 1998).
In general, the information about the structure of single pyrimidine
bulges in RNA is limited. In an early NMR study a uridine bulge was
found in an extra-helical conformation (van den Worm et al., 1998). A
more recent NMR study reveals a cytosine bulge located outside the
helix in the iron response element (IRE) RNA where the cytosine bulge
makes specific interactions with IRE protein (Addes et al., 1997;
Gdaniec et al., 1998) and in the HIV-1 genomic RNA (Puglisi and
Puglishi, 1998).
Single adenosine bulges in RNA were studied more extensively. From the
x-ray study of a viral coat protein/RNA complex, the adenosine bulge is
outside the helix (Valegård et al., 1997; van den Worm et al., 1998).
However, the NMR study of the RNA binding sequence for the coat protein
bacteriophage R17 reveals the adenosine bulge inside the helix (Borer
et al., 1995). In the crystal structure of a hybrid DNA/RNA the
adenosine bulge is located outside the helical track (Portmann et al.,
1996). Single purine bulges, where the bulged nucleotide was a
2-aminopurine, a fluorescent adenosine analogue, were studied using
time-resolved spectrofluorimetry. The results point to the coexistence
of two conformational states: one with 2-aminopurine looped out of the
helix, and another where the bulged base is stacked within helix
(Kulinski et al., 1996).
For that reason, we have built two initial models: in one structure the bulge was modeled inside, and in the other outside the helix. We used them to examine the stability of both configurations and their influence on the rest of the structure. To improve conformational sampling in the simulations, we carried out several trajectories for each of the initial structures with different initial conditions rather than a single long trajectory. The resulting structures kept the initial general bulge orientation and differed mainly in the bulge accommodation and the conformation of the stem part opposite the bulge. No tendency to continuous migration of the intrahelical U base toward the outside of the helix, nor in the reversed direction, was observed in any trajectory, indicating that there is a free energy barrier for this transition. From the present trajectories it is not possible to estimate neither the size of the barrier, nor the free energy difference between the two types of conformation. Structures with the bulge inside can be divided into two groups according to the position of the bulge nucleotide U4 relative to the flanking basepairs. For the structures with the bulge outside the helix a high backbone flexibility was observed in the bulge region.
Zacharias and Sklenar (1999) applied a hierarchical method and
energetic evaluation with a continuum solvent model for the conformational analysis of the bulge nucleotide. This method was based
on the assumption that the main conformational distortion in the double
helix due to the bulge encompasses only the bulge nucleotide and
nucleotides directly flanking the bulge side.
In our study, the way in which the bulge nucleotide is accommodated
into the helix affected the stem part opposite the bulge. In this
region the backbone conformation is extended. Also a sequence of the
stem fragment contributes to the difference in stability between the 5'
and the 3' sides. In the 5' strand there are successive purines and in
the 3' strand successive pyrimidines. The purines stack better than
pyrimidines (Saenger, 1984), which, on the one hand, makes the 3'
strand less rigid; on the other hand, this sequence seems to be more
susceptible to the conformational changes due to bulge accommodation.
Our results concerning the loop part are in agreement with what has
been obtained up to now. The 200-ps-long MD simulation of the GCAA
tetraloop performed by Zichi (1995) shows a good agreement of the
structural features derived from simulation with the NMR data. In that
simulation the formation of water-mediated hydrogen bonds between the
bases of the sheared G-A basepair in the tetraloop was observed. The
GUGA tetraloop as a part of a hammerhead ribozyme was investigated by
MD by Hermann et al. (1998). We have shown that the tetraloop can be
stabilized by a network of hydrogen bonds, which is dynamic and is
maintained even when some conformational rearrangements take place. In
our simulations we observed a direct as well as water-mediated
N6-H...N3 hydrogen bond between bases G7 and A10 in the tetraloop.
We also observed transitions from an arrangement with a direct hydrogen
bond to an arrangement with a water-mediated bridge.
The structure of two plant 5S rRNA species has been studied by
nonspecific lead-induced hydrolysis (Ciesiolka and Krzyzosiak, 1996).
This method is considered to be sensitive to the secondary structure
(Ciesiolka et al., 1998; Zagorowska et al., 1998). The 5S rRNA fragment
that was the subject of that study displayed a relatively weak cleavage
on both sides of the bulge nucleotide U4 and in the region spanning
nucleotides U8 and C12.
For the Pb2+-dependent hydrolysis of the RNA
backbone, the mechanism of intramolecular phosphoester transfer has
been proposed (Zhou and Taira, 1998). Such a reaction involves a
nucleophilic attack of the 2'O oxygen on the adjacent phosphorus center
and the production of a transient 2'-3' cyclic phosphate and 5'-OH terminus. It is believed that susceptibility to
Pb2+-induced hydrolysis of the polynucleotide
chain is determined by its flexibility and the ability to adopt an
in-line conformation (Ciesiolka, 1999). For that reason, the static
structure model is not sufficient to explain the experimental cleavage
pattern and a dynamic study can provide more useful information.
We can see some correlation between the dynamics of our models and the
cleavage pattern. Experiments show a lack of lead induced cleavage at
the three guanosine nucleotides from G5 to G7 (Ciesiolka and
Krzyzosiak, 1996). In our simulations this fragment appears as the most
rigid part of the molecule. In all structures independent of the bulge
arrangement, this fragment kept a classical A-RNA conformation, and in
this region we did not observe any transitions for the backbone torsion
angles. This fragment also shows the lowest RMS atomic fluctuations
around the average positions of the atoms. The region opposite the
bulge is where the cleavage is observed. For this region we have found
more possible conformations, and several changes of the conformations
during the course of the simulations were observed (rotations of the
torsion angles). Rotation of the backbone torsion angle from region
+sc to ap causes loss of C-H...O contacts, which normally stabilize
the structure, and increases the solvent exposure of O5'.
Soukup and Breaker (1999) observed a correlation between the
spontaneous RNA cleavage rate and the in-line fitness of the nucleotide
linkage. The conformational change that takes place at the cleavage
site of the hammerhead ribozyme during self-cleavage was captured using
x-ray crystallography (Murray et al., 1998). An MD study of the
hammerhead ribozyme showed that inducing a flip of the ribose pucker
from C3'-endo to C2'-endo at the ribose which holds the cleavable
phosphate moved a linkage to in-line conformation (Hermann et al.,
1998). That study suggested that conformational change at the cleavage
site, which takes place during the hydrolysis, can be induced by
transition in the sugar pucker.
In the course of our simulations we did observe spontaneous transitions of the sugar pucker. These transitions occurred in the sites susceptible to lead ion-induced hydrolysis. We have not observed spontaneous formation of the in-line conformation, but from the analysis of the O'2(i)... P(i + 1)-O5'(i + 1) angles and O'2(i)...P(i + 1) distances, differences in the accessible conformation between cleavage susceptible linkages and resistant linkages can be noticed.
The present simulations compare internal interactions and flexibility of the various regions of the models with the bulge localized inside the helix and the bulge pointing out from the helix. Our study does not definitively answer the question of whether the bulge base is stacked into the helix or remains outside. The two forms may exist in an equilibrium: the structure model with the bulge inside the helix better explains the cleavage observed in the part of the stem opposite the bulge than the model with the bulge outside, whereas the flexible backbone in the bulge region observed for the structures with the bulge outside the helix can explain cleavage localized in this part of the structure.
| |
ACKNOWLEDGMENTS |
|---|
This work was supported by the Swedish Institute, the Swedish Natural Science Research Council, the Magnus Bergvall Foundation, and the Swedish Research Council for Engineering Sciences.
| |
FOOTNOTES |
|---|
Received for publication 22 March 2000 and in final form 31 May 2000.
Address reprint requests to Lennart Nilsson, Center for Structural Biochemistry, Department of Bioscience at NOVUM, Karolinska Institutet, Hälsovägen 7, SE-141 57 Huddinge, Sweden. Tel.: 46-8-6089228, Fax: 46-8-6089290; E-mail: Lennart.Nilsson{at}biosci.ki.se.
| |
REFERENCES |
|---|
|
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
|
|---|
Biophys J, September 2000, p. 1213-1227, Vol. 79, No. 3
© 2000 by the Biophysical Society 0006-3495/00/09/1213/15 $2.00
This article has been cited by other articles:
 |
|
 |
|
  A. Barthel and M. Zacharias Conformational Transitions |
