1:
2:
3:
4: \section{The extended lattice model}
5: \label{sec:lattice}
6:
7:
8: As was discussed above, the elastic rod model is not generally
9: sufficient for describing the conformational dynamics of the DNA.
10: Important, it does not allow for its intrinsic degrees of freedom
11: corresponding to the structure of the double helix. To build an
12: adequate model to the effect, is a difficult problem, and an
13: attempt to manufacture it 'from first principles' is doomed to
14: failure. Thus, there is a need for drastic simplifications, and
15: it is necessary to take into account: (1) the DNA having the two
16: strands; (2) the base-pairs being linked by the hydrogen bonds;
17: (3) the helical symmetry of the DNA. The problem still waits its
18: general solution, but specific cases are nonetheless tractable. In
19: this section we are going to see what could happen if elastic
20: modes that can be expected within the rod model, may interact with
21: internal motions of the double helix.
22:
23: We consider short pieces of the DNA, of several persistence lengths, so that the spatial
24: conformation of the molecule on the mesoscale , is not of primary importance. We focus on the
25: internal dynamics, trying to accommodate the above requirements through a one-dimensional lattice
26: model of the DNA. The key point in this respect is the wise choice of dynamical variables that could
27: give a picture of the DNA dynamics, both simple and adequate. El Hasan and Calladine,
28: \cite{calladine}, give the framework for such analysis by setting up the scheme for the internal
29: geometry of the double helix of the DNA. They describe the relative position of one base with respect
30: to the other in a Watson-Crick base-pair, and also the positions of two base-pairs, by introducing
31: local frames for the bases and the base-pairs, and translation-slides along their long axes.
32:
33: We follow the guidelines of paper \cite{calladine}, but aiming at
34: a qualitative description of the DNA dynamics use a simplified set
35: of variables. We shall describe the relative position of the bases
36: of a base-pair by means of the vector $\vec Y$ directed along the
37: axis of orientation for complimentary bases inside the base pair;
38: $\vec Y$ being equal to zero when the base-pair is at equilibrium.
39: The relative position of the base-pairs is described by the
40: torsional angles $\phi_n$, which give deviations from the standard
41: equilibrium twist of the double helix. Thus a twist of the DNA
42: molecule, which does not involve inter-strand motion or mutual
43: displacements of the bases inside the pairs, is determined by the
44: torsional angles $\phi_n$ that are the angles of rotation of the
45: base-pairs about the axis of the double-helix. The twist energy of
46: the molecule is given by the equation
47: $$
48: \sum_n\, \left[
49: \frac{I}{2} \, \dot{\phi}_n^2
50: + \displaystyle{\frac{\tau}{2a^2}} \,
51: (\phi_{n+1} - \phi_n)^2
52: \right]
53: $$
54: in which $I$ is the moment of inertia, and $\tau$ is the twist
55: coefficient, which for the sake of simplicity and taking into
56: account the qualitative picture at which we aim, are assumed the
57: same for all the base-pairs. Inter-strand motions should
58: correspond to the relative motion of the bases inside the
59: base-pairs, therefore the kinetic energy due to this degree of
60: freedom may be cast in the form
61: $$
62: \sum_n\, \frac{M}{2}\, \dot{\vec Y}_n^2
63: $$
64: where $M$ is the effective mass of a couple.
65:
66: For each base-pair we have the reference frame in which z-axis
67: corresponds to the axis of the double helix, y-axis to the long
68: axis of the base-pair, x-axis perpendicular to z- and y- axes. At
69: equilibrium the change in position of adjacent base-pairs is
70: determined only by the twist angle $\Omega$ of the double helix.
71: We shall assume $\Omega = 2 \pi / 10$. To determine the energy due
72: to the inter-strand displacements we need to find the strain
73: taking into account the constraint imposed by the helical
74: structure of our system. For this end one may utilize the method
75: employed by G.Kirchhoff for the twisted rod, that is the covariant
76: derivative, as was done in paper \cite{kats} for the DNA molecule.
77: But a more simple and straightforward approach is possible.
78:
79: Let us confine ourself only to the torsional degrees of freedom of
80: the double lattice and assume the vectors $\vec Y_n$ being
81: parallel to x-y plane, or two-dimensional.
82: Consider the displacements $\vec Y_n,\, \vec Y_{n+1}$
83: determined within the frames of the two consecutive base-pairs, n, \, n+1.
84: Since we must compare the two vectors in the same frame,
85: we shall rotate the vector
86: $\vec Y_{n+1}$ to the frame of the n-th base pair,
87: $$
88: \vec Y^{\, back}_{n+1} = R^{-1}(\phi)\, \vec Y_{n+1}
89: $$
90: Here $R^{-1}(\phi)$ is the inverse matrix
91: of the rotation of the n-th frame to the (n+1)-one given by the equation
92: \begin{equation}
93: R(\phi) = \left[
94: \begin{array}{ll}
95: \cos \phi & - \sin \phi \\
96: \sin \phi & \cos \phi
97: \end{array}
98: \right] \label{rot}
99: \end{equation}
100: The matrix $R$ is 2 by 2 since the vectors $\vec Y_n$
101: are effectively two-dimensional.
102: Then the strain caused by the displacements of the base-pairs
103: is determined by the difference
104: $$
105: \vec Y^{\, back}_{n+1} - \vec Y_n
106: $$
107: For this argument I am indebted to D.I. Tchertov.
108:
109:
110: It is important that the angle $\phi$ is given by
111: the twist angle, $\Omega$, describing the double helix,
112: in conjunction with the torsional angles $\phi_n$, so that
113: $$
114: \phi = \Omega + \phi_{n+1} - \phi_n
115: $$
116: Therefore, the energy due to the {\it inter-strand} stress reads
117: $$
118: \sum_n \left\{
119: \frac{M}{2}\, \dot{\vec Y_n}^2
120: + \displaystyle{\frac{K}{2a^2}} \,
121: \left[ R^{-1}(\Omega + \phi_{n+1} - \phi_n)\, \vec Y_{n+1}
122: - \vec Y_n
123: \right]^2
124: \right \}
125: $$
126:
127: It corresponds with the fact that the equilibrium position of the
128: double helix is the twisted one determined by $\Omega$ and all
129: $\phi_n$ being equal to zero. We suppose that the size of DNA
130: molecule is small enough that it can be visualized as a straight
131: double helix, that is not larger than the persistence length.
132: Hence the number of base-pairs, $ N \le 150 $, approximately.
133: Combining the formulas given above we may write down the total
134: energy of the DNA molecule in the form, \cite{microwave},
135: \begin{widetext}
136: \begin{eqnarray}
137: {\cal H} &=&
138: \sum_n\, \left[
139: \frac{I}{2} \, \dot{\phi}_n^2
140: + \displaystyle{\frac{\tau}{2a^2}} \,
141: (\phi_{n+1} - \phi_n)^2
142: \right] \nonumber \\
143: &+& \sum_n \left\{
144: \frac{M}{2}\, \dot{\vec Y_n}^2
145: + \displaystyle{\frac{K}{2a^2}} \,
146: \left[ R^{-1}(\Omega + \phi_{n+1} - \phi_n)\, \vec Y_{n+1}
147: - \vec Y_n
148: \right]^2
149: + \frac{\epsilon}{2}\, \vec Y_n^2
150: \right \} \label{main}
151: \end{eqnarray}
152: \end{widetext}
153: in which $K$ and $a$ are the torsional elastic constant and the
154: inter-pairs distance, correspondingly. In summations given above n
155: is the number of a site corresponding to the n-th base-pair, and $
156: n= 1,2, \ldots, N $, $N$ being the number of pairs in the segment
157: of the DNA under consideration. The last term, $\epsilon / 2 \,
158: \vec Y^2$ accommodates the energy of the inter-strand {\it
159: separation} due to the {\it slides of the bases inside the
160: base-pairs}.
161:
162: It should be noted that the dynamical variables $\phi_n$ and
163: $\vec Y_n$ are of the same order of magnitude, that is the first.
164: Consequently, preserving only terms up to the third order, we may
165: transform Eq.(\ref{main}), so that it takes on the form
166: \begin{widetext}
167: \begin{eqnarray}
168: {\cal H} &=&
169: \sum_n\, \left[
170: \frac{I}{2} \, \dot{\phi}_n^2
171: + \displaystyle{\frac{\tau}{2a^2}} \,
172: (\phi_{n+1} - \phi_n)^2
173: \right] \nonumber \\
174: &+& \sum_n \left\{
175: \frac{M}{2}\, \dot{\vec Y_n}^2
176: + \displaystyle{\frac{K}{2a^2}} \,
177: \left[ R^{-1}(\Omega)\, \vec Y_{n+1} - \vec Y_n
178: \right]^2
179: + \frac{\epsilon}{2}\, \vec Y_n^2
180: \right \} \nonumber \\
181: &+& \frac{K}{a^2} \sum_n\, (\phi_{n+1} - \phi_n)\,
182: \left[ R^{-1}(\Omega)\, \vec Y_{n+1} \times \vec Y_n
183: \right]_3 \label{main2}
184: \end{eqnarray}
185: \end{widetext}
186: We have used the fact that the axis of the double-helix is
187: directed along Oz-axis.
188:
189:
190: Let us simplify Eq.(\ref{main2}) by diagonalizing it with
191: the help of the unitary transformation
192: $$
193: \vec Y_n = S\, \vec u_n; \qquad
194: S = \left[
195: \begin{array}{lll}
196: \frac{1}{\sqrt{2}} & \frac{i}{\sqrt{2}} \\
197: \frac{i}{\sqrt{2}} & \frac{1}{\sqrt{2}}
198: \end{array}
199: \right]
200: $$
201: which is a two by two matrix, for the vectors $\vec Y_n$ and
202: $\vec u_n$ are effectively two-dimensional, their third coordinates being
203: equal to zero.
204: The equation for the energy (\ref{main2}) takes on the form
205:
206: \begin{widetext}
207: \begin{eqnarray*}
208: \cal{H} &=&
209: \sum_n \left[
210: \frac{I}{2}\, \dot{\phi}_n^2
211: + \frac{\tau}{2 a^2}\, (\phi_{n+1} - \phi_n)^2
212: \right] \\
213: &+& \sum_n \left[
214: \frac{M}{2}\, \dot{\vec u}_n \cdot \dot{\vec u}^*_n
215: + \frac{\epsilon}{2}\, \vec u_n \cdot \vec u_n^*
216: + \frac{K}{2 a^2}\,
217: \left( \mid e^{i \Omega}\, u^1_{n+1} - u^1_n \mid^2
218: + \mid e^{- i \Omega}\, u^2_{n+1} - u^2_n \mid^2
219: \right)
220: \right] \\
221: &-& \frac{K}{a^2} \, \sum_n \, (\phi_{n+1} - \phi_n) \,
222: \left[ - i e^{i \Omega}\, u^1_{n+1}\, \stackrel*{u}^1_n
223: + i e^{- i \Omega}\, u^2_{n+1}\, \stackrel*{u}^2_n
224: \right]
225: \end{eqnarray*}
226: \end{widetext}
227:
228: The star $*$ signifies complex conjugation.
229:
230: We can further simplify the equation
231: for the energy by applying the Fourier transform
232: given by the equations
233:
234: \begin{widetext}
235: \begin{eqnarray*}
236: f_n &=& \frac{1}{\sqrt{N}}\, \sum_q\, e^{- i n a q}\, f_q \\
237: f_q &=& \frac{1}{\sqrt{N}}\, \sum_{n = -N/2}^{n = + N/2}\,
238: e^{ i n a q}\, f_n \quad
239: q = \frac{2 \pi}{N a}\, m; \quad
240: m = 0, \pm 1, \pm 2, \ldots, \pm \frac{N}{2};
241: \end{eqnarray*}
242: \end{widetext}
243:
244: It is important that after the Fourier transform the variables $\vec u_n$
245: verify the following equations for their complex conjugates
246: \begin{equation}
247: \stackrel*{u}^1_q = i u^2_{-q}, \quad
248: \stackrel*{u}^2_q = i u^1_{-q} \label{cconj}
249: \end{equation}
250: The equation for the energy can be cast in the form
251:
252: \begin{widetext}
253: \begin{eqnarray}
254: \cal{H} &=&
255: \sum_q \left[
256: \frac{I}{2}\, \dot{\phi}_q\, \dot{\phi}_q^*
257: + \frac{\tau}{2 a^2}\, \sin^2\, \frac{a q}{2}\, \phi_q\, \phi_q^*
258: \right] \nonumber \\
259: &+& \sum_q \left[
260: \frac{M}{2}\, \dot{\vec u}_q \cdot \dot{\vec u}^*_q
261: + \frac{\epsilon}{2}\, \vec u_q \cdot \vec u_q^*
262: + \frac{2 K}{a^2}\,
263: \left(
264: \sin^2\, \frac{\Omega - a q}{2} \,
265: u^1_q \stackrel*{u}^1_q
266: + \sin^2\, \frac{\Omega + a q}{2} \,
267: u^2_q \stackrel*{u}^2_q
268: \right)
269: \right] \nonumber \\
270: &+& \frac{K}{a^2} \, \sum_{q' q'' }\, i \frac{e^{- i aq}}{\sqrt{N}}\,
271: \phi_{q'}\,
272: \left[ - e^{i \Omega}\, u^1_{q''}\, \stackrel*{u}^1_{q'+q''}
273: + e^{- i \Omega}\, u^2_{q'}\, \stackrel*{u}^2_{q'+q''}
274: \right] \label{main3}
275: \end{eqnarray}
276: \end{widetext}
277:
278: \noindent
279: in which
280: $$ q_* = \Omega / a$$
281:
282:
283: The above equation serves as well a Hamiltonian that describes the dynamics of a molecule of the DNA,
284: within the framework of the present model, up to terms of the third order. It is worth noting that the
285: latter is based on the assumptions given above concerning the basic structural properties of the DNA,
286: that is it takes into account its two-strand structure, the lattice formed by the base pairs, and the
287: helical symmetry. The specific feature of the Hamiltonian is the presence of the interaction term that
288: describes three-wave interaction, \cite{microwave}, and may result in resonance. We shall utilize the
289: fact for deriving the parametric maintenance of the $u_q$ modes, i.e. the HBS (hydrogen-bond-stretch)
290: modes, (see below).
291:
292:
293: One can obtain, in the usual way, the equations of motion for
294: $u_q^{\alpha}, \, \alpha = 1,2$ and $\phi_q$, from the equation
295: for the energy indicated above. The essential point is the effects
296: of dissipation, which are due to ions in the close neighborhood
297: of the molecule and water effects, see \cite{zandt}. The
298: dissipation could be accommodated by writing down terms linear in
299: $\dot{u}_q^{\alpha},\, \dot{\phi}_q$. We shall take into account
300: external force, or torque ${\cal T}_q$, only in the equation for
301: $\phi_q$, for it corresponds to external degrees of freedom of our
302: model. Thus, the equations of motion can be cast in the form
303:
304: \begin{widetext}
305: \begin{eqnarray}
306: \ddot{u}^{\alpha}_q + \omega^2_{\alpha\,q} u^{\alpha}_q
307: + \gamma_u\, \dot{u}^{\alpha}_q
308: &+&
309: \frac{4 K \sin \Omega }{Ma^2\, \sqrt{N}}
310: \sum_{q'}\, e^{- i a q'}\, \phi_{q'} u^{\alpha}_{q-q'} = 0,
311: \label{motion_f1} \\
312: \ddot{\phi}_q + \omega^2_q\, \phi_q
313: + \gamma_{\phi} \dot{\phi}_q
314: &+&
315: i \frac{4 K \sin \Omega \, e^{ i a q}}{I a^2\, \sqrt{N}}
316: \sum_{q'}\, u^1_{q'} u^2_{q-q'}
317: = {\cal T}_q \label{motion_f}
318: \end{eqnarray}
319: \end{widetext}
320:
321: \noindent
322: Here
323:
324: \begin{widetext}
325: \begin{equation}
326: \omega^2_{\alpha q} = \frac{4 K}{M a^2}\,
327: \sin^2 \frac{\Omega + (-1)^{\alpha} aq}{2}\, +\, \frac{\epsilon}{M},
328: \quad
329: \omega^2_q = \frac{4 \tau}{I a^2}\, \sin^2\frac{aq}{2}
330: \label{freqs}
331: \end{equation}
332: \end{widetext}
333:
334: \noindent are the dispersion laws for the fields $u_q^{\alpha},\,
335: \alpha=1,2$, and $\phi_q$. We see that the spectrum of $\phi_q$
336: has a typical acoustic character, whereas that for $u^{\alpha}_q$
337: has a local minimum determined by the helical twist, $\Omega$.
338: Thus, the spectrum of our model is in qualitative agreement with
339: conclusions of \cite{proh1}. The specific nature of the torque is
340: to be specified elsewhere, for the moment, we shall consider
341: general dynamical phenomena to which the torque may be conducive.
342:
343: Let us suppose that for one thing the amplitudes of the HBS-modes
344: given by $u^{\alpha}_q$ be so small that the quadratic term in
345: Eq.(\ref{motion_f}) can be neglected, and for another the
346: external torque ${\cal T}_q$ be appreciable enough to maintain the
347: vibration of the torsional mode $\phi_q$. Thus, we can visualize
348: the latter as a pump mode that interacts with the HBS-mode
349: $u^{\alpha}_q$ through the non-linearity in
350: Eq.(\ref{motion_f1}). We shall confine ourself to the case of
351: the torque ${\cal T}_q$ being non-zero only at $q = q_*$ and
352: having a frequency $2\omega$. Therefore, the forced wave, or the
353: pump wave for the HBS-mode, has the form
354: \begin{equation}
355: \phi_{q^*} = e^{i 2 \omega t}\, \Phi\, \delta_{qq^*}, \quad
356: \phi_{-q^*} = e^{- i 2 \omega t}\, \Phi^*\, \delta_{-qq^*}
357: \label{ppp}
358: \end{equation}
359: To obtain larger values for the pump wave, $\phi_q$, the resonance
360: condition
361: $$
362: \omega_{q^*} = 2 \omega
363: $$
364: should be verified, even though the resonance behavior of the
365: torsional $\phi_q$-mode itself could be attenuated by
366: dissipation, i.e. it may be a mode of small amplitude.
367:
368:
369: The equations of motion for $u^{\alpha}_q$ in the pumping regime read
370: \begin{widetext}
371: $$
372: \ddot{u}^{\alpha}_q + \omega^2_{\alpha q} u^{\alpha}_q
373: + \gamma_u \dot{u}^{\alpha}_q
374: + \frac{2K}{Ma^2}\frac{\sin \Omega}{\sqrt{N}}
375: \left( A\, e^{i 2 \omega t}\, u^{\alpha}_{q - q*}
376: + A^*\, e^{- i 2 \omega t}\, u^{\alpha}_{q + q*}
377: \right)
378: = 0
379: $$
380: \end{widetext}
381: here
382: $$
383: A = e^{-i \Omega}\, \Phi
384: $$
385: Note that the momentum conservation in the q-values is preserved,
386: as required by the three-wave interaction.
387: The equations indicated above can be cast in the matrix form
388: \begin{equation}
389: \ddot{\vec u}_{\alpha} + \hat{\omega}_{\alpha}^2 \vec u_{\alpha}
390: + \gamma_u \dot{\vec u}_{\alpha} =
391: \left( e^{i 2 \omega t}\, {\cal K} +
392: e^{- i 2 \omega t}\, {\cal K}^{+} \right) \vec u_{\alpha}
393: \label{matr}
394: \end{equation}
395: where ${\cal K}$ and ${\cal K}^{+}$ are hermitian conjugate, and
396: $$
397: {\cal K}^{+}{\cal K} = {\cal I}\, \left( \frac{2 K}{M a^2}
398: \frac{\sin \Omega}{\sqrt{N}}
399: \right)^2\, \mid A \mid^2, \qquad
400: {\cal I}_{ij} = \delta_{ij}
401: $$
402: It is worth noting that Eq.(\ref{matr}) is a kind of matrix
403: Mathieu equation. In fact, we can apply to it Rayleigh's method
404: for studying parametric resonance, \cite{Rayleigh}. For this end
405: let us look for the solution to Eq.(\ref{matr}) in the form
406: of a series
407: $$
408: \vec u(t) =
409: \vec A_1\, e^{i \omega t} + \vec B_1\, e^{- i \omega t} +
410: \vec A_3\, e^{i 3 \omega t} + \vec B_3\, e^{- i 3 \omega t} + \ldots
411: $$
412: On substituting the expression given above into
413: Eq.(\ref{matr}) and preserving only the terms corresponding
414: to $e^{\pm i \omega t}$, we obtain the equations
415: \begin{eqnarray*}
416: \left[ (- \omega^2 + i \gamma_u \omega)\, {\cal I} + \hat{\omega}_{\alpha}^2
417: \right]\, \vec A_1 &+& {\cal K}\, \vec B_1 = 0 \\
418: \left[ (- \omega^2 - i \gamma_u \omega)\, {\cal I} + \hat{\omega}_{\alpha}^2
419: \right]\, \vec B_1 &+& {\cal K}^{+}\, \vec A_1 = 0
420: \end{eqnarray*}
421:
422: The compatibility condition of the equations indicated above can be cast
423: in the form of determinant for the block matrix
424: \begin{equation}
425: det\, \left[
426: \begin{array}{cc}
427: \hat{\omega}_{\alpha}^2 - \omega^2 + i \gamma_u\, \omega
428: & {\cal K} \\
429: {\cal K}^{+}
430: & \hat{\omega}_{\alpha}^2 - \omega^2 - i \gamma_u\, \omega
431: \end{array}
432: \right] = 0
433: \label{rc}
434: \end{equation}
435: Here $\hat{\omega}^2$ is the matrix of frequencies given by
436: Eq.(\ref{freqs}), and $\omega^2$ and $\gamma_u\, \omega$ are
437: the scalar ones. We can transform Eq.(\ref{rc}) into a more
438: amenable form. Let us notice that it is equivalent to the equation
439:
440: \begin{widetext}
441: $$
442: det \, \left \{
443: \left[
444: \begin{array}{cc}
445: \hat{\omega}_{\alpha}^2 - \omega^2 + i \gamma_u \omega
446: & {\cal K} \\
447: {\cal K}^{+}
448: & \hat{\omega}_{\alpha}^2 - \omega^2 - i \gamma_u \omega
449: \end{array}
450: \right]
451: \left[
452: \begin{array}{cc}
453: {\cal I}
454: & - (\hat{\omega}_{\alpha}^2 - \omega^2 + i \gamma_u \omega)^{-1} \\
455: 0
456: & \rho^2 {\cal K}^{+}
457: \end{array}
458: \right]
459: \right \}
460: = 0
461: $$
462: \end{widetext}
463:
464: \noindent
465: in which
466: \begin{equation}
467: \rho = \frac{Ma^2}{2K} \frac{\sqrt{N}}{\sin \Omega} \label{rho}
468: \end{equation}
469: and the matrices ${\cal K}^{+}$ and ${\cal K}$ verify the equation
470: $$
471: - {\cal I} + \rho^2 \, {\cal K} \, {\cal K}^{+} = 0
472: $$
473: We have used the fact that for the range of frequencies we are considering,
474: the matrix
475: $$
476: \hat{\omega}_{\alpha}^2 - \omega^2 + i \gamma_u \, \omega
477: $$
478: is not degenerate. Therefore, the equation given above is equivalent to
479: the following one
480:
481: \begin{widetext}
482: $$
483: det \left[
484: - {\cal I} + \rho^2 \,
485: ( \hat{\omega}_{\alpha}^2 - \omega^2 + i \gamma_u \, \omega )
486: {\cal J}
487: ( \hat{\omega}_{\alpha}^2 - \omega^2 - i \gamma_u \, \omega )
488: {\cal J}^{+}
489: \right] = 0
490: $$
491: \end{widetext}
492:
493: \noindent
494: in which the matrix ${\cal J}$ is given by
495: $$
496: {\cal J}_{qq'} = \delta_{q'\, q - q_*}
497: $$
498: We may cast the last equation into the form
499:
500: \begin{widetext}
501: \begin{equation}
502: (\omega_{\alpha q}^2 - \omega^2 - i \gamma_u \, \omega )
503: (\omega_{\alpha\, q-q_*}^2 - \omega^2 + i \gamma_u \, \omega )
504: - \left( \frac{2 K}{M a^2} \frac{\sin \Omega}{\sqrt{N}}
505: \right)^2 \, |A|^2 =0
506: \label{rc2}
507: \end{equation}
508: \end{widetext}
509:
510: \noindent which is quite similar to the usual condition for
511: parametric resonance. Solutions to Eq.(\ref{rc2}) are
512: generally complex and therefore correspond to attenuated regimes.
513: But there is a specific wave number, $q_{res}$, for which the
514: solution gives the real frequency $\omega$, and it is easy to see
515: that it should satisfy the constraint
516: \begin{equation}
517: \omega^2_{\alpha q-q_*} = \omega^2_{\alpha q}, \qquad \mbox{at }\qquad q = q_{res}
518: \label{resQ}
519: \end{equation}
520: Thus, we may cast the condition for parametric resonance
521: in the familiar form, \cite{Rayleigh},
522: \begin{equation}
523: (\omega^2 - \omega^2_{\alpha q_{res}})^2
524: + \gamma^2\, \omega^2
525: - \left( \frac{2 K}{M a^2} \frac{\sin \Omega}{\sqrt{N}}
526: \right)^2\, |A|^2 = 0
527: \label{pr}
528: \end{equation}
529:
530:
531: The existence of the parametric resonance discussed above presupposes that the bands of the torsional
532: acoustic (TA), the $\phi_q$ modes of the present paper, and the hydrogen-bond-stretch (HBS) modes,
533: that is the $u_q^{\alpha}$ modes, of the DNA interpenetrate each other. So far, the knowledge of the
534: spectra of the DNA vibrations relies mainly on the computer analysis within the framework of quantum
535: chemistry, \cite{proh1}. According to these results, it is accepted that the conformational dynamics
536: of the DNA is confined to elastic vibrations of the DNA molecule in the range of $10^9 \pm 10^{12} \,
537: Hz$, \cite{star}. Kim and Prohofsky, \cite{proh1}, claim that the region comprises two domains, which
538: correspond with different degrees of freedom of the molecule: (1) acoustic modes, which do not involve
539: the hydrogen bonds; (2) modes that stretch the hydrogen bonds between the base-pairs (the HBS modes).
540: Local minimum of the frequency is characteristic of the HBS-modes, \cite{proh1}; its position
541: depending on the choice of the band. The vibrations of the DNA, which are ascribed to the inter-strand
542: modes, were observed in the low-frequency Raman scattering, \cite{urabe}, \cite{urabe2}, and the
543: Fourier-transform infra-red absorption experiments, \cite{powell}. Globus et al, \cite{globus1},
544: report the existence of internal modes generated by the interaction of artificial DNA-type molecules
545: with electromagnetic radiation in sub-millimetre range. It should be noted that the type of modes
546: observed depends on the kind of DNA samples, i.e. in aqueous solutions, or films and filaments,
547: \cite{globus1}. The experimental data, \cite{star}, is not conclusive as to the relative positions of
548: the acoustic and the HBS modes. The inter-strand and the acoustic modes of the DNA are alleged to be
549: overdamped, \cite{proh1}. But the opinion of scientific community in this respect is not unanimous.
550: First of all, they are observed and measured, fairly well, in experiment, \cite{edwards},
551: \cite{urabe}, \cite{urabe2}, \cite{edwards}. Second, according to paper \cite{georghiou} it is the
552: modes related to motion of the sugar-phosphate backbone that are overdamped, that is the $\phi_q$
553: modes discussed above, whereas the inter-strand modes, or the above $u_q^{\alpha}$ are not. Third,
554: Davis and VanZandt, \cite{zandt}, had shown that the dissipative effects suffered by the modes are
555: greatly diminished in case the GHz-frequency range is taken into account; in the region it is
556: necessary to employ the so-called Maxwell model of hydrodynamics, or the Leontovich theory,
557: \cite{fabelinsky}. Thus, we see that there are serious arguments against the inter-strand modes being
558: overdamped and absent.
559:
560:
561:
562: We are in a position to assess the action of mw-radiation on the molecule of the DNA. The question
563: which has aroused controversy during the last fifteen years. We wish to make it clear that there is no
564: "scare-mongering", but merely {\it a suggestion to employ microwave radiation for studying the
565: biophysics of the DNA}. In fact, we feel that the technic of Raman scattering, which had been
566: successfully used for detecting the inter-strand modes, in conjunction with the microwave radiation
567: could be instrumental in studying intrinsic motions of the DNA.
568:
569: The key point of the theoretical analysis of the interaction of
570: electromagnetic radiation with the DNA is accommodating the fact
571: that the wavelength of radiation is by many orders of magnitudes
572: larger than the characteristic size of the region of the molecule
573: involved in the process. It was Chun-Ting Zhang, \cite{zhang}, who
574: suggested a mechanism to overcome this difficulty. The main point
575: of Zhang's argument is that the helical configuration of the
576: electric dipoles corresponding with the base-pairs makes the
577: interaction of the dipole $\vec P$ and the field $\vec E$
578: $$
579: U = -\vec P \cdot \vec E
580: $$
581: dependent on angle.
582: Therefore, different torsional momenta are applied at the base-pairs.
583: The equation for the energy of interaction between the dipoles of DNA
584: and an incident micro-wave reads
585: $$
586: - \sum_n\, \vec E \cdot R(n \Omega + \phi_n) \vec P_{o}
587: $$
588: Here $R(n \Omega + \phi_n)$ is the rotation matrix given by Eq.(\ref{rot}), and $\vec P_o$ is the
589: dipole at site $n=0$. Consequently, even though on the molecular scale the radiation has a plane wave
590: configuration, it still twists the DNA molecule about the axis of the double-helix. Since the momenta
591: changes periodically in time with the incident wave, the irradiation results in a periodic stress that
592: may produce elastic vibrations in the DNA molecule. Zhang suggested that the force may generate
593: resonance vibrations, resulting in a cross-over mechanism which takes up initial torsion excitations
594: and transforms them into longitudinal acoustic vibrations.
595:
596: In the present paper we will try to combine Zhang's mechanism,
597: \cite{zhang}, and the excitations of the double-helix studied by
598: Prohofsky and Kim, \cite{proh1}, with the view of generating
599: inter-strand waves in the DNA by mw-irradiation. In contrast to
600: the original idea by Zhang, we do not utilize a cross-over into
601: longitudinal acoustic vibrations, but employ the interaction
602: between torsional oscillations and the inter-strand ones, i.e. the
603: three-wave, given by Eq.(\ref{main3}).
604:
605: The main point is that
606: by expanding the rotation matrix $R(n \Omega + \phi_n)$ in the angles
607: $\phi_n$ and keeping only the first order terms, we may cast
608: Zhang's interaction in the form
609: \begin{equation}
610: {\cal H}_Z = - \sum_n \, \phi_n \, (\vec E \times \vec P_n )_3
611: \, + \, const, \quad \vec P_n = R(n \Omega) \, \vec P_o
612: \label{Zinter}
613: \end{equation}
614: in which $\vec P_o$ is the dipole vector at site $n = 0$. Next, by
615: using Eq.(\ref{rot}) for the matrix $R(n \Omega)$ and
616: neglecting the constant term we may cast Eq.(\ref{Zinter})
617: in the form
618:
619: \begin{widetext}
620: $$
621: {\cal H}_Z = \frac{1}{2}
622: \sum_n \, \phi_n \, \left \{
623: e^{i n \Omega} \, [ (\vec E \times \vec P_o)_3
624: - i (\vec E \cdot \vec P_o) ] \,
625: + \, e^{- i n \Omega} \, [ (\vec E \times \vec P_o)_3
626: + i (\vec E \cdot \vec P_o) ]
627: \right \}
628: $$
629: \end{widetext}
630:
631: On applying the Fourier transform for the $\phi_n$, and utilizing
632: the equation
633: $$
634: \frac{1}{N} \, \sum_n \, e^{i(\Omega \pm aq) n} =
635: \delta_{\Omega, \pm aq}
636: $$
637: we obtain the following expression for Zhang's interaction
638:
639: \begin{widetext}
640: $$
641: {\cal H}_Z = \frac{N}{2} \left \{
642: \phi_{q = \frac{\Omega}{a}} \, [ (\vec E \times \vec P_o)_3
643: - i (\vec E \cdot \vec P_o) ] \,
644: + \, \phi_{q = - \frac{\Omega}{a}} \, [ (\vec E \times \vec P_o)_3
645: + i (\vec E \cdot \vec P_o) ]
646: \right \}
647: $$
648: \end{widetext}
649:
650: Hence, the torque ${\cal T}_q$ in Eq.(\ref{motion_f})
651: corresponding to ${\cal H_Z}$ is given by the equation
652: \begin{equation}
653: {\cal T} = \frac{{\cal Z}}{I} \, \delta_{q, - q^*} + \,
654: \frac{{\cal Z^*}}{I} \, \delta_{q, q^*} \qquad
655: q_* = \Omega / a
656: \label{torqueZ}
657: \end{equation}
658: in which
659: $$
660: {\cal Z} = \frac{N}{2}\,
661: \left[
662: ( \vec E \times \vec P_o)_3 + i (\vec E \cdot \vec P_o)
663: \right]
664: $$
665: It should be noted that $\pm q_*$ are the local minima of the
666: HBS-modes. From Eq.(\ref{freqs}) we infer that $q$ verifies
667: the constraint given by Eq.(\ref{resQ}) reads
668: \begin{equation}
669: q_{res} = \frac{3}{2}\, q_{*} \label{pump}
670: \end{equation}
671: It is worth noting that the wave numbers $q_{*}$ and $q_{res}$
672: correspond to the wavelengths of one and $\frac{2}{3}$ turns of
673: the double-helix.
674:
675:
676: Equations given above provide an opportunity for making numerical,
677: order of magnitude, estimates, which enable us to assess the
678: effect of mw-radiation on the HBS-modes. From
679: Eq.(\ref{torqueZ}) we infer that the torque ${\cal T}$ has
680: the size
681: $$
682: {\cal T} \propto e^{2 i \omega t}\, E \, P
683: $$
684: where $E$ and $P$ are the external field and the dipole moment of
685: the base-pair, respectfully. Next, suppose that the resonance condition
686: $$
687: \omega_q = 2 \omega, \qquad q = q_* = \frac{\Omega}{a}
688: $$
689: be true, so that the action of the radiation on the torsional
690: modes should be the largest possible. Then the amplitude of the
691: pumping wave, $\phi_{q_*}$, according to
692: Eq.(\ref{motion_f}), is of the order
693: \begin{equation}
694: \Phi \propto \frac{\sqrt{N}}{I}\, \frac{E P}{2 \omega \gamma_{\phi}}
695: \label{Phi}
696: \end{equation}
697: Next, we turn to Rayleigh's condition for the parametric resonance
698: of the HBS-mode given by Eq.(\ref{pr}). For the pumping wave
699: corresponding to Eq.(\ref{Phi}), it gives
700: $$
701: ( \omega^2 \, - \, \omega^2_{\alpha q_*})^2 +
702: \gamma^2_u \, \omega^2 \approx
703: 4 \left ( \frac{K \sin \Omega}{M a^2}\,
704: \frac{E P}{I \gamma_{\phi}} \right )^2
705: $$
706: Hence we have the threshold
707:
708: \begin{equation}
709: \gamma_u \, \gamma_{\phi} \le
710: \frac{2 K \sin \Omega}{M a^2 \omega^2} \, \frac{E P}{I}
711: \label{effect}
712: \end{equation}
713:
714: \noindent which is {\em the condition that the energy supplied to
715: a DNA molecule is greater than that dissipated}, so that the
716: maintaining of the HBS-mode can take place. We suppose that the
717: frequency of the HBS-modes, as given by Eq.(\ref{freqs}), is
718: generally determined by the gap term $\epsilon / M$ in the
719: equation for $u_{\alpha q}$ and the first factor in
720: Eq.(\ref{effect}) does not differ much from unity. It
721: signifies that the energies of the inter-strand separation per
722: base-pair and the twist of the relative positions of the two
723: adjacent base-pairs, should be comparable. At any rate, the
724: hypothesis appears not to contradict the data reproduced in paper
725: \cite{proh1}. If so, we could have the estimate for the
726: dissipative constants, at least by orders of magnitude,
727: \begin{equation}
728: \gamma_u \, \gamma_{\phi} \le \frac{E P}{I}
729: \label{estimateE}
730: \end{equation}
731: On utilizing the relation
732: $$
733: E \propto 2 \, \sqrt{\frac{\pi \, S}{c}}
734: $$
735: which follows from the expression for Pointing's vector
736: $$
737: \vec S = \frac{c}{4 \pi} \vec E \times \vec H,
738: $$
739: in which $c$ is the velocity of light, we cast the estimate given
740: by Eq.(\ref{estimateE}) in the form
741: \begin{equation}
742: \gamma_u \, \gamma_{\phi} \le 2 \, \frac{P}{I} \,
743: \sqrt{\frac{\pi\, S}{c}}
744: \label{estimateW}
745: \end{equation}
746: in which $S$ is the power density of the interaction.
747: If we assume
748: $$
749: P \propto 1 \, Debye \quad \mbox{or} \quad 10^{-18} \, CGS
750: $$
751: and the inertia coefficient $I \propto 10^{-36} \, gr \, cm^2$,
752: corresponding to the mass of the base-pair $\propto 10^{-22} \,
753: gr$, and the size $ \propto 10 \, \AA $, then for the power
754: density $S \propto 100 \, mW/cm^2$, we have
755: $$
756: \gamma_u \, \gamma_{\phi} \le 10^{16} \, Hz^2
757: \quad \mbox{or} \quad
758: \gamma_u , \gamma_{\phi} \le 10^8 \, Hz
759: $$
760: The estimate suggests that the effect produced by mw-radiation is
761: to be looked for at the edge of the GHz zone, for in this case the
762: requirement on the line-width is less stringent. It should be
763: noted that the crucial point in assessing the feasibility of
764: experiments on mw-irradiation of the DNA, and its possible
765: influence, is the part played by ambient solvent and ions
766: contained in it. In fact, the irradiation may result in just
767: heating the solvent, so that the dissipation due to the ions takes
768: up all effects on the molecules of DNA. Generally, the thin
769: boundary layer of water and ions close to the DNA-molecule may
770: have an important bearing on the dynamics initiated by the
771: incident mw-radiation and result in the overdamping of the
772: molecule's torsional oscillations.
773:
774: Davis and VanZandt, \cite{zandt}, put forward arguments that the
775: ions contained in a layer close to the DNA molecule should have an
776: influence small enough to allow the survival of the effect due to
777: mw-irradiation. The part played by the dissipation caused by water
778: is more subtle.
779:
780: The current arguments,\cite{adair}, about the overdamping of the DNA elastic modes, rely on the Stokes
781: law for frictional force, $F= 6 \pi \eta R v$, for a sphere of radius $R$ moving in a fluid of
782: viscosity $\eta$ at speed $v$. It is important that for the specific case of the DNA it should involve
783: the GHz region of frequencies. But, the classical hydrodynamics, that is the Navier-Stokes theory,
784: breaks down in the region, as can be inferred from the phenomenon of light-scattering in liquids,
785: which is characterized by the triplet structure: the central Rayleigh line, $\nu$, due to the elastic
786: scattering, and the Mandelstam-Brillouin doublet, $\nu \pm f$, of the inelastic one; $f$ being the
787: frequency of elastic waves in liquid. The classical hydrodynamics gives the width of a line in the
788: Mandelstam-Brillouin doublet larger than the distance between this line and the maximum of the central
789: line of the triplet, so that the discrete triplet structure should not be observable; in fact, it is,
790: \cite{fabelinsky}. Mandelstam and Leontovich, \cite{fabelinsky}, brought about the solution to this
791: problem by using the relaxational theory of hydrodynamics in which liquid is considered as a viscous
792: elastic medium characterized by a coefficient $\eta$ of viscosity and a shear modulus $G$, the
793: so-called Maxwell model. In fact, the theory also takes into account effects of anisotropy,
794: \cite{fabelinsky}. It predicts that in the region of hypersound, a few GHz or more, the attenuation
795: coefficient for sound waves , $\alpha_{\eta}$, ceases to depend on frequency $\omega$, whereas in the
796: low frequency classical region, in which the Stokes law is valid, the dependence reads $\alpha_{\eta}
797: \sim \omega^2$. Davis and VanZandt, \cite{zandt}, used the approach of the Maxwell theory to find
798: estimates for the damping of the DNA elastic modes, taking into account the first and the second
799: hydration layers, and the quasi-crystallin structure of water in a neighborhood of the DNA. They found
800: the attenuation to be two orders of magnitude smaller than that given by the Stokes law.
801:
802:
803: In fact, there is a further reason for rejecting the approach
804: based on the Stokes law. The water molecules form hydration shells
805: of DNA, \cite{tao}, \cite{tao2}. The primary hydration shell
806: comprises the water molecules immediately adjacent to the DNA,
807: about 20 molecules per nucleotide pair, which constitute a medium
808: different from bulk water. The secondary hydration shell is
809: generally considered to be similar to bulk water. But, at the
810: spatial scale of the diameter of the DNA-molecule, that is several
811: tens $ \AA $, the water in the second hydration shell, is hardly a
812: condensed medium. Indeed, in this case one should have
813: accommodated its local quasi-crystalline structure, described by
814: the icosahedral model, \cite{chaplin}, \cite{mull}, which is to
815: result in sophisticated dynamical equations. The conclusion is
816: that, presently, it is difficult, if possible at all, to construct
817: accurate theoretical estimates for the attenuation of DNA-modes.
818:
819:
820: From the experimental point of view the situation is more
821: advanced. The DNA helical modes were observed in the experiments
822: on the Raman,\cite{urabe}, \cite{urabe2}, \cite{tao}, and the
823: far-infrared, \cite{powell}, scattering. Therefore, one may
824: suggest that the attenuation effects due to viscosity should not
825: preclude elastic modes of the DNA. At the same time small relaxation
826: times for damping between the DNA and the first hydration layer,
827: of order of several tens ps, (see \cite{tao2}) should result in
828: the double helix of DNA concerted motion with surrounding layer of
829: water. The circumstance could be accommodated within the framework
830: of the semi-phenomenological model of the present paper. In fact,
831: the DNA molecule and its first hydration layer still form a helix
832: structure, and the mutual motion of constituent bases of a pair
833: together with hydration water molecules could be described with
834: the field $\vec Y$. Of course, the values of the model's
835: constants, $K, \tau$, should be changed, and for the time being
836: there is lack of information as to their size.
837:
838: It is also worth noting that the effects of dissipation in aqueous
839: solutions, where a certain form of the Stokes law could be
840: possible, and in films, or fibers, should be quite different. So
841: far there has been no comprehensive theoretical analysis of the
842: dissipation, which would allow for comparing the DNA dynamics in
843: solutions and in films . Nonetheless, the interplay of internal
844: vibration modes and sub-millimeter electromagnetic irradiation
845: was registered in paper \cite{globus1}, using Fourier transform
846: spectroscopy and films of the double-stranded homopolymers
847: poly[A]-poly[U] and poly[C]-poly[G] . Employing the concept of
848: normal modes, or oscillators, of macromolecules, developed earlier
849: for proteins, \cite{go}, and used later for DNA, \cite{zakrz},
850: Globus et al, \cite{globus1}, made a numerical simulation of their
851: experimental results, and thus obtained an estimate for the
852: relaxational parameter $\gamma$, which has the meaning of
853: oscillators dissipation. It turned out that in the range of
854: frequencies several $10 cm^{-1}$, the best fit for $\gamma$ is
855: less than $1 cm^{-1}$, depending on the conformation of an
856: external electric field and a sample. This value of $\gamma$ is
857: too large for Eq.(\ref{estimateW}), but the region of
858: frequencies studied in \cite{globus1} is far from the edge of GHz
859: region, so that one may consider the question of acceptable rate
860: of dissipation as still open, and suggest that studying the
861: effects of mw-radiation on the DNA modes may be instrumental for
862: understanding the phenomenon.
863:
864: We see that the elastic dynamics of the double helix could have
865: enough structure for providing a means for stretching the
866: hydrogen bonds of the base-pairs of DNA, or generating the
867: HBS-modes. If the vibrational modes of the DNA are not overdamped
868: by the ambient solvent, and the balance between energies supplied
869: and dissipated is favourable, the maintenance of the HBS-modes
870: could be expected at the edge of the HBS-zone. The best technique
871: for studying the H-bond stretching still remains the Raman
872: spectroscopy on which certain improvements have been made (see
873: \cite{moliveanu} and references therein). Thus, the HBS-modes, and
874: also the breathing modes, are well accessible from the
875: experimental point of view.
876:
877:
878: The choice of specific means for generating torsional excitations
879: of the DNA is important and interesting. In this paper we have
880: envisaged mw-irradiation of the DNA. In case the interpenetration
881: of the acoustic and the HBS-modes takes place, mw-radiation could
882: maintain the HBS-modes, if the power density is sufficiently
883: large, $100 \, mW/cm^2$ or more. It is important that there is no
884: need for long exposures of the sample to the radiation. If the
885: effect be sufficiently pronounced, it may result in the formation
886: of the bubbles of broken H-bonds. At this point it is worth noting
887: that our estimate for the critical power density, $100 \,
888: mW/cm^2$, is by orders of magnitude larger than that officially
889: prescribed, i.e. $0.2 \, - \, 0.1 \, mW/cm^2 $.
890: