close

Вход

Забыли?

вход по аккаунту

?

Structural Principles of Biomembranes.

код для вставкиСкачать
E
Volume 11 Number 7
July 1972
Pages 551-652
a
International Edition in English
Structural Principles of Biomembranes[**]
By Werner Kreutz“]
Dedicated to Professor 0.Kratky on the occasion of his 70th birthday
The determination of the electron density distribution of the membrane cross sections of nerve
myelin and the discs of the outer segments of retinal rods by a new X-ray evaluation method
indicates, as in the case of the cross-sectional structure of the photosynthetic membrane, an
“asymmetric” mass distribution. In connection with the detection of planar protein lattices in
the membranes of photosynthetic bacteria, the chloroplasts of higher plants, erythrocytes,
and retinal discs as we11 as the detection of a new category of lipid characteristics, comprising
isothermal conformational and phase changes by specific ligand binding, a basis is established
for the development of a general concept of biomembrane structure.
1. Introduction
The phenomenological capabilities of biomembranes comprise an impressive range of functions, such as nerve excitation, sensory transduction, gland functions, energy
transformation (photosynthesis and respiration), osmotic
regulation, etc. This quite apart from the fact that, as purely
morphological elements, they are responsible for the compartmentation of “internal milieus” and thus create one
of the essential prerequisites for life per se. For membranologists this fascinating complexity of biomembranes represents a great challenge, namely that of understanding
this functional variety on the molecular level. It is this
p] Prof. Dr. W. Kreutz
Institut fur Biophysik und Strahlenbiologie der Universitat
78 Freiburg, Albertstrasse 23 (Germany)
[“I The following abbreviations have been used:
DGD = digalactosyl dlglyceride
MGD = monogalactosyl diglyceride
PA
= phosphalidic acid
= phosphatidylcholine
PC
PE
= phospharidylethanolamine
PG
= phosphatidylglycerol
PI
= phosphatidylinositol
PS
= phosphatidylserine
Angew. Chem. internat. Edit. 1 Vol. I 1 (1972) 1 No. 7
aspect as manifested in the dynamic molecular architecture,
in the molecular characteristics, and in molecular interactions with which this progress report is concerned.
The following four questions outline the essential problems
which confront us on the molecular level :
1. What is the steric state (conformation) of the proteins
and what is their planar arrangement within the membrane? Are there variable protein states?
2. What is the conformation of the lipids and what is their
planar distribution? Do variable lipid conformations
exist?
3. What interactions occur between proteins and lipids?
What is their steric relationship (coupling)?
4. What possibilities of structural variation and what
physicochemical capabilities associated therewith are
encountered in the molecular membrane architecture
as a whole for the accomplishment of the various functions such as e.g.: ensuring a specific permeability,
regulation of permeability changes, passive-, facilitated-,
and active transport mechanisms and their control,
generation of concentration gradients or potentials,
regulation of potential changes (e.g . light-induced or
551
chemically induced changes), transmission of excitation,
potential-linked chemical syntheses (phosphorylation),
redox reactions, and probably also information storing
capacities.
Our knowledge gained hitherto indicates that the extensive
variability of membrane functions derives from a flexible
"basis-membrane'' whose versatile structural adaptability
permits the association of various special functional proteins
on both sides of the membrane and thus makes vectorial
processes feasible in both directions. It therefore appears
reasonable to distinguish between a "basis membrane"
and a "function membrane". With this in mind, it will be
our initial endeavor to develop a general concept of the
basis-membrane structure. If we are to proceed in this
manner, then the stereochemical and physicochemical
characteristics of the basis-membrane proteins and of the
lipids will have to be discussed as a first step. An exposition
of the qualities of the two groups of substances will provide
the pre-conditions essential for an examination of the
possibilities of coupling, i. e. the joining of these substances
to form a membrane. The second part of the paper pursues
the conversion of the basis membrane into the function
membrane and an attempt is made to explain some functional capabilities on the basis of this structural concept.
3. Lipids can undergo phase transitions induced by ligand
binding" 1' .
4. The miscibility of lipids with each other is highly selective[14.151
In view of these qualities one is led to speak of an intrinsic
lipid dynamic behavior. In considering these intrinsic
dynamics we must also allow for transverse and lateral
lipid movements in .lipid layers['61 as well as for cationinduced lipid-organization changes' ''].
2.1.1. Conformational States and Micelle Types
The steric characteristics and the miscibility behavior of
lipids can be examined by using low angle X-ray diffraction
methods. Different conformational states and micelle types
produce very distinct X-ray diffraction patterns. Further
information can be obtained by quantitative titrations of
the polar groups['*' in different conformational states, by
ESR spin labeling["], NMR measurements['21,enzymatic
fragmentation, and measurements of calorimetric and
kinetic data of conformational changes[''.
2. The "Basis Membrane"
2.1. The Intrinsic Dynamics of Lipids
Hitherto, in the assessment of membrane phenomenology
functional predominance has been attributed to the proteins. The reason for this lies in the fact that the lipid
characteristics so far established did not allow an appreciable functional variance. In this respect it will be
necessary in future to make a fundamental adjustment.
The physico-chemical data available today are essentially
the result of observations of p,?lphase transitions between
laminar cubic and hexagonal lattices['- 31, of measurements
of surface potentials, of surface pressure or surface tension[4.51, as well as surface requirement of monomolecular
lipid layers on an aqueous phase (summary : [6]).To these
must be added the determination of permeability data and
electrical qualities of bimolecular lipid layers" and
also ESR-I' 'I and NMR measurements[''] with lipids.
Furthermore, it has commonly been accepted that phosphoand galacto-lipid molecules occur in only one conformational state, i. e. the one shown in Figure l a. (There is only
one exception : the lipid conformational state postulated
by F i n e ~ n " ~in' his model of myelin-membrane structure.)
It is as the result of the following qualities that the lipids
acquire actual functions :
1. In principle lipids are able to exist in two different conformational states[14](Figs. l a and 1b).
2. These conformational states can be transformed from
one into the other by ligand binding. In other words,
lipids can undergo chemically induced conformational
changes" '1.
552
Fig. 1. a) p,,-conformational state of lipids represented by a StuartBriegleb model and by the corresponding symbol used further on in
this paper, b) representation of the p,,-conformational state.
The essential difference between the two conformational
states of the lipids can most probably be ascribed to the
fact that in the one case the polar group is located outside,
and in the other it is incorporated in the fatty acid domains
(Figs. 1a and 1b). In the following, the conformations shown
in Fig. 1 will be referred to as the p,,-conformation (polar
group external conformation) and the pi,-conformation
(polar group internal conformation). These two conformational states permit the formation of laminar and tubularshaped lipid superstructures shown in Fig. 2. So far we
have observed structures a, b, c, f of Fig. 2, but not e; d is
questionable. It would appear that laminar micelles can
be formed both by lipids in the p,,-conformation and by
lipids in the pi,-conformation, whereas the tubular structures, which are always arranged in hexagonal lattices,
can only be formed by lipids in the pi,-conformation.
All saturated lipids, i. e. lipids comprising saturated fatty
acids, occur at room temperature exclusively in the pexconformation. This can be demonstrated very simply by
Angew. Chem. internat. Edit. / Vol. I1 (1972) / No. 7
hydrogenating unsaturated natural lipids (n-lipids) thereby
transforming them into saturated lipids (s-lipids).All lipids
of this category so far studiedc3]form laminar structures at
room temperature, and accordingly reveal periods corresponding to twice the length of a fatty acid plus the polar
groups in X-ray experiments (Fig. 2a)"41.
Fig. 2. a), b), c), d), e) conceivable laminar structures of lipids in the pi.and p,,-conformation, f) cross section of a tubular lipid superstructure
formed by four lipid molecules in the pi,-conformation.
On the other hand the natural lipids which are partially or
completely unsaturated exhibit a much more differentiated
behavior. This category includes molecules of a static
nature, i. e. which occur at room temperature invariantly
in either the pex- or the p,-conformation, and dynamic
molecules which, depending upon physico-chemical parameters or ligand binding, acquire either the pex-or the pi,conformation. Thus, for instance, PC (phosphatidylcholine)
and DGD (digalactosyl diglyceride) represent typical static
molecules, whereas PE (phosphatidylethanolamine), PA
(phosphatidic acid), PS (phosphatidylserine), PG (phosphatidylglycerol), and PI (phosphatidylinositol) show a
dynamic behavior.
2.1.2. Phase Transitions Caused by Ligand Binding
In this case the term "phase transition" signifies the cooperative. transformation from one type of lipid lattice
into another. Temperature-dependent phase transitions of
lipids are well documented in the literat~re".~'and since
they take place at temperatures about 40°C"', they can
scarcely play a significant role in physiological processes
in viuo"]. Of great physiological significance, however,
could be the temperature-independent reversible phase
transitions which can be performed by PE. The phospholipid PE possesses the unique quality of accomplishing
temperature-independent phase transitions by bicarbonate
and carbonic acid binding['51. Since only bicarbonate and
carbonic acid are bound in the ratio 1:1, this process
also represents a pH-dependent phase transition. In
[*] The biological significanceof the very well regulated body temperature of 36-37°C in idiothermous animals could lie in the prevention
of irregular thermal phase-transitions.
Angew. Chem. internat. Edit.
Vol. I 1 (1972) N o . 7
other words, it constitutes a phase transition which is
indirectly controlled via the auxiliary system :
cNaf
H,CO,
-=t
+H+
NaHCO,
According to the apparent pK-value of carbonic acid at
pH=6.3 the range of control extends from pH>5.5 to
pH <: 7. In the pH range< 5.5 only tubular structures of
PE in hexagonal arrangements are found (Fig. 3). In the
range of regulation between pH= 5.5 and 7.0 the tubular
structures are transformed into lamellae depending on the
extent of bicarbonate and carbonic acid binding (max. at
pH=6.3). Our experiments also indicate that at pH>8
the carbonic acid system no longer has any ligand function
for PE, since in this pH-range PE again forms tubular
micelles. It is an interesting fact that in the said pHrange PE mainly binds Ca2+, its tubular micelles being
transformed into laminar ones in the process (Fig. 3).
Ca2+ and the carbonic acid system thus possess similar
competing ligand functions; dominance of one over
the other is in each case determined by a definite
pH-range. Both in the case of bicarbonate (pH=5.5-8)
binding and in that of Ca2+ binding (pH>7) PE occurs
in the pi,-conformation and forms symmetrical bimolecular
micelles of 44 .& or 45 .& thickness.
H.55-8
HNaCO,.H,CO,
pH.7
'
+~a'+
b
a
ip88D11
Fig. 3. Illustration of the bicarbonate- and Cazi-dependent phase
transition of PE between tubular hexagonal lattices and laminar linear
lattices.
A further significant characteristic is that Ca2+ is in part
also bound in the acidic pH-range by PE, the PE undergoing
transition from the pin- to the p,,-conformation and forming laminar bimolecular micelles that are 54 .& thick.
pH=3
pH=8
-C
lasso.41
Fig. 4. pH-dependent cross-section construction of tubular PE-micelles
consisting of five, four, and three PE molecules.
It is also noteworthy that in the absence of CO, and Caz+
the diameter of the tubular structures decreases as the pH
increases. It drops gradually from 48 -t 1 .& at pH z 3, via
43 1 .&, to 37 f1 A at pH = 8. Since the tubular diameters
are in a ratio of
@, the tubular cross sections
apparently consist of 5, 4, or 3 PE moIecules (Fig. 4)[151.
fl:fl:
553
2.1.3. pH-Dependent and Ligand-Regulated Conformation
Changes
In contrast to PE, which is outstanding for its ability to
perform regulated phase transitions in which its conformational state essentially remains unchanged, PA is distinguished by the unique ability to undergo pH-, C0,-, and
Ca2+-dependent conformational changes while maintaining the same laminar micelle type"51. A decisive parameter
in the conformational transition of PA is the pK-value of
the second dissociation step of the phospho group at
pH 7. In the acidic pH-range PA is found in the p,-conformation, in the alkaline pH-range it exists in the pex-conformation (Fig. 5).
mm ,I
!I?
1
Fig. 7. Amino-acid induced conformational changes of PS. a) Alanine
(0)
transforms the p..-conformation quantitatively into the pi,-conformation, b) other amino-acids ( 0 )are associated with the per-conformation.
Fig. 5. pH-dependent conformational and micelle change of PA: The
pi,-micelles are converted into p,,-micelles. By the addition of bicarbonate at p H > 7 the p,,-micelles are partly transformed into pi.micelles.
In the presence of CO, both conformational states occur
in about equal proportions in the alkaline range. On
addition of Ca2+ throughout the entire pH scale, PA is
quantitatively transformed from the pin- into the pex-conformation (Fig. 6). In the acidic pH-range two different
kinds of laminar micelle formation occur (Fig. 6a) (in
varying proportions), whereas in the alkaline range only
one laminar micelle type is formed (Fig. 6 b). In conclusion
it can be said that protonation of the secondary PO- group
of PA favors the pi,-conformation, whereas the binding of
Na' and Ca2+ stabilizes the p,,-conformation.
Fig. 6. Ca' +-dependent conformational and micelle changes of PA :
in the acidic range the p,,-micelles are converted into two types of pexmicelles, in the alkaline only one type of p,,-micelle is formed.
PS shows neither the qualities of PE nor those of PA.
Throughout the entire pH-scale, PS is found in the pexconformation. Nor can this conformational state be
modified by addition ofCa2+-ions.However, PS is provided
with another special property : The p,,-conformation can
be totally converted into the pin-conformation by the addition of the amino acid alanine (Fig. 7a). This conversion
cannot be achieved in such a perfect manner by other
amino acids. Most amino acids are associated with the
p,,-conformation (Fig. 7 b)["l.
554
Comparable specific qualities are also to be anticipated
in the case of PI and PG. However, in both instances no
final results are yet available.
2.1.4. The Miscibility Behavior of Lipids
If phospholipids are mixed in organic solvents, dried by
evaporation, and studied at room temperature by X-ray
diffraction methods the following overall picture of lipid
behavior is obtained : n-PC, n-PG, and sphingomyelin
form laminar micelles, whereas n-PA, n-PE, and n-PS
form tubular micelles. Tubular superstructures are also
yielded by the combination PS/PA. A common characteristic of mixed micelle formation is that the mixing partners
adapt their conformational state so that mixed micelles of
homogeneous conformation are
' I.
In an aqueous environment, lipids show a far more complex
behavior. This is already evident from the above description
of the behavior of the pure lipids. The examination of
aqueous lipid mixtures is rendered more difficult by the
tendency of such aqueous suspensions to produce very
diffusex-ray diagrams, thus limitingevaluation possibilities.
The data thus far obtained indicate that lipids are partly
miscible and partly immiscible under such conditions. A
decisive criterion for miscibility or immiscibility would
appear to be an identical or divergent ligand-governed
conformational state of the mixing partners. Immiscibility
implies that the various lipid types are capable of forming
separate domains within the lipid layer of biomembranes.
Such a mosaic structure of the lipid layer may be regarded
as an essential precondition for the possible performance
of cooperative conformational and micellar transformations in lipid domains of identical population.
These viewpoints should be of particular interest with
regard to the regulation of permeability in biomembranes.
However, these problems cannot be considered exclusively
from the lipid aspect. The lateral distribution of lipids
must be decisively dependent upon the superstructure
of the protein layer. A knowledge of the protein layer
structure therefore seems to be indispensable for any
further discussion of all these questions.
Angew. Chem. internat. Edit. 1 Vol. II (1972) N o . 7
2.2. The Membrane Proteins
Each membrane is equipped with a special set of proteins
according to its specific functional role. Some of these
proteins are loosely associated with the membrane, whereas
by far the larger part is firmly bound to the membrane or,
in other words, actually constitutes part of the membrane
itself. We do not propose to deal at this stage with the loosely
attached proteins-they will be discussed in connection
with the function membrane (Section 3)-but will limit
ourselves to a consideration of the membrane-bound
proteins. These have hitherto been referred to as “structural
protein” in the literature. However, this general meaning
of the term “structural protein” is too complex for the
purpose of our discussion. At this point we only wish to
deal with that part of the structural protein which can be
defined by purely morphological physicaI criteria and
which is distinguished by three main characteristics :
1. Its capability of forming two-dimensional lattices ;
2. its particular solubility behavior;
3. its high volume occupancy.
The most notable quality is the first one. Therefore, we will
henceforce designate this part of the structural protein
“2d-lattice protein”. The interrelationships between the
“2d-lattice protein” and the structural protein will be established in Section 2.2.2.
2.2.1. The Tertiary Structure of the 2d-Lattice Protein
The best known 2d-lattice protein is that of the photosynthetic membrane. Moreover, this protein is the only
one for which it has been possible to calculate a Gauss
synthesis of the electron density distribution of the subunit
(protomer) from X-ray data[142
19]. For reasons which
cannot be explained in detail here, this Gauss synthesis is
still imperfect. Nevertheless, it is adequate to indicate
certain basic structural features.
A clearer picture is obtained if the amino acid composition
is taken into account. Based on the amino acid determinaand Baileyrz3’,
tions of Weber[”], Menke[”], Criddle[221,
the subunit should consist of 94+2 amino acids (Table 1).
Table 1. Amino-acid composition of one lattice-protein subunit. The
subunit consists of 94 k 2 amino acids.
Amino acid
moles of amino
acid per subunit
Asp-NH,
Thr
Ser
Pro
Glu-NH,
GlY
Ala
Val
Cys-SH
Met
Ile
Leu
TY r
Phe
LY s
His
Arg
TrP
7.45
4.07
4.80
5.08
8.31
9.35
8.17
7.0
1.o
1.46
3.76
9.25
1.51
5.19
3.2
0.92
3.16
1.22
Amide
8.83
The sum of these amino acids represents a molecular weight
of about IOOOO for the subunit shown in Figure 8, i. e. 5000
per ring. The same result is obtained in estimating the
molecular weight on the basis of the calculated mass
distribution (Fig. 8).
Fig. 8 illustrdtes the extent of our information so far
gathered from the electron density maps[14v19]. A possible
interpretation is that two ring-like protein molecules,
whose mass is unequally distributed between the upper
and lower horizontal section, are arranged orthogonally
to each other to form a subunit. Although the “ring”
dimensions amount to 27 and 36 only, they cover the
whole subunit plane of 42 x 93
the only possible explanation is that linking side chains must be present between the two rings and between the subunits.
A
A
A;
Fig. 9. Three possible conformations of a polypeptide chain consisting
of 94 amino acids which are in agreement with the X-ray data.
\L
/-
93 H
+\
Fig. 8. Schematic representation of the mass distribution of the tertiary
structure of the 2d-lattice protein subunit.
Angew. Chem. internat. Edit. 1 Vol. II (1972j 1 No. 7
In principle there are three conformational variations
which might be in accordance with the X-ray data. A polypeptide chain consisting of 94 amino acids would suffice
to form two separate rings (Fig. 9a), a closed figure-eight
structure (Fig. 9 b) or an open figure-eight structure (Fig. 9c).
555
The first two structures possess no amino acid end groups.
The lowest identifiable molecular weight would be SO00
in the case of Figure 9a, and 1OO00 if the configuration of
Figure 9b applies. The configuration of Figure 9c has two
end groups and a molecular weight of IOOOO.
A combined end-group and molecular-weight determination could ascertain which of these three structures is the
correct one. According to Menke12*1the number of Cterminal amino acids amounts to 13; however, other
a ~ t h o r s [ ~ have
~ - ~ found
~l
no amino acid end groups,
so that at present this problem remains unsolved. However,
for the purpose of a discussion of the membrane structure
as envisaged in this paper such details concerning the
tertiary structure are of little importance.
2.2.2. The Aggregation Behavior of the "2d-Lattice Protein"
Of predominant significance is the aggregation behavior
of the protein subunit shown in Fig. 8, i.e. the quaternary
structure or the superstructure of the 2d-lattice protein. In
principle the margin of variation in aggregation is restricted
by the circumstance that only a limited number ofsymmetry
operations is admissible in planar protein lattices. We
would stress that here, as in other proteins, only L-amino
acids occur.
potential number of planar space groups from 17 to 5,
since all space groups with coplanar two-fold rotation- and
screw axes are thereby excluded.
1
Fig. 11. Two-dimensionallattice of primitive polarity: One asymmetric
protomer type forms an infinite 2d-lattice. The two thin lines and the
bold one arranged at right angles within each protomer signify the
position of the peptide rings.
Finally, allowing for tangential polarity, the transverse
rotation axes are likewise eliminated so that out of the
potential 17 space groups only one remains, namely that
of the oblique planar lattice type as illustrated in Figure 11.
This lattice type then would constitute the most complete
but also the most primitive polarity possible: One single
asymmetric protein subunit (protomer) would homogeneously cover the total membrane plane. However, this type
of polarity does not seem to be desirable in biomembranes,
since it would not allow any variability. The X-ray and
electron-microscopic data of membranes so far available
rather indicate that biomembranes are to be viewed as
associations of linear superstructures which themselves
may consist of a homogeneous distribution of subunits.
a
A second possibility is that two or more oligomeric superstructures may be arranged alternately in a linear pattern.
Naturally, such a planar membrane design requires a
higher number of different protomers (subunits). It is,
however, quite admissible that such a variety is attributable
to minute deviations in the primary structure of one and
the same type of protomer. The photosynthetic membrane
may serve to demonstrate this principle.
Fig. 10. a) Lamella with a twofold coplanar rotation axis, b) lamella
without a coplanar rotation axis.
From the outset this very fact excludes symmetry centers
and mirror planes. The only possible symmetry elements
are limited to translations, transverse rotations (i. e. rotation axis vertical to the plane), and two-fold coplanar
rotation- and screw axes (Fig. 10a). The combination of
these symmetry elements permits the definition of 17 space
groups[25!
In the case of biological membranes two essential restrictions in respect of possible symmetry operations are presupposed : it is alleged that biomembranes possess transverse and coplanar polarity. From our above discussion
concerning the cross-sectional structure of the protein
subunit it emerges that the 2d-lattice protein itself also
possesses the property of transverse polarity. As regards
the coplanar polarity of membranes it appears that this
quality can only exist if it is provided by a two-dimensional
lattice. On the other hand, two-dimensional lattices can
only be imagined when formed by proteins. Therefore, we
may also envisage tangential polarity in the protein layer.
Transverse asymmetry (Fig. lob) already reduces the
556
To permit a better understanding we must first of all
explain that in the photosynthetic membrane of higher
plants two types of light reactions or two photosynthetic
systems (electron flow systems), known as system I and
system 11, cooperate. On the other hand, in the case of the
photosynthetic bacteria (e.g. rhodopseudomonas uiridis)
only system I applies. A comparison of the structure of
these two objects is most appropriate to demonstrate how
on the molecular level a relatively simple structure, namely
that of the bacteria, can develop into a complex structure.
designed to couple two systems. An electron-microscopic
picture of the planar structure of a bacterial photosynthetic
membrane recorded by Giesbrecht'261reveals a homogeneous protein structure with two principal directions (Fig. 12).
These two privileged directions enclose an angle of ~ 5 6 "
and are accounted for by protein double strands 90 8, wide
forming the membrane plane.
Further evidence is furnished by X-ray diagrams of the
same object["]. Two main interferences emerge which
correspond to Bragg spacings of 77 and 51 A. These interferences may be interpreted by means of the linear structure
shown in Figure 13. The interferences corresponding to 77
and 51 A and the angle of cx=56" (see Fig. 13a) indicate
Angew. Chem. internat. Edit. Vol. I 1 (1972) No. 7
chains to about 170 A. The X-ray diagram of such objects
consists of very strong interference corresponding to 41 A,
a strong interference corresponding to 31 A, and a very
weak interference corresponding to 62 A which is, furthermore, considerably broader than the other two.
If for evaluation purposes one starts from the assumption
that both in bacteria and in higher plants system I possesses
a similar basic structure, then at least part of these reflections ought to be capable of interpretation by a modification of the lattice in Figure 13. The presence of a sharp 41 A
reflection and the absence of a 93 A reflection indicates
that in this case aggregation occurs in the x2 direction but
not in the x1 direction. Moreover, the existence of a 62
reflection points to the fact that the aggregation in the x2
direction forms an orthogonal lattice (Fig. 15). However,
while the sharp 41 A reflection provides evidence of a high
Fig. 12. Electron-microscopic plan view of thylakoids of rhodopseudomonas uiridis.The surfaces are characterized by a regular planar arrangement of proteins in which two predominant directions exist (arrows),
(After P. Giesbrecht [26].)
dimensions of the protein subunit of 41 and 93 A. The
existence of a planar structure, i. e. a structure possessing
many net-planes in the x,-direction as well as in the x2direction, can be ruled out because of the lack of a 41 Ainterference (Fig. 13a).
a
Fig. 14. Eiecrron-microscopic picture providing the plane view of a
thylakoid of higher plants showing linear superstructures. (After R . B.
Park and N . G . Pon [28].)
aggregation state in the x2 direction, the broad 62 A reflection indicates that the aggregation breaks off after 4-5
periods. Although these findings may appear paradoxical
we are able to furnish below a plausible explanation.
Fig. 13. Lattice arrangement which explains the occurrence of a 77 and
a 51 reflection. Lattlce b differs from a in the lack of the 41 periodicity. b corresponds to the real diffraction pattern.
A
A
A linear structure of the type shown in Figure 13b is, on
the other hand, only plausible providing it is equipped with
2 twofold transverse rotation axes. In other words, this
would presuppose preferential coupling between two rows
of protein subunits. If we allow for the mass distribution
of the structural protein subunit described earlier in the
paper we come to a superstructure of the lattice protein of
bacteria as depicted in Figure 13b.
Fig. 1.5.Planar arrangements of protomers of the photosynthetic system
I giving an explanation for the 62 reflection.
We can now consider the construction principle of the
lattice protein of the photosynthetic membrane of higher
plants. Here also electron-microscopic pictures[2s-301,e. g.
that in Figure 14, reveal linear superstructures (chains),
which, however, possess quite different dimensions from
those in bacteria. The lateral distance between the chain
axes amounts to about 190A, and the period along the
To this end we must first again touch upon the 31 A reflection. Since it cannot be assumed that system I (Fig. 15)
produces a 31 A reflection, it seems justified to impute the
said reflection to system 11. There are two possible forms
of aggregation of the protein subunit in the x2 direction
which could account for a 31 A interference, the one being
Angew. Chem. internat. Edit. 1 Vol. I I (1972)/No.7
557
polar and the other nonpolar (Figs. 16a and 16b). In the
case of the polar structure in Fig. 16a there is no evidence
as to why there should be a trend for a linear aggregation
in the x2 direction. One would far sooner expect there to
be a two-dimensional association of any size, as shown in
Figure 11. On the other hand, the association in Figure 16b
satisfies the criteria for a linear structure and is therefore
to be regarded as the most likely one. In addition to the
31 .& reflection, the lattice of Figure 16b should also produce
a 41 A reflection, the shape of which depends on the size
of the linear aggregation in direction x2.
We are now faced with the question of the coupling of
systems I and I1 and the relative proportions of the two
systems. As already pointed out, coupling can only take
place under specific symmetry conditions, the ratios of the
two systems being determined either by a definite number
of different protomers, i. e. through the formation of finite
associations (oligomers),or else through a definite ratio in
the production rates of the subunits of system I and 11. In
the case under consideration it would appear that both
possibilities apply. System I is turned into a finite structure
314
.
\
m
a
b
\
'
w
Fig. 16.Two possibilities for the lanar distribution of system I1 protomers which can produce a 31 reflection. Configuration b is considered to be the true one.
by the input of several different protomers; however, the
ratio between system I1 and I is determined by the production rate of each system.
Besides the X-ray criteria for the existence of a finite structure of system I (broad 62 A reflection) and a fundamentally
infinite structure of system I1 (sharp 31 A reflection), there
are also criteria provided by chemical experiments: It is
possible to isolate system I oligomers from the membrane
and to define their structural characteristics, whereas the
remaining system I1 protomers join to give large superstructures, i. e. in the formal sense they constitute infinite
structure^'^'-^^^. X-ray investigations show that the
dimensions of the system I oligomers in the x1 and x2
directions are 164 and 186 A. Furthermore, we know from
chemical experiments that the proportion of system I
protein to system I1 protein is about l:2[3i1. Thus, in
conjunction with the aforementioned symmetry conditions,
the limits of the quaternary structure are traced.
From the dimension of the system I oligomers it can be
deduced that they consist of eight protomers. For these
oligomers to undergo coupling with the linear arrangements 11, they would have to be equipped as a minimum
with one twofold transverse rotation axis. On the other
hand, the definite size of the system I oligomers can only
be ensured if they consist of at least four different kinds
of protomers, i. e. a definite set of binding sites. The symmetry and binding conditions are therefore in conformity.
558
The outcome of these considerations is the quaternary
structure shown in Figure 17a, which as a whole will
produce a broad reflection at 62 A, a sharp one at 31 A,
and a very sharp one at 41 A. The pseudo-contradiction
concerning the sharpness of these reflections mentioned
above is thus explained. This quaternary structure consists of five different protomers, four in system I and one
in system I1 (Fig. 17b).
Fig. 17. a) Combined linear arrangement of system I and system I1
protomers which agrees with the X-ray diffraction data (41, 31, and a
62 A reflection), b) the same arrangement demonstrating the necessary
binding sites (number of different protomers) by different symbols.
In the case of an open tertiary structure as illustrated in
Figure 9c one might expect a maximum of ten different
terminal end groups in the 2d-lattice proteins. The ring
structure and the closed figure-eight structure shown in
Figure 9a and 9b would not allow for any end groups at
all. A decision between these alternatives is not yet possible
because of the aforementioned discrepancies in the amino
acid end-group determinations.
Support. for the general validity of these arguments is
forthcoming from the X-ray investigations of the discs of
rod outer segments (ROS) of the retina and the erythrocyte
membranes. In ROS-discs the lattice protein provides the
same interferences as were recorded in the case of photosynthetic bacteria, i. e. 77 and 51 .& reflections[341,and the
erythrocyte planar lattice produce X-ray reflections corresponding to Bragg spacings of 41 and 33 A.
The examples of the lattice protein configuration described
above give an indication of the scope of variation of possible
structural modifications. However, they also set limitations
upon the principal realization possibilities: The given
criteria allow for the existence of two-dimensional lattice
types with primitive polarity (Fig. Il), linear lattice types
with twofold rotation axes as shown in Figure 13b (ROSdiscs, photosynthetic membranes of bacteria), in addition
to more complex lattice types comprising several protomers
and oligomer associations capable of performing most
complicated functional tasks (Fig. 17) (photosynthetic
membranes of higher plants). There is, however, one factor
which appears to be of fundamental significance: For all
these lattice types one single type of basic protomer, i.e.
one type with identical or almost identical tertiary structure,
suffices for every possible variation. In this context the
chemical experiments on structural protein referred to
earlier acquire particular significance since they have
demonstrated that the structural protein of the membranes
of various objects show only minute deviations in their
Angew. Chem. internat. Edit. 1 Vol. 11 (1972) No. 7
primary structure and that by the action of proteases or
detergents they can be decomposed to basis peptides with
a molecular weight of 5000, 10000, or 20000[20-24335-371.
This being so, what can be said about the required tangential polarity? The introduction of polarity into the anisotropic (linear) homogeneous structural elements obviously is brought about by the association of functional
proteins. This creates periodically distributed linear bipolarity. That is to say, energy fluxes, impulses, or actions
emanating from the polarity centers are propagated in a
bipolar fashion. In this manner radial energy distribution,
which would mean a waste of energy in this case, is prevented. A primitive polarity as shown in Figure 11 would permit
an energy flux in only one direction, which would be less advantageous. In the more sophisticated linear structures
consisting of several oligomers, strong polarity is also
created in the phase boundaries between each pair of
different oligomers.
A further feature of great importance for the construction
of membranes is the distribution of hydrophobic and
hydrophilic amino acids within the 2d-lattice protein
subunit. We have learned from the Gauss synthesis of the
cross-sectional structure of the protein unit that the ring
centers represent hydrophobic areas[”’. In addition, that
horizontal part of the ring with less mass would have to
contain at least a predominance of hydrophobic amino
acids. The latter statement derives from the sandwich
behavior of the lattice protein in aqueous solution and the
collapsing process in membranes dealt with in Section 3.2.
This unusual distribution of amino acids will serve to
clarify the fact that lipids are only associated with one side
of the protein-layer (a feature considered in Section 2.3).
And last, but not least, we must once again draw attention
to the truly remarkable mass distribution inside the lattice
protein subunit. The impression is gained that the aim is
not to fill up the subunit with protein mass (and consequently also the protein layer),but rather as far as possible
to keep the subunit free of protein mass. Such a structure
opens up interesting possibilities with respect to the
coupling of proteins with lipids and also of the lipid
distribution and lipid superstructure.
2.3. Protein-Lipid Coupling
The answer to the question of protein-lipid coupling in
biomembranes lies ultimately in their cross-sectional structure. It is only in the last decade that significant experimental
progress has been made in this field. The first speculative
concept advanced was that of a symmetrical membrane
cross section[3s! However, it is evident that biophysicaIIy
speaking such a membrane structure could give rise to no
activity whatsoever, i. e. it would represent the biologically
most unfavorable membrane. Indeed, our X-ray results
on three different membranes point to an asymmetric
membrane cross section. The most pronounced asymmetry
seems to exist in the photosynthetic membrane of higher
plants and the nerve myelin structure. On the other hand,
the results so far obtained from rod outer segments of
vertebrate retina indicate an alternate asymmetric memAngew. Chem. internuf. Edit. / Vot. I1 (1972) / N o . 7
brane construction. It is to be expected that erythrocytes
have the same type of membrane.
A fundamental asymmetry can be established in all membranes so far investigated by means of a fairly simple X-ray
experiment, i. e. by a comparison of the X-ray data obtained
in this experiment with electron-microscopic pictures of
the same membrane system. The electron-microscopic
pictures of membrane systems reveal that membranes
always form closed vesicles. In flattened vesicIes (e.g. ROSdiscs, myelin membrane, photosynthetic membrane) two
membranes are therefore arranged mirror-symmetrically
and in this form give rise to the identity period observed
in the membrane system (Fig. 18a).
T
8
5
c
170 I!
P
rn
b
I
C
Fig 18 Relationship between electron-microscopic and X-ray periods
in lamellar systems a) Electron-microscopic cross-section view of a
flattened vesicle, b)X-ray period in the case ofa symmetrical membrane,
c) X-ray period in the case of an asymmetrical membrane
The X-ray beam “senses” electron-density projections of
these vesicles, such as for instance the electron-density
projection onto the stacking axes which is under consideration here. This means that if the electron-microscopic
periodicity is compared with that obtained by X-ray diffraction, the X-ray identity period would-in the case of a
symmetric membrane (Fig. 18b)-amount to half of the
periodicity visible on electron-microscopy. However, in
the case ofan asymmetric membrane cross section (Fig. 18a)
the X-ray and electron-microscopic periods would be
identical.
This comparison can best be made by using partly dehydrated membrane systems (up to 30% H,O) of flattened
vesicles, since the identity period then corresponds to the
thickness of a single or of a double membrane. It is possible
to make this comparison using chloroplasts of higher
plants, nerve myelin, and ROS-discs. Each of these three
membrane types yields anX-ray period which is comparable
with that found by electron microscopy, that is to say, it is
equal to or greater than twice a membrane thickness. (The
situation is somewhat more complicated regarding membranes forming spherical vesicles. This aspect will be dealt
with in Section 2.3.1.)
This result alone already justifies the statement that at least
these three membranes have an asymmetric cross-sectional
structure. Apart from this common characteristic, however,
there are considerable deviations with regard to the magnitude of the period or the thickness of the vesicle. The period
found in chloroplasts amounts in the dehydrated state to
150-170& in ROS to 210-230,&, and in myelin to
559
130-150 A. The
referred to in the
sections. Let us
structure of the
asymmetric one.
significance of these differences will be
detailed discussion on membrane cross
first of all discuss the cross-sectional
photosynthetic membrane, the most
2.3.1. The Cross-Sectional Structure of the Photosynthetic
Membrane
The structure of the photosynthetic membrane has been
comprehensively dealt with in an earlier paper['4]. We may
limit ourselves here to some essential aspects. The photosynthetic membrane of higher plants consists of three
layers which are distinguished by three different sections in
the electron-density diagram (Fig. 19a). The layer with the
highest electron-density has proved to be a protein layer,
that with the lowest density to be a porphyrin ring layer,
and the layer with medium density to be a monomolecular
1-
neutral lipids IMGDl
structural protein
1-
Porphyrin rings
Fig. 19. Structure of the photosynthetic membrane of higher plants.
a) Electron-density (p) distribution of the cross section of a vesicle
(thylakoid) in the living state, h) its interpretation (see designations),
c) plane view of one asymmetric membrane.
lipid layer (Fig. 19b). The mirror plane between the two
membranes of the vesicle (thylakoid) is so arranged that
the proteins form the outer layers.
The following points are essential for the discussion on the
protein lipid coupling in Section 2.3.2 :
1. Both chlorophyll a and b are not miscible with the
principal lipid component MGD, with constitutes about
two-thirds of the total lipids.
2. Phytol shows the same behavior.
560
3. Upon fragmentation of the membrane by detergents
the major part of the chlorophyll adheres to the protein~[~'].
4. In chlorophyll-protein complexes the lamellae are
about 12 A thicker than pure protein layers[31!
These results, which will not be further elucidated here,
support the concept that chlorophylls are anchored in the
2d-lattice protein by means of the phytol chain and that
the contact between protein and lipid is established by the
porphyrin rings, which apparently partially penetrate into
the unsaturated fatty acid zones of the lipid layer"41. An
over'ill view of the photosynthetic membrane of higher
plant> i< given in Figure 19c.
2.3.2. General View of Protein-Lipid Coupling
Protein-lipid coupling by means of chlorophyll cannot
apply to membranes containing no chlorophyll. However,
the question arises as to whether we are confronted here
with a special instance of a general coupling principle.
Remarkably, most membranes studied hitherto have
revealed a constant proportion of 30k 5% PC16].The only
exception encountered is that of the photosynthetic membrane which has yielded only 1-2% of PC. The other
phospholipids vary considerably from one membrane to
another, depending upon their function. Once again the
photosynthetic membrane provides an exception : The
phospholipids are extensively replaced by galactolipids
(MGD and DGD). Moreover, taking into account the fact
that PC shows the least conformation and phase variability
of all phospholipids, it emerges as the foremost candidate
for a special structural role.
On the basis of these lipid qualities and experimental
results obtained in respect of the photosynthetic membrane
we postulated a general principle concerning protein-lipid
coupling in bi~membranes['~]
: According to this postulation PC assumes the function of coupling between proteins
and lipids in membranes containing no chlorophyll. The
coupling is thought to be accomplished in such a manner
that the fatty acids of PC interact hydrophobically with
the peptide chains in the protein layer, while the choline
groups penetrate the fatty-acid zone of the lipid layer (Fig.
20a). The stability of this type of coupling is assured on the
one hand by the purely hydrophobic coupling between
fatty acids and protein and on the other by the polar interaction of the choline group of PC with the polar groups of
the coupled lipids (and the proteins). Maximum stability
of the polar interaction is achieved by the fact that the
interacting polar groups are embedded in hydrophobic
fatty-acid phases (Fig. 20b). This coupling principle will be
supplemented in this paper only to account for partial
assumption of the role of PC by sterols in some membranes.
2.3.3. The Cross-Sectional Structure of the Nerve Myelin
Membrane
In the past various phase combinations have been applied
by different authors for calculation of the electron-density
distribution of the myelin membrane cross section; consequently, different electron-density distributions were
obtained (for summary, see [39]). In the Iast year we have
Angew. Chem. internat. Edit. 1 Vol. 11 (1972) / N o . 7
likewise made a calculation of the electron-density distribution of the myelin membrane cross section[40! The evaiuation method used by us (the Q,-method), however, differs
fundamentally from that employed by other authors. Our
a
CI- CI-
protein side
Ca"
b
lipid layer. If it were possible to chemically label these
double bond planes, for instance with OsO,, our membrane
concept could be tested experimentally.A difference-Fourier
or difference-Q,-evaluation would then directly yield the
0 s distribution in the membrane. Systematic incorporation
experiments of this kind were first carried out by Parsons
and A k e r ~ [ ~The
~ ] Fourier-evaluation
.
of these experiments
by H ~ r k e r [resulted
~ ~ ] in the same 0 s distribution as our
evaluation via the Q,-function (Fig. 21 c)[,O1. These evaluations reveal that each membrane possesses two planes with
different 0 s concentrations and an 0 s enrichment between
the membrane pairs. The best access for OsO, molecules
is offered by the C=C bond layer within the protein layer,
T
36 1
I
PC
0.5 d
1
I
22 8,
I
I
I
I
I
I Protein 1
I
I
I
Lipids
I
I
PA, PS
lipid side
K'
@&@J
Fig. 20. General coupling concept of lattice proteins and lipids by PC.
a) Spatial situation at one ring of the subunit (compare Fig. 8), b) charge
distribution within such a coupling unit.
Q,-method, which was developed for and applied in the
structural investigation of the photosynthetic membrane['41,
directly yields an unequivocal electron-density distribution
from the intensity (and not as usual from the amplitude)
distribution by straightforward calculation, leaving out the
phase problem entirely. The electron-density distribution
obtained for nerve myelin membrane in vivo is shown in
Figure 21 a.
We have interpreted these resuIts as shown in Figure 21 b.
According to this concept the symmetry center of the
electron-density projection is flanked by the cross-sectional
profiles of mirror-symmetrically arranged lattice protein
layers. These layers are first covered on the outside by a
layer of the coupling choline groups and sterols, then by a
pure fatty-acid layer, and lastly by a layer of polar groups
of the lipids. In principle this structure corresponds to the
cross-sectional structure of the chloroplast thylakoids if one
replaces the chlorophylls by PC and sterols. There is,
however, one important distinction : In the myelin structure
the mirror plane is located between the two protein layers,
whereas in the case of the photosynthetic membrane
vesicles (thylakoids) it is positioned between the two lipid
layers. We will refer to this again later in Section 3.1.
According to our membrane concept of nerve myelin each
membrane possesses two C=C bond planes : one is formed
by the C=C bonds of the unsaturated fatty acids of the PC
located within the lattice protein, the other plane of the
fatty acid C=C bonds being situated within the actual
Angew. Chem. internat. Edit.f Voi. I 1 ((972) / No. 7
Fig. 21. Structure of the myelin membrane. a) Computer diagram of the
electron-density (p) distribution of the cross section of a myelin membrane pair, h) its interpretation : Each membrane is composed of an
inner protein layer and an outer lipid layer. The coupling is achieved
by PC and sterols (hatched). A water interspace has to be envisaged
between the two protein layers as well as between the two lipid layers,
c)computer diagram ofthe pure Osdistribution in the myelin membrane
pair, d) 0 s distribution indicated by diagram c. 0 s atoms stain the two
C=C bond planes within each membrane.
since the lattice protein is hydrated up to 70%. The action
ofthe OsO, molecules on these double bonds can be carried
out in an aqueous reaction medium. Therefore, the corresponding 0 s peak in Figure 21 c is the most dominant one.
561
The situation is quite different with the second plane of
C=C bonds. These double bonds are incorporated in a
hydrophobic environment, which makes penetration and
reaction of 0 s molecules more difficult. Accordingly, the
corresponding 0 s peak in Figure 21c is considerably
smaller. The 0 s distribution established within the membrane cross section agrees with our membrane concept of
nerve myelin and consequently also with our concept of
protein-lipid coupling in biomembranes.
A Fourier-transformation of the electron density distribution shown in Figure 21 a and 21 c provides the correspondingamplitude functions. They yield aphase set of + - - - for the five reflection amplitudes of the nerve myelin in uiuo
and a phase set of + - - + + for the five reflection amplitudes of the pure 0 s distribution. The mirror-symmetrical
phase set to the above phase set for nerve myelin in uiuo,
namely - + + + +, was already reported by Moody in
1963[421as one of two possible phase sets (and - + + - -). In the case of the 0 s distribution Harker14‘’
found the same phase set as we did. Harker proposes the
same phase set for the in uiuo structure as for the 0 s distribution. Interestingly, these phase sets, which come nearest to,
or are identical with, those determined unambiguously by
us, were gained by systematic experimental phase determinations.
++++
2.3.4. The Definition of the Basis Membrane
The calculation of the electron-density distribution of the
cross section of the photosynthetic and myelin membranes
has revealed totally asymmetric cross-sectional structures.
Based on the particular cross-sectional structure of the
photosynthetic membrane we were able to postulate a
general coupling principle for lipids and proteins in biomembranes. And finally X-ray investigations on the myelin
membrane have furnished support for this postulation. This
structural knowledge combined with the intrinsic qualities
of lipids and proteins and with the results of comparative
chemical investigations of membranes permits the definition of a basis structure for biomembranes.
ed by the two-dimensional lattice formed by the proteins,
while the coupling between the proteins and lipids is
accomplished by PC.An area measuring42 x 93 A2,covered
by a single protein unit, accommodates 7&80 lipid molecules depending upon the conformational state. For the
coupling of these molecules, 30 PC molecules are available,
15 per protein ring or 7-8 on each side of the rings. When
arranged side by side these 7 or 8 lipid molecules can extend
over a distance of ca. 35 A or 70 A, depending on whether
the fatty acids are assembled in single or double file. In
view of the fact that each structural protein ring has a
diameter of ca. 30 A, the double file assembly appears the
more probable. This suggests a structural configuration as
illustrated in Figure 20a. Each PC molecule couples one
other lipid molecule. Over an area of 42 x 93 A2 there are
thus left 40-50 uncoupled lipid molecules. These might
be expected to form one of the micelle structures shown
in Figure 2. Consequently the following possibilities for
the cross-sectional structure of the basis membrane may
be envisaged according to the type of uncoupled lipids
present in the protein interspaces (Fig. 22):
a) Basis membrane with lipid regions of monomolecular
symmetric construction and of homogeneous pex-conformation. This can apply to PG, PA, and PE (Fig. 22a).
b) Basis membrane with lipid regions of bimolecular symmetric construction and of homogeneous pex-conformation.
This micelle structure can be formed by PA, PS, or PI
(Fig. 22 b).
c) Basis membrane with lipid regions of bimolecular symmetric construction and of homogeneous pi,-conformation.
This micelle type could consist of PE or of PA (Fig. 22c).
d) Basis membrane with lipid regions of tubular construction. This configuration could be composed of PE
alone or of mixtures of PE with PA and PS (Fig. 22d).
The specific type of micelles involved will depend upon
the pH of the surrounding aqueous phase, whether CO,
is present, and lastly upon the concentration of different
cations (e.g. Na’ or Ca”). This brings us to the question
n
n
n
m
Fig. 22. Possible cross-sectional views of the basis membrane differing in their lipid interspaces
(see text).
We may define the basis membrane as that part of the
biomembrane which is exclusively constituted by 2d-lattice
protein and the membrane lipids. Its cross-sectional structure is totally asymmetric and its planar structure is govern5 62
of the influence of the aqueous phases with their different
constituents on the membrane structure. Section 3 deals
with these functional aspects of biomembranes.
Angew. Chem. internat. Edit. / Vol. 11 (1972)
/ No. 7
3. The Function Membrane
Before discussing actual functions of the membrane it is
necessary to complete the structure of the basis membrane
to give a function membrane. This completion may be
achieved in three ways : By a loose association of functional
proteins ; by the incorporation of non-proteinaceous substances ; and by the direct binding of prosthetic groups
(e.9. heme groups) to the lattice protein, thus yielding
modified or “converted lattice protein”.
functional, i. e. apart from the coupling function they also
have a role in the actual membrane process as characterized
by the coupled protein. A typical example would appear
to be the ATPase system. It is conceivable that ATP acts
n
n
n
3.1. The Structure of the Function Membrane
The category of proteins which are loosely associated with
the basis membrane comprises special functional proteins
such as the ATPases which are indispensable for active
transport of Na’ and C a 2 + or the rhodopsin of the ROS
membrane, etc. Some examples of non-proteinaceous substances of this category are NADP, ADP, acetylcholine,
etc. We assume that the other category of membrane-bound
functional proteins includes mainly converted lattice proteins (Fig. 23a). It is thought probable that they embrace a
large proportion of the redox enzymes, e. 9. the cytochromes
f and b of the photosynthetic membrane.
At this point it may be noted that the mixture of pure lattice
protein, converted lattice protein, and certain other coupled
proteins has up to now been referred to as “structural
protein”. The converted lattice proteins should not differ
essentially from the normal lattice protein ; they should
only vary slightly in their primary structure. Hence it is
not surprking to find that the “structural proteins” of
various membranes reveal only minute differences in their
amino acid composition, that they can all be chemically
fragmented in the same manner, and that they can be
decomposed to the same peptide size. So far the structural
proteins of the photosynthetic membrane[20-231
, mitochondria[’2*241,nerve myelin[36.371, the retina, and erythroc y t e ~ [ ’ 3~5.1 have been analyzed by these procedures.
The coupling of the first category of proteins to the basis
membrane may be achieved in two ways: on the protein
side (Fig. 23 b) or on the lipid side (Fig. 23c). The coupling
to the protein side may occur specifically to the proteins
or to the lipids, whereas on the lipid side the coupling can
only occur specifically to the lipids. The lipid specificity is
determined by the lipid composition of the membrane, and
by the circumstance that coupling is only conceivable in
those lipid regions which are not blocked by coupling PC
molecules, i. e. in lipid regions of symmetrical micelle
structure. These lipid regions are thereby converted from
a symmetrical to an asymmetrical structure. Thus, the
coupling of further addenda (e.9. proteins) to the basis
membrane results either in a totally asymmetric function
membrane (Figs 23 a and 23 b), in an alternatingasymmetric
membrane (Fig. 23c), or in a mixture of both types.
It is to be expected that a coupling molecule is involved in
the coupling of protein to the lipid side which is comparable
in its action to PC in the case of the lattice-protein coupling.
Presumably, however, these coupling molecules are biAngew. Chem. inrernat. Edit. 1 Vol. I 1 11972) 1 No. 7
.
-
Fig. 23. Three possibilities for the association of functional proteins to
the basis membrane. a) Conversion of a lattice protein subunit by the
binding of a prosthetic group (a),
b) coupling of an “other” protein
molecule to the protein side by a coupling molecule (hatched) or by
pure protein interaction, c) coupling of an “other” protein to the lipid
side by a coupling molecule.
as the coupling molecule between the lipids and the ATPase,
simultaneously serving as energy donor. An indication of
this is certainly provided by experiments in which A D P
and ATP were mixed with lipids[’51.
An example of the coupling of redox enzymes to the protein
side is to be found in the electron transport chain in photosynthesis and respiration. The photosynthetic membrane
therefore represents a typical asymmetric function membrane (Figs. 19, 23a, 23b). The same applies to the myelin
membrane. Both these membranes have been dealt with
earlier in this paper and only require completion in
conformity with Figures 24a and 24 b. Finally, it is necessary
to discuss the second type of function membrane. Our
experiments on ROS-discs have revealed that this membrane
is representative of this category.
The straightforward calculation of the electron-density
distribution of the ROS-disc membrane cross section cia
the Q , - f ~ n c t i o n results
~ ~ ~ ~ in the distribution function
shown in Figure 24a. [Recently, an electron-density distribution of the cross section of the ROS-discs was also calculated by Blaurock and W i l k i n ~ by
[ ~ ~choosing
]
a phase
combination that most closely approached a symmetrical
membrane cross section (‘‘unit membrane”). For a further
discussion, see [34].] The symmetry center of the electrondensity diagram of Figure 24a is flanked by two peaks of
high electron density followed by electron-density troughs
563
and then by two electron-density peaks of medium height.
This electron-density distribution is, in our opinion, to be
construed as illustrated in Figure 24 b. Unlike the chloroplast thylakoids but in accordance with the myelin
structure the lattice protein forms the two inner layers
(the layers adjoining the symmetry center) of the discs,
each layer being flanked on the outside by a monomolecular lipid layer. Furthermore, we assume that
to the outside of the lipid layers the rhodopsin molecules
are attached, which, together with the aqueous phase,
account for the electron density in this region.
0.5
d
i
b
found in those retina preparations in which the superlattice
is disintegrated. The disintegration can be caused by operations resulting in the decomposition of the lipid layer or by
light-induced uncoupling of the rhodopsin molecules[34].
Apparently the rhodopsin molecules are arranged in an
oblique planar lattice with plane distances of 154 A and
102 A which is indirectly determined by the lattice protein
on the opposite lipid side (Fig. 24d). These findings are in
agreement with the results of Blasie et al. who could
immunologically demonstrate that the rhodopsin molecules are arranged on the outside surface of the ROSThe ROS-disc membrane may, therefore, be
classified as an alternating asymmetric type of membrane
(Fig. 23c). For the sake of completeness it may be added
that the electron-density distribution of Fig. 24a corresponds to a phase set of + - - - - - - for the first seven
reflection amplitudes.
Concluding these considerations on the structure of
biomembranes it should be mentioned that the protein
lattice structures and thus also the defined planar structure
of the entire membranes discussed above are to be regarded
as exceptional crystalline states which can exist only
under special experimental conditions. I n uivo these structures should exist in para-crystalline or still more fluidal
crystalline states.
3.2. The Collapse of the Function Membrane
,Rhodopsin
d
Fig. 24. Structure of the ROS-disc. a) Computer diagram of the electrondensity distribution of the cross section of ROS-disc, b) its interpretation: two inner lattice protein layers flanked by lipid layers on the outer
sides. Rhodopsin molecules are coupled to PE in the pi,-conformation
by retinene, c) planar distribution of the lattice protein combined with
a rhodopsin superlattice, d) overall view of the ROS-disc membrane.
Support for the concept of the rhodopsin molecules being
attallied to the outside of the lipid layer is provided by
the following observations : Two different lattice types have
been identified within the plane of the retina
One
lattice characterized by Bragg spacings of 51 and 7 7 A ,
and the other by Bragg spacings of 154,102, and 77 A. The
second lattice represents a superlattice of the first one
(Fig. 24c). The first type oflattice has already been described
in connection with photosynthetic bacteria in Section 2.2
and was attributed to the 2d-lattice protein. It is always
564
A feature providing significant information concerning the
thermodynamic stability of function membranes and therefore also concerning the structure itself, is the phenomenon
of function membrane collapse. In actual fact experiments
carried out over the last ten years indicated the existence
of this phenomenon, but it had not been recognized as such.
The interrelationships became apparent when we recorded
X-ray diffraction patterns of the ROS membranes. What
happens is that in particular membranes of spherical vesicles, e. g. erythrocyte^^^'], photosynthetic bacteria[461,but
also those of m i t ~ c h o n d r i a ' ~
and
~ ] the flattened vesicles of
the ROS[34i,in the partially dehydrated state (20--30%
H,O), produce a 58 i2 reflection in the X-ray experiment
which is mostly accompanied by a 116+4 A reflection.
Evidently these reflections must relate to a structural
quality which is common to a wide variety of membranes.
The said reflections occur at the meridian when the three
first mentioned membranes are orientated by drying on an
object slide, i. e. they represent diffraction patterns analogous to those produced by membrane stackings. The
quantitative analysis of the diffraction patterns of the
photosynthetic membranes of rhodopseudoinonasspheroides
conducted by Paper4'] has revealed that the basic lamellar
unit is a 115 A thick double layer whose two asymmetric
sublayers are arranged in mirror symmetry.
In our X-ray experiments on dehydrated orientated ROSdiscs'341we also encountered the 51 A reflection but the
amazing fact was that this reflection definitely came from
the plane structure of the ROS-discs, i. e. the diffraction
effects as compared with the corresponding effects in other
membranes were swivelled through 90".
Angew. Chem. internat. Edit.
Val. I1 (1972) 1 No. 7
For an interpretation of these findings it is necessary to
divert briefly from the point. The original concept of the
molecular membrane structure was that of a symmetrical
membrane, i . e. a membrane construction of the highest
thermodynamic stability was intuitively postulated[38! This
holds true, for instance, in the case of the photosynthetic
membrane and the myelin membrane, however not within
one single membrane but in a mirror symmetrical arrangement
o f a membrane pair. It should therefore not be surprising
that these two membranes do not alter appreciably upon
dehydration.
The situation is different with regard to membrane systems
which form spherical vesicles or, in more general terms, in
the case of those membranes,whose two lipid layers cannot
adopt a mirror symmetrical arrangement, i. e. cannot
stabilize upon dehydration. This formulation also applies
to the ROS. In spherical vesicles contact between the lipid
layers is inhibited by the “filling” of the vesicles, whereas
in the case of the retina it is prevented by the rhodopsin
molecules attached to the lipid side. The result is that when
the water phase is eliminated upon drying, i. e. when the
is
thermodynamic basis of the “hydrophobic
eliminated, and when the monomolecular lipid layers of
the membranes are not stabilized by adjoining lipid layers,
a protein-lipid dissociation takes place followed by a
structural rearrangement of the proteins and the lipids
(Fig. 25). However, it is to be expected that at this hydration
stage of 20-30% water the PC molecules will still remain
attached to the lattice protein and that only those lipids
which form the proper monomolecular lipid layer of the
membrane are disintegrated. The original thickness of the
membrane is thereby reduced by about 25 A, i. e. from 70 A
to 45 A (45 A = 36-38 A lattice protein + 8 A towering
choline groups). A suspension of densely packed, flattened
collapsed membrane vesicles which are 20-30% hydrated
must contain 90 A thick membrane pairs in mirror-symmetrical arrangement spaced by 20-30 A water layers.
When drying is complete the stacking period will shrink
to about 90A (Fig. 25). Such a behavior was found by
HZO 30%
Lipid extraction will result in a further state of collapse, or,
more adequately expressed, state of denaturation : The
elimination of the PC molecules from the lattice protein
will be followed by a stabilization process of the pure lattice
proteins. This stabilization is achieved by a tilting away
of the protein subunits from the original lattice plane. In
spite of the loss of the PC molecules this process leads to a
thickening of the protein layer from 45 or 37 to about 55 A.
This membrane behavior is well established by electronmicroscopic observations by Fleischer and S t o c k e n i ~ s ‘ ~ ~ ~
as well as by X-ray investigation^^^^] (Fig. 25c, and 25c,).
In electron-microscopic procedures, employing staining
and alcohol dehydration, only collapsed membranes will be
pictured, i. e. the pictures represent protein layers stained at
their outer surfaces (Fig. 25d). There is no“unit membrane”
but a common appearance ofdenatured membranes. Sjoestrand
also comes to the same result in a comprehensive electronmicroscopic studyr5
’’.
The ROS-discs represent an exception in so far as the
orientating pressure is swivelled through 90” with respect
to other membrane systems in such dehydration experiments. Whereas a transversal pressure is applied in the
membrane systems discussed above, lateral forces have to
be considered in the ROS-discs (Fig. 25 b). In all X-ray
experiments a strong 56 A reflection appears when the 77
and 51 A reflections of the initial 2d-lattice protein disappear on dehydration, indicating a breakdown of the 2dlattice and the formation ofa new linear lattice. We suppose
that this 56 A lattice is formed by the tilting away of the
protein subunits from the lattice plane to arrange in a
“pickaback” structure (Fig. 25 b). (Blasie and Worthington
attributed the 56 A reflection to a liquid planar distribution
of rhodopsin molecules[531.)
3.3. Perspectives of Molecular Membrane Functions
On the basis of the membrane structure as projected in the
preceding sections new possibilities are opened up concerning membrane functions on the molecular level. As an
introduction to a consideration of this subject the following
three directive postulations should be premissed :
\
a
Finean et al. in X-ray experiments on mitochondria and
erythro~ytes[~~*~~!
1
1. The capability of the lipids to exist in two different con-
lipids
y i n i n g
-
formational states and the control of conformational
transitions by ligand binding may be regarded as the
carrier basis of facilitated and active transport processes.
J
d
v
Fig. 25. Representation of the collapse process and the origin of the
“unit-membrane” appearance in biomembranes. a) Collapse of membrane vesicles which are orientated by dehydration, b) collapse in ROSdiscs on dehydration, c) thickening of the protein layers by rearrangement of the subunits after total lipid extraction, d) stained collapsed
membranes: “unit-membranes”.
Angew. Chem. internat. Edit. 1 Vol. I 1 (1972) No. 7
2. The maintenance of a definite permeability and its adjustability is founded on controllable phase transitions
of phospholipids through binding or attachment of
ligands as well as on the different binding capacities of
individual lipids with respect to univalent and bivalent
ions.
3. Proteins have the task of solving stereospecificstructural
problems, they are responsible for the enzymatic regulation
of membrane processes, and in particular for governing
the energy transmission between the outside aqueous
phases and the membrane.
565
3.3.1. General Remarks on Carrier-Transport Mechanisms
The newly found intrinsic lipid characteristics point to the
possibility of carrier-transport mechanisms based on conformational changes of phospholipids. (A “carrier” is considered to be a molecule which is able to effect a mechanical
movement of substrate across the permeation barrier.) It
is thought that the diversity in the mechanisms is due to
the availability of different carrier molecules, and to the
adaptability of various induction and energy-input systems
to these individual carriers. In each instance, however, the
initiation of the conformational change occurs by means
of basically the same physico-chemical process, namely
ligand binding or ligand exchange.
Despite the basic similarity of the carrier mechanisms, two
categories are distinguishable from the energetic point of
view. In the one case the ligands are only weakly associated,
in the other they form a true chemical bond with the carrier.
In conformity with the divergent energy requirements there
are also two different types of energy supplies available.
The energy is provided either by an efflux (source flux), i. e.
by the maintenance of a concentration gradient, or in the
form of chemically stored energy (e.g. in the form of energyrich phosphate bonds).
The supply of chemically stored energy to the transport
process, i. e. the conversion into a vectorial mechanical
movement, is considered to be performed uia the
dissociation equilibria of the acid groups of the
carriers. Thus, for instance, the energy supplied by the
phosphoester bond of ATP enables exchange of protons
from the carrier phospho acid groups for Na’ or Ca2+.
The resulting alteration of intra- and intermolecular interaction energy results in a steric displacement of the polar
carrier group which represents the actual mechanical
transport process.
This type of mechanism naturally demands a considerably
greater effort, as regards the substrate coupling and energy
input directly at the membrane, than in the case of the
efflux-driven transport mechanism. These two molecular
mechanisms are probably identical with the so-called
facilitated and active transport processes which are already
well documented in the literature from the phenomenological point of view.
3.3.2. The Regulation of Permeability
The actual permeation barriers of biomembranes are formed
by lipids. Consequently, the Permeability parameters must
ultimately depend on the lipid characteristics. Three
fundamental aspects having an important bearing on permeability may be deduced from these lipid characteristics :
The first aspect concerns the thickness of the permeation
barrier in biomembranes. It is certainly of decisive importance whether the permeation barrier is formed by a monomolecular lipid layer about 22 A thick or by a bimolecular
lipid layer at least 44 thick. The fact that the “outer” and
“inner” aqueous phases surrounding biomembranes in vico
possess different pH-values as well as different ion concentrations makes it appear very improbable that bimolecular lipid layers exist in biomembranes. Structures as
566
illustrated in Figure 22 will only be stable provided the two
outer phases are identical, i.e. in membranes without a
gradient. Such membranes are in a state of thermodynamic
equilibrium, i. e. they are “dead”. The “basis membrane”
might therefore also be considered as the inactive basis
form of biomembranes in thermodynamic equilibrium, the
fictive ideal forms of which are shown in Figure 22. A
membrane adapted to different “outer” and “inner” phases,
i. e. a membrane in the biological steady state can, however,
only exist if it forms monomolecular lipid layer domains
which constitute, together with a one-sided protein association, a completely asymmetric structure. The asymmetric
structures shown in Figures 23a, 23 b, and 23c must therefore be regarded as the most stable membrane structures
in the biologically steady state. In analogy to the definition
of the basis membrane the “functional membrane” can be
defined as the fully active biomembrane in the biological
steady state.
These statements can be understood from the lipid characteristics. Let us assume that the pH of the cell interior is 6.5
and that of the cell exterior 7.5. In a bimolecular lipid
micelle of PA for instance, the inner lipid layer would have
to exist in the pi,-conformation, while the outer lipid layer
would have to adopt the p,,-conformation. According to
our experience this is an unstable steric situation which
results in a decomposition of the micelle. The reconstituted
micelle of uniform conformation would, however, undergo
the same process. This decomposition process would be
repeated until a protein could become associated with one
side of a monomolecular lipid layer. In other words, this
process of reconstitution would again lead to a stable
totally asymmetric structure of the kind shown in Figure 23.
Conversely, it can also be envisaged that membranes which
are removed from their in viuo milieu (e.y. by isolation in
ritro) assume a tendency to undergo symmetrization and
start symmetrization operations, i. e. they aim towards the
basis membrane state.
We consider the jumplike permeability alterations, induced
and regulated by pH-changes of the cell interior, to be a
second essential aspect of permeability behavior. For the
accomplishment of such unsteady permeability changes
the C0,- and pH-dependent phase transition of PE
seems to be an appropriate system. It is considered that
with this system the regulation of permeability is performed
in the physiologicalIy relevant pH-range of pH = 5.5
to 7.5, the steepest range of control lying at pH=6.3
corresponding to the pK-value of the carbonic acid
system. A further system which reacts even more sensitively
to pH-changes than the PE/bicarbonate/carbonic acid
system does is the complex formed by PS, Ca”, and
inorganic phosphate. The normal binding capacity of PS
for Ca2’ is enhanced many times in the small pH-range of
pH = 6.3 to 6.4 if inorganic phosphate is available[541.
In conclusion a third aspect deserves brief consideration :
just as important as the unsteady permeation-which
most likely plays an essential role in the sensory signal
conversion processes-is the establishment and control
of a steady permeation flux (leak-flux). A constant
permeation flux should be upheld by largely pH-independent C a Z + binding to the phospholipids PA, PC,
PE, and PS. The decisive parameter on which such a
A n g e w Chem. internat. Edit. 1 Vol. I 1 (1972) 1 No. 7
leak-flux will largely depend is provided by the CaZ+concentration in the cell interior which for its part is
determined by the active CaZ transport.
+
4. Conclusion
The general outline of membrane structure and functions
given above is intended as an account of experimental
results and of postulates based thereon. Taken as a whole
it forms a comprehensive membrane hypothesis. This paper
does not set out to furnish a complete membrane concept
but rather to expound our standpoint regarding membrane
problems and to advance new aspects which may lead to a
widening of our understanding of molecular biology and
molecular medicine.
I am especially indebted to Dr. H . Emrich for his valuable
assistance in the preparation of this paper and for many
discussions which have made a most helpful contribution to
this study. Thanks are also due to I)$.-Phys. S. Stange and
to Dip/.-Ing. D. Walter who have provided many experimental
data in connection with this work and to Dip/.-Phys. H . Pape
for carrying out the computer calculations. I am also grateful
to Miss G. Zippan for her technical assistance and for preparing the illustrations.
Received: June 2, 1971 ;
supplemented: April 27,1972 [A 880 IE]
German version: Angew. Chem. 84, 597 (1972)
[l] V Luzzati and F . Husson, J. Cell Biol. 12, 207 (1962).
[2] H . Puubiuble, Naturwissenschaften 58, 277 (1971).
[3] D. Chapman, R. M . Williams, and B. D. Ladbrooke, Chem. Phys.
Lipids I , 445 (1967);B. D. Ludhrooke and D. Chapman, ibid. 3.304 (1969).
[4] E. Rojas and J . M . Tobias, Biochim. Biophys. Acta 94, 398 (1965).
[ 5 ] D. Papahadjopoulos, Biochim. Biophys. Acta 163, 240 (1968).
[6] A . D. Bangham, Progr. Biophys. Mol. Biol. 18,29 (1968).
[7] 'I:Hainai, D. A . Hayndon, and J . Taylor, J. Theor. Biol. 9,278 (1965)
[S] S. Ohki, Biophys. J . 9, 1195 (1969); J. Colloid Interface Sci. 30,413
( 1969).
[9] P. Lcuger, Proc. Coral Gables Conf. on Physical Principles of Biol.
Membranes, 1970, p. 227.
[lo] K . S. Cole, Proc. Coral Gables Conf. on Physical Principles of
Biol. Membranes, 1970, p. 1.
[11] J . C. P. Smith, Chimia 25, 349 (1971); 0. H . Grqfith, L. J . Libertin,
and G. B. Birnell, J. Phys. Chem. 75, 3417 (1971).
[I21 D. Chapman and A. Morrison, J. Biol. Chem. 241, 5044 (1966);
E. G. Finer, A . G. Flook, and H . Hauser, FEBS Lett. 18, 331 (1971).
[13] J . B. Finean, Experientia 9, 17 (1953).
[14] W Kreutz, Adv. Bot. Res. 3, 54 (1970).
[15] W Kreutz and D. F. Walter, to be published.
Angew. Chem. internat. Edit.
1 Vol. 11 (1972) / N o . 7
[16] M . Blank and J . S. Britten, Proc. Coral Gables Conf. on Physical
Principles of Biol. Membranes, 1970, p. 143.
[I71 K . W Butler, H . Dugas, I . C . P. Smith, and H . Schneider, Biochem.
Biophys. Res. Commun. 40, 770 (1970).
[IS] U.Keller, Diplomarbeit, Technische Universitat Berlin (1971).
1191 R. Hosemann and W Kreutz, Naturwissenschaften 53, 298 (1966).
[20] P. Weber, Z. Naturforsch. 176, 683 (1962); Z. Naturforsch. 186,
1105 (1963).
[21] W Menke and E. Jordan, 2. Naturforsch. 14b, 234, 393 (1959).
[22] R. S. Criddle and L. Park, Biochem. Biophys. Res. Commun. 1,
74 (1964); R . S. Criddle, R. M . Bock, D. E. Green, and H . nsdale, Biochemistry 1, 827 (1962).
[23] J. L.Bailey, personal communication.
[24] M . 'I: Laico, E. I. Ruoslahti, D. S. Papermaster, and W J . Dreyer,
Proc. Nat. Acad. Sci. USA 67, 120 (1970).
[25] International Tables for X-Ray Crystallography. Kynock Press,
Birmingham 1952.
1261 P. Giesbrecht and G. Drews, Arch. Mikrobiol. 54, 297 (1966).
[27] W Kreutz, P. Giesbrecht, and A. Paulick, in preparation.
[28] R . B. Park and N . G. Pon, J . Mol. Biol. 6, 105 (1963).
[29] K . Miihlethaler, H . Moor, and J . W Szarkowski, Planta 67, 305
(1965).
[30] St. H . Howell and E. N . Moudrianakis, Proc. Nat. Academy Sci.
USA 58, 1267 (1967).
[31] J . L. Bailey and W Kreutz, Int. Congr. Photosynthesis, Freudenstadt 1968.
[32] J . S. C. Wessels, Int. Congr. Photosynthesis, Freudenstadt 1968.
[33] L. P. Vernon, B. Ke, H . H . Mollenhauer, and E. R. Shaw, Int. Congr.
Photosynthesis, Freudenstadt 1968.
[34] S. G. Stange, H . M . Emrich, and W Kreutz, to he published.
[35] L. J . Schneiderman and I . G. Janga, Biochemistry 7, 2281 (1968).
[36] E. H . Eylar, Proc. Nat. Acad. Sci. USA 67,1425 (1970); E . H . Eylar,
J . Salk, G. Beuerdige, and L. Brown, Arch. Biochem. Biophys. 132, 34
(1969).
[37] J . London, Biochem. Biophys. Acta 249, 188 (1971).
1381 J . F. Danielli and A. Dauson, J. Cell. Comp. Physiol. 5,495 (1935).
[39] C. K . Akers and D. F. Pearsons, Biophys. J . 10, 101, 116 (1970).
[40] W Kreutz and S. G. Stange, to be published.
[41] D. Harker, Science, in press.
[42] M . F.Moody, Science 142, 1173 (1963).
[43] A. E. Blaurock and M . H . F. Wilkins, Nature 223,906 (1969).
[44] J. K. Blasie and C. R. Worthington, J. Mol. Biol. 39,407 (1969).
[45] J. B. Finean, R. Coleman, W A . Green, and A. R . Linbrick, J. Cell.
Sci. 1, 287 (1969).
[46] W Menke, Brookhaven Symps. in Biol. 19, 328 (1966).
[47] J . E. Thompson, R . Coleman, and J . B. Finean, Biochim. Biophys.
Acta 150, 405 (1968).
[48] H . Pape, Thesis, Technische Universitat Berlin 1971.
[49] G. Nhmethy and H . A . Scheraga, J . Chem. Phys. 36, 3382, 3401
(1962).
[so] S. Fleischer, B. Fleischer, and W Stoeckenius, J. Cell Biol. 32, 193
(1967).
[51] W Kreutz and W Menke, Z. Naturforsch. 15b, 402 (1960).
[52] F. S. Sjostrand and L. Barajas, J. Ultrastruct. Res. 25, 121 (1968).
1531 J. K. Blasie and L. R. Worthington, J . Mol. Biol. 39,417 (1969).
[54] J . M . Cotmore, G. Nichols, and R . E . Wuthier, Science 172, 1339
(1971).
567
Документ
Категория
Без категории
Просмотров
0
Размер файла
2 266 Кб
Теги
structure, biomembranes, principles
1/--страниц
Пожаловаться на содержимое документа