Scott M.

O’Grady ANSC/PHSL 5700/PHSL 4700 Cell Physiology Lecture 1

Diffusion, mean square displacement and partition coefficients

Objectives
1. Understand Fick’s law of diffusion and the factors that constitute the diffusion coefficient.

2. Understand the random walk concept for diffusion of a single molecule and why calculating
the average position of a population of molecules does not provide an adequate description of
diffusion.

3. Understand how a determination of the root-mean-squared (RMS) displacement offers an

effective measure of diffusion.

4. Understand why the square root of time dependence of diffusion makes the process
inefficient for transport of molecules over macroscopic distances.

5. Know the factors that affect rates of solute diffusion within the cytoplasm and how they can
be measured using recovery from fluorescence photobleaching.

6. Understand the compensations to the rate limiting aspects of diffusion within cells.

7. Understand the meaning of partition coefficients and their importance in describing the
diffusion of a solute across a membrane.

8. Understand the definition of the permeability coefficient and how it is used in the constant
field (Goldman-Hodgkin-Katz) equation.

Readings

1. Miller, C., Ionic hopping defended. J. Gen. Physiol., 113:783-787, 1999.

2. Verkman, A.S. Solute and macromolecule diffusion in cellular aqueous compartments.

Trends Biochem Sci., 27:27-33, 2002.

3. Brangwynne, C. et al., Cytoplasmic diffusion: molecular motors mix it up. J. Cell Biol.,
183:583-587, 2008.

4. Gorman, J. and E.C. Greene, Visualizing one-dimensional diffusion of proteins along DNA.
Nature Struct Mol Biol., 15(8):768-774, 2008.

5. Helenius, J. et al., The depolymerizing kinesin MCAK uses lattice diffusion to rapidly target
microtubule ends. Nature, 441(4):115-119, 2006.

1
Scott M. O’Grady ANSC/PHSL 5700/PHSL 4700 Cell Physiology Lecture 1

A. Diffusion from the continuum perspective

Diffusion in dilute aqueous solution can be viewed macroscopically as a mechanical process
where solute molecules behave as rigid spheres moving through a viscous fluid. A mathematical
description was initially formalized by Adolf Fick (1855) who recognized that diffusion was
analogous to heat transfer and that Fourier’s law of heat conduction could be adapted to describe
the process. Fourier’s law states that the rate of heat transfer is dependent on the temperature
gradient, surface area and thermal conductivity of the materials engaged in heat exchange. The
sum of Fick’s law of diffusion and Ohm’s law constitute a more general description of electrolyte
movement in dilute solutions which is known as the Nernst-Planck electrodiffusion equation. In
the Nernst-Planck view, solvent molecules are treated as a continuum, offering frictional
resistance to the motion of rigid sphere solute molecules (hydrated ions) as a result of an applied
force (an electric field or concentration gradient). The concept of rate from the continuum
perspective is perceived macroscopically as movement reflected by the average velocity. Thus
when the force is zero, solute flow stops and the average velocity is zero.
1. Diffusion and Fick’s law

Figure 1: Solute diffusion between compartments separated by a semi-permeable membrane

Fick’s law of diffusion:

C or J    D C when expressing the flux per unit area
Eq. 1 J   AD
x x
moles cm 2 moles
 cm 2  
sec sec cm 4
Where: J = flux (moles/sec), A = area over which diffusion occurs (cm2),
D = diffusion coefficient and C / x = concentration gradient
This equation defines the rate at which solute molecules move past some point along the path of
diffusion. If the concentration gradient is linear, then the driving force is the same at all points
along the diffusion pathway. A non-linear concentration gradient indicates that the driving force
changes as a function of position. The derivative of the equation that describes how
concentration changes as a function of position defines the driving force at that position.

2
Scott M. O’Grady ANSC/PHSL 5700/PHSL 4700 Cell Physiology Lecture 1

a. Jnet= -AD ( c)

dc
b. Jnet= -AD ( dx)
x
c dc
.

driving force = driving force (x1) =

x
Concentration (c)

Concentration (c)
dx

driving
dc
force (x2) =
dx
membrane membrane
thickness thickness

X1 X2 X1 X2
Position (x) Position (x)

Figure 2: Linear and non-linear changes in solute concentration vs. position

B. Definition of the diffusion coefficient

The diffusion coefficient is comprised of terms that define the average kinetic energy of the
solute (which is the Boltzmann constant (k) multiplied by the absolute temperature, T) and a
coefficient of friction () between the solute particle and the solvent.

kT RT
Eq. 2 D or
 NA 
cm 2
Recall that R = (NA · k) and that the final units are .
sec
A more complete description of the diffusion coefficient requires a definition of the coefficient of
friction (). Note that Stokes law states:
v 1
Eq. 3 F f  6rvd and d 
F f 6r
Where Ff = frictional force, vd = drift velocity, r = molecular radius and  = viscosity. By
v sec
definition, d is the mobility () of the solute and has the units of . The inverse of mobility
F gm
gm
is the coefficient of friction ( ) so that:
sec
J gm  cm 2
 K
kT RT 2 cm 2
Eq. 4 D or K  sec 
6r N A  6r cm 
gm gm sec
cm  sec sec
Where: kT = the average kinetic energy of a solute molecule, r = molecular radius, NA =
Avogadro’s number and  = viscosity of the medium

The diffusion coefficient is proportional to the speed with which a solute molecule can move in
the surrounding medium. The term takes into account the size of the solute, its average kinetic
energy and the viscosity of the surrounding medium.

3
Scott M. O’Grady ANSC/PHSL 5700/PHSL 4700 Cell Physiology Lecture 1

C. Einstein’s random walk concept of diffusion as an explanation of Brownian motion

In the early part of the 19th century, Robert Brown, a British botanist, described the random
movement of pollen grains in solution, a phenomenon that would later be known as Brownian
motion. Albert Einstein envisioned that a Brownian particle in solution behaves like a large atom
constantly colliding with much smaller particles (solvent), moving at random resulting in
movement of the Brownian particle. Einstein demonstrated in his 1905 paper that a precise
connection exists between diffusive processes on a macroscopic scale and random movement of
particles on a microscopic scale. Since Brownian motion could be observed and measured, it was
possible to confirm Einstein’s theoretical calculations and predictions by careful measurement of
Brownian motion. In 1926, the French physicist Jean Perrin would receive the Nobel Prize for his
determination of Avogadro’s number (NA) by applying Einstein’s relation to his measurements of
Brownian motion.
1. Diffusion from the random walk/rate theory perspective
Unlike the continuum perspective, the random walk/rate theory viewpoint recognizes the
molecular nature of the diffusion process. Rate theory visualizes diffusion as a series of discrete
steps and allows us to consider the various factors that influence movement at each step. It also
emphasizes the statistical nature of diffusion, focusing on the transitions from one site to another
without the need for identifying a driving force as is the case when the more mechanical
continuum perspective is applied. In addition, the solvent (e.g. water) is thought of as a matrix
through which a solute moves by jumping from one point to another. The jumping rate is
determined by the intrinsic thermal energy and by any external force acting on the solute.
Opposition to movement is not viewed as the frictional resistance of the solvent as reflected in
the continuum perspective, but is instead thought of as energy barriers that must be overcome in
order for the solute to jump to the next site.
Another important distinction between the continuum perspective and rate theory is the concept
of rate itself. As mentioned earlier, the continuum approach views the concept macroscopically,
such that the net rate of solute flow is dependent on the average velocity. In contrast, the rate
theory perspective is derived from an analysis of chemical reactions, where the equilibrium state
is dynamic. In other words, translocation events are thought of as having two components, a
forward rate and a backwards rate analogous to a chemical reaction. Thus two rate coefficients
(sec-1) can be defined that describe the forward and backward unidirectional movements of the
solute.
k1
Site 1 Site 2
k-1
Although the units of rate coefficients k1 and k-1 are not the same as permeability coefficients
(cmsec-1; see details on page 9), the two are comparable if the rate coefficients are multiplied by
the distance separating the sites (x):
Eq. 5 J net  J12  J 21  J net  k1 (x)[S ]1  k1 (x)[S ]2

Where Jnet represents the net flux, J12 and J21 represent the unidirectional fluxes and [S] x
represents the solute concentrations at either site 1 or at site 2.

4
Scott M. O’Grady ANSC/PHSL 5700/PHSL 4700 Cell Physiology Lecture 1

The overall goal of rate theory is to provide a molecular and energetic interpretation of the rate
coefficients for a single jump, and more generally, to provide a description of transport from a
more molecular perspective.
2. Analysis of the random walk: mean squared displacement in one dimension
# of molecules # of molecules # of molecules
1000 50 20

0 0 0
-100 -50 0 50 100 -100 -50 0 50 100 -100 -50 0 50 100
Position Position Position
Figure 3: Movement of individual molecules as a “random walk”

Individual molecules moving in one dimension (as equal steps, either left or right along the x
axis) in an independent random fashion spread out symmetrically from the initial position (0).
Note however that as time proceeds, there is no change in the average position of the population
(still 0) because movement is random and an equal chance exists for each individual molecule to
take a positive or negative step.
Analysis of the root mean squared (RMS) displacement however provides a measure of how
spread out the initial cluster of molecules become after a certain amount of time has elapsed (For
details, see derivation provided in Lecture Supplement 1: RMS displacement).
y
Eq. 6 x  2Dt
dy d(x,y)
Figure 4: RMS vectors
Where: d = displacement, D = diffusion coefficient and t = elapsed
1D 2D
time

Vector analysis can now be used to determine the solution of the RMS x
0 1D dx
displacement from a point source in two dimensions.

Figure 4 represents diffusion from point zero to position d(x,y) in two dimensions. As illustrated,
this is simply calculated as the vector sum of diffusion in the x and y dimentions using the
Pythagorean theorm:

Eq. 7 xy x x d 2  y d 2  (2Dt) x  (2Dt) y  4Dt

For three dimensions, the displacement is combined for x, y and z to give:

x2
Eq. 8 xyz x  (2 Dt ) x  (2 Dt ) y  (2 Dt ) z  6 Dt ;
t
6D
Let D = 10-6 cm2/sec, when: x = 100 m; t = ~17 sec

5
Scott M. O’Grady ANSC/PHSL 5700/PHSL 4700 Cell Physiology Lecture 1

RMS displacement can be used to define the amount of time required for a solute to diffuse over
a given distance. The relation shows that the time required for diffusion increases as the square of
the distance. This imposes an important physical constraint on biological systems since diffusion
is slow over macroscopic
100
distances.
Figure 5: This concept 80

is illustrated by
comparing a linear

Distance m
60

(slope = 1) vs. square-

40
root dependence of
distance as a function of 20
time. In the linear case,
as time increases, 0

distance increases by 0 20 40 60 80 100

equal increments.
Time (ms)

However, when distance increases as the square root of time then the distance traveled is less as
time increases. Also note that as the distance travelled decreases, average velocity must also be
decreasing as shown on the right.
D. Fick’s law of diffusion from the perspective of RMS displacement
Consider two adjacent sites (“a” and “b”) located at x1 and x2 that are separated by a distance x
as indicated in figure 6. Suppose that n1 is the number of particles at site “a” and n2 at site “b”.
Particles are restricted to jumps between positions x
x1 and x2 and no interaction between particles
x
occurs. Let f define the mean jump
 # jumps 
frequency   , the average number of
 sec  x
times that a particle jumps in any one direction, to
a b
x1 x2 x
the right and to the left. The unidirectional flux of
particles moving from site “a” to site “b” (Ja-b) and
from “b” to “a” (Jb-a) is J a b  f  n1 ; J ba  f  n2 .
Therefore, the net flux from site “a” to site “b” is Ja-b
given by J net  J a b  J ba . Jb-a
Figure 7: Diffusion as a series of “jumps”
Substituting the unidirectional flux terms as defined above yields:
# jumps
Eq. 9 J net   f (n2  n1 ) , where f 
tsec
Rewriting the expression in terms of concentration:

 ( n( 2 x )n31 )   (C  C1 )   C   C 
Eq. 10 J net   f (x)  
2
   f (x) 2   2    f (x) 2     J net   D  
 x   x   x   x 
# jumps cm 2
Note the units of the product f  (x) 2   cm 2   D !!
sec sec

6
Scott M. O’Grady ANSC/PHSL 5700/PHSL 4700 Cell Physiology Lecture 1

The physical meaning of the diffusion coefficient from the view point of a one dimensional
random walk is revealed by further analysis of RMS displacement.

Recall that (x) 2  2Dt . In the context of random walk analysis, the mean squared displacement
after N jumps is given by (x) 2N  Nd 2 , where d = mean distance per jump.
N f
Combining these equations yields D   d 2   d 2 and thus the diffusion coefficient can be
2t 2
f
envisioned as depending on jump frequency and the average size of the jump. The term
2
represents the frequency of jumps in one direction, assuming a random distribution of jumps
between sites. Interestingly, the units of the diffusion coefficient have greater intuitive meaning
in this context as the product of the squared displacement and the
jump frequency.
This distinction has important applications especially when ion
permeation through a channel is considered. In fact, continuum
models at this time as represented by the Nernst-Planck equation
do not adequately describe certain aspects of ion movements
through channels, particularly saturation and ion-ion interactions
in multi-ion pores. However, diffusion represented as a series of
hops or jumps from one site to an adjacent site through the
channel can be analyzed by applying principles of rate theory
(Eyring Rate Theory) which incorporates an important concept of
ion interaction with the channel itself in addition to other ions that
may be present within the pore.
Figure 8: Diffusion time course

E. Diffusion of solutes and macromolecules in aqueous cellular compartments:

Perspectives on the properties of the cell interior

have evolved from the concept of a viscous gel to
one of a water-like, but crowded compartment.
Measurements of fluorescent probe diffusion and
fluorescence correlation microscopy have
revealed that diffusion of small molecules within
cells is significantly slower than in saline
containing films (comparable to extracellular
fluid). Figure 9: Fluorescence recovery after photobleaching
1. Measurement of diffusion within cells: Fluorescence recovery after photo-bleaching
This approach involves the use of fluorescently labeled molecules loaded into cells or expressed
as green fluorescent protein (GFP) chimeras. Irreversible spot photobleaching of fluorophores in
a known volume is produced by exposure to a brief intense laser light pulse. The diffusion of
unbleached fluorophores into the bleached volume is then measured as a quantitative index of
translational diffusion.

7
Scott M. O’Grady ANSC/PHSL 5700/PHSL 4700 Cell Physiology Lecture 1

Determinants of translational mobility of solutes like second messengers, nucleotides and various
metabolites have been identified by using small fluorescent probes like BCECF (often used as an
intracellular pH indicator). Diffusion of BCECF in cytoplasm was found to be 4 fold slower than
in water (Dcyto/DH2O). Reasons for the slower diffusion rate include:
1. Fluid phase viscosity: cytoplasm has greater microviscosity compared to water or saline films.
2. Binding to intracellular constituents: small solutes can bind to macromolecules or organelles.
3. Molecular crowding: collisions with other solutes
within the compartment.
The primary factor that accounts for the reduced rate
of diffusion within the cytoplasm was molecular
crowding.
Another important factor is the size of the solute. As
shown in Figure 10, large macromolecules can have
dramatically reduced translational diffusion rates due
primarily to molecular crowding and binding to
intracellular constituents. Figure 10: Effects of size on diffusion rate

2. Compensations to the limits of diffusion: Cytoplasmic diffusion and molecular motors:

Recent studies of the motion of tracer particles within the cytoplasm indicate that random motion
has an unexpectedly greater magnitude than what
would be predicted from thermal fluctuations
(responsible for Brownian motion) indicating that
biological activity within the cell gives rise to random
fluctuating motion that significantly influences
diffusion of small molecules within the cell.
Fluctuation motion within the cell is associated with
the activity of molecular motors (e.g. myosin II
motors) that interact with actin networks.
Microtubules embedded within these networks exhibit
random bending motions that are driven by motor
activity, producing mixing-type movements that
appear to enhance rates of diffusion within the cell
(Brangwynne et al. 2008). Figure 11: Microtubule network
An additional compensation to the limits of the diffusion within cells is the organized assembly
of multi-protein complexes that can include enzymes, receptors, ion channels and other proteins
associated with cell signaling into compartmentalized microdomains. Such compartmentalization
positions interacting molecules in close proximity, thus reducing the time required for diffusion.
3. One-dimensional Diffusion Accelerates Molecular Motors
The depolymerization dynamics of microtubules is regulated by MCAK, a member of the
kinesin-13 family of motor proteins which controls microtubule length during cell division.
Interestingly, MCAK is capable of targeting the ends of microtubules in a one-dimensional (1D)
random walk. The transition from three-dimensional diffusion to 1D diffusion corresponds to a
"reduction in dimensionality" that has been proposed as the search strategy by which DNA

8
Scott M. O’Grady ANSC/PHSL 5700/PHSL 4700 Cell Physiology Lecture 1

enzymes find specific binding sites (see Gorman and Greene, 2008). MCAK uses this strategy to
target both microtubule ends more rapidly than direct binding from solution.

Figure 12: Mean-squared displacement of MCAK is plotted as a function of. The linear fitted
curve reveals a diffusion coefficient,
D (obtained from the slope), of 0.38
mm2/s;  x 2   2 Dt .

E. Partition Coefficients
Since biological membranes are
characteristically hydrophobic and
non-polar while the bathing
solutions are highly polar solutes
often exhibit different solubility in
the membrane relative to aqueous
solution which can affect concentration differences across the membrane. Partition coefficients
() are quantities that express the ratio of a solute concentration just inside the membrane (Cm)
relative to the adjacent solution (Caq):
C
Eq. 11   m so that; Cmi  Ci and Cmo  Co
Caq
The diffusion equation can be written to account for concentration differences in the membrane:
C C o  C mi C  Ci
Eq. 12 J  D  D m   D o
x x x
Inclusion of the partition coefficient serves to more accurately represent the solute concentration
gradient within the membrane. Partitioning can alter the driving force for movement through the
membrane and therefore the net flux across the membrane. Note when = 1, the solute dissolves
equally well in aqueous solution and in the membrane, thus partitioning does not contribute to
Jnet. However, when  > 1, the solute = 1 > 1 < 1
preferentially enters the membrane resulting in a
steeper concentration gradient within the
Co Co Co
membrane and an increase in Jnet. When < 1,
the solute is polar and prefers the aqueous phase,
resulting in a reduced gradient and a lower Jnet. Ci Ci Ci
When = zero, there is no entry of solute into the Jnet Jnet Jnet
membrane and no transport across the membrane. out in out in out in
F. Permeability
By definition, the permeability coefficient (P which is used in the Goldman-Hodgkin-Katz
equation) has the units of cm/sec and is the product of the partition coefficient and diffusion
coefficient divided by the thickness of the membrane.
D
Eq. 13 P
x

Rearrangement of the Stokes-Einstein equation yields: D = kT

9
Scott M. O’Grady ANSC/PHSL 5700/PHSL 4700 Cell Physiology Lecture 1

s J
Recalling that  = ion mobility, k is Boltzmann’s constant ( ) and T is temperature (K).
gm K
s J
 K
kT gm K cm
Eq. 14 Therefore: P   
x cm s

This is the definition of permeability as it is used in the GHK voltage equation:

Eq. 15 RT PNa Na o  PK  K o  PCl  Cl i

Vm  ln
zF PNa Na i  PK  K i  PCl  Cl o

10