Download as pdf or txt
Download as pdf or txt
You are on page 1of 35

ANNUAL

REVIEWS Further
Quick links to online content

Copyright 1973. All rights reserved

LONG-RANGE PHYSICAL FORCES IN THE 9029


BIOLOGICAL MILIEU
V. ADRIAN PARSEGIAN
Annu. Rev. Biophys. Bioeng. 1973.2:221-255. Downloaded from www.annualreviews.org

National Institutes of Health, Bethesda, Maryland

I. BACKGROUND

Despite enormous progress in understanding the genetics and biochemistry of


by George Mason University on 09/04/13. For personal use only.

molecular synthesis we still have only primitive ideas of how linearly synthesized
molecules form the multimolecular aggregates that are cellular structures. We assume
that the physical forces acting between aggregates of molecules and between indi­
vidual molecules should explain many of their associative properties; but available
physical methods have been inadequate for measuring or computing these forces in
solids and liquids.
During the past several years there has been some success in determining several
features of these forces, in particular their range and magnitude. One purpose of
this review is to summarize current understanding of long-range physical forces as
they have been formulated to attack problems of biomolecular organization.
Another is to develop criteria for deciding when the theory of forces and energies is
an appropriate vehicle for analyzing biological structure. Third I would like to
point out several areas needing experimental or theoretical work for a better physical
understanding of biological organization.
For present purposes, "long-range forces" are those physical forces acting at
distances large compared with interatomic or intermolecular spacing (much greater
than 5 A for solids and liquids). The reason for this restriction is simply that no
reliable physical theory of forces in liquid water exists for shorter distances.
There are two classes of such long-range interactions: electrostatic (or Coulomb)
forces due to electric charge on bodies, and electrodynamic (or van der Waals) forces
which can arise between electrically neutral bodies. The behavior of these two is of
contrasting character.
Coulombic interactions are repulsive between bodies of like charge and are
usually attractive between bodies of opposite charge. They die exponentially (or
faster) with separation and the rate of decay incrcases with the concentration of
salt in the intervening medium. The van der Waals force is attractive between like
bodies, and the strength of interaction at very long distances decreases relatively
slowly, as a power of the separation.' The magnitude of long-range electrostatic

1 These properties of long-range forces form the basis of the Derjaguin-Landau-Verwey­


Overbeek theory of colloid stability. The book (Ref. 1), written before 1948 by Verwey and
Overbeek, is still an excellent introduction to this theory.

221
222 PARSEGIAN

forces is proportional to the product of electric charge on two bodies; that of


electrodynamic forces depends on the strength and similarity of electromagnetic
absorption spectra of the two bodies and of the intermediate medium.
Historically, the theoretical treatment of charge-charge interactions is based on
the Debye-Huckel model of salt solutions (2). That analysis has been extended,
without appreciable modification, to formulate the interactions between multiply
charged polyelectrolytes (3-5) and large colloid particles ( 1 , 6, 7). A slowly convergent
series of critiques, modifications, improvements, and justifications of the many
assumptions made in the Debye-Huckel approach suggests that it is probably
qualitatively correct for applications in the long-distance regimes, where it will be
Annu. Rev. Biophys. Bioeng. 1973.2:221-255. Downloaded from www.annualreviews.org

discussed here (4, 8).


Electrodynamic forces have traditionally been formulated using principles
appropriate for almost-ideal gases. Consequent estimates of van der Waals forces in
aqueous milieux were ambiguous. It is only in recent years that a more appropriate
by George Mason University on 09/04/13. For personal use only.

formulation was developed, mainly by Lifshitz (9), Rytov (10), and Dzyaloshinskii,
Lifshitz & Pitaevskii ( 1 1). They followed in spirit the suggestion in 1 894 by Lebedev
( 1 2), who pointed out that there is a common physical basis for the absorption
spectrum of a substance and its tendency to undergo the spontaneous charge
pisplacements that underlie electrodynamic interactions. Their physical theory
provides an explicit expression for the relationship between the force and the ab­
sorption spectra of the materials composing each of the interacting bodies and the
intervening medium. Subsequent theoretical work has been directed toward develop­
ing practical and accurate methods for unambiguously converting spectral properties
into forces, testing these methods successfully against experimental measurements,
and developing a catalog of expressions for the interactions between bodies of
varying shape and separation.
Sections II and III briefly review the physical assumptions and resulting formulas
used in the computation of long-range forces; the more mathematical parts of these
sections may be passed over in a first reading for an intuitive picture. Section IV
summarizes and illustrates the contrasting principal properties of electrostatic and
electrodynamic forces. In Section V these properties are examined in the context of
several biological phenomena----cell contact, lipid membrane formation and arrange­
ment, viral assembly and aggregation-to consider when it might be reasonable to
think about these systems in terms of the physical forces between their components.
The last section, VI, points out some problems requiring work in the theoretical
understanding of physical forces or in collecting experimental information.

n. ELECTROSTATIC FORCES; THEORETICAL SUMMARY

One may conveniently distinguish two types of electric charge in a biological


system. The more prevalent are in the form of mobile ions, principally Na+, K + ,
Cl-, Ca+ +, Mg+ +, Mn+, H +, and OH - , which are present in virtually all aqueous
substances. There are also significant amounts of charge covalently bound to macro­
molecules, to aggregates of molecules, or to broad regions such as membrane
surfaces. The origin of the charge on these "fixed" moieties is chemical-acidic or
LONG-RANGE PHYSICAL FORCES IN BIOLOGY 223

basic dissociation or association, the magnitude of total charge having to do with


the chemical equilibrium. What concerns us is the electrostatic interactions that
take place between macromolecular (or multimolecular) bodies bearing fixed charges
and acting across a saline medium of electrolyte solution.
Confinement of groups bearing charges of like sign to a defined region such as to
one molecule or to a surface, in opposition to their Coulombic repulsion, will result
in a net electrostatic potential whose sign is that of the bound species. The effect of
this potential will be to attract mobile ionic "counterions" (species of charge sign
opposite to the fixed charge) and to repel "coions" (ions of the same charge sign).
We speak of an electrostatic "double-layer" potential 'P(r), where ze 'P(r) is the
Annu. Rev. Biophys. Bioeng. 1973.2:221-255. Downloaded from www.annualreviews.org

electrostatic energy of a charge (ze) at the position r. The fixed charge imposes the
boundary conditions initially responsible for the potential while the ionic medium
both adapts to that imposition and modifies IJ' by redistribution of the mobile
by George Mason University on 09/04/13. For personal use only.

charges.
There is a vast literature on the problem of finding 'P(r) in ionic solutions, par­
ticularly near surfaces bearing fixed charge ( 1 3), between colloidal particles ( 1 ), and
around polyelectrolyte molecules (3, 4, 5,14, 15). The present summary is restricted
to long-range electrostatic interactions that are likely to occur between biological
structures across intervening media. More specifically, in the continuum model that
we use, the suspending solvent is imagined to be a continuum dielectric whose
molecular structure is ignored, wherein also the spatial distribution of discrete
electric charges is averaged over time, and assumed to vary smoothly with position.
The physical theory used here is not as rigorous as that used for determining van
der Waals interactions. This is unavoidable at present since even if a better physical
theory were available, experimental information would still be lacking on the struc­
ture of water as solvent and on the arrangement of charges on biological molecules.
We consider first the role of the mobile charge, where it is necessary to recognize
immediately that the position of these charges determines, as well as depends on,
the shape of 'P(r).
Poisson's equation prescribes how the density of charge per) at a given point
determines the divergence of the electric field, E(r) - V'P(r), from that point
=

V2'1'(r) =
_ 4np(r) 2.1
e

Here V2 is the Laplacian operator and F. the dielectric "constant" of the medium
(roughly 80 for water at room temperature).2 The use of a constant £ implicitly
assumes that the medium is a continuum dielectric substance, an assumption that is
questionable both for high potential gradients and for high ionic concentrations.
But the density of charge p(r), itself carried on mobile ions, depends on energies
lv,,(r) of each ion species, z. Since lv,,(r) is a statistically averaged potential energy
reflecting all interactions of the ionic particles [e.g. with the solvent, the discrete
nearby charges of opposite sign (4, 8)], its determination is a major problem. When

2 See Slater & Frank, "Electromagnetism" (17) for a good description of Poisson's

equation.
224 PARSEGIAN

the ionic energies are not subject to wide fluctuations one may assume, following
Debye (2), that the physically important part of �(r) is essentially ionic charge (ez)
times the time-averaged electrostatic potential 'P(r) at position r. Debye's assumption
is highly approximate but will be satisfactory for present purposes as long as:
(a) the energies ez'P are small compared with the thermal energy kT [where It is
Boltzmann's constant, T absolute temperature (4, 8, 15, 16)] or when (b) the ions in
a region of strong potential are predominantly of like charge sign so that they will
not experience strong transient interactions with discrete ions of opposite charge.
Then the concentration distribution nAr) of ion species of valence z will be given by
a Boltzmann distribution
Annu. Rev. Biophys. Bioeng. 1973.2:221-255. Downloaded from www.annualreviews.org

2.2
Here y(r) is the "reduced potential"
by George Mason University on 09/04/13. For personal use only.

y(r) = e'P(r)/kT 2.3

and n? is the concentration of species z at a point in the salt solution where the
potential is most conveniently taken to be zero. Equation 2.2 simply says that an
electrochemical potential, Jl.o + ze'P(r) + kT In (nz(r», is the same everywhere or
that all points accessible to mobile ions are in electrochemical equilibrium.
Under these assumptions, the charge density is a sum over all charged species,
p(r) = Lnz(r)ez. This is introduced into Poisson's equation (Equation 2. 1 ) to give a
" Poisson-Boltzmann" differential equation for the electrostatic potential,

4ne2
V2y = ---I nzO e-zy(r) 2.4
ekT z
This nonlinear differential equation for the electrostatic potential is difficult to
solve mathematically because of the exponentials e±zy from the charge density. I f
the reduced potentials yare s o small that z yi s much less than 1 , then the exponentials
can be written in the approximate expanded form 1 ± zy + . . . Substitution of these
.

expansions yields a linearized P-B equation (also due to Debye)

2.5
where K is the very useful "Debye constant,"

2 8nne2
/( =-- 2.6
skT
For a neutral salt solution, L n?z is zero; n is the ionic strength.
When this linearized form of Equation 2.5 is accurate (i.e., when zy � 1), it can
be used to solve problems in several geometries where the full P-B Equation 2.4 is
mathematically intractable. This is especially useful in cylindrical and spherical
geometries; most planar and lamellar configurations can be solved using the non­
linear Equation 2.4. Comparing the exact and approximate solutions in planar
geometry gives a useful indication of the accuracy of linearizing the P-B equation.
LONG-RANGE PHYSICAL FORCES IN BIOLOGY 225

What may be the magnitude of these potentials, and over what distances will they
extend from a region of fixed charge? The natural units of'¥ are kT/e � 25 mY.
For y = e,¥/kT � 1,'¥will be �25 mV. All distances can be reduced to dimensionless
quantities by dividing them by the "Debye length" I/K, the natural unit of length for
a salt solution. For physiologic saline, approximately a . 14 molar solution, the Debye
length (1/K) is approximately 8.3 A.
Gaussian boundary conditions for the second-order P-B equation, due to the
distribution of fixed charge, are chosen to suit the geometric nature of the assembly
under study. The number of fixed charges is some fraction of the total number of
Annu. Rev. Biophys. Bioeng. 1973.2:221-255. Downloaded from www.annualreviews.org

potentially ionizable groups bound to a molecule or large aggregate (18-21). Imagine,


for example, an acid dissociation reaction HA � A- + H+ producing a bound
anionic charge A - plus a proton H+ from the neutral species HA. Since both A­
and H+ are at the same electrostatic potential at the locus of dissociation, one may
by George Mason University on 09/04/13. For personal use only.

use a dissociation constant Kd for this reaction that is characteristic of the species
HA in free solution. Then we can write the local concentration of H+ based on
chemical dissociation

2.7

where the brackets denote concentrations; r:x is the degree of dissociation, [A-)/
([A -) + [HA), which must be determined. [H+)} decreases with r:x.
There is a second, physical, relation between [H+) and 11.. At the position of HA,
O
the concentration [H+)f is related to [H +) , the H+ concentration or activity in
the bathing medium (where'¥ is usually taken as zero, infinitely far from HA) by a
Boltzmann relation

2.8

Here YHA = e'¥HA/kT is the reduced potential at HA.


Now the electrostatic potential field lJI{r) is initially due to the dissociation of
fixed ionizable groups and is therefore a function of r:x,'¥(r; IX). There is physical
electrostatic attraction of H + to the region of fixed charge. [H+); increases with IX.
The functions 10gIO[H+)j and 10gIO[H+)J are shown for one particular problem
in Figure 1. Since [H+]j and [H+]; refer to the same quantity, they must be equal.
The correct value of r:x is where the lines cross. This value will depend on several
variables, namely: Kd, the ionic concentrations in the salt solution, the surface
density of ionizable groups, the pH of the reference solution, as well as the shape and
location with respect to other fixed charge of the ionizable surface region. Its deter­
mination requires solution of the governing P-B equation for '¥(r; a) and an iterative
determination of IX from that solution. The restriction to one kind of ionizable
group may be removed by considering simultaneous chemical equilibria for several
acidic or basic groups.
In practical applications, recognition of the variable degree of dissociation
provides a natural link between the chemical description of a charged surface and
its physical electrostatic properties.
226 PARSEGIAN

7
pHo= 7

A = 40 �2
n =.1 M, I-I electrolyte

..
..:s
9
(!) 5
Annu. Rev. Biophys. Bioeng. 1973.2:221-255. Downloaded from www.annualreviews.org

0
...J
f

r
a.
by George Mason University on 09/04/13. For personal use only.

.2 .4 .6 . 8
a= FRACTION OF GROUPS IONIZED

FIGURE 1. Graph showing pH, = -log lOrH+ Js vs degree of dissociation. Acid dissociation
(Equation 2.7) and electrostatic attraction (Equation 2.8) predict two different [H+Js con­
centrations for a given Cl. Example is for a planar surface bearing one ionizable group per
4o_A2 area, facing a O.IM solution of 1-1 electrolyte, pH 7. Curves are given for two different
dissociation constants 1.5 x 10-5 M and 1.5 x 10-4 M (pK 4.8 and 3.8). Subscript s
(rather than f) emphasizes dissociation at a surface as in section A(i) below.

For the electrostatic forces between bodies bearing fixed charge we use the poten­
tial as expressed in Equations 2.4 or 2.5 to compute the electrostatic interaction
between fixed-charged bodies A and B immersed in an ionic solution. In most cases
we can take the zero of energy to be at infinite separation of A and B, and the electro­
static free energy to be the work required to move the bodies to mutually proximate
positions.3
Since by definition of the problem all mobile ionic charges are in electrochemical
equilibrium with the bathing solution, the free energy of ions will not change with
displacement of the fixed-charge bodies A and B. It is the electrostatic energy of the
fixed charges that changes with separation.
Most generally one may perceive the force between A and B via the electric and
osmotic stress set up in the intervening medium by the charged bodies (1, 4, 1 9, 24).
One imagines any conveniently chosen surface enclosing A or B and computes the

3 This definition of zero will not always work when there is no salt in the bathing medium.

In these exceptional cases (for example, polyelectrolytes in distilled water) it is preferable to


define the energy via a Debye-type charging process from the zero of a discharged state
(22, 23). The equivalence of the two definitions is clearly described in the book by Verwey
& Overbeek (1).
LONG-RANGE PHYSICAL FORCES IN BIOLOGY 227

stress across that surface. We can write down the momentum flux through each
incremental piece of surface via an electroosmotic stress matrix. Mathematically the
stress is written as a sum of two matrices. One is the electric or Maxwell stress
tensor [a full discussion is given in (25)]. The second is an osmotic pressure matrix
with n(r) = kT"'[, n z(r) on the diagonal elements and zeros off the diagonals. In
practice, the stress matrix may be drastically simplified by choosing an enclosing
surface such that most or all of the elements are zero or constant.

A. PLANAR GEOMETRIES
Annu. Rev. Biophys. Bioeng. 1973.2:221-255. Downloaded from www.annualreviews.org

When the interacting charges on bodies A and B are on fiat surfaces facing each
other across a planar region (see Figure 2), the electrostatic potential will vary only
in the direction normal to these surfaces. Choosing this normal direction as the x
axis, we have 'I' 'I'(x). Any electrical fields set up by the charged surfaces will be
=
by George Mason University on 09/04/13. For personal use only.

in the x direction only.


(i) An isolated planar surface that lies between a region of ionic solution and a region
that excludes ions :-Let there be ionizable groups of surface density 11A and disso­
ciation constant Kd as defined above (Equation 2.7). For a degree of dissociation (X
the surface charge density will be as = -e(XIA. The slope of the potential at the
surface is proportional to as, (d'l'/dx) =-4naje. Infinitely far from the surface the
electrostatic potential will be zero, as will its slope. Using these two conditions, we
find that the reduced potential Ys at the surface is given by (20)

2n0'2
-- = L nO(e-ZY, - 1) 2.9
£kT zZ

Equation 2.9 for Ys can be used to determine (X for an isolated surface when the
surface H + concentrations predicted by dissociation and the Boltzmann distribution
are combined as described above and in Figure 1. In practice, rather than use the

impenetrable ionic impenetrable


region solution region

reference solution

FIGURE 2. Geometric scheme for considering the interaction between two planar charged
surfaces bounding ion-penetrable bodies. The region between bodies is in equilibrium with
a defined salt solution.
228 PARSEGIAN

graphical method of Figure 1, one may consider Equation 2.9 as a polynomial in


powers of (e Yo) and solve for (e- Ys) in any mathematicalJy convenient way depending
-

on the coefficients n� .
When the bathing medium is a uni-univalent electrolyte solution or a divalent
cation-univalent anion (2-1) electrolyte, the final Ct. can be found analytically (20).

(ii) Layer of finite thickness containing charges:-Consider next a modification


wherein the fixed-charge groups are not in a mathematically ideal plane but occupy
a region of finite thickness 7: into which mobile charge can also permeate. We speak
Annu. Rev. Biophys. Bioeng. 1973.2:221-255. Downloaded from www.annualreviews.org

of the surface potential as an average y, over this region. On one side of the y, layer
the space is impermeable to ions while on the other there is an aqueous medium in
ionic equilibrium with a defined salt solution. An electrostatic double-layer potential
will develop on the aqueous side of the fixed-charge region.
by George Mason University on 09/04/13. For personal use only.

The net charge within the fixed-charge layer will be the sum of (negative) bound
charges, again taken at a surface density of Ct.IA, and mobile charges, 7:' Lz n�(ze).
So the net surface charge density at the aqueous interface where the double layer
begins is

a, = { -� + r . L n�z e-ZY,] 2.1 0

Solution for net charge density (J" mean potential y, (or Ys), and degree of disso­
ciation Ct. follows from substituting Equation 2.10 into 2.9. Note that the magnitude'
of (J will always be less than when layer thickness r = 0 because the sign of y, will
be the same as that of the fixed charge (as long as no additional binding of mobile
species is assumed).
Computation of effective charge densities for any other models of charged surface
[e.g., (26) and (27)J may be similarly treated.

(iii) Two parallel planar charged surfaces :-Exact formulas exist for the one­
dimensional electrostatic P-B potential Equation 2.4 (a) between equally charged
parallel plates in equilibrium with a solution containing uni- and divalent cations
(+ charges) plus univalent anions (- charges) (19, 20) and (b) between unequally
charged plates confronting a uni-univalent electrolyte solution (28, 29). These
expressions are written in terms of Jacobi elliptic functions expanded as theta
functions (30). One value of these formulas is the opportunity to check the accuracy
of the simpler, linearized, approximation Equation 2.5.
When the origin of surface charge is an acid dissociation, the degree of dissociation
Ct., Equation 2.7, is essentially constant upon mutual approach from infinite separation.

This holds until the separation distance is of the order of the Debye length 1 1K,
Equation 2.5 (19, 20).
The mutual force is computed by considering the stress across a surface in the
intervening medium parallel to the charged plates when the plates are at finite
separation minus the stress at infinite separation. Some pressure vs distance curves
using the treatment of Ninham & Parsegian (20) are shown in Figure 3. The solid
lines are for repulsions across uni-univalent electrolyte solutions. The dashed curves
LONG-RANGE PHYSICAL FORCES IN BIOLOGY 229

107 105

r
Annu. Rev. Biophys. Bioeng. 1973.2:221-255. Downloaded from www.annualreviews.org

PRESSURE
(dynes/cm2 )

106 104
by George Mason University on 09/04/13. For personal use only.

(dynes/cm2 )

105 103

FIGURE 3. Repulsive pressure between parallel charged surfaces bearing ionizable groups.
Forces are exponential and very sensitive to the fraction t/ of positive mobile charge on
divalent ions. Data are for a reference solution of O.l4-M ionic strength and acid dissocia­
tion constant Kd 1.5 X 10-5 M (20).
=

t/ = .05 give the pressure when 5 % of the (positive mobile charge is made of divalent
cations.
These curves were drawn for surface charge densities and salt concentrations
characteristic of cells in physiologic saline, and the charge was assumed to reside on
carboxyl groups. The force is clearly exponential down to distances near the Debye
length 11K::::: 10 A. The small amount of divalent cation decreases the force by some
25%.
Using the linearized P-B equat io n Equation 2.5, Parsegian & Gingell (31) found
,

that between two unequally charged surfaces 0"A and 0"B separated by distance t the
pressure is
p = 8n O"� + O"AO"B(e,<t + e-Kt ) + O"� 2.11
8 (eKt e-Kt)2
_
230 PARSEGIAN

This assumes that ions are excluded from the regions outside the plates. Sarkics &
Sammut (29) have solved the nonlinear Equation 2.4 for this case and extended the
work of Derjaguin (28). They find that the approximate expression Equation 2.11
gives an accurate representation of the pressure for small surface potentials (;S25mV).
This last relation offers wide flexibility in applications because any model for
surface charge densities aA, aB can be introduced. All considerations of acid (or base)
dissociation, finite thickness of the charge layer, and distribution of fixed charge
within the fixed charge region, may be built into a's, which are themselves dependent
on separation t and potentials YA or YB at the planar surface.
Note that P can be repulsive for aA and aB of opposite sign. Also P reduces exactly
Annu. Rev. Biophys. Bioeng. 1973.2:221-255. Downloaded from www.annualreviews.org

to the correct limit, 21[a2/1"., when charges are equal and opposite (aA = -fJB = fJ)
and there is no salt in the bathing medium. In the limit of high salt or large distances
(Kt » 1), P reduces to the form of an exponentially screened interaction of two
by George Mason University on 09/04/13. For personal use only.

charged plates, (8n/£)0"AO"B e-Kt.


The linearization approximation formally restricts application to cases where
potentials yare small compared to 1. In practice, this restriction may often be relaxed
by computing effective 0"A and aB for infinitely separated surfaces using the full
nonlinear expression. These "exact" a's may then be fed into the equation for P.
As long as changes in Y with finite separation t are smaller than unity, a linearized
expression for these changes in potential, and consequently for changes in the a's,
will closely approximate the exact expressions (20, 31).
We may relax the restriction that ions cannot permeate either side of the charged
layer. If both sides are in equilibrium with a reference electrolyte solution, then
potentials characteristic of isolated surfaces will exist on the outsides of the layers.
The charge densities will be proportional to the difference in electric fields on either
side of the charged layer. For the linearized treatment, the pressure becomes

2.11a

B . SPHERICAL PARTICLES

Because of mathematical difficulties, most formulations of the electrostatic


properties of charged spheres in ionic solution have been simplified by making
strongly restrictive approximations peculiarly appropriate to the specific problem at
hand. Consequently, a rather extensive literature exists for particular cases of inter­
actions between spherical particles. This literature has dealt primarily with problems
in colloid sciences ; the conditions on the formulation [e.g., strong potentials (y > 1 )
held constant over the spherical surface during the approach o f two particles] are not
necessarily conditions that apply to the biological system. Bell, Levine & McCartney
(24) have recently reviewed this literature. Comparatively little has been done under
assumptions suitable to the electrostatic interaction of spherical bodies in a biological
context (32). For present purposes it will be enough to point out the major features
of spherical systems, especially how they differ from the properties of planar
geometries.
LONG-RANGE PHYSICAL FORCES IN BIOLOGY 231

(i) Isolated sphere.-The electrostatic potential about a singlc spherical body will
depend only on the distance r from the center of the sphere. For very weak potentials,
z· Y < 1, the solution about a sphere of radius a is
e-Kr

y = const x -­ 2.12
r

where the constant is found either from the surface charge density (] (take r = a at
the surface) or from the surface potential y(a).
For stronger potentials, numerical methods (33-35) or approximations specific to
Annu. Rev. Biophys. Bioeng. 1973.2:221-255. Downloaded from www.annualreviews.org

the problem must be used. For example, when the radius is much greater than the
Debye length, the potential very near the sphere surface can be solved as if the surface
were planar (36a). This approximation would be appropriate near a biological cell
where radii of curvature are in microns while the Debye length in physiologic saline
by George Mason University on 09/04/13. For personal use only.

is some IDA.
All methods for determining surface-charge density from dissociation or a par­
ticular model can be carried over from the planar analysis to apply to an isolated
sphere.

(ii) Two charged spheres.-The electrostatic potential created by two spheres


requires inclusion of angular terms in V2qt and has not been solved analytically.
The two-sphere problem is greatly simplified by assuming that the potential has
one value over the entire surface of each sphere4 and that these two surface potentials
change negligibly as the spheres are brought together.
In that regime, Bell, Levine & McCartney (24) use a linear superposition approxi­
mation to describe the mutual approach of spheres 1 and 2 having radii ai' a2'
effective surface potentials Y1, Y2, and center-to-center distance R. They assume
that potential y(r) in a region between spheres is a sum of potentials YI(r) + Y2(r) due
to the isolated spheres and that very near either sphere the single-sphere potential
dominates, that is, Y :::::: YI or Yz near sphere 1 or 2, respectively.
The individual reduced potentials are

2.13

where 'I' '2 are distances from sphere centers and, for present purposes,
K2 = 8nne2/6k T, n = 1/2 I no Z2 reflecting the ionic composition of the medium.
(These forms for YI and Y2 are for the small Y approximation. According to Bell,
Levine and McCartney, when the isolated sphere bears a large surface potential,
one may choose effective surface potentials at effective radii where the small Y
approximation is valid.)
The sum of YI and Y2 is used to compute the osmotic and electric stress over a
planar surface that bisects the line between spherical centers. (The stress computation
surface may be closed by imagining a hemisphere of infinite radius surrounding one

4 The uniform surface potential will be reasonable when surface charge is able to spread
freely over the spherical surface.
232 PARSEGIAN

of the spheres. The stress is constant on this hemisphere.) The force is


a a
force = £'I's1 'I's2 (1 + KR � e-"(R�al -az) 2.14
'ji2
The free energy of bringing the two spheres together at constant surface potential is
then the (-) integral of this force from infinite separation. This expression is at
least qualitatively correct when the distance between the outer edges of the two
spheres is much greater than the Debye length, i.e. when K(R - at - a2) » 1. In the
case where the medium bears no salt (K = 0), it reduces to Coulomb's law between
Annu. Rev. Biophys. Bioeng. 1973.2:221-255. Downloaded from www.annualreviews.org

point particles.
C. CYLINDRICAL PARTICLES

Problems in cylindrical geometry have been worked out mainly for linear poly­
by George Mason University on 09/04/13. For personal use only.

.electrolytes (4, 5, 36-38). There is still no exact expression for the electrostatic poten­
tial using the nonlinear P-B Equation 2.4 about a single charged rod in salt solution,
but methods for matching different formulas near and far from the rod seem to give
good results (39, 40). For charge-lined cylindrical pores there are formulas appro­
priate to both very high concentrations and zero salt in the bathing medium (41).
Brenner & McQuarrie (42, 43) recently reviewed and developed expressions for
potentials and free energies of an isolated rod and two parallel interacting rods.
They used the linearized P-B Equation 2.5. Charge densities were derived self­
consistently as due to acid dissociation. It was not necessary for them to assume a
constant surface potential or surface charge density over the whole rod surface
during interaction of two rods. They found also that the degree of dissociation IX of
charges on electrostatically repulsive cylinders varied negligibly with separation
distances greater than the Debye length.
The force had the approximate form
e-KR

force = canst x CD 2.15


y KR
for KR » 1 , where the constant is independent of the axis-to-axis distance R. The
range e-KRIJIdi of cylindrical interactions at distances KR » 1 is therefore inter­
mediate between planar, e-Kt, and spherical, e-KRIR, forces.

III. ELECfRODYNAMIC FORCES, THEORETICAL SUMMARY

At every point in a material body there arise spontaneous transient electric and
magnetic fields due to the transitory displacement or fluctuation of electric charge.
These charge fluctuations occur because of thermal agitation or because of natural
uncertainties in the positions and momenta of electrons and atomic nuclei. It is the
interaction of the electric charges and currents during these fluctuations that con­
stitutes the van der Waals or electrodynamic interaction.
One may think of the local transient fluctuations in terms of underlying con­
tributions from oscillations at all possible frequencies. It is physically intuitive
[and can be shown rigorously ( 1 O)J that the strength of contribution to the local
LONG-RANGE PHYSICAL FORCES IN BIOLOGY 233

fluctuation from a given frequency is proportional to the absorption of light at that


frequency by the material. Both phenomena are a measure of the ease with which
charges in a material can move either in responding to an applied disturbance
(incident radiation) or in following their own laws of motion.
The field fluctuations set up by local oscillations act to move charges at other
positions and in that way influence or perturb the fluctuations occurring elsewhere.
Each point in the system undergoes charge fluctuations that are the sum of its own
spontaneous motion and of its response to charge fluctuations elsewhere. In effect,
each point simultaneously transmits and receives radiation to and from all other
Annu. Rev. Biophys. Bioeng. 1973.2:221-255. Downloaded from www.annualreviews.org

points. The fluctuations occurring throughout the system are thereby correlated.
Since absorption wavelengths are typically much greater than the interatomic
distances observed in solids and liquids, the correlation of electromagnetic fluctua­
tions implicitly requires the simultaneous participation of many atoms or molecules.
by George Mason University on 09/04/13. For personal use only.

This requirement is especially important when one considers the correlation


between fluctuations arising in two large (i.e., multiatomic) bodies separated by
great (i.e., greater than interatomic) distances. On this "macroscopic" scale the
correlated electric and magnetic fields can be pictured schematically as oscillatory
waves within and between the bodies. Since the change in energy of an electromagnetic
field is the change in energy of the charges creating that field, one may write the energy
of correlating charge fluctuations as the sum of energies of electromagnetic waves set
up at each frequency by those fluctuations. The physically important part of the total
wave energy is its variation with separation of the two bodies. And it is this part one
uses to evaluate the van der Waals force between them (44--46).5
The problem is then to determine the pertinent electromagnetic waves or modes
as they depend on the size, shape, and separation of the interacting bodies. When the
bodies are large and distant compared with interatomic spacings, it is possible to
analyze these waves in terms of the macroscopic Maxwell wave equations. This
macroscopic-continuum view has two immediate virtues: the electromagnetic
fluctuations can be formulated via a rigorous physical theory; and answers are
expressed in terms of macroscopic electric and magnetic polarizabilities experimen­
tally accessible from spectroscopic measurements.
The historical alternative to this macroscopic-continuum approach was based on
formulas for the attraction of two isolated point particles. These were derived by
Keesom (48) for two permanent dipoles, by Debye (49) for a permanent dipole with
a nonpolar particle, and by London (50) and Casimir & Polder (50a) for two nonpolar
molecules (5 1, 52, 52a). In all cases the energy of interaction goes as the inverse
sixth power of separation when the finite velocity of light is ignored. In order to
treat problems involving solids and liquids, Hamaker (53) assumed that one could
add up the interactions of all incremental points two at a time and that these incre­
mental interactions followed an r-6 law. These assumptions can be shown to be

5 This is only one convenient way to look at the electrodynamic energy. It does give

identical answers to those found using the exact method of Dzyaloshinskii, Lifshitz &
Pitaevskii (II) or the earlier work of Lifshitz (9) (mentioned in the introduction). The best
description of the fundamental physical theory is found in references (II) and (47).
234 PARSEGIAN

inappropriate for condensed media on straightforward quantum mechanical grounds


without resorting to comparison with the Lifshitz macroscopic-continuum theory
(54a-54d).
A. INTERACTION BETWEEN Two LARGE BODIES SEPARATED
BY A PLANAR GAP
The first rigorous analysis of the interaction between two macroscopic bodies
was derived by Dzyaloshinskii, Lifshitz & Pitaevskii using quantum field theoretic
techniques (9, 11, 47). [Since their original derivation there have been several alter­
nate methods suggested (45, 46, 55-58) for more easily obtaining their results or
Annu. Rev. Biophys. Bioeng. 1973.2:221-255. Downloaded from www.annualreviews.org

small variations on it.] They treated the case of two large bodies of uniform isotropic
substances Land R separated by a relatively small planar slab of thickness ( filled
with medium m. Their expression for the electrodynamic force between the two bodies
can be integrated to give the work or free energy bringing them to the distance ( from
by George Mason University on 09/04/13. For personal use only.

infinite separation (59).


It is instructive to write this electrodynamic (ed) free energy in the highly approx­
imate form (60)

Ged(t) =
kT
-- ",
00 foo x In (1 - A(n) A(n) e-X) dx
Lm Rm 3.1a
L.... 0
11-'0 n:=

[
8,"p2 rn

00
kT A(O)
Lm A(O
)
J
::::::; ___ Rm + A(n) A(n) (1
L Lm Rm
+ rn) e-rn 3.th
8nt2 2 n= 1

The quantities /lt� and /lW� here are differences in the dielectric susceptibilities of
the component substances

3.2

The susceptibilities are evaluated at frequencies i�n where

)' _ 2nkT
'on - n n, n = 0,1, 2··· 3.3

21th, k, T are Planck's constant, Boltzmann's constant, and absolute temperature.


The quantity fn is the ratio of (2(8:':2 (i�n)/c), the travel time of an electric signal
across the gap t and back, to the characteristic period (1/�n) of the frequency �n'

3.4
and c is light velocity in vacuum. The prime in summation, I', indicates that the
n = 0 term is multiplied by t.
Equation 3.1 explicitly asserts that the van der Waals interaction is due to differ­
ences in the material polarizabilities as measured by the /l's at each frequency �n'
The first term in the brackets in Equation 3.1 refers to electrostatic polarizabilities
(or dielectric constants when one neglects conductivity, see below). For example, if
substances Land R are both oils of dielectric eL eR 2 and the medium m is = =

water em 80, then /ll�) = /lR<:!.)= 0.95.


=

The sum in the second term of the brackets in Equation 3.1 refers to /l's evaluated
at higher frequencies. The factor (1 + fn)e-'· takes into account the loss in correlation
LONG-RANGE PHYSICAL FORCES IN BIOLOGY 235

of fluctuations because of the finite travel time of an electric signal across the gap t.
The result is always a "retardation damping" of the energy. When travel time across
and back equals the characteristic period 1/�n' then rn = 1 and the (1 + rn)e-rn is
74%. When rn 2, the factor is 41 %.
=

Because the ratio rn is proportional both to separation t and to frequency �n'


each term in the free energy Ged(t) of Equation 3.1 has a different distance dependence.
Since the LiLm(�n)' LiRm( �n) are experimental quantities peculiar to each pair of materials
at the particular �n' the relative weights of the terms in Ged will differ from one prob­
lem to the next. There is no general rule for the distance dependence of Ged(t).
Annu. Rev. Biophys. Bioeng. 1973.2:221-255. Downloaded from www.annualreviews.org

This is most emphatically true in systems with aqueous components since the first
term Li\:.o� Li�� is often comparable to the entire sum I:� 1 in the second term of the
brackets.
Somc rules for the t-dependence of Ged can be given for special cases:
by George Mason University on 09/04/13. For personal use only.

If, for example, one neglects the finite velocity of light and sets c = 00, then rn
in Equation 3.4 is zero and the bracketed terms in Ged are independent of t. In this
"nonretarded" limit, Ged goes as t-z.
In the case where the 8'S are virtually independent of frequency c;n
frequency term in the brackets is small compared with the sum I;;x'= 1, then Ged will
vary with the inverse third power, r3, of distance.
A third qualitative feature expressed in Equation 3.1 is the specificity of van der
Waals attraction between like substances. The attraction Gab of two different sub­
stances A and B (which can sometimes even be a repulsion) is weaker than the
arithmetic mean of attractions Gaa and Gbb between the same substances. On the
basis of Equation 3.1, bodies of materials A or B suspended in medium m would
minimize their free energy by making aggregates of their own kind.

B. AN EXAMPLE OF THE INTERACTION PREDICTED BY EQUATION 3.1

The general expression for the interaction between two flat semiinfinite media
has been tested and successfully applied to several systems: quartz (12, 61-63),
mica (64-66), and glass (67) across air or vacuum, and thin films of liquid helium
(68, 69) or hydrocarbon (70) between air and a solid or liquid substrate.
A measurement by Haydon & Taylor (71) for the attraction of two bodies of
water across a thin glycerol mono-oleate hydrocarbon film has been especially
helpful in examining the behavior of van der Waals forces. Their technique was to
observe Newton's rings about small drops of oil that had failed to drain from the
film during formation. This observation gives the drops' angle of deviation from the
planar film surface; from this angle one may infer the consequent pressure due to a
van der Waals force acting across the film. The two advantages of this system are the
absence of air or vacuum and the presence of water, which is unusual in that it has
strong absorption over the entire spectrum.
Partly for reasons of tradition (53), the free energy Ged(t) is generally discussed in
terms of a quantity A, the Hamaker coefficient, such that

A(t, T )
3.5
236 PARSEGIAN

And from the approximate form used here for Ged(t), Equation 3.1 , we have for
water acting across hydrocarbon

3.6

where substances Land R are water wand medium m is hydrocarbon h. At t' ;::;; 50 A
the measured value for Ged is -3.94 X 1 0-3 erg/cm2 and consequently

Aexpt 10-14 erg


Annu. Rev. Biophys. Bioeng. 1973.2:221-255. Downloaded from www.annualreviews.org

:;::;; 4.7 X

Since this result depends on an estimate of film thickness t in addition to any other,
systematic, error, the ambiguity in the experimental A(t, T) is probably about 20%.
by George Mason University on 09/04/13. For personal use only.

Computing A(t) requires use of the "imaginary frequency" dielectric suscep­


tibilities e(i�.). This has been a major obstacle to more general application of the
theory. Discussion of a(i,") is given in (72-76a). Parsegian & Ninham (59, 74)
suggested writing e of frequency w in the form used by spectroscopists:

for frequencies up through the ultraviolet. 6 The imaginary part of e is related to the
absorption spectrum of a material; absorption and reflection spectra, together with
refractive indices and dielectric constants 8 (
Cd, 'd, Ct, Wt, and Yt. The required quantity e(i�n) is obtained by introducing w = i�n
into the form Equation 3. 7 after determination of coefficients.
For the water-hydrocarbon-water system, Gingell & Parsegian (76) estimated
A (50 A) in the range 3.4 .:-:;: A .:-:;: 6.8 X 1 0-14 erg. Ambiguities arise from limited
knowledge of ultraviolet absorption spectra and of the film's refractive index.
These results are much smaller than the order-of-magnitude A 1 0-12 erg expected
-

from the original DLVO theory ( 1 ) and somewhat larger than the 1967 estimate
A = - 8.8 x 10- 15 inferred by Fowkes from interfacial energies (77).
More than half the observed free energy of interaction can be accounted for in
terms of the static polarizabilities of hydrocarbon and water, when ew;::;; 80,
eh ;::;; 2, (A�h/2) ;::;; 1 /2. Since kT;::;; 4 x 1 0-14 erg, the first term in Equation 3.6
yields A = (3/2)kT( A;h/2) � 3 x 1 0-14 erg. Because this first term is proportional to
temperature T (exclusive of the change in e's with temperature), most of its contri­
bution to the free energy Ged(t) is due to an increase in entropy rather than a decrease
in energy (78). This will be true of any nonpolar substance immersed in water. We
still need a clearer picture of any structural changes that may accompany this
entropy increase.

6 Contributions from the far ultraviolet to X-ray regions are not important in the present

instance. A different form for E may be used when these very high frequencies are to be
included (59,75).
LONG-RANGE PHYSICAL FORCES IN BIOLOGY 237

The remaining terms from the sum II are essentially independent of temperature
(59, 78) but are subject to retardation (which is very small at 50-A separation). The
individual quantities /).2 in these terms are much less than 1 and change sufficiently
slowly with frequency �n to be shown as a smooth plot. From the curve given in
Figure 4, some 10% of the total energy GeV) comes from �n' corresponding to
infrared frequencies, while the remaining 40% is due to fluctuations at ultraviolet
frequencies (76).
Prior to observations on this system, one usually assumed that van dcr Waals
interactions were dominated by fluctuations in the ultraviolet. At least for systems
Annu. Rev. Biophys. Bioeng. 1973.2:221-255. Downloaded from www.annualreviews.org

including aqueous media but not gases, this assumption is qualitatively misleading.
In those cases one must not neglect either the effects of temperature on low­
frequency polarization forces, or the contribution of molecular vibration at infrared
frequencies, or the different energy-vs-distance relations of the several contributions
by George Mason University on 09/04/13. For personal use only.

to the electrodynamic energy from fluctuations at different frequencies.

C. MORE FORMULAE FOR OTHER PLANAR SYSTEMS

Many generalizations of the formula for two flat uniform isotropic media have
appeared in the past few years. Complete expressions for isotropic materials including

100
<.!)
>-
a::
w
z
w
12
z
0
i=
::>
<0 50
ii:
f-
Z
0
U
w
>
i=
.q
...J
W
a: 0

Hydrocarbon
2.0
--------....
......
Water ......
,�
E(i�) 1.5

1.0
Vis-
- Infrared ible Ullroviolet --

14 15 16 17
LOG,o FREQUENCY � (Rodions per Sec)
-3 -4 -5 -6
t ! I I

LOG,o WAVELENGTH (em)


FIGURE 4. Spectrum of fluctuations contributing to the electrodynamic attraction of water
across a hydrocarbon film. Curve is for infrared and ultraviolet fluctuations, which account
for half the total interaction (76).
238 PARSEGIAN

all effects of finite light velocity and temperature can be derived by using art electro­
magnetic mode summation analysis for triple-layered films (75), for planar bodies
coated with any number of layers (79), and for infinite multilayers in which planar
bodies of finite thickness are stacked interspersed with layers of medium (80).
Similar results are found by a many-particle integration method that neglects
finite light velocity and temperature (81, 82).
Anisotropic media (where the I: are tensors) can exert a mutual torque as well as
an attractive or repulsive force with distance (83-85).
When the dielectric susceptibility changes continuously near a boundary surface,
Annu. Rev. Biophys. Bioeng. 1973.2:221-255. Downloaded from www.annualreviews.org

rather than stepwise as usually assumed, there can be qualitative changes in the
energy-distance dependence (86).
Interactions between planar systems can generally be written in the form approx­

oo x In (1
imated in Equation 3.1 (79):
by George Mason University on 09/04/13. For personal use only.

Ged(t) = -
kT 00
2 L
8nt n=O Jrn
- �Lm�Rm e-X) dx 3.8

but the quantities I'lLm' I'lRm are themselves functions of x and of the geometric
variables describing the problem, as well as of material polarizabilities I:.

D. Two THIN HYDROCARBON FILMS ATTRACTING IN WATER


For example, in the attraction of two thin hydrocarbon films immersed in water,
I'lLm = I'lRm = I'l

(1 ��;h-:XI:XIU)
has the approximate form (87)

� = �wh 3.9

where a t = film separation and b thickness. We again speak of a coefficient


= =

A(a, b) such that the free energy Ged(a, b) = -A/12na2 and again speak of contri­
butions Amw, Air> Auv from each frequency region. Figure 5 shows wide differences in
the relative rates of change of each component with separation a.
Here the low-frequency term (which is the largest term) changes most slowly.
The Air infrared contribution has essentially the behavior one would have obtained
from pairwise summing inverse-sixth-power interactions between incremental
pieces of the system. Auv, from the ultraviolet, decays even faster than Air because
higher frequencies are affected sooner by retardation damping. If the intervening
medium were a salt solution rather than pure water, the low-frequency term Amw
would be subject to an approximate factor (1 + 2KU) e- 2KO, where K is the charac­
teristic Debye constant Equation 2.6 of the salt water (88). This is again different
from the behavior of Air and Auv. (Note that when I'lwh is approximately equal to
1 (ew » eh), then the expression in brackets in Equation 3.9 is also close to 1 ; Equation
3.9 and the energy Ged(a, b) are only very weakly dependent On thickness b.)

E. SPHERE-SPHERE INTERACTIONS, SMALL PARTICLES IN SOLUTION

Because of mathematical difficulties, there are still unresolved problems in the


formulation of sphere-sphere interactions. Mitchell & Ninham (89) have used a
LONG-RANGE PHYSICAL FORCES IN BIOLOGY 239

100%

Zero Frequency,
80% - ' - -_� An.O
"
"

60%
A (0,50)\)
I A ( O , 501l)
40%

Ultrav iolet,
20%
Annu. Rev. Biophys. Bioeng. 1973.2:221-255. Downloaded from www.annualreviews.org

10 20 50 100 200 500 1000


o ( )\ ) -
by George Mason University on 09/04/13. For personal use only.

FIGURE 5. Relative decay with separation of contribution to van der Waals forces for two
interacting planar slabs of thickness b and separation a. For two hydrocarbon films in
pure water the microwave, infrared, and ultraviolet contributions are roughly in the ratio
50 % : 10 % : 40 % when the slabs are very close (b » a) (87).

mode of summation analysis that ignores finite light velocity to obtain asymptotic
series for the leading term plus correction terms when spheres are either very close or
far apart compared with their radii. For two spheres of radii al and a 2 with center-to­
2 2
center distance R they define the quantity b such that R = ai + a� + 2b and
quantities csch , = al/b, csch 'I = a2lb. The interaction energy for close separation is
(L'1im = (Ei - Em)/(Ei + Em»
kT 00 fOC!
2(, + '1 ) 2 nI= '0
G(R) = X dx In (1 - � l m� 2 m e-X )

[ ' In ' I00 '


0

kT
+ (1 - � l m ) I n (1 - � l m� Z m) 3. 1 0
2(, + '1) n= 0

+ '1 In '1 �' (1 - �zm) In (1 - �lm�zm) J


+ unknown higher-order terms
The leading term is the same as one would obtain in the close distance limit by
pairwise adding incremental inverse-sixth-power interactions between spheres (53)
if one uses the planar result Equation 3. 1 to match the correct coefficient. In the
limit of close approach (R ;;:: al + a 2 ), this term goes as the inverse first power of

)
separation. The full expression from Hamaker's summation is

G6(R) = { (
- � a l aZ
1
+
RZ - (al + az)2 RZ - (al - a zf
1

}
3
I
3. 1 1
R Z - (al + az )Z
+ - In -=-:;,---,----=--- =_:._
2 R Z - (al - az)Z
240 PARSEGIAN

where we use

from Equations 3.1a and 3.5 with r. = O.


For R » a1 + a2 the expansion is
Annu. Rev. Biophys. Bioeng. 1973.2:221-255. Downloaded from www.annualreviews.org

3.12
by George Mason University on 09/04/13. For personal use only.

+ unknown terms
There is some question when further, unformulated, terms become important.
Smith, Mitchell & Ninham (90) computed numbers using Equations 3.10 and 3.12
for two oil drops in water and made graphical interpolation between these long- and
short-distance expansions. In this way they found significant deviations from Equation
3.11 for the temperature-dependent low-frequency contributions.
Langbein (91) has used a method of summing over sets of atoms to improve on
Hamaker by including many-atom interactions. Working under the assumption of
zero temperature as well as infinite light velocity he derived an infinite summation
series for all separations that had to be simplified before evaluation. He has also
introduced a second-order perturbation technique for treating finite light velocity
that explicitly assumes that the �im are much less than unity (92), probably a valid
assumption for high-frequency fluctuations.
Pitaevskii (93) found that the nonretarded interaction between solute molecules in

3�T6 f' [Of,s/ONJ2


dilute solution, i.e. when solute separation R greatly exceeds solute size, is

G(R) = - 3.13
811: R .=0 f,w

Here (aeJaN) is the experimentally measurable solute dielectric increment or de­


crement of the solution ; N is the number density of solute molecules ; ew is the solvent
("water") dielectric susceptibility. This formula is written to include finite tem­
perature as slightly modified by Parsegian and Ninham, who discussed its use in
computing protein-protein interactions (78). Inclusion of retardation requires
multiplying each term in the sum I'� o over frequency �n by the factor

where r. = (2';.Rew 1/2/C) from Equation 3.4 (94).


LONG-RANGE PHYSICAL FORCES IN BIOLOGY 241

F. CYLINDRICAL PARTICLE INTER ACTIONS

Only fragmentary solutions exist for the electrodynamic interaction energy


between rodlike particles of arbitrary radii, lengths, polarizabilities, and mutual
angle. Greatest progress has been made for cases where radii are very small compared
with center-to-center distance. The leading term in the attraction of two rods of
radius a skewed at mutual angle () at large separation R » a, and of length L effec­
tively infinite, is (95-97)

g(R ) = _(3:2f ��� n�� [�.L (N + � II ) + 136(�11 - �.L)2] 3.14


Annu. Rev. Biophys. Bioeng. 1973.2:221-255. Downloaded from www.annualreviews.org

per unit length for parallel rods, and

3a4 .nkT 4 I' [�.L(�.L + � II) + 1 + 216cos2 e (�.L - � 11 )2


g(R, 0) = _

J
by George Mason University on 09/04/13. For personal use only.

8 sm OR n=O

-
per interaction for skewed rods. Note the R 5 and R-4 dependence for parallel and
skewed rods. The angular dependence in g(R, 9) implies .a torque. The quantities
3.15

�.L (I:; - I:m)/(e ; + em) and �II (1:;1 - I:m)/(2I:m) depend on . transverse and
= =

parallel rod polarizabilities.


The next correction to account for slightly larger (a/R) for isotropic rods is to
multiply by [1 + (25aZ/1 6RZ)] (98). Some effects of retardation (99) and of simultaneous
attraction amorig several parallel thin rods have also been discussed (100, 101).
Parsegian (95) has pointed out that the observable birefringence and dichroism of
arrays of parallel rods can give useful information for computing the van der Waals
force between them.
Langbein (102) has applied his integration technique to treat parallel isotropic
cylindrical rods (but neglecting finite temperature). As with his analysis of sphere­
sphere attraction, the result is in the form of a lengthy summation, which must be
evaluated numerically. In the limits of wide separation or close approach the sum
can be approximated in order to be evaluated analytically.
The spatial dependence of the rod-rod attraction of isotropic rods for both very
close and distant separation seems to approach the algebraic form of a result based
on pairwise summation of incremental inverse sixth-power interactions ( 1 02). One
may use the coefficient A defined for planar systems in order to give an approximate
expression that includes the influence of finite temperature but stilI neglects retarda­
tion. We let (�im (ei - em)/(ei + em))
=

A � 12kT ", " (� 1m �2 m)q/q3


£..., £...,
n=O q= 1

(from Equations 3.1 and 3.5 with r. 0) ; then for rods whose surface-to-surface
=

separation d is much less than their radii at , az and their axis-to-axis distance

ata2 )
d + a t + az , we have ( 102)

(
R =

1 /2
ed(R) � -A
G 3.16a
32 (2d)3 R
per rod per unit length. Note the d-3/2 dependence on separation.
242 PARSEGIAN

At distances R » Q1 + Q2 the approximate energy of interaction per rod per unit


length is ( 102 103)

Ged(R ) = _ � � � r2(i + j + !) (a1) 2i (a2 ) 2i


,

3R i = 1 i = l i !j !(i - l ) !U - 1) ! R R
3.1 6 b

DeRocco and Hoover ( 104) have made r - 6 pairwise additive summations for
several rectangular systems.
Annu. Rev. Biophys. Bioeng. 1973.2:221-255. Downloaded from www.annualreviews.org

G. OTHER PROBLEMS

Some progress is being made in including charge fluctuations by ionic electrolytes


in the general theory of van der Waals forces. Preliminary arguments, still based on
by George Mason University on 09/04/13. For personal use only.

ad hoc assumptions, suggest that the zero frequency contributions will be the most
significantly modified. For planar systems, the dielectric constant Gi(O) in the L1
functions should be replaced by Gi(O)JX2 + (2K/)2 ; here Ki is the Debye constant
Equation 2.6 for the electrolyte solution and x is the variable of integration used in
Equations 3.1, 3.8, and 3.9 ( l 05). Also, the lower limit of integration in x for the zero
frequency contribution is (2Kmt) rather than zero. The double result is to increase the
magnitude of charge fluctuation and to introduce exponential screening of the
low-frequency interaction. Several effects of ionic conduction-near linear poly­
electrolytes, around point particles, in the region of an electrostatic double layer-are
reviewed and discussed elsewhere (88, 1 06).
There have beed several attempts to use the macroscopic continuum approach to
problems of particle adhesion ( 1 07, 1 08).

IV. COMPARATIVE STRENGTH AND RANGE OF ELECTROSTATIC


AND ELECTRODYNAMIC FORCES

The relative strength and range of Coulombic and van der Waals forces is most
easily considered for planar geometries where mathematical limitations are least
restrictive. Also, a great number of numerical computations have been performed to
model the attraction and repulsion between biological cell surfaces when they
flatten against each other to make strong contacts ( 109, 1 10).
Electrostatic interactions between charged bodies decrease exponentially at
distances large compared with the Debye length I/K. Under physiologic conditions,
I/K is approximately 8 A where the bathing medium is a . 14-M electrolyte solution.
The constant K increases with salt concentration and with electrolyte valence,
Equation 2.6. For large interparticle distances the electrostatic forces go as e - Kt
between planes of separation t (Equation 2. 1 1), e - KR/R for spheres (Equation 2. 1 4),
and e - KR/jR for parallel cylinders of center-to-center distance R (Equation 2.15).
Examples of planar exponential interaction between like surfaces are given in
Figures 3 and 6 for several surface fixed charge densities. Repulsive pressures are
strong but die quickly. For example, when two surfaces bearing a net value (J of one
ionic charge per 500 A 2 are brought together, the repulsive pressure in aqueous
LONG-RANGE PHYSICAL FORCES IN BIOLOGY 243

CELL MEMBRANES
,
,
,
,
,
,
, ,
, ,
, ,
\ ,
, "
, '
\
\
, '
\ \
, '
\ \
,
\
\
Annu. Rev. Biophys. Bioeng. 1973.2:221-255. Downloaded from www.annualreviews.org
by George Mason University on 09/04/13. For personal use only.

\
10-4 O\C--�---;!,:---,f;c----;�---;""-�- iA
1 DISTANCE (AI
FIGURE 6. Electrostatic repulsion energy between parallel planar surfaces bearing fixed
charge in layers of 20-A thickness (109, 1 10). Regions f are layers having the polarization
properties of 30 % or 60 % sucrose solutions ; he are hydrocarbon layers. The intervening
region is a . 14 M salt solution.

solution is (Equation 2. 1 1) 2.9 x 107 e - K( dyne/cm 2 or about .6 atm aU ' 30 A but =

only .05 atm at t =50 A ( 11K 7.8 A). =

When electrostatic forces depend on the product (8n/e)uAUB e-K( (Equation 2.11),
the energy of interaction G:�(t) between unlike bodies compared with G:�(t) and
G�W') always obeys the inequality

The decay of electrodynamic interaction energies is more closely described as


some power of the separation distance rather than exponential. Figure 7 shows
examples of the attraction between like layered structures. Differences in the curves
in this example are due to variations in composition of the two bodies.
Since the electrodynamic energy is a sum over terms of the form (A · B) (Equations
3.1 and 3.8), the interaction G��(t) between unlike bodies will satisfy an inequality

opposite to that for electrostatic interactions. This feature has been emphasized by
Jehle (52) as a cause for attractive specificity in biological systems.
244 PARSEGIAN

f he f w f he f

\
\
\
\ \
\ \
\ \
':; \ \\ \
� \ \
\
\\
li! \
\
\
\
Annu. Rev. Biophys. Bioeng. 1973.2:221-255. Downloaded from www.annualreviews.org

!:!
>- - 10-2 \ \
lil \ \
\
W \
z \
w
ci
u.i
by George Mason University on 09/04/13. For personal use only.

FIGURE 7. Electrodynamic attraction energy between parallel planar structures. Calculations


neglect effects of ionic fluctuations ( 1 09, 1 1 0).

When attraction and repulsion are combined as in Figure 8, one obtains energy
vs distance curves qualitatively resembling those of classical colloid theory. Parsegian
& Gingell ( 109, 1 10) found that the balance of electrostatic and electrodynamic
forces will cause relatively weak "secondary" minima at separations of 50 A to 80 A
for lipid membranes coated with saccharide layers. The depths of the minima are
of the order of 10- 3 to 10-4 erg/cm2•
Since the repulsive force is so sudden compared with attraction, the depth of the
energy minimum is approximately the strength of the van der Waals energy alone.
This approximation should hold also for spherical and cylindrical geometries.
It allows one to think mainly of the electrodynamic energy in deciding whether long­
range forces are strong enough to hold bodies together (next section).
At distances of the order of 10 to 50 A the positive electrostatic energy dominates.
The height of the peak is very sensitive to fixed-charge densities. At very short
distances (dotted portions) the long-range methods of computation are inapplicable.

V. CRITERIA FOR APPLICABILITY OF FORCES TO BIOLOGICAL


SYSTEMS

When does it make sense to use the theory of physical forces to explain the assembly
of molecules into functioning biological structures? This must, of course, be answered
case-by-case as we learn more experimentally and simultaneously increase theoretical
LONG-RANGE PHYSICAL FORCES IN BIOLOGY 245

I I
+ 0..1 I 'I
I I
I

I I I
I
I I ,
I
I II II I
+ 0..0.9 I
I I I
I I
I I I

+0..0.8 I I I CELL MEMBRANES


I I
I I
I
I
I I
+ 0..0.7 I
I I
I I
Annu. Rev. Biophys. Bioeng. 1973.2:221-255. Downloaded from www.annualreviews.org

I I
+0..0.6 I
OJ I
I
:E
U I
.... +0.50. I
(!) Surface Fuzz Is 30. percent sugar.
� I
by George Mason University on 09/04/13. For personal use only.

!!:! I
\
(!) + 0..40.
>-
I

w
I a 20.0. 1>.2 Per Charge
z I
w I b 4001>.2
+ 0..0.3 "
I I
\ 600A2
\
\
\
\
I
+ 0.0.2 \
I
\
\
+ 0..0.1

0.

-0..0.1

0. 10.

FIGURE 8. Net electrostatic and electrodynamic energy curves, Figures 6 and 7. Energy
minima occur between 40 A and 80 A (109, 1 10).
techniques. Still, it is worth trying to set out some guidelines for thinking abDut
fDrces in real systems.
It may help to see this question of forces and organization in the context of a
hierarchy of processes. Molecular genetics and biochemistry have taught us how
biological materials are synthesized. Simply stated, the current dogma is that the
genetic material directs the synthesis o.f pro.tein mo.lecules. Each o.f these materials
is a polymer or chain molecule whose links are chosen from an alphabet o.f subunits.
Genetic informatio.n is sequential or coded information. After synthesis, linear
proteins "kno.w" how to. fold into. three-dimensional objects to. become structural
units or catalytic enzymes. Enzymes in turn direct the further synthesis of all bio­
logical materials. These materials then arrange themselves to make bio.logical
structures as we see them.
In this hierarchy of genetic information, linear synthesis, and three-dimensional
organization there is a major conceptual leap between the biochemistry of making
246 PARSEGIAN

the materials and the cytology that observes them in the assembled state. The data
of molecular cytology are still a vast collection of weakly correlated observations.
At this level of organization the study of forces and energies is a natural means for
introducing physical logic into a formation proccss.
One must now learn enough of physical forces acting between biological materials
to relate structural observation with biochemical synthesis, to the extent that this is
an appropriate means of relation. There are still major limitations both in physical
theory and in molecular description. But we have learned enough to know when it
may be worthwhile to pursue a force-energy analysis. Here are three criteria derived
Annu. Rev. Biophys. Bioeng. 1973.2:221-255. Downloaded from www.annualreviews.org

in part from our present understanding of forces :


1 . When the size of two like bodies is much bigger than the minimum distance of
their separation, then an electrodynamic attractive energy of interaction is likely to
be strong enough to qold them together.
by George Mason University on 09/04/13. For personal use only.

2. Processes of assembly that do not require extra energy sources are likely to be
driven by physical forces.
3. When the spacing or packing of bodies changes as a function of physicochemical
variables (e.g., pH, ion concentration, temperature), the association of these bodies
may be influenced by long-range physical forces.

A. LONG-RANGE ATTRACTION : How BI G ? How SMALL ?

After making a formulation and computation o f a long-range force, how does one
know that this force can be responsible for an observed long-range ordering?
The first requirement of an energy causing ordering is that the difference in energy
between ordered and disordered states be big enough to overcome the difference in
entropic contributions to the free energy. This entropic factor will be of the order of
kT per particle or aggregate body (e.g. virus, cell). The required comparison is easy
for aqueous systems since it is natural to express the van der Waals energy as a
coefficient of k T (cf Equations 3.1, 3.8, 3. 10, 3.12-3.1 6). In general, the total attractive
energy between two like bodies increases with body size and decreases with separation.
A cursory examination of the dominant interaction for several geometries will give
some idea of the relative size-to-separation requirements for obtaining energies
greater than or comparable with Brownian energy kT.
All attractive energies are roughly proportional to S = L';;"= o Ll2, as discussed in
Section III. For nonpolar, nonconducting materials in an aqueous environment,
S is of the order of 1 ; this value is used below.
For large planar bodies the attractive energy per unit area relative to infinite
separation is Ged � -(kTjSnt2)S (Equation 3.1). Considering an interaction between
two faces of area d we require that G x d be greater than kT. In terms of distances
t and dimensions d this is (with S = 1) d > 8 nt2 For two bodies at 100 A sep­
.

aration, the area of contact must be d > 30 x (10- 2 11)2 = (0.05 11)2= (500 A)2
.

Two rods at a separation R much greater than their radius a will enjoy an attraction
(Equation 3.14)
LONG-RANGE PHYSICAL FORCES IN BIOLOGY 247

per unit length for parallel rods. Two rods of length L (taken to be much greater
than R) must have an attraction (S1) =

9 n a4 L
G 11 · L = - kT- - -
16 R4 R
with the condition G il ' L > kT. If we take a 10 A and R 50 A for two linear
= =

molecules, say, then we require L > 15,000 A 1.5 11 ! =

3n a4
Two thin perpendicular rods have an energy (Equation 3.15)
Annu. Rev. Biophys. Bioeng. 1973.2:221-255. Downloaded from www.annualreviews.org

G.l ::::: 4kT 4 S


R
When a « R this energy will not be comparable to kT.
For two parallel thick rods that almost touch (that is, R ::::: 2a), the energy per rod
by George Mason University on 09/04/13. For personal use only.

S
( )
is (Equation 3. 14a), with A= (3/2) ,

::::: 3SkT _a_ 1 / 2


G L
64 1 6d3
This will be large compared with kT when L is greater than (236/3) · d3/2jal/2• For
example, if a = 100 A and d = 30 A, L must be greater than 1400 A.
Two spheres at separations R very large compared with radii a attract as (Equation
3. 1 2)

This will always be a negligible interaction even neglecting retardation damping.

-
But when the nearest distance between two surfaces is very small compared with the
spherical radius we have (d = R' 2a)
d kT a
C se ::::: - -· S
8 d
and strong interactions will occur in the region
a
d< -
8
For example, two spherical particles of 1 -11 radius will have strong attraction (neg­
lecting retardation effects) when their closest separation d satisfies the condition
d « 0. 1 J.l = 103 A
In sum, if long-range van der Waals interactions are to hold bodies together,
their separation must be much less than their size. One will not expect structure­
forming forces between point particles, i.e. bodies whose size is negligible compared
with separation. Viruses and cells may well respond to van der Waals forces calcu­
lable (at least in principle) by the Lifshitz approach. Strong attractions between
individual molecules would have to occur at distances too close for a macroscopic
continuum picture to be accurate.
248 PARSEGIAN

B. SPONTAN EOUS LONG-RANGE ORGANIZATION IN LIPID


LAMELLAR STRUCTURES

Perhaps the strongest cases of self-organizing systems involving biological


materials occur in lipid-plus-water mixtures. The electrically neutral phospholipid
lecithin, for example, when placed in water will insistently form lipid bilayers which
in turn stack up to make "myelin figures" ; these are essentially planar lamellar
arrays alternating layers of water and hydrocarbon material. These arrays form a
separate phase existing in equilibrium with the aqueous bathing medium (1 1 1-1 14).
The lateral extent of the lamellae in this equilibrium state is indeterminately larger
Annu. Rev. Biophys. Bioeng. 1973.2:221-255. Downloaded from www.annualreviews.org

than the lipid thickness, approximately 50 A, or the water thickness, about 20 A.


The structure of such arrays may be precisely and systematically studied by X-ray
diffraction ( 1 1 4). Both theoretical and experimental evidence suggest that observed
equilibrium spacings are due to a balance of attractive and repulsive forces between
by George Mason University on 09/04/13. For personal use only.

lipid lamellae. When small amounts of electrically charged lipids are incorporated
in the lamellae, the distance between layers increases but remains finite ( 1 1 5) ; highly
charged phospholipid lamellae in pure water appear to move apart beyond limit
( 1 1 6). The spacings vary also when the polarizability of the aqueous medium is
changed by addition of sugars ( 1 1 7).
Parsegian & Gingell (109, 1 10) have computed the balance of long-range forces
as it may occur between the membrane layers surrounding biological cells. Making
minimal assumptions, they concluded that, under physiological conditions, energy
minima can very easily be deep enough to hold two cells together [as had been
suggested by Curtis ( 1 1 8)]. The forces can be cell-specific where surfaces differ in
protein or sugar or in the lipid layer [as emphasized previously by Jehle (52)].7
Aggregation of dead biological cells has been observed repeatedly. Formalin
treatment (1 19) still allows specific aggregation of sponge cells ; minced chick liver
and neural retinal cell fragments are able to stick together even preferentially to
their own tissue type (1 20). It would appear that to a significant extent physical
forces can act to hold cells together. It is similarly likely that long-range attractive
forces between lipid layers contribute to the stable envelopment of nerve axon by
myelin sheath (121).
C. VI RAL ARRA YS

In 1941, Bernal & Fankuchen (122) reviewed and described the mutual arrange­
ment of cylindrical tobacco mosaic virus (TMV) particles in acid-buffered salt water.
These particles of protein and nucleic acid some 3000-A long and 1 80 A in diameter
(122, 1 23) will spontaneously come together to form an anisotropic phase where rods
are packed in a two-dimensional hexagonal array. The lattice spacing is a strong
function of the salt concentration and pH of the medium. The salt water space
between TMV rods varies from contact to more than 1 20 A, putting it within the
purview of "long-range" interaction. A second isotropic phase containing rods at
a very much lower concentration remains in equilibrium with the ordered array.

7 For some analyses of physical forces between biological cells using older methods see
( 1 1 8a-1 1 8d).
LONG-RANGE PHYSICAL FORCES IN BIOLOGY 249

Because of the mathematical difficulties of working in cylindrical coordinates,


there is still no adequate explanation of Bernal and Fankuchen's data. The conjecture
that a balance of attractive and repulsive forces is responsible for the spacings of the
ordered phase is so strongly held, though, that several analogies have been drawn
between displacements of cylindrical protein that cause muscle contraction and the
variation in putative minimum-energy spacings of TMV rods (124-126).
Onsager ( 127) has shown that only repulsive forces are necessary for phase
separation and ordering so that electrodynamic attraction might be relatively
unimportant. But such attractions do exist. Brenner & McQuarrie (128) have com­
Annu. Rev. Biophys. Bioeng. 1973.2:221-255. Downloaded from www.annualreviews.org

puted energy minima between two rods that are somewhat less than kTper interaction.
The influence of this energy minimum on Onsager's original conclusion is not known.
The spherical Tipula Iridescent Virus in water packs in a face-centered cubic
lattice. The center-to-center distance, > 2000 A, is somewhat larger than the mean
by George Mason University on 09/04/13. For personal use only.

hydrated diameter, 1 800 A, suggesting the possibility of stabilization by long-range


forces ( 129). The existence of thin interviral filaments stabilizing the array has not
been eliminated, however (1 30).

In most cases, spontaneous assembly involves physical interactions at short


distances. Such systems pose theoretical problems outside the range of the techniques
described in this review but are examples of phenomena where physical forces
spontaneously create multimolecular structures from component particles. Several
of these phenomena have been reviewed recently by Kushner (131) and in two
symposia (1 32, 1 33).
As a class, lipid molecules aggregate when confronted with an aqueous medium
(108, 109). The nonpolar moieties merge to minimize contact with water. Some
progress has been made to explain the size, shape, and packing of aggregates in
terms of interaction free energies (23, 1 34, 1 35), but physical arguments are necessarily
ad hoc. Improved understanding of the reasons for molecular arrangement of lipids
can be valuable in predicting their organization in cellular membranes.
There is a substantial experimental literature on the formation of a whole virus
out of its constituent proteins and nucleic acids. For example, TMV viruses can be
broken down into one kind of protein and nucleic acid, then reconstituted into an
infectious whole again, all by adjusting the temperature and pH ( 1 36). The viral
structure (1 23, 1 3 7, 1 3 8) and thermodynamic changes (1 39) are known. Several of
the T-even phages have been studied by inducing genetic defects in their parts ;
whole viruses form from the spontaneous combination of healthy parts derived from
different parent strains (140).
Caspar, Klug, and collaborators have performed elegant analyses of the geometric
principles governing the juxtaposition of identical particles to form spherical and
cylindrical aggregates ( 1 23, 14 1-143). This juxtaposition assumes the existence of
attractive forces between particles.
A large number of proteins ( 144)-for example, actin ( 145), myosin (146), micro­
tubular proteins (147)-similarly aggregate under appropriate environmental con­
ditions to undergo reaction or form cellular organelles. Many of the phenomena
observed as salting-in and -out proteins from solutions have their origin in molecular
250 PARSEGIAN

interactions. It is likely that changes in the three-dimensional conformation of linearly


synthesized molecules are strongly influenced by nonbonded interactions between
parts of the same molecule ( 1 48).

VI. SOME THINGS TO DO

A. EXPERIMENT

There is an urgent need for direct measurement of long-range forces between


lipids in
Annu. Rev. Biophys. Bioeng. 1973.2:221-255. Downloaded from www.annualreviews.org

bodies in aqueous media. The lamellar lattices of the "myelin figures" of


water (Section V) are particularly suitable for measurement. One should be able to
test theoretical predictions of forces as they depend on temperature, on polarizability
of component materials, on the presence of neutral and ionic solutes. Besides
by George Mason University on 09/04/13. For personal use only.

elucidating general understanding of long-range forces, measurements on lamellar


structures would be immediately applicable to problems of contacts between cells
in tissue (109, 1 10). Similarly, force measurements in cylindrical geometry, such as
the viral arrays mentioned in the previous section or the hexagonal lattices sometimes
observed for lipid cylinders in water, are crucial to any physical explanation of
these arrangements.
Spectroscopic data, necessary for the computation of van der Waals forces, are
an essential property of biological materials. Until we know absorption spectra
and indices of refraction (or dielectric susceptibilities) over the full spectrum including
the ultraviolet region, we will be limited in applying the theoretical methods now
available.

B. THEORY

In studying the long-range ordering of bodies that are themselves the result of
aggrega ting cons ti t uent molecules, one is examining only the relatively weak but
theoretically accessible long-range consequence of forces that are much more
important in the regime of molecular contact. In order to deal with short-term
interaction, it is still found necessary to compute van der Waals interactions by
making pairwise summation assumptions that are clearly invalid at long distances
( 1 48-1 5 1 ).
A formulation for the surface or interfacial tension between two media similar
in spirit to Lifshitz' analysis for the electrodynamic interaction of two media, if
possible, would allow us to relate measurable polarizabilities with surface energies.
[The useful work of Fowkes (17) begins to consider this problem by relating inter­
facial tension to the older ideas on long-range van der Waals attraction.] In par­
ticular the role of temperature and the consequent importance of entropic changes
upon creation of a material interface is a persistent question in most problems of
assembly (1 36, 1 52). For example, the attraction between nonpo lar parts of pro tein
squeezed out from the attracting groups ( 152). Such "hydrophobic bonds" are
thought to be typical of oil/water systems and characteristically increase in strength
with temperature ( 1 52) ; yet when measured as oil/water or air/water interfacial
energies they decrease with temperature ( 1 53).
LONG-RANGE PHYSICAL FORCES IN BIOLOGY 251

The systematic elaboration of the Lifshitz approach should produce formulas


for the electrodynamic interaction of bodies of different shapes having spatially
varying polarizabilities within them. The aim is to avoid the incorrect but popular
assumption that the total interaction is the pairwise sum of forces between parts of
the two bodies. Second-order perturbation approximations for solving such problems
are tantamount to making the pairwise approximation.
With regard to Coulombic or electrostatic interactions, the underlying theory is
so approximate that great effort in solving the governing P-B equation (Equation
2.4) may be inappropriate. Nevertheless these solutions are instructive, especially in
Annu. Rev. Biophys. Bioeng. 1973.2:221-255. Downloaded from www.annualreviews.org

giving the qualitative features of force laws between electrically charged bodies of
various shapes at distances that are large compared with the characteristic Debye
lengths of the suspending medium. Very little work has been done on electrostatic
interactions at relatively short distances ( ;5 10 A in physiologic saline), where it is
by George Mason University on 09/04/13. For personal use only.

probably incorrect to neglect the discreteness of the charge-bearing particles, the


consequent distortion of the water solvent by strong electric fields near these charges,
the image forces on mobile charges near the boundaries of nonaqueous media, etc.
Far more important, though, than making extensions and improvements on
existing methods, the theoretician must if possible choose problems in biological
assembly sufficiently well-defined to allow physical analysis. The experimental
system can teach us the right approximations for developing a physical view of matter
at the m ultimolecular level. As we learn to evaluate the invisible force field around
molecules in an aqueous milieu, we will be better able to recognize the geneticaJly
determined properties causing them to act in biological structures.

ACKNOWLEDGMENT

I have benefited greatly from contact with Donald Caspar, Aharon Katchalsky,
and Shneior Lifson, who gave me an early introduction to many of the questions
considered here. Also I have enjoyed collaboration with David Gingell, George H.
Weiss, and especially Barry Ninham in working on several of these problems during
which we had many useful conversations with Andrew De Rocco, Herbert Jehle, and
Arthur Forer. I am happy to acknowledge conversations and correspondence with
N. G. van Kampen, B. R A. Nijboer, and K. Schram on formulation of the Lifshitz
theory ; and I thank S. Brenner, D. Langbein, M. Renne, B. Ninham, and P. Richmond
for sending me preprints of their work. My wife Valerie gave considerable assistance
in preparing the text.
252 PARSEGIAN

LITERATURE CITED

I . Verwey, E. J. W., Overbeek, J. Th. G. 22. Lifson, S., Katchalsky, A. 1 954. J.


1948. Theory of the Stability of Lyo­ Polym. Sci. 1 3 : 43
phobic Col/oids. Amsterdam : Elsevier 23. Parsegian, V. A. 1966. Trans. Faraday
2. Debye, P., Huckel, E. 1 923. Phys. Z. Soc. 62 : 848
24 : 1 85-305 24. Bell, G. M . , Levine, S., McCartney,
3. Katchalsky, A., Alexandrowicz, Z., L. N. 1 970. J. Call. Int. Sci. 33 : 335
Kedem, O. 1 966. Chemical Physics of 25. Landau, L., Lifshitz, E. M . 1 95 1 .
Annu. Rev. Biophys. Bioeng. 1973.2:221-255. Downloaded from www.annualreviews.org

Ionic Solutions, ed. B . E. Conway, R. Classical Theory of Fields, trans. M .


G. Barradas, Chap. XV, 295. New Hammermesh. Reading, Mass. :
York : Wiley Addison-Wesley
4. Rice, S. A., Nagasawa, M . 1 96 1 . 26. Gingell, D. 1 967. J. Theor. Bio!. 1 7 : 4 5 1
Polyelectrolyte Solutions. New York : 27. Gingel l , D . 1 968. J. Theor. Biol. 1 9 : 340
by George Mason University on 09/04/13. For personal use only.

Academic 28. Derjaguin, B. V. 1954. Discuss. Faraday


5. Oosawa, F. 1 97 1 . Polyelectrolytes. New Soc. 1 8 : 85
York : Dekker 29. Sarkies, K. W., Sammut, R. 1 973. J.
6. Derjaguin, B. V. 1 940. Trans. Faraday Call. Int. Sci. Submitted for publication
Soc. 36 : 203 30. Whittaker, E. T., Watson, G. N. 1 952.
7. Derjaguin, B. V., Landau, L. 1 94 1 . A Course ofModern Analysis. London :
A cta Physicochim. 14 : 633 Cambridge Univ. Press
8. Stillinger, F., Kirkwood, J. G. 1960. 3 1 . Parsegian, V. A., Gingell, D. 1 972.
J. Chem. Phys. 33 : 1282 Biophys. J. 1 2 : 1 192 .
9. Lifshitz, E. M. 1 955. Zh. Eksp. Tear. 32. Hill, T. L. 1 956. J. Am. Chern. Soc.
Fiz. 29 :95 ; 1 956. Sov. Phys. JETP 2 : 78 : 1 57 7 ; 1 958. J. Am. Chem. Soc.
73 80 : 3241
1 0. Rytov, S. M. 1953. Theory of Electric 33. Loeb, A. L., Overbeek, J. Th . G.,
Fluctuations and Thermal Radiation. Wiersema, P. H. 1 96 1 . The Electrical
Moscow : Academy of Sciences Press Double Layerarounda Spherical Colloid
(transI. H. Erkku ( 1 959), Air Force Particle. Cambridge, Mass. : MIT Press
Cambridge Research Center, Bedford, 34. Wall, F. T., Berkowitz, J. 1957. J.
Mass.) Chem. Phys. 26 : 1 14
1 1 . Dzyaloshinskii, I. E., Lifshitz, E. M . , 35. Honig, E. P., Mul, P. M. 1 97 1 . J.
Pitaevskii, L. P. 1 959. Advan. Phys. Coli. Int. Sci. 36 : 258
1 0 : 1 65 36. Alexandrowicz, Z., Katchalsky, A.
1 2. Derjaguin, B. V., Abrikosova, I. I., 1 963. J. Polym. Sci. 1 : 2093
Lifshitz, E. M. 1 956. Quart. Rev. 36a. Abraham-Shrauner, B. 1 973. J. Call.
(London) 1 0 : 295 Inte�race Sci. In press
1 3. Derjaguin, B. V., Ed. 1 963. Research 37. Alexandrowicz, Z., Katchalsky, A.
in Surface Forces. New York : Con­ 1 963. J. Polym. Sci. 1 : 323 1
sultants Bureau 38. Fuoss, R. M . , Katchalsky, A., Lifson,
14. Veis, Arthur, Ed. 1970. Biological S. 1 95 1 . Proc. Nat. A cad. Sci. USA
Polyelectrolytes. New York : Dekker 3 7 : 579
15. Manning, G. S., Zimm, B. H. 1965. 39. Alamov, Yu. M. 1 963. Colloid J.
J. Chem. Phys. 43 : 4250 ; Manning, G. ( USSR) 25 : 3 1 3
S. 1965. J. Chem. Phys. 43 : 4260 40. Philip, J . R., Wooding, R . A . 1970. J.
1 6. Hogg, R., Healy, T. W., Fuerstenau, Chem. Phys. 52 : 953
D. W. 1966. Trans. Faraday Soc. 41. Dresner, L. 1 963. J. Phys. Chern.
62 : 1 638 67 : 990
17. Slater, J. C., Frank, N. H. 1 947. 42. Brenner, S. L., McQuarrie, D. A. 1 973.
Electromagnetism. New York : J. Call. Int. Sci. Submitted for publi­
McGraw-Hill cation
18. Levine, S., Jones, J. E. 1969. Kolloid-Z. 43. Brenner, S. L., McQuarrie, D. A. 1973.
Z. Polym. 230: 306 Biophys. J. Submitted for publication
1 9. Parsegian, V. A. 1969. Biophys. J. 44. Casimir, H. B. G. 1948. Konkinkl.
(Suppl.) 9 : A4 1 (Abs.) Ned. Akad. Wetenschap. Proc. (Ser. B)
20. Ninham, B. W., Parsegian, V . A. 1 97 1 . 60 : 793
J. Theor. Bioi. 3 1 : 405 45. van Kampen, N. G., Nijboer, B. R. A. ,
21. Timasheff, S. N. 1970. See Ref. 14 Schram. K. 1 968. Phys. Lett. 26A : 30
LONG-RANGE PHYSICAL FORCES IN BIOLOGY 253

46. Ninham, B. W . , Parsegian, V. A . , 69. Richmond, P., Ninham, B. W. 1 97 1 . J.


Weiss, G. H . 1 970. J. Stat. Phys. 2 : 323 Low Temp Phys. 5 : 1 77
47. Abrikosov, A. A., Gorkov, L. P., 70. Richmond,P., Ninham, B. W . ,Ottewill,
Dzyaloshinskii, I. E. 1 963. Methods of R. H. 1973. J. Coli. Interface Sci.
Quantum Field Theory in Statistical Submitted for publication
Physics, trans. R. A. Silverman. Engle­ 7 1 . Haydon, D. A., Taylor, J. 1968.
wood Cliffs, N. J. : Prentice-Hall Nature (London) 2 1 7 : 739
48. Keesom, W. H. 1 9 12. Cornrnun. Phys. 72. Landau, L., Lifshitz, E. M. 1 960.
Lab., Suppl. 24-26, Univ. of Leiden, Electrodynamics of Continuous Media.
Leiden, Holland Reading, Mass. : Addison-Wesley
49. Debye, P. 1 920. Phys. Z. 2 1 : 1 78, 73. Parsegian, V. A. 1973. Advan. Coli.
22 : 302 Interface Sci. In press
50. London, F. 1930. Z. Phys. Chern. 74. Parsegian, V. A., Ninham, B. W. 1 969.
Annu. Rev. Biophys. Bioeng. 1973.2:221-255. Downloaded from www.annualreviews.org

B I I : 222, 1 936. Trans. Faraday Soc. Nature (London) 224 : 1 197


33 : 8 ; 1 942. J. Chern. Phys. 46 : 305 75. Ninham, B. W . , Parsegian, V. A. 1970.
50a. Casimir, H. G. B., Polder, D. 1 948. J. Chern. Phys. 52 : 4578
Phys. Rev. 73 : 360 76. Gingell, D . , Parsegian, V. A. 1972.
5 1 . Kauzmann, W. 1 957. Quantum Chern-' J. Theor. Bioi. 3 6 : 4 1
by George Mason University on 09/04/13. For personal use only.

istry. New York : Academic 76a. Nir, S . , Rein, R . , Weiss, L. 1 97 1 . J.


52. Jehle, H. 1 969. Ann. N Y A cad. Sci. Theor. Bioi. 36: 4 1
1 58 : Art. 1 . 240 7 7 . Fowkes. F. M . 1 967. Surfaces and
52a. Jehle, H., Yos, J. M., Bade, W. L. 1 958. Interfaces, I: Chemical and Physical
Phys. Rev . 1 10 : 793 Characteristics. ed. Burke, Reed. &
53. Hamaker, H. C. 1 937. Physica 4: 1058 Weiss. 1 97. Syracuse : Syracuse Univ.
54a. Margenau, H., Stamper, J. 1 967. Advan. Press
Quant. Chern. 3 : 1 29 78. Parscgian, V. A., Ninham, B. W. 1970.
54b. Margenau, H., Kestner, N. R. 1 969. Biophys. J. 1 0 : 664
Theory of Intermolecular Forces. New 79. Parsegian. V. A., Ninham. B. W. 1 973.
York : Pergamon J. Theor. Bioi. 38 : 1 0 1
54c. Zwanzig, R. 1 963. J. Chern. Phys. 39 : 80. Ninham, B. W., Parsegian, V. A. 1970.
225 1 J. Chern. Phys. 53 : 3398
54d. Yorke, E. D., De Rocco, A. G. 1 970. 8 1 . Langbein, D. 1 972. J. Adhes. 3 : 2 1 3
J. Chern. Phys. 53 : 764 82. Renne, M . J. 1 97 I . PhD thesis. Utrecht
55. Langbein, D. 1 969. J. Adhes. 1 : 237 83. Parsegian, V. A., Weiss, G. H. 1972.
56. Nijboer, B. R. A., Renne, M. J. 1 97 1 . J. A dhes . 3 : 259
Phys. Norv. 5 : 243 84. Kats, E. I. 1 97 1 . Sov. Phys. JETP
57. Ninham, B. W., Richmond, P. 1 97 1 . 33 : 634
J. Phys. C . (Solid State Physics) 4 : 1 988 85. Smith, E. R . , Ninham, B. W. 1 973.
58a. Langbein, D. 1 97 1 . J. Phys. Chern. Physica. Submitted for publication
Solids 32: 1 33 86. Parsegian, V. A., Weiss. G. H. 1 972.
58b. Renne, M. J., Nijboer, B. R. A. 1 967. J. Coli. Int. Sci. 40 : 35
Chern. Phys. Lett. I : 3 1 7 87. Parsegian, V. A., Ninham, B. W. 1 97 1 .
58c. Cerlach, E. 1 97 1 . Phys. Rev. B . 4 : 393 J. Coli. Interface Sci. 37 : 332
59. Ninham, B. W . , Parsegian, V. A. 1970. 88. Parsegian, V. A., Ninham, B. W. 1 973.
Biophys. J. 1 0 : 646 In preparation
60. Gingell, D. Parsegian, V. A. 1 973. J. 89. Mitchell, D. 1., Ninham, B. W. 1972.
Coli. Int. Sci. In press J. Chern. Phys. 56 : 1 1 1 7
6 1 . Derjaguin, B. V. 1 960. Sci. Am. 203 : 47 90. Smith, E. R., Mitchell, D. J., Ninham,
62. Black, W., deJongh.l. G. V., Overbeek, B. W. J. Coli. Interface Sci. Submitted
1. Th. G., Sparnaay, M. 1. 1 960. for publication
Trans. Faraday Soc. 56 : 1 597 9 1 . Langbein, D. J. Chern . Phys. Solids
63. van Silfhout, A. 1967. Proc. Kon. Ned. 32 : 1 657
Ak. Wetensch. B69 : 50 1 , 5 1 6, 532 92. Langbein, D. 1 970. Phys. Rev. B. 2: 3371
64. Tabor, D., Winterton, R. H. S. 1 969. J. 93. Pitaevskii, L. P. 1959. Zh. Eksp. Teor.
Coil. Int. Sci. 3 1 : 304 Fiz. 37 : 577 ; 1 960. Sov. Phys. JETP
65. Tabor, D., Winterton, R. H. S. 1969. 1 0 : 408
Proc. Roy. Soc. London Ser. A 3 1 2 :435 94. Parsegian, V. A. 1 973. To be published
66. Richmond, P., Ninham, B. W. J. 95. Parsegian, V. A. 1972. J. Chern. Phys.
Coli. Int. Sci. In press 56 : 4393
67. Kitchener, J. A., Prosser, A. P. 1 957. 96. Mitchell, D. J., Ninham, B. W., Smith,
Proc. Roy. Soc. London Ser. A 24 2 : 403 E. R. 1973. J. Theor. Bioi. In press
68. Anderson, C. H . , Sabisky, E. S. 1 970. 97. Mitchell, D. J., Ninham, B. W. 1 973.
Phys. Rev. Lett. 24 : 1049 J. Chern. Phys. In press
254 PARSEGIAN

98. Mitchell, D. J., Ninham, B. W., Rich­ 121 .Caspar, D. L. D., Kirschner, D. A.
mond, P. 1 973. Biophys. J. In press 1 97 1 . Nature New Bioi. 23 1 ( l a ) : 46
99. Mitchell, D. J., Ninham, 8. W., Rich­ 1 22. Bernal, J. D., Fankuchen, I. 1 94 1 . J.
mond, P. 1 973. Biophys. J. In press Gen. Physiol. 25 : I I I
100. Smith, E. R . , Mitchell, D. J., Ninham, 1 23. For review, see Caspar, D. L. D. 1963.
8. W. 1 973. J. Theor. Bioi. Submitted Advan. Protein Chern. 1 8 : 37
for publication 1 24. Shear, D. 1 968 . Physiol. Chern. Phys.
1 0 1 . Smith, E. R . , Mitchell, D. J. 1973. J. 1 : 495
Theor. Bioi. Submitted for publication 1 25 . Elliott, G. 1 968. J. Theor. Bioi. 21 : 7 1
102. Langbein, D. 1 972. Phys. kondens. 1 26. Miller, A., Woodhead-Galloway, J.
Materie 1 5 : 6 1 1 97 1 . Nature 229 : 470
103. Brenner, S . L., McQuarrie, D . A . 1 973. 1 27. Onsager, L. 1949. Ann. N Y Acad. Sci.
Annu. Rev. Biophys. Bioeng. 1973.2:221-255. Downloaded from www.annualreviews.org

Biophys. J. In press .
5 1 : 627
104. De Rocco, A. G., Hoover, W. G. 1960. 1 28. Brenner, S. L., McQuarrie, D. A. 1973.
Proc. Nat. A cad. Sci. USA 46 : 1057 . Biophys. J. In press
105. Davies, 8., Ninham, 8. W. 1 972. J. 1 29. Klug, A., Franklin, R. E . , Humphreys­
Chern. Phys. 56 : 1 2 Owen, S. P. F. 1 959. Biochem. Biophys.
Davies, 8., Ninham, B . W., Rich­
by George Mason University on 09/04/13. For personal use only.

106. Acta 32 : 203


mond, P. 1 973. J. Chern. Phys. In 1 30. Klug, A. Personal communication
press 1 3 1 . Kushner, D. J. 1969. Bacteriol. Rev.
107. van den Tempel, M. 1 972. Advan. Coli. 33 : 302
Int. Sci. 3 : 1 37 1 32. Wolstenholme, G. E. W., O'Connor,
108. Krupp, H. 1 967. Advan. Coli. Interface M . , Eds. 1 966. PrinciplesofBiomolecular
Sci. I : I I I Organization. London : J . & A . Churchill
109. Parsegian, V. A., Gingell, D. 1 972. 133. Hayashi, T., Szent-Gyorgi, A. G . , Eds.
Recent Advances in Adhesion, ed. L. 1 966. Molecular Architecture in Cell
Lee. London : Gordon & Breach. In Physiology. Englewood, Cliffs, N . J . :
press Prentice-Hall
1 10. Parsegian, V. A., Gingell, D. 1 972. 1 34. Parsegian, V. A. 1 967 . Science 156 : 939
J. Adhes. 4 : 283 1 35. Parsegian, V. A. 1 968. Membrane
III. Small, D. M., Bourges, M . 1 966. Mol. Models and the Formation of Biological
Crysl. 1 : 54 1 Membranes ed. L. Bolis, B. A. Pethica.
1 1 2. Small, D . M . 1 967. J. Lipid Res. 8 : Amsterdam : North-Holland
551 1 36.Lauffer, M. A., Stevens, C. L. 1 968 .
1 13. Luzzati, V., Reiss-Husson, F., Rivas, Advan. Virus Res. 1 3 : I
E. , Gulik-Krzywicki, T. 1 966 Ann. N Y 1 37. Barrett, A. N., et al. 1 97 1 . Cold Spring
Acad. Sci. 1 37 : 409 Harbor Symp. Quant. Bioi. 36 : 433
1 14. Reiss-Husson, F. 1 967. J. Mol. Bioi. 1 38. Klug, A., Durham, A. C. H . 1 97 1 .
25 : 363 Cold Spring Harbor Symp. Quant.
1 1 5. Gulik-Krzywicki, T., Tardieu. A., Bioi. 36 : 449
Luzzati, V. 1 969. Mol. Crysl. Liquid 1 39. Lauffer. M . A. 1 97 1 . Subunits in
Crystals 8 : 285 Biological Systems A , ed. S. N. Tima­
1 16. Rand, R. P., Luzzati, V. 1968. Biophys. sheff, G. D. Fasman, 149. New York :
J. 8 : 1 25 Dekker
1 1 7. Neveu, D., Rand, R. P. 1972. Personal 1 40. Wood, W. 8., Edgar, R. S., King, J.,
communication, to appear Lielausis, I., Henninger, M . 1 968.
1 18 . Curtis, A. S. G. 1 967. The Cell Fed. Proc. 27 : 1 160
Surface: lis Molecular Role in Morpho­ 1 4 1 . Klug, A., Finch, J. T., Leberman, R.,
genesis. London : Academic Lanley, W. 1 966. See Ref. 1 32, p. 1 58
1 I 8a. Pethica, B. A. 1 96 1 . Exp. Cell Res. 142. Caspar, D. L. D. 1 966. See Ref. 1 32,
Suppl. 8 : 1 23 p. 7
1 I 8b. Weiss, L. 1 967. The Cell Periphery, 143. Caspar, D. L. D., Klug, A. 1962.
Metastasis, and Other Contact Phenom­ Cold. Spring Harbor Symp. Quant.
ena. Amsterdam : North-Holland Bioi. 27 : I
1 1 8c. Gingell, D. 1 97 1 . In Membrane Metab­ 144. Timasheff, S. N., Fasman, G. D., Eds.
olism and Ion Transport, V ol . 3, Chap. 1 97 1 . Subunits in Biological Systems A .
II. New York : Wiley . New York : Dekker
1 1 8d. Weiss, L., Harlos, J. P. 1972. Progr. ; 145. Oosawa, F., Kasai, M. 1 97 1 . See Ref.
Surface Sci. 1 (4) : 355 144, Chap. 6
1 1 9. Moscona, A. A. 1968. Develop. Bioi. 146. Lowey, S. 1 97 1 . See Ref. 144, Chap. 5
1 8 : 250 1 47. Stephens, R. E. 1 97 1 . See Ref. 144,
1 20. Roth, S. 1968. Develop. Bioi. 1 8 : 602 Chap. 8
LONG-RANGE PHYSICAL FORCES IN BIOLOGY 255

1 48. Liquori, A. M. 1 969. Quart. Rev. 1 5 1 . Ramachandran, G. N., Sasisekharen,


Biophys. 2 : 65 V. 1 968. Advan. Protein Chern. 23 : 283
149. Katz, L., Levinthal, C. 1 972. Ann. 1 52 . Kauzmann, W . 1959. Advan. Protein
Rev. Biophys. Bioeng. 1 : 465 Chern. 1 4 : I
ISO. Waugh, D. F. 1 954. Advan. Protein 1 53 . Werst, R. c., Ed. 1 969. Handbook of
Chern. 9 : 326 Chemistry & Physics, 50th ed., p. F-30.
Cleveland, Ohio : Chemical RubberCo.
Annu. Rev. Biophys. Bioeng. 1973.2:221-255. Downloaded from www.annualreviews.org
by George Mason University on 09/04/13. For personal use only.

You might also like