Sei sulla pagina 1di 9

R

E
S
E
A
R
C
H
A
R
T
I
C
L
E
Copyright 2005 American Scientic Publishers
All rights reserved
Printed in the United States of America
Journal of
Computational and Theoretical Nanoscience
Vol. 2, 19, 2005
Peyrard-Bishop-Dauxois Model of
DNA Dynamics and Impact of Viscosity
S. Zdravkovi c,
1
J. A. Tuszy nski,
2
and M. V. Satari c
3
1
Faculty of Technical Sciences, University of Pritina, Kosovska Mitrovica, Yugoslavia
2
Department of Physics, University of Alberta, T6G 2J1, Canada
3
Faculty of Technical Sciences, 21000 Novi Sad, Yugoslavia
In this paper we try to elucidate the physical signicance of the analytical solutions in the Peyrard-
Bishop-Dauxois (extended Peyrard-Bishop) model of DNA dynamics. We discuss the impact of
some parameters of the model, especially the harmonic constant of the helicoidal springs. We study
both the case of DNA dynamics when viscosity is ignored and when it is taken into consideration.
Keywords: DNA, PBD Model, Solitons, Viscosity.
1. INTRODUCTION
Deoxyribonucleic acid (DNA) is doubtlessly one of the
most important biomolecules. Its double standard heli-
cal structure undergoes a very complex dynamics and the
knowledge of that dynamics provides insights into var-
ious related biological phenomena such as transcription,
translation and mutation. The key problem in DNA bio-
physics is how to relate functional properties of DNA with
its structural and physical dynamical characteristics. The
possibility that nonlinear effects might focus the vibration
energy of DNA into localized soliton-like excitations was
rst contemplated by Englander et al.
1
Although several
authors
29
have suggested that either topological kink soli-
tons or bell-shaped breathers would be good candidates to
play a basic role in DNA nonlinear dynamics, there are
still several unresolved questions in this regard. A hier-
archy of the most important models for nonlinear DNA
dynamics was presented by Yakushevich.
10
The local openings can be analytically described as
breather-like objects of small amplitude, which have nev-
ertheless interesting properties; as long as their amplitude
is small enough they can move along the chain. This is
unidimensional, which allows local openings of the hydro-
gen bonds and formation of denaturation bubbles.
In the present paper we deal with the extended model for
DNA dynamics, rst proposed by Peyrard and Bishop
7
and

Author to whom correspondence should be addressed.


further developed by Dauxois.
11, 12
This Peyrard-Bishop-
Dauxois model will be henceforth referred to as the PBD-
model for short.
The paper is organized in the following way. In Sec-
tion 2 we outline the PBD model primarily intended to
describe the process of local opening of base pairs (or
local melting of the double helix). Then we attempt to shed
new light on the still somewhat vague parameter values of
the PBD model, especially discussing the harmonic con-
stant of helicoidal springs. This section represents a certain
extension of two of our previous papers.
13, 14
However, we
introduce a new procedure which yields qualitatively new
results and solves a couple of problems.
In Section 3 we present our consideration of DNA
breather dynamics in the context of the PBD model when
viscosity is taken into consideration. Section 4 puts for-
ward our discussion and concluding remarks.
2. THE PBD MODEL OF DNA
The B-form DNA in the Watson-Crick model is a double
helix, which consists of two strands s
1
and s
2
(Fig. 1),
linked by the nearest-neighbour harmonic interactions
along the chains. The strands are coupled to each other
through hydrogen bonds, which are supposed to be respon-
sible for transversal displacements of nucleotides.
It was argued
7, 11, 12
that we could safely assume a com-
mon mass m for all the nucleotides and the same coupling
constant k along each strand. We will get back to this
J. Comput. Theor. Nanosci. 2005, Vol. 2, No. 2 1546-198X/2005/2/001/009/$17.00+.25 doi:10.1166/jctn.2005.110 1
R
E
S
E
A
R
C
H
A
R
T
I
C
L
E
Peyrard-Bishop-Dauxois Model of DNA Dynamics and Impact of Viscosity Zdravkovi c et al.
s
1
s
2
k
k
m
m
v
n1
v
n+1
v
n
u
n1
u
n+1
u
n
Morse
Potential
Fig. 1. Graphical representation of the simple model for DNA strands.
point later. As was pointed out, we rely on the PBD model,
which takes into account the fact that the DNA molecule
is twisted. This helicoidal structure of the DNA chain
implies that nucleotides from different strands become
close enough so that they interact through water laments.
This means that a nucleotide at the site n of one strand
interacts with both (n+h)th and (nh)th nucleotides of
the other strand. In Refs. [13, 14] we used h = 4
11, 12
for all calculations. However, h = 5 might be better.
15, 16
This comes from the fact that there are approximately ten
nucleotides per one turn. Introducing the transversal dis-
placements u
n
, v
n
of the nucleotides from their equilibrium
positions along the direction of the hydrogen bonds, the
Hamiltonian for the DNA chain becomes
11, 12
H =
_
_
m
2
( u
2
n
+ v
2
n
) +
k
2
_
(u
n
u
n1
)
2
+(v
n
v
n1
)
2
_
+
K
2
_
(u
n
v
n+h
)
2
+(u
n
v
nh
)
2
_
+D|e
o(u
n
v
n
)
1]
2
_
(1)
Here k (respectively K) is the harmonic constant of the
longitudinal (respectively helicoidal) spring. The last term
in the Hamiltonian represents a Morse potential approxi-
mating the potential of the hydrogen bonds while D and o
are the depth and the inverse width of the Morse potential
well, respectively.
It is more convenient to describe the motion of two
strands by making a transformation to the centre-of-mass
coordinates representing the in-phase and out-of-phase
transversal motions, namely
x
n
=(u
n
+v
n
),

2, ,
n
=(u
n
v
n
),

2 (2)
From relations (1) and (2) one can obtain the follow-
ing dynamical equations describing linear waves (phonons)
and nonlinear waves (breathers):
m x
n
=k(x
n+1
+x
n1
2x
n
) +K(x
n+h
+x
nh
2x
n
) (3)
and
m ,
n
= k(,
n+1
+,
n1
2,
n
) K(,
n+h
+,
nh
+2,
n
)
+2

2oD
_
e
o

2,
n
1
_
e
o

2,
n
(4)
As was explained in a couple of articles
1113, 17
, we can
apply the transformation
, =a4; (a 1) (5)
This means that we assume that oscillations of nucleo-
tides are large enough to be anharmonic but still small
enough so that the particles oscillate around the bottom of
the Morse potential. Equations (4) and (5), and the expan-
sion of exponential terms in Eq. (4), yield

4
n
=
k
m
(4
n+1
+4
n1
24
n
)
K
m
(4
n+h
+4
nh
+24
n
)
w
2
g
(4
n
+ao4
2
n
+a
2
p4
3
n
) (6)
where
w
2
g
=
4o
2
D
m
, o =
3o

2
and p =
7o
2
3
(7)
To solve Eq. (6) we use the semi-discrete approxi-
mation.
17
This means that we look for wave solutions of
the form
4
n
(t) = E
1
(anI, at)e
i0
n
+a|E
0
(anI, at) +E
2
(anI, at)e
i20
n
]
+cc +O(a
2
) (8)
and
0 0
n
=nqI wt (9)
Here, I is the distance between two neighbouring nucleo-
tides in the same strand, w is the optical frequency of
the linear approximation of their vibrations, q is the
wave number whose role will be discussed later, cc are
conjugate-complex terms and the function E
0
is real.
Before we proceed with solving Eq. (6) we give some
explanations for Eq. (8). If there were not the last term in
Eq. (6), the one with w
2
g
, which comes from the nonlinear
term in Eq. (4), we would expect the solution in the form
E
1
e
i0
n
+cc instead of Eq. (8). This would be a modulated
wave with a carrier component e
i0
n
and an envelope E
1
. We
will see later that the modulation factor E
1
will be treated
in a continuum limit while the carrier wave will not. In
other words, the carrier component of the modulated wave
includes the discreteness and the procedure is called semi-
discrete approximation. As if there are terms with 4
2
n
and
4
3
n
in Eq. (6) we can not expect solution of this equation in
the simple form E
1
e
i0
n
and nonexponential term as well as
terms with e
i20
n
should be incorporated into the expression
for the solution. A whole method is explained in much
more details in Ref. [18], while its mathematical basis is
given in Refs. [19, 20].
Now, we are ready for solving Eq. (6). It was already
pointed out that the functions E
i
would be treated in the
continuum limit. So, taking this limit (nI z) and apply-
ing the transformations
Z =az; T =at (10)
2 J. Comput. Theor. Nanosci. 2, 19, 2005
R
E
S
E
A
R
C
H
A
R
T
I
C
L
E
Zdravkovi c et al. Peyrard-Bishop-Dauxois Model of DNA Dynamics and Impact of Viscosity
yields the following continuum approximation
E (a(nh)I, at) E (Z, T ) E
Z
(Z, T )aIh
+
1
2
E
ZZ
(Z, T )a
2
I
2
h
2
(11)
where E
Z
and E
ZZ
mean corresponding derivatives with
respect to Z. This allows us to obtain a new expression
for the function 4
n
(t):
4
n
(t) E
1
(Z, T )e
i0
+a|E
0
(Z, T ) +E
2
(Z, T )e
i20
] +cc
=E
1
e
i0
+a|E
0
+E
2
e
i20
] +E

1
e
i0
+aE

2
e
i20
(12)
where stands for complex conjugate and E
i
E
i
(Z, T ).
From Eqs. (9)(12) one can straightforwardly obtain
expressions for all the terms in Eq. (6).
18
From the con-
tinuum version of Eq. (6), equating the coefcients for
the various harmonics, we can get a set of important
relations.
1113
For example, equating the coefcients for e
i0
one obtains a dispersion relation
w
2
=w
2
g
+
2k
m
(1cos(qI)) +
2K
m
(cos(qhI) +1) (13)
This can be used to nd the corresponding group velocity
Jw,Jq as
V
g
=
I
mw
|k sin(qI) Khsin(qhI)] (14)
In the same way, equating the coefcients for e
i0
= 1,
we can easily obtain
E
0
=jE
1

2
(15)
where
j =2o
_
1+
4K
mw
2
g
_
1
(16)
However, for the next three harmonics (e
i20
, e
i30
and
e
i40
) the matter becomes more complicated. All of these
harmonics give a relation
E
2
=oE
2
1
(17)
but with different values for the parameter o. For example,
equating the coefcients for e
i20
one can get
_
4w
2
+
2k
m
(cos(2qI) 1)

2K
m
(cos(2hqI) +1) w
2
g
_
E
2
=w
2
g
oE
2
1
(18)
which means that o is not constant but a function of qI.
However, coefcients for e
i30
and e
i40
give constant values
o,
p
2o
and
3p
o
respectively. We will return to this point
later.
As if we can express functions E
0
and E
2
through the
function E
1
we can derive an equation for E
1
. Using new
coordinates again:
S =ZV
g
T , t =aT (19)
we can nally obtain the nonlinear Schrdinger equation
(NSE) for the function E
1114
1
iE
1t
+PE
1SS
+QE
1

2
E
1
=0 (20)
where the dispersion coefcient P and the coefcient of
nonlinearity Q are given by
P =
1
2w
_
I
2
m
|k cos(qI) Kh
2
cos(qhI)] V
2
g
_
(21)
and
Q =
w
2
g
2w
|2o(j+o) +3p] (22)
Those derivations can be found in some more details in
Ref. [18].
For PQ > 0 the solution of the NSE (20) is
11, 12, 21
E
1
(S, t) =A sech
_
S u
e
t
L
e
_
exp
iu
e
(S u
c
t)
2P
(23)
where the envelope amplitude A and its width L
e
have the
forms
A =
_
u
2
e
2u
e
u
c
2PQ
(24)
and
L
e
=
2P
_
u
2
e
2u
e
u
c
(25)
and u
e
and u
c
being the velocities of the envelope and the
carrier waves, respectively. Using Eqs. (23), (8), (9), (15),
(17), (10), and (19) we obtain
1114
4
n
(t) = 2Asech
_
a
L
e
(nI V
e
t)
_

_
cos(OnI Dt)+aAsech
_
a
L
e
(nI V
e
t)
_

_
j
2
+ocos|2(OnIDt)]
__
+O(a
2
) (26)
where
V
e
=V
g
+au
e
(27)
O =q +
au
e
2P
(28)
and
D=w+
au
e
2P
(V
g
+au
c
) (29)
In Ref. [13] we suggested discrete values for a wave
number q. For h =4 we obtained the following four values
for the wave length \: 6I, 7I, 8I, and 9I or, as if q =2r,\,
J. Comput. Theor. Nanosci. 2, 19, 2005 3
R
E
S
E
A
R
C
H
A
R
T
I
C
L
E
Peyrard-Bishop-Dauxois Model of DNA Dynamics and Impact of Viscosity Zdravkovi c et al.
the following values for qI: 1.05 rad, 0.90 rad, 0.78 rad =
r,4 and 0.70 rad. We gave an argument to reject the rst
value (\ =6I). For h =5, using the same procedure,
13
one
obtains the following values for \: 7I, 8I, 9I, 10I, and 11I
and the following values for qI: 0.90 rad, 0.78 rad, 0.70 rad,
0.63 rad =r,5 and 0.57 rad. For the most favourable mode
we suggested the one for which qI = r,4 for h = 4. For
h =5 that would be qI =r,5.
For all the calculations we chose the following set of
values characterizing the DNA molecule:
k =3K =24 N/m (30a)
I =3.4 10
10
m (30b)
m=5.1 10
25
kg (30c)
o =2 10
10
m
1
(30d)
D =0.1 eV (30e)
The value for the mass is an average value for all four
nucleotides. Namely, molecular masses of nucleotides are
from 340 g/mol (Cytosine) to 380 g/mol (Guanine). There-
fore, for the average value, after subtraction the masses of
three water molecules, one obtains 305.75 g/mol, which
gives the above value for the mass of the single nucleotide.
All other constants were taken from the papers.
11, 12
It was pointed out that there were two possibilities for
the parameter o, constant and nonconstant. In Ref. [13] we
accepted the constant o coming from e
i30
. Therefore, we
did not use the lowest possible harmonic e
i20
! In Ref. [14],
however, we made a choice having more physical sence.
This is nonconstant o given by Eqs. (17) and (18). In that
article we suggested that the value of the parameter K
was overestimated. It was shown that there should be K
4.6 N/m for qI = 0.78 rad and h = 4 and all the other
constants given by (30). For h =5 and qI =0.63 rad one
obtains K 4.4 N/m.
Of course, one can argue that those values for qI,
0.78 rad and 0.63 rad, were obtained for K = 8 N/m in
Ref. [13]. This is correct. However, using the same proce-
dure for the possible qI values, explained in Ref. [13], we
obtain, for h =4, \ =6I, 7I, 8I, and 9I, i.e., qI =1.05 rad,
0.9 rad, 0.78 rad and 0.7 rad for both K =8 N/m and K =
6 N/m. The value K = 4 N/m gives the same four values
for qI and two more (\=10I and 11I, or qI =0.63 rad and
qI =0.57 rad) in addition. For K =3 N/m we obtain \ =
6I, 7I, . . . , 13I while K =2 N/m gives \ =6I, 7I, . . . , 19I.
Hence, the value qI = 0.78 rad, as well as qI = 0.63 rad
for h = 5, corresponds to both large and small values of
the parameter K. This means that qI = 0.78 rad can be
suitable choice for all acceptable values of K, but does not
mean that this is the optional value for our calculations.
Thus, we need one more criterion for qI to be determined
and we show in the following that this exists.
The breather-type solution in Eq. (26) represents a sort
of a modulated solitonic wave. From hyperbolic and cosine
terms in Eq. (26) we can recognize wave numbers of both
the envelope and the carrier wave. In other words, we can
see that the length of the envelope, A, and the wavelength
of the carrier wave, \
c
, are
A=
2rL
e
a
(31)
and
\
c
=
2r
q +
au
e
2P
(32)
This allows us to calculate the number of wavelengths
of the carrier wave contained within the length of the enve-
lope as
D
o

A
\
c
(33)
In what follows, we call D
o
a density of internal oscilla-
tions (density of carrier wave oscillations) for short even
though, strictly speaking, this is not density, i.e., the num-
ber of wavelengths per unit length. From Eqs. (31)(33)
and (25) we can easily obtain
D
o
=
u
e
_
u
2
e
2u
e
u
c
_
1+
2qIP
aIu
e
_
(34)
We should keep in mind that the dispersion coefcient P
is dened by Eq. (21) and is a function of qI.
If we choose the following set of values for parameters
characterizing a traveling wave solution
11, 12
u
e
=10
5
m/s (35a)
u
c
=0 (35b)
a =0.007 (35c)
we can plot the function D
o
vs. qI and this is done in
Figure 2 for h = 5 and three values for the parameter K.
From this gure we see that the density of the inter-
nal oscillations of the soliton D
o
reaches a maximum for
10
8
6
4
0.4 0.5 0.6 0.7 0.8 0.9
ql (rad)
a
b
c
D
0
Fig. 2. Density of the carrier wave oscillations D
o
as a function of qI.
(a: K =4.5 N/m, b: K =4 N/m, c: K =3.5 N/m).
4 J. Comput. Theor. Nanosci. 2, 19, 2005
R
E
S
E
A
R
C
H
A
R
T
I
C
L
E
Zdravkovi c et al. Peyrard-Bishop-Dauxois Model of DNA Dynamics and Impact of Viscosity
qI = 0.636 rad, which is extremely close to the value of
0.628 rad = r,5 that was earlier selected as the most
favourable one. For h = 4 the highest value of the func-
tion occurs at about qI =0.77 rad, which is very close to
0.78 rad =r,4. Should we accept this value, correspond-
ing to the highest D
o
, for our calculations? Certainly yes,
since a higher D
o
probably means a more efcient modu-
lation. It is required for modulation, when both signals are
cosine functions, that the frequency of the carrier compo-
nent be much higher then the frequency of the envelope.
In our case, however, there are no two frequencies, but
we introduced D
o
instead. Our intuition strongly suggests
that nature wants modulation with D
o
as high as possi-
ble, which one might call the most efcient modulation.
According to Figure 2, the maximum value of the function
D
o
is around 8 to 10. We should emphasize that this does
not have too profound meaning since this value simply
depends on the choice of the somewhat arbitrary parame-
ters a and u
e
. In fact, we are looking for the maximum of
the product function qI P(qI).
So, we can make two important conclusions based on
the analysis of how D
o
depends on qI. One can see from
Figure 2 that the highest values of the density of the internal
oscillations D
o
increases when the parameter K increases.
What is also important is the fact that an optimal qI, for any
reasonable K, does not differ signicantly from our previ-
ously accepted values of qI =0.63 rad and qI =0.78 rad for
h =5 and h =4. Otherwise, for K =0 one obtains D
om
2
for qI = 0.53 rad for both values of h. This means that,
from the criterion of modulation, the PBD model
11, 12
is bet-
ter then the original PB model,
7
which could be obtained
from the former one by letting K =0.
Now, we can plot our solitonic function 4
n
(t) given by
Eq. (26), or the function , dened by Eq. (5). Before we
do this, however, we want to study how the amplitude A
depends on qI for various K. We should keep in mind that
the value A is the amplitude of the function E
1
(S, t) but
is not the amplitude of the soliton 4
n
(t). From Figure 3
0.3 0.4
0.6
0.9
1.2
1.5
1.8
0.5 0.6 0.7 0.8 0.9
ql (rad)
A

(
n
m
)
Fig. 3. Amplitude A as a function of qI for o =const (K =4 N/m).
0.4 0.5 0.6 0.7 0.8
ql (rad)
A

(
n
m
)
a
b
c
1
2
3
4
5
Fig. 4. Amplitude A as a function of qI for o =o(qI). (a: K =4.5 N/m,
b: K =4.3 N/m, c: K =3.8 N/m).
we can see how A depends on qI for constant o, i.e., for
o = o,2p, the value that was previously accepted for
our calculations,
13
and for h =5. We selected K =4 N/m,
but for different K, both larger and smaller, the curves are
almost the same.
However, for o given by Eqs. (17) and (18) the situation
looks completely different. Figure 4 shows how the ampli-
tude A depends on qI for the three values of K and for
h =5. Instead of minimum (Fig. 3) the function A(qI) now
has a maximum. The highest values are 4.75 nm at qI
0.62 rad for K = 4.3 N/m, 2.31 nm at qI 0.62 rad for
K =4 N/m and 1.78 nm at qI 0.64 rad for K =3.7 N/m.
One can see that those maxima decrease with smaller K.
For K = 4.58 N/m the maximum of the amplitude A is
even 16 nm, which probably does not make physical sense,
while for K = 3.4 N/m the function A(qI) does not have
any maximum at all.
Therefore, we can state three facts concerning the
parameter K. First, there is a maximum value K
m
=
4.39 N/m for qI =0.63 rad. Second, higher K means larger
D
0
, i.e., a better transmission of a signal through the DNA
chain, as was explained above. Third, K should be small
enough to ensure that the amplitude A be small enough so
that the oscillating nucleotide does not exceed the depth
of Morse potential well.
Finally, we can plot our soliton 4
n
(t) given by Eq. (26).
In Figures 57 we show the elongation 4
n
(t) vs. time
for K = 4.3 N/m, 4 N/m and 2.5 N/m, respectively. We
chose h = 5 for all of them. According to the shape of
the curve in Figure 5, and the maximum of the function
4
n
(t), we can conclude that this value for K, i.e., K =
4.3 N/m, would not be acceptable. For K = 4.39 N/m,
4
n
(t) has only positive values with the maximum reach-
ing up to 360 nm which is totally unacceptable because
the expansion of the Morse potential, Eq. (6), presumes
that the amplitude may not reach the Morse plateau! For
4
n
= 360 nm, from Eq. (5) it follows that ,
n
= 2.5 nm
which is far beyond the plateau. However, Figures 6 and 7
J. Comput. Theor. Nanosci. 2, 19, 2005 5
R
E
S
E
A
R
C
H
A
R
T
I
C
L
E
Peyrard-Bishop-Dauxois Model of DNA Dynamics and Impact of Viscosity Zdravkovi c et al.
15
10
5
0
5
48 52 56 60 64
t (ps)


(
n
m
)
Fig. 5. Elongation of the out-of-phase motion as a function of time
(n =300, qI =0.628 rad, h =5, K =4.5 N/m, o =o(qI)).
4
2
0
2
4
6
48 52 56 60 64
t (ps)


(
n
m
)
Fig. 6. Elongation of the out-of-phase motion as a function of time
(n =300, qI =0.628 rad, h =5, K =4 N/m, o =o(qI)).
2
1
0
1
2
3
52 56 54 50 58 60 62
t (ps)


(
n
m
)
Fig. 7. Elongation of the out-of-phase motion as a function of time
(n =300, qI =0.628 rad, h =5, K =2.5 N/m, o =o(qI)).
suggest that the values K =4 N/m and K =2.5 N/m might
be acceptable. Of course, we should keep in mind that
a higher K means more oscillations of the carrier wave
in one envelope, i.e., a higher D
0
. For K = 0 only three
internal oscillations can hardly be discerned which shows,
as we stated above, the superiority of the PBD model
11, 12
over the original PB version.
7
At this point we can speak of a certain range for
accepted values for the parameter K. Also, as was pointed
out earlier, K should be big enough to ensure a large D
0
,
but still not too large to bring about a very large amplitude.
Finally it might be interesting to compare solitonic solu-
tions 4
n
(t) given in Figures 57 with the same function
obtained for constant o.
13
In Figure 8 we show the soli-
tonic wave 4
n
(t) for K =4 N/m and constant o, i.e., o =

p
2o
. Comparing Figures 6 and 8, both plotted with the
same K, we see that the amplitude for the non-constant
o (Fig. 6) is about three times larger than that for the
constant o (Fig. 8). In both cases positive amplitudes are
higher than negative, which is a result of the term with
the parameter o in Eq. (26). One can see that the shapes
of both curves are almost the same. This is not surprising
since D
0
, the density of the carrier wave oscillations of the
breather, does not depend on o.
Finally, we need to point out a couple of advantageous-
ness of this procedure (nonconstant o) over the one when o
was constant. First, the amplitude A(qI) has the maximum
having very big values in a limit when the parameter K
approaches its critical value, which suggests a certain res-
onance behaviour. This procedure solves the problem of a
small amplitude discussed in Ref. [13].
Also, the existence of the upper limit of the parame-
ter K, which we explained in this chapter and in Ref. [14],
solves one more problem. Namely, in Ref. [11] optical
and acoustical frequencies were compared. There are four
crossing points in the rst Brillouin zone for h =4. How-
ever, for K - K
m
those crossing points do not exist and
w
o
>w
o
in the whole zone.
22
In that article,
22
a resonance
48 52 56 60 64
t (ps)


(
n
m
)
1.5
1.0
1.0
0.5
0.5
0.0
Fig. 8. Elongation of the out-of-phase motion as a function of time
(n =300, qI =0.628 rad, h =5, K =4 N/m, o =const).
6 J. Comput. Theor. Nanosci. 2, 19, 2005
R
E
S
E
A
R
C
H
A
R
T
I
C
L
E
Zdravkovi c et al. Peyrard-Bishop-Dauxois Model of DNA Dynamics and Impact of Viscosity
mode was dened. It might be interested to point out
that the resonance mode occurs at qI = 0.785 rad/s and
0.628 rad for h =4 and h =5 respectively. Those are well
known values rst obtained in Ref. [13] and conrmed in
this section by maximization of the function D
0
(qI).
3. THE INFLUENCE OF VISCOSITY
In the previous paper on this topic
13
we attempted to deal
with a more realistic description of DNA dynamics, which
takes into account the impact of the medium, surround-
ing a DNA molecule. It is well known the importance of
hydrating water for the biochemical activities of proteins
and especially DNA. Since water molecules are highly
polar they form an ordered network on the surface of a
protein along which protons can be transferred. The forma-
tion of this network with long-range connectivity has been
detected as a percolation transition when the water con-
tent approaches 0.5 g per 1 g of protein.
23
We suggest that
the helicoidal spring parameter K in the model considered
here may reect the mediating role of the water molecules.
Moreover, the helicoidal structure itself arises as the result
of optimization of the interplay between hydrogen bond-
ing, hydrophobicity and long-range connectivity. On the
other hand, we must take into consideration the fact that
the solvating water does act as a viscous medium that
damps out DNA dynamics, favouring energy expenditure.
We took this effect in Ref. [13] by adding the viscous
force on the nucleotide pairs, i.e., in Eq. (4) of this paper.
In that treatment
13, 24
viscous force was considered as com-
petitive with other forces arising from Hamiltonian (1).
The consequence of that approach was the outcome which
showed the impact of viscosity being so strong that the
breather solution (26) decays almost instantaneously into
its asymptotic form which is localized bell-shaped mode
given by expression:
,
n
(t) =a
2
A
2

j sech
2
_
a
L

(nI V
e
t)
_
(36)
where A

, L

, V
e
are parameters dened by Eqs. (24),
(25), and (14), respectively, renormalized by viscosity
impact. This procedure was shown in Refs. [13, 24] where
we dened so called big and small viscosities. So, the
rst one, big viscosity, was explained in Refs. [13, 24],
while the second one will be in what follows.
Here we start from probably more realistic and favour-
able approach that viscous force has features of small per-
turbation. The viscous forces exerted on the bases within
a pair n are a u
n
and a v
n
. The small a, the same as
in Eq. (5), indicates that viscous force has the character of
small perturbation. It leads to the effective damping force
acting on the out of phase base pair motion as follows:
E
v
=a ,
n
(37)
Now starting from the perturbed equation of motion
m ,
n
= k(,
n+1
+,
n1
2,
n
) K(,
n+h
+,
nh
+2,
n
)
+2

2oD(e
o

2,
n
1)e
o

2,
n
a ,
n
(38)
and performing the expansion according to Eqs. (5), (8),
(10) and (19) one nds
iE
1t
+PE
1SS
+QE
1

2
E
1
=a

2mw
E
1t
(39)
This equation could be solved by the method of slowly
varying parameters developed in Ref. [25]. The essence
of the method is that the carrier wave number O=q +a
u
e
2P
slowly changes with time through change of u
e
.
The breather solution, Eq. (23), can be written in more
suitable form:
E
1
(S,;)=Asech
_
1
L
e
(S;)
_
exp
_
i
_

_
S
u
c
u
e
;
_
+
__
(40)
where =
u
e
2P
is the carrier wave number shift, which is
expected to depend of viscosity, ; =u
e
t is the position of
centre of the breather, and is time phase.
According to Ref. [25], the above parameters should
slightly depend on time on the following ways:
(t) =
0
+] ()t (41)
;(t) =;
0
+
u
e0

0
t (42)
(t) =
u
e0
2
0
_
1
L
2
e
+
2
_
t
=
u
e0
2
0
_
1
L
2
e
+
2
0
_
t +
_
u
e0
] ()t +u()
_
t (43)
where
0
, ;
0
, and u
e0
are corresponding unperturbed
values, while u() denotes possible phase shift of the
breather caused by viscosity.
The wave function of the breather perturbed by viscosity
E
1v
could be expressed by expansion with respect to small
parameter a:
E
1v
=E
1
(S, ;) +a
__
oE
1
o
_
+L
e
_
oE
1
o;
___
u
e0
2
0
_
] ()t
2
(44)
The above approximation holds for time t smaller than
characteristic time scale t
0
=

0
a] (
0
)
. It can be shown on
the basis of Eqs. (41)(43) that this expansion diverges
when a 0 except if the secular sum in square brackets
vanishes. If we use denitions, Eqs. (41)(43), and take
into account that the derivative E
1t
on the right hand side
of Eq. (39) could be transformed as follows
oE
1
ot
=
_
oE
1
o
_
J
Jt
+
_
oE
1
o;
_
J;
Jt
=
oE
1
o;
_
oE
1
o
oE
1
o;
J
Jt
+
J;
Jt
_
(45)
J. Comput. Theor. Nanosci. 2, 19, 2005 7
R
E
S
E
A
R
C
H
A
R
T
I
C
L
E
Peyrard-Bishop-Dauxois Model of DNA Dynamics and Impact of Viscosity Zdravkovi c et al.
the effective perturbation caused by viscosity could be
written as
(, ;, S) = i
_
oE
1
o
] () +
oE
1
o
u()
_

u
e0
w
0
_

m
+] ()
_
oE
1
o;
(46)
The unknown functions from Eqs. (41), (43) ] () and
u() now play the role of Lagrange multipliers, which
could be determined by orthogonality conditions:
_

JS(, ;, S)
_
oE
1
o;
_

=0 (47a)
_

JS(, ;, S)
_
oE
1
o
_

=0 (47b)
where stands for complex conjugates. This removes the
mentioned divergence of Eq. (44). Taking into account the
derivatives
oE
1
o
=iSE
1
, for u
c
=0, and
oE
1
o
=iE
1
, contained
in Eq. (46), the second condition, Eq. (47b), gives that the
phase shift u() is zero in this approximation, while the
rst condition yields the rate of change of carrier wave
number in the form:
J
Jt
=] () =

m
_
D
1+D
_
(48)
where
D =
1
3
u
e0
wL
e
=
1
6
u
2
e0
Pw
(49)
From Eq. (48) easily follows
=
0
exp
_

m
D
1+D
t
_
(50)
Having in mind the denition =
u
e
2P
we get
u
e
=u
e0
exp
_

m
D
1+D
t
_
(51a)
u
e0
=2P
0
(51b)
If we estimate P and w from Eqs. (21) and (13) taking
K = 4 N/m and qI = 0.78 rad, we readily get P = 0.89
10
6
m
2
/s and w =8.6 10
12
rad/s.
If we choose, as before, u
e0
= 10
5
m/s, dimensionless
parameter D, Eq. (49), becomes
D =217 1 (52)
Under such circumstances of high launching breathers
velocity, the envelope velocity, Eq. (27), by using Eqs. (10)
and (19), becomes
V
e
=V
g
+au
e0
exp
_

m
a
2
t
_
(53)
From the absolute viscosity of water (T 300 K), q =
7 10
4
kg/ms, and considering a base as thin rigid rode, it
could be roughly estimated 10
12
kg/s. With a =0.007
the breathers decay time is t
J
=
m
a
2
10
8
s. Starting with
V
e
1.9 10
3
m/s the path passed by the breather for t
J
=
10
8
s reaches approximately 20 10
6
m or 6 10
4
base
pairs along DNA chain. This is quite favourable regarding
expected role of breathers as long-range effects mediators
in DNA.
Finally, on the basis of Eqs. (24) and (25) it follows
that the breathers amplitude decays with the same rate as
envelope velocity, and the width of the breather spreads
exponentially.
4. CONCLUSION AND DISCUSSION
In this paper we have dealt with the PBD model applied
to an idealized DNA chain. The hydrogen bonds in this
DNA chain are represented by the Morse potential, which
results in the nonlinear Schrdinger equation describing
the breather excitations.
In Section 2, we briey described the model proposed.
The nonlinear dynamics in the context of this model
depends on several parameters and the values of some of
them are still rather vague and uncertain. We rst consid-
ered our attention on the issue of choosing the parameter
o, which relates the amplitudes of the rst and second har-
monics in the expansion in Eq. (8). Usually, this parameter
has been taken to be a constant, but a richer physical situ-
ation arises if we take the condition in Eqs. (17) and (18),
which seems more signicant emerging from a lower har-
monic in the perturbation procedure. This brought about
the qualitatively different dependence of the amplitude A
from qI (Figs. 3 and 4). Our second concern was the dis-
cussion of the helicoidal spring constant K.
In Section 3 we introduced viscosity term, i.e., the inter-
action of DNA nucleotides with the surrounding water
molecules. Here, we have expanded our earlier approach
to this problem
13
by considering viscous term as small per-
turbation, Eq. (43). We found that the envelope velocity
of breather exponentially decays eventually reaching the
group velocity, Eq. (14). We estimated that the decay time
for breather with parameters used in this paper is of the
order of t
J
10
8
s. This leads to the path of about 20 jm.
If the electric eld of appropriate frequency is applied on
DNA
26
the energy loss due to viscous dissipation should
be balanced by the parametric resonance effect preventing
breathers decay.
It should be stressed that a great deal of attention
was given earlier to the numerical analysis of the PBD
model but this was done without viscous perturbation.
2730
The so-called discrete breathers arising in these numerical
treatment exhibit a number of very interesting properties,
especially regarding their mutual interactions.
For example, the modulation instability of the wave put
in the discrete PBD Hamiltonian splits it into wave pack-
ets, which are small breathers. Subsequently, the interaction
8 J. Comput. Theor. Nanosci. 2, 19, 2005
R
E
S
E
A
R
C
H
A
R
T
I
C
L
E
Zdravkovi c et al. Peyrard-Bishop-Dauxois Model of DNA Dynamics and Impact of Viscosity
of these breathers tends to favour the largest excitations,
which grow at the expense of others. This process satu-
rates when the breathers, that get narrower as they grow
in amplitude, are so narrow that they are trapped by lattice
discreteness and no longer propagate in the lattice. It still
remains to be demonstrated numerically how the helicos-
ity and viscosity affect the dynamics behaviour of discrete
breathers. It is expectable that the viscosity which shows
opposite tendency of above described growing of ampli-
tude and narrowing of discrete breather could establish
subtle balance which enables breather to self sustain sta-
bility and maintain capability of reaching long way along
DNA chain.
Acknowledgments: This project was supported by
funds from NSERC, MITACS, the Institute of Theoreti-
cal Physics at the University of Alberta and from Serbian
Ministry of Sciences, project No H1822.
References and Notes
1. S. W. Englander, N. R. Kalenbach, A. J. Heeger, J. A. Krumhansl,
and S. Litwin, Proc. Natl. Acad. Sci. USA 77, 7222 (1980).
2. S. Yomosa, Phys. Rev. A 30, 474 (1984).
3. S. Homma and S. Takeno, Prog. Theor. Phys. 72, 679 (1984).
4. Ch. T. Zhang, Phys. Rev. A 35, 886 (1987).
5. V. Muto, A. C. Scott, and P. L. Christiansen, Phys. Lett. A 13, (1989).
6. S. N. Volkov, J. Theor. Biol. 143, 485 (1989).
7. M. Peyrard and A. R. Bishop, Phys. Rev. Lett. 62, 2755 (1989).
8. Ch. T. Zhang, Phys. Rev. A 40, 2148 (1989).
9. L. V. Yakushevich, Phys. Lett. A 136, 413 (1989).
10. L. V. Yakushevich, Nonlinear Physics of DNA, Wiley Series in Non-
linear Science, John Wiley, Chichester (1998).
11. T. Dauxois, Phys. Lett. A 159, 390 (1991).
12. T. Dauxois and M. Peyrard, Dynamics of Breather Modes in a Non-
linear Helicoidal Model of DNA, Lecture Notes in Physics 393,
Dijon (1991), p. 79.
13. S. Zdravkovi c and M. V. Satari c, Phys. Scripta 64, 612 (2001).
14. S. Zdravkovi c, J. A. Tuszy nski, and M. V. Satari c, J. Comput. Theor.
Nanosci. 1, 171 (2004).
15. Peyrard, private communication.
16. G. Gaeta, Phys. Lett. A 172, 365 (1993).
17. M. Remoissenet, Phys. Rev. B 33, 2386 (1986).
18. S. Zdravkovi c, Nonlinear Dynamics of DNA Chain, will be pub-
lished in Finely Dispersed Particles: Micro-, Nano-, and Atto-
Engineering, Edited by A. Spasi c and Jyh-Ping Hsu, Marcel Dekker,
Inc., New York (2005).
19. T. Kawahara, J. Phys. Soc. Japan 35, 1537 (1973).
20. R. K. Dodd, J. C. Eilbeck, J. D. Gibbon, and H. C. Morris, Soli-
tons and Nonlinear Wave Equations, Academic Press, Inc., London
(1982).
21. A. C. Scot, F. Y. F. Chu, and D. W. McLaughlin, Proc. IEEE 61,
1443 (1973).
22. S. Zdravkovi c and M. V. Satari c, About optical and acoustical fre-
quencies in a nonlinear helicoidal model of DNA molecule, submit-
ted to Chin. Phys. Lett.
23. G. Carreri, M. Geraci, A. Ginansanti, and J. A. Ruply, Proc. Natl.
Acad. Sci. USA 82, 5342 (1985).
24. S. Zdravkovi c and M. V. Satari c, Int. J. Mod. Phys. B 17, 5911
(2003).
25. N. N. Bogoliubov and Y. A. Mitropolskii, The Asymptotic Method
in the Theory of Nonlinear Vibrations, Nauka, Moscow (1974), (in
Russian).
26. M. V. Satari c, Physica D 126, 60 (1999).
27. T. Dauxois, M. Peyrard, and C. R. Willis, Phys. Rev. E 48, 4768
(1993).
28. O. Bang and M. Peyrard, Phys. Rev. E 53, 4143 (1996).
29. K. Forinach, T. Cretegny, and M. Peyrard, Phys. Rev. E 55, 4740
(1997).
30. M. Peyrard, Physica D 119, 184 (1998).
Received: 7 December 2004. Accepted: 5 January 2005.
J. Comput. Theor. Nanosci. 2, 19, 2005 9

Potrebbero piacerti anche