1: \documentclass[aps
2: ,rmp
3: %prl%
4: %,preprint%
5: ,twocolumn%
6: ,amsmath
7: ,amssymbol
8: ,showpacs%
9: ,superscriptaddress,floats%
10: ]{revtex4}
11:
12: \usepackage{graphicx}
13: \usepackage{times}
14: %\usepackage{showlabels}
15:
16: \setlength{\textfloatsep}{1.2\parindent plus 2pt minus 2pt} %%% DELETE %%%
17: \setlength{\arraycolsep}{1pt}
18: \newcommand{\be}{\begin{equation}}
19: \newcommand{\ee}{\end{equation}}
20: \newcommand{\bea}{\begin{eqnarray}}
21: \newcommand{\eea}{\end{eqnarray}}
22: \newcommand{\kT}{k_{\rm B}T}
23: \newcommand{\e}{{\rm e}}
24: \newcommand{\dd}{{\rm d}}
25:
26: \newcommand{\affila}{
27: Dept.\ of Biological Physics, E\"otv\"os University,
28: P\'azm\'any P.\ stny.\ 1A, H-1117 Budapest, Hungary
29: }
30: \newcommand{\affilb}{
31: Laboratoire de Physico-Chimie Th\'eorique, UMR 7083 CNRS, ESPCI,
32: 10 rue Vauquelin, F-75231 Paris C\'edex 05, France
33: }
34:
35:
36: \begin{document}
37:
38: \title{
39: Effects of intermediate bound states in dynamic force spectroscopy
40: }
41:
42: \author{Imre Der\'enyi}
43: \email{Derenyi@elte.hu}
44: \affiliation{\affila}
45:
46: \author{Denis Bartolo}
47: \email{Denis.Bartolo@espci.fr}
48: \affiliation{\affilb}
49:
50: \author{Armand Ajdari}
51: \email{armand@turner.pct.espci.fr}
52: \affiliation{\affilb}
53:
54: %\date[]{\protect\today}
55: %\date[]{submitted to where..., when...}
56:
57: \begin{abstract}
58: We revisit here some aspects of the interpretation of dynamic force spectroscopy experiments. The standard theory predicts
59: a typical unbinding force $f^*$ linearly proportional to the
60: logarithm of the loading rate $r$ when a single energetical barrier
61: controls the unbinding process; for a more complex situation of $N$
62: barriers, it predicts at most $N$ linear segments for
63: the $f^*$ vs.\ $\log(r)$ curve, each segment characterizing a different
64: barrier. We here extend this existing picture using a refined approximation,
65: we provide a more general analytical formula,
66: and show that in principle up to $N(N+1)/2$ segments can show up experimentally. As a consequence the interpretation of data can be ambiguous,
67: for the characteristics and even the number of barriers. A further
68: possible outcome of a
69: multiple-barrier landscape is a bimodal or multimodal
70: distribution of the unbinding force at a given loading rate, a feature
71: recently
72: observed experimentally.
73: \end{abstract}
74:
75: \pacs{82.37.-j, 87.15.-v, 82.20.Kh, 33.15.Fm}
76: % 02.50.-r Probability theory, stochastic processes, and statistics
77: % 05.40.-a Fluctuation phenomena, random processes, noise, and Brownian motion
78: % 33.15.-e Properties of molecules
79: % 33.15.Fm Bond strengths, dissociation energies
80: % 68.43.-h Chemisorption/physisorption: adsorbates on surfaces
81: % 68.43.Mn Adsorption/desorption kinetics
82: % 82.20.-w Chemical kinetics and dynamics
83: % 82.20.Db Transition state theory and statistical theories of rate constants
84: % 82.20.Kh Potential energy surfaces for chemical reactions
85: % 82.37.-j Single molecule kinetics
86: % 82.37.Gk STM and AFM manipulations of a single molecule
87: % 82.37.Np Single molecule reaction kinetics, dissociation, etc.
88: % 82.37.Rs Single molecule manipulation of proteins and other biological molecules
89: % 87.15.-v Biomolecules: structure and physical properties
90: % 87.15.By Structure and bonding
91: % 87.15.He Dynamics and conformational changes
92: % 87.15.La Mechanical properties
93: % 87.64.-t Spectroscopic and microscopic techniques in biophysics and medical physics
94:
95: \maketitle
96: \section{Introduction}
97: The last decades have witnessed a
98: revolution in the methods to observe and manipulate single bio-macromolecules or bio-molecules complexes. New micromanipulation techniques have especially been put forward to probe the folded structure of proteins and to quantify the strength of adhesion complexes~\cite{titin,Poi01,Sim99,Nis00,Pie96}. An important
99: step in this direction
100: is the proposal of the group of Evans
101: to use soft structures to pull on adhesion complexes
102: or molecules at various loading rates (dynamic force spectroscopy or DFS)~\cite{Eva}.
103: Moving the other end of the soft structure at constant velocity
104: induces on the complex a pulling force that
105: increases linearly in time $f=rt$.
106: Measuring the typical unbinding time $t^*$
107: yields an unbinding force $f^*=rt^*$ that depends
108: on the pulling rate $r$. The typical outcome of such experiments is a plot of $f^*$ vs.\ $\log r$ composed by a succession of straight lines with increasing slopes (force spectrum).
109: It has been argued and that it is possible to deduce the value of some relevant structural parameters of the complex by analysing force spectra thanks to an adiabatic Kramers picture. This picture consist in considering the unbinding process as the thermally activated escape from bound states over a succession of barriers along a one-dimensional path crossing a mountainous energy landscape~\cite{Eva}. Within this scheme, each straight line of the force spectrum witnesses the overcome of an energy barrier and its slope maps the barrier to a distance $x$ along the pulling direction. This procedure has been shown to yield reasonable values for a few systems, and has been conforted
110: by numerical simulations \cite{Hey00}.
111:
112:
113: Subsequently, theoretical studies have refined the above original model, e.g., by inclusion of rebinding events~\cite{Sei02}, study of time dependent loading rates~\cite{Eva}, or consideration of more complex
114: topographies ~\cite{Str00} and topologies of the energetical landscape ~\cite{Bar02}.
115:
116: In the present paper we explore the potential influence of the
117: existence of intermediate bound states on the experimental dynamic response of adhesion complexes as probed in DFS. To achieve this goal, we first revisit the analysis of the escape from a bound state consisting of an arbitrary number of barriers along a 1D path under the application of an external load
118: (in line with earlier studies of Strunz et al. \cite{Str00}),
119: and then discuss the implications of this analysis for the interpretation of experimental data.
120: In (\ref{sec_std}) the standard picture is recalled, together with its two underlying assumptions. In (IV), we first relax the {\it a priori} assumption of a deep fundamental bound state and provide a general expression that relates the typical rupture force to the loading rate (within a single escape rate approximation).
121: The practical implications of this new formula [Eq.\ \eqref{eq_fr}] are discussed, and in particular we comment upon intrinsic ambiguities
122: in inferring informnation from a $[\log(r),f^*]$ plot.
123: Then, we show in (V) that in the presence of multiple bound states
124: it may be necessary to relax the other assumption
125: (a single typical rupture force for each loading rate)
126: as multimodal rupture force distributions naturally show up,
127: a feature recently observed in lipid extraction experiments~\cite{Eva02}.
128: \section{Model and Notations}
129: \begin{figure}[!t]
130: \centerline{\includegraphics[width=.8\columnwidth]{fig1_imre}}
131: \caption{
132: Sketch of the one-dimensional energy landscape describing the unbinding pathway projected along the pulling direction. %
133: %The location $x_0$ and energy $E_0$
134: %of the fundamental bound state are arbitrarily taken as origins of the spatial and %energetical axes.
135: }\label{fig1}\end{figure}
136: Figure \ref{fig1} illustrates the energy landscape of a one-dimensional
137: escape path with $N$ energy barriers and wells. The position and energy
138: of the $i$th energy well ($i=0$ marking the fundamental bound state,
139: and $1\leq i \leq N-1$ the intermediate ones) are denoted by $x_i$ and
140: $E_i$, respectively. Similarly, the position and energy of the $j$th
141: energy barrier are denoted by $\widehat{x}_j$ and $\widehat{E}_j$,
142: respectively (where $1\leq j \leq N$). For convenience, without loosing
143: generality, we set $x_{0}=0$ and $E_{0}=0$ for the fundamental bound
144: state. The unbound ``state'' is on the right hand side of the $N$th
145: barrier. If the energy differences $(\widehat{E}_{i}-E_i)$ and $(\widehat{E}_{i+1}-E_i)$ exceed $k_{\rm B}T$ the transition rates $k_i^-$ (and $k_i^+$) from the $i$th energy well over the left $i$th barrier (and right $(i+1)$th barrier, respectively) can be written according to the Kramers formula
146: \bea
147: k^-_i &=& \omega_0 \, \alpha_i \, \widehat{\alpha}_i \,
148: \e^{-(\widehat{E}_i-E_i)/\kT} ,
149: %\qquad \mbox{and}
150: \label{eq_k-}\\
151: k^+_i &=& \omega_0 \, \alpha_i \, \widehat{\alpha}_{i+1} \,
152: \e^{-(\widehat{E}_{i+1}-E_i)/\kT} ,
153: \label{eq_k+}
154: \eea
155: where
156: $\omega_0$ is a typical attempt frequency,
157: $\alpha_i$ and $\widehat{\alpha}_j$ are geometric factors
158: characterizing the shape of the $i$th energy well and $j$th energy
159: barrier, respectively. Note that there is no transition from the
160: fundamental bound state to the left, therefore, $k^-_0\equiv 0$.
161:
162:
163: We assume throughout the paper that the energy wells and barriers are sharp,
164: so that for any loading force $f$ their locations remain constant,
165: and their energies change as
166: $E_i(f)=E_i(0)-fx_i$ and
167: $\widehat{E}_j(f)=\widehat{E}_j(0)-f\widehat{x}_j$.
168: To simplify the notations, wherever the argument of the energies and
169: transition rates is omitted, a loading force $f$ is implicitly
170: assumed.
171:
172: Finally it will prove convenient to introduce a few compact notations.
173: For any $0\leq i<j\leq N$ we denote
174: the distance between the $i$th well and the $j$th barrier
175: (on the right) by
176: $\Delta x_{i,j}=\widehat{x}_j-x_i$,
177: and their energy difference by
178: $\Delta E_{i,j}=\widehat{E}_j-E_i$.
179: We also define a formal (effective) rate constant
180: from the $i$th well over the $j$th barrier on the right as
181: \be
182: k_{i,j}=\omega_0 \, \alpha_i \, \widehat{\alpha}_j \,
183: \e^{-\Delta E_{i,j}/\kT} .
184: \label{eq_k_eff}
185: \ee
186: Obviously $\Delta E_{i,j}$ and $k_{i,j}$ implicitly depend on $f$,
187: whereas $\Delta x_{i,j}$ are constants given the assumption of the previous paragraph.
188:
189: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
190: \section{Stochastic kinetics of unbinding under external forces: standard description and corresponding approximations}\label{sec_std}
191: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
192:
193: We first recall the standard description of the "force spectrum",
194: which relies on two major assumptions, namely the single escape rate (SER) and the deeply bound fundamental state (DBFS) approximations.
195:
196: %%%%%%
197: \subsection{Single escape rate approximation}
198: %%%%%%
199: If for any experimentally relevant load $f$ the
200: equilibration within the bound states is much faster than the escape to the
201: unbound state, the unbinding can be described by a single load
202: dependent escape rate $k(f)$. Following the calculation of Evans
203: \cite{Eva01}, if rebinding is negligible (which is the
204: case in most experimental situations), the probability $P(t)$ of
205: remaining in the bound state at time $t$ (the survival probability
206: of the bond) then decreases as
207: \begin{equation}
208: \frac{\dd P(t)}{\dd t} = -k(rt) P(t).
209: \end{equation}
210: The solution of this differential equation is
211: $P(t) = \exp[-\int_0^t k(rt') \, \dd t']$.
212: The probability density for unbinding between times
213: $t$ and $t+\Delta t$ is
214: $p_t(t)=- \dd P(t)/\dd t= k(rt) P(t)$, from which, after changing the variable from $t$
215: to $f$, one gets the probability density
216: %\be
217: %p_f(f) = (1/r) k(f) \exp\left[-(1/r)\int_0^f \! k(f') \, \dd f' \right]
218: %\ee
219: for the distribution of the unbinding force: $p_f(f)=(1/r) k(f)P(f/r)$.
220: The typical unbinding force $f^*$ is defined as the peak of this
221: probability density:
222: $\dd p_f(f)/\dd f|_{f=f^*} =0$, which yields
223: the simple formula
224: \be
225: \left.\frac{\dd \tau(f)}{\dd f}\right|_{f=f^*}=-\frac{1}{r} ,
226: \label{eq_f}
227: \ee
228: where $\tau(f)\equiv 1/k(f)$ denotes the load dependent mean escape time.
229: This formula gives the loading rate $r$
230: at which the typical unbinding force is $f^*$. For practical
231: purposes it is often necessary to invert this relation to, e.g., predict the typical unbinding force for an experimentally imposed loading rate.
232:
233: To set a reference for further comparison, we explicitely invert the above relation in case of a single barrier, i.e.\ when $\tau(f)=\tau(0)\exp(-f {\widehat x}_1)$, and obtain
234: \begin{equation}
235: f^* = \frac{\kT}{{\widehat x}_1} \ln\left[\frac{r\tau(0){\widehat x}_1}{\kT}\right].
236: \end{equation}
237: As mentioned in the introduction,
238: the escape over a single barrier results in a single straight line in the force spectrum.
239: The experimental observation of a linear segment consequently
240: gives hints as to the structure of the energy landscape,
241: in particular the slope of the segment permits to deduce
242: a distance $\widehat{x}_1$ between the energy well and the barrier.
243:
244: %%%%%%%%
245: \subsection{Deeply bound fundamental state approximation}
246: %%%%%%%%
247:
248: Assuming further that the fundamental bound state is much deeper than the
249: intermediate ones: $E_i(f)-E_0(f)\gg\kT$ for any experimentally
250: relevant load $f$ (i.e. before unbinding has statistically almost certainly occurred, see Fig.~\ref{fig2} a.),
251: Evans has shown that the mean escape time from the
252: fundamental bound state to the unbound state is well approximated by
253: \cite{Eva01,Eva02}
254: \be
255: \tau(f)
256: =\sum_{j=1}^N
257: \frac{1}{k_{0,j}(f)}
258: =\sum_{j=1}^N
259: \frac{\e^{-f\Delta x_{0,j}/\kT}}{k_{0,j}(0)}
260: \label{eq_evans}
261: \ee
262: This allows one to obtain an explicit $r$ vs.\ $f^*$ relationship
263: by plugging
264: Eq.\ (\ref{eq_evans}) into Eq.\ (\ref{eq_f}),
265: which yields the compact formula
266: \begin{equation}
267: r=\left[
268: \sum_{j=1}^N
269: \frac{\Delta x_{0,j}}{\kT} \,
270: \frac{\e^{-f^*\Delta x_{0,j}/\kT}}{k_{0,j}(0)}
271: \right]^{-1}.
272: \label{eq_fr_dbfs}
273: \end{equation}
274: This equation predicts a spectrum $f^*$ vs. $\log (r)$ consisting of a succession of
275: at most $N$ segments of increasing slopes, each of which yielding
276: an information $\Delta x_{0,j}$ about an intermediate barrier.
277:
278: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
279: \section{Beyond the deeply bound fundamental state approximation}
280: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
281:
282: In this section we relax the approximation made in the last subsection (IIIB),
283: generalize accordingly equations (7) and (9), and discuss the experimental implications of this generalization.
284:
285: %%%%%%%%%%
286: \subsection{Refined theory}
287: %%%%%%%%%%
288: \begin{figure}%[!b]
289: \centerline{\includegraphics[width=\columnwidth]{fig2mod}}
290: \caption{
291: Sketch of two energy landscapes with one intermediate well. Dotted drawings: no external force. When a constant force is applied energies are lowered by $fx$ (dashed lines), the resulting landscapes appear in solid lines. The dotted arrows indicate which pair of wells and barriers control the kinetics at zero load. The solid arrows indicate the new limiting effective escape process at higher forces.
292: (a) The escape from the fundamental bound state remains the limiting process whatever the pulling force.
293: (b) The escape form the intermediate bound state energy becomes the limiting process at high forces.
294: }\label{fig2}\end{figure}
295:
296: In general, it is possible that for large enough forces one or more of the
297: intermediate bound states become deeper than the fundamental bound
298: state before unbinding has occured (see Fig.~\ref{fig2} b.).
299: In such cases the above DBFS approximation breaks down.
300: However, we show below that it is still possible to compute rather simply
301: the escape time $\tau$ from the
302: "bound state" to the "unbound state", provided
303: we maintain the assumption of
304: a single escape rate $1/\tau(f)$.
305:
306: Let us put the system into its fundamental bound state, and let it
307: evolve according to the transition rates given in Eqs.\ (\ref{eq_k-})
308: and (\ref{eq_k+}). Whenever the system gets into the unbound state (by
309: making a transition over the outermost barrier) let us place it back into
310: the fundamental bound state. The stationary state of an ensemble of
311: such systems is characterized by a probability current, which is
312: constant everywhere and equal to $1/\tau$ by definition.
313: To calculate $\tau$ we have to
314: solve the following system of equations:
315: \bea
316: P_i k^+_i - P_{i+1} k^-_{i+1} &=& 1/\tau \qquad 0\leq i \leq N-2 ,
317: \label{eq_i}\\
318: P_{N-1} k^+_{N-1} &=& 1/\tau ,
319: \label{eq_N-1}\\
320: \sum_{i=0}^{N-1} P_i &=& 1 ,
321: \label{eq_norm}
322: \eea
323: where $P_i$ denotes the probability of being in the $i$th bound state
324: ($0\leq i \leq N-1$). The first $N$ equations describe the probability
325: current over each of the $N$ barriers, and the last equation is just
326: the normalization condition. These $N+1$ linear equations uniquely
327: determine the $N+1$ variables ($P_i$ and $\tau$), and can be solved
328: easily in a recursive way. First, $P_{N-1}\tau$ can be expressed from
329: Eq.\ (\ref{eq_N-1}), and then $P_{N-2}\tau$, $...$, $P_0\tau$
330: recursively from Eq.\ (\ref{eq_i}) yielding
331: \bea
332: P_i \tau &=&
333: \frac{1}{k^+_i} +
334: \frac{k^-_{i+1}}{k^+_i k^+_{i+1}} + ... +
335: \frac{k^-_{i+1} ... k^-_{N-1}}{k^+_i k^+_{i+1} ... k^+_{N-1}}
336: \nonumber\\
337: &=&
338: %\frac{1}{k_{i,i+1}} + \frac{1}{k_{i,i+2}} + ... + \frac{1}{k_{i,N}}
339: %=
340: \sum_{j=i+1}^N \frac{1}{k_{i,j}} \label{eq_Pitau},
341: \eea
342: where Eqs. (\ref{eq_k-}), (\ref{eq_k+}), and the definition
343: (\ref{eq_k_eff}) have been
344: used. Note that because the $k_{i,j}$ are only formal definitions,
345: constructed as products
346: and ratios of the single-barrier rates (\ref{eq_k-}) and (\ref{eq_k+}),
347: they are meaningful even if
348: $\Delta E_{i,j}<0$.
349: From the normalization (\ref{eq_norm}) one can easily express $\tau$ as
350: \be
351: \tau
352: =\sum_{i=0}^{N-1} P_i \tau
353: =\sum_{i=0}^{N-1} \sum_{j=i+1}^N
354: \frac{1}{k_{i,j}} .
355: \label{eq_tau_int}
356: \ee
357: The sum is dominated by the smallest effective rates, which are the
358: bottlenecks of the unbinding process.
359: Consequently, this formula remains a good
360: approximation for $\tau$ even if some of the barriers disappear at big
361: loads, because the corresponding formal transition rates make negligible
362: contributions. By indicating the load force $f$ explicitly, we arrive
363: at
364: \begin{equation}
365: \tau(f)
366: =\sum_{i=0}^{N-1} \sum_{j=i+1}^N
367: \frac{1}{k_{i,j}(f)}
368: =\sum_{i=0}^{N-1} \sum_{j=i+1}^N
369: \frac{\e^{-f\Delta x_{i,j}/\kT}}{k_{i,j}(0)} .
370: \label{eq_tau}
371: \end{equation}
372: which generalizes (7).
373: An {\em analytic} formula can be given for the
374: $f^*$ vs.\ $r$ relationship by plugging
375: Eq.\ (\ref{eq_tau}) into Eq.\ (\ref{eq_f}) :
376: \begin{equation}
377: r=\left[
378: \sum_{i=0}^{N-1} \sum_{j=i+1}^N
379: \frac{\Delta x_{i,j}}{\kT} \,
380: \frac{\e^{-f^*\Delta x_{i,j}/\kT}}{k_{i,j}(0)}
381: \right]^{-1},
382: \label{eq_fr}
383: \end{equation}
384: This generalization of equation (8) is one of the main results of this paper.
385: %However, we keep in mind that the SER approximation may not hold since the equilibration between the bound states
386: %is not necessarily fast compared to the escape any more (see
387: %discussion later)
388: %(I would simply remove this remark, DB.).
389: Let us briefly comment on immediate features of this new formula.
390:
391:
392: First, Eq.~\eqref{eq_fr_dbfs} is easily recovered from (15) assuming a DBFS. Indeed, the assumption $E_i(f)\gg E_0(f)$ implies $k_{0,j}\gg k_{i,j}$, if $i>0$ [see Eq.~\eqref{eq_k_eff}] and therefore, the relation $P_i/P_0\ll 1$, if $i>0$ is deduced from Eq.~\eqref{eq_Pitau}. The probability to find the system in the fundamental bound state is close to $1$. So, the sum over $i$ in Eq.~\eqref{eq_tau} is dominated by the contributions of the effective escape rates from the $0$th well only. Finally, the sum over $i$ (labeling the intermediate states) is reduced to its sole first term too, and Eq.~\eqref{eq_fr} becomes identical to Eq~\eqref{eq_fr_dbfs}.
393:
394:
395: Second, each of the $N(N+1)/2$ terms of Eq.\ (\ref{eq_fr}) alone would yield
396: a straight line in the $f^*$ vs.\ $\log(r)$ plot. However, at any
397: loading rate the highest force value (the uppermost segment,
398: corresponding to the most
399: difficult transition) limits the unbinding process, therefore, the
400: $f^*(r)$ curve is expected to closely follow the upper envelope of these
401: segments [see Fig.\ \ref{fig3}(a)]. Depending on the position of the
402: lines, this upper envelope can consist of up to $N(N+1)/2$ linear segments.
403:
404: Third, this last point is clearly at odds with the
405: prediction within the DBFS approximation.
406: Indeed assuming a DBFS corresponds to
407: forbidding the display in the force spectrum of
408: the $N(N-1)/2$ segments
409: corresponding to the probing of the escape from an intermediate bound state (see Fig. \ref{fig2}. b.).
410:
411: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
412: \subsection{Practical implications: Ambiguity in the determination of "structural" parameters}
413: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
414:
415: We now insist on some practical implications
416: of the above general description.
417: We do not attempt a full inspection of all the possible dynamic responses of arbitrarily complex systems, but rather focus on two simple examples
418: in order to stress that the main features of the energy landscape can
419: in general {\em not} be unambiguously inferred from $[\log(r),f^*]$ plots.
420: To emphasize the experimental relevance of this discussion, we use
421: for the parameters values comparable to those observed in experimental systems.
422: Specifically, we take the geometric factors
423: $\alpha_i$s and $\widehat{\alpha}_j$s to be all equal to $1$, $\omega_0=10^8$~s$^{-1}$ and $\kT=4\times10^{-21}$J.%
424: %, and express distances
425: %in nanometer, energies in $k_BT$, forces in pN and loading rates in pN/s.
426:
427:
428: \begin{figure}[!t]
429: \centerline{\includegraphics[width=\columnwidth]{figambigu1}}
430: \caption{Two very similar force spectra corresponding to different energy landscapes with one intermediate well.\\
431: Curves in solid lines: Force spectra plotted using Eq.~\eqref{eq_fr}. It closely follows the upper envelope of the straight lines corresponding to the transitions: $k_{0,2}$ (dotted line), $k_{0,1}$ (dashed line) and $k_{1,2}$ (dash-dotted line).\\ Insets: shape of the energy landscapes at low and high forces, each drawn with the same line style as the straight line associated with the limiting transition.\\
432: Parameter values:
433: (a)
434: $(\widehat{x}_1,\widehat{E}_1)=(1\,{\rm nm},11\,\kT)$,
435: $(x_1,E_1)=(1.5\,{\rm nm},8 \,\kT)$, and
436: $(\widehat{x}_2,\widehat{E}_2)=(2\,{\rm nm},20\,\kT)$,
437: (b)
438: $(\widehat{x}_1,\widehat{E}_1)=(0.5\,{\rm nm},12\,\kT)$,
439: $(x_1,E_1)=(1\,{\rm nm},9 \,\kT)$, and
440: $(\widehat{x}_2,\widehat{E}_2)=(2\,{\rm nm},20\,\kT)$.
441: }
442: \label{fig3}
443: \end{figure}
444: \begin{figure}%[!t]
445: \centerline{\includegraphics[width=\columnwidth]{figambigu2}}
446: \caption{Two very similar force spectra corresponding to energy landscapes with different numbers of intermediate wells.\\ The rule of the line styles is the same as in Fig.~\ref{fig3}:
447: (a) dotted lines: $k_{0,3}$, dash-dotted lines: $k_{0,2}$, dashed lines: $k_{0,1}$. (b) dotted lines: $k_{0,2}$, dash-dotted lines: $k_{1,2}$, dashed lines: $k_{0,1}$.\\
448: Parameter values:
449: (a)
450: $(\widehat{x}_1,\widehat{E}_1)=(0.6\,{\rm nm},14\,\kT)$,
451: $(x_1,E_1)=(0.7\,{\rm nm},12 \,\kT)$,
452: $(\widehat{x}_2,\widehat{E}_2)=(1.1\,{\rm nm},19\,\kT)$,
453: $(x_2,E_2)=(2\,{\rm nm},16\,\kT)$ and
454: $(\widehat{x}_3,\widehat{E}_3)=(2.5\,{\rm nm},24\,\kT)$.
455: (b)
456: $(\widehat{x}_1,\widehat{E}_1)=(1.1\,{\rm nm},19\,\kT)$,
457: $(x_1,E_1)=(1.9\,{\rm nm},10 \,\kT)$, and
458: $(\widehat{x}_2,\widehat{E}_2)=(2.5\,{\rm nm},24\,\kT)$.
459: }
460: \label{fig4}
461: \end{figure}
462:
463: %%%%%%%%%%%%%%%%%%
464: \subsubsection{Ambiguity in determining the barriers positions}
465: %%%%%%%%%%%%%%%%%%
466:
467: Fig~\ref{fig3}\ (a) and Fig~\ref{fig3}\ (b)
468: display two force spectra as obtained from Eq.~\eqref{eq_fr}. Both correspond to energy landscapes with two barriers.
469: Though the two $[\log(r),f^*]$ plots are almost identical
470: they are related to very different
471: sets of values for the energy levels and positions (along the pulling direction)
472: of the wells and the barriers.
473:
474: Fig~\ref{fig3}\ (a) corresponds to the situation where the standard picture to account for the two segments is well suited~\cite{Eva}. At low force, the escape from the fundamental $0$th state over the outermost barrier is the limiting process. The slope of the first segment is proportional to $k_BT/\widehat{x}_2$. For the highest forces (above $\sim 30{\rm pN}$) the energy of the external barrier is reduced below $\widehat{E}_1$ and the deepest bound state remains located at $x_0=0$. The process that mostly impedes the unbinding is the overcome of the innermost barrier $\widehat{E}_1$ with a rate $k_{0,1}$. The slope of the curve is now larger and proportional to $k_BT/\widehat{x}_1$.
475:
476: Fig. \ref{fig3}\ (b) corresponds to an energy landscape for which the above explanation is inappropriate. At low force the unbinding kinetic is controlled by the escape from the fundamental state over the outermost barrier again. But, for pulling forces larger than $\sim 30{\rm pN}$ this outer barrier remains the highest [see inset in Fig~\ref{fig3}.(b)]. However the slope of the spectrum increases as in the (a) case. The reason is that the deepest (and most occupied) bound state is now located at $x=x_1$ and the presence of the second segment actually witnesses the escape from this intermediate state to the unbound state with a rate $k_{1,2}$. The value of the second slope scales therefore with $k_BT/(\widehat{x}_2-x_1)$. Since the escape rate $k_{1,2}$ in the (b) case is equal to the escape rate $k_{0,1}$ in the (a) case, the two spectra in Fig.~\ref{fig3} turn out to be indistinguishable and cannot be {\it a priori} associated with one of the two possible landscapes.
477:
478: %%%%%%%%%
479: \subsubsection{Ambiguity in determining the number of barriers}
480: %%%%%%%%%%
481:
482: After having shown with the simple example above that ambiguity
483: can exist in determining distances from dynamic force spectra, we show here
484: that even more strikingly it is impossible in general to assess the number of wells and barriers. Again we use a simple example to do so.
485:
486: Fig. \ref{fig4} displays two force spectra obtained using Eq.~\eqref{eq_fr}. They are both well approximated by a succession of three segments with increasing slopes. Again, the two $[\log(r),f^*]$ curves are very similar although they are constructed from landscapes that do not even comprise the same number of peaks and wells.
487:
488: In Fig.~\ref{fig4}\ (a) the three segments describe
489: the escape from the same fundamental state over the three distinct energy barriers. The larger the pulling force the closer the limiting barrier to the fundamental state [see inset in Fig.~\ref{fig4}\ (a)].
490:
491: In Fig.~\ref{fig4}\ (b), the landscape consists of only two barriers.
492: However, the force spectrum reveals that three different escape processes can limit the unbinding kinetic. At low forces ($f\lesssim50 {\rm pN}$) the two observed linear segments results from the escape form the fundamental state over the two peaks at $\widehat{x}_1$ and $\widehat{x}_2$ respectively. Conversely, at high forces it is the escape from the deeply lowered intermediate state over the outer barrier that determines the escape rate (see inset in Fig. \ref{fig4}\ (b), drawing with dash-dot line). With the chosen parameters the effective rate $k_{0,3}$, $k_{0,2}$ and $k_{0,1}$ in Fig. 4(a) case correspond respectively to $k_{0,2}$, $k_{0,1}$ and $k_{1,2}$ in Fig. 4(b) case. Thus the two plots are indistinguishable and cannot be used to predict the number of barriers along the 1D escape path.
493:
494:
495: In conclusion of this subsection, we suggest great care in inferring features
496: of the underlying energy landscape from dynamic force spectroscopy experiment.
497: Our generalized equation may
498: be helpful in dealing with the corresponding ambiguity
499: as it allows (with some work) to generate various landscapes
500: that can account for the observed data, whereas
501: Eq.~\eqref{eq_fr_dbfs} can only yield a single set of parameters (
502: e.g. those used for the plots in Fig.~\ref{fig3} (a) and Fig.~\ref{fig4} (a)).
503:
504: %In summary the two major implications of this first step toward a generic description of the unbinding %kinetic are:
505: %(i) It is impossible to extract the absolute position of the barriers along the pulling direction from a %force spectrum composed of more than one segment without complementary knowledges on the topography of the %energy landscape. Only the separations $\Delta x_{i,j}$ can be inferred unambiguously.
506: %(ii) The number of distinct relevant barriers along the pulling direction do not necessarily corresponds to %the number of linear segments observed in the $[\log(r),f^*]$ plots but only to the number of different %effective escape processes that mostly limit the unbinding kinetic.
507:
508: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
509: \section{Beyond the single escape rate approximation: Multimodal unbinding force distributions}
510: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
511: \begin{figure}[!t]
512: \centerline{\includegraphics[width=\columnwidth]{fig5completedef}}%{fig5Pab22}}
513: \caption{A scenario that yields a multimodal unbinding force distribution.(a) Three snapshots of the energy landscape for the pulling forces: $f=0\,{\rm pN}$ (dotted line), $f=18\,{\rm pN}$ (dahs-dotted line) and $f=60\, {\rm pN}$ (dashed line). The three arrows indicate the corresponding most difficult transitions. Parameters values: $(\widehat{x}_1,\widehat{E}_1)=(0.5\,{\rm nm},16\,\kT)$,
514: $(x_1,E_1)= (1.1\,{\rm nm},4 \,\kT)$, and
515: $(\widehat{x}_2,\widehat{E}_2)=(2.3\,{\rm nm},24\,\kT)$.
516: (b) Force spectrum (full line) associated with the landscape described in (a) plotted using Eq.~\eqref{eq_fr}.
517: The full probability density $p_f(f)$ for unbinding at force $f$ is plotted in gray scale in the inset.
518: $p_f(f)$ has been obtained following the procedure described in~\cite{Bar02}. The circles in the main plot represent the local maxima of the unbinding force distribution $p_f(f)$
519: }
520: \label{fig5}
521: \end{figure}
522:
523: Up to this point we have been considering a generalized theory in which
524: the deeply bound fundamental state (DBFS) approximation is dropped, but
525: the unbinding is still approximated as a simple first-order escape
526: process. Indeed, the validity of Eq.\ (\ref{eq_fr}) relies on the
527: assumption that at any moment the distribution of the populations of
528: the bound states can be well approximated by the distribution
529: corresponding to a homogeneous stationary current.
530:
531: This is, however, not always the case. As we stated earlier, the sum of
532: the
533: $1/k_{i,j}(f)$ terms in Eq.\ (\ref{eq_tau}) is dominated by the
534: smallest effective rate constant
535: $k_{i',j'}(f)$
536: corresponding to the slowest effective transition. A consequence of
537: this is that all the bound states located to the left of barrier $j'$
538: are close to equilibrium (because of the slow outflow over barrier $j'$),
539: and the population of any state located to the right is negligible
540: (because they practically belong to the unbound state). Now, If the
541: slowest transition rate changes from
542: $k_{i',j'}(f)$ to $k_{i'',j''}(f)$ as the loading force $f$ is
543: increased, and
544: if $j''<j'$, then a considerable population might remain in the
545: intermediate bound states between the new and the old limiting
546: barriers, $j''$ and $j'$ respectively. This residual population is
547: incompatible with the new stationary current dominated by
548: $k_{i'',j''}(f)$, and must escape in a different
549: way, yielding a secondary maximum of the unbinding force distribution
550: (see Fig.\ \ref{fig5}b).
551:
552: The escape of the majority of the population (located to the left of
553: the new limiting barrier $j''$) can still be characterized by Eq.\
554: (\ref{eq_tau}) of our generalized theory. On the other hand, we have to
555: slightly modify this formula to describe the escape of the residual
556: population (trapped between the new and old limiting barriers). Since
557: $j''$ is the limiting barrier now, almost the entire residual
558: population can escape without ever jumping backward over barrier $j''$.
559: Therefore, for the residual population we can consider barrier $j''$ as
560: a reflecting boundary, and describe the escape by our general theory in
561: this modified potential. Eq.\ (\ref{eq_tau}), e.g., changes accordingly:
562: \begin{equation}
563: \tau(f)
564: =\sum_{i=j''}^{N-1} \sum_{j=i+1}^N
565: \frac{1}{k_{i,j}(f)}
566: =\sum_{i=j''}^{N-1} \sum_{j=i+1}^N
567: \frac{\e^{-f\Delta x_{i,j}/\kT}}{k_{i,j}(0)} .
568: \label{eq_tau_bis}
569: \end{equation}
570:
571: Consequently, the absolute maximum of the unbinding force distribution
572: always follows the upper envelope of the $N(N+1)/2$ lines, however,
573: some secondary maxima might also appear at lower forces, which follow
574: the upper envelope of only a subset of the lines [comprising
575: $(N-j'')(N-j''+1)/2$ elements]. Such secondary maxima of a multimodal
576: force distribution give important information on the internal structure
577: of the energy landscape of the unbinding path, and makes the
578: determination of the number and positions of the energy wells and
579: barriers less ambiguous.
580: It is actually a nice achievement of our
581: generalized theory to be able to make sense of the segments
582: of secondary maxima in a unique frame for fitting parameters
583: (see e.g. Fig. 5, where the segment corresponding to the secondary maximum
584: corresponds to the transition from the well 1 over the barrier 2,
585: a step neglected in the DBFS approximation).
586: The possibility of a bimodal distribution for
587: the case of a two-state system
588: has already been reported by Strunz et al.
589: \cite{Str00}, and our description systematizes and generalizes their findings.
590:
591: To provide a simple illustration for the somewhat formal discussion
592: above,
593: we also focus on a system consisting of two
594: bound states, as depicted in Fig.\ \ref{fig5} (a).
595: Increasing the force, the limiting transition rate changes from
596: $k_{0,2}$ to $k_{1,2}$ and then to $k_{0,1}$. In the range of the
597: loading rate $r$ between about $10^4$ and $10^5$~pN/s the intermediate
598: bound state (1) has enough time to accumulate a large population, which is
599: then flushed by the $k_{1,2}$ transition before the transition
600: $k_{0,1}$ flushes the rest from the fundamental bound state (0). In the
601: range above $10^6$~pN/s the intermediate bound state (1) cannot
602: accumulate much of the population, but it still possesses a small
603: fraction of the initial equilibrium distribution, which
604: is again flushed by the $k_{1,2}$ transition first.
605:
606: Very recently the group of
607: Evans actually reported the experimental
608: occurrence of a bimodal force distribution
609: \cite{Eva02}.
610: The corresponding experiment consisted in pulling on "diC14 PE" lipids
611: from a bilayer made of "C18:0/1 PC" lipids.
612: With the help of our generalized theory,
613: Evans and Williams were able to fit their data and
614: interpret the results in terms of an energy landscape with
615: two barriers [personal communication,
616: see Ref.\ [3] in Ref.\ \cite{Eva02}].
617:
618: \section{Conclusion}
619: In this paper, we have revisited the standard theory used to account for the dynamic response of molecular stickers. Our refined description, valid for an arbitrarily complex one-dimensional energy landscape, has allowed us to highlight several practical consequences of the diversity of the possible unbinding scenarios. For example several markedly different energy landscapes can yield the same rupture force distribution. To resolve this ambiguity other experimental technics, e.g. flow chamber experiments~\cite{Pie02}, are then required.
620: We have also identified the physical origin of multimodal unbinding force distributions and shown how their analysis provides informations on the unbinding pathways.
621: %We hope that this theoritical analysis will prove usefull in the planning of future dynamic %force spectroscopy experiments.
622:
623:
624: %(i) It is impossible to extract the absolute position of the barriers along the pulling direction from a force spectrum composed of more than one segment without complementary knowledges on the topography of the energy landscape. Only the separations $\Delta x_{i,j}$ can be inferred unambiguously. Hence, other investigation method need to be
625:
626: %(ii) The number of distinct relevant barriers along the pulling direction do not necessarily corresponds to the number of linear segments observed in the $[\log(r),f^*]$ plots but only to the number of different effective escape processes that mostly limit the unbinding kinetic.
627: %(iii)multimodal
628:
629: \vspace{-5pt} %%% DELETE %%%
630: \begin{thebibliography}{}
631: \bibitem{Str03}
632: T. R. Strick, M.-N. Dessinges, G. Charvin, N. H. Dekker, J.-F. Allemand, D. Bensimon and V. Croquette
633: %Streching of macromolecules and proteins
634: Rep. Prog. Phys. {\bf 66}, 1 (2003).
635: %
636: \bibitem{titin}
637: M. Rief, M. Gautel, F. Oesterhelt, J. M. Fernandez, and H. E. Gaub,
638: % Reversible Unfolding of Individual Titin Immunoglobulin Domains by AFM
639: Science {\bf 276}, 1109 (1997);
640: %
641: M. S. Kellermayer, S. B. Smith, H. L. Granzier, and C. Bustamante,
642: % Folding-Unfolding Transitions in Single Titin Molecules
643: % Characterized with Laser Tweezers
644: {\em ibid.}\ {\bf 276}, 1112 (1997).
645:
646: \bibitem{Poi01}
647: M. G. Poirier, A. Nemani, P. Gupta, S. Eroglu, and J. F. Marko,
648: % Probing Chromosome Structure with Dynamic Force Relaxation
649: Phys.\ Rev.\ Lett.\ {\bf 86}, 360 (2001).
650:
651: \bibitem{Sim99}
652: D. A. Simons, M. Strigl, M. Hohenadl, and R. Merkel,
653: %Statistical breakage of single proteine A-IgG Bonds reveals
654: %Crossover from spontaneous to Force-Induced Bond Dissociation
655: Phys.\ Rev.\ Lett.\ {\bf 83}, 652 (1999).
656:
657: \bibitem{Nis00}
658: T. Nishizaka, R. Seo, H. Tadakuma, K. Kinosita, and S. Ishiwata,
659: %Characterization of Single Actomyosin Rigor Bonds: Load Dependence of Lifetime
660: %and Mechanical Properties
661: Biophys.\ J.\ {\bf 79}, 962 (2000).
662:
663: \bibitem{Pie96}
664: A. Pierre, A. M. Benoliel, P. Bongrand, and P. A. van der Merwe,
665: %Determination of the lifetime and force dependence of interactions of single bonds
666: %between surface-attached CD2 and CD48 adhesion molecules
667: Proc.\ Natl.\ Acad.\ Sci.\ USA {\bf 93}, 15114 (1996).
668:
669: \bibitem{Eva}
670: E. Evans and K. Ritchie,
671: % Dynamic strength of molecular adhesion bonds
672: Biophys.\ J.\ {\bf 72}, 1541 (1997);
673: %
674: R. Merkel, P. Nassoy, A. Leung, K. Ritchie, and E. Evans,
675: % Energy landscapes of receptor-ligand bonds explored with dynamic
676: % force spectroscopy
677: Nature {\bf 397}, 50 (1999).
678:
679: \bibitem{Hum01}
680: G. Hummer and A. Szabo,
681: %Free energy reconstruction from nonequilibrium single-molecule pulling experiment
682: Proc.\ Natl.\ Acad.\ Sci.\ USA {\bf 98}, 3658 (2001).
683:
684: \bibitem{Hey00}
685: H. Grubm\"uller, B. Heymann and P. Tavan,
686: % Ligand binding: molecular mechanics calculation of the streptavidin-biotin rupture force.
687: Science,\ {\bf 271}, 997 (1996).
688:
689: \bibitem{Eva01}
690: E. Evans,
691: % Energy landscapes of biomolecular adhesion and receptor anchoring at
692: % interfaces explored with dynamic force spectroscopy
693: Faraday Discuss.\ {\bf 111}, 1 (1998);
694: %
695: E. Evans,
696: % Probing the relation between force lifetime and chemistry in single
697: % molecular bonds
698: Annu.\ Rev.\ Biophys.\ Biomol.\ Struct.\ {\bf 30}, 105 (2001).
699:
700: \bibitem{Sei02}
701: U. Seifert,
702: Europhys. Lett.\ {\bf 58}, 792 (2002).
703:
704: \bibitem{Bar02}
705: D. Bartolo, I. Der\'enyi and A. Ajdari,
706: %Dynamic response of adhesion complexes: Beyond the single path picture
707: Phys. Rev. E. {\bf 65}, 051910-1 (2002).
708:
709: \bibitem{Str00}
710: T. Strunz, K. Oroszlan, I. Schumakovitch, H.-J. G\"untherodt, and M. Hegner,
711: % Model Energy Landscapes and the Force-Induced Dissociation of
712: % Ligand-Receptor Bonds
713: Biophys.\ J.\ {\bf 79}, 1206 (2000).
714:
715: \bibitem{Eva02}
716: E. Evans and P. Williams,
717: % Dynamic Force Spectroscopy,
718: in {\it Physics of bio-molecules and cells},
719: edited by H. Flyvbjerg {\it et al.}
720: (Springer, Berlin, 2002), pp. 145--204.
721:
722: \bibitem{Pie02}
723: A. Pierres, D. Touchard, A.-M. Benoliel, and P. Bongrand,
724: %Dissecting steptavidin-biotin interaction with laminar flow chamber
725: Biophys.\ J.\ {\bf 82}, 3214 (2002).
726: \end{thebibliography}
727:
728: %\bibliography{p}
729:
730: \end{document}
731: