Maturation of siRNA by strand separation: Steered Molecular dynamics study
Extracted source text · arXiv 1910.03778
\documentclass[preprint]{revtex4}
%\documentclass[aps,onecolumn,amssymb]{revtex4}
\topmargin .0in
\textheight 8.0in
\usepackage{graphicx}
\usepackage{epsfig}
\pdfoutput=1
\begin{document}
\title{Maturation of siRNA by strand separation:
Steered Molecular dynamics study}
\author{ Rakesh K Mishra$^{1}$, Sanchita Mukherjee$^{2}$, Dhananjay Bhattacharyya$^{*1}$ }
\affiliation{
$^{1}$ Computational Science Division, Saha Institute of
Nuclear Physics Kolkata, 700064, India.\\
$^{2}$ Department of Biological Sciences and Centre for Climate and Environmental Studies,
Indian Institute of Science, Education and Research-Kolkata, Mohanpur 741246, India }
\begin{abstract}
RNA interference, particularly siRNA induced gene silencing is becoming
an important avenue of modern therapeutics. The siRNA is delivered to the
cells as short double helical RNA which becomes single stranded for forming the
RISC complex.
Significant experimental evidence is available for
most of the steps except the process of the separation
of the two strands. We have attempted to understand the pathway for
double stranded siRNA (dsRNA) to single stranded (ssRNA) molecules using
steered molecular dynamics simulations. As the process is completely
unexplored we have applied force from all possible directions restraining
all possible residues to convert dsRNA to ssRNA. We found pulling one
strand along helical axis direction restraining far end of the other
strand demands excessive force for ssRNA formation.
Pulling a central
residue of one strand, in a direction perpendicular to the helix axis,
while keeping the base paired residue fixed requires intermediate force
for strand separation. Moreover we found that in this process the force
requirement is quite high for the first bubble formation (nucleation energy)
and the bubble propagation energies are quite small. We hypothesize this could
be the mode of action adopted by the proteins in the cells.
\end{abstract}
\maketitle
\section{introduction}
Ribonucleic acid (RNA) interference (RNAi) was first time discovered
in plants, but it was not noted in animals until Fire and Mello
demonstrated that double-stranded RNA (dsRNA) can cause greater
suppression of gene expression than single-stranded RNA (ssRNA) in
Caenorhabditis elegans \cite{Fire}. Due to the excellent gene silencing
potential of RNAi, it has attracted broad attention to
exploit its capabilities. In recent years, RNAi has become more
and more important in gene silencing and drug development because of
its high specificity, significant effect, minor side effects and ease
of synthesis \cite{Dogini}. When dsRNA enters the cell, it is first
cleaved into short double stranded fragments of $20-23$ nucleotide
silencing RNAs. These cleaved products have been recognized as
the small interfering RNAs (siRNAs) in the form of double stranded helices. They
are generally named as passenger strand and guide strand. Presumably the double helix
also remains complexed with
RNA-Induced Silencing Complex (RISC). In the RISC, the guide strand
of siRNA pairs with a complementary sequence in a messenger RNA (mRNA)
molecule and induces cleavage of mRNA by enzyme Argonaute. Thus, the
process of mRNA translation can be interrupted by siRNA \cite{Sashi, Martin, Meist, Li}.
Since rational design of siRNA can specifically inhibit endogenous and
heterologous genes, it can modulate any disease-related gene
expression. Following this strategical revelation, several synthetic
siRNA are being designed with desirable sequences to inhibit
any target gene expression \cite{Lopez,David,Fell}. The siRNAs undergoe
further processing inside the cell, where, one strand or part of one
(say guided strand ) gets separated from the other strand (say passenger strand).
This separation takes place by the force generated by some enzymatic
reaction. The naturally coded microRNA (miRNA) also goes through similar
steps for their action. Thus, the structure and force involved in the
separation of the strands become one of the important aspects
to deal with the efficiency of siRNA.
Double-helical deoxyribonucleic acid (dsDNA) has been widely studied
with respect to strand separation based on experiments and theory \cite{Garima,Kumar,
Essvaz,Bockel,Strunz,Schum,Danilo,Hatch,Cdan,Kuher,Marko}, where
researchers have studied the effects of mechanical force on the
structural changes of dsDNA. It has been shown that differences
in the chemical structure of dsDNA and dsRNA molecules affects the
intra-strand distances \cite{Alberto}. Recardo {\it et al.} \cite{Recardo}
have studied the effects of force using optical and magnetic
tweezers on the stretching of dsRNA and compared it with dsDNA
results. Lipfert {\it et al.} \cite{Lipfert} have studied the
effects of force and torque on the structural changes of long
stretch of dsRNA and pointed out striking differences between dsRNA
and dsDNA. Unfolding of compact structure of RNA was also studied
recently by various groups using experiment and simulation \cite{Mandal,Vishnu, Savinov, Manju}.
Nevertheless, siRNA evolves as one of the unprecedented small
bio-molecule, unlike large polymeric DNA or various RNA motifs,
requiring broad study of structural changes under the application
of external mechanical force. Oligomeric siRNA, having 20 to 22
base-pairs and UU overhang in both the strands requires special
attention in its structural changes during unzipping. It may be
mentioned that natural microRNAs (miRNA) also require unzipping
after they are processed by DICER protein, which may require some
assistance from proteins. We approach this novel problem of siRNA
strand separation by focusing our study on the opening of both the
strand of siRNA under the application of external mechanical force.
This mechanical force might come from some protein, such as Argonaute, DICER
or some other RNA binding protein, within the cell. However, we could
not find any report on action of such protein in the literature.
It may be noted that partial strand separation of dsDNA also takes place during
transcription initiation. It was found that the sigma-factor of RNA polymerase
is responsible for that \cite{Saecker, Zuo, Aayatti}. Thus, separation of strands of
double helical nucleic acid chains is an important aspect for understanding
several biochemical pathways.
In this report,
we present the effects of the pulling with the constant velocity
steered molecular dynamics (SMD) simulations under different
protocols as shown in Figure 1, broadly classifying them as axial
pulling and unzipping. We provided a comparative study using all
atom MD simulation, which may help in characterizing the structural
changes that occur during the pulling of double-stranded siRNA. We
address the problem with extensive studies of structural parameters,
hydrogen bonds (H-bonds) disruption and stacking interactions.
\begin{figure}[t]
\includegraphics[height=3.2in, width=3.3in]{fig1} \\
\caption{Schematic diagram showing different protocols studied in this
report: (a) Shows the Axial Rupture system, where force is applied along the
helical direction. In this case force is applied at $3'$-end of one strand
keeping $3'$-end of it's complementary strand fixed. (b) Shows the Axial
Stretch model, where force is applied along the axial direction of the
system. But, in this case pulling is applied at $3'$-end of one strand
while $5'$-end of the same strand being fixed. (c) Shows Terminal Unzip model,
where force is applied perpendicular to the helix direction. In this case,
pulling is applied at $5'$-end of one strand keeping $3'$-end of its
complementary strand fixed. Here, both pulling and fixed ends are in the
same side of the duplex. (d) Corresponds to the Central Unzip model, where
force is applied perpendicular to the helical direction at the central
residue of the system.}
\label{fig-1}
\end{figure}
The present studies, which are involved in calculation of path dependent
force and disruption of H-bonds by various protocols may shed the light to
provide future perspective of binding proteins with siRNA for gene silencing.
The analyses may also be important for designing more effective siRNA sequences.
The paper is organized as follows: in Section II, we describe the computational
protocol for simulation and analysis. Section III contains analysis of the
measured force on the strands and the reason behind such forces. In Section IV,
we have discussed variation of different structural parameters of the RNA.
Section V describes implication of the studies and how our results are
consistent with the available experimental data.
\section{Model and Method}
\subsection{Equilibrium MD simulation:}
{\bf Starting system :}
Well studied siRNA crystal structure with PDB ID 2F8S \cite{Yuan} is taken for all the
simulations. The structure is comprised of 22 nucleotides on each strand of the
duplex with characteristic UU overhang in both the $3'$-ends. The self-complementary
sequence is, $ 5' (AGACAGCAUAUAUGCUGUCUUU)_{2} $. The dimension of the duplex is
$\approx$ $8.0$ nm in length in normal double helical form.
{\bf Protocol:}
All molecular dynamics simulations were carried out using GROMACS 5.1 package
\cite{Abraham} and CHARMM36 Force Field \cite{Best}. The complete siRNA double
helical structure, without the Argonaute protein interacting through one end of
the RNA, was considered including the $5'$-terminal phosphate groups. The siRNA is solvated with
TIP3P water model in cubic box with sufficient dimension (10 nm $\times$ 10 nm $\times$ 10 nm)
and neutralized with 44 Na$^{+}$ ions. The system is then subjected to energy
minimization by steepest descent method to eliminate initial stress. For
initial equilibration, standard protocol of 100 ps each of NVT and NPT simulations
were done \cite{Abhijit}. Position restraints was applied to the RNA atoms
during equilibration of the system for both NVT and NPT processes. The siRNA
and non siRNA atoms were coupled to separate temperature coupling baths,
maintaining 300 K using Berendsen weak coupling method \cite {Beren}.
Final 100 ns NPT production MD run was conducted in absence of any restraints.
In the production run, the Nos´e-Hoover thermostat \cite{Nose,Hoover} was used
to maintain temperature and the Parrinello-Rahman barostat \cite{Nose, Hoover, Nose2,Rahman}
was used to isotropically regulate pressure. Periodic boundary conditions
(PBC) were employed for all simulations and the particle mesh Ewald (PME)
method \cite{Darden} was used for long-range electrostatic interactions.
The simulation time step was set to 2 fs with LINKS algorithm to maintain
bond lengths involving hydrogen atoms. Final structure from the end of 100 ns
equilibrium trajectory was used as starting configurations for pulling simulations.
{\bf Steered MD simulation:}
Four different steered molecular dynamics simulations (SMD) were carried out with
equilibrated conformations of siRNA along with solvents at 300 K. Henceforth we would
refer to them as: (i) Axial Rupture, (ii) Axial Stretch, (iii) Terminal
Unzip and (iv) Central Unzip. In case of Axial Rupture, shown in Figure 1a, force
is applied at the $3'$-terminal residue of one chain keeping the $3'$-terminal residue
of it’s complementary strand fixed (immobile) to their original position. For Axial
Stretch (Figure 1b), we have fixed the $5'$-terminal residue of one strand and
applied force on the $3'$-terminal residue of the same strand along the
helical direction of the system. In case of Terminal Unzip, shown in Figure 1c, we
have fixed the $3'$-terminal residue of one strand and force is applied on the $5'$-end
of its complementary strand along a direction perpendicular to the helical axis.
Here both the $3'$- and $5'$-terminal residues are in the same side of the double
helix. In Central Unzip case (Figure 1d) the force has been applied in the central point
(residue 11) of one strand keeping the central residue of the complementary strand
fixed. This is also the case, where force is applied perpendicular to the helical
direction. In case of pulling simulations, big rectangular boxes with dimensions
sufficient to satisfy minimum image convention for complete separation of
siRNA were generated. This provided space for the nearly elongated single stranded RNA along the Z-axis for
the rupture and along Y-axis (perpendicular to the Z-axis) for the unzipping.
We have adopted the boxes of the size 15 nm $\times$ 15 nm $\times$ 54 nm,
15 nm$\times$ 15 nm$ \times$ 54 nm, 15 nm $\times$ 40 nm $\times$ 15 nm and
15 nm$\times$ 40 nm $\times$ 15 nm, for Axial Rupture, Axial Stretch, Terminal Unzip
and Central Unzip respectively. These boxes were filled with TIP3P model of
explicit water with adequate Na+ counterions to neutralize the systems.
Equilibration was performed for 200 ps NVT and 10 ns NPT simulations,
using the same methodology described above prior to SMD simulations.
{\bf Protocol:}
SMD is based on applying external forces to particles in a selected direction
by adding a spring-like restraint, thus imitating directly the basic idea of
an AFM experiment through optical or magnetic tweezers.
The SMD simulations with constant velocity (CV) stretching (SMD-CV protocol)
were carried out by fixing one of the residues and applying external forces
to the dummy atoms attached to center of mass of another residue (SMD residue)
with a virtual spring. After several test simulations, we adopted a spring constant value of
$1000 \times {kJ} \times {{mol^{-1}} \times {nm^{-2}}}$
and a pulling rate of $ 0.0008 \times nm \times {ps^{-1}} $.
The force experienced by the pulled terminal residue, $ F $ is defined
as $F(t) = k(vt - x)$ where, $ x $ is the displacement of the pulled atom
from its original position, $ v $ is the pulling velocity, and $ k $ is
the spring constant.
The direction of pulling was such
that the end-to-end distance always increased, {\it i.e.}, the SMD residue
was pulled away from the fixed residues.
{\bf Analysis:}
Analysis of the trajectories, including finding number of H-bonds were done
by GROMACS 5.1 \cite{Abraham}. Base-pair orientation parameters and stacking geometry
were analysed by NUPARM \cite{Bansal,Shayantani,Pingali}.
\begin{figure}[t]
\includegraphics[height=3.5in, width=5.0in]{fig2}
\caption{Shows the variation of force experienced by the system with the
time under different protocol as shown in Figure 1. (a) Corresponds to the
variation of force with time for the Axial Rupture model. (b) Corresponds the
variation of force for the pulling of the Axial Stretch model. (c) Corresponds
the variation force for the pulling of the Terminal Unzip model and (d)
Corresponds the variation force applied for the Central Unzip model of siRNA.}
\label{fig-2}
\end{figure}
\section{Results analysis}
We first look at the stability of the equilibrium MD simulation. Previous
equilibrium molecular dynamics studies by several groups had observed that separation of the strands of
siRNA took place during interaction with graphene or carbon
nanotube \cite{Jung, Prabal,Santosh, Landry, Ghosh}. It was also demonstrated
that such separation of strands did not happen in case of double stranded DNA\cite{Prabal, Santosh}, thus attributed this siRNA separation to specific interaction between
carbon nanomaterial and siRNA. Our recent study, however indicate siRNA remains in double
helical form in physiological environment \cite{Abhijit}. This indicates that
external force is possibly needed to compel the separation of strands in siRNA.
The results compliments our equilibrium simulations, which also demonstrates
only moderate RNA breathing in equilibrium MD throughout the $100$ ns production run.
We have taken the final equilibrated structure of siRNA duplex from the equilibrium
MD simulation for further force induced SMD simulations. In SMD or center of
mass (COM) pulling the system is biased to demonstrate the behavior toward a
particular phenomenon. Application of an external force to cause displacement
in the simulated system allows for the calculation of work, a path-dependent
quantity. For opening of the strands of the siRNA we adopted two standard
protocols as described in method section (Figure 1), one is to apply force along
the helical direction of the system and the other is to apply the force
perpendicular to the helical direction of the system, the later can be generalised
as unzipping.
We have measured the forces experienced by the siRNA, which vary with time
for the different model systems (Figure 2). Figure 2a shows the variation of
force with time for the Axial Rupture model. We observe nearly linear variation
of extension with simulation time (Figure SI 1), hence the force {\it vs}. extension
curves also look very similar. In the case of Axial Rupture one strand is pulled
along helical axis direction keeping the far end of the other strand constrained.
During the initial phase of pulling, the $3'$-terminal single stranded UU residues
adopt stretched out conformation, which does not require any extra effort.
The double helical structure is also not affected by such conformational change
of the single stranded region.
After this phase, varying peaks of the curve reveal that structural transition
occurs and the system starts to change from its double-helical form. One expects
the hydrogen bonds (H-bonds) between the complementary bases to break at this
moment. The stacking interactions between successive base-pairs also may get
affected during this second phase of simulation with large force. The third
option is that both take place simultaneously during this phase. In order
to understand the mechanism, we have analysed number of H-bonds present in the
system at each time frame (Figure 3). Total number of H-bonds continue to
decrease with time in the first phase, {\it i.e.} upto 10 ns (Figure 3a, blue
curve). After about 10 ns, the number of H-bonds do not reduce significantly
with time, possibly indicating most of the phase transitions took place by 10 ns.
However, the number of H-bonds in the double helix does not reduce to zero value
within this time. As seen in Figure 2a, the force increases at this point of
time when number of H-bonds between the two strands appear to increase slightly.
After this phase transition, number of H-bonds slowly reduces to zero when
the strands dissociate completely (Figure 3a, blue line). But, after this
critical interaction the system breaks and the requirement of an additional
force starts to decrease. In this time duration, force
reduction from maximum to minimum reveals the structural change of the system
from the bound double helical state to the completely unbound ssRNA state. Both the strands
become almost separated, where most of the base-pairs are broken. It may be
noted that the unfolded single stranded chains has capability to form intra
strand H-bonds and hence, the blue line does not reach to zero value. This can
be visualised by the various snapshots shown in the first row of Figure 4.
Variation of force with time for the Axial Stretch model system is shown in
Figure 2b, where the constrained point and pulling end are on the same strand
and pulling is done along the helical direction (Figure 1b) somewhat similar
to the Axial Rupture model. Variation of force with time, however, is found to differ significantly
from that of Axial Rupture model. Here, slight increase in the force during
the last phase of SMD simulation is presumably due to conformational transition
of the pulled strand to somewhat all-trans geometry, which is obviously not
energetically favorable, especially for nucleic acids \cite{Bourne}.
\begin{figure}[t]
\includegraphics[height=2.0in, width=5.0in]{fig3}
\caption{ Variation of number of H-bonds in between the two strands of siRNA
with time during (a) Axial Rupture (blue line) and Axial Stretch (yellow
line) SMD simulations and (b) Terminal Unzip (green line) and Central
Unzip (red line) SMD simulations.}
\label{fig-3}
\end{figure}
Nevertheless, in this case completely separated strands do not arise
because pulling is done on the strand which is also fixed at the other
end. But, it is noticeable that, even then force induced breakage of
most of the base-pair H-bonds between the complementary strands take
place (yellow line in Figure 3a). This can also visualised by the
snapshots shown in second row of Figure 4. The helical structure is
converted to ladder like form (Figure 4), but most of the bases remain close
to their complementary ones of the opposite strand. Thus, some H-bonds
were retained till the end of the SMD simulation and further pulling
was impossible as that would need to break or stretch the covalent bonds. Possible such effort took place after around $8$ ns, showing increase in the measured force.
Variation of unzipping force with time for the Terminal Unzip model system
is shown in Figure 2c. Here, unzipping starts from one end and propagates to
the far side progressively. This kind of strand separation was also observed
by several groups in MD simulations of dsDNA at elevated temperature
\cite{Wong, Zgarbova,Sangeeta} or partially even at physiological condition. This can
be termed as extension of fraying or peeling effect. In this case
intact base-pairs are breaking gradually and progressively one by one, which
causes the system to experience almost equal small force in steps until
complete separation of the strand. And hence, variation of force is
nearly at constant small value at all time steps. The progression of simulation can be seen
by the different snapshots shown in first row of Figure 5. The base-pairs break
continuously as time progresses, which is also reflected in number of H-bonds
(Figure 3b, green curve). Significant reduction of number of H-bonds was
not observed till 8 ns, as the single stranded UU residues were changing
to stretched out conformation during this phase.
We observe unique result on the Central Unzip model system (Figure 1d),
where force is also applied perpendicular to the helical direction of the RNA
but in this case at the central residue (11th residue) of one of the strands
keeping the paired residue of the complementary strand immobile. Here, also
the variation of force experienced by the system with time is qualitatively
similar to that of terminal-unzip model system. However, it is notable that the measured force is significantly larger
(nearly 300 {kJ} {{mol$^{-1}$} {nm$^{-1}$}}) in the initial phase (Figure 2d) as compared to the other systems.
After the initial phase
the force reduces to smaller magnitude around 100 {kJ} {{mol$^{-1}$} {nm$^{-1}$}}
and even smaller values. This initial increase of force can be explained as
initiation of base-pair opening and the later as propagation of base pair
opening to both sides of the central one. This is equivalent to nucleation
energy for cooperative transition from helix to coil state. The nucleation
energy is quite high, as it would disrupt a base pair (at least two hydrogen
bonds) and two stacking interactions between the pulled base pair and its two
neighboring base-pairs on both sides. The second type of force is supposedly
stronger than the base pairing energy and it is doubled also \cite{Peter}.
Once the H-bonds of the central base-pair break and the stacking between the
central base-pair and its neighbouring base-pair are disrupted, the neighbouring
base-pairs can have fraying like effect. In other word these neighbouring base-pairs
come to contact with solvent water. Hence, These bases can form H-bond with the
complementary bases or with solvent water molecules in a competitive manner. The trajectory of this
model system can be visualised by the snapshots shown in the second row of Figure 5.
Furthermore, the propagation stage is quite faster as compared to the other SMD
results, as two base-pairs break together, {\it i.e.}, C+1 and C-1 after base
pair C breaks (where C is the central base-pair) then C+2 and C-2 break and
so on. Hence, they can now easily become single stranded breaking the Watson-
Crick base pairing and stacking interactions, requiring small amount of force.
\begin{figure}[htp]
\includegraphics[height=3.6in, width=6.7in]{fig4}
\caption{ Panel of axial force application: First row corresponds to the snapshots of
structural transition for Axial Rupture model system. Second row corresponds to the
snapshots of structural transition for the Axial Stretch model system. }
\label{fig-4}
\includegraphics[height=3.6in, width=6.7in]{fig5}
\caption{ Panel of force applied to the unzipping model systems: First row
corresponds the snapshots of structural transition for the Terminal Unzip model
system. Second row corresponds the snapshots of structural transition for
the Central Unzip model system. }
\label{fig-5}
\end{figure}
\section {structural transition}
As indicated above, the H-bonds between the complementary bases in a base-pairs
breaks during force induced SMD simulations. Thus the bases do not remain coplanar
to each other and the other degrees of freedom of the bases also increase beyond
their regular values. Quantitative analysis of these degrees of freedom of the bases
with respect to paired ones can be done by the six IUPAC-IUB recommended intra
base-pair parameters \cite{Olson}. We have therefore looked at shear, open
angle and stretch values, which are related to the H-bonding features. Similarly
relative orientations of a base-pairs with respect to their neighbouring stacked
ones also change significantly when the stacking interactions are disrupted.
These can be analysed by tilt, roll, etc., inter base-pair local parameters.
Effect of all these inter base-pair parameters can also be analysed by a
composite parameter, namely stacking overlap, and we have analysed that also.
Variation of shear, base-pair overlap and twist, as representative
parameters, are shown in Figure 6 and Figure 7 to compare all the systems.
\begin{figure}[htp]
\includegraphics[height=4.0in, width=7.2in]{fig6}
\caption{Variation of base pair parameters during equilibrium and axial pulling
simulations. Residue numbers are shown in x-axis, time in y-axis and the
graded color for values of (a) shear ({\r{A}}), (b) overlap ({\r{A}}$^{2}$)
and (c) twist ($^\circ$), for equilibrium simulation. The shear, overlap and
twist for Axial Rupture model are shown in (d), (e) $\&$ (f) while (g), (h)
$\&$ (i) are the same for Axial Stretch.}
\label{fig-6}
\includegraphics[height=3.2in, width=7.2in]{fig7}
\caption{Base pair parameters for unzipping SMD simulations. The base-pair
numbers are shown in x-axis, time (ns) in y-axis and the graded color represents
values of: (a) shear ({\r{A}}), (b) overlap ({\r{A}}$^{2}$) and (c) twist ($^\circ$)
for the for Terminal Unzip model and (c) shear ({\r{A}}), (d) overlap
({\r{A}}$^{2}$) and (e) twist ($^\circ$) for Central Unzip
model.}
\label{fig-7}
\end{figure}
The shear parameter gives information about relative movement of the bases with
respect to the paired ones, indicating disruption of hydrogen bonds in a base
pair, overlap provides information about stacking between two base-pairs and
twist indicates ladder like structure formation, and hence is often related to
stretching \cite{Alberto}. These are illustrated as three dimensional plots.
It may be noted that shear values of good Watson-Crick base-pairs are around zero \cite{Shayantani},
twist value of A-RNA stretches are around 33$^\circ$ \cite{Bourne} and stacking overlap between
successive base-pairs in RNA double helices are around 45 to 50 {\r{A}}$^{2}$ \cite{Pingali}
depending on base sequence. We have also analyzed variation of these parameters
during equilibrium MD simulations of the siRNA to understand the force induced
effects on the double helical structure (Figure 6a, b $\&$ c).
From the
figures, it is clear that the structure maintains almost ideal values of these
parameter throughout the simulation, indicating that no separation of the
strands took place in absence of any force or molecules like graphene or
carbon nanotube. Furthermore, terminal fraying is also found to be minimum as
compared to other simulations using CHARMM force-field \cite{Sangeeta}, possibly
due to somewhat capping effect by the two single stranded residues at the two ends
of the double helix.
The changes in the parameters for the Axial Rupture system are shown in Figure 6d, e $\&$ f.
Figure 6d illustrates that initially upto 4 ns shear values of all the 20
base-pairs are near $0$ (sky blue color), indicating no disruption of any
base-pair. After that minor disruption of base pairing can be seen with minor
fluctuation of shear values until 10 ns. Here, the shear values of 5th, 6th
and 7th base-pairs become more negative and can be seen to be more affected
by the force. Base-pairs of $5'$-terminal residue adopt large positive shear
and all the base-pairing near $3'$-terminal assume large negative shear. All
the shear values start changing because of structural transition after 12 ns
and afterwards huge fluctuation of shear (Yellow color corresponding to values
around $10 - 15$ {\r{A}} or deep blue color for shear values around $- 15$
{\r{A}} ) indicate disruption of initial base pairing. Comparing with disruption
of H-bonds (Figure 3a for Axial Rupture model) it can be concluded that, after
breaking half of the total possible H-bonds (from $\approx$ 48 to 25) the
disruption is steep linear after around 12 ns. Variation of overlap parameters
(Figure 6e) illustrates continuous disruption of stacking for all the base-pair
steps within 8 ns, and it is seen to initiate after 2 ns at the central region.
However, even at this time (8 ns) most of the bases are paired to their
complementary ones, as reflected from the analysis of shear. Thus, base-pairs opening and stacking disruption appear
to be independent and unrelated events. For few instances, we found that,
after complete breakdown of the stacking, again overlap value increases
due to single stranded helix like structure formation. Such single stranded
structure formation leads to stacking between successive bases, instead of
stacking between successive base-pairs in double-stranded RNA.
From Figure 6f it appears
that the twist values of the base-pairs at different helical positions start
to fluctuate at diverse time points. The middle base pair steps (9th and 10th)
faces twist disruption at as early as 2 ns. Comparing with the two other
parameters, shear and overlap, we found twist to get disrupted earliest and
become most sensitive to the applied force. It is noted that the central base-pairs
are getting effected earlier possibly due to weaker base-pairing in the central
region (AU rich sequence). These structural transitions can be seen with snapshots throughout the
simulation in first row of Figure 4.
Structural parameter variations for the Axial Stretch model of same strand
pulling are presented by Figure 6g, h $\&$ i. Interestingly, it can be
observed that in this case values of the parameters have complete different
signature of variation as compared to the Axial Rupture model. We found that
the regular shear variation is maintained for much longer time of pulling
for all the base-pairs as compared to Axial Rupture model (Figure 6d).
However, somewhat larger fluctuation of shear values is observed for few
base-pairs (14th to 17th) after 4 ns. These base-pairs are also seen to be
unstacked with respect to their neighbours from 4 ns (Overlap values reduce
to around 20 {\r{A}}$^{2}$). As expected twist value of these base-pairs
also reduce at the same time. Disruption in stacking overlap is also found
less for Axial Stretch model system (Figure 6h) as compared to Axial Rupture
model system (Figure 6e). Even at the end of Axial Stretch model simulation
significant stacking overlap around 30 {\r{A}}$^{2}$ is found between successive
base-pairs, indicating separation of the strands did not take place.
Comparing with the breaking of H-bonds for this case (Figure 3a, yellow line)
showing that even after the system is melted the number of H-bonds are still significant
($ > 10$ number), which can be observed by different snapshots shown in second
row of Figure 5. We find twist values (Figure 6i) are sensitive to the force
as compared to shear and overlap in Axial Stretch model. However, the
fluctuation starts late as compared to Axial Rupture model. These differences
in variations can be related to lesser disruption of H-bonds in Figure 3a
for axial-stretch model, indicating the situation that complete force induced
rupture is not taking place.
Variations of base pair parameter for Terminal Unzip model system are shown in
Figure 7a, b $\&$ c. In this case the fixed end and the end at which pulling force
is applied are on the same terminal side of the duplex and this unzipping mode
is prominent in the signature of parameter variations. Variation of shear
(Figure 7a) indicates that the disruptions of base-pairs are completely
dependent on the base positions. The terminal to the pulling end is first
disrupted by the unzipping force, which starts to perturb the system from the
pulling end gradually (19th base pair). Similar gradual disruption of overlap
values can be observed from Figure 7b. This illustrates that the unzipping
effect are progressive in nature and minimal effect is transmitted to the
bases far from the current unzipped base pair. This is almost equivalent to
extension of fraying effect seen earlier \cite{Sangeeta}.
Shear variation for the Central Unzip (Figure 7d), clearly indicates melting
of the central base pair occurs just after 1 ns. The base-pairs next to the
central pulled one (12th one) acquire large positive shear, while previous one
(10th base pair) gets large negative shear and this feature continuous till the
end. The base-pairs, which are away from the pulling point, maintain almost
same value (near zero, cyan color) for much longer time. Similar structural
changes can be observed in terms of stacking overlap value for all base
positions as shown in Figure 7e. The overlap values of the 6th residue and
14th residue appear to increase to value close to 35 {\r{A}$^{2}$} after 7 ns,
indicating formation of secondary helix-loop-helix like structure within the
separated single strands. This can be visualized by the snapshots (at 9 ns,
12 ns and 18 ns) in lower panel of Figure 5. In terms of variation of twist
value (Figure 7f) similar trend of separation of the strands starting from
central pulling positions is observed.
\section{Discussion }
In this study, we have tried to integrate structural transition of the
system with different protocols of applied force. We found that, when
pulling is applied at the axial direction (Axial Rupture) of the system,
it experiences highest force compared to all other protocols. In this
case highest force ($ \sim $ 600 {kJ} {{mol$^{-1}$} {nm$^{-1}$}}) builds
up to a point until certain critical interaction is broken. On the other
hand, for the same direction pulling (Axial Stretch) the force continuously
increases because the fixed end and pulled end are situated at the same
strand. Nevertheless, ultimate disruption of most of the H-bonds in
this case indicates that separation of the strands can also be achieved
by this way of steering. In case of pulling perpendicular to the helix
axis, the system unzips in usual way of strand separation. Here, opening
of base-pairs are progressive and is achieved one by one from the pulling
terminal. Due to this, during initial simulation time, force increases
and then maintains almost same value over the pulling time. Among the
protocols of pulling, Central Unzipping looks most different as compared
to others. In this case, initially system experiences strong force,
because of opening of first intact base-pairs and simultaneously unstacking
the central base pair from both the sides of pulled nucleotides. But,
eventually successive opening of progressive H-bonds requires much less
force compare to all other cases of pulling. This is also revealed by the
measurement of breaking of number of H-bonds. Hence this mode of opening
can be viewed as most feasible as compared to all other possible protocols
of siRNA strand separation.
We have also looked at the crystal structures of RNA double helical
fragments bound to argonaute protein from Protein Data Bank \cite{pdb} to evaluate
the most interactive structural signature of RNA. We found the protein bound
double helical RNA strands
are of significant length (10 or more base-pairs) in 3HJF,
3HK2, 3HM9, 4N47, 4NCB, 5AWH and 5UXO. We have analyzed hydrogen bonds between
the protein and RNA using PyrHBfind software \cite{PyrHBfind} and have focused
on the strong ones.
The middle portion of the RNA duplex is found to be mostly interacting
with protein residues by formation of very strong H-bonds involving
negatively charged phosphate group of RNA and positively charged Lys or Arg
residues of argonaute. In most of the cases, one of the ends of each
strand is also found similarly anchored. e.g. 3HJF.pdb \cite{Wang} and
3HK2.pdb \cite{Wang} have length of double helical region are 12 and 14
respectively and around 3 to 4 bases of both the strand residing at the
middle region of the helix in total forms H-bonds with the Protein. Hence,
accepting the idea that more crowding will enable higher grip by protein
and subsequently lead to rupture start-point, we can conclude that, central
pulling must be most feasible phenomena in nature. However, crowding at one
of the terminals may indicate that other mechanisms could also take place,
though with lesser probability. The complete understanding of anchoring the
siRNA at multiple points (center as well as the terminals) may require new methodological development to analyze the system.
Again, the role of UU overhang cannot be
detected in the unzipping of siRNA duplex, which might be involved in some
other process. Among the helical parameters, we observed that for all the case
of pulling, twist parameter is most sensitive during the opening of strands
of siRNA duplex. Experimental determination of siRNA structure in Protein
Data Bank is so far inadequate in number. In future perspective, it is
necessary to study the effects different sequences and length for the
system of siRNA and observe the changes associated.
\section{acknowledgements}
We gratefully acknowledge SERB and DST, New Delhi India for their
financial supports through project numbers PDF/2015/000308 and PDF/2017/002110/CS. We also acknowledged CDAC for their computing
support. Most of the simulations were performed in the cluster of
CAPP-II project of DAE. We thank Prof. Sanjay Kumar of Banaras Hindu
University and Prof. Rituparna Sinharoy of IISER-Kolkata for discussion
and useful suggestions.
\begin{thebibliography}{4}
\bibitem{Fire} Fire, A., Xu, S., Montgomery, M.K., Kostas, S.A., Driver, S.E. and Mello, C.C. (1998)
Potent and specific genetic interference by double-stranded RNA in Caenorhabditis elegans. {\it Nature}, {\bf 391}, 806-811.
\bibitem{Dogini} Dogini, D.B., Pascoal, V.D.A.B., Avansini, S.H., Vieira, A.S., Pereira, T.C. and Lopes-Cendes, I.
(2014) The new world of RNAs. {\it Genet. Mol. Biol.}, {\bf 37}, 285-293.
\bibitem{Sashi} Sashital, D.G. and Doudna, J.A. (2010) Current Opinion in Structural Biology. {\it Curr. Opin. Struct. Biol.}, {\bf 20}, 90-97.
\bibitem{Martin} Martinez, J., Patkaniowska, A., Urlaub, H., Lührmann, R., Tuschl, T. (2002) Single-stranded antisense siRNAs guide target RNA cleavage in RNAi.
{\it Cell}, {\bf 110}, 563-74.
\bibitem{Meist} Meister, G. and Tuschi, T. (2004) Mechanisms of gene silencing by double-stranded RNA. {\it Nature}, {\bf 431}, 343-9.
\bibitem{Li} Li, J., Xue, S. and Mao, Z.W. (2016) Nanoparticle delivery systems for siRNA-based therapeutics. {\it J. Mater. Chem. B}, {\bf 4}, 6620-6639.
\bibitem{Lopez} Wu, S. Y., Lopez-Berestein, G., Calin, G.A. and Sood, A.K. (2014) RNAi therapies: drugging the undruggable. {\it Sci. Transl. Med.}, {\bf 6}, 240.
\bibitem{David} Davidson, B.L. and Jr. McCray, P.B. (2011) Current prospects for RNA interference-based therapies. {\it Nat. Rev. Genet.}, {\bf 12}, 329-40.
\bibitem{Fell} Fellmann, C. and Lowe, S.W. (2014) Stable RNA interference rules for silencing. {\it Nat. Cell Biol.}, {\bf 16}, 10-18.
\bibitem{Garima} Kumar, S., Jensen, I., Jacobsen, J.L. and Guttamann A.J. (2007) Role of conformational entropy in force-induced biopolymer unfolding.
{ \it Phys. Rev. Lett.}, {\bf 98}, 128101.
\bibitem{Kumar} Kumar, S. and Li, M.S. (2010) Biomolecules under mechanical force. {\it Phys. Rep.486}, {\bf 1}, 1-74.
\bibitem{Essvaz} Essevaz-Roulet, B., Bockelmann, U. and Heslot, F. (1997) Mechanical separation of the complementary strands of DNA.
{\it Proc. Natl. Acad. Sci. USA}, {\bf 94}, 11935-11940.
\bibitem{Bockel} U. Bockelmann, Essevaz-Roulet, B. and Heslot, F. (1997) Molecular Stick-Slip Motion Revealed by Opening DNA with Piconewton
Forces. {\it Phys. Rev. Lett.}, {\bf 79}, 4489.
\bibitem{Strunz} Strunz, T., Oroszlan, K., Sch\"{a}fer, R. and G\"{u}ntherodt, H.J. (1999) Dynamic force spectroscopy of single DNA molecules.
{\it Proc. Natl. Acad. Sci. USA}, {\bf 96}, 11277-82.
\bibitem{Schum} Schumakovitch, I., Grange, W., Strunz, T., Bertoncini, P., G\"{u}ntherodt, H.J. and Hegner, M. (2002)
Temperature dependence of unbinding forces between complementary DNA strands. {\it Biophys. J.}, {\bf 82}, 517-521.
\bibitem{Danilo} Danilowicz, C., Kafri, Y., Conroy, R.S., Coljee, V.W., Weeks, J. and Prentiss, M. (2004 Measurement of the Phase
Diagram of DNA Unzipping in the Temperature-Force Plan. {\it Phys. Rev. Lett.}, {\bf 93}, 078101.
\bibitem{Hatch} Hatch, K., Danilowicz, C., Coljee, V. and Prentiss, M. (2008) Demonstration that the shear force required to separate
short double-stranded DNA does not increase significantly with sequence length for sequences longer than 25 base pairs. {\it Phys. Rev. E}, {\bf 78}, 011920.
\bibitem{Cdan} Danilowicz, C., Limouse, C., Hatch, K., Conover, A., Coljee, V.W., Kleckner, N. and Prentiss, M. (2009)
The structure of DNA overstretched from the $5'5'$ ends differs from the structure of DNA overstretched from the $3'3'$ ends.
{\it Proc. Natl. Acad. Sci. USA}, {\bf 106}, 13196-13201.
\bibitem{Kuher} K\"{u}hner, F., Morfill, J., Neher, R.A., Blank, K. and Gaub, H.E. (2007) Force-Induced DNA Slippage. {\it Biophys. J.}, {\bf 92}, 2491-2497.
\bibitem{Marko} Cocco, S., Monasson, R. and Marko, J.F. (2001) Force and kinetic barriers to unzipping of the DNA double helix.
{\it Proc. Natl. Acad. Sci. USA}, {\bf 98}, 8608-8613.
\bibitem{Alberto} Marin-Gonzaleza, A., Vilhenaa, J.G., Perezb, R. and Moreno-Herreroa, F. (2017) Understanding the mechanical
response of double-stranded DNA and RNA under constant stretching forces using all-atom molecular dynamics. {\it Proc. Natl. Acad. Sci. USA}, {\bf 114}, 7049-7054.
\bibitem{Recardo} Herrero-Gal\'{a}n, E., Fuentes-Perez, M.E., Carrasco, C., Valpuesta, J.M., Carrascosa, J.L., Moreno-Herrero, F.
and Arias-Gonzalez, J.R. (2013) Mechanical Identities of RNA and DNA Double Helices Unveiled at the Single-Molecule Level.
{\it J. Am. Chem. Soc.}, {\bf 135}, 122.
\bibitem{Lipfert} Lipfert, J. Skinner, G.M. {\it et al.} (2014) Double-stranded RNA under force and torque: Similarities to and
striking differences from double-stranded DNA. {\it Proc. Natl. Acad. Sci. USA}, {\bf 111}, 15408-413.
\bibitem{Mandal} Mandal, M., Lee, M. Barrick, J.E., Weinberg, Z., Emilsson, G.M., Ruzzo, W.L., Breaker, R.R. (2004)
A glycine-dependent riboswitch that uses cooperative binding to control gene expression. {\it Molecu. Bio.}, {\bf 306}, 275-279.
\bibitem{Vishnu} Chandra, V. Hannan, Z., Xu, H. and Mandal, M. (2017) Single-molecule analysis reveals multi-state folding of a
guanine riboswitch. {\it Nature Chem. Biol.}, {\bf 13}, 194-201).
\bibitem{Savinov} Savinov, A., Perez, C.F. and Blockc, S.M. (2014) Single-molecule studies of riboswitch folding.
{\it Biochim. Biophys. Acta.}, {\bf 1839}, 1030-1045.
\bibitem{Manju} Gupta, A. and Bansal. M. (2017) The Role of Sequence in Altering the Unfolding Pathway of an RNA Pseudoknot:
A Steered Molecular Dynamics Study. {\it Phys. Chem. Chem. Phys.}, {\bf 18}, 28767-28780.
\bibitem{Zuo} Zuo, Y. and Steitz, T.A. (2015) Crystal Structures of the E. coli Transcription
Initiation Complexes with a Complete Bubble {\it Molecular Cell}, {\bf 58}, 534-540.
\bibitem{Aayatti} Mallick Gupta, A., Mukherjee, S., Dutta, A., Mukhopadhyay, J., Bhattacharyya, D. and Mandal, S. (2017) Identification of a suitable promoter for the sigma
factor of Mycobacterium tuberculosis. {\it Mol. BioSystems} {\bf 13}, 2370-2378.
\bibitem{Saecker} Saecker, R.M., Record Jr, M.T. and deHaseth, P.L. (2011) Mechanism of Bacterial Transcription Initiation:
RNA Polymerase - Promoter Binding, Isomerization to Initiation-Competent Open Complexes, and
Initiation of RNA Synthesis {\it J. Mol. Biol.} {\bf 412}, 754-771.
\bibitem{Yuan} Yuan, Y.R., Pei, Y., Chen, H.Y., Tuschl, T., and Patel, D.J. (2006) A Potential Protein-RNA Recognition Event
along the RISC-Loading Pathway from the Structure of A. aeolicus Argonaute with Externally Bound siRNA. {\it Structure} {\bf 14}, 1557-1565.
\bibitem{Abraham} Abraham, M.J., Murtola, T., Schulz, R., Pall, S., Smith, J.C., Hess, B. and Lindahl, E. (2015)
GROMACS: High performance molecular simulations through multi-level parallelism from laptops to supercomputers. {\it SoftwareX}, {\bf 1-2}, 19-25.
\bibitem{Best} Best, R.B., Zhu, X., Shim, J., Lopes, P.E.M., Mittal, J., Feig, M. and MacKerell, A.D. (2012) Optimization of the Additive
CHARMM All-Atom Protein Force Field Targeting Improved Sampling of the Backbone $\psi$, and Side-Chain $\chi^{1}$ and $\chi^{2}$ Dihedral Angles.
{\it J. Chem. Theory Comput.}, {\bf 8}, 3257-3273.
\bibitem{Abhijit} Biswas†, A., Chakraborty, K., Dutta, C., Mukherjee, S., Gayen, P., Jan, S., Mallick, A.M., Bhattacharyya, D. and Sinha Roy, R. (2019)
Engineered Histidine-Enriched Facial Lipopeptides for Enhanced Intracellular Delivery of Functional siRNA to Triple Negative Breast Cancer Cells.
{\it ACS Appl. Mater and Interfaces}, {\bf 11}, 4719-4736.
\bibitem{Beren} Berendsen, H.J.C., Postma, J.P.M., van Gunsteren, W.F., DiNola, A. and Haak, J.R. (1984)
Molecular dynamics with coupling to an external bath. {\it J. Chem. Phys.}, {\bf 81}, 3684.
\bibitem{Nose} Nos\'{e}, S. (1984) A unified formulation of the constant temperature molecular dynamics methods. {\it J. Chem. Phys.}, {\bf 81}, 511.
\bibitem{Hoover} Hoover, W.G. (1985) Canonical dynamics: Equilibrium phase-space distributions. Canonical dynamics:
Equilibrium phase-space distributions. {\it Phys. ReV. A}, {\bf 31}, 1695.
\bibitem{Nose2} Nos\'{e}, S., Klein, M.L. (1983) Constant pressure molecular dynamics for molecular systems. {\it Mol. Phys.}, {\bf 50}, 1055-1076.
\bibitem{Rahman} Parrinello, M. and Rahman, A. (1981) Polymorphic transitions in single crystals: A new molecular dynamics method.
{\it J. Appl. Phys.}, {\bf 52}, 7182.
\bibitem{Darden} Darden, T., York, D. and Pedersen, L. (1993) Particle mesh Ewald: An N⋅log(N) method for Ewald sums in large systems.
{\it J. Chem. Phys.}, {\bf 98}, 10089.
\bibitem{Bansal} Bansal, M., Bhattacharyya, D. and Ravi, B. (1995) NUPARM and NUCGEN: software for analysis and generation of sequence
dependent nucleic acid structures. {\it Bioinformatics}, {\bf 11}, 281-287.
\bibitem{Shayantani} Mukherjee, S., Bansal, M. and Bhattacharyya, D. (2006) Conformational specificity of non-canonical base pairs and
higher order structures in nucleic acids: crystal structure database analysis. {\it J. Comput. Aided Mol. Des.}, {\bf 20}, 629-45.
\bibitem{Pingali} Pingali, P.K, Halder, S, Mukherjee, D., Basu, S. Banerjee, R. Choudhury, D. and Bhattacharyya, D. (2014)
Analysis of stacking overlap in nucleic acid structures: algorithm and application. {\it J. Comput. Aided Mol. Des.}, {\bf 28}, 851-867.
\bibitem{Jung} Jung, S., Cha, M., Park, J., Jeong, N., Kim, G., Park, C., Ihm, J. and Lee, L. Dissociation of Single-Strand DNA:
Single-Walled Carbon Nanotube Hybrids by Watson−Crick Base-Pairing. {\it J. Am. Chem. Soc.}, {\bf 132}, 10964-10966.
\bibitem{Prabal} Santosh, M., Panigrahi, S., Bhattacharyya, D., Sood, A. K. and Maiti, P. K. (2012) Unzipping and binding of small
interfering RNA with single walled carbon nanotube: a platform for small interfering RNA delivery. {\it J. Chem. Phys.}, {\bf 136}, 065106.
\bibitem{Santosh} Santosh, M., Panigrahi, S., Bhattacharyya, D., Sood, A.K. and Maiti, P.K. (2012)
Unraveling siRNA unzipping kinetics with graphene. {\it J. Chem. Phys.}, {\bf 137}, 054903.
\bibitem{Landry} Landry, M.P., Vukovic, L., Kruss, S., Bisker, G., Landry, A.M., Islam, S., Jain, R., Schulten, K. and Strano, M.S.
(2015) Comparative Dynamics and Sequence Dependence of DNA and RNA Binding to Single Walled Carbon Nanotubes. {\it J. Phys. Chem. C.}, {\bf 119}, 10048-10058.
\bibitem{Ghosh} Ghosh. S. and Chakrabarti, R. (2016) Spontaneous Unzipping of Xylonucleic Acid Assisted by a Single-Walled
Carbon Nanotube: A Computational Study. {\it J. Phys. Chem. B.}, {\bf 120}, 3642-3652.
\bibitem{Bourne} Needle, S., Schneider, B., and Berman, H.M. {\it Structural Bioinformatics, 2nd ed} (P.E. Bourne, H. Weissig, eds), pp.41-76 (2009).
\bibitem{Wong} Wong, K.Y. and Pettitt, B.M. (2008) The Pathway of Oligomeric DNA Melting Investigated by Molecular
Dynamics Simulations. {\it Biophys. J.}, {\bf 95}, 5618-5626.
\bibitem{Zgarbova} Zgarbova, M., Otyepka, M., Sponer, J., Lankas, F. and Jurecka, P. (2014) Base Pair Fraying in Molecular Dynamics
Simulations of DNA and RNA. {\it J. Chem. Theory Comput.}, {\bf 10}, 3177-3189.
\bibitem{Sangeeta} Kundu, S., Mukherjee, S. and Bhattacharyya. D. (2017) Melting of polymeric DNA double helix at elevated temperature:
a molecular dynamics approach. {\it J. Mol. model}, {\bf 23}, 226.
\bibitem{Peter} Yakovchuk, P., Protozanova, E. and Frank-Kamenetskii, M.D. (2006) Base-stacking and base-pairing contributions into
thermal stability of the DNA double helix. {\it Nucl. Acids Res.}, {\bf 34 }, 564-574.
\bibitem{Olson} Olson, W.K., Bansal, M., Burley, S.K., Dickerson, R.E., Gerstein, M., Harvey, S.C., Heinemann, U., Lu, X.J., Neidle, S., Shakked,
Z., Sklenar, H., Suzuki, M., Tung, C.S., Westhof, E., Wolberger. C. and Berman, H.M. (2001) A standard reference frame for the description of nucleic acid
base-pair geometry. {\it J. Mol. Biol}, {\bf 313}, 229-237.
\bibitem{Wang} Wang, Y., Juranek, S., Li, H., Sheng, G., Wardle, G.S., Tuschl, T. and Patel, D.J. (2009) Nucleation, propagation and cleavage
of target RNAs in Ago silencing complexes. {\it Nature}, {\bf 461}, 754-761.
\bibitem{pdb} Berman, H.M., Westbrook, J., Feng, Z., Gilliland, G., Bhat, T.N., Weissig, H., Shindyalov, I.N., Bourne, P.E. (2000) The Protein Data Bank.{\it Nucl. Acids Res.} {\bf 28}, 235-252.
\bibitem{PyrHBfind} Mukherjee, S., Majumdar, S. and Bhattacharyya, D. (2005) Role of Hydrogen Bonds in Protein-DNA Recognition: Effect of Nonplanar Amino Groups. {\it J. Phys. Chem. B} {\bf 109}, 10484-10492.
\end{thebibliography}
\end{document}