The role of internal chain dynamics on the rupture kinetics of adhesive contacts
Extracted source text ยท arXiv 0805.1011
%\documentclass[twocolumn,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}
\documentclass[preprint,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}
% Some other (several out of many) possibilities
%\documentclass[preprint,aps]{revtex4}
%\documentclass[preprint,aps,draft]{revtex4}
%\documentclass[prb]{revtex4}% Physical Review B
\usepackage{graphicx}% Include figure files
\usepackage{dcolumn}% Align table columns on decimal point
\usepackage{bm}% bold math
\usepackage{epstopdf}
%\nofiles
%\documentstyle[aps,amssymb,amstex,epsf,floats]{revtex}
%\oddsidemargin 0in \evensidemargin 0in \textwidth 6.5 in
%\renewcommand{\topfraction}{.99}
%\renewcommand{\textfraction}{.01}
%\renewcommand{\baselinestretch}{1.6}
%\newcommand{\sign}{\mathop{\rm sign}}
%%\input psfig
%%\psfigurepath{Figures/}
%\psdraft
\begin{document}
\title{The role of internal chain dynamics on the rupture kinetics\\
of adhesive contacts}
\author{V. Barsegov$^1$, G. Morrison$^{2}$ and D. Thirumalai$^{2}$}
\affiliation{$^1$Department of Chemistry, University of Massachusetts, Lowell, MA 01854\\
$^2$Biophysics Program, Institute for Physical Science and Technology,
University of Maryland, College Park, MD 20742}
%\pacs{}
%\date{\today}
%%\maketitle
\begin{abstract}
\noindent
We study the forced rupture of adhesive contacts between monomers that are not
covalently linked in a Rouse chain. When the applied force ($f$) to the chain end
is less than the critical force for rupture ($f_c$), the {\it reversible} rupture
process is coupled to the internal Rouse modes. If $f/f_{c}$$>$$1$ the rupture
is {\it irreversible}. In both limits, the non-exponential distribution of contact
lifetimes, which depends sensitively on the location of the contact, follows the
double-exponential (Gumbel) distribution. When two contacts are well separated
along the chain, the rate limiting step in the {\it sequential} rupture kinetics is the
disruption of the contact that is in the chain interior. If the two contacts are
close to each other, they cooperate to sustain the stress, which results in an
``all-or-none'' transition.
\end{abstract}
\maketitle
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Nanomanipulation of biological molecules by force can be used to explore and control
their energy landscapes at the single molecule level
\cite{BlockAnnRevBiophys2007TinocoPNAS07FernandezPNAS07}.
%The ability to probe the
%features of the landscape that are inaccessible in conventional experiments illustrates
%the power of single molecule methods.
The challenge is to solve the inverse problem, namely, to extract the unbinding pathways
and structural features pertaining to the systems of interest from measurements such as the
force-extension curves, lifetime distribution of adhesive contacts, and unbinding force distributions.
From these data, many characteristics of the energy landscape (roughness, barrier height,
and movement of the transition states) have been obtained
using theoretical models \cite{HyeonPNAS03DudkoPNAS03,BarsegovPRL05}.
In these models, the force-induced unbinding transitions from the bound state (for
example folded conformations of RNA) are presumed to occur \textit{irreverisbly} along a
one dimensional reaction coordinate, $R$, that is conjugate to the applied force, $f$. In this
picture, $f$ ``tilts" the energy landscape, thus rendering the bound state unstable when $f$
exceeds a critical value. On the other hand, detailed simulations of forced unfolding of proteins
and RNA \cite{Hyeon06BiophysJ} show that collective topology-dependent events that
describe the internal dynamics \cite{BarsegovPRL05} are coupled to $R$. Thus, it is important
to develop solvable models for force-induced disruption of ``folded" structures starting
from a well-defined Hamiltonian.
In this letter, we illustrate the interplay between internal polymer dynamics and the rupture
kinetics of intramolecular adhesive contacts between monomers. Such a model is a caricature of
forced-unfolding of RNA hairpins \cite{BlockAnnRevBiophys2007TinocoPNAS07FernandezPNAS07}
where the adhesive contact mimics the stability of the stem.
Using an exactly solvable Rouse model, with one or two adhesive contacts between monomers
that are not covalently linked, we address the following questions: (i) For a single adhesive
contact, how do the lifetime distribution functions vary with the backbone distance between
monomers $r$ and $s$ that are in adhesive contact? (ii) To what extent is the binding of a
ruptured contact influenced by $f$ and the internal polymer motions?
%(iii) What are the sequence of unbinding
%events when the locations of two adhesive contacts, one between monomers $r$ and $s$ and
%the other between $r^{\prime}$ and $s^{\prime}$, are alter
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
{\it The model:}
Consider a Rouse chain with $N$ identical monomers, connected by a harmonic spring with
the equilibrium distance $a$, with a single adhesive contact $(r,s)$ between
two non-covalently linked monomers $r$ and $s$, $1$$\le$$r$$<$$s$$\le$$N$. The first monomer
is fixed, and a constant force ${\bf f}$$=$$f$${\bf x}$ is applied to the $N^{th}$ monomer in the
direction parallel to the end-to-end vector ${\bf X}$. Letting ${\bf R}_m$$=$$\{ R_m^x,R_m^y,R_m^z \}$
be the monomer position, ($m$$=$$1,2,\ldots,N$),
the equation of motion is,
%%
%\begin{equation}\label{1.1}
$\xi {{d}\over {dt}}{\bf R}_m$$=$$-\nabla_{{\bf R}_m}$$H(\{ {\bf R}_m \})$$+$${\bf F}_m$
%\end{equation}
%%
where $\nabla_{{\bf R}_m}$$\equiv$$\partial/\partial {\bf R}_m$, $\xi$ is the friction cefficient,
and ${\bf F}_m$ is the random force with zero mean and
$\langle F_m^{\alpha}(t)F_n^{\beta}(0)\rangle $$=$$2\xi k_B T\delta_{mn}\delta_{\alpha \beta}\delta(t)$
($\alpha$, $\beta$$=$$x$, $y$, $z$). We model the adhesive contact $(r,s)$
using a harmonic potential with a spring constant, $\kappa$$=$$3k_B T/b^2$, where $b$ is the
equilibrium contact distance. The Hamiltonian with a single contact under tension is taken to
be $H$$=$$H_0$$+$${\bf f_{N}}({\bf R}_N-{\bf R}_1)$ where
%%
\begin{equation}\label{1.3}
H_0 = {{3k_B T}\over {2a^2}}\sum_{n=2}^{N}({\bf R}_n-{\bf R}_{n-1})^2 + {{3k_B T}\over {2b^2}}\Theta(B-|{\bf R}_{rs}|){\bf R}_{rs}^2
\end{equation}
%%
with ${\bf R}_{rs}$$=$${\bf R}_{r}$$-$${\bf R}_{s}$, and $\Theta(x)$ is the Heaviside step function. The
contact is ruptured when $|{\bf R}_{rs}|$$\ge$$B$. The probability density function (pdf) of the
$(r,s)$-contact lifetimes is obtained as $p_{rs}(t,f)$$=$$-d P_{rs}(t,f)/dt$,
%$p_{rs}(t)$, is obtained as
%%
%\begin{equation}\label{1.5}
%p_{rs}(t,f)=-{{d}\over {dt}}P_{rs}(t,f),
%\end{equation}
%%
where $P_{rs}(t)$, the probability that the adhesive contact remains intact, is
%%
%\begin{equation}\label{1.6}
$P_{rs}(t,f)$$=$${{\int_{|{\bf R_{rs}}|\le B} d^3{\bf R}_{rs}
\int d^3{\bf R}_{rs}^0 Q_{f}({\bf R}_{rs},t;{\bf R}_{rs}^0) P_{I}({\bf R}_{rs}^0)}
\over {\int d^3{\bf R}_{rs}
\int d^3{\bf R}_{rs}^0 Q_{f}({\bf R}_{rs},t;{\bf R}_{rs}^0)P_{I}({\bf R}_{rs}^0)}}$.
%\end{equation}
%%
Here, $Q_{f}({\bf R}_{rs},t;{\bf R}_{rs}^0)$, the conditional probability for the
contact distance ${\bf R}_{rs}$ with the initial distribution $P_{I}({\bf R}_{rs}^0)$ of the
contact distance, ${\bf R}_{rs}^0$, is given in terms of ${\bf C_1}(t)$$=$$\langle {\bf R}_{rs}(t)\rangle$,
and $C_2(t)$$=$$\langle[{\bf R}_{rs}(t)-\langle {\bf R}_{rs}(t)\rangle]^2\rangle$ by
%, and $d{\bf R}$$=$$4$$\pi$${\bf R}^2$$dR$.
%For the Ornstein-Uhlenbeck process with a fixed ${\bf R}_{rs}^0$
%(Eqs. (\ref{1.1})-(\ref{1.3})),
%%
\begin{equation}\label{1.7}
Q_{f}({\bf R}_{rs},t;{\bf R}_{rs}^0)=\left[ {{3}\over {2\pi C_2(t)}}\right]^{{3}\over {2}}
\exp{\left[ -{{3({\bf R}_{rs}-{\bf C_1}(t))^2}
\over {2 C_2(t))}} \right]}
\end{equation}
%%
We define the $N$$\times$$3$-matrices for positions,
${\bf R}(t)$$\equiv$$({\bf R}_1(t),{\bf R}_2(t),\ldots ,{\bf R}_N(t))^\dagger$, random
forces, ${\bf F}(t)$ $\equiv$$({\bf F}_1(t),{\bf F}_2(t),\ldots ,{\bf F}_N(t))^\dagger$, and the applied force,
${\bf f}$$\equiv$$({\bf 0},{\bf 0},\ldots ,{\bf f}_N)^\dagger$, where ``$\dagger$'' denotes
the adjoint $3$$\times$$N$ matrix. The equation of motion in matrix form is
%%
%\begin{equation}\label{2.1}
${{d}\over {dt}}{\bf R}+{\bf M}{\bf R}={{1}\over {\xi}}{\bf F}+{{1}\over {\xi}}{\bf f}$
%\end{equation}
%%
%where in the $N$$\times$$N$ matrix ${\bf M}$$=$$w_0({\bf A}_0+\gamma {\bf A}_1)$, $w_0$${\bf A}_0$
%describes the dynamics of free Rouse chain, $w_0$$\gamma$${\bf A}_1$ accounts for the adhesive contact,
%and $\gamma$$=$$(a/b)^2$ is the strength of the $(r,s)$-contact relative to the harmonic bonds
%between two covalently connected monomers. The tridiagonal matrix ${\bf A}_0$$\equiv$$[a^{0}]_{mn}$ has
%the non-zero elements $a^{0}_{1,1}$$=$$a^{0}_{N,N}$$=$$1$, $a^{0}_{m,m}$$=$$2$, and
%$a^{0}_{m,m+1}$$=$$a^{0}_{m-1,m}$$=$$-1$ \cite{20}. The non-zero elements of matrix
%${\bf A}_1$$\equiv$$[a^{1}]_{mn}$ are $a^{1}_{r,r}$$=$$a^{1}_{s,s}$$=$$1$ and
%$a^{1}_{r,s}$$=$$a^{1}_{r,s}$$=$$-1$.
with the $N$$\times$$N$ matrix ${\bf{M}}$$=$$\omega_0({\bf{A}}_0$$+$$\gamma{\bf{A}}_1)$. The
tridiagonal Rouse matrix is ${\bf{A}}_0$, with $a^{0}_{1,1}$$=$$a^{0}_{N,N}$$=$$1$, $a^{0}_{m,m}$$=$$2$,
and $a^{0}_{m,m+1}$$=$$a^{0}_{m-1,m}$$=$$-1$ \cite{20}, and $\omega_0$$=$$3k_B T/2a^2$.
The interaction between monomers $r$ and $s$ are included in ${\bf{A}}_1$, with
$a^{1}_{r,r}$$=$$a^{1}_{s,s}$$=$$1$ and $a^{1}_{r,s}$$=$$a^{1}_{r,s}$$=$$-1$. The
strength of the adhesive contact, relative to the harmonic bond is $\gamma$$=$$(a/b)^2$.
Integration of the equation of motion gives
${\bf R}(t)$$=$${\bf G}(t)$${\bf R}(0)$$+$$(1/\xi)$$\int_0^tdt'{\bf G}(t')({\bf F}(t')$$+$${\bf f})$, where
${\bf G}(t)$ is the Greens function; ${\bf G}(\omega)$$\equiv$$\left[ i\omega {\bf I}+{\bf M}\right]^{-1}$ in the Fourier domain.
To compute ${\bf C_1}(t)$ and $C_2(t)$, we must invert ${\bf G}(\omega)$. We rewrite
${\bf A}_1$$=$${\bf P}{\bf P}^{\dagger}$, where $\bf{P}$ is an $N$$\times$$1$ vector with non-zero elements $P_r$$=$$1$
and $P_s$$=$$-1$,
%where the $1$$\times$$N$
%adjoint ${\bf P}^{\dagger}$ of the projection vector ${\bf P}$ is given by the row vector
%${\bf P}^{\dagger}$$=$$(0, \ldots, 1, \ldots, 0, \ldots -1, \ldots, 0)$ with the non-zero entries
%$P_r$$=$$1$ and $P_s$$=$$-1$ for the $(r,s)$-contact.
%Using the matrix identity in refs \cite{24,25,22},
and
${\bf G}(\omega)$$=$${\bf G}_0(\omega)$$-$${\bf G}_0(\omega){\bf P}$$[ {{1}\over {\gamma w_0}}{\bf I}+
{\bf P}^\dagger {\bf G}_0(\omega){\bf P}]^{-1}$${\bf P}^\dagger {\bf G}_0(\omega)$ \cite{242522},
where the matrix elements
$G_0(\omega)_{mn}$$=$$[i\omega {\bf I}$$+$${\bf M}_0]_{mn}^{-1}$$=$${{1}\over {N}}$$\sum_p{{\cos{[p\pi n/N]}\cos{[p\pi m/N]}}\over {i\omega+\lambda_pw_0}}$ and
$\lambda_p$$=$$\pi^2 p^2/N^2$ are the eigenmodes ($p$$=$$0,1,\ldots,N$) \cite{20}.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
{\it The pdf of contact lifetimes for a single contact:}
Constraining the first residue of the chain is equivalent to excluding
$\lambda_0$$=$$0$, which defines the overall translation of the chain. The results are presented by including
the first ten modes
%($\lambda_1,\lambda_2,\ldots,\lambda_{10}$)
for $N$$=$$100$, $a$$=$$0.4nm$, the diffusion coefficient, $D$$=$$10^6$$nm^2/s$, and $k_B T$$=$$4.1$$pN$$nm$.
We chose $P_{I}({\bf R}_{rs}^0)$$=$${{1}\over {4\pi (B/2)^2}}$$\delta (|{\bf R}_{rs}^0|-B/2)$,
with the rupture distance set to $B$$=$$2a$.
%The distance at which the contact breaks was set to $B$$=$$2$$a$.
To assess the effect of the strength of the adhesive contact
on the contact rupture kinetics, we used $\gamma$$=$$1$ (strong contact) and $\gamma$$=$$0.01$ (weak contact).
The critical equilibrium force required to rupture the contact is $f_c$$=$$(3 k_B T B/b^2)$; $f_c$$=$$31pN$
for $\gamma$$=$$1$, and $f_c$$=$$0.31pN$ for $\gamma$$=$$0.01$. The values of $f$ were set to $f$$=$$1pN$
(weak force) and $f$$=$$40pN$ (strong force). At $f$$=$$1pN$, $f/f_c$$\ll$$1$ for
$\gamma$$=$$1$, and $f/f_c$$\sim$$1$ for $\gamma$$=$$0.01$; at $f$$=$$40pN$, $f/f_c$$>$$1$ for
$\gamma$$=$$1$ and $f/f_c$$\gg$$1$ for $\gamma$$=$$0.01$.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\begin{figure}[t]
\includegraphics[width=.48\textwidth]{Figure1NewB.png}
\caption{The survival probabilities $P_{rs}(t)$ for a single contact $(r,s)$$=$$(1,100)$
({{$\mathbf{a}$, $\mathbf {d}$}}), $(20,80)$ ($\mathbf {b}$, $\mathbf {e}$), and $(40,60)$ ($\mathbf {c}$,
$\mathbf {f}$), with $ {f}$$=$$1$pN ($\mathbf {a}$$-$$\mathbf {c}$) and $40$pN ($\mathbf d$$-$$\mathbf f$),
and with $\gamma$$=$$1$ (solid lines) and $0.01$ (dashed lines). The lifetime distributions $p_{rs}(t)$
are shown in the insets. Notice the variations in time scales in the panels. The chain structures for the intact single contact $(20,80)$,
before and after the rupture are shown.}
\end{figure}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
The survival probabilities $P_{rs}(t)$ (Fig. 1) of the adhesive contact at $f$$=$$1 pN$ and $f$$=$$40 pN$
for $\gamma$$=$$0.01$ and $\gamma$$=$$1$ show striking variations with respect to $|r - s|$.
For a fixed $f$ (low forces with $f/f_c$$\ll$$1$ or $\sim$$1$) the decay of $P_{rs}(t)$ not
only depends on the strength of the contact, but also on its location. At $f$$=$$1pN$, $P_{rs}(t)$
for strong and weak bonds display markedly different kinetics, with $P_{rs}(t)$ for weak contact
$(r,s)$$=$$(1,100)$ and $(20,80)$ decaying to values close to zero on $30 \mu s$ timescale
(Figs. 1a and 1b). However, for $(r,s)$$=$$(40,60)$ the asymptotic value of
$P_{rs}(t$$\to$$\infty)$$=$$P_{rs}^{eq}$$\approx$$0.12$ (Fig. 1c). Surprisingly,
$P_{rs}(t)$s for $(r,s)$$=$$(20,80)$ and $(40,60)$ for strong bonds are qualitatively similar (Fig. 1b and 1c)
and approach higher values of $P_{rs}^{eq}$$\approx$$0.42$ and $0.85$ respectively, whereas
$P_{rs}^{eq}$$\approx$$0.26$ for $(r,s)$$=$$(1,100)$ (Fig. 1a). Thus, the rupture of contacts
in the middle of the chain is strongly dependent on the coupling between the internal modes of the chain
and the dynamics of instability due to the applied force. We refer to this as the
{\it ``internal motion dominated'' (IMD)} regime. The stochastic nature of the bond rupture kinetics is also reflected
in the decay of $P_{rs}(t)$ for strong and weak bonds. For $(r,s)$$=$$(1,100)$, $P_{rs}(t)$ decays faster to
lower $P_{rs}^{eq}$, due to larger chain
fluctuations at the chain ends compared
to contacts $(r,s)$$=$$(20,80)$ and $(40,60)$ in the middle of the chain. In contrast to low
forces, for $f/f_c$$>$$1$, $P_{rs}(t)$ for all contacts decay to zero at long times, regardless of the contact strength
(Figs. 1d-1f). However, $P_{rs}(t)$ for strong contact $(r,s)$$=$$(40,60)$, which
is close to the midpoint of the chain, decays to zero on a much slower $40\mu s$ timescale, compared
to $10\mu s$ and $1\mu s$ timescale for $(r,s)$$=$$(20,80)$ and $(1,100)$. At $f$$=$$40pN$,
$P_{rs}(t)$ for weak contacts decay to zero on similar time scales (Figs. 1d-1f) which shows that at large
forces rupture kinetics is solely determined by the instability caused by the applied tension, with chain
dynamics playing relatively minor role. Thus, as $f/f_c$ increases, the rupture kinetics become increasingly
{\it ``tension dominated''}, i.e. contact rupture is due to the applied force. Interestingly, there is a significant
plateau in the decay of $P_{rs}(t)$ especially for $\gamma$$=$$1$. The duration of the plateau increases
as $|r-s|$ decreases and $f$ increases (Figs. 1d-1f), and shows that the kinetics of rupture is non-exponential.
The lifetime distributions, $p_{rs}(t)$ (insets in Fig. 1), show longer tails for the stronger contacts;
$p_{rs}(t)$ for the interior contact $(40,60)$ (Figs. 1c and 1f), does not have a pronounced dependence
on $f$ compared to $p_{rs}(t)$ for contacts $(20,80)$ and $(1,100)$ (Figs. 1a, 1b,
and 1d and 1e). In particular, $p_{rs}(t)$ at $f$$=$$1pN$ and $f$$=$$40pN$ \textit{agree quantitatively}
for strong and weak $(40,60)$-contacts. This implies that the disruption of the interior contacts is
mediated by internal chain motions, and hence is in the IMD regime even at
high $f$. The influence of internal motion of the chain on the contact rupture kinetics is also reflected in the
width of $p_{rs}(t)$s which broadens for remote contacts $(20,80)$
and $(40,60)$.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\begin{figure}
\includegraphics[width=.48\textwidth]{side_Figure2.png}
\caption{The $Q$$-$$Q$ plots (empty circles) of lifetimes $p_{rs}(t)$ for single contact $(r,s)$$=$$(1,100)$
($x$-axis) versus lifetimes $p_{exp}(t)$ ($y$-axis), for $f$$=$$40$pN and $\gamma$$=$$1$ ($\mathbf a$)
and $\gamma$$=$$0.01$ ($\mathbf b$). The rate $K$ in $p_{exp}(t)$ is set to the inverse of the average
contact lifetime (the maximum likelihood estimate) \cite{1919a}.}
\end{figure}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
{\it Contact lifetime pdf is non-exponential:}
Forced rupture of noncovalent bonds is often described using two-state kinetics, $(rs)$$\to$$0$, which
corresponds to the exponential pdf of contact lifetimes, $p_{exp}(t)$$=$$K$$\exp{[-Kt]}$ where $K$ is
rupture rate. The results for $p_{rs}(t)$ in Fig. 1 deviate from $p_{exp}(t)$, especially when $f/f_c$$<$$1$
or when the contact $(rs)$ is in the chain interior. The extent of deviation is assessed using the
quantile-quantile ($Q$$-$$Q$) plot \cite{1919a,Barsegov07BiophysJ}. A quantile is a number $x_p$ such that
$100$$\times$$p\%$ of the probability values are $\le$$x_p$. The $Q$$-$$Q$ scatterplot
compares two sets of probability values: ${x_p}$, the quantile for $p_{rs}(t)$, and ${y_p}$, the quantile for $p_{exp}(t)$.
% with the lifetime of the bond determined from $K^{-1}=\int_0^\infty dt\, tp_{rs}(t)$
%(corresponding to the maximum likelihood estimate for $p_{exp}(t)$ \cite{19}).
If the two sets are similarly distributed, the points fall on the reference line \cite{1919a}. The
$Q$$-$$Q$ plots of the lifetime distribution for the contact $(r,s)$$=$$(1,100)$ at $f$$=$$40pN$ and for
$\gamma$$=$$1$ and $\gamma$$=$$0.01$ show that $p_{rs}(t)$ is larger (smaller) than $p_{exp}(t)$ in
the middle range (tail) of the lifetimes (Fig. 2). The non-exponentiality arises because it takes time
($\tau_{tp}$) for the force-induced tension to propagate to the contact \cite{Structure}.
%As a
%result, the decay of $P_{rs}(t)$ is delayed, and the observed peaks of contact lifetimes are shifted along
%the time axis by $\tau_{tp}$, which in Figs. 1d-1f, is in the range $0.1\mu s$$-$$15\mu s$.
{\it Interplay of timescales:}
The non-exponential nature of $p_{rs}(t)$ is determined by the interplay of two
timescales, $\tau_{tp}$ and the spectrum of relaxation times $\{\tau_p\}$
($p$$=$$1,2,\ldots, N$). These determine the evolution of
${\bf C}_1(t)$ due to the applied force ${\bf f}$, and the broadening of the Gaussian
density, $C_2(t)$, due to the random force ${\bf F}$ (Eq. (\ref{1.7})). To reveal the richness in the rupture
kinetics, we consider $P_{rs}(t)$ for the $(r,s)$-contacts close to the
chain ends in the limits $f/f_c$$<$$1$, and $f/f_c$$>$$1$. For these contacts,
the contribution from even Rouse modes is negligible \cite{20}, and only the slowest mode with relaxation
time $\tau_1$$=$$\xi/\lambda_1$ contributes significantly. As a result,
%%
%\begin{equation}\label{P}
$P_{rs}(t)$$=$$c_0\int_{0}^{B}d{\bf R}_{rs}$$\exp{\left[ - {{3({\bf R}_{rs}-{\bf C_1}(t))^2} \over {2 C_2(t))}}\right]}$,
%\end{equation}
%%
where $c_0$ is the constant, ${\bf C}_1(t)$$\approx$${\bf R}_{rs}^{eq}$$(1-e^{-t/\tau_1})$,
$C_2(t)$$\approx$$(\Delta{\bf R}_{rs}^{eq})^2$$(1-e^{-2t/\tau_1})$, and ${\bf R}_{rs}^{eq}$ and
$\Delta{\bf R}_{rs}^{eq}$ denote the equilibrium $(r,s)$-contact vector and fluctuations, respectively.
{\it 1. Weak force, $f/f_c$$<$$1$:} In this regime, $|{\bf C}_1(t)|$$<$$B$, and hence the kinetics of
$P_{rs}(t)$ is dominated by the broadening of $C_2(t)$. By evaluating $P_{rs}(t)$ and using the series
expansion of the error function,
%$Erf(x)$$=$${{2e^{-x^2}}\over {\sqrt{\pi}}}$$\sum_{k=0}^{\infty}$$2^kx^{2k+1}/(2k+1)!!$, ($B$$<$$\sqrt{2C_2}$),
we obtain $P_{rs}(t)$$\sim$$\exp{[-B^2/2C_2(t)]}$. At short times, $t$$\ll$$\tau_1$, and we can Taylor
expand the exponential function in $C_2(t)$, which to the first order in $t$$\ll$$\tau_1$ reads
$e^{-2t/\tau_1}$$\approx$$1$$-$$2t/\tau_1$. Then, $P_{rs}(t)$$\sim$$e^{-c_1\tau_1/t}$ ($c_1$
is a constant). At longer times, when $t$$\gg$$\tau_1$, $e^{-2t/\tau_1}$$\ll$$1$, and
$P_{rs}(t)$$\sim$$\exp{[-c_2e^{-2t/\tau_1}]}$ ($c_2$ is a constant). This implies that for a weak
force, in the long time limit, $P_{rs}(t)$ of the contacts close to the chain
ends is described by the double-exponential (Gumbel type) distribution of the largest value for the
exponential density \cite{Gumbel}.
{\it 2. Strong force, $f/f_c$$>$$1$:} The kinetics of $P_{rs}(t)$ is dominated by the
the dynamics of ${\bf C}_1(t)$. However, the decay of
$P_{rs}(t)$ is delayed by $\tau_{tp}$. The tension propagation timescale $\tau_{tp}$$\approx$$\tau_1$.
{\it a. Tension propagation regime:} ($t$$<$$\tau_{tp}$). Since $|{\bf C}_1(t)|$$<$$B$, the contact lifetime
is strongly linked to the random motions of the polymer, which results in broadening of the Gaussian
density due to increase in $C_2(t)$. Evaluating $P_{rs}(t)$, and assuming
%using series expansion of the error function
%with the dominant $k$$=$$0$ term (
$|{\bf C}_1(t)|$$\approx$$B$$<$$C_2(t)$,
%followed by Taylor expansion
%of the exponential functions in ${\bf C}_1(t)$ and $C_2(t)$ and retaining the first order terms,
we obtain $P_{rs}(t)$$\sim$$(t/\tau_1)^3$$e^{-c_3(t/\tau_1)^{1/2}}$ ($c_3$ is a constant). Thus,
$P_{rs}(t)$ scales with force as $P_{rs}(t)$$\sim$$f^3$$e^{-c_4 f}$ ($c_4$ is a constant).
{\it b. Kinetic regime: ($t$$>$$\tau_{tp}$)}. The contact lifetime is
still linked to the random motions of the polymer (broadening of $C_2(t)$) but is dominated by
the dynamics of ${\bf C}_1(t)$$>$$B$.
%We integrate Eq. (\ref{P}) and use series expansion of the
%error function with the $k$$=$$0$ term ($|{\bf C}_1(t)|$$\gg$$C_2(t)$).
In this regime, $e^{-t/\tau_1}$$\ll$$1$ in ${\bf C}_1(t)$ and $e^{-2t/\tau_1}$$\ll$$1$ in $C_2(t)$,
and $P_{rs}(t)$ is again double-exponential (Gumbel) density \cite{Gumbel}, i.e.
$P_{rs}(t)$$\sim$$\exp{[-c_5 e^{-2t/\tau_1}]}$, where $c_5$ is a constant. We find
$P_{rs}(t)$$\sim$$f^2$$e^{-c_6 f}$, where $c_6$ is a constant.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
{\it The contact lifetimes for two contacts:}
The Hamiltonian of the chain with two contacts $(r,s)$ and $(r',s')$ is given by
%%
%\begin{eqnarray}\label{4.1}
$H$$=$${{3k_B T}\over {2a^2}}$$\sum_{n=2}^{N}{\bf R}_{n,n-1}^2$$+$${\bf f}$${\bf R}_{N1}$$+$
$\sum_{(i,j)}$${{3k_B T}\over {2b_{ij}^2}}$$\Theta(B_{ij}$$-$$|{\bf R}_{ij}|){\bf R}_{ij}^2$,
%\end{eqnarray}
%%
where $b_{ij}$ ($B_{ij}$) are the equilibrium (critical rupture) distance,
%For two well-separated contacts, $(r,s)$ and $(r',s')$,
which we rewrite as
$H$$=$$H_{rs}$$+$$H_{r's'}$, where
$H_{ij}$$=$${{1}\over {2}}$${{3k_B T}\over {2a^2}}$$\sum$${\bf R}_{n,n-1}^2$$+$${{{\bf f}}\over {2}}$${\bf R}_{N1}$$+$${{3k_B T}\over {2b_{ij}^2}}$$\Theta(B_{ij}$$-$$|{\bf R}_{ij}|)$${\bf R}_{ij}^2$.
%($ij$$=$$(r,s)$ and $(r',s')$). %This decomposition shows that the rupture
%kinetics for well-separated contacts $(r,s)$ and $(r',s')$ are independent.
The population is
%$P_{tot}(t)$ is,
$P(t)$$=$$P_{rs,r's'}(t)$$+$$P_{0,r's'}(t)$$+$$P_{rs,0}(t)$, where
$P_{rs,r's'}(t)$$=$$P_{rs}(t)$$P_{r's'}(t)$ denotes that both contacts are intact,
$P_{0,r's'}(t)$$=$$P_{rs\to 0}(t)$$P_{r's'}(t)$ means that $(r,s)$ is ruptured but $(r',s')$
is intact, and $P_{rs,0}(t)$$=$$P_{rs}(t)$$P_{r's'\to 0}(t)$ implies that $(r',s')$ is disrupted
but $(r,s)$ is intact. $P_{ij \to 0}(t)$$=$$1$$-$$P_{ij}(t)$ is the population
of conformations with the disrupted contact $(i,j)$.
%The case of the two neighbor contacts
%$(r,s)$ and $(r,s)$, such that $r'$$=$$r$$\pm$$1$ and $s'$$=$$s$$\pm$$1$, could be a model for
%multiple bonds between nucleic acids RNA or in single stranded DNA.
Using $r'$$\approx$$r$, $s'$$\approx$$s$, and $b_{rs}$$\approx$$b_{r's'}$$=$$b$, the Hamiltonian
for two nearest neighbor contacts is
%$H$$=$${{3k_B T}\over {2a^2}}$$\sum$$({\bf R}_n-{\bf R}_{n-1})^2$$+$${\bf f}$$({\bf R}_N-{\bf R}_1)$
%$+$${{3k_B T}\over {2\bar{b}^2}}$$h(|{\bf R}_{rs}|)$$({\bf R}_r-{\bf R}_s)^2$,
$H$$=$${{3k_B T}\over {2a^2}}$$\sum$${\bf R}_{n,n-1}^2$$+$${\bf f}$${\bf R}_{N1}$
$+$${{3k_B T}\over {2\bar{b}^2}}$$\theta(B$$-$$|{\bf R}_{rs}|)$${\bf R}_{rs}^2$,
where
$\bar{b}$$=$$b/\sqrt{2}$ is the rescaled distance. $P_{tot}(t)$ ($p_{tot}(t)$)
is computed using $P_{rs}(t)$ ($p_{rs}(t)$) for single contact $(r,s)$.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\begin{figure}[t]
\includegraphics[width=.48\textwidth]{Figure3NewB.png}
\caption{The survival probabilities $P_{tot}(t)$ for contact pairs $(1,100)$ and $(20,80)$
($\bf{a}$), $(1,100)$ and $(40,60)$ ($\bf{b}$), and $(20,80)$ and $(40,60)$ ($\bf{c}$),
$(1,100)$ and $(2,99)$ ($\bf{d}$), $(20,80)$ and $(21,79)$ ($\bf{e}$), and $(40,60)$
and $(41,59)$ ($\bf{f}$), for $f$$=$$40$pN (solid lines) and $80$pN (dashed lines), all with $\gamma$$=$$1$.
The lifetime distributions $p_{tot}(t)$ are shown in the insets. Time scales vary greatly in all panels. Structures are for the intact pair $(20,80)$
$(40,60)$, disrupted contact $(20,80)$ and intact contact $(40,60)$, and both contacts disrupted are shown.}
\end{figure}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Plots of $P_{tot}(t)$ for two separated strong contacts at $f$$=$$40pN$ and $80pN$
show that, in general, increasing $f$ facilitates more rapid rupture of the binary contacts (Figs. 3a-3c).
The presence of interior contacts delays their rupture
(compare Figs. 3a with 3b and 3c) even though $f/f_c$$>$$1$.
%The long-time behavior of $P_{tot}(t)$ for two separated contacts is the same as the single contact
%population for the interior point $(r',s')$, $P_{r's'}(t)$ (compare to the long-time behavior of Fig 1d-f),
The time dependence of $P_{tot}(t)$ for two separated contacts (Figs 3a-c), after a lag phase, can be
analyzed using two distinct exponential functions, which shows the rupture occurs by sequential kinetics,
with $\{(r,s),(r',s']\}$$\to$$\{(0),(r',s')\}$$\to$$\{(0),(0)]\}$.
%Comparison of $P_{tot}(t)$ for two separated contacts (Fig. 3a-3c) with the curves of $P_{rs}(t)$ for
%single contacts at the same $\gamma$ and $f$ shows $(r,s)$$+$$(r',s')$$\to$$(0)$$+$$(r',s')$$\to$$(0)$$+$$(0)$,
%i.e rupture occurs by sequential kinetics.
The {\it rate limiting second step} is the rupture of the more interior contact. Comparison of $P_{tot}(t)$,
for $\gamma$$=$$1$ and $f$$=$$40pN$, for the nearest neighbor contact pairs $(r,s)$$=$$(1,100)$ and
$(r',s')$$=$$(2,99)$, $(20,80)$ and $(21,79)$, and $(40,60)$ and $(41,59)$ (Fig. 3d-3f) with $P_{rs}(t)$
for single contact $(r,s)$$=$$(1,100)$, $(20,80)$, and $(40,60)$
(Fig. 1d-1f) shows that the double bond increases stability of the contact, especially for the
interior contact $(40,60)$. The chain with double bonds ruptures in a single step,
$\{(r,s),(r',s')\}$$\to$$\{(0),(0)\}$, i.e. the rip occurs cooperatively, so that $P_{tot}(t)$ decays with
a single rate constant at long times.
% By comparing the values of $P_{tot}^{eq}$ for separated contacts
%and for double bonds, at $f$$=$$40pN$, we see that the separated contacts enhance the stability of
%the contact architecture (Fig. 3).
The decay of $P_{tot}(t)$ for separated contacts,
$(r,s)$$=$$(20,80)$ and $(r',s')$$=$$(40,60)$, is slower than $P_{tot}(t)$ for the nearest neighbor
contacts, $(1,100)$ and $(2,99)$ (Figs. 3c and 3d), and $P_{tot}^{eq}$ is larger, which
shows that the persistence of binary interactions critically depends on the location of contacts
in the chain.
The lifetime distributions show that increasing the force from $f$$=$$40pN$ to $f$$=$$80pN$ results
in the decreased stability of contacts and shorter lifetimes (Fig. 3).
%However, this
%effect is less pronounced for interior contact pairs $(20,80)$ and $(40,60)$, and $(40,60)$ and $(41,59)$
%(Fig. 3c and 3f).
The contact lifetimes increase when they are located in the chain interior (Fig. 3).
% which is seen by comparing the curves of $p_{tot}(t)$ for contacts $(1,100)$ and $(20,80)$ with the curves
%of $p_{tot}(t)$ for contacts $(1,100)$ and $(40,60)$, or contacts $(20,80)$ and $(40,60)$.
The lifetimes of the nearest neighbor contacts, $(1,100)$ and $(2,99)$ (Figs. 3d and 3e), are
shorter than that for the pair, $(40,60)$ and $(41,59)$ (Fig. 3f), which shows
the importance of the location of binary contacts with respect to the point of application of the force.
%By comparing the width of the distributions we see that the contact rupture kinetics is
%more stochastic at a lower force especially for separated contacts in the chain interior.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
{\it Conclusions:} The present work shows that the shape of the free energy landscape can only be
discerned by analyzing the topology-dependent lifetime distributions of adhesive contacts that reflect
the structure the molecules. The continued increase in the temporal resolution in laser optical
tweezer experiments should allow the prediction that, the internal motions of biomolecules are coupled to global fluctuations, to be tested.
The present theory provides a conceptual framework for interpreting such experiments \cite{BlockAnnRevBiophys2007TinocoPNAS07FernandezPNAS07}. In contrast, the rich behavior predicted here cannot
be obtained by solving for the dynamics in the exactly calculable equilibrium free energy profile in the presence of non-zero force. Accounting for the non-exponential distribution of unbinding lifetimes
will require models that reflect the interplay between local chain motions and tension-induced global
unfolding.
{\bf Acknowledgments:} This work was supported by National Science Foundation Grant NSFCHE-05-14056.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\begin{thebibliography}{99}
\bibitem{BlockAnnRevBiophys2007TinocoPNAS07FernandezPNAS07}
W. J. Greenleaf, M. T. Woodside, S. M. Block, {\em Annu. Rev. Biophys. Biomol. Struct.} {\bf 36}, 171 (2007);
P. T. X. Li, C. Bustamante, and I. Tinoco, Jr., {\em Proc. Natl. Acad. Sci. USA} {\bf 104}, 7039 (2007);
K. A. Walther, F. Graeter, L. Dougan, C. L. Badilla, B. J. Berne, and J. M. Fernandez,
{\em Proc. Natl. Acad. Sci. USA} {\bf 104}, 7916 (2007).
\bibitem{HyeonPNAS03DudkoPNAS03}
C. Hyeon and D. Thirumalai, {\em Proc. Natl. Acad. Sci. USA} {\bf 100}, 10249 (2003);
O. K. Dudko, A. E. Filippov, J. Klafter, and M. Urbakh, {\em Proc. Natl. Acad. Sci. USA} {\bf 100},
11378 (2003); O. K. Dudko, G. Hummer and A. Szabo, {\em Phys. Rev. Lett.} {\bf 96}, 108101 (2006).
\bibitem{BarsegovPRL05}
V. Barsegov and D. Thirumalai, {\em Phys. Rev. Lett.} {\bf 95}, 168302 (2005).
\bibitem{Hyeon06BiophysJ}
C. Hyeon and D. Thirumalai, {\em Biophys. J.} {\bf 90}, 3410 (2006).
\bibitem{20}
M. Doi and S. F. Edwards. {\em The Theory of Polymer Dynamics}, (Oxford University Press, New York, 1994).
\bibitem{242522}
P. Lancaster and M. Tismenetsky. {\em Theory of Matrices}, (Academic Press, Orlando, 1985);
H. V. Henderson and S. R. Searle. On Deriving the Inverse of a Sum of Matrices. {\em SIAM Rev.}, {\bf 23},
53-60 (1981); M. P. Solf and T. A. Vilgis. {\em J. Phys. A: Math. Gen}, {\bf 28}, 6655 (1995).
\bibitem{1919a}
G. Casella and R. L. Berger. {\em Statistical Inference}, (Pacific Grove, Duxberry, 2002);
Davison, A. C. {\em Statistical Models}, (Cambridge University Press, Cambridge, 2003).
\bibitem{Barsegov07BiophysJ}
E. Bura, D. K. Klimov and V. Barsegov, {\em Biophys. J.} {\bf 93}, 1100 (2007); {\bf 94}, 2516 (2008).
\bibitem{Structure}
C. Hyeon, R. I. Dima and D. Thirumalai, {\em Structure} {\bf 14}, 1633 (2006).
\bibitem{Gumbel}
E. J. Gumbel. {\em Statistics of Extremes}, (Dover Publications, New York, 2004).
\end{thebibliography}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%\newpage
%\section*{\bf FIGURE CAPTIONS}
%{\bf Fig. 1.}
%The survival probabilities $P_{rs}(t)$ for a single contact $(r,s)$$=$$(1,100)$ ({{$\mathbf{a}$, $\mathbf {d}$}}),
%$(20,80)$ ($\mathbf {b}$, $\mathbf {e}$), and $(40,60)$ ($\mathbf {c}$, $\mathbf {f}$), with $ {f}$$=$$1$pN
%($\mathbf {a}$$-$$\mathbf {c}$) and $40$pN ($\mathbf d$$-$$\mathbf f$), and with $\gamma$$=$$1$ (solid
%lines) and $0.01$ (dashed lines). The lifetime
%distributions $p_{rs}(t)$ are shown in the insets.
%\bigskip
%{\bf Fig. 2.}
%The $Q$$-$$Q$ plots (empty circles) of the density of contact lifetimes $p_{rs}(t)$ for single contact $(r,s)$$=$$(1,100)$
%($x$-axis) versus the density of contact lifetimes sampled from the exponential density $p_{exp}(t)$ ($y$-axis), for
%$f$$=$$40$pN and $\gamma$$=$$1$ ($\mathbf a$) and $\gamma$$=$$0.01$ ($\mathbf b$). The deviation from the dashed
%reference line shows the non-exponential nature of $p_{rs}(t)$. The rate $K$ in $p_{exp}(t)$ is
%set to the inverse of the average contact lifetime corresponding to the maximum likelihood estimate for $p_{exp}(t)$. \cite{19}
%\bigskip
%{\bf Fig. 3.}
%%The survival probabilities $P_{tot}(t)$ with $\gamma=1$ for well-separated contacts $(r,s)$$=$$(1,100)$ and $(r',s')$$=$$(20,80)$
%%($\mathbf a$), $(1,100)$ and $(40,60)$ ($\mathbf b$), and $(20,80)$ and $(40,60)$ ($\mathbf c$). Also shown are the probabilities for nearest neighbor
%%contacts (with $r'=r-1$, $s'=s-1$ and $\gamma=1$) for $(r,s)$$=$$(1,100)$ ($\mathbf d$), $(20,80)$ ($\mathbf e$), and $(40,60)$ ($\mathbf f$), for $f$$=$$40$pN (solid lines) and 80pN (dashed lines). The lifetime distributions $p_{tot}(t)$ are shown in the insets.
%The survival probabilities $P_{tot}(t)$ for well-separated contacts $(r,s)$$=$$(1,100)$ and $(r',s')$$=$$(20,80)$
%($\bf{a}$), $(1,100)$ and $(40,60)$ ($\bf{b}$), and $(20,80)$ and $(40,60)$ ($\bf{c}$), and the nearest neighbor
%contacts $(r,s)$$=$$(1,100)$ and $(r',s')$$=$$(2,99)$ ($\bf{d}$), $(20,80)$ and $(21,79)$ ($\bf{e}$), and $(40,60)$
%and $(41,59)$ ($\bf{f}$), for $f$$=$$40$pN (solid lines) and $80$pN (dashed lines), all with $\gamma$$=$$1$. The lifetime distributions $p_{tot}(t)$ are shown in the insets.
%\newpage
%\begin{figure}
%\includegraphics[width=7.3in]{Figure1New.eps}
%\end{figure}
%\newpage
%\begin{figure}
%\includegraphics[width=7.0in]{Figure2New.eps}
%\end{figure}
%\newpage
%\begin{figure}
%\includegraphics[width=7.3in]{Figure3New.eps}
%\end{figure}
\end{document}