1: \documentclass[11pt]{article}
2: \usepackage{graphicx}% Include figure files
3: \usepackage{bm}% bold math
4: \usepackage{epsfig}
5: \usepackage{amsmath}
6: \usepackage{fullpage}
7: \usepackage[round,numbers,sort&compress]{natbib}
8: \usepackage{endfloat}
9: \bibliographystyle{achemso}
10: %\bibliographystyle{biophysj}
11: %\renewcommand{\bibnumfmt}[1]{#1.}
12:
13:
14: \begin{document}
15:
16: %\addtolength{\oddsidemargin}{-.875in}
17: %\addtolength{\evensidemargin}{-.875in}
18: %\addtolength{\textwidth}{1.75in}
19: %\addtolength{\topmargin}{-.5n}
20: \addtolength{\textheight}{0.5in}
21: %\doublespacing
22:
23: \title{Entropic elasticity of DNA with a permanent kink}
24: \author{Jinyu
25: Li$^1$, Philip C. Nelson$^2$, and M. D. Betterton$^{3*}$\\
26: $^1$Department of Applied Mathematics,
27: University of Colorado at Boulder, \\ Boulder, CO, jinyu.li@colorado.edu\\
28: $^2$ Department of Physics and Astronomy,
29: University of Pennsylvania, \\ Philadelphia, PA, pcn@physics.upenn.edu\\
30: $^3$Department of Physics,
31: University of Colorado at Boulder, \\ Boulder, CO, mdb@colorado.edu \\
32: $^*$Corresponding author. Address: Department of Physics, University
33: of Colorado at Boulder,\\ 390 UCB, Boulder, CO~80309 USA}
34:
35: \maketitle
36:
37: \begin{abstract}
38: Many proteins interact with and deform double-stranded DNA in cells.
39: Single-molecule experiments have studied the elasticity of DNA with
40: helix-deforming proteins, including proteins that bend DNA. These
41: experiments increase the need for theories of DNA elasticity which
42: include helix-deforming proteins. Previous theoretical work on bent
43: DNA has examined a long DNA molecule with many nonspecifically
44: binding proteins. However, recent experiments used relatively short
45: DNA molecules with a single, well-defined bend site. Here we develop
46: a simple,
47: %analytically tractable
48: theoretical description of the
49: effect of a single bend. We then include the description of the
50: bend in the finite worm like chain model (FWLC) of short DNA
51: molecules attached to beads. We predict how the DNA force-extension
52: relation changes due to formation of a single permanent kink, at all
53: values of the applied stretching force. Our predictions show that
54: high-resolution single-molecule experiments could determine the bend
55: angle induced upon protein binding.
56: \end{abstract}
57:
58: Keywords: DNA elasticity, force-extension measurements,
59: helix-deforming proteins, transcription factors, bent DNA, theory.
60:
61: \section{Introduction}
62:
63: In cells, many different interactions between DNA and proteins occur,
64: processes which are essential to gene expression, genome replication,
65: and cellular DNA management. One major class of proteins interacts
66: with DNA and mechanically deforms the double helix by wrapping,
67: looping, twisting, or bending DNA \cite{dicker98,luscom00}. Examples
68: include DNA-packaging proteins and transcription factors which
69: regulate gene expression. The mechanical deformation of the DNA may
70: be important for gene expression: it has been suggested that DNA
71: deformation by transcription factors may help other proteins bind to
72: the DNA and initiate transcription.
73:
74: The deformation of DNA by proteins can be detected in single-molecule
75: force microscopy. In this experimental method, force is applied to
76: individual DNA molecules and the DNA end-to-end extension is measured
77: (figure \ref{bendgeom}). Single-molecule force microscopy has been
78: used to detect the deformation of DNA caused by protein binding
79: \cite{noort04,skoko04,yan04a,broek05,dixit05,mccaul05}. In these
80: experiments the DNA end-to-end extension changes when a
81: deformation-inducing protein binds. Varying the applied force allows
82: one to probe the deformation and better understand the details of the
83: protein-DNA interaction.
84:
85: In this paper we focus on proteins that bend the DNA backbone and
86: develop theoretical predictions of the force-extension behavior of
87: bent DNA. Our description is based on the worm-like chain theory
88: (WLC) \cite{bustam94,marko95,bouch99}. The WLC predicts the average
89: end-to-end extension $z$ of a semiflexible polymer, given the force
90: $F$ applied to the ends of the chain and the values of two constant
91: parameters (the contour length $L$ and the persistence length $A$).
92: However, DNA elastic behavior is altered by backbone-deforming
93: proteins, an effect that is not included in the traditional WLC.
94: Extended theories have been developed which combine the WLC treatment
95: of DNA elasticity with local bends. Rivetti \textit{et al.}\
96: addressed the case of zero applied force \cite{rivet98}, while Yan and
97: Marko have described the changes in the force-extension behavior of a
98: long polymer to which many kink-inducing proteins can bind
99: nonspecifically \cite{yan03}. Similarly, Popov and Tkachenko studied
100: the effects of a large number of reversible kinks \cite{popov05},
101: Metzler \textit{et al.}\ studied loops formed by slip-rings
102: \cite{metz02a} , and Kuli\'c \textit{et al.} studied the high-force
103: limit of a kinked polymer \cite{kulic05}.
104:
105: Previous theoretical work has focused on large numbers of reversible
106: kinks or the limit of low or high applied force. However, recent
107: single-molecule experiments have examined relatively short DNA
108: molecules with a single specific kink site, over a range of applied
109: force \cite{dixit05}. Therefore a theory is needed which applies to
110: ({\it i}) one kink site and ({\it ii}) a polymer of finite contour
111: length ($L/A \sim 1-10$). Recently, we introduced a modified solution
112: of the WLC applicable to polymers of this length, and demonstrated
113: that applying the traditional WLC solution to molecules with $L/A \sim
114: 1-10$ can lead to significant errors \cite{li05}. Our finite worm-like
115: chain solution (FWLC) includes both finite-length effects, often
116: neglected in WLC calculations, and the effect of the rotational
117: fluctuations of a bead attached to the end of the chain.
118:
119: \begin{figure}
120: \begin{center}
121: \includegraphics*[width=2.5in]{bendgeom}
122: \caption{Typical experimental geometry of single-molecule force
123: microscopy measurements. The DNA molecule is attached at one
124: end to a surface and at the other end to a bead (radius $R$).
125: A force $F$ is applied to the bead. (a) DNA molecule in the
126: absence of bound protein. The mean end-to-end extension is $
127: z $. (b) DNA molecule with a single bend-inducing protein
128: bound. The protein bends the DNA backbone through the
129: external angle $\alpha$ at the bend site. As a result, the
130: mean extension decreases to $ z_{bend}$.}
131: \label{bendgeom}
132: \end{center}
133: \end{figure}
134:
135: Here we formulate a theoretical description of a single kink induced
136: by a protein, and extend the FWLC treatment to include such local
137: distortions. Our theory has a simple analytical formulation for the
138: case of a force-independent bend angle, i.e., a rigid protein-DNA
139: complex. Our predictions are relevant to experiments like those of
140: Dixit \textit{et al.}\ \cite{dixit05}, which detect with high
141: resolution a single bend induced in a relatively short DNA molecule.
142: Although we will primarily focus on the case of a single bend angle,
143: our method can also describe a kink which takes on different angles
144: with different probabilities. This model could be relevant to a
145: binding protein that can fluctuate between different binding
146: conformations with different kink angles \cite{parkh01}.
147:
148: \section{Theory}
149:
150: \begin{figure}
151: \begin{center}
152: \includegraphics*[width=3in]{bc-coord}
153: \caption{Boundary conditions and coordinates. (a)
154: Unconstrained. (b) Half-constrained. (c) Normal. (d)
155: Coordinates.}
156: \label{bc-coord}
157: \end{center}
158: \end{figure}
159:
160: \subsection{FWLC theory of unkinked DNA}
161:
162: The classic WLC model \cite{marko95} and the FWLC theory \cite{li05},
163: which includes finite-length effects, describe an inextensible polymer
164: with isotropic bending rigidity. The bending rigidity is
165: characterized by the persistence length, $A$, the length scale over
166: which thermal fluctuations randomize the chain orientation. We assume
167: that the twist is unconstrained and can be neglected (as is the case,
168: for example, in optical tweezer experiments).
169:
170: The chain energy function includes terms which represent the bending
171: energy and the work done by the applied force:
172: \begin{equation}
173: E=\int_0^\ell ds \left( \frac{\kappa^2}{2} - f {\bf \hat{\bf z}}\cdot{\bf
174: \hat{t}} \right),
175: \end{equation}
176: where $E$ is the energy divided by the thermal energy $k_BT$,
177: $\ell=L/A$, $s$ denotes arc length divided by the persistence length
178: $A$, and all other lengths are similarly measured in units of $A$. The
179: quantity $f$ is the force multiplied by $A/k_BT$, and we assume the
180: force is applied in the $\hat{\bf z}$ direction. The total extension
181: of the chain is $z=\int ds\ {\bf \hat{\bf z}}\cdot{\bf \hat{t}}$. The
182: curvature $\kappa$ can be defined in terms of arc-length derivatives
183: of the chain coordinate (figure \ref{bc-coord}d). If the chain
184: conformation is described by a space curve ${\bf r}(s)$ and the unit
185: vector tangent to the chain is ${\bf \hat{t}}(s)$, then $\kappa =
186: \left| \frac{\partial^2 {\bf r}}{\partial s^2}\right|= \left|
187: \frac{\partial {\bf \hat{t}}}{\partial s}\right|$.
188:
189: The chain partition function weights contributions from different
190: polymer conformations \cite{fixman73,yamak76}. If the ends of the
191: chain are held at fixed orientations, we have
192: \begin{equation}
193: Z({\bf \hat{t}}_f,\ell;{\bf \hat{t}}_i,0)
194: =\int D{\bf \hat{t}} \ \exp \left[- \int_0^\ell ds \left( \frac{1
195: }{2}(\partial_s{\bf \hat{t}})^2 - f {\bf \hat{\bf z}}\cdot{\bf
196: \hat{t}} \right)\right],
197: \end{equation}
198: where the integral in $D{\bf \hat{t}}$ is over all possible paths
199: between the two endpoints of the chain with the specified
200: orientations. The partition function can be interpreted as a
201: propagator which connects the probability distribution for the tangent
202: vector at point $s$, $\psi({\bf \hat{t}},s)$ to the same probability
203: distribution at point $s'$:
204: \begin{equation}
205: \psi({\bf \hat{t}},s) = \int d{\bf \hat{t}}' \ Z({\bf \hat{t}},s;{\bf
206: \hat{t}}',s')\ \psi({\bf \hat{t}}',s').
207: \label{propagator}
208: \end{equation}
209: From this relation, one can derive a Schr\"odinger-like equation, which
210: describes the $s$ evolution of $\psi$ \cite{marko95}:
211: \begin{equation}
212: \frac{\partial \psi}{\partial s}=\left(\frac{\nabla^2}{2}+f \cos
213: \theta \right) \psi.
214: \label{hamilt}
215: \end{equation}
216: Here $\nabla^2$ is the two-dimensional Laplacian on the surface of the
217: unit sphere and $\cos \theta= \hat{\bf z} \cdot \hat{\bf t}$.
218:
219: For relatively short DNA molecules ($\ell \sim 1-10$), the boundary
220: conditions at the ends of the chain \cite{li05,samuel02} and
221: bead rotational fluctuations become important. The boundary
222: conditions are specified by two probability density functions,
223: $\psi({\bf \hat{t}},s=0)$ and $\psi({\bf \hat{t}},s=L)$. The boundary
224: conditions modify the force-extension relation, and enter the full
225: partition function matrix element via
226: \begin{equation}
227: \label{fullpart}
228: Z_{tot}=\int d{\bf \hat{t}}_i \ d {\bf \hat{t}}_f \ \psi({\bf
229: \hat{t}}_i,0) Z({\bf \hat{t}}_i,0;{\bf \hat{t}}_f,L) \psi({\bf
230: \hat{t}}_f,L).
231: \end{equation}
232: Rotational fluctuations of the bead(s) attached to the end of the DNA
233: complicate the analysis of experiments.
234: %are a source of experimental error in single-molecule force
235: %microscopy; the error increases for shorter DNA molecules. Rotational
236: %fluctuations of the bead are explicitly included in the FWLC theory
237: %as
238: What is observed and controlled is not the endpoint of the (invisible)
239: DNA chain, but rather the bead's center. The relation between these
240: distinct points fluctuates as the bead performs rotational Brownian
241: motion. The FLWC theory accounts for these fluctuations via
242: %
243: an effective boundary condition at the end(s) of the chain, which
244: depends on applied force, bead radius, and the nature of the link
245: joining the bead to the polymer \cite{li05}. We will study boundary
246: conditions that are azimuthally symmetric; thus our end boundary
247: conditions will be functions of $\hat{\bf t}\cdot \hat{\bf z}$ only.
248:
249: \subsection{Fixed-angle bend}
250:
251: %Consider a bend where the angle is fixed, independent of applied
252: %force. Note that we neglect force dependence of the bend angle and
253: We now suppose that our chain contains a permanent bend, whose
254: location along the DNA, and angle, are fixed, independent of applied
255: force. In this paper we will also neglect
256: %
257: force-induced unbinding of the deforming protein. (These effects are
258: straightforward to incorporate into our analysis.) In addition, we
259: neglect twist stiffness, which is legitimate since we wish to study a
260: single bend in a polymer with unconstrained twist. (Twist stiffness
261: effects will be important for experiments in which multiple bends
262: occur or twist is constrained.)
263:
264: The Schr\"odinger-like equation \eqref{hamilt} must be modified by the
265: inclusion of a ``bend operator'' which transforms $\psi$ at the bend.
266: Suppose that the kink occurs at position $s_o$. Given the
267: tangent-vector probability distribution $\psi$ at $s_o-\epsilon$
268: (where $\epsilon$ is infinitesimal), our goal is to determine
269: $\psi({\bf \hat{t}},s_o+\epsilon)$, the distribution just after the kink. If we
270: denote the exterior angle of the kink by $\alpha$, then ${\bf
271: \hat{t}}_{s_o-\epsilon} \cdot {\bf \hat{t}}_{s_o+\epsilon}=\cos
272: \alpha$. Because twist is unconstrained, we may average over
273: rotations; effectively, the bend occurs with uniform probability in
274: the azimuthal angle: if ${\bf \hat{t}}_{s_o-\epsilon}$ points directly
275: along the $\hat{\bf z}$-axis, then ${\bf \hat{t}}_{s_o+\epsilon}$ is
276: uniformly distributed in a cone at angle $\alpha$ to the $\hat{\bf
277: z}$-axis.
278:
279: The bend operator then can be written using the kernel
280: \begin{equation}
281: K_{\alpha}({\bf \hat{t}},{\bf \hat{t}}')=\frac{1}{2 \pi} \delta({\bf
282: \hat{t}}\cdot {\bf \hat{t}}'-\cos \alpha).
283: \label{bendop}
284: \end{equation}
285: The probability distribution $\psi({\bf \hat{t}}',s_o-\epsilon)$ of
286: tangent-vector angles just before the kink is related to the
287: distribution $\psi({\bf \hat{t}},s_o+\epsilon)$ just after the kink by
288: \begin{equation}
289: \psi({\bf \hat{t}},s_o+\epsilon)= \int d {\bf \hat{t}}'\
290: K_{\alpha}({\bf \hat{t}},{\bf \hat{t}}') \ \psi({\bf
291: \hat{t}}',s_o-\epsilon).
292: \label{psitrans}
293: \end{equation}
294: Below (section \ref{bendopcalc}) we show that spherical harmonics
295: diagonalize the operator \eqref{bendop}.
296: %transformation from $\psi({\bf \hat{t}},s_o-\epsilon)$
297: %to $\psi({\bf \hat{t}},s_o+\epsilon)$.
298:
299: \subsubsection{Distribution of bend angles}
300:
301: Suppose that the bend occurs not for a single fixed angle, but a
302: distribution of angles. We assume that ${\bf \hat{t}}_{s_o-\epsilon}
303: \cdot {\bf \hat{t}}_{s_o+\epsilon}=\cos \alpha=u$ is distributed
304: according to the probability density function $h(u)$, where $h$ is
305: normalized so that $\int d \phi \int du \ h(u)=1$. Then the
306: bend-operator kernel can be written as an integral over the
307: probability distribution:
308: \begin{equation}
309: K_{h}({\bf \hat{t}},{\bf \hat{t}}')=\frac{1}{2 \pi} \int_{-1}^{1} du \
310: h(u)\ \delta({\bf \hat{t}}\cdot {\bf \hat{t}}'-u).
311: \label{bendopdist}
312: \end{equation}
313:
314:
315:
316: \section{Calculation}
317:
318: The main quantity of interest in single-molecule experiments is the
319: force-extension relation, which can be determined by solving equation
320: \eqref{hamilt} for the tangent-vector probability distribution
321: $\psi({\bf \hat{t}},s)$. The Schr\"odinger-like equation is solved
322: using separation of variables in $s$ and ${\bf \hat{t}}$, where the
323: angular dependence is expanded in spherical harmonics \cite{marko95}.
324: \begin{equation}
325: \psi({\bf \hat{t}},s) = \sum_{j=0}^{\infty} \Psi_j(s) Y_{j0}({\bf
326: \hat{t}}).
327: \label{psiser}
328: \end{equation}
329: (By azimuthal symmetry, only the $m=0$ terms will enter in our
330: formulae.) In the basis of spherical harmonics, the operator in
331: equation (\ref{hamilt}) is a symmetric tridiagonal matrix $H$
332: %added for referee
333: with diagonal terms
334: \begin{equation}
335: H_{j,j}=-\frac{j(j+1)}{2},
336: \end{equation}
337: and off-diagonal terms
338: \begin{equation}
339: H_{j,j+1}=\frac{f(j+1)}{\sqrt{(2j+1)(2j+3)}}.
340: \end{equation}
341: %
342: The vector of coefficients at $s$ is ${\bf \Psi}(s)=e^{sH}{\bf
343: \Psi}(0)$ \cite{marko95,li05}. This expression for $\psi({\bf
344: \hat{t}},s)$ is exact if the infinite series of spherical harmonics
345: is used.
346:
347: \subsection{Force-extension relation}
348:
349: Given the boundary conditions ${\bf \Psi}(s=0)$ and ${\bf
350: \Psi}(s=\ell)$, the partition function is
351: \begin{eqnarray}
352: Z&=&{\bf \Psi}^T(s=\ell)e^{\ell H}{\bf \Psi}(s=0),\\
353: &=&\sum_{j,k} \Psi_j(s=\ell) [e^{\ell H}]_{jk} \Psi_k(s=0).
354: \label{zinnprod}
355: \end{eqnarray}
356: The fractional extension of the chain is
357: \begin{equation}
358: \frac{z}{L}=\frac{1}{\ell}\frac{\partial \ln Z}{\partial f}.
359: \label{forcextf}
360: \end{equation}
361: We work in the ensemble relevant to most experiments, where the
362: extension is determined for fixed applied force (different ensembles
363: are not equivalent for single finite-length molecules
364: \cite{dhar02,keller03,sinha05}). Equation \eqref{forcextf} applies
365: for a chain of any length. However, we can show the structure of the
366: partition function more clearly by separating $\ln Z$ into two terms:
367: one representing an infinite chain and a finite-length correction
368: \cite{li05}. Let $B=e^H$, denote by $\lambda_*$ the largest
369: eigenvalue of $B$, and let ${\cal B}=B/ \lambda_*$. Then ${\cal B}$
370: has eigenvalues with magnitude less than or equal to 1 and the
371: logarithm of the partition function can be written
372: \begin{equation}
373: \ln Z= \ell \ln \lambda_* + \ln [{\bf \Psi}^T(s=\ell){\cal B}^{\ell}
374: {\bf \Psi}(s=0)].
375: \label{znew}
376: \end{equation}
377: Only the first term is considered in the usual WLC solution; the
378: second term is the finite-length correction \cite{li05}.
379: % added to address the referee's comments
380: Equation \eqref{znew} is an exact expression for $\ln Z$ which is
381: difficult to evaluatae analytically. We numerically calculate the
382: force-extension relation by using equation \eqref{znew} with the
383: series truncated after $N$ terms. This expression can be accurately
384: numerically calculated, and the truncation error determined by
385: comparing the results with different $N$. Our calculations use $N=30$
386: unless otherwise specified.
387: %
388:
389: \subsubsection{Boundary conditions and bead rotational fluctuations}
390:
391: The boundary conditions at $s=0$ and $s=\ell$ affect the
392: force-extension relation, because they alter the partition function as
393: shown in equation \eqref{znew}. Some experiments appear to implement
394: ``half-constrained'' boundary conditions, where the polymer is
395: attached to a planar wall by a freely rotating attachment point, and
396: the wall is perpendicular to the direction of applied force
397: \cite{nelson05}. In this case the tangent vector at the end of the
398: chain can point in any direction on the hemisphere outside the
399: impenetrable surface (figure \ref{bc-coord}(b)). The effects of
400: different boundary conditions on the force-extension relation are
401: considered in detail in reference \cite{li05}. In the
402: ``unconstrained'' boundary condition the tangent vector at the end of
403: the chain is free to point in any direction on the sphere (in $4 \pi$
404: of solid angle, figure \ref{bc-coord}a). In this case $\psi(\hat{\bf
405: t})$ is independent of $\cos \theta$ and ${\bf
406: \Psi}=(1,0,\cdots,0)$. In the ``half-constrained'' boundary
407: conditions (figure \ref{bc-coord}b), the tangent vector at the end of
408: the chain can point in any direction on the hemisphere outside the
409: impenetrable surface; then the leading coefficients of ${\bf \Psi}$
410: are 1, 0.8660, 0, -0.3307, 0, 0.2073, 0. In the ``normal'' boundary
411: condition, the tangent vector at the end of the chain is parallel to
412: the $\hat{\bf z}$ axis, normal to the surface (figure
413: \ref{bc-coord}c). Then the coefficients of ${\bf \Psi}$ are all equal
414: to 1 \cite{normnote}.
415:
416: The FWLC formulation can also average over rotational fluctuations of
417: spherical bead(s) attached to one or both ends of the polymer chain.
418: The result is an effective boundary condition that depends on applied
419: force and bead radius \cite{li05}. Both the case of perpendicular wall
420: attachment and bead attachment generate boundary conditions that are
421: invariant under rotations about the $\hat{z}$ axis (the direction in
422: which force is applied), and hence give boundary states of the form
423: given in equation \eqref{psiser}.
424:
425:
426:
427: \subsection{Bend operator}
428: \label{bendopcalc}
429:
430: We wish to represent the bend operator (equation \eqref{psitrans}) in
431: terms of spherical harmonics; the operator is diagonal in this basis.
432: Denote $x={\bf \hat{t}}\cdot {\bf \hat{t}}'$ and note that any
433: function of $x$ with $-1\le x\le 1$ can be written as a series of
434: Legendre polynomials \cite{jackson}:
435: \begin{equation}
436: K_{\alpha}(x)=\sum_{l=0}^{\infty} k_l P_l(x).
437: \label{k1}
438: \end{equation}
439: The $k_l$ are determined by projecting the kernel $K$ onto the
440: Legendre polynomials, using
441: the
442: normalization relation
443: $\int_{-1}^{1} P_{l'}(x) P_l(x) dx = \frac{2}{2l+1} \delta_{l l'}$.
444: Therefore
445: \begin{eqnarray}
446: k_l&=&\frac{2l+1}{4 \pi} \int_{-1}^{1} \delta(x-\cos \alpha) P_l(x) dx,\\
447: &=&\frac{2l+1}{4 \pi} P_l(\cos \alpha). \label{coeff}
448: \end{eqnarray}
449: Next we use the addition theorem for spherical harmonics
450: \cite{jackson}
451: \begin{equation}
452: P_l({\bf \hat{t}}\cdot {\bf \hat{t}}')=\frac{4 \pi}{2l+1} \sum_{m=-l}^{l}
453: Y_{lm}^*(\hat{\bf t}') Y_{lm}(\hat{\bf t}), \label{addthm}
454: \end{equation}
455: Substituting equations \eqref{addthm} and \eqref{coeff} in equation
456: \eqref{k1}, we have
457: \begin{equation}
458: K_{\alpha}({\bf \hat{t}},{\bf \hat{t}}')= \sum_{l=0}^{\infty} P_l(\cos
459: \alpha) \sum_{m=-l}^{l}
460: Y_{lm}^*(\hat{\bf t}' ) Y_{lm}(\hat{\bf t} ).
461: \label{ksph}
462: \end{equation}
463: Note that if $\alpha=0$, the kink operator reduces to the identity
464: because $P_l(1)=1$.
465:
466: The probability distribution $\psi$ just before the bend is
467: \begin{equation}
468: \psi({\bf \hat{t}}',s_o-\epsilon) = \sum_{j=0}^{\infty}
469: \sum_{k=-j}^{j} \Psi_{jk}(s_o-\epsilon)
470: Y_{jk}(\hat{\bf t}').
471: \label{series2}
472: \end{equation}
473: Note that in the case of azimuthal symmetry, the terms with $k \neq 0$
474: are zero. To determine $\psi$ just after the bend, we substitute the
475: expressions in equations \eqref{ksph} and \eqref{series2} into the
476: formula
477: \begin{equation}
478: \psi({\bf \hat{t}},s_o+\epsilon)= \int d {\bf \hat{t}}' \ K_{\alpha}({\bf
479: \hat{t}},{\bf \hat{t}}') \ \psi({\bf \hat{t}}',s_o-\epsilon).
480: \label{trans}
481: \end{equation}
482: The expression simplifies by the orthonormality of spherical
483: harmonics:
484: \begin{eqnarray}
485: \psi({\bf \hat{t}},s_o+\epsilon)&=& \sum_{l=0}^{\infty} P_l(\cos
486: \alpha) \sum_{m=-l}^{l} Y_{lm}(\hat{\bf t} )
487: \sum_{j=0}^{\infty} \sum_{k=-j}^j \Psi_{jk}(s_o-\epsilon) \int d {\bf
488: \hat{t}}' \ Y_{lm}^*(\hat{\bf t}' )
489: Y_{jk}(\hat{\bf t}' ) \\
490: &=& \sum_{l=0}^{\infty} P_l(\cos
491: \alpha) \sum_{m=-l}^{l} Y_{lm}(\hat{\bf t} )
492: \sum_{j=0}^{\infty} \sum_{k=-j}^j \Psi_{jk}(s_o-\epsilon) \delta_{jl}
493: \delta_{mk} \\
494: &=& \sum_{l=0}^{\infty}\sum_{m=-l}^{l} P_l(\cos
495: \alpha) \Psi_{lm}(s_o-\epsilon) Y_{lm}(\hat{\bf t} ).
496: \end{eqnarray}
497: The transformation can thus be written
498: $\Psi_{lm}(s_o+\epsilon)=P_l(\cos \alpha) \Psi_{lm}(s_o-\epsilon)$.
499: The probability distribution just after the kink differs from the
500: distribution before the kink only in the multiplication of each term
501: in the series by $P_l(\cos \alpha)$. We can represent the
502: transformation by a diagonal matrix $W$ such that
503: \begin{equation}
504: {\bf
505: \Psi}(s_o+\epsilon)=W {\bf \Psi}(s_o-\epsilon).
506: \end{equation}
507: Because $\psi$ is azimuthally symmetric (only the $m = 0$ terms appear
508: in the series expansion), $W$ has entries $W_{l,l}=P_{l-1}(\cos
509: \alpha)$.
510:
511: \subsubsection{Distribution of bend angles}
512:
513: The representation of the bend operator in terms of spherical
514: harmonics remains simple when the bend contains a distribution of
515: angles described by $h(u)$ (equation \eqref{bendopdist}). As above,
516: we expand $K_h(x)$ in Legendre polynomials, $
517: K_{h}(x)=\sum k_l P_l(x). $ The $k_l$ are the projection of $h(x)$
518: onto Legendre polynomials:
519: \begin{equation}
520: k_l=\frac{2l+1}{4 \pi} \int_{-1}^{1} dx\ h(x)
521: P_l(x). \label{coeff_dist}
522: \end{equation}
523: The calculation is then identical to the case of a single bend angle,
524: with the result $\Psi_{lm}(s_o+\epsilon)=k_l \Psi_{lm}(s_o-\epsilon)$.
525: We can represent the transformation by a diagonal matrix $W_h$ such
526: that
527: \begin{equation}
528: {\bf
529: \Psi}(s_o+\epsilon)=W_h {\bf \Psi}(s_o-\epsilon).
530: \end{equation}
531:
532:
533: \subsection{Force-extension relation with bend}
534:
535: Once the matrix $W$ (which represents the bend operator in the basis
536: of spherical harmonics) has been determined, calculation of the
537: force-extension relation is straightforward. Suppose a single bend
538: occurs at fractional position $a$ along the chain. The partition
539: function with a bend is
540: \begin{equation}
541: Z_b={\bf \Psi}^T(s=\ell)e^{(1-a) \ell H} W e^{a \ell H}{\bf \Psi}(s=0),\\
542: \label{zbend}
543: \end{equation}
544: As before, we let $B=e^H$, denote by $\lambda_*$ the largest
545: eigenvalue of $B$, and define ${\cal B}=B/ \lambda_*$. Using $e^{\ell
546: H} = \lambda_*^{\ell} {\cal B}^{\ell}$, the logarithm of the
547: partition function is
548: \begin{equation}
549: \ln Z_b= \ell \ln \lambda_* + \ln [{\bf \Psi}^T(s=\ell)\ {\cal
550: B}^{(1-a) \ell} W {\cal B}^{a \ell} \ {\bf \Psi}(s=0)].
551: \label{znewbend}
552: \end{equation}
553: As above, the extension is
554: $z/L=\ell^{-1}\ \partial \ln Z/\partial f.$
555:
556:
557: \section{Results}
558:
559:
560: Here we predict the magnitude of extension change induced by a single
561: bend, in order to understand when such single-bending events will be
562: experimentally detectable. We describe how the extension change
563: induced by a bend depends on applied force, bend angle,
564: contour length, and the position of the bend.
565:
566:
567: In figure \ref{forcext} we show the change in extension induced by a
568: bend: the extension of the chain without the bend minus the extension
569: of the chain with the bend. As expected, the extension change is
570: larger when the bend angle is larger. In addition, we find that the
571: change in extension has a maximum near an applied force of 0.1 pN. At
572: this force, the change in extension due to the bend is a significant
573: fraction of the persistence length (10-30 nm for $A=50$ nm).
574:
575:
576: \begin{figure}
577: \begin{center}
578: \includegraphics*[width=2.5in]{forcext_bend_b1}
579: \includegraphics*[width=3in]{forcext_bend_b2}
580: % \includegraphics*[width=2.75in]{forcext_bend_b3}
581: \caption{Change in extension due to a bend as a function
582: of force, determined by subtracting the extension of the chain
583: with the bend from the extension without the bend. The contour
584: length is (A) $L$=200 nm and (B) $L$=500 nm. For larger
585: contour lengths, the prediction is similar to (B). We assume
586: $A=50$ nm, the bend is at the midpoint of the polymer, a bead
587: of radius 250 nm is attached to one end of the chain, and
588: half-constrained boundary conditions. For the largest bend
589: angle, we show for comparison the prediction of Kuli\'c
590: \textit{et al.} \cite{kulic05}, which is valid in the
591: high-force limit.}
592: \label{forcext}
593: \end{center}
594: \end{figure}
595:
596: %KULIC COMPARISON
597: For the largest bend angle, we show for comparison the prediction of
598: Kuli\'c \textit{et al.} \cite{kulic05}. The Kuli\'c \textit{et al.}
599: result is valid in the high-force limit, and we find that their
600: prediction and our result converge as the force becomes large. The
601: Kuli\'c \textit{et al.} result is valuable because it is a simple
602: analytical expression. Although our results are obtained numerically,
603: they are valid over the entire force range.
604: %KULIC COMPARISON
605:
606: As the applied force increases, the polymer becomes more stretched and
607: aligned with the force, decreasing the effect of the bend. For a
608: classical elastic rod where thermal fluctuations are a weak
609: perturbation the characteristic propagation length of elastic
610: deformations is $\sqrt{k_b T\ A /F}$. Therefore, as the force
611: increases, the region of the chain experiencing a significant
612: deflection due to the bend drops. By this argument, one might expect
613: that the largest change in extension due to the bend will occur for
614: the lowest values of the applied force. However, as the force applied
615: to the ends of the polymer goes to zero, the extension also approaches
616: zero (on average, there will be no separation of the two ends). In
617: this case the change in extension due to the bend approaches zero.
618: The effect of the bend is therefore largest at intermediate force,
619: where the molecule is extended by the force but not fully extended.
620:
621: We predicted the change in extension due to the bend with and without
622: a bead attached to one end of the DNA, and for different values of the
623: bead radius. In all cases, we predict similar values for the change in
624: extension due to a bend (not shown).
625:
626: \begin{figure}
627: \begin{center}
628: \includegraphics*[width=2in]{anglext_bend_b1}
629: \includegraphics*[width=2in]{anglext_bend_b2}
630: % \includegraphics*[width=2in]{anglext_bend_b3}
631: \caption{Change in extension due to a bend as a function
632: of angle. (A) F=0.1 pN. (B) F=1 pN.
633: %(C) F=10 pN.
634: We assume $A=50$ nm, the bend is at the midpoint of the
635: polymer, a bead of radius 250 nm is attached to one end of the
636: chain, and half-constrained boundary conditions.}
637: \label{anglext}
638: \end{center}
639: \end{figure}
640:
641:
642: In figure \ref{anglext} we show how the change in extension induced by
643: the bend varies with bend angle. The dependence of the extension
644: change on angle is strong, suggesting that high-resolution experiments
645: could measure the bend angle by measuring the change in extension due
646: to a bend. For larger values of the applied force ($F\ge$1 pN), the
647: result is independent of contour length of the polymer. However at low
648: force ($F$=0.1 pN), where the change in extension due to a bend is
649: largest, the results depend on the polymer contour length.
650:
651:
652: \begin{figure}
653: \begin{center}
654: \includegraphics*[width=2in]{bend_pos_1}
655: \includegraphics*[width=2in]{bend_pos_2}
656: % \includegraphics*[width=2in]{bend_pos_3}
657: \caption{Change in extension due to a bend as a function
658: of the position of the bend along the polymer. (A) F=0.1 pN.
659: (B) F=1 pN. %(C) F=10 pN.
660: %
661: Note the difference in scale between the two panels.
662: %
663: We assume $A=50$ nm, $L=$500 nm, a bead of radius 250 nm is
664: attached to one end of the chain, and half-constrained
665: boundary conditions. }
666: \label{bendpos}
667: \end{center}
668: \end{figure}
669:
670: The dependence on the position of the bend is weak, unless the bend is
671: within a few percent of one end of the polymer (figure
672: \ref{bendpos}).
673: %
674: We note that the curves in figure \ref{bendpos} are
675: not reflection symmetric about the middle of the polymer. This occurs
676: because we assume one end of the polymer ($s=0$) is attached to a fixed
677: surface, while the other end of the polymer ($s=1$) is attached to a
678: bead which can undergo rotational fluctuations. We chose to plot this
679: case because it is a typical experimental geometry; in the case that
680: both ends of the polymer experience identical boundary conditions,
681: then the effects of a bend obey reflection symmetry about the middle
682: of the polymer.
683: %
684:
685:
686: \section{Discussion}
687:
688:
689: We have described a theory of DNA elasticity applicable to bent DNA
690: molecules. The finite worm-like chain model (FWLC) of polymer
691: elasticity extends the WLC to polymers with $L/A \sim 1-10$
692: \cite{li05}. The FWLC includes chain-end boundary conditions and
693: rotational fluctuations of a bead attached to the end of the polymer,
694: modifications which are important for polymers with contour length a
695: few times the persistence length.
696:
697: This work allows predictions of DNA force-extension behavior when a
698: single bend occurs at a specified point along the chain. When the bend
699: angle is constant (independent of applied force) the bend operator is
700: diagonal in the basis of spherical harmonics, allowing straightforward
701: calculation of the effects of a bend. This mathematical description of
702: a bend is suitable both for a bend with a single angle and for bends
703: with a distribution of different bend angles.
704:
705: We demonstrate that the change in polymer end-to-end extension induced
706: by the bend can be a significant fraction of the polymer persistence
707: length: $\Delta z/A \sim 0.2-0.7$ for bend angles of 90-180$^o$, or
708: $\Delta z \sim 10-35$ nm for dsDNA, which has persistence length of
709: approximately 50 nm. The change in extension due to the bend is
710: predicted to show a maximum for applied force around 0.1 pN; for
711: larger force the polymer conformation becomes highly extended and the
712: influence of the bend decreases, while for low force the polymer
713: extension approaches zero, independent of the presence of the bend.
714:
715:
716: The alterations in polymer extension induced by the bend should be
717: detectable in high-resolution single-molecule experiments. Since
718: recent work in single-molecule optical trapping with DNA has
719: demonstrated a resolution of a few nm\cite{perkin04,nugen04}, DNA
720: extension changes of 10-35 nm due to a bend should be detectable.
721: Furthermore, the predicted change in extension strongly depends on the
722: bend angle, suggesting that high-resolution single-molecule
723: experiments could directly estimate the angle of a protein-induced
724: bend.
725:
726:
727: \section*{Acknowledgements}
728:
729: We thank Igor Kuli\'c, Tom Perkins, Rob Phillips, and Michael Woodside
730: for useful discussions, and the Aspen Center for Physics, where part
731: of this work was done. PCN acknowledges support from NSF grant
732: DMR-0404674 and the NSF-funded NSEC on Molecular Function at the
733: Nano/Bio Interface, DMR-0425780. MDB acknowledges support from NSF
734: NIRT grant PHY-0404286, the Butcher Foundation, and the Alfred P.
735: Sloan Foundation. MDB and PCN acknowledge the hospitality of the
736: Kavli Institute for Theoretical Physics, supported in part by the
737: National Science Foundation under Grant PHY99-07949.
738:
739: %\bibliography{fwlc_bend}
740:
741: \providecommand{\refin}[1]{\\ \textbf{Referenced in:} #1}
742: \begin{thebibliography}{10}
743:
744: \bibitem{dicker98}
745: Dickerson,~R.~E. \textit{Nucleic Acids Research} \textbf{1998,} \textsl{26,}
746: 1906-1926.
747:
748: \bibitem{luscom00}
749: Luscombe,~N.;\ \ Austin,~S.;\ \ Berman,~H.;\ \ Thornton,~J. \textit{Genome
750: Biology} \textbf{2000,} \textsl{1,} 1.
751:
752: \bibitem{noort04}
753: van Noort,~J.;\ \ Verbrugge,~S.;\ \ Goosen,~N.;\ \ Dekker,~C.;\ \ Dame,~R.~T.
754: \textit{Proceedings of the National Academy of Sciences of the United States
755: of America} \textbf{2004,} \textsl{101,} 6969-6974.
756:
757: \bibitem{skoko04}
758: Skoko,~D.;\ \ Wong,~B.;\ \ Johnson,~R.~C.;\ \ Marko,~J.~F.
759: \textit{Biochemistry} \textbf{2004,} \textsl{43,} 13867-13874.
760:
761: \bibitem{yan04a}
762: Yan,~J.;\ \ Skoko,~D.;\ \ Marko,~J.~F. \textit{Physical Review E}
763: \textbf{2004,} \textsl{70,} 011905.
764:
765: \bibitem{broek05}
766: van~den Broek,~B.;\ \ Noom,~M.~C.;\ \ Wuite,~G. J.~L. \textit{Nucleic Acids
767: Research} \textbf{2005,} \textsl{33,} 2676-2684.
768:
769: \bibitem{dixit05}
770: Dixit,~S.;\ \ Singh-Zocchi,~M.;\ \ Hanne,~J.;\ \ Zocchi,~G. \textit{Physical
771: Review Letters} \textbf{2005,} \textsl{94,} 118101.
772:
773: \bibitem{mccaul05}
774: McCauley,~M.;\ \ Hardwidge,~P.~R.;\ \ Maher,~L.~J.;\ \ Williams,~M.~C.
775: \textit{Biophysical Journal} \textbf{2005,} \textsl{89,} 353-364.
776:
777: \bibitem{bustam94}
778: Bustamante,~C.;\ \ Marko,~J.~F.;\ \ Siggia,~E.~D.;\ \ Smith,~S.
779: \textit{Science} \textbf{1994,} \textsl{265,} 1599-1600.
780:
781: \bibitem{marko95}
782: Marko,~J.~F.;\ \ Siggia,~E.~D. \textit{Macromolecules} \textbf{1995,}
783: \textsl{28,} 8759-8770.
784:
785: \bibitem{bouch99}
786: Bouchiat,~C.;\ \ Wang,~M.~D.;\ \ Allemand,~J.~F.;\ \ Strick,~T.;\ \
787: Block,~S.~M.;\ \ Croquette,~V. \textit{Biophysical Journal} \textbf{1999,}
788: \textsl{76,} 409-413.
789:
790: \bibitem{rivet98}
791: Rivetti,~C.;\ \ Walker,~C.;\ \ Bustamante,~C. \textit{Journal of Molecular
792: Biology} \textbf{1998,} \textsl{280,} 41-59.
793:
794: \bibitem{yan03}
795: Yan,~J.;\ \ Marko,~J.~F. \textit{Physical Review E} \textbf{2003,} \textsl{68,}
796: 011905.
797:
798: \bibitem{popov05}
799: Popov,~Y.~O.;\ \ Tkachenko,~A.~V. \textit{Physical Review E} \textbf{2005,}
800: \textsl{71,} 051905.
801:
802: \bibitem{metz02a}
803: Metzler,~R.;\ \ Kantor,~Y.;\ \ Kardar,~M. \textit{Phys. Rev. E} \textbf{2002,}
804: \textsl{66,} 022102.
805:
806: \bibitem{kulic05}
807: Kulic,~I.~M.;\ \ Mohrbach,~H.;\ \ Lobaskin,~V.;\ \ Thaokar,~R.;\ \
808: Schiessel,~H. \textit{Physical Review E} \textbf{2005,} \textsl{72,} 041905.
809:
810: \bibitem{li05}
811: Li,~J.;\ \ Nelson,~P.~C.;\ \ Betterton,~M.~D. ``DNA entropic elasticity for
812: short molecules attached to beads'', 2005
813: http://arxiv.org/abs/physics/0601185.
814:
815: \bibitem{parkh01}
816: Parkhurst,~L.~J.;\ \ Parkhurst,~K.~M.;\ \ Powell,~R.;\ \ Wu,~J.;\ \
817: Williams,~S. \textit{Biopolymers} \textbf{2001,} \textsl{61,} 180-200.
818:
819: \bibitem{fixman73}
820: Fixman,~M.;\ \ Kovac,~J. \textit{Journal of Chemical Physics} \textbf{1973,}
821: \textsl{58,} 1564-1568.
822:
823: \bibitem{yamak76}
824: Yamakawa,~H. \textit{Pure and Applied Chemistry} \textbf{1976,} \textsl{46,}
825: 135-141.
826:
827: \bibitem{samuel02}
828: Samuel,~J.;\ \ Sinha,~S. \textit{Physical Review E} \textbf{2002,} \textsl{66,}
829: 050801.
830:
831: \bibitem{dhar02}
832: Dhar,~A.;\ \ Chaudhuri,~D. \textit{Physical Review Letters} \textbf{2002,}
833: \textsl{89,} 065502.
834:
835: \bibitem{keller03}
836: Keller,~D.;\ \ Swigon,~D.;\ \ Bustamante,~C. \textit{Biophysical Journal}
837: \textbf{2003,} \textsl{84,} 733-738.
838:
839: \bibitem{sinha05}
840: Sinha,~S.;\ \ Samuel,~J. \textit{Physical Review E} \textbf{2005,} \textsl{71,}
841: 021104.
842:
843: \bibitem{nelson05}
844: Nelson,~P.~C.;\ \ Brogioli,~D.;\ \ Zurla,~C.;\ \ Dunlap,~D.~D.;\ \ Finzi,~L.
845: ``Quantitative analysis of tethered particle motion'', 2005 submitted.
846:
847: \bibitem{normnote}
848: Note that for the computation of the force-extension relation, it is not
849: necessary to properly normalize the probability distribution, because we are
850: computing the derivative of the logarithm of $Z$. Our expressions for the
851: probability distribution vectors will neglect the constant normalization
852: factor.
853:
854: \bibitem{jackson}
855: Jackson,~J.~D. \textit{Classical Electrodynamics;} John Wiley and Sons: New
856: York, second ed.; 1975.
857:
858: \bibitem{perkin04}
859: Perkins,~T.~T.;\ \ Li,~H.~W.;\ \ Dalal,~R.~V.;\ \ Gelles,~J.;\ \ Block,~S.~M.
860: \textit{Biophysical Journal} \textbf{2004,} \textsl{86,} 1640-1648.
861:
862: \bibitem{nugen04}
863: Nugent-Glandorf,~L.;\ \ Perkins,~T.~T. \textit{Optics Letters} \textbf{2004,}
864: \textsl{29,} 2611-2613.
865:
866: \end{thebibliography}
867:
868:
869:
870: \end{document}