Problem Set 3 Solutions


1. We have a formula for the thermally-averaged dipole-dipole interaction potential energy, for two identical dipoles of moment p:
< U > = -  2
3
k2 p4
kB T r6
 
We need the dipole moment for a water molecule, which must be on the order of an electron charge times about 0.5 A or about 0.8 ×10-29 C·m. This estimate is a bit too large because the electron charges are rather spread out in the H2O; the real value is about 0.6 ×10-29 C·m, which is 2 Debye units (recall 1 Debye = 3×10-30 C·m).

We just plug in p = 6×10-30 C·m, k = 9×109 N·m2/C2, kB T = 4×10-21 J, and r = 3×10-10 m. The resulting potential is

< U > = - 2
3
(9×109)2 (6×10-30)4
(4×10-21) (3×10-10)6
 
= 2.4 ×10-20 J = 6 kB T

This estimate neglects the dielectric screening caused by the water molecules in the 3 A between our two interacting water molecules. At a range of 3 A, a value of dielectric constant of e = 5 is reasonable, which reduces k to k/e. Thus our estimated interaction potential ends up reduced by a factor of 25, to < U > = 9.6 ×10-22 J = 0.24 kB T.

By comparison, a hydrogen bond has a cohesive energy somewhat larger than this, about 5 kB T.


2. We will consider the electric field due to the point charge a distance r from the charge, and the dipole whose orientation is subject to thermal fluctuation.

The interaction energy between a point dipole p and a static electric field E is U = -p·E = -p E cosq, where q is the angle between the electric field and the dipole direction. The probability of each orientation is proportional to the Boltzmann distribution, exp[pE cosq/(kB T)]. We will call the combination p E/(kB T) = a.

Integrating over all directions of the dipole using the Boltzmann distribution, including the normalization, gives the averaged potential

< U > = - p E 
ó
õ 
p

0 

dqsinqcosqexp[a cosq]

ó
õ 
p

0 

dqsinqexp[a cosq
 
 
These are just the integrals encountered in the theory of the random flight polymer. Here their meaning is different, but the calculation is just a repeat of the random-flight polymer elasticity calculation that you can find in the notes...
= - p E  é
ê
ë 
cotha -  1
a
ù
ú
û 
 
For U << kB T, we have a << 1, so that we can keep just the linear term in the power series expansion of the above,
< U > = -  p E a
3
= -  p2 E2
3 kB T
 
Finally we should plug in the value of E = k q / er2 for the E-field a distance r from the charge, giving
< U > = - p k2 q2
3 kB T e2 r4 
 

3. The charge q simply induces a dipole moment proportional to the field at the molecule, p = aE = ak / (er2).

Note that this induced dipole moment is always in the direction of the electric field, no matter what the orientation is of the molecule.

Therefore the interaction potential energy is simply

U = - p·E = - ak2
e2 r4 
 
(b) Averaging over the orientations of the molecule does not change this, since the induced dipole moment is always in the same direction:
< U > = - ak2
e2 r4
 

4. (a) This problem involves a `two-state system', where one state the random (open) coil, and where the other is the (closed) hairpin. The Boltzmann distribution tells us
Prandom
Phairpin
exp[-Grandom/(kB T)]
exp[-Ghairpin/(kB T) 
= exp[DG/(kB T)]
 
Since Prandom + Phairpin = 1 we have then
exp[DG/(kB T)] =  1 - Phairpin
Phairpin
 
or
DG
kB T
= ln æ
ç
è 
1 - Phairpin
Phairpin 
ö
÷
ø 
 
Reading off the hairpin probabilies at 40, 50 and 60 C (see graph), Phairpin = 0.9, 0.5 and 0.1, and plugging in gives us DG / (kB T) = -2.2 , 0, and +2.2. The change in kB T over the 20 degree range of these measurements is only a few percent, so we just take kB T = 4×10-21 J, which means
DG = -8 ×10-21 J, 0 and +8×10-21 J at 40, 50 and 60 C.

(b) The dependence of DG on temperature is nearly linear,
DG = (8 ×10-21 ) (T - 323 )/10 J, where T is now in Kelvins (recall 50 C = 273+50 K = 323 K). Rewriting this

DG = -2.6 ×10-19 J + (0.8×10-21 J/K) T = DE - T D
 
The simple identification of this linear dependence with energy and entropy differences as indicated above gives us

DE = -2.6 ×10-19 J = 65 kB T.

The energy (or enthalpy depending on your taste) difference is about 65 kB T, or about 8 kB T per base pair.

DS = -8 ×10-22 J/K = -58 kB.

This entropy difference can be interpreted in terms of the ratio of the number of states of the random coil vs the hairpin:

Wrandom
Whairpin 
= exp[ (Srandom - Shairpin)/kB ] = exp[ -DS ] = exp[58] = 1.5 ×1025
 
Rather mind-bogglingly, the random coil has about 1025 times as many states as the hairpin. Each base in the random coil therefore has something like 103 times the number of states as when it is a hairpin.


5. (a) We suppose G1 - G0 = DG. The relative probability of these states is given by the Boltzmann distribution as
P1
P0
= exp[-DG / (kB T)] 
 
Using P1 + P0 = 1 tell us the individual probabilities,
P1 1
1 + exp[DG / (kB T)]
 
and
P0 1
1 + exp[-DG / (kB T)]
 
(b) The next step is to follow the directions, and to suppose that the energies of the two states are E0 = 0 and E1 = -Ebinding, and the entropies are S0 = Scst - kB logc and S1 = 0. Therefore the free energies are
G0 = E0 - T S0 = - T Scst + kB T logc 
 
 
G1 = E1 - T S1 = -Ebinding 
 
and the relative free energy is
G1 - G0 = -Ebinding + T Scst - kB T logc
 
Finally we find the probability for the protein being bound,
P1 1
1 + exp[-Ebinding/(kB T) + Scst/kB] / c 
 
(c) The entropy goes as -logc since 1/c is proportional to the volume available per free protein, which is in turn proportional to the total number of states. Entropy is then given by kB times the log of the number of states.

(d) If the DNA site is occupied with probability P1 = 0.95 at concentration c = 10-6 M, we will need (106 M) exp[-Ebinding/kB T + Scst/kB] = 0.05. This determines exp[-Ebinding/kB T + Scst/kB] = 5×10-8 M, and therefore

P1 1
1 + (5 ×10-8 M)/c 
 
The plot of P1 vs. concentration on log scale is below:

(e) Kd = 5 ×10-8 M = 50 nM, a typical dissociation constant for a DNA-binding protein.


Note: John apologizes for the bad wording, and messy notation of problem 5 in the assignment.


File translated from TEX by TTH, version 2.53.
On 5 Mar 2001, 23:05.