Download as pdf or txt
Download as pdf or txt
You are on page 1of 51

Biochirnica et Biophysica .

4cta, 415 (1975) 29-79


Elsevier Scientific Publishing Company, Amsterdam - Printed in The Netherlands

BBA 85143

SOLUBILIZATION OF MEMBRANES BY DETERGENTS

ARI HELENIUS and KAI SIMONS*


Department of Serology and Bacteriology, University of Helsinki, Haartmaninkatu 3, 00290 Helsinki 29
(Finland)
(Received October 28th, 1974)

CONTENTS

I. Scope of the review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29


II. Properties of detergents in aqueous solutions . . . . . . . . . . . . . . . . . . . 30
A. Detergents - - a special group of lipids . . . . . . . . . . . . . . . . . . . . 30
B. Detergents used in membrane studies . . . . . . . . . . . . . . . . . . . . . 34
C. Physical properties of detergent solutions . . . . . . . . . . . . . . . . . . . 35
IlL Solubilization of lipid bilayers . . . . . . . . . . . . . . . . . . . . . . . . . . 41
IV. Interaction between detergents and proteins . . . . . . . . . . . . . . . . . . . . 46
A. Binding of detergents to water-soluble proteins . . . . . . . . . . . . . . . . 47
B. Binding of detergents to membrane proteins . . . . . . . . . . . . . . . . . . 50
1. Membrane protein classification. . . . . . . . . . . . . . . . . . . . . . . 50
2. Binding of 'denaturing' detergents . . . . . . . . . . . . . . . . . . . . . 51
3. Binding of 'mild' detergents . . . . . . . . . . . . . . . . . . . . . . . . 52
V. Effects of detergents on biological membranes . . . . . . . . . . . . . . . . . . . 54
A. Effects at low detergent/membrane ratios . . . . . . . . . . . . . . . . . . . . 55
1. High affinity binding . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
2. Changes in membrane properties . . . . . . . . . . . . . . . . . . . . . . 57
B. Solubilization of membranes into lipoproteins, proteins and mixed micelles . . . . . . 59
C. Separation of lipids and proteins . . . . . . . . . . . . . . . . . . . . . . . 62
D. Effects on protein-protein interactions . . . . . . . . . . . . . . . . . . . . . 66
E. Selective solubilization of membrane components . . . . . . . . . . . . . . . . 68
VI. Detergent removal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72

I. SCOPE OF THE REVIEW

Biological m e m b r a n e s form supramolecular aggregates composed of lipids


a n d p r o t e i n s to w h i c h c a r b o h y d r a t e m a y be covalently b o u n d . To be able to study

Abbreviations: HLB, hydrophile-lipophile balance; CMC, critical micellar concentration;


ANS, l-anilino-8-naphthalene sulphonate; SF virus, Semliki Forest virus.
* Present address: European Molecular Biology Laboratory, 69 Heidelberg, Box 10.2209,
G.F.R
30

membrane structure and function it is usually necessary to dissociate the membrane


into its components. The techniques for extraction and analysis of the membrane
lipids are well worked out, and are mainly based on the use of organic solvents.
These have also been used to extract the proteins, but the resultant irreversible
aggregation and denaturation of many membrane proteins limit their usefulness.
Partial and selective protein solubilization can be obtained by a number of methods
involving chelating agents, manipulation of ionic strength and pH. Protein pertur-
bants (i.e. urea, guanidine and chaotropic agents) and enzymatic digestion have also
been used. However, these procedures do not lead to solubilization of the proteins
which are more strongly bound to the lipid matrix of the membrane. For such proteins
the use of detergents (synthetic detergents, bile salts and saponins) appears to provide
the most generally useful extraction method presently available.
The literature on the use of detergents in membrane solubilization is vast, and
scattered. Their use has largely been empirical, and few guidelines on the mech-
anisms of action have been available [1-6]. To make things even more difficult for a
reviewer, there is another large body of data on the properties of detergents in aqueous
solutions. This field constitutes a branch of science in itself, the knowledge of which
is essential for the proper use of detergents.
It is not the intention of this article to review systematically the literature on
detergents, and try to compile an exhaustive catalogue of facts and references. This
would be quite a hopeless task. Instead, we will focus our attention on the work that
deals with the mechanisms of detergent action. Useful generalisations are beginning
to emerge from recent work. From a short general survey of the properties of deter-
gents we proceed to model lipid systems and to detergent-protein interactions before
turning to the possible sequence of events when detergents interact with biological
membranes.

II. PROPERTIES OF DETERGENTS IN AQUEOUS SOLUTION

IIA. Detergents - - a special group o f lipids


Lipids form a very heterogeneous group of molecules including hydrocarbons,
pigments, cholesterol, phospholipids, glycolipids and detergents. They all contain
apolar groups of aliphatic or aromatic nature and most lipids also have polar groups.
The aromatic and aliphatic moieties are hydrophobic; they are readily soluble in
many non-polar solvents but sparingly soluble in water. The low solubility of apolar
groups in aqueous solution is mainly due to the strong interactions between water
molecules [6-9]. These have to be broken or distorted if a solute molecule is intro-
duced. If, as for a hydrocarbon, little compensating attraction occurs between water
and the solute, the water molecules closest to the solute are forced into ordered
cage-like structures. In addition, the internal mobility of hydrocarbon chains is
reduced [10]. This results in loss of entropy and is energetically unfavourable. Ex-
perimentally it has been shown that the molar free energy of transfer of hydrocarbons
31
t

from pure liquid hydrocarbon to water (which is a good quantitative measure of their
hydrophobicity) can be related to the size of the hydrocarbon, or rather to the sur-
face area of the cavity formed by the hydrocarbon in the water milieu [11]. On a
molar basis large apolar groups are therefore more hydrophobic than smaller ones.
The polar groups may be charged such as phosphate, amino, sulphate and
carboxyl groups, or neutral such as hydroxyl, carbonyl, ester or protonated carboxyl
groups. Being polar these are hydrophilic (water-loving), forming strong non-
covalent bonds with the surrounding water which are more than sufficient to compen-
sate for the loss of hydrogen bonds between the water molecules. The non-ionic
groups are weaker hydrophiles than the charged groups as their energy of interaction
with water is lower [6]. The hydrophilic groups in lipids are often called heads and
the hydrophobic groups tails, particularly when they are alkyl chains. Molecules
which are partly hydrophilic and partly hydrophobic are called amphiphiles.
Lipids differ greatly in the relative balance between their hydrophilic and hydro-
phobic moieties. This is reflected in their behaviour in water and provides a basis for
their classification (Table I) [12]. The soluble amphiphiles, which include the de-
tergents used for membrane solubilization, differ from the insoluble and the swelling
amphiphiles (which constitute the major lipid groups in biological membranes)
essentially only in having a more hydrophilic character. For instance, they have a
much higher monomer solubility in water; sodium dodecylsulphate, a typical soluble
amphiphile, has a monomer solubility of about 10-2 M [13], whereas the solubility

TABLE I
OPERATIONAL CLASSIFICATION OF LIPIDS ACCORDING TO THEIR BEHAVIOUR IN
AQUEOUS SOLUTION
From Small [12].
Class Surface behaviour Bulkbehaviour Examples
Non-polar lipids No monolayer Insoluble Hydrocarbons, cholesteryl
esters of fatty acids

Polar lipids
I. Insoluble non-swelling Stable monolayer Insoluble Triacylglycerols, diacylglyce-
amphiphiles rols, long-chain protonated
fatty acids, cholesterol
II. Insoluble swelling Stable monolayer Pure liquid Phospholipids, monoacylgly-
amphiphiles crystals cerols, glycolipids with less
(e.g. bilayers) than four carbohydrate units
IlI. Soluble amphiphiles
(A) With lyotropic Unstable Micelles above Sodium and potassium salts of
mesomorphism monlayer CMC, liquid long chain fatty acids, most
crystals at high anionic, cationic and non-
concentrations ionic detergents, lysolecithin,
gangliosides
(B) Withoutlyotropic Unstable Micelles above Bile salts, saponins and chlor-
mesomorphism CMC promazine
t,O

TABLE II

SOME SURFACTANTS OF TYPE A

Chemical name Examples of trade names


Structural
formula

Anhmic detergents
0 Sodium dodecylsulphate
VVVVV~o-~-o" .o"
o
o
Sodium dodecylsulphonate
vvvvv~-o- .o"
0 C,H3
VVVVV~c/.\~.{coo" "o" Sodium dodecyI-N-sarcosinate
0

Cationic detergents
CH 3
Cetyltrimethylammonium bromide
V V V V V V V X . . , "- c., B,-
CH:~
Tetradecylammonium bromide
VVVVVVV~.; B,"
Dodecylpyrimidinium chloride
VVVVV~.~ c,-

Ampholytic detergents
OH 0 CH3
~/~/~/~C-O CH2-CHg'CHz'O-,P-O-CHz-CH;,-N,,"CH~ Palmitoyllysolecithin
0 O" CH~,
CH3
Dodecyl-N-betaine
X/VVVVX/?'- CH~ CH~-COO-
CHa
Non-ionic surfactants

~ 0 [CHECHiO]n H Polyoxyethylene alcohol Brij series, Lubrol W, AL series


Polyoxyethylene isoalcohol Sterox A J, AP series
~ o - [c~-c.p]..
Emulphogen BC series
Renex 30 series
~L-~ O-[C.2-CH2-0], h Polyoxyethylene p-t-octyl phenol Triton X series
Igepal CA series
Nonidet P 40
V ~ O - [CH2-CHz-O]nH
Polyoxyethylene nonylphenol Triton N series
lgepal CO series
0 Surfonic N series
/ k / X / V V V V ~ ~o-[cE-c~,-o]~. Polyoxyethylene esters of fatty acids Sterox CO series
Myrj series
0 O.(C~. CHEO)vH
Span series
Polyoxyethylene sorbitol esters* Tween series
o I
~CH2-CH2O- [CH2-CH2-O]ZH Emasol series
x~y*z*w:n

* The formula shown is just one molecular type in a complex mixture of different possible structures, n = average number of ethylene oxide units per
molecule.
34

for monomers of dipalmitoyllecithin (swelling amphiphile) [14] and cholesterol


(non-swelling amphiphile) [I 5] are 10-t and 10-8 M, respectively.
The soluble amphiphiles can be divided into two groups, types A and B, as
seen in Table I. Type A includes all those amphiphiles which can form liquid crystals
(cubic, hexagonal and lamellar structures) at high concentrations, i.e. they display
lyotropic mesomorphism. The hydrophobic moieties of the type A amphiphiles
are aliphatic or arylaliphatic. Type B includes those amphiphiles which do not
display lyotropic mesomorphism, presumably because of their bulky and complicated
cyclic or aromatic hydrophobic moieties. In this review, soluble amphiphiles will be
referred to both as surfactants and detergents. The term surfactant is used synonym-
ously with soluble amphiphile, whereas the term detergent will be restricted to those
soluble amphiphiles that effectively bring about the solubilization of membrane lipids.
The distinction between type A and B soluble amphiphiles will also be preserved as
they appear to solubilize biological membranes in somewhat different ways.

liB. Detergents used in membrane studies


The number of different surfactants available is large and hundreds of them have
been used in biochemical studies. Tables II and III show the structure and some trade
names of commonly used surfactants. In much of the biochemical literature only
trade names are used which is very confusing, since different trade names may,
for instance, refer to practically identical products. For proper evaluation of work
involving surfactants the chemical names should always be stated.

TABLE Ill
SOME SURFACTANTS OF TYPE B

Structural formula Name

OH
Sodium cholate
coo No" (trihydroxy bile salt)
HO'"

OH
o
C -NH-CH -CH - S-0- Na*
Sodium taurodeoxycholate
(dihydroxy bile salt)
HO'"

Digitonin
2 GALACTOSE "" OH
GLUCOSE
L I XYLOSE J H
35

Commercial surfactants are as a rule chemically impure. They may contain


varying amounts of water and additives; one batch may differ from the next; and
after prolonged storage of liquid non-ionic surfactants the composition at the bottom
of the container may differ from that on the top. To obtain reproducible results it
is advisable to purify the surfactants used whenever possible. In the case of non-
ionic surfactants, which are difficult to purify [16], the degree of purity of the surfac-
tant used should be determined [17]. Bile salts and other ionic surfactants can
usually be purified by crystallization. It should be observed that even trace impurities
can sometimes be a nuisance in biochemical studies. Non-ionic surfactants, for
instance, may contain enough phosphorus contaminants to interfere with phospho-
lipid determinations [18].
Another problem is the heterogeneity of non-ionic surfactants. Unless frac-
tionated these have polydisperse polyoxyethylene head groups due to the purely
statistical polymerization of ethylene oxide [19]. The number of ethylene oxide
units per molecule given by the manufacturer is thus only a mean value. There is also
heterogeneity in the hydrophobic portion of synthetic surfactants since inhomo-
geneous fatty acids and alcohols are usually used in their synthesis [17,20].
To obtain a measure for the balance of size and strength of the opposing
hydrophilic and hydrophobic groups in non-ionic surfactants, Griffin has introduced
an arbitrary empirical quantity called the hydrophile-lipophile balance (HLB) [21 ]. The
HLB value helps to predict the usefulness of surfactants for particular applications.
The most hydrophobic materials have a low HLB value (1-10) and increasing HLB
value corresponds to increasing hydrophilic character. For many of the synthetic
non-ionic surfactants, the HLB can be calculated from the average structure [22].
The HLB values of some widely used non-ionic surfactants are listed in Table IV.
The membrane solubilizing power of non-ionic detergents depends on their HLB
value (see p. 61).

IIC. Physical properties of detergent solutions


When small quantities of soluble amphiphile are added to water, part of it
is dissolved as monomers and part forms a monolayer at the air/water interphase.
The molecules in the monolayer are in equilibrium with monomers in the bulk solu-
tion and each monomer concentration (or rather monomer chemical potential)
corresponds to a characteristic surface tension. When the monomer concentration
reaches a critical value, added amphiphile begins to associate to form micelles (Fig. 1).
Micelles are defined as thermodynamically stable colloidal aggregates, spontaneously
formed by amphiphiles above a narrow concentration range (the critical micellar
concentration, CMC) at temperatures above the critical micellar temperature [23].
The effect of temperature on micelle formation and the meaning of the critical
micellar temperature may best be understood from the temperature vs concentration
phase diagram of sodium dodecylsulphate in Fig. 2. At low temperatures the de-
tergent forms insoluble crystals in which the hydrocarbon region as well as the polar
groups are ordered. The monomer concentration in equilibrium with the crystals is
36

TABLE IV

HLB-VALUE A N D CMC FOR SOME NON IONIC SURFACTANTS


The HLB-data are taken from refs 329 and 330, and the CMC-values from refs 13, 31, 42. PEG,
polyoxyethyleneglycol.

Surfactant Commercial name HLB-number CMC (mM) CMC(mg/I)

Sorbitan monostearate Emasol 310 5


Sorbitan monolaurate Span 20 9
PEG (4.5) p-t-octylphenol Triton X-45 10.4 0.11
PEG (7-8) p-t-octylphenol Triton X-114 12.4 0.20
PEG (10) stearyl alcohol Brij 76 12.4 0.03
PEG (10) oleyl alcohol Brij 96 12.4 <0.04
PEG (10) cetyl alcohol Brij 56 12.9 0.002
PEG (9) p-t-octylphenol Nonidet P 40 13.1 0.29
PEG (9-10) nonyl phenol Triton N-101 13.4 0.085
PEG (9-10) p-t-octylphenol Triton X-100 13.5 0.240
PEG (11) tetradecyl alcohol Sterox 67-K 13.8
PEG (12-13) p-t-octylphenol Triton X-102 14.6 0.3-0.4
PEG (17) cetyl-stearyl alcohol Lubrol WX 14.9 0.02-0.06
PEG (20) sorbitol monostearate Tween 60 14.9 - 27
PEG (20) sorbitol monooleate Tween 80 15.0 - 13
PEG (29) oleyl alcohol Brij 98 15.3 0.025
PEG (20) sorbitol monopalmitate Tween 40 15.6 - 29
PEG (20) cetyl alcohol Brij 58 15.7 0.077
PEG (16) p-t-octylphenol Triton X-165 15.8 0.43
PEG (20) sorbitan monolaurate Tween 20 16.7 - 60
PEG (40) p-t-octylphenol Triton X-405 0.810

Monoloyer

Monomer

Micelle

Fig. 1. Schematic representation of the equilibrium of surfactant between monomeric, monolayer


and micellar forms.

below the CMC. The equilibrium monomer concentration increases, however, with
t e m p e r a t u r e a n d r e a c h e s t h e C M C level a t t h e critical m i c e l l a r t e m p e r a t u r e w h i c h is
t h u s t h e l o w e s t t e m p e r a t u r e a t w h i c h m i c e l l e s c a n f o r m [24]. T h e critical m i c e l l a r
t e m p e r a t u r e is o b s e r v e d as a s u d d e n c l e a r i n g o f t h e c l o u d y c r y s t a l l i n e s u s p e n s i o n .
T h e K r a f f t p o i n t is t h e t e m p e r a t u r e a t w h i c h c l e a r i n g o c c u r s in s o l u t i o n s w h e r e t h e
37

40

e
/CMT
Crystalline Micellar
suspension solution
E 4
v
g
ch
Krafft p o i n t ~

1~3 2'0 3'0 4'0


Temperer u re (C)
Fig. 2. Temperature-concentration phase diagram of sodium dodecylsulphate (SDS) in 0.1 M NaC1/
0.05 M sodium phosphate buffer, pH 7.4 (CMC, critical micellar concentration and CMT, critical
micellar temperature). Taken from Becker et al. [202].

concentration of amphiphile is at the CMC. For most surfactants the critical micellar
temperature and the Krafft point are synonymous but the distinction is useful
because there are cases, e.g. the lithocholates, where the critical micellar temperature
is concentration dependent [25]. The critical micellar temperature coincides roughly
with the melting of the hydrocarbon chains in the crystals [16]. The critical micellar
temperature is very sensitive to impurities. This explains the range of values (10-23 C)
published for sodium dodecylsulphate [26,27]. The critical micellar temperatures of
non-ionic surfactants and the common bile salts are below 0 C [16,25].
The driving force for the spontaneous aggregation of surfactant molecules
to form micelles is hydrophobic. The interior of the micelle consists of the hydro-
phobic groups. They are sequestered from the water by the polar groups which
cover the surface of the micelle. When the hydrophobic groups are hydrocarbon
chains (as in most type A soluble amphiphiles) the micellar interior is in a fluid
liquid-like state approaching that of liquid hydrocarbon [28-30]. Thermodynamic-
ally, micelle formation may therefore be treated as a phase separation phenomenon.
The free energy of micelle formation, i.e. the free energy of transferring one surfactant
molecule from water to a micelle, can be expressed as
AG = RT'InCMC (l)

where it is practical to give the CMC in mole fraction units. However, the equation
should be corrected for the free energy required to mix the micelles formed with the
solvent, which yields (Eqn 7-3, ref. 6)

AG = RT'lnXmo. -- ~ . ln (2)

where N is the average number of amphiphile molecules per micelle (the aggregation
38

1.5
E AmphiDhfle present
oE as m o n o m e r
/
cmc / .
__V'" I / / .
o
c c:~
01 /
/;;onome /
conc : c m / ~ / A
:p,~-
uu~
~_ ~ 0 . 5
/ Arnphiphile
g~
u / pPesent in
r" ~ cr~c/" micellar
U
o Ol
05 1.0 15 2.0
Total amphiphile c o n c e n t r a t i o n

Fig. 3. The monomeric and micellar surfactant concentrations in relation to total surfactant con-
centration (based on Eqn 2 with ffz - 50). The dashed lines show empirical procedures for deter-
mining the CMC. The concentration units are arbitary. Taken from Tanford [6].

number), A1mic the mole fraction of amphiphile in the micelles and Xmo n the equili-
brium monomer concentration. Since ~ is large for most amphiphiles of type A
(50-150) the second term remains small, and X .... does not increase much beyond
the CMC with increasing micellar concentration. In practice, the free monomer
solubility of these surfactants can therefore be taken as equal to the CMC.
This is seen in Fig. 3 where the monomer concentration in equilibrium with micelles
( ~ = 50) is shown as a function of the total surfactant concentration. For bile salts
and other surfactants with low aggregation numbers, the monomer concentration
increases, however, significantly above the CMC. It should be pointed out that in
the above simplified treatment the acitivity coefficients are taken as unity and the
effect of counter ions neglected. For recent reviews on the thermodynamics of
micelle formation and of the properties of surfactant molecules in aqueous solution
see refs. 6,13,16 and 31-38.
CMC values are usually determined by extrapolation as shown by the dotted
lines in Fig. 3, but in reality the C M C is not a unique concentration but rather a
narrow concentration range [13]. Numerous CMC determination methods have
been used, many of which can be performed in laboratories equipped for biochemical
work. The C M C values reported in the literature for 720 surfactants have recently
been collected by Mukerjee and Mysels [13]. Tables IV and V give the CMC values
of some ionic and non-ionic surfactants. In line with what is known about the hydro-
phobic effect, the CMC decreases as the hydrophobic character of the tail increases
(see refs 16 and 32-39). Introduction of double bonds or branching points increases
the CMC. So do additives known to break up water structure such as urea.
The hydrophilic groups oppose micelle formation [16,36,39]. Because of their
charged head groups, ionic surfactants often have CMC values about 100 times higher
than those of their non-ionic analogues. Their CMC is reduced by increased counter
ion concentration. This is due mainly to the reduction of electrostatic repulsion
39

TABLE V
MICELLAR WEIGHTS, AGGREGATION NUMBERS AND CMC FOR SOME
SURFACTANTSa
Surfactant Aggregation Micellar CMC (raM) Conditions References
number weight

Sodium dodecylsulphate 62 18 000 8.2 H20 13, 331


126 36 000 0.52 0.5 M NaCI 13, 131
Sodium tetradecylsulphate 138 44000 0.1 MNaCl 332
2.1 H20 13
Sodium dodecylsulphonate 54 15 000 9.8 H20 13, 332
Tetradecyltrimethyl 64 19 000 4.5 H20 13, 333
ammonium chloride
Cetyltrimethylammonium 169 62 000 0.013 M KBr 334
bromide
0.92 H20 13
Lysolecithin (egg) 181 92 000 (0.01-0.001~w/w) H20 58
Triton X-100 140 90 000 0.240 H20 68
PEG (10) nonyl phenolb 100 66 000 0.075 H20 31
PEG (10) tridecyl alcohol 88 56 000 0.125 H20 31
PEG (20) cetyl alcohol 70c 82 000c 0.077 H20 31
PEG (14) stearyl alcohol 370 330 000 0.060 H20 31
Digitonin 60 70 000 186
Sodium cholate 2-4 900-1800 13-15 HzO d 49, 52, 69
Sodium taurocholate 4 2200 10-15 H20 d 49, 52, 69
Sodium deoxycholate 4--10 1 7 0 0 - 4 2 0 04-6 H20 a 49, 52, 69
Sodium taurodeoxycholate 5 2000 2-6 H20 d 49, 52, 69

a Measured at room temperature.


b PEG, polyoxyethyleneglycol.
c Values obtained for derivativeswith homogeneous PEG-chain length(the other values for non-ionic
surfactants in the table are for polydisperse materials).
a pHi8.

between the head groups allowing association at a lower m o n o m e r c o n c e n t r a t i o n


[6,36]. The C M C of n o n - i o n i c surfactants also decreases somewhat with increasing
c o n c e n t r a t i o n s of salts depending on the lyotropic character of the ions [40-42].
The effect of temperature between 0 and 37 C on the C M C of ionic surfactants is
fairly small, but there is significant lowering of the C M C of n o n - i o n i c surfactants
with increasing t e m p e r a t u r e [36,43].
For most practical purposes micelles can be considered h o m o g e n e o u s in size
a l t h o u g h in fact their molecular weights are distributed a r o u n d a mean [6]. The
mean micellar size a n d shape depend on the characteristics of the amphiphile
molecule a n d the conditions, but in a less predictable way than the C M C [6]. In
general, low C M C values of the n o n - i o n i c surfactants correspond to high micellar
weights whereas the high C M C values of ionic surfactants correspond to low micellar
weights [31 ]. The aggregation n u m b e r a n d micellar weights of some surfactants are
seen in Table V.
The restricting forces in micellar growth depend m a i n l y on (1) the electrostatic
repulsion between head groups, (2) head group hydration a n d configuration and
40

(3) the amphiphile concentration (see refs 6,16,32,35,39 and 44). An increase in
counter ion concentration, surfactant concentration and, with non-ionic surfactants,
higher temperatures lead to an increase in micellar size. At a characteristic tempera-
tures (called the cloud point), solutions of non-ionic surfactants become turbid and
a phase rich in surfactant eventually separates out of solution [36,45]. The cloud
point effect is caused by a decrease in head group hydration. The cloud point of
Triton X-100 in water is 64C, but is lowered by higher ionic strength and by addi-
tives [46].
The physical properties of the type A surfactant micelles present at low sur-
factant concentrations suggest a nearly spherical shape (see ref. 6). Calculations,
which take into consideration the surface areas requirements of the head groups and
the volume and length of the hydrophobic groups, indicate that most micelles
must in fact be somewhat ellipsoidal in order to be able to accommodate the
number of molecules that they contain [6,47,48 ].
The micellar properties of type B soluble amphiphiles, especially the bile salts,
differ in many respects from those of type A. The physical properties of bile salts
have been studied in many laboratories, because of their physiological role in lipid
absorption in the gut [12,25,49,50-52]. In the bile salts, the polar groups are distrib-
uted in different parts of the molecule and there is no well-defined 'head group'
(Table III). The hydroxyl groups are, however, all on one side of the rigid cyclo-
pentheno-phenantrene ring structure and the terminal ionic group is situated at the
end of a short flexible branched aliphatic chain. The bean shaped molecule thus
possesses a polar and an apolar face, and consequently orientates itself 'fiat' on air/
water interphases. In water above a critical concentration, bile salts form small
aggregates (from dimers to octamers) in which the molecules lie back to back (Fig. 4).
At higher counter ion concentrations larger aggregates may form. Small and collab-

PRI MARY SECON DARY

section

Hydr'ogen
Cross bonding
section

,88
AggregGtion 2-10 12-100
number
Fig. 4. Proposed structures for the "primary" and "secondary" bile salt micelles. Taken from Carey
and Small [25].
41

orators [25] call these 'secondary micelles', in contrast to the smaller 'primary micelles'.
The 'secondary micelles' are probably made up of 'primary micelles' bound to each
other by polar interactions forming relatively globular clusters. Table V shows the
concentrations at which aggregation starts (we call it CMC for convenience, see ref.
49) for different di- and trihydroxy bile salts. Both the CMC and the aggregation
number of trihydroxy bile salts are resistant to the counter ion concentration whereas
with dihydroxy bile salts the aggregation number increases and CMC decreases with
increasing counter ion concentration [25].
The properties of bile salts which have carboxyl groups are very dependent on
pH [25,49,52]. When the pH is lowered to values approaching the pKa, bile acid is
formed, which is insoluble in water. The acid can be solubilized to some extent by the
bile salt micelles present but when these are saturated, the acid starts to precipitate.
Sodium cholate precipitates at pH 6.5 and sodium deoxycholate at 6.9. In the case
of deoxycholate, the occurrence of the acid form is coupled to a dramatic increase
in the micellar size. At a pH just above the precipitation limit deoxycholate forms a
gel. In membrane work it is advisable to use slightly alkaline pH or conjugated bile
salts. The conjugated bile salts are not precipitable at a neutral pH but in other re-
spects resemble their unconjugated counterparts. The pH has little effect on the
micellar properties of non-ionic surfactants or ionic surfactants with strongly acidic
or basic groups.

III. SOLUBILIZATION OF LIPID BILAYERS

Lipid bilayers and liposomes are frequently used as experimental models for
membranes [53]. The results from these model systems have proved useful for the
understanding of phenomena occurring in the more complex biological membranes.
Similarly, use can be made of the interactions known to occur between detergents
and membrane lipids in aqueous solution to facilitate the interpretation of detergent
effects on membranes.
When detergent is added to a suspension of phospholipid liposomes, part of it
interacts with the bilayer lipids and part of it remains free in solution. The available
data on this partition (taurocholate and lecithin [25], sodium dodecylsulphate and
lecithin [54] suggest that the free concentration remains below the CMC of the pure
detergent and decreases with decreasing detergent/lipid ratio (see however ref. 55).
A general understanding of the fundamental properties of such partitioning can be
obtained from work with simple mixtures of two surfactants. Shinoda et al. [16],
Becher [31] and Tanford [6] have reviewed the literature on mixtures of two sur-
factants with different head groups, or with different tails. The results can be sum-
marized as follows: (I) The CMC and the aggregation number change gradually
from that of one component to the other as the ratio of the surfactants changes.
(2) The monomer concentration of each of the two components in equilibrium with
the mixed micelle is always lower than the CMC of the pure component. (3) The
42

component with a higher CMC in the pure state is enriched in the monomeric phase
compared to its mole fraction in the micellar phase. All this makes sense if mixed
micelles and the corresponding equilibrium monomer concentrations are considered
analogous to liquid mixtures and their vapour-pressure equilibria. Assuming thus
that the micelles form a separate phase in which there is ideal mixing, we can employ
Raoult's law

X'i,w ~= Xi, mic "CMCi (3)


where Xumic is the mole fraction of amphiphile i in the mixed micelle, Xuw its mole
fraction in monomeric solution and CMC~the CMC of the pure amphiphile (in mole
fraction units). Although the assumptions involved (the liquid-like nature of the
mixed phase and ideal mixing of components) may be unjustified for most complex
mixtures of amphiphiles, these thermodynamic considerations provide a basis for
understanding surfactant interactions with membrane lipids [6]. They are also
important when working with surfactant mixtures in membrane biochemistry.
The thermodynamic considerations thus suggest that the free detergent con-
centration in detergent/phospholipid mixtures remains below the CMC of the pure
detergent and decreases with the detergent/phospholipid ratio. What then is the
nature of the phospholipid-detergent complexes in which the bound detergent
participates? Whether they are detergent micelles containing phospholipids, phos-
pholipid bilayers containing detergent, other mixed structures or mixtures of different
complexes depends in each case on the ratio of the components. This is illustrated
by the available three component phase diagrams (Figs 5 and 6). The diagram in
Fig. 5 by Small [56] shows the phases occurring in all possible combinations
of egg lysolecithin (type A soluble amphiphile), egg lecithin (swelling amphiphile)
and water. The structures of the lamellar, hexagonal and micellar phases

/I / / Hexogem~
I
~ / t~~ z/ LYSOLEGITHIN
~!~/,, A
~'" ~ / "1"

a ~ ~ 52"G
/ - v / - " , / \ t/ \

Fig. 5. Lecithin/lysolecithin/waterternary phase diagram at 52C. The structures of the lamellar,


hexagonal and micellar phases are indicated by the insets. The lecithin molecules have been drawn
in darker characters. Taken from Small [56].
43

NOCHOLAT~

rling (TT)

a ('in"B)

LECITHIN

He~/
/ ,"

Fig. 6. Lecithin/sodium cholate/water ternary phase diagram. Hypothetical structures for the lamel-
lar, hexagonal and micellar phases are shown. The structure of the cubic phase (VI) is not known.
Figure taken from Small [56]. For a more detailed phase diagram see Small et al. [59].

are indicated in the insets. The corners represent the pure substances while the
sides of the triangle show all the possible binary mixtures of the components.
Within the triangle each point represents a unique mixture of the three com-
ponents. The area in the left corner of the figure which contains information
on mixtures with high water content is most relevant to liposome solubilization.
The amount of lysolecithin that has to be added to a dilute suspension of lecithin in
water in order to obtain a clear isotropic mixed micellar solution can be derived
from this region of the diagram: about I0 molecules of lysolecithin per lecithin are
needed. The results of Robinson [57] and Saunders [58] suggest that highly asym-
metrical micelles with molecular weights of about 1.5 l06 are already formed when
"

there are 2 mol of lysolecithin present per mol of egg lecithin. Substances like
lecithin which themselves are insoluble or sparingly soluble in water are said to be
solubilized when the addition of soluble amphiphile results in the formation of a
thermodynamically stable isotropic solution.
The phase diagram in Fig. 6 (sodium cholate/egg lecithin/water) by Small
et al. [59] is for a type B soluble amphiphile. As before there are lamellar, hexagonal
and isotropic micellar phases and also a small area with a cubic phase (V.I.). The area of
the isotropic micellar phase is much larger than that seen in the lysolecithin/lecithin/
water diagram, which illustrates the great capacity of bile salts to solubilize lecithin
and other swelling amphiphiles. About 2 mol of lecithin can be solubilized by 1 mol
of cholate. Hypothetical structures for the lamellar, hexagonal and micellar phases
44

are shown in the insets of Fig. 6. The bile salt molecules found in the lamellar phase
probably occur as dimers, trimers and tetramers with hydroxyl groups turned towards
each other and the hydrophobic surface of the molecules turned outwards (Fig. 6).
Unfortunately there are as yet no phase diagrams of systems including phospholipids
and other detergents used in membrane solubilization. Qualitatively, however,
such diagrams may be expected to resemble the diagrams presented. Dennis
and Owens [60] have studied the solubilization of egg lecithin and dipalmitoyl-
lecithin by Triton X-100 in 2H20 at 37C. Their N M R results indicate that
2 mol of Triton X-100 per tool of lecithin are required for solubilization, Unlike
the solubilization of egg lecithin, dipalmitoyllectithin solubilization was
temperature dependent; more than 20 tool Triton X-100 was needed at 20C,
which is below the melting temperature of the hydrocarbons of this lipid [61].
Only 0.5 mol of Triton X-100 is required to solubilize 1 mol of brain sphingomyelin
in water solution [62]. Preliminary results from our laboratory show that about
2 mol of sodium dodecylsulphate per mol of egg lecithin suffice to clear up egg leci-
thin/water suspensions. The solubilization limits are by no means absolute since
solubilization like micelle formation depends on many factors such as ionic strength,
pH etc.
The diagrams in Figs 5 and 6 and the solubilization data give little information
on the phases occurring at detergent/phospholipid ratios lower than needed to obtain
complete phase transition from the lamellar to the isotropic micellar phase. There
is, however, reason to believe that when small amounts of detergent are added to
liposomes some of it will be incorporated into bitayers without disrupting them.
This evidence comes from permeability studies on black lipid films in the presence
of detergents [63,64] and from the binding of fluorescent probes [65] and anaesthetics
[66] (both are soluble amphiphiles) to liposomes. That the lamellar phase of swelling
amphiphiles can accomodate surfactants to some degree is illustrated by the lamellar
phases depicted in Figs 5 and 6 and it is also indirectly supported by the general
features of a number of ternary amphiphile/water systems that have been studied in
detail and recently reviewed by Ekwall [71 ]. When the bilayers become saturated with
detergent additional detergent will induce the formation of mixed micelles. These
will have the highest possible phospholipid content (i.e. they will be saturated with
phospholipid). At intermediate detergent/lipid ratios bilayers saturated with deter-
gents and micelles saturated with phospholipid will coexist in different proportions.
When enough detergent is added to reach the phase transition limit all the phos-
pholipid will be converted to the mixed micellar form. Additional detergent will
then cause a gradual increase in the detergent/phospholipid ratio in the mixed
micelles.
Mixed micelles differ greatly in size and structure depending on the substance
solubilized (the solubilizate), the detergent, and the detergent/solubilizate ratio.
Thus lysolecithin micelles, which in the absence of lecithin contain about 180 mo-
nomers, grow to contain about 2600 lysolecithin molecules and 280 lecithin molecules
in 10:1 lysolecithin/lecithin mixtures [67]. The aggregation number of pure Triton
45
o o o o o o
48A 48A 56A 56A 40A~ 90A

a b c

tTritnX-100 ! Sphingomyelin
Fig. 7. Proposed structures for Triton X-100 micelles and brain sphingomyelin-Triton X-100 mixed
miclles in a cut-away representation. (a) Pure Triton X-I00 mice]le 034 molecules of Triton X-100
per micelle). (b) Small mixed micelle (192 molecules of Triton and 50 molecules of sphingomyelin).
(c) Large mixed micelle (210 molecules of Triton X-100 and 442 molecules of sphingomyelin).
Taken from Yedgar et al. [62].

X-100 is about 140 (Table V) [68]. Mixed micelles saturated with sphingomyelin
contain, according to recent studies by Yedgar et al. [62], 196 mol of Triton and 442
mol of spingomyelin. There is also a minimal size to the Triton/sphingomyelin
mixed micelles, 196 molecules of Triton and 50 molecules of sphingomyelin. These
small-sized mixed micelles occur together with pure Triton micelles when there are
more than four Triton molecules per sphingomyelin present. The fact that the
number of Triton molecules is constant (~200) in all the mixed micelles indicates,
according to Yedgar et al., that each Triton X-100 molecule contributes a fixed
measure of curvature regardless of how many sphingomyelins are present in the
micelle. The role of Triton in solubilizing bilayers may thus be understood on a
geometrical basis; they act by "overcoming the low surface curvature of the bilayer
so that particles can be formed with sufficient curvature to close upon themselves
within a small radius". Models for sphingomyelin/Triton mixed micelles are seen in
Fig. 7. The pure Triton X-100 is pictured as being roughly spherical [62] (Fig. 7a)
and so is the smallest possible mixed micelle (Fig. 7b). The saturated mixed micelle
is pictured as an oblate ellipsoid with Triton X-100 enriched in the parts with higher
curvature (Fig. 7c).
Bile salts have an enormous capacity to solubilize swelling amphiphiles as seen
in Fig. 6, but they solubilize cholesterol and other large apolar molecules ineffi-
ciently (for review, see ref. 69). The presence of swelling amphiphiles, however,
enhances the cholesterol solubility [69,70]. The molecular structure of the bile salts
makes it difficult to consider a curvature-enhancing role for them similar to that pro-
posed for the type A soluble amphiphiles in the mixed micellar structure. Dervichian,
Small and co-workers have proposed a hypothetical structure for the bile salt/
lecithin mixed micelles (Fig. 8) in which a cylinder of bilamellar lecithin is surrounded
on its perimeter by bile salt molecules [72,73]. The size of the mixed micelles de-
creases progressively with decreasing lecithin/bile salt ratio until the minimum size is
reached after which mixed micelles and pure bile salt micelles will occur simultane-
ously.
46

Longitudinal
section

Cross section

Fig. 8. Schematic view of the proposed model for lecithin/sodium cholate mixed micelles. On the
left are small micelles containing a large proportion of bile salt to lecithin, on the right larger micelles
with less bile salt and more phospholipid. Taken from Small et al. [72].

In summary, we suggest the following sequence of events as increasing amounts


of detergent are added to a suspension of liposomes prepared from pure phospho-
lipids:
Stage 1: Detergent binding. Detergent is incorporated into the bilayer and
causes changes in its physical properties.
Stage II: Lamellar-micellar phase transition. When the bilayers are saturated
with detergent, mixed micelles begin to form resulting eventually in complete phase
transition.
Stage Ill: Size decrease of mixed micelles, After completed phase transition
the detergent/phospholipid ratio in the mixed micelles increases and their size de-
creases.
The effect of type A detergents may be pictured as resulting from their 'wedge'
shape [74] which causes a pressure for increasing curvature in the bilayer thereby
inducing the formation of smaller mixed structures. The bile salts, on the other
hand, may chop up the bilayers into small disc-like pieces where the bile salt molecules
cover the hydrophobic edges.
The above treatment concerns systems in equilibrium and the kinetic aspects
have not been considered. Micelle formation and dissociation are fast reactions with
rate constants higher than 10 s -1 [75]. Solubilization in dilute aqueous solutions also
seems to occur quite rapidly. In more concentrated mixed systems the time to reach
equilibrium is longer. An interesting and important question is how fast surfactant
molecules move (flip-flop) through bilayers. Black lipid membrane experiments with
dodecylsulphate and Triton X-100 seem to indicate that the movement is fairly
slow [63,64]. More data on this aspect are, however, needed.

IV. INTERACTIONS BETWEEN DETERGENTS AND PROTEINS

Most data on the interaction between detergents and proteins have been ob-
tained with water-soluble proteins, mainly with serum albumin. These studies provide
47

the basis for our present understanding of detergent action on proteins (for reviews,
see refs. 6 and 76).

IVA. Binding of detergents to water-soluble proteins


The binding of detergents to proteins follows the usual thermodynamic laws
of equilibrium. Binding has been found to occur to discrete binding sites (at least at
low detergent concentrations) and is a function of the free detergent concentration in
equilibrium with the protein. It is influenced by temperature, pH, ionic strength
and other environmental factors, which should be controlled. A general discussion
of the methods employed to determine binding is provided by Steinhardt and Rey-
nolds [76]. Analysis of the binding data is usually performed as presented by
Scatchard [77] and Klotz [78]. In this way, binding sites with different levels of
affinity can be revealed and analyzed.
It has been demonstrated for a number of detergents that it is the monomeric
species that is bound and not the micellar form, when both ligand and protein are
present at relatively low concentrations [20,79,80]. It is therefore the free monomeric
concentration of the detergent that determines the amount bound to the protein.
Little further binding is observed upon increasing the free detergent concentration
beyond the CMC. This also means that the binding of detergents to proteins has to
compete with the self-association of detergent molecules to miceiles. This competitive
feature of the binding process effectively limits the acquisition of analyzable data
to situations where the free energy gain associated with binding to the protein
exceeds the free energy gain in miceIle formation (for a more detailed discussion,
see ref. 6). The binding of protein molecules to pre-existing detergent micelles may,
however, be possible in some instances. Preliminary data cited by Tanford et al.
[81,82] indicate that with membrane proteins and lipoproteins, cooperative incre-
ments can be observed above the CMC, and such processes may be accompanied by
changes in the state of aggregation of the protein as well. Also pancreatic colipase
binds to bile salt micelles [83].
The studies of detergent binding to serum albumin have shown that there are
different levels of affinity characterizing the association [20,80,84-90]. The native
protein has a small number of binding sites (~< 10) with high affinity for all detergents
studied (Table VI). Anionic long-chain amphiphiles like sodium dodecylsulphate are
bound in greater number and with higher affinity than cationic and non-ionic
detergents, and bile salts (Table VI). This is in conformity with the major biological
role of this protein, which is to function as a carrier for fatty acid anions in the
circulatory system. The high affinity binding sites on serum albumin are generally
visualized as hydrophobic patches or crevices on the surface of the protein [6].
Apart from serum albumin, few other native proteins appear to have such
sites. (The only known exception is [3-1actoglobulin [91-95]). Hydrophobic surface
crevices or patches large enough to bind detergents with high affinity are apparently
not present on most water-soluble proteins.
When the free concentration of anionic and cationic detergents is increased
48

TABLE VI
BINDING OF DETERGENTS TO ALBUMIN" IN RELATION TO CMC

Concentration (M)
Sodium - ~4~4 Me3 "i'riion X--100~- Deoxycholate "
dodecylsulphate
50~ saturation of high
affinity sites of native I " 10-6 5 10-5 5 [0-5 1.5' 10--~
albumin (10 sites: (4 sites) (4 sites) (4 sites)
Critical concentration 3 10-4
" 3 " 10-3 Not observed
for massive cooperative
binding
CMC b l " 10 -3 4' 10 -3 3' 10-4 3" 10-3

Data for dodecylsulphate, Triton X-100 and deoxycholate from ref. 20 and for tetradecyltrimethyl
ammonium chloride (C14NMe3), from ref. 80.
b CMC values determined in the same buffer as used for the binding experiments.
The number of high affinity sites is shown in parentheses.

above that required to saturate the high affinity sites on serum albumin, binding to
other sites occurs [6,76]. This binding is cooperative, and is accompanied by a con-
formational change of the protein in which presumably many previously buried
hydrophobic groups become exposed.
For sodium dodecylsulphate, the cooperative mode of association is c o m m o n
virtually to all proteins, and the m a x i m u m a m o u n t o f dodecylsulphate that is bound
per g of protein is the same for most of them [79,96]. Multi-chain proteins (if re-
duced) are usually dissociated into their constituent polypeptide chains during the
cooperative binding process [97-100]. For reduced polypeptides, there are actually
two types of dodecylsulphate-protein complexes, one containing about 0.4 g dode-
cylsulphate per g protein which is formed at a free dodecylsulphate concentration
between 5"10 -4 and 8 ' I 0 - 4 M , and another which is saturated at about 1.4 g dodecyl-
sulphate per g protein and which is observed at free dodecylsulphate concentrations
above 8" 10-4 M [79]. The sodium dodecylsulphate-polypeptide complexes form
extended rod-like particles, the length of which is roughly proportional to the mole-
cular weight of the polypeptide [101 ].
It should be observed, however, that proteins differ in their intrinsic stability
towards dodecy[sulphate denaturation. At room temperature, some proteins,
including pepsin, papain and glucose oxidase (if unreduced) do not bind sodium
dodecylsulphate in a cooperative fashion with accompanying unfolding even if the
dodecylsulphate concentration is increased to the C M C (or above) [100]. Many
viral protein capsids (stabilized mainly by protein-protein interactions) share this
resistance towards sodium dodecylsulphate [102]. The massive cooperative mode
of binding can be induced for all these proteins by heating in the presence of dode-
cylsulphate.
49

Nozaki et al. [80] have shown that tetradecyltrimethylammonium chloride


forms rod-like complexes with a number of proteins similar to those formed by
sodium dodecylsulphate. However, the cooperative process occurs close to the
CMC of this cationic detergent and the concentration needed is 10-fold higher than
usually required for dodecylsulphate (Table VI). One consequence of this difference
between the detergents is that saturation of the cooperative binding with tetra-
decyltrimethylammonium chloride is apparently prevented by the onset of micelle
formation [80]. One would therefore expect that more proteins would be resistant
to the denaturing effect of the cationic detergents than with dodecylsulphate for which
there is more leeway before the CMC is reached. Ribonuclease A, for instance, is
denatured by sodium dodecylsulphate but not by dodecyltrimethylammonium bro-
mide at25 C [103].
The non-ionic detergents and the bile salts do not usually induce the coopera-
tive mode of binding with accompanying denaturation characteristic to the anionic
and cationic detergents. Triton X-100 binding to a number of water-soluble proteins
has been studied and no binding to sites other than the high affinity sites of serum
albumin has been detected [20,87,104,105]. Neither does Triton X-100 usually
appear to induce conformational changes in proteins leading to loss of their bio-
logical properties [89,106-111 ]. Similar results have been obtained with deoxycholate.
Apart from the four high affinity sites on serum albumin, deoxycholate binds to only
fourteen additional weaker sites with no detectable change in protein conforma-
tion [20]. Studies with a number of other proteins show no binding or only low
binding of this detergent, the highest binding being observed for thyroglobulin, which
bound 0.15 g deoxycholate per g protein [20,104]. Like Triton X- 100, deoxycholate
does not usually denature proteins [20,106,108,112]. Both of these detergents also
appear to be very inefficient in breaking protein-protein interactions. Most proteins
preserve their quaternary structure in the presence of high concentrations of Triton
X-100 and deoxycholate [104,107-113]. However, examples of dissociation are
also known. Haemocyanin [104], alkaline phosphatase and haemoglobin [109] are
partially dissociated into subunits by Triton X-100.
Table VI shows a summary of the binding of dodecylsulphate, tetradecyltri-
methylammonium chloride, Triton X-100 and deoxycholate to bovine serum albumin.
A common feature for the denaturing detergents appears to be a combination of a
charged head group and a flexible apolar tail. The nature of the charged head group
and the length of the detergent alkyl tail are important and influence both the critical
concentration needed to induce cooperative binding and the resulting conformational
change [6,76]. The mild detergents, Triton X-100 and deoxycholate, have rigid and
bulky apolar moieties, which probably do not penetrate the crevices of the protein
surfaces as efficiently as the flexible alkyl chains.
Furthermore, Triton X-100 is non-ionic which may explain why Triton X-100
in general shows even less interaction with proteins than anionic deoxycholate
(less binding, see also p. 66). It is likely that both deoxycholate and Triton X-100
could induce cooperative binding and denaturation if high enough concentrations
50

of the free monomeric form of the detergent could be reached. However, this is not
possible because of micelle formation; the monomer concentration is limited ap-
proximately to the CMC of the detergent (more so for Triton X-100 than for deo-
xycholate, for which the monomer concentration continues to rise above the CMC,
see p. 38).

IVB. Binding of detergents to membrane proteins


IVB-1. Membrane protein classification. The interactions of detergents with
membrane proteins will differ from those seen with water-soluble proteins due to
their differences in structure. It has been suggested that the proteins associated with
membranes can be roughly classified into two categories termed peripheral [114]
(extrinsic [115] or membrane associated [116]) and integral [114] (intrinsic [115]).
This classification is operational and based on the way these proteins can be detached
from the membrane [117]. The peripheral proteins can be dislodged, usually in lipid-
free form, from membranes by chelating agents, by lowering or increasing the ionic
strength or the pH without dissociating the lipid matrix of the membrane. Peripheral
proteins are usually water soluble after extraction, and include proteins like cyto-
chrome e, the tertiary structure of which is known and is quite similar to those of
other water-soluble globular proteins having a hydrophilic surface and a hydrophobic
interior [118]. These proteins are presumably bound by predominantly polar inter-
actions to the proteins or the lipids in the membrane [1 ! 9].
The integral proteins are tightly bound to the membrane and can only be sol-
ubilized by disrupting the membrane with organic solvents or by detergents. These
proteins have been suggested to have an amphiphilic structure [120,121]. Evidence
supporting this view is beginning to appear. Cytochrome b5 of the endoplasmic
reticulum has a hydrophilic globular part [122] with a molecular weight of 11 000
which is anchored to the lipid matrix by a 5000 dalton peptide segment enriched in
apolar amino acids [123,124]. The MNglycoprotein (glycophorin) of the erythrocyte
membrane has a stretch of 23 apolar amino acids spanning the membrane with
hydrophilic portions of the peptide chain on both sides of the membrane [125-127].
The Semliki Forest (SF) virus membrane glycoproteins span the viral membrane in a
similar fashion with about 90 % of the protein protruding as spikes into the external
aqueous medium [128,129]. Rhodopsin in the rod outer segments of the retina seems
to be fairly deeply immersed in the lipid matrix [130,13l] forming intramembranous
particles with a diameter of 110 ~ [132], but this protein is also presumably amphi-
philic having hydrophilic regions exposed to the aqueous phase [133]. A special
case is presented by the proteolipids in myelin, which like phospholipids are soluble
in chloroform/methanol [134] and have a very hydrophobic amino acid composi-
tion [135]. To what extent these proteins are exposed outside the lipid region of the
membrane is not yet known.
From this classification, it is apparent that the peripheral membrane proteins
can be expected to interact with detergents in a similar manner as with water-soluble
proteins. However, only one peripheral protein has yet been studied. Cytochrome c
51

is denatured by dodecylsulphate and binds maximally 1.4 g sodium dodecylsulphate


per g protein [101]. No binding of Triton X-100 could be detected [104] whereas
only 12 molecules of deoxycholate are bound per molecule of cytochrome e [136]
apparently by electrostatic interactions. More studies are clearly needed, especially
to reveal whether high affinity sites (hydrophobic surface patches or crevices) will
turn out to be a more general feature in this class of proteins than in ordinary water-
soluble proteins.
The integral membrane proteins have been more extensively studied. For
binding studies the proteins have to be released from the membrane preferably in lipid-
free form. We will describe below (p. 62) how integral proteins have been obtained
lipid free in soluble form using detergents. The integral proteins as such are usually
poorly soluble in aqueous solution. The complete binding isotherms may therefore
be difficult to obtain. However, at higher detergent concentrations, the proteins are
usually soluble and at least the maximal detergent binding capacity of the proteins
can be measured.
IVB-2. Binding of "denaturing" detergents. Most membrane proteins are
denatured by sodium dodecylsulphate, and it is usually assumed that these proteins
show a cooperative mode of binding identical to that of water-soluble proteins.
This is indeed the basis of the use of sodium dodecylsulphate gel electrophoresis
[97,98,137] to estimate the molecular weights of reduced membrane polypeptide
chains. However, this assumption may be invalid for some integral membrane pro-
teins. Sodium dodecylsulphate interacts cooperatively with the MN membrane
g[ycoprotein at the CMC with binding ratios as high as 5-7 g sodium dodecylsul-
phate per g protein [82]. It is likely that the observed high binding ratios reflect
interaction of the detergent with the sugar residues as well as with the polypeptide
chain. There are also reports for the SF virus glyeoproteins [138], for cytochrome bs
[124] and cytochrome b5 reductase [139], for the polypeptide chains of (Na + + K+) -
ATPase [140] and for cytochrome oxidase [111 ] which indicate that the binding of
sodium dodecylsulphate to these integral membrane proteins may differ quantita-
tively from the binding by water-soluble proteins.
Some membrane proteins are resistant to the denaturing action ofdodecylsulphate
at room temperature. These include membrane-bound neuraminidase of influenza virus
[141 ], alkaline phosphatase in liver cell plasma membranes [142] and phospholipase
AI of Eseheriehia cob cell membranes [143]. It is most important to note that some
still undefined proteolytic enzymes (either membrane associated or released in active
form by cellular fragmentation) may digest most membrane proteins to fragments
with molecular weights of less than 10 000 even in the presence of dodecylsulphate
concentrations above the CMC [5,144].
The cationic detergents seem to denature many membrane proteins in a
similar fashion as sodium dodecylsulphate (see refs 80,145-148), but no binding
studies are available. Some proteins like rhodopsin, which can be extracted from
the retinal rod outer segments in a lipid-free form complexed to the cationic deter-
gent, are more resistant to denaturation [132]. At 4C the rhodopsin molecule
52

TABLE VII

PROPERTIES OF TRITON X-100-MEMBRANE PROTEIN COMPLEXES


Triton X-100 (mg Molecular Sedimentation Stokes Ref.
per mg protein) weight coefficient (S) radius (~.)

SF virus membrane 0.54 1.5' 105 4.5 53


glycoproteins 0.21 11 ' 105 23,5 94 110
Rhodopsin 1.46 1.02" 105 2.7 49 178
Cholinergic receptor ~0.3 4.7" 105 12.5 73 108
protein
Cytochrome oxidase ~0.3 3.08 " 105 7.7 75 111
ATPase (sarcoplasmic
reticulum) 0.6 - - 177
Low density lipoprotein a 0.52 - - - 104

a The apoprotein of serum low density lipoprotein has lipophilic properties similar to those of mere-
brahe proteins.

does not undergo any great conformational changes as judged from its spectral
properties, but at r o o m temperature the protein is rapidly denatured by the detergent.
The thermal stability of the detergent-rhodopsin complex increases as the
chain length in the alkyltrimethylammonium bromide series used increases
[1321.
IVB-3. Binding of "mild' detergents. Triton X-100 and other non-ionic deter-
gents of type A (mostly polyoxyethylene alcohols like the Lubrol and the Brij series)
have been used to solubilize a number of integral membrane proteins without loss
of their biological activities [2]. This has been done with receptor proteins [107,
149-157], transplantation antigens [158,159], viral membrane glycoproteins [110,
160--162], and numerous membrane enzymes [ 123,124,147,163-174]. The resulting
detergent-protein complexes that have been characterized are shown in Table VII.
These studies show that large amounts of Triton X-100 are bound to the proteins.
The SF virus glycoproteins (El and E2 b o t h w i t h a molecular weight of about 50 000,
and E3 with a molecular weight of about 10 000 [175]) can be extracted with Triton
X-100 f r o m the viral membrane as a soluble lipid-free detergent-protein complex
(Fig. 11). This complex has a sedimentation coefficient of 4.5 S and a Stokes radius
of 5 3 A and contains about 75 molecules of Triton X-100 bound to about
100000 daltons of glycoprotein which probably corresponds to one spike on the
viral surface [176]. The 4.5 S complexes aggregate readily to form a 23.5 S complex,
which has a Stokes radius of 94 A, and contains eight molecules each of El, E2 and
E3 and about 260 molecules of Triton X-100 [110]. Thus about 340 molecules of
Triton X-100 are released from the protein-detergent complex for each 23.5 S
complex from the 4.5 S complexes suggesting that some of the Triton binding sites
on the protein in the 4.5 S complex are engaged in protein-protein contacts in the
23.5 S aggregate. Sucrose alters the equilibrium between the 4.5 S and 23.5 S forms:
53

removal of sucrose leads to aggregation of the 4.5 S form to the 23.5 S form, and
addition of sucrose to dissociation [ll0]. The Triton is apparently bound to the
hydrophobic domains of the glycoproteins, and not to their hydrophilic parts [128].
Other studies show that serum low density lipoprotein [104], a mixture of
erythrocyte membrane proteins [104], a calcium-independent ATPase from sarco-
plasmic reticulum [177], rhodopsin [178] and cytochrome b5 [179] can be delipidated
with Triton X-100 to form soluble protein-detergent complexes (Table VII). Qualita-
tive immunological analysis indicated that the antigenic properties of the low-density-
lipoprotein apoprotein were preserved in the protein-Triton X-100 complex [180].
An interesting feature of the rhodopsin study was that bleaching of the rhodopsin
resulted in a drastic reduction of the Triton X-100 bound and in aggregation of the
opsin formed [178]. This aggregation may explain in part why the opsin cannot be
regenerated into rhodopsin in the presence of Triton X-100. The calcium-independent
ATPase of the sarcoplasmic reticulum lost activity when delipidated by Triton
X-100 but could be reactivated by a variety of anionic amphiphiles and phospho-
lipids [177]. Optical measurements indicated no change in protein structure in cyto-
chrome b5 complexed to Triton X-100. The detergent was found to be bound to the
hydrophobic domain of cytochrome bs and not to the hydrophilic globular part [179].
The cholinergic receptor protein [108], cytochrome oxidase [l 1l] and o-alanine
carboxypeptidase [109] have also been solubilized as Triton X-100 complexes. In
these studies Triton X-100 binding was measured indirectly without recognizing the
pitfalls later pointed out by Tanford et al. [81 ]. Both the cholinergic receptor protein
and o-alanine carboxypeptidase preserved their activities when solubilized with
Triton X-100 whereas cytochrome oxidase was inactivated, but could be reactivated
by added phospholipid.
A typical feature of the protein-Triton X-100 complexes is a combination of a
low sedimentation coefficient and a high Stokes radius. This is mainly caused by
the high partial specific volume of the complex due to bound Triton X-100 (partial
specific volume of Triton X-100 ~ 0.908) [81], but asymmetry and abnormal hy-
dration of the complexes may also contribute. Such hydrodynamic properties appear
to be typical also for other membrane proteins solubilized with Triton X-100 and
other non-ionic detergents but for which the contribution of detergent (and lipid)
content has not yet been evaluated [ 107,153,156,166,171,181 ].
The type B detergents (bile salts and digitonin) have also been successfully used
to solubilize many membrane proteins without loss of biological activity [112,108,
123,124,148,154,168,180,182-189]. Workers in the mitochondrial field especially
use these surfactants [1,2,4,190]. Few of the resulting soluble lipid-free complexes
have been analyzed in any detail. In the first detailed characterization of a membrane
protein-detergent complex, Hubbard showed that rhodopsin can be extracted by
digitonin as a complex containing one rhodopsin molecule (possibly some residual
lipid) and approximately 180 digitonin molecules [186]. The lipid-free digitonin-
rhodopsin complex can be regenerated after bleaching [132]. Bile salts have been
found to bind to Bacillus licheniformis penicillinase [147]. This enzyme can occur
54

in two forms, one membrane bound and the other water-soluble, the latter being
secreted into the extracellular medium. The membrane-bound form can be solubilized
with bile salts, as a complex in which one penicillinase molecule is bound to about 37
molecules of taurodeoxycholate [191]. The secreted form does not bind bile salts.
The structural difference between these two penicillinase forms is not yet known.
Other studies have shown that SF virus glycoproteins [104], the apoproteins of the low-
[ 104] and high-density lipoproteins [81 ], and a mixture of delipidated erythrocyte mem-
brane proteins [104] bind from 0.29-0.64 mg deoxycholate per mg protein. It should be
noted that this class of detergents have partial specific volumes (for deoxycholate 0.778
[81 ]) which are lower than thoseof most non-ionic detergents, therefore a similar combi-
nation of low sedimentation coefficients and high Stokes radii (low diffusion coefficients)
due to bound detergent is not to be expected for the bile salt-protein complexes.
For a useful discussion of the methods available to characterize the hydrodynamic
properties of detergent-protein complexes, see ref. 81.
These studies show that the integral membrane proteins so far studied bind
appreciable amounts of the mild detergents Triton X-100 and deoxycholate. In a
few cases the binding is associated with loss of activity. We will return to the mech-
anisms involved in a later section (see p. 65).

V. EFFECTS OF DETERGENTS ON BIOLOGICAL MEMBRANES

Compared to the simple two and three component systems described in the
preceeding sections, biological membranes constitute extremely complex mixtures of
lipids, proteins, ions, etc. In spite of the greater complexity of membranes it seems
clear to us that the general principles of detergent action revealed by studies on model
lipid bilayers and isolated proteins, apply to membranes as well, and that the effects
of detergents on membranes can in fact be understood as a combination of the data
described in the preceeding sections. There will be modifying factors due to special
structural features of biological membranes. The lipid may be asymmetrically dis-
tributed between the two leaflets of the bilayer in the membrane [192,193]. Different
regions of the bilayer may be in different physical state (rigid vs fluid) and have
different lipid composition (see ref. 117). Divalent cations complexed to the lipids
may alter their properties (see ref. 194). Both electrostatic and hydrophobic inter-
actions between the lipids and the membrane proteins may further affect the state of
the lipids [195-199]. The lipid composition in close vicinity of membrane proteins
may differ from the rest of the membranes. The presence of a carbohydrate surface
coat (glycochalyx) [200], of highly charged proteins or glycoproteins [192], or a
dense protein network [201] on the surface of the membrane may affect the pene-
tration and the effects of detergents on the membrane.
We shall divide our survey of the effects of increasing detergent concentrations
on membranes into three sections: (1) Effects at low surfactant to membrane ratios.
(2) Disruption of membranes into mixed micelles+ proteins and lipoprotein complexes.
(3) Separation of lipid and protein.
55

This division corresponds roughly to Stages I-III in the solubilization of


phospholipid bilayers by detergents (see p. 46). Separate sections are devoted to
the effects of detergents on protein-protein interactions in membranes and to the se-
lective nature of the solubilization of different membrane components. We shall
stress the data obtained using sodium dodecy]sulphate, Triton X-100 or deoxycholate
as the physical properties of these detergents are relatively well characterized and
since they have become the most commonly used representatives of the three impor-
tant detergent groups used in membrane work, i.e. ionic detergents (type A), non-
ionic detergents (type A) and bile salts (type B).

VA. Effects at low detergent/membrane ratios


Detergents and other soluble amphiphiles seem to bind to membranes even
at very low concentrations. As with lipid bilayers and proteins, such binding affects
the membrane properties in many ways most of which are still unknown. These
changes may involve subtle alterations in the permeability, or at higher detergent
concentrations more drastic effects such as membrane lysis and fusion. All the effects
described in this section occur at lower detergent concentrations than those required
for the solubilization of the membrane.
VA-1. High affinity binding. Of the traditional membrane solubilizing surfac-
tants, the binding of dodecylsulphate and lysolecithin to biological membranes (the SF
virus membrane and rat brain microsomes) have been determined under equilibrium
conditions spanning a large concentration range. Fig. 9 shows the isotherm of the
sodium dodecylsulphate monomer binding to the SF virus [202]. The region of
binding at which lysis and breakdown of the viral membrane occurs is indicated.
The binding is biphasic exhibiting saturable binding (binding to a finite number of
sites) at low dodecylsu]phate concentrations and apparently non-saturable at the

>

5
E
15000

a 10000
B i n d i n g of SDS t o SFV a t 4 " C


!.
g
5000
: Breakdownof SF~/
5
Z
i i I i i ii i i i i I I i ii
-5 -4
Log conc. f r e e S D S ( m )

Fig. 9. Binding of sodium dodecylsulphate (SDS) to SF virus. Taken from Becker et al. [202].
SFV = SF virus.
56

Cfree(13M)
10 5 4 3 2 . 5 2 0 151.3
I , " ' , , , , , , , , , , , , , ]

Protection / 7Q012
70I ~/YS~ / t 0015 }

bi ~ K=165000(M-1)40.025
3o/l ,[ / ' - ..... ~003.
~Iu~- ~-/ -.~ 40.04 E
1 g t5
oK . . . . . . . P
0.2 0.4 0.6 0.8
frce(JUM-1)
Fig. 10. Reciprocal plot of the concentration of chlorpromazine bound to erythrocyte membranes
(cmcmb,a,e) against the free equilibrium concentration (cf,e~) at 22C. cmemb.... is expressed as
mol per I of hydrated or wet membrane. Taken from Kwant and Seeman [209].

higher lytic concentrations. Scatchard plots of the binding at sub-lytic concentrations


suggest about 11 000 binding sites in the virus with affinities close to 105 M ~.
This corresponds to 1 binding site per 1.5 phospholipids in the membrane or to about
40 sites per spike glycoprotein of 100 000 daltons. Binding of Triton X-100 to the
SF virus membrane also occurs at concentrations well below the CMC but there are
not enough data to calculate binding affinities [203]. Lyso[ecithin is avidly bound to
microsomes even at 10-5 M concentrations, and the maximal binding with high
affinity corresponds to about 1 binding site per 2 phospholipids [204].
Anaesthetic agents and fluorescent probes provide another group of compounds
for which there are detailed membrane-binding data [65,66,205,206]. The anaesthetic
drugs are a heterogenous group of molecules many of which are amphiphilic, surface
active and form aggregates at higher concentrations [207,208]. The reciprocal plot
of the binding of chlorpromazine (often used as a model substance for the surface-
active anaesthetics) to erythrocyte membranes is shown in Fig. 10 [209]. The linear
region at free chlorpromazine concentrations below 10-s M gives 0.06 mol of binding
sites per 1 wet membrane (approximately 1 binding site per 5 phospholipids)
with an average affinity constant of 0.165" l06 M -1. Similar binding relative to
membrane phospholipid has been observed for chlorpromazine in isolated sarco-
plasmic reticulum [210]. At higher concentrations of chlorpromazine (known to
be lyric to red blood cells), a large number of lower affinity binding sites can be de-
tected. The thermodynamic parameters suggest that the high affinity binding is
mainly hydrophobic in nature [209].
Many of the fluorescent probes commonly used in membrane work, such as
l-anilino-8-naphthalene sulphonate (ANS) and 2-toluidinylnaphthalene 6-sulphonate
form aggregates and can be classified as type B soluble amphiphiles. Their
binding to membranes and lipoproteins has been thoroughly studied (see refs 65 and
57

206). They bind with affinity constants between 10 a and l0 s M -1 depending on


different parameters such as the nature of the membrane, the presence of cations and
the membrane potential. The bound ANS profoundly alters the structure of serum
concentrations lipoproteins, liposomes and probably also of biological mem-
branes at higher than those normally used in fluorescence measurements
[208].
Whether the high affinity binding sites for dodecylsulphate, anaesthetics and
fluorescent probes involve the membrane protein, lipid or both is an open question.
Water-soluble, lipid-free octamers of the SF virus spike glycoproteins (obtained
after Triton X-100 delipidation and subsequent Triton removal) have been shown to
possess binding sites with affinities similar to those of the intact membrane [202].
But they are not present in high enough numbers to account for all the high affinity
binding sites observed in the virus. It is not known, however, how many sodium
dodecylsulphate binding sites are shielded by protein-protein interactions in the
octameric aggregates (see Table IX). Single shelled liposomes of isolated membrane
lipid, on the other hand, have the capacity to account for all the viral high affinity
binding sites for dodecyl-sulphate [202].
ANS and other fluorescent probes have also been shown to bind to both isolated
lipids and proteins, and in biological membranes they apparently bind to both lipid
and protein (see ref. 65). The recent work of Haynes and Staerk [211] shows that
saturable binding similar to that observed in biological membranes can be observed
with pure lipid bilayers and that the head group region of four lecithin molecules
may provide one binding site. In model lipid-protein-water phases, the most likely
location of ANS has been shown to be the region of hydrophobic contacts between
the protein and the lipids [212]. The high affinity binding of surface-active anaesthet-
ics is usually thought to involve the lipid part of the membranes but binding to
proteins cannot be ruled out [66].
VA-2. Changes in membrane properties. The high affinity binding of anaes-
thetics has been shown to induce a number of changes in the membrane (for a review,
see ref. 66). One change is the expansion of erythrocyte membranes which may result
in an increase in the membrane area up to 7 ~. The increase in membrane area is much
larger than expected from the bulk volume of the anaesthetics in the membrane and
the reasons for the large expansion remain unclear. The stabilization of membranes
against osmotic, mechanical and acid lysis observed at low anaesthetic and surfactant
concentrations is probably a consequence of the increased membrane area. Chlor-
promazine [213], Triton X-100 [214] and sodium dodecylsulphate [214] protect red
blood cells against osmotic shock at concentrations of 1"10-5-6 "10-5 M but higher
concentrations of these agents cause lysis (cf. Fig. 10). Not all surfactants have this
stabilizing property; the saponins which form specific complexes with cholesterol
do not stabilize red blood cell membranes at any concentration. Anaesthetics at
high affinity binding concentrations also cause chap.ges in membrane permeability
to ions and water. Relatively high concentrations (about half the concentration
required for complete lysis) cause a marked increase in the permeability of charged
58

and neutral solutes but not of macromolecules. This concentration range is called
the prelytic region.
When proteins and other macromolecules begin to pass through the membrane,
lysis has occurred. The available binding isotherms show that lysis of membranes is
connected to a massive increase in binding (Figs 9 and 10). Lysis by surfactants has
been studied mainly using erythrocytes, since the process can be measured quantita-
tively by following the release of haemoglobin. In spite of intensive research in this
field the molecular mechanism of lysis is not yet known [215,337]. The lytic process can,
however, be divided into five stages [215]: (1) The surfactant monomers adsorb
to, and (2) penetrate into the membrane, where (3) they induce a change in molecular
organization. This leads to an alteration in permeability (4) and in the osmotic equi-
librium, and finally (5) to the leakage of haemoglobin. Stages 2, 3 or 4 are rate lim-
iting. It is generally believed that lysis results from an interaction between surfactant
and the lipids of the membrane. Haydon and Taylor [74] have suggested that sur-
factants may act as 'wedges' which destroy the natural orientation of the lipid bilayer.
Similar mechanisms of localized membrane disordering have been proposed to ex-
plain the fusion of biological membranes induced by lysolecithin and other surfac-
rants [216]. Fusion frequently occurs at lyric surfactant concentrations and does not
involve complete disruption of the lamellar membrane [217,218].
Of the other low concentration effects on the physical properties of biological
membranes, the following should be mentioned: displacement or the increase of
membrane-bound Ca z+ by various anaesthetics [66], possible membrane fluidization
(disordering) by anaesthetics [66] and reduction of membrane buoyant density by
sodium dodecylsulphate [202] and Triton X-100 [203].
Low surfactant concentrations affect most membrane-bound enzymic activities
in one way or another (see refs 4 and 219). These may be activated, inhibited or
modified. Phosphorylation in mitochondria and chloroplasts is uncoupled at low
detergent concentrations, facilitated diffusion and carrier-mediated transport through
membranes are affected, and some enzymic activities increase more than 10-fold.
As these effects have not usually been correlated with equilibrium binding studies it is
not possible to say if they are a consequence of high affinity binding or if they occur
at higher detergent concentrations. In many cases the effects may stem from "assay
problems", due to the particulate nature of the enzyme preparations [219]. In mem-
brane vesicles the enzymes are accessible from the inside, the outside or from both
sides. By making the vesicles permeable to the substrate, products, effectors, etc.
and by changing the state of aggregation of the membranes, surfactants may merely
alter the conditions of the assay. It is therefore difficult to ascertain whether the
changes in activity result from the direct action of surfactant on the enzyme molecule
or on the membrane environment around the enzyme. In some cases the surfactant
may also affect the state of the substrate [220]. The effect of detergents on membrane
enzymes is sometimes biphasic; activation is observed at low detergent concentrations
and inhibition at higher. Optimal activation occurs usually when the enzyme is still
membrane bound.
59

VB. Solubilization of membranes into lipoproteins, proteins and mixed mieelles


There is direct [145,202,203,221,222] and indirect [223-228] evidence showing
that increasing amounts of detergent are bound to membranes as the detergent
concentration is raised beyond lytic concentrations. Apparently saturation, anal-
ogous to that postulated for the pure lipid bilayers, is reached at some point and
gradual disintegration of the membrane starts. In this section, we shall restrict
ourselves to the stage in membrane breakdown at which the original lamellar structure
disappears, i.e. when the membrane is "solubilized". There are no generally accepted
criteria for membrane solubilization. In pure lipid systems solubilization can be
defined as the formation of mixed micelles (see p. 43). The more complex nature
of the membrane solubilization process does not permit such a simple definition [1,
3-5,229]. The criteria used are operational, and are usually based on the decrease in
turbidity of the membrane solution, the increase in non-sedimentable material and
the disappearance of continuous lamellar membranes as seen in electron microscopy.
As in the case of surfactant binding to proteins, it seems quite clear that it is
primarily the monomer form of the detergent that is bound to the membrane and
not the micellar [202,227]. At total concentrations of dodecylsulphate below the
CMC (i.e. in conditions in which by definition no micelles can occur), solubilization
of the membrane of SF virus has been observed [202]. Binding of micelles
to membranes as an alternative binding mechanism when membranes and micellar
solutions are mixed cannot, however, be completely ruled out [230,231 ]. There is also
evidence indicating that bile salts may begin to bind to membranes first at the CMC
or just below it [232,233]. Bile salts cause lysis, however, at concentrations far below
the CMC [234,235] which suggests that at least some binding must occur at these
concentrations. How the effects of bile salts relate to their membrane binding,
and to the CMC, is still an open question.
The free equilibrium concentrations of sodium dodecylsulphate and Triton
X-100 have been shown to remain below or equal to the CMC in the presence of
membranes, the excess being bound to the membrane or to components released
from the membrane [202,203]. The binding of detergent monomers to membranes
prevents the formation of pure detergent micelles by reducing the free detergent
concentration below the CMC in the same way as observed with lipid bilayers
and isolated proteins. Therefore micelles of these detergents do not occur together
with membranes in solution unless the detergent is present in a large excess, which
completely solubilizes the membrane [227]. With surfactants that form small micelles
(such as cholate), the monomer concentration keeps rising significantly at concen-
trations above the CMC (see p. 38). In these cases membranes and micelles may in
theory occur together at concentrations above the CMC although actual examples
are lacking.
The extent of solubilization depends mainly on the amount of detergent bound
and can therefore best be correlated with the ratio of bound detergent to membrane.
This ratio is often rather laborious to determine. It is, however, a common experience
that the effects of detergents correlate quite well also with the total detergent/mere-
60

TABLE VIII
DETERGENT: LIPID RATIO REQUIRED FOR THE SOLUBILIZATION OF MEMBRANES
Triton X-100 Sodium Deoxycholate
dodecylsulphate

Detergent: lipid (w/w) 1.9 0.9 a 1.6 0.3 1.6 1.3


References (148, 164, 203, 252, (148,224, 311,336) (148, 182, 249, 297,
295, 311,335) 250, 261,311,328)
Detergent: lipid (w/w)
(pure phospholipid 0.4 - 1.7b 1.8c
liposomes)

a Standard deviation.
b Refs 60, 62.
c Unpublished observations (Helenius, A.).

brane ratio in the sample especially if the concentration of membranes is relatively


high. The reason is that under these conditions the free unbound detergent will
constitute just a small portion of the total detergent in the sample. The lower the
CMC of the detergent and the higher the concentration of the membrane the closer
will the total detergent/membrane ratio be to the bound detergent/membrane ratio.
It is clear that for evaluation and reproduction of solubilization results, it is not
enough to state only the total detergent concentrations used without mention of the
membrane concentrations, as so often occurs in the literature. Only if the bound
detergent is a negligible part of the total detergent, as when working with trace
amounts of radioactively labeled membranes, should the detergent concentration
alone be stated. In this case it will be approximately equal to the equilibrium free
detergent concentration.
In Table VII1 we have collected some data on the amounts of Triton X-100,
dodecylsulphate, and deoxycholate needed to solubilize various isolated biological
membranes. For the reasons given above, only cases in which the membrane concen-
tration was higher than 2 mg/ml were included. Values for the phase transitions in pure
phospholipid bilayers are included for comparison. The end point of solubilization
is taken as the ratio at which no significant additional release of membrane material
is observed upon further addition of detergent. In a few cases the data were based on
the reduction of turbidity seen with increasing detergent concentrations. The values
are quite approximate as the exact membrane lipid and protein content necessary
for the calculations was not exactly known in all cases and as the end points of sol-
ubilization were not always easy to pinpoint. It is seen, however, that with sodium
dodecylsulphate and Triton X-100, the ratio of detergent to membrane lipid
corresponds roughly to that needed for phase transition in phospholipids.
Comparative studies involving all the lipids extracted from membranes and the
membranes themselves are clearly needed to elaborate on this point. The amount of
deoxycholate (for which we have no liposome data) varies from one report to the other
61

more than for the other detergents, which may partly reflect the well-documented de-
pendence of bile salt effectiveness on the presence of salts and the pH [2].
The detergents that have been singled out empirically as effective membrane
solubilizers are probably the ones which have high affinity for the membrane compared
to their tendency to form micelles, and have in addition molecular properties that
effectively bring about the dissociation of membranes when bound.
For non-ionic surfactants there is an interesting correlation between structure
and solubilizing potency. Studies with mitochondria [236], microsomes [237], viral
membranes [238] and bacterial membranes [171,173] show that almost all effective
surfactants are in the 12.5-14.5 HLB range. Those with higher HLB values (such as
the Tweens), have also found many applications in membrane work [239-242].
They release material (mainly peripheral protein) from many membranes but fail to
dissociate the lamellar membrane structure at the concentrations normally used for
solubilization ( < 5 ~). The lipids and the integral proteins are released only if the
extraction is done at very high pH [239] or if it is done repeatedly [240,242].
The increase of solubilization power with repeated extractions may be explained
by a preferential binding to the membrane of surfactant molecules with shorter
ethyleneoxide chain length [46,243]. This would lead to a progressive decrease of
the H LB value of the membrane-bound surfactant, when the membranes are isolated
and new surfactant is added. Non-ionic surfactants with low HLB numbers are
not water soluble and therefore their use is seldom attempted. Our preliminary results
show that the solubilization of liposomes prepared from isolated membrane lipids
follows a very similar HLB dependence. This suggests that it is the ability to disperse
the membrane lipid that determines the effectiveness of non-ionic surfactants in
membrane solubilization. The effectiveness of non-ionic surfactants in solubilizing
membranes probably also depends on other factors than the HLB value. The chemical
structure of the hydrophilic and the hydrophobic groups has been shown to be im-
portant in some cases [238].
There is little detailed information on the structure of the various complexes
that are released from membranes as the result of the phase transition. The main
reason is the complexity of the solubilized mixtures and the lack of good methods to
study the intermediates in the solubilization process. Although useful for the analysis
and isolation of membrane components, such methods as gel filtration, ion-exchange
chromatography, gradient centrifugation, electron microscopy, neutral salt pre-
cipitations, electrophoretic methods, isoelectric focusing and affinity chromatography
may yield little information about the composition and structure of the complexes
originally released from the membrane during phase transition. This is because the
free equilibrium monomer concentration in the solubilized sample must be maintained
throughout the procedure in order to prevent changes in the complexes due to chang-
ing detergent binding. Reduction of the free monomer concentration will induce
reassociation of membranes, and an increase will result in the further disruption or
even complete separation of lipid and protein of possible complexes.
There is, however, enough evidence to suggest that at least part of the membrane
62

proteins are originally released from the membranes as lipoprotein-detergent com-


plexes. It is well known that particles containing lipid, protein and presumably also
detergent, are derived from mitochondria [4], chloroplasts [24d 248] and sarco-
plasmic reticulum [249,250] after treatment with intermediate concentrations of bile
salts and non-ionic detergents. In their study on the breakdown of Acholeplasma
membranes, Engelman et al. [251] used sucrose gradient centrifugation where the
gradient dodecylsulphate concentrations were adjusted approximately to the free
concentrations in the samples. The initial step in the detergent attack on the Achole"
plasma membrane was the release of some protein (presumably peripheral) followed
by a drastic decrease in particle size with the protein and lipid sedimenting together.
Only at high sodium dodecylsulphate concentrations ( ~ 4 mg dodecylsulphate/mg
lipid) was there complete separation of lipid and protein.
It has been shown by similar sucrose gradient methods that the disruption of
the simple SF virus membrane by dodecylsuiphate [202] and Triton X-100 [203]
proceeds through intermediate lipoprotein-detergent complexes gradually to full
delipidation of the membrane protein. Whereas lysis of the viral membranes re-
quired 5000 mol of bound Triton X-100 per mol of virus, 40 000 mol were needed
to induce solubilization. This amounts to more than 2 mol of detergent per mol of
phospholipid in the membrane. The viral membrane was shown to disintegrate into
fairly homogeneous complexes containing glycoprotein, phospholipid and Triton,
and into separate lipid/Triton mixed micelles (Fig. 11). The complexes contained
protein, phospholipid and detergent in the weight ratio 25:5:20 but increasing
Triton concentrations gradually reduced the lipid content, the sedimenta-
tion coefficient and the size of the globular particles seen in electron micros-
copy.

VC. Separation of lipids and proteins


According to the view presented in the preceeding section solubilization of
membranes with detergents involves the formation of mixed micelles with membrane
lipid and of mixed micellar structures containing both lipids and proteins. Some
peripheral proteins may be released lipid free. When higher concentrations of deter-
gent are added to membranes than are needed to induce the lamellar-micellar phase
transition, separation of the lipids from the proteins takes place. Enough detergent
has to be added to solubilize the lipids into mixed micelles and to saturate the binding
capacity of the proteins that bind detergent. The mixed micelles formed between
the lipids and the detergent should also preferably be saturated with detergent
(see p. 44) to facilitate separation. The quantities of detergent needed will depend
both on the composition of the membrane and on the detergent itself. Almost com-
plete separation of the membrane lipids from the proteins has been achieved with
ionic and non-ionic detergents of both type A and B [104,110,124,177,178,182,189,
247,251-260]. Instances in which delipidation has been less complete have also
been reported [111,221,261-263]. Whether this has been due to the conditions used,
i.e. insufficient detergent present and (or) sub-optimal separation methods, or to the
63

TxIOO
BINDING LYSIS SOLUBILIZATION DELIPIDATION TXIO0 REMOVAL

@-@ -.~.
~ -

+
r LIPID-PROTEIN-7 FPROTEIN-
~"LTX,OO.COMPLEXJ"
+
LTxIOO-COMPLEXJ- LCOMPLEXJ
+
7 FPROTEIN-
7
NUC LEOCAPSID LIPID-TXIOO LIPID-TXIOO
(NC) MIXED MICELLES MIXEDMICELLES

o 20 I0 20 I0 20 I0 20 5 I0 15
I0
FRACTIONNUMBER
I ii !

~ ~:~ " ":4~

Fig. 11. The stepwise dissociation ofSF viruswith increasingTriton X-100 (TX 100). The various stages
in the dissociation are schematically shown at the top of the figure. The bound Triton X-100 is
indicated by the ! symbol, the grey area represents the internal nucleocapsid (NC) consisting of RNA
and protein, and the black spikes represent the spike proteins which penetrate through the bilayer
membrane. The five sucrose gradients in the centre of the figure demonstrate isolation of the various
intermediates. The RNA and phospholipid of the virus used was labeled with 32p ((3 . . . . (3)
and the protein by all-labeled lysine ( 0 O). 3H-labeled Triton X-100 ( - - x ) was used
in the gradient together with unlabeled virus to demonstrate the Triton binding. The buoyant den-
sities of some of the intermediates are indicated. In all the gradients except the first Triton X-100
(usually 0.135 rag/g) was included. Electron micrographs of the SF virus and various dissociation
products obtained from the sucrose gradients are also shown (negative staining with phosphotung-
state). The enlargement is the same and the bars represent 50 nm. For details, see refs I10 and 203.

residual lipid being so tightly b o u n d t h a t detergent d i s p l a c e m e n t is impossible (cf.


ref. 264) is n o t clear.
The m e t h o d s used to achieve r e m o v a l o f lipid/detergent mixed micelles f r o m
the m e m b r a n e p r o t e i n complexes have been based on differences in size, b u o y a n t
density, charge, binding affinity a n d solvent partitioning, The presence o f micellar
c o n c e n t r a t i o n s o f detergent during the s e p a r a t i o n stage is essential to ensure a suf-
ficient c o n c e n t r a t i o n o f the m o n o m e r i c species o f the detergent in e q u i l i b r i u m with
the s e p a r a t i n g mixed micelles a n d the p r o t e i n - d e t e r g e n t complexes. Gel filtration
has been used for s e p a r a t i o n when the Stokes radii o f the mixed micelles are different
f r o m the proteins. M i x e d micelles o f p h o s p h o l i p i d and T r i t o n X-100 are fairly large
(Stokes radius > 50 A) [62,104], mixed micelles with d o d e c y l s u l p h a t e are i n t e r m e d i -
ate in size ( > 35 A) [180,265] while the bile salts form smaller mixed micelles ( > 15 ,~)
64

[104,265]. The size of the mixed micelles formed during delipidation decreases with
the detergent/lipid ratio until the smallest attainable size is reached (see p. 44).
The buoyant density of the mixed micelles are usually lower than for the protein
or the protein-detergent complexes, and this has been utilized for successful separa-
tion by density gradient centrifugation [251]. The mixed lipid micelles and the
detergent-protein complexes formed by non-ionic detergents can be separated con-
veniently by ion-exchange chromatography [266]. In some cases affinity chromato-
graphy can be used both with non-ionic detergents [267-269] and with bile salts [270].
An interesting new method has been introduced by Albertsson, who combined
phase partition separation with Triton X-100 delipidation to remove the mixed
micelles from the protein [105].
To give an example from our own experience, we have used sodium dodecyl-
sulphate, Triton X-100 and deoxycholate to delipidate the SF virus membrane-
proteins. With sodium dodecylsulphate either gel filtration [138] or density gradient
centrifugation [202] were used. A free sodium dodecylsulphate concentration of
about 0.75 mM (CMC, 2 mM in the buffer used) was enough to obtain delipidation
at 30C whereas below the Krafft point (Fig. 2), total lipid removal could not be
achieved at any concentration tested [202]. With sodium deoxycholate, gel filtration
was employed [104], about 10 mg detergent was added to 1.5 mg of membrane
(0.5 mg lipid) and the column was equilibrated with 4 mM sodium deoxycholate,
pH 9.7 at 4C. With Triton X-100 either density gradient centrifugation or ion-
exchange chromatography but not gel filtration could be used for successful separa-
tion [110]. Essentially complete delipidation was obtained by adding 6.5 mg Triton
X-100 to 1.5 mg membrane and by having a concentration of 0.5 mg Triton X-100
per ml present during separation. If the detergent concentration in the density gradient
was lowered to about the CMC (0.15 mg per ml), more Triton X-100 (10 rag) had
to be added to the same amount of membrane to achieve delipidation (unpublished
observations).
The delipidation can be understood as involving progressive removal of the
lipids from the lipoprotein-detergent complexes formed as intermediates during
membrane disruption. As the detergent concentration is increased, the lipid in
contact with protein is exchanged for detergent [271] and transferred into mixed
micelles of lipid and detergent resulting in total separation of the lipid from the
protein. Dissociation of the last lipids bound to the protein may take place only
when the mixed lipid and detergent micelles are being separated from each other by
any of the methods described above.
The detergent-protein complexes formed depend on the nature of the detergent
used. The anionic and cationic detergents of type A such as dodecylsulphate, not
only displace the lipids from those membrane proteins that bind to lipids but due
to the high detergent monomer concentration needed for delipidation, the cooperative
mode of binding to the membrane proteins (see p. 48) cannot usually be avoided.
Sodium dodecylsulphate binds to both peripheral and integral proteins which thereby
usually undergo drastic conformational changes and loss of biological activity.
65

Fig. 12. Hypothetical mechanism for the two-phase extraction (delipidation) of an amphiphilic
membrane protein with a mild detergent like Triton X-100. In the soluble detergent-protein complex
seen to the right the bound detergent molecules form a micelle-likeregion around the hydrophobic
domain of the protein (shaded). The detergent does not interact with the bydrophilic parts of the
protein.

In contrast to the denaturing detergents, Triton X-100 and deoxycholate


appear to interact predominantly with those proteins which are bound to the mem-
brane lipids by hydrophobic interactions. Usually the binding of these detergents
does not lead to major conformational changes of the protein and loss of activity
(see p. 52). The action of these mild detergents is depicted in schematic form in
Fig. 12. According to this view the detergent molecules are bound to the hydro-
phobic domain and not to the hydrophilic parts of the amphiphilic protein. More
data are needed to decide whether the detergent molecules are all bound to individual
sites or whether only a part of the bound molecules would interact directly with the
protein and the rest bind cooperatively to form a micelle-like region on the surface
of the protein. A micelle-likeinteraction appears more likely [179]. The important
feature of this scheme is that the milieu around the hydrophobic domain of the
protein remains apolar, and the milieu around the hydrophilic parts aqueous through-
out the delipidation. Therefore the orientation of the protein in two different phases
is preserved during solubilization, and in many cases the protein-bound detergent
mimics the lipid environment in the membrane sufficiently well to support continued
activity of the protein. Such gentle two-phase extraction is difficult to achieve for
amphiphilic proteins using any other solubilization method. Direct support for this
scheme has been obtained mainly from studies of the SF virus membrane proteins
[128] and cytochrome bs [179].
Although Triton X-100 and the bile salts do not usually function as protein
denaturants, there are a number of membrane proteins which lose their biological
activities when solubilized with these detergents, e.g. (Na+q - K+)-ATPase [237,
272,273], Ca2+-ATPase [177,258], cytochrome oxidase [111,264], glucose-6-phos-
phatase [274], and hormone responsiveness of adenylate cyclase [275]. This may be
due to (1) these proteins being less resistant to denaturation (the constraints that hold
the proteins in their active conformation may be affected by the disruption of essential
protein-lipid or protein-protein interactions during solubilization), (2) detergent
binding to the functional sites on the proteins, (3) removal of cofactors necessary for
activity, (4) to inactivation because of 'assay problems' (see p. 58). Among these
different possibilities, the effects of breaking protein-lipid interactions on enzyme
66

activity have been studied most extensively. A number of membrane enzymes have
been found to depend on specific lipids for activity [219,276]. When these lipids are
removed, the enzyme is inactivated. However, many of these studies have used fairly
harsh methodology for delipidation (organic solvent extraction, impure phospholi-
pases) and in some cases the requirements obtained may depend more on the life
history of the protein than on specificity. Different investigators working with the
same membrane protein (e.g. (Na++ K+)-ATPase [277], rhodopsin [132,278,279]
have arrived at different conclusions concerning their lipid specificities.
Among the presently available methods, Triton X-100 (and probably other
non-ionic detergents) appears to provide the most generally applicable solubilization
agent to extract and purify an integrally bound membrane protein lipid free in its
native conformation or something close to it. Overall Triton X-100 appears to be
more gentle in its effects on proteins than the bile salts [160,168,252,280,281] but
there are examples for which the reverse is true [190]. In special cases, the anionic
(e.g. ref. 143) and cationic detergents (e.g. ref. 132) of type A can be used too, but
with these the likelihood of obtaining a grossly denatured product is great. Mixtures
of detergents can also be tried. Solibilization can in such cases perhaps be achieved
with lower free monomeric concentrations of the detergents (cf. Raoult's law, p. 42)
and a lower risk for denaturation (see ref. 172).

VD. Effects on protein-protein interactions


Like soluble proteins, many membrane proteins possess quaternary structure.
The non-covalent interactions between membrane proteins may be permanent or
reversible depending on their functions in the membrane (see refs 115,117 and 282).
Mitochondrial and chloroplast ATPase and electron transport chain components [4,
283,284], the cholinergic receptors [285,286], viral spike proteins [176,287,288],
the acetylcholine esterase [289,290] and the spectrin of erythrocyte membranes [117,
282] are just a few examples of membrane proteins which have been claimed to have
oligomeric structures. Mild detergents are often unable to dissociate the protein-
protein interactions which keep such complexes together. The complexes are there-
fore solubilized, delipidated and kept in solution as units. A common strategy in the
characterization of active protein complexes has been to isolate the smallest active
protein unit by mild detergents and then study its polypeptide composition using
sodium dodecylsulphate which dissociates the complex. Whether all the peptides in
the complex are essential for biological activity is usually uncertain. Neither can one
be completely sure that artefactual association of polypeptides does not occur.
The 23.5 S complex (octameric spike glycoprotein + 260 Triton X-100 molecules,
see p. 52 and Fig. 11) obtained from the SF virus membrane after delipidation re-
presents such an artefact [110].
The junctions connecting plasma membranes of neighbouring cells in animal
tissue are probably characterized by strong non-covalent protein-protein inter-
actions [l 17]. They are resistant to the action of deoxycholate, non-ionic detergents
~nd lauryl sarcosinate and can thus be easily isolated after detergent treatment
67

[291-293]. Synaptic membranes have also proved more resistant to Triton X-100
than the surrounding neuronal membranes [294,295]. Yu et al. [296] have recently
shown that Triton X-100 causes incomplete solubilization of erythrocyte membranes
when the ionic conditions are such that they favour associations between the mole-
cules of the major peripheral protein called spectrin (which occurs as fibrillar aggre-
gates on the inner surface of the membrane). The ghost-shaped filamentous residue
left after Triton X-100 treatment contained the spectrin, a number of other peripheral
protein components and some lipid. All major integral glycoproteins were solubiliz-
ed. When extraction is performed with Triton X-100 at low ionic strength, which
dissociates the spectrin, practically no such residue is observed. Other filamentous
structures known to occur in conjunction with plasma membranes or specialized
areas of the plasma membranes in animal cells may be similarly resistant to mild
detergents. Mitochondria treated with high concentrations of deoxycholate give a
protein-rich residue which looks like a loosely woven network in electron micro-
scopy [297]. Extraction of Micrococcus lysodeikticus cell membranes with deo-
xycholate also yields a protein residue with the appearance of membraneous sheets
[298]. Generally, non-ionic detergents solubilize the cytoplasmic membranes of
bacteria effectively [299-301]. However, the outer membrane layers of the Gram-
negative bacteria are very resistant to detergents presumably because of protein-
protein interactions. Triton X-100 removes practically no cell wall protein from E.
coli and only part of the lipopolysaccaride and phospholipid [300]. When extracted
with Triton X-100 and EDTA about half of the protein and all the lipid is solubiliz-
ed [301]. The membrane structure appears to be stabilized by divalent cations.
Many animal membrane viruses possess a layer of nonglycosylated peripheral
membrane proteins on the inner surface of the viral membrane, between the bilayer
and the internal nucleoproteins [302,303]. These protein shells are also quite
resistant to mild detergents [303,304,305]. The marine bacteriophage PM2 is un-
usual among membrane viruses in having an external coat of membrane protein
covering most of the bilayer lipid [201 ]. The PM2 is resistant to non-ionic detergents,
unless the protein coat is removed. The external protein is basic and interacts
electrostatically with the phosphatidylglycerol enriched in the outer bilayer leaflet
[306]. The inefficiency of non-ionic surfactants in solubilizing the membrane
!
may in
this case be due either to inability to disrupt and penetrate the outer protein coat or
to dissociate the electrostatic bonds between lipid and protein.
There is an interesting study by Lanyi showing that the effectiveness of Triton
X-100 to solubilize the membranes of Halobacterium cutirubrum depends on the
respiratory state of the organism [228]. The spectrum of the bacteriorubin pigment
is changed by the presence of Triton X-100 in the lipid region of the membrane and
this serves as an indicator for Triton-binding. When the Triton is added to respiring
cells little detergent is bound, and the membrane is not solubilized. When respiration
is inhibited, binding is observed and the membrane is completely solubilized. The
molecular reason for this effect is not known.
Conditions known to enhance the dissociation of proteins, such as high pH
68

[2,164,173], low [164] and high [2] ionic strength, metal chelators [301] and urea
[152,307], as a rule also increase the solubilization efficiency of detergents. These
conditions may change the properties of the surfactant as well, and consequently
the C M C and micellar properties may change. The C M C is, for instance, reduced
by increasing counter ion concentration, which may partly explain the increased
solubilizing activity of ionic detergents with increasing salt. For this reason bile
salts are frequently employed in the presence of high concentrations of NaCI or
KC1 [2]. Sucrose has in some cases been observed to increase the efficiency of non-
ionic detergents [167], an effect for which there is no obvious explanation. The
denaturing detergents which are able to dissociate proteins into their polypeptide
chains can solubilize most membranes up to 100 percent.

VE. Selective solubilization of membrane components


A very useful and interesting phenomenon in the solubilization of membranes
is the selectivity with which different components are released from membranes.
With mild detergents, the selectivity is partly due to the insolubility of large continu-
ous structures stabilized by protein-protein interactions as described above. How-
ever, also the other components (those that can be solubilized) have their own specific
solubilization properties at low and intermediate surfactant-membrane ratios.
An increase in surfactant concentration leads to a more or less sequential release of
different membrane components [2,308-311 ]. Alternatively, selective extraction can
be obtained by changing the ionic strength [140], the HLB value of the surfactant
[242,312,3 l 3 ], or some other crucial solubilization parameter.
Selective solubilization has proved extremely valuable in the purification of
membrane proteins and active membrane complexes [2,4,140,249,308,310,314-316].
Treatment of mitochondria with the appropriate amounts of various detergents
results in the separation of a red supernatant rich in cytochromes b, c, and cl and a
green precipitate rich in cytochromes a and a3 [297,317]. Depending on the detergent
used, the composition of the supernatant can be further varied. Photosystems I and
I! from the chloroplast can be separated by extraction with different detergents [244,
318]. Photosystem I is then released into the supernatant as lipoprotein particles,
whereas the photosystem I1 remains in the membraneous residue.
In the isolation of specific integral membrane proteins, other membrane proteins
are often removed by solubilization at low detergent: membrane ratios, and the re-
maining membrane fraction, now enriched in the active component, can be solubilized
by adding more detergent. A recent method by J6rgensen [316] for obtaining highly pure
Na+K+ATPase from rabbit renal medulla microsomes relies entirely on the selective
removal of inactive protein from the membrane by careful sodium dodecylsulphate
extraction. The CaZ+ATPase of sarcoplasmic reticulum [249] and the mitochondrial
cytochrome oxidase [314] can be enriched similarly, albeit less dramatically, using
bile salts and other detergents. The sarcoplasmic reticulum ATPase appears as
70-90/~ intramembraneous particles in freeze-etching electron microscopy [319,320].
These particles become enriched in the membrane residue when the Ca2+-ATPase
69

is purified by selective removal of other membrane proteins with lysolecithin [222].


The rules governing the selective extraction of membrane proteins are probably
very complex. Selectivity depends on the nature of the detergent, on the conditions,
on possible pretreatment of the membranes, etc. It is often necessary, for instance,
to test optimal conditions for selectivity separately for each membrane batch used for
protein purification. To some extent, however, the selectivity can be rationalized
on the basis of the interactions between the proteins and the membrane matrix.
Thus integral proteins assumed to penetrate deeply into the hydrocarbon region of
the membrane (e.g. animal cell Na+K+ATPase, CaE+-ATPase of sarcoplasmic reti-
culum, cytochrome oxidase), seem to require higher surfactant/membrane ratios than
other membrane proteins to be solubilized.
Membrane lipids can also be selectively extracted. Cholesterol is left behind
in the residue when erythrocyte membranes are extracted with intermediate concen-
trations of deoxycholate [311], whereas in addition to cholesterol, sphingomyelin
and glycolipids also become enriched when extraction is performed with Triton X- 100
[296]. With erythrocyte membranes, sodium dodecylsulphate displays little lipid
selectivity [3t l ], but dodecylsulphate extraction of renal medulla microsomes leads
to increased cholesterol content in the residual membrane [316]. Lysolecithin solu-
bilizes lecithin more effectively than phosphatidylethanolamine from sarcoplasmic
reticulum [222] and Triton X-100 extracts choline-containing phospholipids better
than amino phospholipids from the SF virus membrane [321 ]. It is also of interest
that phospholipids and free cholesterol are more easily solubilized from plasma low
density lipoprotein by deoxycholate [180], sodium dodecylsulphate [180], ANS [208]
and decyl sulphate [322] than are the cholesterol esters and triglycerides. The
separation of photosystems I and I1 from chloroplasts results in a clearcut separation
of various lipid pigments thought to be functionally connected to the different
photosystems in the membrane [246].
The selectivity of lipid solubilization may have many different explanations.
(l) It may result from specific interactions between lipids and other membrane com-
ponents (mainly proteins) which would influence their extractability. (2) It could be
due to differences in chemical composition and physical properties and hence to
differences in susceptibility to detergent solubilization between different regions in
the membrane. The two leaflets of the bilayer have been shown to differ in lipid
composition [192] and there is also indirect evidence for phase separation in the
lateral plane of membranes (see ref. 323). (3) The selectivity may reflect properties of
the detergent rather than the membrane. Detergents have different solubilizing
power for different lipids. Cholesterol, for example, is not effectively solubilized by
bile salts unless swelling amphiphiles are present [69,70] Other explanations are also
possible.
The main difficulty in interpreting lipid solubilization data is to know when the
phenomena observed are determined by the membrane rather than the detergent.
The situation is reminiscent of the difficulties in interpreting the differential extraction
of membrane lipids by organic solvents of different polarity [324,325]. The selective
70

nature of membrane solubilization reveals more than anything how complex the
solubilization process is. More knowledge about membrane structure and of the
detergents is needed for complete understanding of this aspect of solubilization.

VI. DETERGENT REMOVAL

Much useful information about the molecular characteristics of isolated de-


tergent-protein complexes and detergent-lipoprotein complexes can be obtained by
conventional physical and biochemical methods [81]. Experience is, however, still
rather limited in this field and the methodology must be further developed to allow
refined analysis. The availability of chemically homogeneous detergents (especially
non-ionic ones) would be important for exact physicochemical studies.
If the proteins are active in solubilized form, their functional properties can
be studied as such, but if inactivated or modified the factors and conditions needed
to regain the original activity have to be defined. This is usually done by reconstitu-
tion studies. Like lipid phase transitions in general, the solubilization of membranes
is in principle a reversible process. Therefore, bilayer membranes are easily formed
when detergent is removed from solubilized membrane proteins in the presence of
phospholipids or membrane lipid mixtures (see refs 3,4,326). One of the problems
involved in this approach to reconstitution is that the vectorial orientation of the
molecules in the bilayer is usually not regained in the random reassembly process
[326]. Another problem concerns the difficulty of removing the detergents (especially
the non-ionic ones) from solubilized lipids and proteins [3]. For instance, the reason
for the slow dialyzability of non-ionic surfactants is mostly due to their low monomer
concentration which gives an insufficient chemical potential difference between the
the inside and the outside of the dialysis bag. When the ratio of detergent to mem-
brane lipid decreases, the monomer concentration also decreases (see p. 42). Bile
salts, having a high CMC are easier to remove than non-ionic detergents and have
been successfully used in reconstitution studies [3,4,326], but are not always as mild
in their effects on proteins as the non-ionic detergents.
Instead of exchanging the detergent in the complexes for membrane lipid it is
also possible to exchange it for another detergent. The new detergent may be chosen
so that it is easier to remove than the one used in the original solubilization. It may
also be one that is less harmful to the protein.
The removal of detergents from detergent-protein complexes is possible but
leads to protein aggregation and often to the formation of amorphous precipitates.
The aggregation tendency is presumably due to the hydrophobic domains on the
amphiphilic proteins being uncovered as bound detergent disappears. There are,
however, examples of proteins which aggregate to form water-soluble and often
homogeneous oligomeric complexes after delipidation with mild detergents and their
subsequent removal (Table IX). The formation of such complexes can be understood
on the basis of the general behaviour of amphiphilic molecules in aqueous solution.
71

TABLE IX

"LIPID-FREE" WATER SOLUBLE COMPLEXES OF INTEGRAL MEMBRANE PROTEINS


Average number of Approximate References
protein units a particle weight
Neuraminidase spikes
(influenza virus) 12 (polydisperse) 2 700 000 287
Haemagglutinin spikes
(influenza virus) 5 (polydisperse) 1 100 000 287
Cytochrome b5 8 (homogeneous) 120 000 123, 124
Cytochrome b5 reductase 14 (polydisperse) 600 000 139
Cytochrome f 8 (homogeneous) 270 000 188
SF virus spikes 8 (homogeneous)b 900 000b -
ATPase
(sarcoplasmic reticulum) 3.5 (polydisperse) 350 000 258

a In the case of the viral spikes a protein unit consists of more than one polypeptide chain.
b Unpublished results (Helenius, A., von Bonsdorff, C.-H. and Simons, K.).

By mutually covering the hydrophobic domains the proteins can form a complex
with a sufficiently polar surface to remain soluble.
Our experience with the SF virus spike proteins indicates that the procedure
by which the detergent is removed is critical to the formation of soluble complexes
(unpublished observations). Removal of Triton X-100 by dialysis, ion exchange and
gel filtration resulted in macroscopic precipitates, whereas gradual Triton removal
by sucrose gradient centrifugation proved successful. The lipid-free SF spike protein
octamer thus obtained appeared as a rosette-like particle with an average diameter
of 180 A in electron microscopy suggesting that hydrophobic moieties of the proteins
were in the centre with the hydrophilic spikes projecting radially (Fig. 1 1). Sodium
dodecylsulphate binding studies at 4 C revealed only 40 high affinity binding sites
in the 900 000 dalton aggregates (i.e. 5 sites per spike) suggesting that the surface
was, indeed, rather polar. The neuraminidase and hemagglutinin spike proteins
obtained from other virus membranes also form similar rosette-like aggregates [303].
The most important use for these water-soluble detergent-free membrane pro-
tein complexes will probably be in reconstitution studies. Such studies are beginning
to appear [321,327].

ACKNOWLEDGEMENTS

We are indebted to Dr Krister Fontell for advice. Grants from the National
Research Council for Medical Sciences, the Sigrid Jus61ius Foundation, and Finska
Tekniska Vetenskapsakademien, Helsinki, Finland are gratefully acknowledged.
72

REFERENCES

I Penefsky, H. S. and Tzagoloff, A. (1971) Methods in Enzymology (Jakoby, W. B., ed.), Vol.
XXII, pp. 204-219, Academic Press, New York
2 Tzagoloff, A. and Penefsky, H. S. (1971) Methods in Enzymology (Jakoby, W. B., ed.), Vol.
XXII, pp. 219-230, Academic Press, New York
3 Razin, S. (1972) Biochim. Biophys. Acta 265, 241-296
4 Kagawa, Y. (1972) Biochim. Biophys. Acta 265, 297-338
5 Wallach, D. F. H. and Winzler, R. J. (1974) in Evolving Strategies and Tactics in Membrane
Research, pp. 57-127, Springer, Berlin
6 Tanford, C. (1973) The Hydrophobic Effect, John Wiley, New York
7 Hartley, G. S. (1936) Aqueous Solutions of Paraffin-Chain Salts, Hermann and Cie, Paris
8 Frank, H. S. and Evans, M. W. (1945) J. Chem. Phys. 13, 507-532
9 Kauzmann, W. (1959) Adv. Protein Chem. 14, 1-63
10 Aranow, R. H. (1963) J. Phys. Chem. 67, 556-562
11 Hermann, R. B. (1972) J. Phys. Chem. 76, 2754-2759
12 Small, D. M. (1970) Fed. Proc. 29, 1320-1326
13 Mukerjee, P. and Mysels, K. J. (1971) Critical Micelle Concentrations in Aqueous Surfactant
Systems, National Bureau of Standards, NSRDS-NBS 36, Washington
14 Smith, R. and Tanford, C. (1972) J. Mol. Biol. 67, 75-83
15 Haberland, M. E. and Reynolds, J. A. (1973) Proc. Natl. Acad. Sci. U.S. 70, 2313 2316
16 Shinoda, K., Nakagawa, T., Tamamushi, B. and lsemura, T. (1963) Some Physico-Chemical
Properties of Colloidal Surfactants, Academic Press, New York
17 Schick, M. J., ed. (1967) Nonionic Surfactants, Marcel Dekker, New York
18 Mcllwain, D. L., Graf, L. and Rapport, M. M. (1971) J. Neurochem. 18, 2255-2263
19 Shachat, N. and Greenwald, H. L. (1967) Nonionic Surfactants (Schick, M. J., ed.), pp. 8-42,
Marcel Dekker, New York
20 Makino, S., Reynolds, J. A. and Tanford, C. (1973) J. Biol. Chem. 248, 4926-4932
21 Griffin, W. C. (1949) J. Soc. Cosmet. Chem. 1, 311-326
22 Griffin, W. C. (1954) J. Soc. Cosmet. Chem. 5, 249-256
23 McBain, E. L. and Hutchinson, E. (1955) Solubilization and Related Phenomena, Academic
Press, New York
24 Krafft, F. and Wigelow, H. (1895) Ber. Dtsch. Chem. Ges. 28, 2573-2582
25 Carey, M. C. and Small, D. M. (1972) Arch. Int. Med. 130, 506-527
26 Nakayama, H., Shinoda, K. and Hutchinson, E. (1966) J. Phys. Chem. 70, 3502-3504
27 Corkill, J. M. and Goodman, J. F. (1962) Trans. Faraday Soc. 58, 206-214
28 Shinitzky, M., Dianoux, A. C., Gitler, C. and Weber, G. (1971) Biochemistry 10, 2106-2113
29 Waggoner, A. S., Griffith, O. H., Christensen, C. R. (1967) Proc. Natl. Acad. Sci. U.S. 57,
I 198-1205
30 Wishnia, A. (1963) J. Phys. Chem. 67, 2079-2082
31 Becher, P. (1967) Nonionic Surfactants (Schick, M. J., ed.), pp. 478-515, Marcel Dekker, New
York
32 Hall, D. G. and Pethica, B. A. (1967) Nonionic Surfactants (Schick, M. J., ed.), pp. 516-557,
Marcel Dekker, New York
33 Mukerjee, P. (1967) Adv. Colloid Interface Sci. 1,241-275
34 Anacker, E. W. (1970) Cationic Surfactants (Jungermann, E., ed.), pp. 203-288, Marcel Dekker,
New York
35 Ekwall, P., Danielsson, i. and Stenius, P. (1972) Surface Chemistry and Colloids, MTP Int. Rev.
Sci. Phys. Chem. Ser. 1 (Kerker, M., ed.), Vol. 7, pp. 97-145, Butterworths, London
36 Elworthy, P. H., Florence, A. T. and Macfarlane, C. B. (1968) Solubilization by Surface-Active
Agents, Chapman and Hall, London
37 Hutchinson, E. and Shinoda, K. (1967) Solvent Properties of Surfactant Solutions (Shinoda, K.,
ed.), pp. 1-26, Marcel Dekker, New York
38 Osipow, L. I. (1962) Surface Chemistry, Reinhold, New York
39 Tanford, C. (1974) Proc. Natl. Acad. Sci U.S. 71, 1811 1815
40 Mukerjee, P. (1965) J. Phys. Chem. 69, 4038-4040
41 Schick, M, J. (1962) J. Colloid Sci. 17, 801-813
73

42 Ray, A. and Nemethy, G. (1971) J. Am, Chem. Soc. 93, 6787-6793


43 Goddard, E. D. and Benson, G. C. (1957) Can. J. Chem. 35, 986-991
44 Luzzati,V.(1968) Biological Membranes(Chapman, D.,ed.),pp. 71-123,AcademicPress, London
45 Nakagawa, T. (1967) Nonionic Surfactants (Schick, M. J., ed.), pp, 558-600, Marcel Dekker,
New York
46 Maclay, W. N. (1956) J. Colloid Sci. 11,272-285
47 Tartar, H. W. (1955) J. Phys. Chem, 59, 1195-1199
48 Tanford, C. (1972) J. Am. Chem. Soc. 76, 3020-3024
49 Ekwall, P., Fontell, K. and Sten, A. (1957) Gas/Liquid and Liquid/Liquid Interface, Proc. 2nd
Int. Congr. Surf. Act., pp. 357-373, Butterworths, London
50 Hoffmann, A. F. and Small, D. M. (1967) Annu. Rev. Med. 18, 333-375
51 Fontell, K. (1972) Kolloid Z. 244, 246-252
52 Small, D. M. (1968) Adv. Chem. Ser., pp. 31-52 Washington
53 Bangham, A. D. (1972) Annu. Rev. Biochem. 41,753-776
54 Helenius, A., unpublished results
55 Shankland, W. (1970) Chem. Phys. Lipids 4, 109-130
56 Small, D. M. (1968) J. Am. Oil Chem. Soc. 45, 108-119
57 Robinson, N. (1961) J. Pharm. Pharmacol. 13, 53-57
58 Saunders, L. (1966) Biochim. Biophys. Acta 125, 70-74
59 Small, D. M., Bourges, M. and Derivichian, D. G. (1966) Nature 211, 816-818
60 Dennis, E. A. and Owens, J. M. (1973) J. Supramol. Struct. 1, 165-175
61 Ribeiro, A. A. and Dennis, E. A. (1973) Biochim. Biophys. Acta 332, 26-35
62 Yedgar, S., Barenholz, Y. and Cooper, V. G. (1974) Biochim. Biophys. Acta 363, 98-111
63 Van Zutphen, H., Merola, A. J., Brierly, G. P. and Cornwell, D. G. (1972) Arch. Biochem. Bio-
phys. 152, 755-766
64 Seufert, W. D. (1973) Biophysik 10, 281-292
65 Radda, G. K. and Vanderkooi, J. (1972) Biochim. Biophys. Acta 265, 509-549
66 Seeman, P. (1972) Pharmacol. Rev. 24, 583-655
67 Saunders, L. (1957) J. Pharm. Pharmacol. 9, 834-840
68 Kushner, L. M. and Hubbard, W. D. (1954) J. Phys. Chem. 58, 1163-1167
69 Carey, M. C. and Small, D. M. (1970) Amer. J. Med. 49, 590-608
70 Tamesue, N., Tsuyoshi, I. and Kerrison, J. (1973) Am. J. Digestive Disease 18, 670-678
71 Ekwall, P. (1975) Advances in Liquid Crystal Research (Brown, G. H., e.d.), Vol. 1, Academic
Press, London
72 Small, D. M., Penkett, S. A. and Chapman, D. (1969) Biochim. Biophys. Acta 179, 178-189
73 Dervichian, D. G. (1968) Adv. Chem. Ser. 84, 78-87
74 Haydon, D. A. and Taylor, J. (1963) J. Theoret. Biol. 4, 281-296
75 Ekwall, P. and Stenius, P. MTP International Review of Science, Phys. Chem. Ser. 2, Vol. 7,
Surface Chemistry and Colloids (Kerker, M., ed.), Butterworths, London, in the press
76 Steinhardt, J. and Reynolds, J. A. (1969) Multiple Equilibria in Proteins, pp. 10--82, Academic
Press, New York
77 Scatchard, G. (1949) Ann. N.Y. Acad. Sci. 51,660-679
78 Klotz, I. M. (1953) The Proteins (Neurath, H. and Bailey, K., eds), Vol. 1, pp. 727-806, Academic
Press, New York
79 Reynolds, J. A. and Tanford, C. (1970) Proc. Natl. Acad. Sci. U.S. 66, 1002-1007
80 Nozaki, Y., Reynolds, J. A. and Tanford, C. (1974) J. Biol. Chem. 249, 4452-4459
81 Tanford, C., Nozaki, Y., Reynolds, J. A. and Makino, S. (1974) Biochemistry 13, 2369-
2376
82 Grefrath, S. and Reynolds, J. A. (1974) Fed. Proc. 33, 1531
83 Borgstrtim, B., in the press
84 Yang, J. T. and Foster, J. F. (1953) J. Am. Chem. Soc. 75, 5560-5567
85 Pallansch, M. J. and Briggs, D. R. (1954) J. Am. Chem. Soc. 76, 1396-1403
86 Reynolds, J. A., Herbert, S., Polet, H. and Steinhardt, J. (1967) Biochemistry 6, 937-947
87 Putnam, F. W. (1948) Adv. Protein Chem. 4, 79-122
88 Few, A. W., Ottewill, R. H. and Parreira, H. C. (1955) Biochim. Biophys. Acta 18, 136-137
89 Dowben, R. M. and Koehler, W. R. (1961) Arch. Biochem. Biophys. 93,496-500
90 Rudman, D. and Kendall, F. B. (1957) J. Clin. Invest. 36, 538-542
74

91 McMeekin, T. L., Polis, B. D., Delia Monica, B. S. and Custer, J. H. (1949)J. Am. Chem. Soc.
71, 3606-3616
92 Groves, M. L., Hipp, N. J. and McMeekin, T. L. (1951) J. Am. Chem. Soc. 73, 2790-2793
93 Wishnia, A. and Pinder, T. (1966) Biochemistry, 5, 1534-1542
94 Ray, A. and Chatterjee, R. (1967) Conformation of Biopolymers (Ramachandran, G. N., ed.),
Vol. 1, pp. 235-252, Academic Press, London
95 Seibles, T. S. (1969) Biochemistry 8, 2949-2954
96 Pitt-Rivers, R. and lmpiombato, F. S. A. (1968) Biochem. J. 109, 825-830
97 Shapiro, A. L., Vinuela, E. and Maizel, J. V. (1967) Biochem. Biophys. Res. Commun. 28, 815-
820
98 Weber, K. and Osborn, M. (1969) J. Biol. Chem. 244, 4406-4412
99 Fish, W. W., Reynolds, J. A. and Tanford, C. (1970) J. Biol. Chem. 245, 5166-5188
100 Nelson, C. A. (1971) J. Biol. Chem. 246, 3895-3901
101 Reynolds, J. A. and Tanford, C. (1970) J. Biol. Chem. 245, 5161-5165
102 Boatman, S. (1973) The Generation of Subcellular Structures (Markham, R., Bancroft, J. B.,
Davies, D. R., Hopwood, D. A. and Horne, R. W., eds), pp. 123-134, North-Holland, Amster-
dam
103 Jones, M. N., Skinner, H. A., Tipping, B. and Wilkinson, A. (1973) Biochem. J. 135, 231-236
104 Helenius, A. and Simons, K. (1972) J. Biol. Chem. 247, 3656-3661
105 Albertsson, P. A. (1973) Biochemistry 12, 2525-2530
106 Burkhard, R. K. and Stolzenberg, G. B. (1972) Biochemistry 11, 1672-1677
107 Cuatrecasas, P. (1972) J. Biol. Chem. 247, 1980-1991
108 Meunier, J. C., Olsen, R. W. and Changeux, J. P. (1972) FEBS Lett. 24, 63-68
109 Umbreit, J. N. and Strominger, J. L. (1973) J. Biol. Chem. 248, 6759-6766
110 Simons, K., Helenius, A. and Garoff, H. (I 973) J. Mol. Biol. 80, 119-I 33
111 Rubin, M. S. and Tzagoloff, A. (1973) J. Biol. Chem. 248, 4269-4274
112 Snary, D., Goodfellow, P., Hayman, M. J., Bodmer, W. F. and Crumpton, M. J. (1974) Nature
247, 457-461
113 Crumpton, M. J. and Hayman, M. J., cited in Snary, D., Goodfellow, P., Hayman, M. J.,
Bodmer, W. F. and Crumpton, M. J. (1974) Nature 247, 457-461
114 Singer, S. J. and Nicolson, G. L. (1972) Science 175, 720-731
115 Capaldi, R. A. and Green, D. E. (1972) FEBS Lett. 25, 205-209
116 Fleischer, S., Zahler, W. L. and Ozawa, H. (1971) Biomembranes 2, 105-119
117 Singer, S. J. (1974) Annu. Rev. Biochem. 43, 805-833
118 Dickerson, R. E., Takano, T,, Eisenberg, D., Kallai, B., Samson, L., Cooper, A. and Margvliash,
E. (1971)J. Biol. Chem. 246, 1511-1535
119 Jacobs, E. E. and Sanadi, D. R. (1960) J. Biol. Chem. 235, 531-534
120 Lenard, J. and Singer, S. J. (1966) Proc. Natl. Acad. Sci. U.S. 56, 1828-1835
121 Wallach, D. F. H. and Zahler, P. H. (1966) Proc. Natl. Acad. Sci. U.S. 56, 1552-1559
122 Mathews, F. S., Argos, P. and Levine, M. (1971) Cold Spring Harbor Symp. Quant. Biol. 36,
387-395
123 lto, A. and Sato, R. (1968) J. Biol. Chem. 243, 4922-4923
124 Spatz, L. and Strittmatter, P. (1971) Proc. Natl. Acad. Sci. U.S. 68, 1042-1046
125 Winzler, R. J. (1969) Red Cell Membrane (Jamieson, G. A. and Greenwalt, T. J., eds), pp.
157-171, Lippincott, Philadelphia
126 Bretscher, M. S. (1971) Nat. New Biol. 231,229-232
127 Segrest, J. P., Jackson, R. L. and Marchesi, V. T. (1972) Biochem. Biophys. Res. Commun. 49,
964-969
128 Utermann, G. and Simons, K. (1974) J. Mol. Biol. 85, 569-587
129 Garoff. H. and Simons, K. (1974) Proc. Natl. Acad. Sci. U.S. 71, 3988-3992
130 Blasie, J. K. (1972) Biophys. J. 12, 191-210
131 Blaurock, A. E. and Wilkins, M. H. F. (1972) Nature 236, 313-314
132 Hong, K. and Hubbel, W. L. (1973) Biochemistry 12, 4517-4523
133 Trayhurn, P., Mandel, P. and Virmaux, N. (1974) FEBS Lett. 38, 351-353
134 Folch, J. and Lees, M. (1951) J. Biol. Chem. 191,807-817
135 Eng, L. P., Chao, F. C., Geostl, B., Pratt, D. and Tavaststjerna, M. G. (1968) Biochemistry 7,
4455-4465
75

136 Rowley, G. R. and Wainio, W. W. (1958) J. Am. Chem. Soc. 80, 4384-4386
137 Fairbanks, G., Steck, T. L. and Wallach, D. F. H. (1971) Biochemistry 10, 2606-2617
138 Simons, K. and K/i~ri~inen, L. (1970) Biochem. Biophys. Res. Commun. 38, 981-988
139 Spatz, L. and Strittmatter, P. (1973) J. Biol. Chem. 248, 793-799
140 Kyte, J. (1972) J. Biol. Chem. 247, 7642-7649
141 Laver, W. G. (1963) Virology 20, 251-262
142 Sohn, R. and Marinetti, G. V. (1974) Chem. Phys. Lipids 12, 17-30
143 Scandella, C. and Kornberg, A. (1971) Biochemistry 10, 4447-4456
144 Dreyer, W. J., Papermaster, D. S. and K~hn, H. (1972) Ann. N.Y. Acad. Sci. 195, 61-74
145 Gitler, C., Martinez-Zedillo, G., Martinez-Rojas, D. and Chavez-Diaz, G. (1967) Natl. Cancer
Inst. Monogr. 23, 153-164
146 Soodsma, J. F. and Nordlie, R. C. (1969) Biochim. Biophys. Acta 191,636-643
147 Sargent, M. C. and Lampen, J. O. (1970) Arch. Biochem. Biophys. 136, 167-177
148 Ne'eman, Z., Kahane, I. and Razin, S. (1971) Biochim. Biophys. Acta 249, 169-176
149 Miledi, R., Molinoff, P. and Potter, L. T. (1971) Nature 229, 554-557
150 Meunier, J. C., Olsen, R. W., Menez, A., Fromageot, P., Boquet, P. and Changeux, J.-P. (1972)
Biochemistry 11, 1200-1210
151 Eldefrawi, M. E., Eldefrawi, A. T., Seifert, S. and O'Brien, R. D. (1972) Arch. Biochem. Biophys.
150, 210-218
152 Suginaka, H., Blumberg, P. M. and Strominger, J. L. (1972) J. Biol. Chem. 247, 5279-5288
153 Benzer, T. L. and Raftery, M. A. (1973) Biochem. Biophys. Res. Commun. 51, 939-944
154 Lefkowitz, R. J., Haber, E. and O'Hare, D. (1972) Proc. Natl. Acad. Sci. U.S. 69, 2828-2832
155 Sabet, S. F. and Schnaitman, C. A. (1973) J. Biol. Chem. 248, 1797-1806
156 Charreau, E. H., Dufau, M. L. and Catt, K. J. (1974) J. Biol. Chem. 249, 4189-4195
157 Levey, G. S., Fletcher, M. A., Klein, I., Ruiz, E. and Schenk, A. (1974) J. Biol. Chem. 249,
2665-2673
158 Kandutsch, A. A. (1963) Transplantation 1,201-215
159 Schwartz, B. D. and Nathenson, S. G. (1971) J. lmmunol. 107, 1363-1367
160 Seto, J. T., Becht, H. and Rott, R. (1973) Med. Microbiol. Immunol. 159, 1-12
161 Scheid, A., Caliguiri, L. A., Compans, R. W. and Choppin, P. W. (1972) Virology 50, 640-652
162 Cartwright, B., Talbot, P. and Brown, F. (1970) J. Gen. Virol. 7, 267-272
163 Walker, P. G. and Levvy, G. A. (1953) Biochem. J. 54, 56-65
164 Miller, D. M. (1970) Biochem. Biophys. Res. Commun. 40, 716-722
165 Bendall, D. S. and de Duve, C. (1960) Biochem. J. 74, 444-450
166 Dowhan, W., Wichner, W. T. and Kennedy, E. P. (1974) J. Biol. Chem. 249, 3079-3084
167 Thompson, M. F. and Bachelard, H. S. (1970) Biochem. J. 118, 25-34
168 Soltysiak, D. and Kaniuga, Z. (1970) Eur. J. Biochem. 14, 70-74
169 Singh, J. and Wasserman, A. R. (1971) J. Biol. Chem. 246, 3532-3541
170 Sun, F. F. and Jacobs, E. E. (1967) Biochim. Biophys. Acta 143, 639-641
171 Hengstenberg, W. (1970) FEBS Lett. 8, 277-280
172 Bachorik, P. A. and Dietrich, L. S. (1972) J. Biol. Chem. 247, 5071-5078
173 Umbreit, J. N. and Strominger, J. L. (1973) Proc. Natl. Acad. Sci. U.S. 70, 2997-3001
174 Bonsall, R. W. and Hunt, S. (1971) Biochim. Biophys. Acta 249, 266-280
175 Garoff. H., Simons, K. and Renkonen, O. (1974) Virology, 61, 493-504
176 Garoff, H. (1975) Virology, 62, 385-392
177 Walter, H. and Hasselbach, W. (1973) Eur. J. Biochem. 36, 110-119
178 Osborn, B. H., Sardet, C. and Helenius, A. (1974) Eur. J. Biochem. 44, 383-390
179 Robinson, N. C., Nozaki, Y. and Tanford, C. (1974) Fed. Proc. 33, 1370
180 Helenius, A. and Simons, K. (1971) Biochemistry 10, 2542-2546
181 Konisky, J. and Liu, C. T. (1974) J. Biol. Chem. 249, 835-840
182 Allan, D. and Crumpton, M. J. (1971) Biochem. J. 123,967-975
183 Ernster, L. and Jones, L. C. (1962) J. Cell Biol. 15, 563-000
184 Dallner, G. (1963) Acta Pathol. Microbiol. Scand. Suppl. 166, 1-94
185 Lu, A. Y. H. and Coon, M. J. (1968) J. Biol. Chem. 243, 1331-1332
186 Hubbard, R. (1954) J. Gen. Physiol. 37, 381-399
187 Van 't Riet, J. and Planta, R. J. (1969) FEBS Lett. 5, 249-252
188 Nelson, N. and Racker, E. (1972) J. Biol. Chem. 247, 3848-3853
76

189 Spatz, L. and Strittmatter, P. (1973) J. Biol. Chem. 248, 793-799


190 Hatefi, Y. (1966) Comprehensive Biochemistry (Florkin, M., ed.), p. 203, Elsevier, Amsterdam
191 Crane, L. J. and Lampen, J. O. (1974) Arch. Biochem. Biophys. 160, 655-666
192 Bretscher, M. S. (1973) Science 181,622-629
193 Colley, C. M., Zwaal, R. F. A., Roelofsen, B. and Van Deenen, L. L. M. (1973) Biochim.
Biophys. Acta 307, 74-82
194 Chapman, D., Urbina, J. and Keough, K. M. (1974) J. Biol. Chem. 249, 2512-2521
195 Gulik-Krzywicki, T., Schechter, E., Luzzatti, V. and Faure, M. (1969) Nature 223, II 16-1121
196 Jost, P. C., Griffith, O. H., Capaldi, R. A. and Vanderkooi, G. (1973) Proc. Natl. Acad. Sci.
U.S. 70, 480-484
197 Steir, A. and Sackman, E. (1973) Biochim. Biophys. Acta 311,400-408
198 Warren, G. B., Toon, P. A., Birdsall, N. J. M., Lee, A. G. and Metcalfe, J. C. (1974) Proc. Natl.
Acad. Sci. U.S. 71,622-626
199 Tr~iuble, H. and Overath, P. (1973) Biochim. Biophys. Acta 307, 491 512
200 Rahmbourg, A. (1971) Int. Rev. Cytol. 31, 57-114
20l Franklin, R. M. (1971) Current Topics in Microbiology and Immunology, Springer, Berlin, in
the press
202 Becker, R., Helenius, A. and Simons, K., (1975) Biochemistry, in the press
203 Helenius, A. and S6derlund, H. (1973) Biochim. Biophys. Acta 307, 287-300
204 Leibovitz-BenGershon, Z. and Gatt, S. (1974) J, Biol. Chem. 249, 1525-1529
205 Huunan-Sepp~il~i, A. (1972) Acta Chem. Scand. 26, 2713-2733
206 Papahadjopoulos, D. (1972) Biochim. Biophys. Acta 265, 169-186
207 Scholtan, W. (1955) Kolloid Z. 142, 84-104
208 Muesing, R. A. and Nishida, T. (1971) Biochemistry 10, 2952-2962
209 Kwant, W. O. and Seeman, P. (1969) Biochim. Biophys. Acta 183, 530-543
210 Balzer, H., Makinose, M., Fiehn, W. and Hasselbach, W. (1968) Arch. Exp. Pathol. Pharmacol.
260, 456~300
211 Haynes, D. H. and Staerk, H. (1974)J. Membrane Biol. 17, 313-340
212 Gulik-Krzywicki, T., Shechter, E., lwatsubo, M., Ranck, J. L. and Luzzati, V. (1970) Biochim.
Biophys. Acta 219, 1-10
213 Kwant, W. O. and van Steveninck, J. (1968) Biochem. Pharmacol. 17, 2215 2223
214 Seeman, P. (1966) Biochem. Pharmacol. 15, 1767 1774
215 Reman, F. C., Demel, R. A., De Gier, J., van Deenen, L. L. M., Eibl, H. and Westphal, O. (1969)
Chem. Phys. Lipids 3, 221-233
216 Lucy, J. A. (1970) Nature 227, 815-817
217 Lucy, J. A. (1973) Membrane-Mediated Information (Kent, P. W., ed.), Vol. 2, pp. 117-128,
MTP, Lancaster
218 Poste, G. and Allison, A. C. (1973) Biochim. Biophys. Acta 300, 421-465
219 Coleman, R. (1973) Biochim. Biophys. Acta 300, 1-30
220 Gatt, S. (1973) Metabolic Inhibitors, Vol. IV, pp. 349-387, Academic Press, New York
221 Bont, W. S., Emmelot, P. and Vaz Diaz, H. (1969) Biochim. Biophys. Acta 173, 389408
222 Deamer, D. W. (1973) J. Biol. Chem. 248, 5477-5485
223 Pethica, B. A. and Schulman, J. H. (1953) Biochem. J. 53, 177-185
224 Bishop, D. G., Rutberg, L. and Samuelsson, B. (1967) Eur. J. Biochem. 2, 454-459
225 Salton, M. R. J. (1968) J. Gen. Physiol. 52, 227-252
226 Kamat, V. B. and Chapman, D. (1968) Biochim. Biophys. Acta 163, 411 414
227 Tanford, C. (1972) J. Mol. Biol. 67, 59-74
228 Lanyi, J. K. (1973) Biochemistry 12, 1433-1438
229 Reisfeld, R. A. and Kahan, B. D. (1971) Transplant. Rev. 6, 81-112
230 Auborn, J. J., Eyring, E. M. and Choules, G. L. (1971) Proc. Natl. Acad. Sci. U.S. 68, 1996-1998
231 Choules, G. L., Sandberg, R. G., Steggall, M. and Eyring, E. M. (1973) Biochemistry 12, 4544-
4550
232 J6rgensen, P. L. and Skou, J. C. (1969) Biochim. Biophys. Acta 233, 366-380
233 Phillipot, J. and Authier, M. H. (1972) Biochim. Biophys. Acta 298, 887-900
234 Roelofsen, B., Zwaal, R. F. A., Comfurius, P., Woodward, C. B. and van Deenen, L. L. M. (1971)
Biochim. Biophys. Acta 241,925 929
235 Kreibich, G., Debey, P. and Sabatini, D. D. (1973) J. Cell Biol. 58,436-462
77

236 Roodyn, D. B. (1962) Biochem. J. 85, 177-189


237 Swanson, P. D., Bradford, H. F. and Mcllwain, H. (1964) Biochem. J. 92, 235-247
238 Stromberg, K. (1971) J. Virol. 9, 684-697
239 Hosaka, Y. (1968) Virology 35, 445-457
240 Zorn, M. and Futterman, S. (1973) Arch. Biochem. Biophys. 157, 91-99
241 Shimizu, K., Hosaka, Y. and Shimizu, Y. K. (1972) J. Virol. 9, 842-850
242 Hjert6n, S. and Johansson, K.-E. (1972) Biochim. Biophys. Acta 288, 312-325
243 Ribeiro, A. A. and Dennis, E. A. (1974) Chem. Phys. Lipids 12, 31-88
244 Anderson, J. M. and Boardman, N. K. (1966) Biochim. Biophys. Acta 112, 403-421
245 Shibuya, I., Honda, H. and Maruo, B. (1968) J. Biochem. Tokyo 64, 571-576
246 Vernon, L. P., Mollenhauer, H. H. and Shaw, E. R. (1968) Regulatory Functions of Biological
Membranes (J~irnefelt, J., ed.), pp. 57-71, Elsevier, Amsterdam
247 Nelson, N. and Neumann, J. (1972) J. Biol. Chem. 247, 1817-1824
248 Arntzen, C. J., DiUey, R. A., Peters, G. A. and Shaw, E. R. (1972) Biochim. Biophys. Acta 256,
85-107
249 MacLennan, D. H. (1970) J. Biol. Chem. 245, 4508-4518
250 Meissner, G. and Fleischer, S. (1974) J. Biol. Chem. 249, 302-309
251 Engelman, D. M., Terry, T. M. and Morowitz, H. J. (1967)Biochim. Biophys. Acta 135, 381-390
252 McFarland, B. H. and Inesi, G. (1971) Arch. Biochem. Biophys. 145, 456--464
253 Korte, T. and Hengstenberg, W. (1971) Eur. J. Biochem. 23,295-302
254 Jones, T. H. D. and Kennedy, E. P. (1969) J. Biol. Chem. 244, 5981-5983
255 Heller, J. (1968) Biochemistry 7, 2906-2913
256 Neeman, Z., Kahane, I., Kovartovsky, J. and Razin, S. (1972) Biochim. Biophys. Acta 266,
255-268
257 Salton, M. R. J. and Schmitt, M. D. (1967) Biochem. Biophys. Res. Commun. 27, 529-534
258 Hardwicke, P. M. D. and Green, N. M. (1974) Eur. J. Biochem. 42, 183-193
259 Evans, W. H. and Gurd, J. W. (1973) Biochem. J. 133, 189-199
260 Nicot, C., Nguyen Le, T., Lepretce, M. and Alfsen, A. (1973) Biochim. Biophys. Acta 322,
109-123
261 Phillipot, J. (1971) Biochim. Biophys. Acta 225, 201-213
262 Triplett, R. B., Summers, J., Ellis, D. E. and Carraway, V. L. (1972) Biochim. Biophys. Acta 266,
484~493
263 Cori, C. F., Garland, R. C. and Wang Chang, H. (1973) Biochemistry 12, 3126-3130
264 Awasthi, Y. C., Chuang, T. F., Keenan, T. W. and Crane, F. L. (1971) Biochim. Biophys. Acta
226, 42-52
265 Borgstr/~m, B. (1965) Biochim. Biophys. Acta 106, 171-183
266 Jacobs, E. E., Kirkpatrick, F. H., Andrews, B. C., Cunningham, W. and Crane, F. L. (1966)
Biochem. Biophys. Res. Commun. 25, 96-103
267 Olsen, R. W., Meunier, J. C. and Changeux, J. P. (1972) FEBS Lett. 28, 96-100
268 Franklin, G. I. and Potter, L. T. (1972) FEBS Lett. 28, 101-106
269 Karlsson, E., Heilbronn, E. and Widlund, L. (1972) FEBS Lett. 28, 107-111
270 Hayman, M. J. and Crumpton, M. J. (1972) Biochem. Biophys. Res. Commun. 47, 923-930
271 Yonetani, T. (1959) J. Biochem. Tokyo 46, 917-922
272 Tanaka, R. and Abood, L. G. (1964) Arch. Biochem. Biophys. 108, 47-52
273 Medzihradsky, F., Kline, M. H. and Hokin, L. E. (1967) Arch. Biochem. Biophys. 121,311-316
274 Garland, R. C. and Cori, C. F. (1972) Biochemistry 11, 4712-4718
275 Levey, G. S. (1970) Biochem. Biophys. Res. Commun. 38, 86-92
276 Rothfield, L. and Romeo, D. (1971) Structure and Function of Biological Membranes (Rothfield,
L. I., ed.), pp. 251-285, Academic Press, New York
277 Roelofsen, B. and van Deenen, L. L. M. (1973) Eur. J. Biochem. 40, 245-257
278 Zorn, M. and Futterman, S. (1971) J. Biol. Chem. 246, 881-886
279 Shichi, H. (1973) Exp. Eye Res. 17, 533-543
280 Meyer, Y. P. (1971) J. Biol. Chem. 246, 1241-1248
281 Kirkpatrick, F. H. and Sandberg, H. E. (1973) Biochim. Biophys. Acta 298,209-218
282 Steck, T. L. (1974) J. Cell Biol. 62, 1-19
283 Tzagoloff, A., Rubin, M. S. and Sierra, M. F. (1973) Biochim. Biophys. Acta 301, 71-104
284 Baltscheffsky, H. and Baltscheffsky, M. (1974) Annu. Rev. Biochem. 43, 871-897
78

285 Biesecker, G. (1973) Biochemistry 12, 4403-4409


286 Hucho, F. and Changeux, J.-P. (1973) FEBS Lett. 38, 11-15
287 Laver, W. G. and Valentine, R. C. (1969) Virology 38, 105-119
288 Wrigley, N. G., Skehel, J. J. Charlwood, P. A. and Brand, C. M. (1973) Virology 51,525-529
289 Dudai, Y., Herzberg, M. and Silman, 1. (1973) Proc. Natl. Acad. Sci. U.S. 70, 2473-2476
290 Powell, J. T., Bon, S., Rieger, F. and Massoulie, J. (1973) FEBS Lett. 36, 17-22
291 Emmelot, P., Feltkamp, C. A. and Vaz Diaz, H. (1970) Biochim. Biophys. Acta 211, 43-55
292 Evans, W. H. and Gurd, J. W. (1972) Biochem. J. 128, 691-700
293 Goodenough, D. A. (1974) J. Cell Biol. 557-563
294 De Robertis, E., Azcurra, J. M. and Fizzer, S. (1967) Brain Res. 5, 45-56
295 Cotman, C. W., Levey, W., Banker, G. and Taylor, D. (1971) Biochim. Biophys. Acta 249,
406-418
296 Yu, J., Fischman, D. A. and Steck, T. L. (1973) J. Supramol. Res. 1,233-248
297 Hall, J. D. and Crane, F. L. (1972) Biochim. Biophys. Acta 255,602-619
298 Salton, M. R. J., Freer, J. H. and Ellar, D..1. (1968) Biochem. Biophys. Res. Commun. 33,909-
915
299 DePamphilis, M. L. and Adler, J. (1971) J. Bacteriol. 105, 396-404
300 Schnaitman, C. (1971) J. Bacteriol. 108, 545-552
301 Schnaitman, C. (1971) J. Bacteriol. 108, 553-563
302 Klenk, H. D. (1973) Biological Membranes (Chapman, D. and Wallach, D., eds), Vol. 2, p. 145,
Academic Press, London
303 Lenard, J. and Compans, R. W. (1974) Biochim. Biophys. Acta 344, 51-94
304 Wagner, R. R., Schnaitman, T. C., Snyder, R. M. and Schnaitman, C. A. (1969) J. Virol. 3,
611-618
305 Skehel, J. J. (1971) Virology 44, 409-417
306 Sch~fer, R., Hinnen, R. and Franklin, R. M. (1974) Nature, 248, 681-683
307 Loach, P. A., Sekura, D. L., Hadsell, R. M. and Stemer, A. (1970) Biochemistry 9, 724-733
308 Fowler, L. R., Richardson, S. H. and Hatefi, Y. (1962) Biochim. Biophys. Acta 64, 170-173
309 Hatefi, Y., Haavik, A. G., and Grittiths, D. E. (1962) J. Biol. Chem. 237, 1676-1685
310 Hatefi, Y., Haavik, A. G. and Jurtshuk, P. (1961) Biochim. Biophys. Acta 52, 106-118
311 Kirkpatrick, F. H., Gordesky, S. E. and Marinetti, G. V. (1974) Biochim. Biophys. Acta 345,
154-161
312 Springer, T. A., Strominger, ,1. L. and Mann, D. (1974) Proc. Natl. Acad. Sci. U.S. 71, 1539--
1543
313 Liljas, L., Lundahl, P. and Hjerten, S. (1974) Biochim. Biophys. Acta 352, 327-337
314 Kuboyama, M., Yong, F. K. and King, T. E. (1972) J. Biol. Chem. 247, 6375-6383
315 Sun, F. F., Prezbindowski, K. S., Crane, F. L. and Jacobs, E. E. (1968) Biochim. Biophys. Acta
153,804-818
316 J6rgensen, P. L. (1974) Biochim. Biophys. Acta 356, 36-52
317 Kagawa, Y. (1974) Methods in Membrane Biology (Korn, E. D., ed.), Vol. 1, pp. 201-269, Plenum
Press, N.Y.
318 Vernon, L. P., Shaw, E. R. and Ke, B. (1966) J. Biol. Chem. 241, 4101-4109
319 MacLennan, D. H., Seeman, P., lies, G. H. and Yip, C. C. (1971) J. Biol. Chem. 246, 2702-2710
320 Baskin, R. J. (1971) J. Cell Biol. 48, 49-60
321 Simons, K., Garoff, H., Helenius, A., K~i~iri~iinen, L. and Renkonen, O. (1974) Perspectives in
Membrane Biology (Estrada, O.-S. and Gitler, C., eds), pp. 45-70 Academic Press, New York
322 Gotto, A. M., Levy, R. I., Lindgren, F. T. and Fredrickson, D. S. (1969) Biochim. Biophys. Acta
176, 667-669
323 Petit, V. A. and Edidin, M. (1974) Science 184, 1183-1184
324 Bruckdorfer, K. R. and Green, C. (1967) Biochem. J. 104, 270-277
325 Van Deenen, L. L. M. (1968) in Regulatory Functions of Biological Membranes (J~irnefelt, J.,
ed.), pp. 72-86. Elsevier, Amsterdam
326 Chen, Y. S. and Huddell, W. L. (1973) Exeptl. Eye Res. 17, 517-532
327 Rogers, M. J. and Strittmatter, P. (1973) J. Biol. Chem. 248, 800-806
328 Kreibich, G., Hubbard, A. L. and Sabatini, D. D. (1974) J. Cell Biol. 60, 616-627
329 Becher, P. (1967) Nonionic Surfactants (Schick, M. J., ed.), pp. 604-626, Marcel Dekker, New
York
79

330 Rohm and Haas, Handbook of Physical Properties of Surfactants CS - 16 G/cd, Rohm and Haas
Co., Philadelphia
331 Mysels, K. J. and Princen, L. H. J. (1959) J. Phys. Chem. 63, 1696-1700
332 Tartar, H. V. and Lelong, A. L. M. (1955) J. Phys. Chem. 59, 1185-1190
333 Philippoff, W. (1951) Discuss. Faraday Soc. 11, 96-107
334 Debye, P. (1949) Ann. N.Y. Acad. Sci. 51, 575-592
335 Starofi, K. and Kaniuga, Z. (1974) Acta Biochim. Polon. 21, 55-60
336 Dunnick, J. K., Marinetti, G. V. and Greenland, P. (1972) Biochim. Biophys. Acta 266, 684-694
337 Jain, M. K. (1972) The Bimolecular Lipid Membrane, pp. 209-215, Van Nostrand Reinhold,
New York

You might also like