Published on 01 January 1982. Downloaded by Universidad Autonoma Metropolitana-Iztapalapa on 10/10/2019 11:03:58 PM.

View Article Online / Journal Homepage / Table of Contents for this issue

Faraday Symp. Chem. Soc., 1982, 17, 109-120

Hydrophobic Moments and Protein Structure

BY DAVID EISENBERG, ROBERT c.TERWILLIGER

M. WEISS, THOMAS AND
WILLIAM
WILCOX

Institute of Molecular Biology and Department of Chemistry,

University of California, Los Angeles,
California 90024, U.S.A.

Received 9th August, 1982

The structure of a protein can be analysed in terms of what may be called the “ hydrophobic
moments ” of (1) the entire molecule and (2) of the segments of secondary structure that make up the
polypeptide chain. The zeroth moment is defined as the sum of the hydrophobicities of the amino-
acid residues of the structure under consideration; it is the analogue of the net charge of a cluster of
point charges. The first moment, or hydrophobic dipole moment, is the analogue of the electric
dipole moment of a cluster of charges. Just as the electric dipole moment measures the asymmetry
of the charge distribution, the hydrophobic dipole moment measures the amphiphilicity(asymmetry of
hydrophobicity) of the structure. A large hydrophobic dipole moment indicates that a structure is
predominantly hydrophobic on one side and predominantly hydrophilic on the other. A quadrupole
hydrophobic moment may be similarly defined. It indicates whether a protein is more hydrophobic
in its interior (as for a globular protein in aqueous solution) or at its surface (as for a membrane
protein).
The hydrophobic dipole moment is useful in two separate applications. The first, which we have
discussed elsewhere [D. Eisenberg,R. M. Weiss and T. C. Terwilliger, Nature (London), 1982,299,3711
relates the function and secondary structure of a region of protein structure to its amino-acid sequence.
For example, sequences that form surface-seekinghelices have large hydrophobic dipole moments.
The second application is in the analysis of interactions of a segment or domain of a protein with
neighbouring regions in the protein and with other parts of the environment. In this paper we ex-
amine the hydrophobic dipole moments in the known structures of nine globular proteins and find
that the moments of neighbouring segments tend to point towards each other. This suggests that
the hydrophobic dipole can serve as a simple and pictorial summary of some of the forces at work in
the folding of proteins. We have also found that the interactionsof some macromoleculeswith an
apolar-polar interface can be conveniently described in terms of the hydrophobic moments of the
molecule and a “ hydrophobic field ” which reflects the hydrophobicity of the environment.
~~ ~

Since Kauzmann’s review of 1959 it has been widely appreciated that the hydro-
phobic interaction is of importance in the folding of proteins. In more recent years
there have been several attempts to describe aspects of protein folding quantitatively
in terms of the relative hydrophobic or hydrophilic character of the various amino-acid
residues. These attempts involve assigning a numerical “ hydrophobicity ” to each
type of amino acid, and then relating these hydrophobicities in a particular protein or
fragment with some aspect of structure or function. For example, Capaldi and
Vanderkooi noted that strongly bound membrane proteins are especially rich in
highly hydrophobic amino-acid residues, and Segrest and Feldman pointed out that
trans-membrane segments of proteins have a continuous segment of residues that are
particularly hydrophobic. These studies have dealt with what we call the zeroth
hydrophobic moment; we show here that the higher moments of the hydrophobicity
are also useful in characterizing protein structure and folding.
Published on 01 January 1982. Downloaded by Universidad Autonoma Metropolitana-Iztapalapa on 10/10/2019 11:03:58 PM. View Article Online

110 H Y D R O P H O B I C MOMENTS

CONCEPTS
HYDROPHOBICITY SCALES

The hydrophobicity of an amino-acid residue is not a property that can be defined

easily or measured simply. Nevertheless, several groups have attempted to derive
numerical hydrophobicity scales using a variety of experimental and computational
methods. The resulting values generally correspond to the free energy of transfer of
the side chain of the amino acid from water to an apolar environment. This free
energy thus excludes that due to main-chain atoms in the protein structure. Among
recent comparisons of such hydrophobicity-scales are those of Meirovitch et aZ.,4Kyte
and Doolittle5 and Edsall and McKenzie.6 Because each of the methods used to
derive such a scale reflects some aspect of the hydrophobicity, and because not one of
the methods seems vastly more reliable than the others, our approach has been to take
a consensus of the values determined by some of the more reliable methods.
Our " consensus " hydrophobicity scale is given in table 1 , together with the five
scales from which it has been inferred. One is the well known scale of Nozaki and
Tanford as modified by Segrest and Feldmar~,~ based on free energies of transfer of
amino-acid side chains from water to ethanol. Another is the scale of Wolfenden et
aL8 for the transfer from water to vapour. The scales of Chothia and Janin lo are
from calculated free energy changes for the transfer of amino-acid side chains from
the surface to the interior of a protein, based on the observed distribution (between
surface and interior) of each residue type for globular proteins of known structure.
Another set of semi-empirical calculated values was published by von Heijne and
Blomberg,ll who considered the free energy changes associated with the transfer from
aqueous to non-aqueous media of the amino-acid side chains. They attempted to
include effects of buried non-polar atoms, removal of charges and decrease of hydrogen
bonding during burial.
The consensus scale was derived as follows. We noted that for each of the four
complete hydrophobicity scales on the left of table 1, the value for serine lies at the
mean, or very close to it. For each scale, the standard deviation of the hydropho-
bicities was determined. The " normalized " hydrophobicity of a residue in each
scale was defined as the number of standard deviations that its hydrophobicity lay
above or below the mean. The scales were combined by averaging the normalized
hydrophobicities for each residue over the five scales. To convert this averaged scale
to the units of kcal mol-', all hydrophobicities were arbitrarily multiplied by the
standard deviation of the scale of Janin lo and added to its mean. The result is the
final column in table 1. We used this consensus scale of hydrophobicities for all the
calculations reported in this paper.

E S T I M A T I O N O F T H E H Y D R O P H O B I C D I P O L E MOMENT O F A N A M I N O - A C I D
SEQUENCE

The hydrophobic dipole moment can be estimated from a known amino-acid

sequence if it is assumed that the polypeptide backbone follows some periodic arrange-
ment such as an alpha helix '' or a strand from a beta sheet. For this calculation we
also assumed that the hydrophobicity of each residue i can be represented by a vector
of length Hi, having a direction perpendicular to the axis of the helix or strand of beta
structure [see fig. 1 of ref. (l)]. Then the magnitude of the estimated hydrophobic
dipole moment is
Published on 01 January 1982. Downloaded by Universidad Autonoma Metropolitana-Iztapalapa on 10/10/2019 11:03:58 PM. View Article Online

D . E I S E N B E R G , R . M. W E I S S , T . C . T E R W I L L I G E R A N D W . W I L C O X 111

1 .-HYDROPHOBICITY
TABLE SCALES FOR AMINO-ACID RESIDUES

The order of residues is by decreasing hydrophobicity on the consensus scale, and the
magnitudes may be considered roughly in kcal mo1-' for transfer from a hydrophobic to a
hydrophilic phase.
scale and reference
~

consensus
residue Tanford 397 von Heijne- Janin l o Chothia Wolfenden (this
Blomberg paper)
Ile 5.0 4.4 0.7 0.24 2.15 0.73
Phe 5.0 5.2 0.5 0.0 -0.76 0.61
Val 3.0 3.9 0.6 0.09 1.99 0.54
Leu 3.5 4.2 0.5 -0.12 2.28 0.53
TrP 6.5 3.9 0.3 -0.59 -5.88 0.37
Met 2.5 2.1 0.4 -0.24 - 1.48 0.26
Ala 1.o 2.9 0.3 -0.29 1.94 0.25
GlY 0.0 1.9 0.3 -0.34 2.39 0.16
CYS 0.0 -0.08 0.9 0.0 - 1.24 0.04
TYr 4.5 3.6 -0.4 - 1.02 -6.11 0.02
Pro 1.5 1.1 -0.3 -0.90 - -0.07
Thr 0.5 1.2 -0.2 -0.71 -4.88 -0.18
Ser -0.5 0.36 -0.1 -0.75 -5.06 -0.26
His 1.o -1.5 -0.1 -0.94 - 10.3 -0.40
Glu - -4.0 -0.7 -0.90 - 10.2 -0.62
Asn - 1.5 - 1.0 -0.5 - 1.18 -9.68 -0.64
Gln - 1.0 -0.52 -0.7 - 1.53 -9.38 -0.69
ASP - 5.6 -0.6 - 1.02 - 10.9 -0.72
LYS -2.3 - 1.8 -2.05 -9.52 -1.1
Arg -9.4 - 1.4 -2.71 - 19.9 - 1.8

where 6 is the angle separating side chains along the backbone (e.g. 6 = 100" for an
alpha helix). This is a quantitative expression of the helical wheel proposed by Schiffer
and E d m ~ n d s o n . ' ~

C A L C U L A T I O N OF H Y D R O P H O B I C D I P O L E MOMENTS FROM C O O R D I N A T E S
When the three-dimensional structure of a protein is known, the hydrophobic
dipole moment can be computed from the coordinates (the " structural " moment).
We have used two different functions for this purpose.1 The first is

in which st is a unit vector pointing from the alpha carbon atom of the ith residue to the
centre of the residue's side chain. With this definition, the hydrophobic dipole
moment can be calculated for any segment of a protein structure, whether periodic or
irregular. When computed for a periodic structure, however, like pH it emphasizes
the amphiphilicity perpendicular to the axis of the helix or strand of beta structure.
A second definition of the structural hydrophobic dipole moment is

in which r f is a vector from any origin to the centre of the chain of the ith residue,
Published on 01 January 1982. Downloaded by Universidad Autonoma Metropolitana-Iztapalapa on 10/10/2019 11:03:58 PM. View Article Online

112 H Y D R O P H O B I C MOMENTS

and (Hi) is the average hydrophobicity for the amino-acid side chains in the structure.
This moment, like psi, is independent of the choice of origin. In contrast to psi, it
represents amphiphilicity in directions both parallel and perpendicular to the axis of
secondary structure.
In our analysis of the structures of folded proteins (below) we have used the
moment psi. The sum is normally taken over the residues of one segment of second-
ary structure at a time. We find that the values of pSlfor the corresponding segments
of secondary structure in closely related proteins, such as myoglobin and the alpha and
beta chains of haemoglobin, are similar in magnitude. Also we find that the moments
estimated from the amino-acid sequence (pH) are reasonably close to those calculated
from the actual structure (pJ. These points are illustrated for myoglobin and
haemoglobin in fig. 1.

EI = H b u
=Hbp
[II] = M b

B C D E F G H
helix
FIG.1 .-Comparison of hydrophobic dipole moments of the alpha helices of the myoglobin and haemo-
globin polypeptide chains. The left-hand bar of the histogram gives the magnitude of the moment
calculated from the amino-acid sequence by eqn (1). The right-hand three bars give the magnitudes
of the structure moments computed by eqn (2) for both the alpha and beta chains of haemoglobin,
and for the myoglobin chain.

HYDROPHOBIC INTERACTIONS O F SEGMENTS OF SECONDARY

S T R U C T U R E I N PROTEINS
H Y D R O P H O B I C D I P O L E MOMENTS I N F O L D E D P R O T E I N S
We have computed the hydrophobic dipole moments for each segment of secondary
structure in several proteins, and have examined their magnitudes and directions.
For each of the proteins listed in table 2, we have computed the hydrophobic dipole
moment for each alpha helix, each strand of beta sheet, and each stretch of irregular
backbone that separates any pair of regular structures or any regular segment from a
Published on 01 January 1982. Downloaded by Universidad Autonoma Metropolitana-Iztapalapa on 10/10/2019 11:03:58 PM. View Article Online

D . EISENBERG, R . M . WEISS, T . C . TERWILLIGER A N D W . WILCOX 113

TABLE
2.-cANCELLATION OF THE HYDROPHOBIC DIPOLE MOMENTS OF SEGMENTS OF SECONDARY
STRUCTURE (a HELICES, /I
STRANDS AND IRREGULAR REGIONS) IN FOLDED PROTEINS

Column 1 shows the root-mean-square sum of the lengths of the moments for each protein.
This is the expected length for moments having random directions. Column 2 shows the
length of the vector sum, the actual sum of moments.

expected length of vector sum, the

protein random directions actual kngth

melit t in 11.4 1.9

myoglobin 12.0 8.8
haemoglobin a 9.6 2.0
haemoglobin P 10.0 2.2
haemoglobin dimer (a p) + 13.8 2.6
triosephosphate isomerase 12.1 5.2
carboxypeptidase A 13.2 13.4
flavodoxin 10.1 8.6
lactate dehydrogenase 13.0 3.1
prealbumin monomer 1 7.7 0.1
rhodanese 14.6 3.0
t hermolysin 12.7 1.8

terminus. The moments are calculated from the expression above for psi. Atomic
coordinates and definitions of the segments of secondary structure were taken from
the compilation of the Brookhaven Protein Data Bank.14
A simple example of such a calculation is that for melittin l7 shown in fig. 2.
Melittin consists of four 26 residue peptides, each a bent alpha helix. The hydro-
phobic dipole moment for each helix is large, and points towards the centre of the
tetramer. This illustrates that the centre of the melittin tetramer is far more hydro-
phobic than the outside, and that the moments of the four polypeptide chains nearly
cancel when they join to form a tetramer (see also the first entry in table 2).
A similar tendency for opposition of dipole moments on neighbouring alpha
helices is observed in sperm whale myoglobin l6 (fig. 3). In this protein there are 8

FIG.2.-The melittin tetramer with the hydrophobic dipole moment (pS1) of each of the four alpha
helical polypeptide chains. As in the following stereo pairs, the origin of each hydrophobic moment
vector is placed at the centre of the corresponding segment of secondary structure, and the moment
extends in the direction of greater hydrophobicity. Moments are illustrated with lengths (in A)
numerically six times their actual magnitudes (nominally in kcal).
Published on 01 January 1982. Downloaded by Universidad Autonoma Metropolitana-Iztapalapa on 10/10/2019 11:03:58 PM. View Article Online

114 HYDROPHOBIC MOMENTS

FIG.3.-Myoglobin, with the hydrophobic dipole moments of each of the 8 helices and 7 irregular
segments. The haeme group can be seen edge-on at the top centre of the molecule; helices A and H
are labelled.

alpha helices and 7 irregular segments of polypeptide chain that separate them.
Although the pattern of the 15 hydrophobic dipole moments is more complicated than
for melittin, several points are clear: (1) moments tend to oppose moments from
neighbouring segments of secondary structure; (2) moments from turns and other
irregular regions are not negligible compared to those from helices, although the
largest moments are associated with the long helices B, G and H; (3) the net direction
of all 15 moments is toward the haeme pocket, where the hydrophobic edge of the
haeme touches the protein.
In proteins of the alpha/beta classification of Richardson,17 the beta sheet often has

(b)
FIG.4.-Hydrophobic moments of strands in two beta sheets. (a) 8 strands in carboxypeptidase A.
(b) 6 stands in lactate dehydrogenase. In each figure two strands are labelled according to the
nomenclature of the Brookhaven Protein Data Bank.
Published on 01 January 1982. Downloaded by Universidad Autonoma Metropolitana-Iztapalapa on 10/10/2019 11:03:58 PM. View Article Online

D . EISENBERG, R . M . WEISS, T . C . TERWILLIGER A N D W . W I L C O X 115

one polar side and one non-polar side, as judged by the directions of the hydrophobic
moments of the individual strands. This is illustrated in fig. 4 for the beta sheets in
carboxypeptidase A and lactate dehydrogenase. In the former sheet, six of eight
strands have their moments pointing toward the same side, and in the latter, five of six
strands have moments towards the same side.
This amphiphilicity of beta sheets is not observed in all such proteins, however.
In flavodoxin,16for example, the beta sheet does not have a uniformly polar side.
The hydrophobic dipole moments project from both side of the sheet, and are generally
opposed by moments extending from neighbouring alpha helices. Similarly, in
rhodanese the directions of the hydrophobic moments associated with the strands of
the beta sheets are not regular; moreover the magnitudes of the moments are small.
Some protein structures contain beta sheets twisted into barrels, and we have
examined the hydrophobic moments in two of these. One is prealbumin (fig. 5),
which Richardson l7 classifies as an " antiparallel beta barrel ". Of the eight

Lcc

FIG.5.-Hydrophobic moments of the 8 strands of beta sheet forming the beta barrel of prealbumin.
The region between the sheets if hydrophobic; nearly all of the moments point into this region.

strands of the barrel, seven have hydrophobic moments that point toward the interior
of the barrel, although one of the moments is small. In triose phosphate isomerase
(fig. 6), the moments of the strands forming the barrel tend to point outwards, as
though the interior is more polar than the outside. The outside is surrounded by
alpha helices whose hydrophobic dipole moments tend to point in.

INTERPRETATION OE THE DIRECTIONS OF T H E H Y D R O P H O B I C MOMENTS

I N F O L D E D GLOBULAR PROTEINS
The hydrophobic moments of segments of secondary structure at the surface of
globular proteins tend to point inwards. This is illustrated in fig. 7, which shows for
two proteins the angle between the hydrophobic moments of segments of secondary
structure and the vector from the centre of the molecule to the segment. These
angles are represented by their cosines, weighted by the magnitude of the moments,
and plotted against the distance from the centre of the segment of secondary structure
to the centre of the molecule. A negative value means that the hydrophobic moment
vector points inwards, toward the centre. In myoglobin, the hydrophobic moments of
6 out of 8 alpha helices point inwards. AIso 5 of 7 of the hydrophobic moments of
the irregular segments between helices point inwards. In triose phosphate isomerase,
the moments of most of the alpha helices point inwards. In contrast, the moments of
the beta strands (which are more nearly toward the centre of the molecule) point
outwards.
Published on 01 January 1982. Downloaded by Universidad Autonoma Metropolitana-Iztapalapa on 10/10/2019 11:03:58 PM. View Article Online

116 HYDROPHOBIC MOMENTS

The subjective impression of directions of the hydrophobic moments in folded

proteins is that they tend to cancel. Statistics presented in table 2 confirm this
impression. If the directions and magnitudes of the hydrophobic moments of seg-
ments of secondary structure were truly random, we would expect that the length of
the vector sum of the moments could be represented as the result of a random walk in
three dimensions. The total expected length of the walk can be estimated as the root-
mean-square sum of the lengths of the steps, in the case that the distribution of step

/ /
FIG.6.-Hydrophobic moments in triose phosphate isomerase. (a) The beta barrel, showing that
the moments of 6 of the 8 strands point outwards; (b) moments of both the beta and some of the
alpha segments.

sizes is Gaussian, or the number of steps is very large.15 We assume that one of these
conditions holds approximately. The expected lengths are given in the first column of
table 2, and the magnitude of the vector sum of the moments is given in the second
column. For all proteins other than carboxypeptidase A the actual length is smaller
than the expected length, and for all but three it is smaller than half the expected
length. We conclude that the hydrophobic moments of elements of secondary struc-
ture tend to oppose each other in folded globular proteins.
While we have examined hydrophobic moments in only a few proteins, several
patterns seem to be common. In clusters of alpha helices, moments of neighbouring
helices tend to oppose each other. For beta sheets sandwiched between alpha helices,
the moments of the helices tend to oppose the moments of strands in the sheets. In
some cases, but not all, the strands within the sheet have moments that project mainly
to one side of the sheet.
Published on 01 January 1982. Downloaded by Universidad Autonoma Metropolitana-Iztapalapa on 10/10/2019 11:03:58 PM. View Article Online

D . EISENBERG, R . M . WEISS, T. C. TERWILLIGER AND W . WILCOX 117

-51

-7
'j
0

0
1

-1
-2

-4

-5
i

0
0

FIG.7.-Orientation of hydrophobic dipole moments in myoglobin (top) and triose phosphate isomer-
ase (bottom), plotted as a function of the distance R of the corresponding segment of secondary struc-
ture from the centre of the protein, Negative values mean that the dipole points towards the centre
of the protein. 0 , Helix; x , &strand; +,
irregular.

I N T E R A C T I O N S O F H Y D R O P H O B I C M O M E N T S WITH T H E
ENVIRONMENT
ORIENTATION OF A HYDROPHOBIC DIPOLE AT A SURFACE
It is possible to develop a very crude quantitative description for the energy of
orientation of a hydrophobic dipole. We can think of a surface between a hydro-
phobic phase and a hydrophilic phase, such as the surface of a lipid membrane, as
being described by a hydrophobicity function, M(x), as shown in fig. 8. Because of
Published on 01 January 1982. Downloaded by Universidad Autonoma Metropolitana-Iztapalapa on 10/10/2019 11:03:58 PM. View Article Online

118 HYDROPHOBIC MOMENTS

thermal motion of lipid and water molecules at the surface, we expect that, averaged
over the time of diffusion, this surface is not sharp, but may more accurately be
thought of as a smoothly varying gradient in the hydrophobicity. This diffusionally
averaged structure is similar in conception to the D-averaged structure for a liquid.'*
The value of the hydrophobicity function M ( x ) is defined as +1 for an aqueous
environment and - 1 for an entirely hydrophobic one. As discussed above, the

.
.
.
..........
.............
.. .. .. ......... .. .. .. .. .......
...................
...........
..........................
.......................
...................
... .................. ... .......... ..........
............
...........
...........

FIG.8.-A highly schematic representation of the hydrophobicity function, M ( x ) , plotted against

position perpendicular to a membrane. M(x) is represented by the density of points: it is highest at the
centre of the membrane, and falls off smoothly towards the aqueous interface at the top and bottom
of the figure. At the top left is a surface-seeking helix, positioned so that its hydrophilic side chains
( x ) are mainly in contact with the aqueous solvent and its hydrophobic side chains (0) are mainly in
contact with the more hydrophobic region. At the right is a schematic representation of a membrane
protein with hydrophobic side chains (0) in contact with the more hydrophobic region.

" hydrophobicities " of the various amino acids are related to the free energies of

transfer of their side chains from non-polar to polar environments. Therefore the
product of the hydrophobicity of the environment M ( x ) and the hydrophobicity of an
amino-acid yields, on a relative scale, a very rough estimate of the free energy of the
amino-acid side chain in this environment. Low free energies (negative on this scale)
result from hydrophobic side chains in apolar environments, and from hydrophilic
side chains in polar environments. Thus a rough estimate for the free energy of a
fixed protein structure in an environment with hydrophobicity described by the func-
tion M ( x ) is
GE CHiM(xi) (4)
I

in which the summation is over all amino-acid side chains in the structure. This
expression neglects contributions from main-chain atoms, and assumes that all side-
Published on 01 January 1982. Downloaded by Universidad Autonoma Metropolitana-Iztapalapa on 10/10/2019 11:03:58 PM. View Article Online

D . EISENBERG, R . M . WEISS, T . C . TERWILLIGER A N D W . W I L C O X 119

chain atoms are exposed to solvent. This assumption is not a good one in general,
and it is clear that a more accurate estimate of the free energy could be obtained by
summing only over side chains at least partially exposed to solvent. Alternatively,
one could weight each amino-acid side chain by the fraction of its surface area exposed
to solvent.
If the environment of a protein structure varies slowly and smoothly from apolar to
polar, then we can rewrite eqn (4) in a simpler form. Let us denote the gradient in the
hydrophobicity function as F(x), so that
F(x)= - v M ( x ) . (5)
If the hydrophobicity varies smoothly we can write that, near a point xo,the hydro-
phobicity function is given roughly by
M(x)x M ( x ~
-) (X - xO) F(x0). (6)
Using eqn (6) and the definitions of the hydrophobic moment ps2, eqn (3) and the
hydrophobicity M(x), eqn (4) can then be rewritten as

where xois the “ centre ” of the structure, defined as the mean position of the centres
of the amino-acid side chains. This expression is analogous to that for the potential
energy of a charged object in an electric field. In a slowly varying “ hydrophobic
field,” M(x), the free energy of a structure depends on the location of the “ centre ”
of the structure (xo) as well as on its orientation (the direction of ps2). The free-
energy difference between the state in which the hydrophobic dipole moment is
oriented parallel to the hydrophobic field F(x) and the state in which they are anti-
parallel is then
AG Z - 211~s211 IIF(xo>ll. (8)
Two important assumptions used in this analysis are that (1) the solvent (water and
lipid molecules) diffuses rapidly relative to the protein molecule so that the diffusionally
averaged solvent structure is applicable and (2) the hydrophobicity function M ( x )
varies in roughly linear fashion over the region in which the protein structure is located.
Because the first assumption requires that the protein structure be not too small and
the second requires that it be not too large, this analysis will not apply for all protein
structures at a polar-apolar interface. We suspect that it is most useful in describing
the interaction of a single alpha helical protein structure (such as melittin) with a lipid-
water or air-water interface.
HYDROPHOBIC QUADRUPOLE M O M E N T S

The hydrophobic quadrupole moments are tensor quantities, analogues of the

electric quadrupole moments of a charge distribution.” Their simplest definition is
Qap = IHiriariP. (9)
1

Thus Qxx = c H i x : . A related definition which leads to values independent of the

1

choice of origin is

After diagonalization, it is clear from eqn (10) that a globular protein should have
three negative quadrupole components (since it has predominantly hydrophilic side-
Published on 01 January 1982. Downloaded by Universidad Autonoma Metropolitana-Iztapalapa on 10/10/2019 11:03:58 PM. View Article Online

f 20 H Y D R O P H O B I C MOMENTS

chains, with negative hydrophobicities, on its surface at the greatest distances from the
origin), In contrast, membrane proteins, which are expected 'O to have hydrophobic
residues inside and hydrophobic out, are likely to have at least some positive com-
ponents of the hydrophobic quadrupole. A protein which spans a membrane would
be expected to have one negative and two positive diagonal components.

We thank Walter Kauzmann for the suggestion to consider quadrupole moments,

and we thank the N.S.F. and N.I.H. for support.

W. Kauzmann, Adv. Protein Chem., 1959, 14, 1.

' R. A. Capaldi and G . Vanderkooi, Proc. Natl Acad. Sci. USA, 1972, 69, 930.
J. P. Segrest and R. J. Feldman, J . Mol. B i d , 1974, 87, 853.
H. Meirovitch, S. Rackovsky and H. A. Scheraga, Macromolecules, 1980, 13, 1398.
J. Kyte and R. F. Doolittle, J. Mol. B i d , 1982, 157, 105.
J. T. Edsall and H. A. McKenzie, Adu. Biophys., 1982, in press.
' Y.Nozaki and C. Tanford, J. B i d . Chem., 1971, 246, 2211.
* R. Wolfenden, L. Anderson, P. M. Cullis and C. C . B. Southgate, Biochemistry, 1981,20, 849.
C . Chothia, J . Mol. Biol., 1976, 105, 1 .
lo J. Janin, Nature (London), 1979, 277, 491.
l1 D. Eisenberg, R. M. Weiss and T. C. Terwilliger, Nature (London), 1982, in press.
G. von Heijne and C . Blomberg, Eur. J. Biochem., 1979, 97, 175.
l 3 M. Schiffer and A. B. Edmundson, Biophys. J . , 1967, 7, 121.
l4 F. C. Bernstein,T. F. Koetzle, G. J. B. Williams, E. F. Meyer Jr, M. D. Brice, J. R. Rodgers,
0. Kennard, T. Shimanouchi and M. Tasumi, J. Mol. B i d , 1977, 112, 535.
l5 J. S. Richardson, Adu. Protein Chem., 1981, 34, 167.
l6 J. C. Kendrew, R. E. Dickerson, B. E. Strandberg, R. G. Hart, D. R. Davies, D. C. Phillips
and V. C . Shore, Nature (London), 1960, 185, 422.
l7 T. C. Terwilliger and D. Eisenberg, J. B i d . Chem., 1982, 257, 6010; 6016.
l8 R. M. Burnett, G. D. Darling, D. S. Kendall, M. E. LeQuesne, S. G. Mayhew, W. W. Smith and
M. L. Ludwig, J. B i d . Chem., 1974, 249, 4383.
I9 M. N. Barber and B. W. Ninham, Random and Restricted Walks (Gordon and Breach, New York,
1970), pp. 5-15.
'O D. Eisenberg and W. Kauzmann, The Structure and Properties of Water (The Clarendon Press,
Oxford, 1969).
D. M. Engelman, R. Henderson, A. D. McLachlan and B. A. Wallace, Proc. Natl Acad. Sci.
USA, 1980, 77, 2023.