cond-mat0608458/let8.tex
1: \documentclass[prl,twocolumn,showpacs,superscriptaddress]{revtex4}
2: \usepackage{graphicx}
3: \usepackage{amssymb}
4: \usepackage{amsmath}
5: \usepackage{amsfonts}
6: \usepackage{mathrsfs}
7: \begin{document}
8: \title{DNA bubble dynamics as a quantum Coulomb problem}
9: \author{Hans C. Fogedby}
10: \email{fogedby@phys.au.dk} \affiliation{Department of Physics and
11: Astronomy, University of Aarhus, DK-8000, Aarhus C, Denmark}
12: \affiliation{Niels Bohr Institute for Astronomy, Physics, and
13: Geophysics, Blegdamsvej 17, DK-2100, Copenhagen {\O}, Denmark}
14: \affiliation{NORDITA, Blegdamsvej 17, DK-2100, Copenhagen {\O},
15: Denmark}
16: \author{Ralf Metzler}
17: \email{metz@nordita.dk}
18: \affiliation{NORDITA, Blegdamsvej 17, DK-2100, Copenhagen {\O}, Denmark}
19: \affiliation{Department of Physics, University of Ottawa, 150 Louis Pasteur,
20: Ottawa, Ontario  K1N 6N5, Canada}
21: %\date{\today}
22: \begin{abstract}
23: We study the dynamics of denaturation bubbles in double-stranded DNA
24: on the basis of the Poland-Scheraga model. We demonstrate that the
25: associated Fokker-Planck equation is equivalent to a Coulomb
26: problem. Below the melting temperature the bubble lifetime is
27: associated with the continuum of scattering states of the repulsive
28: Coulomb potential, at the melting temperature the Coulomb potential
29: vanishes and the underlying first exit dynamics exhibits a long time
30: power law tail, above the melting temperature, corresponding to an
31: attractive Coulomb potential, the long time dynamics is controlled
32: by the lowest bound state. Correlations and finite size effects are
33: discussed.
34: \end{abstract}
35: \pacs{05.40.-a,02.50.-r,87.15.-v,87.10.+e}
36: \maketitle
37: 
38: \emph{Introduction.}
39: The dynamics of bubble formation in double-stranded DNA (dsDNA) is a problem
40: of high current interest in biological and statistical physics. Under
41: physiological conditions the Watson-Crick double helix is the equilibrium
42: structure, its stability effected by hydrogen-bonding of base-pairs and
43: stacking between nearest neighbor pairs of base-pairs
44: \cite{Kornberg74,Watson53}. By variation of temperature or pH-value
45: double-stranded DNA progressively denatures, yielding regions of
46: single-stranded DNA
47: (ssDNA), until the double-strand is fully denatured. This is the
48: helix-coil transition associated with the melting temperature
49: $T_{\text{m}}$ \cite{Poland70}.
50: 
51: Subject to thermal fluctuations dsDNA spontaneously unzips and
52: forms flexible single-stranded DNA bubbles ranging in size from a
53: few to some hundred broken base-pairs, depending on $T$ and salt
54: conditions \cite{Poland70,Gueron87,Krueger06}. Assuming that the
55: bubble-forming dynamics takes place on a slower time scale than
56: the equilibration of the ssDNA strands constituting the bubbles,
57: this DNA-breathing can be interpreted as a random walk in the 1D
58: coordinate $x$, the number of denatured base-pairs.
59: 
60: On the basis of the Poland-Scheraga model for DNA-melting \cite{Poland66}
61: DNA-breathing has been studied in terms of a continuous Fokker-Planck
62: equation \cite{Hwa01,Hanke03}, and of a discrete master equation and
63: stochastic simulation \cite{Banik05,Metzler04}. In the present Letter
64: we show that the Fokker-Planck equation for bubble breathing is equivalent
65: to a quantum Coulomb problem with a repulsive
66: potential above $T_{\text{m}}$ and an attractive potential below
67: $T_{\text{m}}$. This mapping allows us to discuss DNA bubble
68: dynamics in terms of the spectrum of a 'hydrogen-like' system and to
69: derive several exact results such as the exact scaling of the bubble
70: survival behavior and the associated correlations.
71: 
72: \emph{Static and dynamic model.} The Poland-Scheraga free energy
73: for the bubble statistics has the form
74: \cite{Poland70,Hanke03,Krueger06}
75: \begin{eqnarray}
76: \label{free}
77: \mathscr{F}=\gamma_0+\gamma x+c\ln x,
78: \end{eqnarray}
79: where $x\ge 0$ is the bubble size in units of base pairs. We here
80: assume a continuum formulation and imply a cutoff for $x\sim 1$.
81: $\gamma_0$ is the free energy barrier for initial bubble
82: formation, $\gamma x$ is the free energy for the dissociation of
83: $x$ base pairs, and $c\ln x$ the entropy loss factor associated
84: with the formation of a closed polymer ring. The free energy
85: density $\gamma=\gamma_1(1-T/T_{\text{m}})$, where $T_{\text{m}}$
86: is the melting temperature. From experimental data one extracts
87: approximate values for the parameters. In units of $kT_{\text{r}}$
88: with reference temperature $T_{\text{r}}=37^{\circ}$C, we have
89: $\gamma_0\approx10 kT_{\text{r}}$, $\gamma_1\approx 4 kT_{\text{r}}$,
90: and $c\approx 2 kT_{\text{r}}$; the melting temperature for standard
91: salt conditions is in the range $T_{\text{m}}\approx 70-100^{\circ}$C,
92: depending on the relative content of weaker AT and stronger GC
93: Watson-Crick base-pairs.
94: 
95: The stochastic bubble dynamics is governed by the Langevin equation
96: with Gaussian white noise $\xi(t)$,
97: \begin{equation}
98: \label{lan}
99: \frac{dx}{dt}=-D\frac{d\mathscr{F}}{dx}+\xi,
100: \quad\langle\xi\xi\rangle(t)=2DkT\delta(t),
101: \end{equation}
102: where the kinetic coefficient $D$ of dimension $(kT_{\text{r}})^{-1}
103: \mathrm{s}^{-1}$ sets the overall time scale of the dynamics:
104: $[DkT_{\text{r}}]^{-1}\sim\mu$s. With dimensionless parameters
105: $\mu=c/2kT$ and $\epsilon=(\gamma_1/2kT)(T/T_{\text{m}}-1)$, and
106: measuring time in units of $\mu s$, the Fokker-Planck equation corresponding
107: to (\ref{lan}) reads
108: \begin{eqnarray}
109: \frac{\partial P}{\partial t}=\frac{\partial}{\partial
110: x}\left(\frac{\mu}{x}-\epsilon\right)P+
111: \frac{1}{2}\frac{\partial^2P}{\partial x^2}.\label{fokker}
112: \end{eqnarray}
113: Note that close to the physiological temperature $T_{\mathrm{r}}$,
114: $\mu\approx 1$ and $\epsilon\approx2(T/T_{\text{m}}-1)$.
115: 
116: \emph{General results.} Eliminating the first order term by means of the
117: substitution $P=e^{\epsilon x}x^{-\mu}\tilde P$, $\tilde P$
118: satisfies
119: \begin{equation}
120: \label{schroedinger}
121: -\frac{\partial\tilde P}{\partial t}=
122: -\frac{1}{2}\frac{\partial^2\tilde P}{\partial x^2}+
123: \left(\frac{\mu(\mu+1)}{2x^2}-
124: \frac{\mu\epsilon}{x}+\frac{\epsilon^2}{2}\right)\tilde P.
125: \end{equation}
126: This is the imaginary time Schr\"{o}dinger equation for a particle with unit
127: mass in the potential
128: $V(x)=\mu(\mu+1)/2x^2-\mu\epsilon/x+\epsilon^2/2$, i.e., subject
129: to the centrifugal barrier $\mu(\mu+1)/x^2$ for an orbital state
130: with angular momentum $\mu$ and Coulomb potential
131: $-\mu\epsilon/x$. Introducing the Hamiltonian
132: $H=-(1/2)d^2/dx^2+V(x)$ and expanding $\tilde P$ on the normalized
133: eigenstates $\Psi_n$, $H\Psi_n=E_n\Psi_n$, the
134: transition probability $P(x,x_0,t)$ from initial bubble size
135: $x_0$ to a final bubble size $x$ at time $t$ yields in closed form,
136: \begin{eqnarray}
137: P(x,x_0,t)=e^{\epsilon(x-x_0)}\left(\frac{x}{x_0}\right)^{-\mu}
138: \sum_ne^{-E_nt}\Psi_n(x)\Psi_n(x_0). \label{prob}
139: \end{eqnarray}
140: Here the completeness of $\Psi_n$ ensures the initial condition
141: $P(x,x_0,0)=\delta(x-x_0)$. Moreover, in order to account for the
142: absorbing boundary condition for vanishing bubble size we choose
143: $\Psi_n(0)=0$. We also note that for a finite strand of length $L$,
144: i.e., a maximum bubble size of $L$, we have in addition the
145: absorbing condition $\Psi_n(L)=0$ for complete denaturation. Expression
146: (\ref{prob}) is the basis for our discussion of DNA-breathing,
147: relating the dynamics to the spectrum of
148: eigenstates, i.e., the bound and scattering states of the
149: corresponding Coulomb problem \cite{Landau59c}.
150: 
151: The transition probability $P$ is controlled by the Coulomb
152: spectrum. Below the melting temperature $T_{\text{m}}$
153: ($\epsilon\propto(T/T_{\text{m}}-1)<0$), the Coulomb problem is
154: repulsive and the states form a continuum, corresponding to a random
155: walk in bubble size terminating in bubble closure $(x=0)$. At the
156: melting temperature ($\epsilon=0$), the Coulomb potential is absent
157: and the continuum of states is governed by the centrifugal barrier
158: alone, including the limiting case of a regular random walk. Above
159: the melting temperature ($\epsilon>0$), the Coulomb potential is
160: attractive and can trap an infinity of bound states; at long times
161: the lowest bound state dominates the bubble dynamics, corresponding
162: to denaturation of the DNA chain. In Fig.~\ref{fig1} we depict the
163: potential in the two cases $\epsilon \gtrless 0$.
164: 
165: \begin{figure}
166: \includegraphics[width=.64\hsize]{fig1.eps}
167: \caption{Schematic of the potential $V(x)-\epsilon^2/2$ of the Schr{\"o}dinger
168: Eq.~(\ref{schroedinger}). a) $T<T_{\text{m}}$: The potential is repulsive,
169: yielding a continuous spectrum. The bubble fluctuations correspond to
170: a Brownian walk process in bubble size $x$ before collapse at $x=0$.
171: b) $T<T_{\text{m}}$. The potential is attractive and can trap a series of
172: bound states. At long times the lowest bound state indicated in the figure
173: controls the behavior. The bubbles increase in size leading to denaturation.
174: } \label{fig1}
175: \end{figure}
176: 
177: \emph{(i) Long times for $T\le T_{\text{m}}$.} At long times and
178: fixed $x$ and $x_0$, it follows from (\ref{prob}) that the
179: transition probability is controlled by the bottom of the energy
180: spectrum. Below and at $T_{\text{m}}$ the spectrum is continuous
181: with lower bound $\epsilon^2/2$. Setting $E_k=\epsilon^2/2+k^2/2$ in
182: terms of the wave number $k$ and noting that
183: $\Psi_k(x)\propto(kx)^{1+\mu}$ for small $kx$ we have
184: $P\propto\exp(-|\epsilon|(x-x_0)) (x/x_0)^{-\mu}
185: \exp(-\epsilon^2t/2)\int_0^\infty dk\exp(-k^2t/2)(k^2xx_0)^{1+\mu}$,
186: and consequently by a simple scaling argument the long-time
187: expression for the probability distribution
188: \begin{equation}
189: \label{longbelow}
190: P(x,x_0,t)\simeq x x_0^{1+2\mu} e^{-|\epsilon|(x-x_0)}
191: e^{-\epsilon^2t/2}t^{-3/2-\mu}.
192: \end{equation}
193: 
194: The lifetime of a bubble of initial size $x_0$ created at $t=0$ follows
195: from Eq.~(\ref{longbelow}) by calculating the first passage time density
196: (FPTD) as time derivative of the survival probability, $W(t)=-\int_0^\infty
197: dx \partial P/\partial t$ \cite{vanKampen92}, or, via Eq.~(\ref{fokker}),
198: $W(t)=(1/2)[\partial P/\partial x+(2\mu/x-2\epsilon)P]_{x=0}$. This produces
199: \begin{eqnarray}
200: W(t)\simeq x_0^{1+2\mu}e^{|\epsilon|x_0}e^{-\epsilon^2t/2}t^{-3/2-\mu}.
201: \label{absorbbelow}
202: \end{eqnarray}
203: Below $T_{\text{m}}$, $\epsilon<0$ and the FPTD $W(t)$ decays exponentially.
204: The characteristic time scale is set by
205: \begin{eqnarray}
206: \tau=2/\epsilon^2\propto (T_{\text{m}}-T)^{-2},
207: \label{timescale}
208: \end{eqnarray}
209: that diverges as one approaches $T_{\text{m}}$. Finally, from
210: Eq.~(\ref{longbelow}) we infer that $P(x,t)\propto\exp(-c_1|\epsilon|(x+c_2
211: |\epsilon|t))$ with constants $c_i>0$, indicating that the profile of the
212: distribution has a drift $\sim|\epsilon|$ towards bubble closure at $x=0$.
213: 
214: At $T_{\mathrm{m}}$ ($\epsilon=0$) the FPTD falls off like a power law,
215:  $W(t)\propto t^{-\alpha}$ with scaling exponent
216: \begin{equation}
217: \label{exp}
218: \alpha=3/2+\mu.
219: \end{equation}
220: The parameter $\mu=c/2kT$ (with $\mu\approx 1$ at $T\approx T_{\text{r}}$)
221: is associated with the entropy loss of a closed polymer loop. Ignoring the
222: logarithmic entropic effects ($\mu=0$) we obtain $\alpha=3/2$, characteristic
223: of an unbiased random walk \cite{Hanke03}. From (\ref{absorbbelow}) we also
224: conclude that the mean bubble lifetime scales like
225: \begin{equation}
226: \label{mean}
227: \tau_{\text{mean}}\propto x_0/|\epsilon|\propto
228: x_0(T_{\text{m}}-T)^{-1},
229: \end{equation}
230: that diverges as the temperature is raised towards $T_{\text{m}}$.
231: 
232: \emph{(ii) Long times for $T>T_{\text{m}}$.} Above $T_{\text{m}}$ ($\epsilon
233: >0$) the transition probability $P$ is controlled by the lowest bound states
234: in the attractive Coulomb potential. For the discrete spectrum we
235: have $E_n=\epsilon^2/2(1-(\mu/(\mu+n))^2)$, $n=1,2,\ldots$. The
236: lowest state is thus given by $E_1=\epsilon^2(\mu+1/2)/(\mu+1)^2$,
237: and the corresponding nodeless normalized bound state by
238: $\Psi_1(x)=Ax^{1+\mu}\exp{(-\mu\epsilon x/(\mu+1))}$ with
239: normalization constant
240: $A^2=[2\mu\epsilon/(\mu+1)]^{2\mu+3}/\Gamma(2\mu+3)$
241: \cite{Landau59c}. This bound state is localized at $\sim
242: 1/(T-T_{\text{m}})$ and thus recedes to infinity as we approach the
243: melting temperature. In Fig.~\ref{fig2} we depict the lowest bound
244: state $\Psi_1$. From (\ref{prob}) we have $P(x,x_0,t)\sim
245: e^{\epsilon(x-x_0)}(x/x_0)^{-\mu} e^{-E_1t}\Psi_1(x)\Psi_1(x_0)$,
246: and we note that the dominant contribution to the distribution
247: originates from the region where the bound state peaks, i.e., at
248: $\sim 1/(T-T_{\text{m}})$. Inserting in (\ref{prob}) we obtain
249: \begin{eqnarray}
250: \nonumber
251: P(x,x_0,t)&=&A^2 x x_0^{1+2\mu}e^{(\epsilon/(1+\mu))(x-x_0(1+2\mu))}\\
252: &&\times e^{-\epsilon^2(1+2\mu)t/2(1+\mu)^2},
253: \label{longabove}
254: \end{eqnarray}
255: after some reduction. Note from (\ref{longabove}) that the profile of the
256: distribution drifts towards larger bubble sizes with velocity $\sim\epsilon$.
257: The associated FPTD becomes
258: \begin{eqnarray}
259: \nonumber
260: W(t)&=& A^2(1/2+\mu) x_0^{1+2\mu}e^{-\epsilon x_0(1+2\mu)/(1+\mu)}\\
261: &&\times e^{-\epsilon^2t(1+2\mu)/2(1+\mu)^2}.
262: \nonumber
263: \label{absorbabove}
264: \end{eqnarray}
265: 
266: \begin{figure}
267: \includegraphics[width=.64\hsize]{fig2.eps}
268: \caption{Lowest bound state of the attractive well in $V(x)$ for
269: $T=10T_{ \mathrm{r}}$ (full line) and $T=2.2T_{\mathrm{r}}$
270: (dashed); with $T_{\mathrm{m}}=2T_{\mathrm{r}}$. Inset:
271: $\log$-$\log$ plot showing the power-law increase at $x=0$.}
272: \label{fig2}
273: \end{figure}
274: 
275: \emph{Exact result at $T_{\text{m}}$.} At the melting temperature ($\epsilon
276: =0$) the bubble dynamics problem is equivalent to the case of a noisy
277: finite-time singularity studied in Ref. \cite{Fogedby02d}. The eigenstates of
278: $H$ are Bessel functions, $\Psi_k(x)=(kx)^{1/2}J_{1/2+\mu}(kx)$, and we obtain
279: after inserting in (\ref{prob}) the distribution
280: \begin{eqnarray}
281: \nonumber
282: P(x,x_0,t)&=&\frac{x^{1/2-\mu}}{x_0^{-1/2-\mu}} \int_0^\infty dk
283: e^{-k^2t/2}k^2J_{1/2+\mu}(kx)\\
284: &&\hspace*{2.6cm}\times J_{1/2+\mu}(kx_0),
285: \label{exact1}
286: \end{eqnarray}
287: or, by a well-known identity \cite{Lebedev72}, the explicit expression
288: \begin{eqnarray}
289: \nonumber
290: P(x,x_0,t)&=&\left(\frac{x}{x_0}\right)^{-\mu}(xx_0)^{1/2} t^{-1}
291: e^{-(x^2+x_0^2)/2t}\\
292: &&\times I_{1/2+\mu}(xx_0/t),
293: \label{exact2}
294: \end{eqnarray}
295: where $I_\nu$ is the Bessel function of imaginary argument
296: \cite{Lebedev72}. From (\ref{exact2}) we infer the FPTD
297: \begin{eqnarray}
298: \label{fptd}
299: W(t)=\frac{2x_0^{1+2\mu}}{\Gamma(1/2+\mu)}e^{-x_0^2/2t}(2t)^{-3/2-\mu},
300: \label{abs}
301: \end{eqnarray}
302: whose maximum at $t=x_0^2/(3+2\mu)$ assumes the value
303: \begin{eqnarray}
304: W_{\text{max}}=
305: \frac{2}{\Gamma(1/2+\mu)}\left(\frac{2e}{3+2\mu}\right)^{-3/2-\mu}
306: x_0^{-1/2}. \label{max}
307: \end{eqnarray}
308: In Fig.~\ref{fig3} we show the zero-size bubble distribution for two
309: different melting temperatures corresponding to different power-law
310: tails of the FPTD $W(t)$.
311: 
312: \begin{figure}
313: \includegraphics[width=.64\hsize]{fig3.eps}
314: \caption{Bubble lifetime distribution $W(t)$ from Eq.~(\ref{fptd})
315: for $T_m=2T_r$ (full line) and $T_m=10T_r$ (dashed). Inset:
316: $\log$-$\log$ plot of the power-law behavior at long $t$, with
317: slopes -2 and -1.6, as indicated by the straight lines.}
318: \label{fig3}
319: \end{figure}
320: 
321: \emph{Correlations and finite size effects.} In typical experiments
322: measuring the fluorescence correlation of a tagged base-pair, DNA
323: bubble dynamics can be measured on the single molecule level
324: \cite{Altan-Bonnet03}. The correlation function $C(t)$ is
325: proportional to the integrated survival probability $C(t)\propto
326: \int_0^LP(x,x_0,t)dx$~, where $L$ is the chain length \cite{REM}.
327: From the definition of the FPTD we see that indeed $C(t)=1-\int_0^t
328: W(t)dt$. We find three cases:
329: 
330: (i) Below $T_{\text{m}}$ ($\epsilon<0$) we obtain from Eq.~(\ref{longbelow})
331: $C(t)=1-x_0^{1+2\mu}e^{|\epsilon|x_0}\int_0^te^{-\epsilon^2t'/2}(t')^{-3/2-\mu}
332: dt'$, or, in terms of the incomplete Gamma function $\gamma$ \cite{Lebedev72},
333: \begin{equation}
334: C(t)=1-x_0^{1+2\mu}e^{|\epsilon|x_0}(\epsilon^2/2)^{1/2+\mu}
335: \gamma(-1/2-\mu,\epsilon^2t/2).
336: \end{equation}
337: With $\gamma(\alpha,x)=\Gamma(\alpha)-x^{\alpha-1}e^{-x}$ for $x\rightarrow
338: \infty$ we obtain
339: \begin{eqnarray}
340: C(t)=\text{const.}+ x_0^{1+2\mu}\epsilon^{-2}e^{|\epsilon|x_0}t^{-3/2-\mu}
341: e^{-\epsilon^2 t/2}
342: \end{eqnarray}
343: for large $t$.  Note that the basic time scale of the correlations
344: is set by $\epsilon^{-2}\propto (T_{\text{m}}-T)^{-2}$. For
345: $t\ll\epsilon^{-2}$ the correlations show a power law behavior,
346: $C(t)\propto t^{-3/2-\mu}$; at long times $t\gg\epsilon^{-2}$ the
347: correlations fall off exponentially. The size of the time window
348: showing power law behavior increases as $T_m$ is approached. In
349: frequency space the structure function
350: $\tilde{C}(\omega)=\int\exp(i\omega t)C(t)dt$ has a Lorentzian line
351: shape for $|\omega|\ll\epsilon^2$, and power law tails for
352: $|\omega|\gg\epsilon^2$:
353: \begin{equation}
354: \tilde C(\omega)\sim\left\{\begin{array}{ll}
355: x_0^{1+2\mu} e^{|\epsilon|x_0}\left(\omega^2+(\epsilon^2/2)^2\right)^{-1},
356: & \mbox{for }|\omega|\ll \epsilon^2,\\
357: x_0^{1+2\mu} e^{|\epsilon|x_0}|\epsilon|^{-2}|\omega|^{1/2+\mu},
358: & \mbox{for } |\omega|\gg \epsilon^2.\end{array}\right.
359: \end{equation}
360: 
361: (ii) At $T_{\text{m}}$ ($\epsilon=0$) the exact expression for the
362: FPTD (\ref{abs}) combined with relation $C(t)=-\int_t Wdt$ yield
363: \begin{eqnarray}
364: C(t)=1-\frac{\Gamma(1/2+\mu,x_0^2/2t)}{\Gamma(1/2+\mu)}.
365: \end{eqnarray}
366: For short times $t\rightarrow 0$ the behavior
367: \begin{eqnarray}
368: C(t)=1-\frac{(x_0^2/2)^{\mu-1/2}}{\Gamma(1/2+\mu)}
369: t^{1/2-\mu}e^{-x_0^2/2t}
370: \end{eqnarray}
371: obtains, while in the long time limit $t\rightarrow\infty$,
372: \begin{eqnarray}
373: C(t)= \frac{2(x_0^2)^{1/2+\mu}}{(1+2\mu)\Gamma(1/2+\mu)}t^{-1/2-\mu}.
374: \end{eqnarray}
375: 
376: (iii) Above $T_{\text{m}}$ ($\epsilon>0$) the DNA chain eventually fully
377: denatures, and the correlations diverge in the thermodynamic limit. We can,
378: however, at long times estimate the size dependence for a  chain
379: of length $L$. From the general expression (\ref{prob}) we find
380: \begin{eqnarray}
381: C(t)\simeq e^{-\epsilon x_0} x_0^\mu\sum_n e^{-E_nt}\Psi_n(x_0)
382: \int_0^L e^{\epsilon x}x^{-\mu}\Psi_n(x)dx.~
383: \end{eqnarray}
384: With the lowest bound state $\Psi_1(x)=Ax^{1+\mu}\exp{(-\mu\epsilon x/
385: (\mu+1))}$ of energy $E_1=\epsilon^2(\mu+1/2)/(\mu+1)^2$, and after
386: integration over $x$, we obtain
387: \begin{eqnarray}
388: \nonumber
389: &&C(t)\propto A^2e^{-\epsilon x_0(2\mu+1)/(\mu+1)}
390: e^{-\epsilon^2((\mu+1/2)/(\mu+1)^2)t}x_0^{1+2\mu}\\
391: &&\times (1+\mu)\epsilon^{-2}
392: \left[1+(L\epsilon/(1+\mu)-1)e^{\epsilon L/(1+\mu)}]\right].
393: \end{eqnarray}
394: The correlations decay exponentially with the time constant $\sim
395: \epsilon^{-2}(\mu+1)^2/(2\mu+1)$. In frequency space the structure
396: function has a Lorentzian lineshape of width
397: $\sim\epsilon^2(2\mu+1)/(\mu+1)^2$, and for the size dependence one
398: obtains
399: \begin{equation}
400: C(t)\sim\left\{\begin{array}{ll}
401: L e^{\epsilon L/(1+\mu)}, & \mbox{for } \epsilon L/(1+\mu)\gg 1,\\
402: L\epsilon/(1+\mu), & \mbox{for } \epsilon L/(1+\mu)\ll 1
403: \end{array}\right..
404: \end{equation}
405: Note that close to $T_m$ the correlation function $C(t)\propto L$.
406: 
407: \emph{Summary and conclusion.} We demonstrated that the breathing
408: dynamics of thermally induced denaturation bubbles forming
409: spontaneously in double-stranded DNA can be mapped onto the
410: imaginary time Schr{\"o}dinger equation of the quantum Coulomb
411: problem. This mapping allows to calculate the distribution of bubble
412: lifetimes and the associated correlation functions, below, at, and
413: above the melting temperature of the DNA helix-coil transition.
414: Moreover, at the melting transition, the DNA bubble-breathing was
415: revealed to correspond to a one-dimensional finite time singularity.
416: 
417: Our analysis reveals non-trivial scaling of the first passage time density
418: quantifying the survival of a bubble after its original nucleation. The
419: associated critical exponents depend on the parameter $\mu$ stemming from
420: the entropy loss factor of the flexible bubble, and therefore on the ratio
421: $T_{\mathrm{r}}/T$ of reference and actual temperature. This correction
422: through $\mu$ decreases with increasing $T$. FPTD and correlations also
423: depend on the difference $T/T_m-1$, and therefore explicitly on the melting
424: temperature $T_{\mathrm{m}}$ (and thus the relative content of AT or GC
425: base-pairs). We also obtained the critical dependence of the characteristic
426: time scales of bubble survival and correlations on the difference $T-T_m$.
427: The finite size-dependence of the correlation function was recovered, as well.
428: 
429: The mapping of the of DNA-breathing onto the quantum Coulomb problem
430: provides a new way to investigate its physical properties, in
431: particular, in the range above the melting transition, $T>T_m$. The
432: detailed study of the DNA bubble breathing problem is of particular
433: interest as the bubble dynamics provides a test case for new
434: approaches in small scale statistical mechanical systems where the
435: fluctuations of DNA bubbles are accessible on the single molecule
436: level in real time.
437: 
438: Discussions with T. Ambj{\"o}rnsson, S. K. Banik, and A. Svane are
439: gratefully acknowledged. The present work has been supported by the
440: Danish Natural Science Research Council, the Natural Sciences and
441: Engineering Research Council (NSERC) of Canada, and the Canada
442: Research Chairs program.
443: 
444: %\bibliography{c:/user/manus/bib/bioarticles,c:/user/manus/bib/articles,c:/user/manus/bib/books}
445: \begin{thebibliography}{14}
446: \expandafter\ifx\csname
447: natexlab\endcsname\relax\def\natexlab#1{#1}\fi
448: \expandafter\ifx\csname bibnamefont\endcsname\relax
449:   \def\bibnamefont#1{#1}\fi
450: \expandafter\ifx\csname bibfnamefont\endcsname\relax
451:   \def\bibfnamefont#1{#1}\fi
452: \expandafter\ifx\csname citenamefont\endcsname\relax
453:   \def\citenamefont#1{#1}\fi
454: \expandafter\ifx\csname url\endcsname\relax
455:   \def\url#1{\texttt{#1}}\fi
456: \expandafter\ifx\csname
457: urlprefix\endcsname\relax\def\urlprefix{URL }\fi
458: \providecommand{\bibinfo}[2]{#2}
459: \providecommand{\eprint}[2][]{\url{#2}}
460: 
461: \bibitem[{\citenamefont{Kornberg}(1974)}]{Kornberg74}
462: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Kornberg}},
463:   \emph{\bibinfo{title}{DNA Synthesis}} (\bibinfo{publisher}{W. H. Freeman},
464:   \bibinfo{address}{San Francisco}, \bibinfo{year}{1974}).
465: 
466: \bibitem[{\citenamefont{Watson and Crick}(1953)}]{Watson53}
467: \bibinfo{author}{\bibfnamefont{J.~D.} \bibnamefont{Watson}} \bibnamefont{and}
468:   \bibinfo{author}{\bibfnamefont{F.~H.~C.} \bibnamefont{Crick}},
469:   \bibinfo{journal}{Cold Spring Harbor Symp. Quant. Biol.}
470:   \textbf{\bibinfo{volume}{18}}, \bibinfo{pages}{123} (\bibinfo{year}{1953}).
471: 
472: \bibitem[{\citenamefont{Poland and Scheraga}(1970)}]{Poland70}
473: \bibinfo{author}{\bibfnamefont{D.}~\bibnamefont{Poland}} \bibnamefont{and}
474:   \bibinfo{author}{\bibfnamefont{H.~A.} \bibnamefont{Scheraga}},
475:   \emph{\bibinfo{title}{Theory of helix-coil transitions in biopolymers}}
476:   (\bibinfo{publisher}{Academic Press}, \bibinfo{address}{Mew York},
477:   \bibinfo{year}{1970}).
478: 
479: \bibitem[{\citenamefont{Gu\'eron et~al.}(1987)\citenamefont{Gu\'eron, Kochoyan,
480:   and Leroy}}]{Gueron87}
481: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Gu\'eron}},
482:   \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Kochoyan}}, \bibnamefont{and}
483:   \bibinfo{author}{\bibfnamefont{J.~L.} \bibnamefont{Leroy}},
484:   \bibinfo{journal}{Nature} \textbf{\bibinfo{volume}{328}}, \bibinfo{pages}{89}
485:   (\bibinfo{year}{1987}).
486: 
487: \bibitem[{\citenamefont{Krueger et al.}(2006)\citenamefont{Krueger, Protozanova,
488:   and Frank-Kamenetskii}}]{Krueger06}
489: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Krueger}},
490:   \bibinfo{author}{\bibfnamefont{E..}~\bibnamefont{Protozanova}},
491:   \bibnamefont{and} \bibinfo{author}{\bibfnamefont{M.~D.}
492:   \bibnamefont{Frank-Kamenetskii}},
493:   \bibinfo{journal}{Biophys. J.} \textbf{\bibinfo{volume}{90}},
494:   \bibinfo{pages}{3091} (\bibinfo{year}{2006}).
495: 
496: \bibitem[{\citenamefont{Hwa et al.}(2003)}]{Hwa01}
497: \bibinfo{author}{\bibfnamefont{T.}~\bibnamefont{Hwa}} \bibnamefont{et al.},
498:   \bibinfo{journal}{Proc. Natl. Acad. Sci. USA} \textbf{\bibinfo{volume}{100}},
499:   \bibinfo{pages}{4411} (\bibinfo{year}{2003}).
500: 
501: \bibitem[{\citenamefont{Hanke and Metzler}(2003)}]{Hanke03}
502: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Hanke}} \bibnamefont{and}
503:   \bibinfo{author}{\bibfnamefont{R.}~\bibnamefont{Metzler}},
504:   \bibinfo{journal}{J. Phys. A} \textbf{\bibinfo{volume}{36}},
505:   \bibinfo{pages}{L473} (\bibinfo{year}{2003}).
506: 
507: \bibitem[{\citenamefont{Banik et~al.}(2005)\citenamefont{Banik, Ambj\"ornsson,
508:   and Metzler}}]{Banik05}
509: \bibinfo{author}{\bibfnamefont{S.~K.} \bibnamefont{Banik}},
510:   \bibinfo{author}{\bibfnamefont{T.}~\bibnamefont{Ambj\"ornsson}},
511:   \bibnamefont{and} \bibinfo{author}{\bibfnamefont{R.}~\bibnamefont{Metzler}},
512:   \bibinfo{journal}{Europhys. Lett} \textbf{\bibinfo{volume}{71}},
513:   \bibinfo{pages}{852} (\bibinfo{year}{2005}).
514: 
515: \bibitem[{\citenamefont{Ambj\"ornsson and Metzler}(2004)}]{Metzler04}
516:   \bibinfo{author}{\bibfnamefont{T.}~\bibnamefont{Ambj\"ornsson}}
517:   \bibnamefont{and} \bibinfo{author}{\bibfnamefont{R.}~\bibnamefont{Metzler}},
518:   \bibinfo{journal}{Phys. Rev. E} \textbf{\bibinfo{volume}{72}},
519:   \bibinfo{pages}{030901(R)} (\bibinfo{year}{2005});
520:   \bibinfo{journal}{J. Phys. Cond. Mat.} \textbf{\bibinfo{volume}{17}},
521:   \bibinfo{pages}{S1841} (\bibinfo{year}{2005});
522: 
523: \bibitem[{\citenamefont{Poland and Scheraga}(1966)}]{Poland66}
524: \bibinfo{author}{\bibfnamefont{D.}~\bibnamefont{Poland}} \bibnamefont{and}
525:   \bibinfo{author}{\bibfnamefont{H.~A.} \bibnamefont{Scheraga}},
526:   \bibinfo{journal}{J. Chem. Phys.} \textbf{\bibinfo{volume}{45}},
527:   \bibinfo{pages}{1456} (\bibinfo{year}{1966}).
528: 
529: \bibitem[{\citenamefont{Landau and Lifshitz}(1959)}]{Landau59c}
530: \bibinfo{author}{\bibfnamefont{L.}~\bibnamefont{Landau}} \bibnamefont{and}
531:   \bibinfo{author}{\bibfnamefont{E.}~\bibnamefont{Lifshitz}},
532:   \emph{\bibinfo{title}{Quantum Mechanics}} (\bibinfo{publisher}{Pergamon
533:   Press}, \bibinfo{address}{Oxford}, \bibinfo{year}{1959}).
534: 
535: \bibitem[{\citenamefont{van Kampen}(1992)}]{vanKampen92}
536: \bibinfo{author}{\bibfnamefont{N.~G.} \bibnamefont{van Kampen}},
537:   \emph{\bibinfo{title}{Stochastic Processes in Physics and Chemistry}}
538:   (\bibinfo{publisher}{North-Holland}, \bibinfo{address}{Amsterdam},
539:   \bibinfo{year}{1992}).
540: 
541: \bibitem[{\citenamefont{Fogedby and Poutkaradze}(2002)}]{Fogedby02d}
542: \bibinfo{author}{\bibfnamefont{H.~C.} \bibnamefont{Fogedby}} \bibnamefont{and}
543:   \bibinfo{author}{\bibfnamefont{V.}~\bibnamefont{Poutkaradze}},
544:   \bibinfo{journal}{Phys. Rev. E} \textbf{\bibinfo{volume}{66}},
545:   \bibinfo{pages}{021103} (\bibinfo{year}{2002}).
546: 
547: \bibitem[{\citenamefont{Lebedev}(1972)}]{Lebedev72}
548: \bibinfo{author}{\bibfnamefont{N.~N.} \bibnamefont{Lebedev}},
549:   \emph{\bibinfo{title}{Special functions and their applications}}
550:   (\bibinfo{publisher}{Dover Publications}, \bibinfo{address}{New York},
551:   \bibinfo{year}{1972}).
552: 
553: \bibitem[{\citenamefont{Altan-Bonnet et~al.}(2003)\citenamefont{Altan-Bonnet,
554:   Libchaber, and Krichevsky}}]{Altan-Bonnet03}
555: \bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Altan-Bonnet}},
556:   \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Libchaber}},
557:   \bibnamefont{and}
558:   \bibinfo{author}{\bibfnamefont{O.}~\bibnamefont{Krichevsky}},
559:   \bibinfo{journal}{Phys. Rev. Lett.} \textbf{\bibinfo{volume}{90}},
560:   \bibinfo{pages}{138101} (\bibinfo{year}{2003}).
561: 
562: \bibitem[{\citenamefont{Amj{\"o}rnsson and Metzler}(2006)
563:   \citenamefont{Amj{\"o}rnsson and Metzler}}]{REM}
564: \bibinfo{author}{\bibfnamefont{T.}~\bibnamefont{Amj{\"o}rnsson}}
565:   \bibnamefont{and}
566:   \bibinfo{author}{\bibfnamefont{R.}~\bibnamefont{Metzler}}
567:   \bibinfo{journal}{(unpublished)}.
568:   %\textbf{\bibinfo{volume}{90}},
569:   %\bibinfo{pages}{138101} (\bibinfo{year}{2003}).
570: 
571: \end{thebibliography}
572: \end{document}
573: