Magnetic Resonance Review 1988, Vol. 13, pp. 135-178
Photocopying permitted by license only
© 1988 Gordon and Breach Science Publishers, Inc.
Printed in the United Kingdom
ELECTRON SPIN RESONANCE OF BIOLOGICAL
MEMBRANES: SPIN-LABELED LIPIDS AND
PROTEINS
JOHANNES J. VOLWERK and O. HAYES GRIFFITH
Institute of Molecular Biology and Department of Chemistry,
University of Oregon, Eugene, OR 97403
Received May 27, 1987
I. Introduction
II. Reviews and Books
III. Freely-Diffusing Spin Labels
A. Lipid Spin Labels
1. Lipid-Protein Interactions
2. Multiple Binding Equilibria of Lipid-Protein Interactions
3. Studies on Pure Lipid Bilayers
4. New Lipid Spin Labels
5. Other Studies Involving Lipid Spin Labels
B. Non-lipid Spin Labels
IV. Spin Labels Covalently Attached to Membrane Proteins
V. Summary
Acknowledgements
References
I. INTRODUCTION
Electron spin resonance (ESR) and saturation transfer ESR (ST-
ESR) have become established as biophysical techniques yielding
valuable information on the structure and dynamics of lipids and
proteins in biological systems. In this paper we review the 1985 to
Spring 1987 literature centered around the topic "ESR of biological
membranes". In reviews published during this period, much of the
literature from before 1985 is treated extensively. These reviews are
135
\z%
136 JOHANNES J. VOLWERK and O. HAYES GRIFFITH
listed in section 2 and for more information on specific topics the
reader is referred to them.
Although computer-assisted literature searches have greatly
alleviated the task of collecting relevant papers it is of course
possible that we have missed one or more important papers in the
field. We apologize for any omissions as well as for the fact that,
unavoidably, emphasis in certain areas and lack thereof in others
has been influenced by our own personal interests.
II. REVIEWS AND BOOKS
There are several recent reviews dealing with ESR applied to bio
logical systems. Acomprehensive series of reviews with emphasis on
modern advances can be found in "EPR and advanced EPR studies
of biological systems" [1], with literature through 1983. Chapters
are dedicated to the theory of magnetic resonance, instrumentation
and methodology of ESR, and applications of ESR spectroscopy to
the study of protein structure and function, lipid membranes, and
DNA. ESR-spin labeling and other methods employing physical
labels are discussed by Likhtenstein et al. [2]. A brief overview of
the use of spin labels in biochemistry is given by Watts [3]. Some of
the theoretical concepts underlying spin label ESR spectroscopy and
molecular mobility in biological systems are discussed by Marsh [4].
The structure and dynamics of biological membranes continues to
be an area where application ofelectron spin magnetic resonance to
a biological system appears most rewarding. This is clearly indicated
by the many reviews specifically dedicated to this topic. Different
aspects of spin label ESR spectroscopy pertaining to the molecular
mobility of lipids and proteins in membranes and interactions
between these membrane constituents are reviewed by Marsh, [5,
6], Davis [7], Devaux [8], Devaux and Seigneuret [9], Seigneuret
et al. [10], Leaver and Chapman [11], Sankaram and Easwaran [12],
and Swartz [13]. Areview by Morse [14] not only provides a useful
introduction to the ESR/spin labeling technique, but also discusses
applications to biological membranes and new methodologies, for
example, electron spin echo spectroscopy. Reviews by Smith and
Butler [15], Fleischer and Mclntyre [16], Hemminga [17], and Freed
[18] contain sections in which special topics related to ESR of
biological membranes are discussed. ST-ESR and its application to
IESR OF BIOLOGICAL MEMBRANES 137
the rotational diffusion of membrane proteins also are extensively
reviewed by Thomas [19, 20, 21] and Thomas et al. [22].
III. FREELY-DIFFUSING SPIN LABELS
A. Lipid Spin Labels
The chemical structures of some examples of spin-labeled lipid
analogues used in ESR studies of the structure and function of
biological membranes are shown in Figures 1 and 2. Useful lipid
spin probes include: spin-labeled fatty acids (Figure 1a, b, c); spin-
labeled, charged amphiphiles (Figure 1d, e, f); spin-labeled steroids
(Figure 1 g, h, i); and spin-labeled phospholipids (Figure 2).
rX.
FIGURE 1 Chemical structures of single-tail and steroid lipid spin labels useful for
investigation of the dynamic behavior of lipids in biological membranes by ESR.
Structures a, b, and c are examples of spin-labeled fatty acids which are the most
commonly used spin probes. By varying the position of the nitroxide moiety along
the alkyl chain, segmental motion and order can be examined as a function of bilayer
depth. The chemically more stable proxyl group (see dand e) can be used instead of
the doxyl group as the nitroxide-bearing moiety. Structures d, e, and f are examples
of spin-labeled, charged amphiphiles. Structures g, h, and i are examples of spin-
labeled steroids which are useful for study of cholesterol and analogues. The names
of the spin labels shown are: a, 5-doxylstearic acid; b, 8-doxylstearic acid; c, 16-
doxylstearic acid; d, 14-proxylstearylmethyl phosphate; e, N-14-proxylstearyl-N.N,N-
trimethylammonium; f, N,N-dimethyl-N-dodecyltempoylammonium; g, 3-doxyl-
androstanol; h, 17, 2'-pyrollidine cholesterol; i. 25-doxylcholestanol.
138 JOHANNES J. VOLWERK and O. HAYES GRIFFITH
=(CH,),-NH,
=(CH,),-N(CHJ),
=CH,- CH-NHj
O"
=CM,-CHOH-CH,OH
H,C -OP-O-R H,C-0-P-C-CH,
6- i
HCOH
O |
H,C-0-P-0- CH,
I o"
-^C-O-CH
FIGURE 2 Chemical structures of spin-labeled phospholipid analogues: PA, pho-
sphatidic acid; PE, phosphatidylethanolamine; PC, phosphatidylcholine; PS, pho-
sphatidylserine; PG, phosphatidylglycerol; CL, cardiolipin or diphosphatidylglycerol(DPG). In all these phospholipids other doxyl- or proxyl-containing fatty acids can be
substituted for the 14-doxyIstearoyl moieties shown here.
1. Lipid-Protein Interactions
Because of its high sensitivity and favorable timescale, conventional
ESR in combination with spin-labeled lipid analogues continues to be
a popular technique to investigate various aspects of lipid-protein
interactions and lipid dynamics in native and reconstituted mem
branes. Most studies focus on the identification, characterization
and quantification ofa motion-restricted component present in ESR
spectra ofmembranes containing lipid spin labels and integral mem
brane proteins. Except for extreme conditions such as low lipid-
protein ratios or temperatures below the hydrocarbon chain melting
transition ofthe lipids, this motion-restricted component arises from
spin-labeled lipids in contact with the hydrophobic protein surface.
[We recommend use of the term "motion-restricted component"
rather than "immobilized component". Although well known in
spectroscopy, the latter term may give rise to the erroneous concept
of a "permanent annulus of immobilized lipids" surrounding a
membrane protein. Quite to the contrary, lipids in the boundary
layer have been shown to retain considerable motion and to
exchange with the bulk bilayer lipids with a rate that is slow to
intermediate on the ESR timescale but fast, for instance, on the
NMR timescale (see ref. 6).]
ESR OF BIOLOGICAL MEMBRANES 139
Seigneuret et al. [23] have studied cholesterol-protein interactions
in the human erythrocyte membrane using a spin label analogue
of cholesterol bearing a nitroxide on the alkyl chain (26-nor-25-
doxylcholestanol). The ESR spectrum is composed of a mobile
component and a highly motion-restricted component (outer hyper-
fine splitting 61-63 G) representing about 45% of the signal.
Removal of the cytoskeletal proteins reduces the motion-restricted
component, while it is no longer observed after chymotrypsin treat
ment which reduces band 3 protein to a smaller fragment. These
results are consistent with a previously proposed selective band 3-
cholesterol interaction and with the known linkage between band 3
and the erythrocyte cytoskeleton.
From the same laboratory there also appeared a series of studies
on the transbilayer diffusion ofphospholipids in human erythrocytes.
Zachowski et al. [24] use spin-labeled analogues of phosphatidyl
choline, phosphatidylethanolamine and phosphatidylserine to
compare the transverse diffusion rates of lipids in normal and sickle
erythrocytes. Rapid incorporation of these lipid spin labels into the
cell membrane was facilitated by the presence of a short, nitroxide-
bearing acyl chain in the sn-2 position, but such a structural modi
fication may, of course, affect transbilayer diffusion as well. By
selective chemical reduction of the nitroxide labels embedded in the
outer leaflet of the membrane it was shown that the aminophospho-
lipid analogues, but not the phosphatidylcholine analogue, diffuse
toward the inner leaflet within 3h. The transverse diffusion rate of
the amino-analogue is significantly reduced in homozygote sickle
cells when compared with normal cells. In a separate report Za
chowski et al. [25] show that a spin-labeled analogue of phosphatide
acid diffuses slowly, similar to the phosphatidylcholine analogue,
while a spin-labeled sphingomyelin analogue shows no inward
transport. More recent papers from Devaux's group provide
evidence that the outside-inside translocation ofaminophospholipids
in the human erythrocyte membrane [26], in the pig lymphocyte
plasma membrane [27], and in human platelets [28] is a protein-
mediated phenomenon. It remains to be shown, however, that
translocation of natural aminophospholipids, which carry two long
acyl chains, occurs in amanner similar to that observed for the short-
chain spin-labeled analogues.
An interesting study is that of Fong and McNamee [29], who
attempt to correlate Torpedo californica acetylcholine receptor
function with structural and dynamic properties of the host lipid
140 JOHANNES J. VOLWERK andO. HAYES GRIFFITH
bilayer. The low to high agonist affinity-state transition in the pre
sence of carbamylcholine and the ion-gating activity of receptor-
containing vesicles in response to carbamylcholine were measured as
a function of the bilayer lipid composition. Dynamic properties of
these membranes were determined by incorporating spin-labeled
fatty acid and measuring the order parameter. It is found that an
optimal membrane fluidity (i.e. lipid segmental motion) is required
for interconversion between the high and low affinity states of the
acetylcholine receptor. Ion-gating activity, however, requires the
presence of both cholesterol and negatively charged phospholipids
in addition to optimal fluidity. The acetylcholine receptor thus
appears to be an example of a membrane protein for which struc
tural and dynamic requirements of the host lipids are clearly
implicated in biological function.
Tanaka and Freed [30] have studied Iipid-gramicidin A interac
tions utilizing macroscopically aligned samples of low (<10%)
water content and ESR ofspin-labeled lipids and acholestane probe
Freed and coworkers continue to express some reservations regarding
the interpretation of the two-component ESR spectra obtained for
protein-containing lipid bilayers in terms of a two-site model. How
ever, from their results on the oriented lipid bilayers Tanaka and
Freed conclude that the main lipid-gramicidin A interaction is that
of a boundary effect. The polypeptide induces disorder at low
temperature and low water content but induces order at high
temperature and high water content, while lipid fluidity is only
slightly affected.
Lipid mobility and order in bovine rod outer segment disk mem
branes have been studied by Pates and Marsh [31] using a series of
eight stearic acid spin label probes labeled at different carbon atom
positions in the chain. All spectra consist of two components; one
closely resembling the spectra obtained from dispersions of the
extracted membrane lipids and the other, characterized by a con
siderably greater degree of motional restriction, induced by the
presence of rhodopsin. The proportion of the motionally restricted
lipid component is approximately constant and independent of the
position of the spin label group. The spectral characteristics of the
motionally restricted component as a function of temperature indi
cate that the mobility of these lipids in contact with protein in
creases with temperature but does not vary greatly along the middle
portion of the chain, between carbons 8 and 14 from the carboxyl
ESR OF BIOLOGICAL MEMBRANES 141
group ESR spectra obtained as a function of the direction of the
magnetic field with oriented samples indicate that the motionally
restricted acyl chains have a broad distribution of orientations. It is
concluded that the spin-labeled lipids in direct association with
rhodoposin are not highly ordered and display motional restriction
along much of the length of their chains. This result is similar to the
early observations on lipid interactions with the mitochondrial
protein, cytochrome c oxidase [32, 33, 6].
The effect on the lipid order parameter of microsomal cyto
chrome P-450 and NADPH-cytochrome P-450 reductase recon
stituted in unilamellar phospholipid vesicles has been investigated
by Kunz et al. [34], employing fluorescence and ESR. The fluore
scent probe used was l,6-diphenyl-l,3,5,-hexatriene and the spin
probes were phosphatidylcholines with 5- or 8-doxylsteanc acid
esterified at the sn-2 position. Both techniques show that these
proteins decrease the order parameter of the phospholipid acyl
chains, contrary to what has been observed for other proteins, e.g.,
bacteriorhodopsin and cytochrome c oxidase.
Volotovski et al. [35] have determined the effect of calmodulin on
the order parameter of lipids in photoreceptor membranes and
rhodopsin-containing phospholipid vesicles by spin label ESR.
Calmodulin did not change the order parameter of lipid spin labels
in bilayers containing only rhodopsin but induced a significant in
crease in the order of the lipids in bovine rod outer segment disk
membranes, which contain rhodopsin and other proteins. This
suggests that the site of calmodulin binding is remote from rhodopsin
itself. „ ,Nishiyama and Kuninori [36] have looked at the effect of oxygen
on wheat flour lipids during dough mixing. Dough was prepared in
the presence of the spin label 5-doxylstearic acid under different
nitrogen/oxygen atmospheres and ESR spectra of the gluten were
recorded The ESR data combined with lipid analysis indicate that
modification of the lipids by oxidation leads to increased fluidity and
a decrease in lipid-protein interactions. These observations may
relate to the effect of oxygen and lipid oxidation on the rheological
properties of dough.
In addition to studies of lipid-protein interactions involving
integral membrane proteins, other groups have employed ESR to
investigate lipid binding of water-soluble proteins and peptides.
Rietveld et al. [37, 38] and Goerrissen et al. [39] have studied the
142 JOHANNES J. VOLWERK and O. HAYES GRIFFITH
association of apocytochrome c, the non-heme precursor of the
mitochondrial protein cytochrome c, with model membranes con
sisting of or containing negatively charged phospholipids. It is
shown that upon electrostatic interaction with the negatively
charged lipids, apocytochrome cis able to penetrate into the hydro
phobic region of the membrane. ESR spectra of phospholipids spin-
labeled at different positions along the sn-2 chain indicate agenera
lized decrease in mobility of the lipids but the characteristic flexibility
gradient toward the terminal methyl end ofthe chain is maintained
With lipids spin-labeled close to the terminal methyl end a second
more motionally restricted component is observed with apocyto
chrome c but not with cytochrome c, indicating a fundamental
difference in the mode of binding of these proteins to negatively
charged lipids. In mixed model membranes consisting of negatively
charged and neutral lipids, the protein interacts preferentially with
the negatively charged lipids but this preferential interaction does
not induce separation of large domains enriched in complexes of
apocytochrome c and negatively charged phospholipids and domains
enriched in neutral lipids.
The activation of the lipolytic enzyme, porcine pancreatic pho-
sphohpase A2, by the presence of negative charges at the lipid
substrate-water interface has been investigated by Volwerk et al
[40] using a combination of kinetic studies, fluorescence, and ESR
experiments. The strong interaction of this enzyme with negatively
charged lipids present in aqueous solution or in mixed micelles was
confirmed using both negatively and positively charged spin-labeled
detergent analogues. These results have physiological implications
because the natural substrates of this digestive enzyme are mixed
micelles containing long-chain phospholipids and negatively charged
detergent-like bile salts. The data show clearly that charge inter
actions between the negatively charged micelle and positive amino
acid residues of the enzyme facilitate efficient hydrolysis of the
solubihzed lipid substrate.
Surewicz and Epand [41] examine the effects of amino acid sub
stitutions in the pentapeptide pentagastrin on the nature of its inter
actions with dimyristoylphosphatidylcholine (DMPC) liposomes by
differential scanning calorimetry (DSC) and ESR. Pentagastrin is a
biologically active, chemically modified form of the carboxyl-
terminal tetrapeptide of the hormone gastrin. In two peptide
analogues, the Asp at position 4in pentagastrin was replaced by Gly
ESR OF BIOLOGICAL MEMBRANES 143
or Phe resulting in uncharged and more hydrophobic peptides.
Three charged analogues were also studied: pentagastrin itself, an
analogue with positions 4 and 5 reversed, and one with Asp-4
replaced by Arg. These peptides mimic the behavior of integral
membrane proteins both in their effect on the DMPC phase behavior
as reported by DSC, and their effect on lipid mobility as reported by
stearic acid spin probes. It is concluded that peptide composition
charge, and sequence are important in determining the nature ot
peptide-lipid interactions.
Ahmad et al. [42] have studied the association of apohpoprotein
C-III (apoC-III) from human very low density lipoprotein with egg
yolk sphingomyelin at the transition temperature of the phospholipid.
The combined results from circular dichroism, gel filtration, sedi
mentation velocity experiments, quenching of the tryptophan fluore
scence by spin-labeled phosphatidylcholine, differential scanning
calorimetry and electron microscopy indicate that apoC-III binds to
the sphingomyelin vesicles inducing a structural transition to much
smaller particles. Amodel is presented featuring a disc-like cylin
drical bilayer whose outermost rings of acyl and alkenyl chains are
covered by apoC-III with the helical axis of this protein perpendi
cular to the long axis of the chains.Mims et al [43] use a series of spin-labeled phosphatidylcholines(PC) and cholesteryl esters (CE) with anitroxide group at fatty acyl
carbon C5 C12, or C16 to study acyl chain motions at the surface or
within the core of microemulsion particles (ME) with or without
apohpoprotein E(apoE) bound to their surfaces. These particles are
about 750 Angstroms in diameter, contain cholesteryl oleate= ancI are
coated with amonolayer of dimyristoylphosphatidylchohne (DMPC).
ESR data obtained with the spin-labeled PC's indicated a gradient
of motion in the ME surface monolayer similar to that observed
with the same probes in abilayer. The 5- and 12-doxyl CE's demon
strated a higher degree of order for the CE-rich core than the
corresponding PC's showed for the phospholipid-nch surface.
Binding of apoE to spin-labeled DMPC vesicles increased the order
of the 5-position of the sn-2 chain but for the ME particles the effect
of protein binding was not as striking. Ascorbate reduction indi
cated that the C5-position of the sn-2 chain of PC in the microemul
sion was less accessible than the corresponding position in the
vesicleNothig-Laslo and Knipping [44] compare lipid dynamics at the
144 JOHANNES J. VOLWERK and O. HAYES GRIFFITH
surface layer of porcine and human high density lipoprotein (HDL)
subclasses by spin labeling. Porcine HDL was artificially subdivided
into buoyant density subclasses corresponding to those of human
HDL and the motion of four different positional isomers of spin-
labeled fatty acids and of spin-labeled androstanol, introduced into
nese HDL subclasses, was determined by ESR. In general, the spin
labels experienced more restricted motion in the porcine HDL than
in the human HDL2 and HDL, subclasses.
Baglioni et al. [45] have studied two-dimensional mixtures of a
polypeptide, poly (Y-methyl-L-glutamate), and the spin probes 5-
doxylstearic acid and 16-doxylstearic acid, in monolayers at the air-
water interface. Interactions between each spin label and the poly
peptide were investigated by analyzing the parameters obtained
from the spreading isotherms and by analyzing ESR spectra of the
collapsed monolayers. The position of the nitroxide group along the
hydrophobic chain affected the interaction of the probe with the
polypeptide, indicated by differences in the Gibbs free energy of
mixing. 5J
2. Multiple Binding Equilibria of Lipid-Protein Interactions
Besides the mostly qualitative studies discussed above, more quanti
tative ESR-spin labeling studies aimed at determining thermo
dynamic parameters of the interactions between integral membrane
proteins and membrane phospholipids continued to appear during
the review period. Pates et al. [46] have employed freely diffusing
phospholipid spin labels to study rhodopsin-lipid interactions in frog
rod outer segment membranes. Each of the spin probes shows a
two-component ESR spectrum with a motionally restricted com
ponent attributed to spin labels in contact with the surface of
rhodopsin and the other component originating from spin labels in
the fluid lipid bilayer region of the membranes. The population of
motion restricted lipids is sufficient to coat the hydrophobic protein
surface while there is little selectivity beween the different spin-
labeled phospholipid classes examined.
From the same laboratory appeared a spin-label study of lipid-
protem interactions in Na+, K+-ATPase membranes from Squalus
acanthias [47, 48]. Using lipids with nitroxide labels at the 14-position
of the sn-2 chain, Esmann and Marsh [47] have investigated the pH
ESR OFBIOLOGICAL MEMBRANES 145
and salt dependence of the interactions of phosphatide acid, pho-
sphatidylserine and stearic acid with the ATPase. For phosphatide
acid and stearic acid, the fraction of motionally restricted spin label
increases with pH, with pKa's of 6.6 and 8.0, respecively, whereas
the fraction of motionally restricted phosphatidylsenne spin label
remains constant between pH 4.7-9.2. Similarly, high salt decreases
the fraction motionally restricted component for phosphatide and
stearic acid but has relatively little effect on that of phosphatidyl-
serine The authors conclude that direct electrostatic effects alone
cannot account for all of these observations. In a second paper
Esmann et al. [48] show that the fraction of motion restricted lipids
corresponds to a stoichiometry of aproximately 66 lipids per
265 000-dalton protein. This number of lipid molecules roughly can
be accommodated within the first shell around the protein dimer.
The ATPase shows some selectivity for association with the various
spin-labeled lipids in the order: cardiolipin >other negatively
charged phospholipids >phosphatidylcholine s androstanol.
The association of spin-labeled cardiolipin with bovine heart
cytochrome coxidase reconstituted in dimyristoylphosphatidylchohne
has been investigated by Powell et al. [49]. They report a hpid-
substituted preparation which contains less than one mole of cardio
lipin per mole of enzyme but, nevertheless, retains oxidative activity.
ESR spectroscopy of these lipid-substituted cytochrome oxidase
preparations demonstrates an average association constant for car
diolipin that is 5.4-times that for phosphatidylcholine. This selec
tivity is interpreted as corresponding to a generalized increase in
specificity for all lipid association sites on the protein.
Some advances toward a better understanding ofthe specificity of
lipid-protein interactions have been made recently [50]. In this study
the model for lipid-protein interactions previously described by
Brotherus et al. [51] is generalized to include lipid binding to an
integral membrane protein solubilized in excess detergent. The central
assumption in this multiple equilibria binding treatment is that in
lipid bilayers and in lipid/detergent mixed micelles there is acompeti
tion between lipid and detergent molecules, for the hydrophobic
protein surface, as diagrammed in Figure 3. Competition between
lipid (L) and detergent (D) molecules in aprotein-containing mixed
micelle is represented as a series of exchange reactions at the
hydrophobic protein-lipid interface:
146 JOHANNES J. VOLWERK and O. HAYES GRIFFITH
FIGURE 3 Schemat.c drawing of the dynamic exchange of phospholipid molecules
at the protein-hpid interface of a membrane protein in a lipid bilayer. Asimilar
exchange occurs between lipid and detergent molecules when amembrane protein is
present in a hpid/detergent m.xed micelle. The crosshatched area indicates the
hydrophobic protein surface normally exposed to the lipid hydrocarbon chains For
phospholipid molecules situated adjacent to a protein molecule, any restriction of
acyl cham mot.on imposed by the protein surface is reported by the nitroxide spin
label moiety attached near the terminal end of one of the acyl chains. This interaction
between l,p,d and protein gives rise to the motion-restricted component observed in
the ESR spectrum.
BDp + L ===± BL + (3 D
where Bis a contact site and |3 represents the number of detergent
molecules replaced by a single lipid molecule. For a protein with
multiple classes of contact sites the following general binding equa
tion is derived:
i=i 1 + KiX^XBr1
where v is moles of lipid bound per mole of protein, N; and K are
the number ofcontact sites and the relative association constant for
class i, and XLf and XDf are the mole fractions of free lipid and free
detergent, respectively. In the limit of high amounts of detergent
relative to the other components, the above equation simplifies to-
NiKiLfD,"
'"2i-i 1 + KjLfDr1
where LfD,-' is the molar ratio of free lipid to total detergent This
binding equation is similar to the one for water-soluble proteins and
hgands, except that LfDt ' is the concentration variable rather than
the aqueous concentration of ligand. Experimentally, the excess
ESR OF BIOLOGICAL MEMBRANES 147
detergent, which is the solvent in this system, allows the lipid/
protein ratio to be varied over a wide range without the risk of
protein aggregation at low lipid/protein ratios. This is an advantage
over previous experiments with lipid bilayers and facilitates detection
of high-affinity contact sites specific for certain lipids. Analysis of
ESR data for cholate-solubilized affinity-purified bovine heart cyto
chrome c oxidase using this binding treatment, suggests that the
selectivity for the negatively charged phospholipid, phosphati-
dylglycerol, is limited to a relatively small number of high-affinity
sites on the protein surface [50].
Several aspects of the problem of lipid-protein interactions in
biological membranes have been investigated by more theoretical
approaches using model calculations. An example is the paper of
Scott [52] who performed Monte Carlo calculations of order para
meter profiles in models of lipid-protein interactions in bilayers. In
these models the lipid chains are 10 carbon atoms long and interact
via van der Waals forces. The chains also interact via van der Waals
forces with a "perturbant", a small cylindrical model polypeptide
with either a smooth or a rough surface. The results show that in all
cases studied there are only slight conformational differences
between the bulk chains and the chains nearest the perturbants, and
Scott believes it is not possible to characterize the boundary chains
as "more" or "less ordered".
Pink and MacDonald [53] report a theoretical study of the inter
actions between glycophorin, an integral membrane protein of the
erythrocyte plasma membrane, and phospholipids in a bilayer
membrane. They calculate the average value of the order parameter
of a spin-labeled hydrocarbon chain in a dimyristoylphosphatidyl-
choline bilayer as a function of the glycophorin concentration for a
temperature above the main lipid phase transition. The results indi
cate that between 200 and 1300 lipid molecules are affected by the
large polar moiety of glycophorin dependent on its conformation. In
a similar approach, the average order parameter of the fluorescent
probe diphenylhexatriene was calculated and used to predict the
value of the limiting anisotropy as a function of the glycophorin
concentration.
The problem of determining the number (N) of lipid molecules in
direct contact with integral membrane proteins in lipid bilayers and
the associated problem of protein lateral distribution also has been
addressed. Laidllaw and Pink [54] argue that a simple model
148 JOHANNES J. VOLWERK and O. HAYES GRIFFITH
commonly used to determine Nfrom ESR data obtained for lipid
bilayers may lead to underestimated values of N because random
protein-protein contacts, which tend to decrease the protein surface
area exposed to the lipids, are not taken into account. They present
an alternative model, supported by Monte-Carlo type computer
simulations, and attempt to fit this model to experimental ESR data
published by other workers. Unfortunately, the fit to their model
heavily relies on data points obtained at low lipid-protein ratios,
where it becomes troublesome to obtain preparations free of irre
versibly aggregated protein. In a different approach to the problem,
Mountain et al. [55] report a molecular dynamics simulation of a
two-dimensional fluid mixture system designed to illustrate how
short-range ordering of lipid and protein molecules might occur in
membranes. In all these models the results are sensitive to the
choice of input parameters. In order to adequately test these the
oretical models new experimental methods must be developed to
detect lipid-protein and protein-protein interactions over a much
wider range of lipid-protein ratios.
3. Studies on Pure Lipid Bilayers
Koole and Hemminga [56] have analyzed the saturation properties
of the in-phase and quadrature second harmonic ESR spectra of
cholestane spin label in oriented lipid multibilayers in the gel phase.
The in-phase and quadrature spectra exhibit different functional
dependencies on the spin-lattice relaxation time T, of the cholestane
spin label and the magnetic component of the microwave field B,,
thus enabling experimental determination of these parameters. The
value of T, in a system of dimyristoylphosphatidylcholine with
33mol% cholesterol increases from 1.2 xlO"6 to 3x 10"6 S on
decreasing the temperature from 20 to -10°C.
Boggs and Rangaraj [57] have investigated the phase behavior of
phosphatidylcholine and phosphatidylglycerol in the presence of
glycerol and polymixin by differential scanning calorimetry and ESR
of fatty acid spin labels. Interdigitation of the lipid acyl chains
induced by glycerol or polymixin causes a large increase in the order
parameter ofa fatty acid spin labeled near the terminal methyl end
(16-doxylstearate). The order parameter becomes similar to that of
ESR OF BIOLOGICAL MEMBRANES 149
a fatty acid with the spin label moiety much closer to the polar head
(5-doxylstearate). Thus the typical flexibility gradient found in non-
interdigitated bilayers is abolished upon interdigitation and 16-
doxylstearate appears useful for detection of such interdigitation. In
a similar approach Boggs and Mason [58] also investigated the
subgel and interdigitated gel phases formed by asymmetric phospha
tidylcholines (PC).
Dynamic properties ofphosphatidylcholine-cholesterol membranes
in the fluid phase and water accessibility to the membranes have
been studied by Kusumi et al. [59] using spin and fluorescence
labeling methods. Unsaturation of the alkyl chains greatly reduces
the ordering effect of cholesterol although unsaturation alone gives
only minor fluidizing effects. By monitoring the dielectric environ
ment around the nitroxide using Q-band second-derivative ESR
spectra it is found that incorporation of cholesterol increases water
accessibility in the hydrophilic loci of the membrane.
Lange et al. [60] have obtained ESR spectra of the l-myristoyl-2-
[6-(4,4-dimethyloxazolidine-N-oxyl)myristoyl]-5n-glycero-3-pho-
sphocholine spinlabel in highly oriented, fully hydrated bilayers of
dimyristoylphosphatidylcholine as a function of temperature and
magnetic field orientation. The oriented spectra shown indications
of slow motional components so that motional narrowing theory is
not applicable to the spectral analysis. The spectra have been simu
lated by a line shape model incorporating trans-gauche isomeriza-
tion in addition to restricted anisotropic motion of the lipid long
axis. Comparison of the results with those of 2H-NMR spectroscopy
of dimyristoylphosphatidylcholine deuterium labeled at the same
position ofthesn-2 chain shows that most oftheparameters governing
chain order and dynamics are in good accord, except that the
gauche population of the spin-labeled chain is considerably higher
and its isomerization rate is much faster in the low-temperature
phases.
Bimolecular collision rates and diffusion coefficients of lipids in
biological and model membranes can be determined by comparison
of experimental line shapes with simulated line shapes obtained
from solutions of the exchange-coupled Blochequations. Sachse and
Marsh [61] show that fitting the normalized line heights gives an
additional method of estimating the exchange frequency and thus
the collision rate and diffusion coefficient. In a related paper King
150 JOHANNES J. VOLWERK and O. HAYES GRIFFITH
and Marsh [62] measure the collision rates between spin-labeled
valeric acid in water, between mixed-chain spin-labeled phosphati
dylcholine in water-methanol mixtures, and between spin-labeled
phosphatidylcholine monomers and micelles in water from the spin-
spin broadening of the ESR spectrum. In each case the second order
rate constants are consistent with a diffusion-controlled process. The
much slower on-rates for association of lipid monomers with lipid
bilayers compared to the monomer diffusion coefficient in water
indicate that insertion into a bilayer is not a diffusion-controlled
process.
In a more recent paper King et al. [63] describe an unconstrained
optimization method for interpreting the concentration and tem
perature dependence of the linewidths of interacting nitroxide spin
labels which allows determination of translational diffusion coeffi
cients. A modified Gauss-Newton optimization procedure is used to
fit an analytical expression for the total peak-to-peak, first-derivative
linewidth to the concentration and temperature dependence of the
ESR spectra. The temperature dependences of the contributions to
the Lorentzian linewidth from both exchange and magnetic dipole-
dipole interactions are inversely related via the translational diffusion
coefficient. The method has been applied to the determination of
the diffusion coefficients of spin-labeled phosphatidylcholine in dio-
leoylphosphatidylcholine over a wide temperature range.
Florine and Feigenson [64] have examined the behavior of fluore
scent and spin-label probes in several fluid and gel phospholipid
phases, in particular the unusual gel phase of phosphatidylserine
(PS) induced by Ca2+ [Ca(PS)2]. Anthroyloxy- and doxyl-labeled PS
exhibit greatly restricted motion in Ca(PS)2 whereas phosphati
dylcholine (PC) and fatty acid derivatives show no apparent change
in probe motion in Ca(PS)2, compared to fluid lamellar lipid (PS).
Experiments employing fluorescence quenching by spin-labeled PC
in PS/PC in excess Ca2+ were used to determine the distribution of
several fluorophore probes between fluid liquid-crystal and Ca(PS)2
gel phases.
ESR and NMR linewidth broadening by spin labels have been
used by Morrot et al. [65] to determine the overall orientation of
spin-labeled analogues of cholesterol and androstanol in egg lecithin
bilayers. Cholesterol analogues were found to have a single orienta
tion in each monolayer with the acyl chain pointing towards the
ESR OF BIOLOGICAL MEMBRANES 151
center of the bilayer, while the androstanol analogue appeared to
experience two opposite orientations in the same monolayer with a
rapid reorientation.
Kar et al. [66] have published a detailed ESR study of spin-
labeled oriented multilayers of dipalmitoylphosphatidylcholine
(DPPC) at low water content. Information regarding ordering and
anisotropic rotational diffusion rates was obtained via ESR line-
shape analysis over the entire motional range. Cholestane (CSL)
and spin-labeled DPPC have been used to probe different depths of
the bilayer. Results from CSL indicate that close to the lipid-water
interface the DPPC molecule is oriented approximately perpendi
cular to the bilayer but the lipid probes indicate that the hydro
carbon chain of DPPC may be bent away from the bilayer normal
by as much as 30°, in agreement with earlier spin labeling studies
[67, 68]. Electron spin echoes (ESE) were observed for the first time
from oriented lipid-water multilayers. The results suggest that for
detection of very slow motions ESE experiments are more sensitive
to dynamics than continuous wave ESR.
Bruno et al. [69] describe a spin label ESR and ST-ESR study of
the bipolar lipids extracted from Sulfolobus solfataricus, an extreme
thermophilic archaebacterium growing at about 85°C and pH 3.
These lipids are cyclic diisopranyl tetraether lipids that are quite
different from the usual fatty acid lipids. The ESR spectra, obtained
using positional isomers of spinlabeled stearic acid, detect phase
transitions at temperatures below the physiological temperatures of
Sulfolobus solfataricus. The head group regions in one lipid fraction
exhibit a more strongly immobilized structure at 85°C than is
observed for the usual monopolar lipids.
The binding of the radioprotective agent, cysteamine to dipal
mitoylphosphatidylcholine (DPPC) membranes has been studied by
Berleur et al. [70], using calorimetry, turbidimetry and spin labeling.
A fading of the pretransition was observed by DSC and turbidi
metry, suggesting that cysteamine interacts with the polar head
region of DPPC bilayers. Spin-labeled fatty acids indicate increased
orderparameter (reduced segmental motion) of the lipid acyl chains
below the gel to liquid crystal phase transition. The effect decreases
above the phase transition.
Raison and Orr [71] compare various methods for detection of
phase transitions in thylakoid polar lipids of chilling-sensitive plants.
152 JOHANNES J. VOLWERK and O. HAYES GRIFFITH
ESR of spin-labeled methyl stearate, calorimetry, and fluorescence
of parinaric acid all give the same temperature for initiation of
the phase transition of thylakoid lipids of various plants, indicating
that all three techniques present valid methods for assessing phase
transitions.
Interaction between the local anesthetic, tetracaine, and egg
phosphatidylcholine multibilayers has been examined by Frezatti
et al. [72]. ESR spectra of a stearic acid methyl ester spin label
indicate that at pH 10.5, the mobility of the spin label increases as
the anesthetic: lipid ratio is increased from 0.2 to 0.6. At higher
tetracaine: lipid ratios there is no increase in mobility, while above a
tetracaine: lipid ratio of 1, a phase separation occurs. At pH 6.2,
where the tetracaine is charged, high concentrations (e.g. 100:1
tetracaine: lipid) disrupt the bilayer to form mixed micelles. New
local anesthetics (heptacaine and carbisocaine) were examined by
Mazur et al. [73]. By using a spin-labeled fatty acid incorporated
into phosphatidylcholine liposomes, the effects of the anesthetics on
lipid order were determined from the ESR spectra. A pH depend
ence of the lipid disordering is observed that differs from that of the
classical local anesthetics procaine and lidocaine.
Lai and Schutzbach [74] use ESR of spin-labeled stearates to
investigate the effects of dolichol on the motion of lipid molecules in
phospholipid membranes. Dolichols comprise a family of long-chain
polyisoprenoid alcohols that are found in the membranes of most
eukaryotic cells but to which few functions have yet been ascribed.
The ESR spectra show that dolichol has little effect on the motion
of the spin probe at carbon-5, which is already significantly motion-
restricted. Dolichol reduces the motion of the spin probes at carbon-
16, suggesting that dolichol molecules penetrate into the lipid core
region of the lipid membranes.
4. New Lipid Spin Labels
Venkataramu et al. [75] describe the synthesis and ESR characteri
stics of a new stearic acid spin label substituted with deuterium in all
positions and 15N in the paramagnetic doxyl group. The new label
displays a 5.5-fold gain in sensitivity in the ESR spectrum and a
60% decrease in linewidth compared to the unmodified analogue.
ESR OF BIOLOGICAL MEMBRANES 153
The increased sensitivity and resolution should make this label a
useful one in studies employing lipid spin labels.
Acquotti et al. [76] describe a new chemical procedure for the
preparation of gangliosides carrying fluorescent or paramagnetic
probes on the lipid moiety. New lipid labels designed for covalent
attachment to membrane proteins are discussed in Section 4.
5. Other Studies Involving Lipid Spin Labels
Hyde et al. [77] describe electron-electron double resonance
(ELDOR) experiments on spin-labeled liposomes using a loop-gap
resonator. The signal-to-noise ratio is improved 20-fold over the
best that has been achieved using a bimodal cavity, permitting
ELDOR experiments on spin-labeled membranes of intact cells. In
a subsequent study Laiet al. [78] use ELDORtechniques employing
the loop-gap resonator to measure the lateral diffusion constant, D,
of lipids in the surface membrane of intact human blood platelets.
Using [l4N], [15N]-16-doxylstearate spin label pairs, D was found to
be 1.0 x lO^cmV at 37°C for freshly prepared platelets but D
was increased about 2.5-fold upon storage of the platelets for 3 days
under routine bloodbank conditions. The increase in D may be
related to loss of cholesterol from the surface membrane during
storage.
Kuroda et al. [79] have performed a kinetic analysis of the fusion
of hemaglutinating virus of Japan (HVJ*) containing 10 mole per
cent spin-labeled phosphatidylcholine with unlabeled erythrocyte
membranes. The peak-height increase in the ESR spectrum due to
relief of spin exchange was caused both by envelope fusion and by
phospholipid exchange catalyzed by the virus-induced hemolyzate.
A similar technique was used by Yamada and Ohnishi [80] to study
the fusion of vesicular stomatitis virus with various cells (HELR 66,
KB and human erythrocytes) and liposomes. Binding to and fusion
with both cells and liposomes was enhanced at acidic pH. Binding to
liposomes was dependent on the lipid head group but not on acyl
chain composition, whereas cis-unsaturated acyl chains were re
quired for efficient fusion. A cell entry mechanism is proposed in
which the virus initially binds to phospholipid domains in the cell
surface membrane, followed by endocytosis and fusion with the
endosome upon acidification.
Keana and Pou [81] have investigated the feasibility of a nitroxide
154 JOHANNES J. VOLWERK and O. HAYES GRIFFITH
regenerating system involving liposomes to solve the "reduction
problem" when nitroxides are used as contrast enhancing agents in
magnetic resonance imaging. It isshown that inclusion of the oxidant
K3Fe(CN)6 entrapped in the aqueous compartment of liposomes
doped with an amphiphilic nitroxide increases the duration of the
nitroxide ESR signal in the presence of an external reductant.
The head group and chain length dependence of phospholipid
self-assembly has been studied by King and Marsh [82] using ESR
spin labeling. The critical micelle concentrations (cmc) of a series of
l-acyl-2-[4-(4,4-dimethyloxazolidine-N-oxyl)valeryl]-sn-glycero-3-
phosphoderivatives were determined by ESR spectroscopy. The
narrow, three-line ESR spectra of the rapidly tumbling monomers
are easily distinguished from the spin-spin broadened spectra of the
micellar aggregates allowing determination of the concentration of
the two species. For phosphatidylcholine, ln[cmc] decreases linearly
with the length of the snA acyl chain with a free energy of transfer of
the monomer from the aqueous phase to the micelle of -1.1RT per
CH2 group. For the negatively charged lipids the cmc's decrease by
1-2 orders of magnitude on increasing the ionic strength from 0 to
2.0M NaCl, while the salt dependence is considerably less for the
zwitterionic lipids. Thus the polar head group can have a marked
effect on the energetics of self-assembly and correspondingly on the
kinetics of lipid transfer between vesicles.
Coan [83] has studied the distribution of the lipophilic spin
probe, 5-doxyl stearate, between the inner and outer halves of
sarcoplasmic reticulum (SR) membranes by titration with the spin
broadening agent Ni-EDTA. It is found that the spin probe distri
bution between outer and inner halves is 35:65, whereas the distri
bution in small sonicated vesicles prepared from SR lipids is 60:40,
consistent with the vesicle geometry. It is concluded that the
observed asymmetry in distribution in the SR membranes is due to
the presence of SR proteins. Changes in the conformation of one of
the SR membrane proteins, the ATPase, induced by Ca2+ and
adenyl-5'-yl imidodiphosphate-Mg binding did not affect the spin
probe distribution but phosphorylation of the enzyme gave some
protection of probe signal from the spectral broadening reagent.
Ascorbate reduction indicated that a small fraction of the spin
probes, and possibly of the SR lipids, may be occluded following
phosphorylation.
IL
ESR OF BIOLOGICAL MEMBRANES 155
Membrane lipids of the general fatty acid auxotrophic bacterium
Butyrivibrio S2 have been characterized by Hauser et al. [84]
employing differential scanning calorimetry and ESR spectro
scopy. Lipid mobility, as determined by ESR, is low compared to
mammalian plasma membranes but is comparable to that of other
bacterial membranes. Membranes of the organism grown with
saturated fatty acids of defined hydrocarbon chain length undergo a
broad endothermic phase transition with the end-point temperature
approximately coinciding with the minimum temperature supporting
growth. ESR and calorimetry evidence together suggest that, at the
growth temperature, the plasma membrane of Butyrivibrio S2 is in
the liquid-crystalline state. The ESR order parameter of cell
membranes of this organism was found to be similar regardless of
whether it was grown on myristic, palmitic or stearic acid. Butyrivi
brio S2 thus has a mechanism enabling it to maintain membrane
packing and fluidity at fairly constant level, although it cannot alter
either the chain length or unsaturation of membrane fatty acids.
Herring et al. [85] examine the fluidity of lipids (i.e., segmental
motion of the spin label, 5-doxyl stearic acid) in membrane pre
parations from a mutant of Escherichia coli that is resistant to the
uncoupler carbonyl cyanide m-chlorophenylhydrazone (CCCP).
Growth in the presence of CCCP results in a decrease in membrane
lipid fluidity but the fluidity of lipids extracted from these mem
branes is increased, concomittant with an increase in the proportion
of unsaturated fatty acids. The lower fluidity of the lipids in these
membranes may be a consequence of an increase in the ratio of
protein to phospholipid.
Cation-induced changes in the membrane fluidity of isolated corn
mitochondria have been investigated by Cooke et al. [86] using
stearic acid spin labels with the nitroxide group close to the polar
head and close to the methyl end (5- and 16-doxyl stearic acids,
respectively). Addition of Ca2+ or La3+ results in a decrease in the
motion of the labels within the mitochondrial membranes. A com
parison of the temperature-induced changes in label motion with
those induced by Ca2+ indicates that the 16-doxyl stearic acid is
more sensitive to Ca2+-induced changes than is 5-doxyl stearic acid.
However, one must keep in mind that the segmental motion in
creases along the fatty acid chains. Thus even before the addition of
divalent cations, there is less motion near the surface because of the
M.R.R.-
156 JOHANNES J. VOLWERK and O. HAYES GRIFFITH
interaction of the polar head groups. It is, therefore, not unreason
able to expect a larger effect with 16-doxyl stearic acid than for
5-doxyl stearic acid.
The effect of the inhibitory neurotransmitter Y-aminobutyric acid
(GABA) on inhalation anesthetic-induced membrane fluidization is
described by Rosenberg and Alila [87]. The anesthetic enflurane at
2mM fluidizes the lipids of synaptic plasma membranes of whole rat
brain or of the straitum, synaptic mitochondrial membranes, and
dipalmitoyl lecithin vesicles, both in the polarheadand core regions.
GABA, added prior to the anesthetic, causes a dose-related inhibi
tion of the fluidization. This restorative effect also occurs in lipid
vesicles and is not prevented by GABA antagonists, and is therefore
probably not related to receptor mechanisms.
Utsumi et al. [88] use spin-labeled stearic acids with the doxyl
group attached at the 5- or 12-carbon atom and the corresponding
methyl esters to estimate the membrane fluidity of rat liver micro
somal membranes after oral administration of carbon tetrachloride.
The results suggest that CC14 or its metabolites increase membrane
fluidity primarily at hydrophobic regions rather than at the surface
layer. Alterations in the hepatocyte plasma membrane following
CC14 poisoning are reported by James et al. [89]. Freeze-fracture
electron microscopy revealed an increase in the individual mean gap
junction size in rats treated with CC14. ESR spectra of hepatocytes
incubated with 12-doxylstearic acid revealed that although the order
parameter was not altered by CC14 exposure in vivo, the rotational
coefficient was significantly smaller. The data suggest that CC14 may
interact with the membrane in a manner which alters membrane
lipid fluidity thus influencing gap junction particle segregation and
aggregation.
Fluidity of the plasma membrane of the bloodstream form of
Trypanosoma brucei has been examined with ESR and fluorescence
spectroscopy by Munske et al. [90]. The temperature dependence
of the order parameter for 5-doxyl stearate and of 8-anilino-l-
naphtalene sulfonate polarization indicates phase alterations near
30°C. Proteolysis of the surface glycoprotein with trypsin increases
fluidity but treatment with human serum, which is trypanocidal, has
no effect.
Hongo et al. [91] report a spin label study of the effect of
ticlopidine on platelets. In mammals, ticlopidine inhibits ex vivo
platelet aggregation through an as yet unknown mechanism. This
ESR OF BIOLOGICAL MEMBRANES 157
study shows that ticlopidine, orally administered to rats, increases
both the order parameter and the apparent rotational correlation
times of 5- and 16-doxylstearic acid spin probes incorporated into
platelet membranes. Thus the inhibitory action of the drug or its
metabolites is accompanied by changes in the segmental motion of
the lipids.
Chattopadhyay and London [92] describe an interesting method
to determine the depth of penetration of a wide variety of fluorescent
molecules embedded in membranes utilizing fluorescence quenching
by spin-labeled phospholipids. The method involves determination
of the parallax in the apparent localization of fluorophores detected
when quenching by phospholipids spin labeled at different positions
along the acyl chain is compared. The method has been applied to
lipids covalently labeled with the fluorophor 7-nitro-2,l,3-benzoxa-
diazoi-4-yl (NBD). Spin-labeled lipids with the nitroxide moiety
located at three different depths were used to quench the fluorec-
ence. Quenching experiments show that, as expected, the NBD
group of head-group-labeled phosphatidylethanolamine is at the
lipid-water interface, while the NBD label on the "tail" of chole
sterol is deeply buried. This method could prove generally useful in
determining the localization of fluorophores, for example tryptophan
residues of membrane proteins, in biological membranes.
Taraschi et al. [93] compare ethanol-induced structural perturba
tions in the membranes of rat hepatic microsomes measured with
the spin probe 12-doxyl stearic acid with those assayed with the
phospholipid spin probes l-palmitoyl-2-(12-doxylstearoyl) phospha
tidylcholine, -phosphatidylethanolamine and -phosphatidic acid.
Both the fatty acid and the phospholipid spin probes show increased
disordering (increased motion) of the membrane lipids with the in
vitro addition of increasing amounts of ethanol but the effect noted
with the phospholipid spin probes is quantitatively less. Both types
of probes show that microsomes obtained from the livers of chroni
cally alcohol-intoxicated animals are resistant to the disordering
effects of ethanol in vitro. These results suggest that fatty acid spin
probes are qualitatively valid for measuring membrane perturba
tions and that ethanol affects microsomal phospholipids regardless
of head group structure.
Membrane mechanisms of ethanol tolerance in Saccharomyces
cerevisiae have been examined by Curtain et al. [94, 95]. Employing
distearoylphosphatidylcholine with a nitroxide group attached to
158 JOHANNES J. VOLWERK and O. HAYES GRIFFITH
either the 5- or 16-carbon atom of one of the acyl chains it is shown
that increasing concentrations of ethanol have a much more fluidizing
effect on the yeast plasma membranes than on phospholipid vesicles
with the same lipid composition. The fluidization is interpreted as
being due to the displacement of annular lipids from the integral
membrane proteins.
The lipid phase of transverse tubule membrane of skeletal muscle
has been probed with fatty acid spin labels by Hidalgo [96]. The
motion of the probe increases as the distance between the spin label
and the polar head increases, as in other membranes. However, the
value of the order parameter for a fatty acid spin label containing
the label attached to the 5-carbon atom is higher than reported for
other mammalian membranes. Order parameters for spin labels
containing the label nearer to the center of the bilayer are closer to
the values reported for other mammalian membranes. The lipid
phase of the transverse tubule membrane is thus less fluid, at least
as far as the motion of the acyl chain segments near the polar head
groups is concerned, than that of other membranes, sarcoplasmic
reticulum membrane in particular. This may be related to the high
cholesterol content of transverse tubules.
The influence of ganglioside insertion into microsomal mem
branes from calf brain on the rate of ganglioside degradation by
membrane-bound sialidase has been investigated by Scheel et al.
[97]. Micelles of exogenously added ganglioside Gma bind to micro
somal membranes in two steps: fast adsorption, followed by slower
uptake (insertion). The product of the sialidase degradation, gan
glioside GMi, is found in the ganglioside pool taken up by the
membranes only. ESR studies using a spin-labeled analogue of
ganglioside GD)a' suggest that exogenously added ganglioside inserts
into the membranes before it is recognized as substrate by the
sialidase. The effect of pH, ionic strength and fluidizing agents on
the rate of insertion should, therefore, be taken into account when
the influence of these parameters on the rate of the sialidase reac
tion is studied.
Herring et al. [98] and Burt et al. [99] have studied the nature of
the membranolytic interaction between crystals of monosodium
urate monohydrate (MSUM) and membranes using ESR. MSUM
crystals produce the inflammatory reaction of acute gouty arthritis
by a mechanism which is believed to involve interactions between
ESR OF BIOLOGICAL MEMBRANES 159
lysosomal membranes and crystals taken up by phagocytosis, re
sulting in membrane rupture. Two spin probe molecules, negatively
charged 5-doxyl stearic acid and positively charged N,N-dimethyl-N-
dodecyltempoylammonium bromide (CAT,2), were incorporated
into intact human erythrocytes and incubated with MSUM crystals.
The stearic acid spin label is located entirely in the membrane and
the (CAT,2) label partitions between the membrane and aqueous
phase. Incubation with MSUM increases the apparent fluidity of the
stearic acid spin label, and reduces the fraction of membrane-bound
CAT12. These observations are interpreted as being caused by elec
trostatic interactions between the negatively charged MSUM crystals
and the charged spin probes.
B. Non-Lipid Spin Labels
The chemical structures of some examples of small, non-lipid spin
probes used in ESR studies of biological membranes are shown in
Figure 4. The top row (structures a-d) shows freely diffusing spin
labels which are all derived from TEMPO (a). Structures e and f are
I-
o
N-0
N—O
FIGURE 4 Chemical structures of small spin probes employed in spin labeling
studies of biological membranes: a, TEMPO; b, TEMPONE; c, TEMPOL; d,
TEMPAMINE. Structures e and f are maleimide spin labels that attach primarily to
sulfhydryl groups in proteins and are commonly used for labeling of membrane
proteins to investigate their dynamic behavior.
160 JOHANNES J. VOLWERK and O. HAYES GRIFFITH
maleimide spin labels that are used for covalent attachment of
nitroxides to membrane proteins (see section 4).
Earnest et al. [100] have examined the effect of the protein
surface potential on the binding of a spin-labeled quaternary amine
local anesthetic to the nicotinic acetylcholine receptor from Torpedo
californica reconstituted into bilayers of phosphatidylcholine. ESR
of membranes labeled with the spin-labeled local anesthetic, pre
viously shown to block the receptor-ion channel function, shows a
spectral component corresponding to the spin label immobilized by
protein. With the receptor reconstituted into the zwitterionic pho
sphatidylcholine, which is assumed to have no surface potential of
its own, it is demonstrate that the spin-labeled local anesthetic binds
to the acetylcholine receptor as a function of the surface potential
on the protein.
Flewelling and Hubbell [101] describe the synthesis of a spin-
labeled analogue of the hydrophobic ion tetraphenylphosphonium
(TPP+). The thermodynamic properties of the interaction of TPP+
with egg phosphatidylcholine vesicles were studied by equilibrium
dialysis and ESR spin labeling. It was found that the binding is
entropy-driven with a positive (repulsive) enthalpy of binding, while
the binding enthalpy is negative for hydrophobic anions. The mem
brane dipole potential may be responsible for this difference.
Sundberg and Hubbell [102] investigate the surface potential asym
metry in phospholipid vesicles by a spin label relaxation method.
The hydrophobic anion tetraphenylboron (TPB) promotes trans
bilayer migration of an amphiphilic spin label probe. Relaxation
data recorded following rapid mixing of the probe with TPB-
containing vesicles provides information about the electrostatic
potentials at both the inner and outer vesicle surfaces. The measured
potentials were found to be in good agreement with theoretical
values calculated using the known surface charge density. The basis
for measurement of the surface potential of the inner mitochondrial
membrane using ESR probes is discussed by Hartzog et al. [103].
The question of the proper choice of models to describe charged
phospholipid bilayers has been addressed by Winiski et al. [104]. By
measuring the adsorption of a fluorescent monovalent anion and a
paramagnetic divalent cation to both positive and negative mem
branes, predictions of a discreteness-of-charge theory were tested
and compared to those of the familiar Gouy-Chapman-Stern (GCS)
theory which assumes that the charge is uniformly smeared over the
ESR OF BIOLOGICAL MEMBRANES 161
surface. All experimental results, including those obtained for
membranes in the gel phase, are in closer agreement with the
predictions of the GCS theory than with a discreteness-of-charge
model. A similar conclusion is reached by Hartsel & Cafiso [105]
based on measurements of the surface potentials of model mem
brane systems using a new series of negatively charged, spin-labeled
alkylsulfonate probes (see Figure 5), in conjunction with positively-
charged spin labels.
Keana et al. [106, 107] describe the preparation of a new series of
9,10-disubstituted-2-anthracenyl rerr-butyl nitroxides aimed at deve
loping a nitroxide based ESR probe for singlet oxygen. One of the
anthracenyl nitroxides reacts with singlet oxygen to give an endo-
peroxide both in organic solvent and when the nitroxide is incor
porated in the bilayers of dimyristoylphosphatidylcholine vesicles.
The reaction is sufficiently rapid and the resulting spectral changes
are sufficiently characteristic for the anthracenyl nitroxide to serve
as an ESR-based probe for singlet oxygen.
The orientation and mobility of a square-planar copper complex
has been investigated by Subczynski et al. [108]. By using the spin
label, TEMPO (2,2,6,6-tetramethylpiperidine-l-oxyl), the copper,
complex, CuKTSM2, a derivative of a potent antitumor drug, was
found to partition favorably into dimyristoylphosphatidylcholine
vesicles. Saturation recovery ESR techniques were used to measure
the effect of bimolecular collisions between Cu complexes and
stearic acid spin labels on the spin-lattice relaxation time of the
nitroxide moiety. The translational diffusion constant of the com
plex was found to be about ten times greater than that of the lipids
in the fluid phase.
Erriu et al. [109] use differential scanning calorimetry in combina
tion with ESR of the spin probe, TEMPO, to examine the phase
~0&CH2)nO—( N— 0
FIGURE 5 Chemical structure of the alkylsulfonate nitroxides described by Hartsel
& Cafiso [105].
162 JOHANNES J. VOLWERK and O. HAYES GRIFFITH
behavior of Y-irradiated lipid membranes. The sharp variation in the
fraction of TEMPO bound to the lipid phase that occurs at the main
transition temperature is much less sharp with irradiated samples.
At greater irradiation doses the products of irradiation lead to the
formation of phase-separated domains.
Boggs and coworkers [110, 111, 112] have published a series of
studies describing the effect of various factors on the permeability
of lipid vesicles using the water-soluble spin label TEMPOcholine
chloride. At the high concentrations of TEMPOcholine trapped in
these vesicles, exchange broadening markedly reduces the height of
the ESR signal. As the TEMPOcholine is diluted into the sur
rounding aqueous environment due to its leakage from the vesicles,
the signal height is increased, providing a measure of permeability.
The ability of charge isomers of myelin basic protein (BP) to in
crease the permeability of multilamellar vesicles composed of pho-
sphatidylserine/phosphatidylcholine and sphingomyelin/cholesterol/
phosphatidic acid was measured by monitoring the release of
TEMPOcholine [110]. The increase in vesicle permeability caused
by BP was taken as a measure of the degree of perturbation of the
bilayer by the protein. All classes of charge isomers of BP were
more effective at increasing vesicle permeability than polylysine, but
a phosphorylated isomer of BP was less effective than the other
isomers. Small changes in protein charge can apparently influence
the interaction of BP with the lipid bilayer. The same technique was
used to measure the influence of lipid environment and ceramide
composition on antibody recognition of cerebroside sulfate in lipo
somes [111]. The surface expression of the glycolipid, i.e., the
reactivity with antibody, was determined by measuring complement-
mediated lysis of the TEMPOchoIine-containing vesicles. The lipid
environment was varied by using phosphatidylcholine of varying
chain length in a mixed vesicle with cholesterol and cerebroside
sulfate. The ceramide structure was varied by using synthetic forms
containing different fatty acids. Both factors influenced the binding
of the various antibodies in the polyclonal antiserum. In a separate
study, Kurantsin-Mills and Boggs [112] show that a low concentra
tion of serum significantly reduces the permeability of liposomes,
whereas serum is generally regarded to increase permeability. It is
found that the effect of serum on liposome permeability depends on
the compound entrapped as well as the type of lipid used.
ESR OF BIOLOGICAL MEMBRANES
Lagercrantz et al. [113] describe the effect of membrane-active
drugs on the entrapping of the spin label TEMPOcholine into
human erythrocytes that occurs when the cells spontaneously reseal
after being subjected to hyposmolar stress. Chlorpromazine, tri
fluoperazine, nicardipine, amperozide and haloperidol give rise to a
dose-dependent decrease of the entrapping of TEMPOcholine. It is
suggested that these substances exert their action on the resealing
process by interacting with the calmodulin system.
Gwozdzinski [114, 115, 116] describes experiments on the effect
of ionizing radiation and thiol-reactive reagents on the permeability
of porcine and fish (carp) erythrocytes to charged and uncharged
small water soluble spin labels. Both factors affect the erythrocyte
permeability dependent on the nature of thespin label. Forexample,
irradiation of porcine erythrocytes increases the permeability of the
spin probes TEMPO and TEMPOL. The thiol reagents N-ethyl-
maleimide and p-chloro-mercuribenzoate decrease the permeablity
of these spin probes and the effect is greater with irridiated ery
throcytes.
Several papers report use of non-lipid spin probes to determine
internal cellular volumes. The method consists of incubating cells or
vesicles with a permeable spin probe, e.g., TEMPONE (2,2,6,6-
tetramethyl-4-oxopiperidine-N-oxyl), and a non-permeable quench
ing agent, e.g., Na2MnEDTA or other complexes containing transi
tion metals. The quenching agent broadens the external spin probe
signal so that the relative signals before and after addition of the
quenching agent give the internal volume as a fraction of the total
aqueous volume. An example is the paper of Ball et al. [117], who
use this method to measure intrathylakoid aqueous volumes to study
ionic permeability properties of thylakoid membranes and the
shrinking and swelling of thylakoids exposed to salt solutions.
Essentially the same technique is used by Anzyr et al. [118] to
measure the intracellular solvent volume of sclerotic cells of the
fungus Claviceps purpurea. These authors also describe a technique
to determine the total cellular volume (the volume including the cell
walls) using a charged impermeable spin probe. Lomax et al. [119]
use paramagnetic spin probes to quantitate both apparent vesicle
volume and the transmembrane pH gradient ofclosed and pH-tight
membrane vesicles prepared from hypocotyls of Cucurbita pepo
(zucchini) seedlings. A weak acid nitroxide probe, TEMPACID
164 JOHANNES J. VOLWERK and O. HAYES GRIFFITH
(2,2,6,6-tetramethyl-4-carboxylpiperidine-N-oxyl), which partitions
preferentially into basic environments, was used to estimate ApH.
These measurements were used to study uptake of the plant growth
hormone indole-3-acetic acid into the vesicles. It is concluded that
uptake is an active process driven by the pH gradient. Measure
ments using nitroxide spin probes to determine internal volumes,
proton gradients, and electrical potentials also are described by
Mehlhorn et al. [120].
IV. SPIN LABELS COVALENTLY ATTACHED TO
MEMBRANE PROTEINS
The period covered in this review has seen a number of publications
in which new spin labels for covalent attachment to membrane
proteins are described. Goals in designing these covalent spin labels
have been two-fold: (i) To obtain information on lipid dynamics in
the region of contact between membrane proteins and lipids; and
(ii) To obtain information on rotational and translational mobility of
membrane proteins in native membranes and lipid bilayers.
Griffith et al. [121] describe the synthesis of a phospholipid photo-
spin label, an analogue of phosphatidylethanolamine (PE) with
tritium labeled 14-proxylstearic acid esterified at the sn-2 position
and a nitroaryl azido group incorporated in the polar head (Figure
6). Reconstitution of bovine heart cytochrome c oxidase with the
photo-spin label, followed by photolysis, results in very efficient
(50%) covalent attachment of the label to the protein. Line shape
analysis of ESR spectra of the labeled protein as a function of the
total lipid/protein ratio in the samples provides additional support
for the notion that the fraction of lipids normally in contact with the
protein, and not an aggregation artifact, accounts for the motion-
restricted component observed in membranes.
From the same laboratory, McMillen et al. [122] describe the
synthesis, characterization and application of a new series of amine-
specific phospholipid analogues based on the benzaldehyde reactive
group (Figure 7). All of these reagents are either radiolabeled or
spin-labeled or both, thus permitting identification ofprotein regions
ESR OF BIOLOGICAL MEMBRANES
0
II
\ C^HJoC¥°CH3-(C^)3-f>(C^) j j-C-C-O-
H
165
-O-P-O-C^-CHj-NH-^ yMj
FIGURE 6 Chemical structure of the phospholipid photo-spin label described by
Griffith et al. [121].
in contact with the phospholipid head groups as well as investigation
of the dynamic behaviour of lipid acyl chains at the protein-lipid
interface. Reaction of the lipid-benzaldehydes with cytochrome c
oxidase results in preferential labeling of some of the polypeptide
components of this integral membrane protein. ESR spectra of a
spin-labeled lipid benzaldehyde covalently attached to cytochrome c
oxidase exhibit a large motion-restricted component with a line
shape and splitting similar to that of freely diffusable lipid spin
labels, providing independent evidence that coupling of the lipid
benzaldehydes occurs at the protein-lipid interface. Recently,
Kuppe et al. [123] extended this approach with the synthesis of a
cardiolipin benzaldehyde label.
The intrinsic membrane protein rhodopsin reconstituted into lipid
bilayers continues to be a system well-suited for investigation of the
effect of the lipid environment on the conformation and function of
a membrane protein. Baldwin and Hubbell [124] show by ESR using
a novel disulfide spin label (Figure 8) that is covalently linked to
rhodopsin reconstituted in dimyristoylphosphatidylcholine, that the
apparent arrest of the protein at the metarhodopsin I stage is
probably not due to aggregation of the protein in this un-physio-
logical lipid environment. Spin-labeling of approximately two sul-
fhydryl groups per opsin molecule demonstrates that the spin label
is in a relatively mobile environment. Only after hydrolysis of about
166 JOHANNES J. VOLWERK and O. HAYES GRIFFITH
)14-C14C]H2-0-P-0-@-C-H
<TNa+
CH3-C-(CH2)13-([:-(CH2)2-^P-0-fJ-C-H
HjC-o-p-o-A-c-[3m
0~Na+
CH3-(CH2)7-CH=CH-(CH2)7-C0-CH2
0 |
CH3-(CH2)?-CH=CH-(CH2)?-CO-CH
h2c-o-p-o-Q>-c-:3h:
o
)7-C0-CH,CH3-(CH2)7-CH=CH-(CH2
CH3-(CH2)3-C-(CH2)u-C-CO-CH
0 0
0"NaT
H2C-0-P-0-^^-C-H
FIGURE 7 Chemical structures of the phospholipid benzaldehyde labels described
by McMillen et al. [122]. The molecule on the lower left is a water-soluble benzal
dehyde used for comparison with the long-chain membrane-soluble lipid benzal-
dehydes.
75% of the lipids by phospholipase C, which produces extensive
two-dimensional lateral aggregates of rhodopsin molecules, is a
highly motion-restricted ESR line shape observed.
Beth and coworkers [125, 126] have studied the dynamics and
interactions of the anion channel (band 3) in intact human erythro
cytes employing a new membrane-impermeant, homobifunctional
spin-labeling reagent, bis-(sulfo-N-succinimidyl) doxyl-2-spiro-4'-
pimelate (BSSDP) (Figure 9). BSSDP reacts with high specificity
with the extracytoplasmic domain of band 3. The conventional
(linear) ESR spectrum of intact BSSDP-labeled erythrocytes shows
that the spin label is immobilized on the protein and spatially iso
lated. ST-ESR of the same sample indicates that the anion channel
(band 3) in intact erythrocytes exhibits rotational dynamics with an
effectivecorrelation time in the range of 0.1-1 ms at 20°C, consistent
with a strong interaction of this protein with the erythrocyte cyto-
skeleton. Upon lysis of the cells these interactions are significantly
disrupted as indicated by an increase in rotational mobility.
Other groups have employed previously described covalent spin
ESR OF BIOLOGICAL MEMBRANES
If
CNHCH,CH,SS
(CH2)10 i CH3
jQcNOa
-J^C^COOH
FIGURE 8 Chemical structure of the disulfide spin label described by Baldwin and
Hubbell [124].
'03S
BSSDP
FIGURE 9 Chemical structure of the bifunctional spin-labeling reagent BSSDP
described by Beth et al. [125, 126].
labels, in particular maleimide spin labels (Figure 4), to study the
rotational dynamics of membrane proteins. A problem in deter
mining the rotational mobility of membrane proteins is that the
presence of small amounts of weakly immobilized probes can result
in large systematic errors in the (effective) rotational correlation
times determined from ST-ESR spectra. This problem is probably
more severe when monofunctional rather than bifunctional covalent
spin labels are employed. The latter can be expected to be more
168 JOHANNES J. VOLWERK and O. HAYES GRIFFITH
tightly bound thus allowing less motion of the spin label moiety with
respect to the protein matrix.
Squier and Thomas [127] describe a new methodology based on
ST-ESR to minimize the interference from weakly immobilized
probes combining measurements of both spectral lineshape and
spectral intensity. In a separate paper [128] the new experimental
methodology is applied to the measurement of the rotational
dynamics of the Ca-ATPase in sarcoplasmic reticulum. The non
linear Arrhenius plot of protein mobility obtained through line
shape analysis of ST-ESR spectra as a function of temperature
appears to be an artifact. The plot is linear when intensity para
meters (spectral integral) are used or when the spectraare corrected
by subtracting out the weakly immobilized component prior to line
shape analysis. The linear Arrhenius plot is consistent with the
theory for the hydrodynamic properties of a membrane protein of
constant size and shape in a fluid bilayer. Interestingly, Bigelow et
al. [129] show that an Arrhenius plot of the Ca-ATPase activity still
exhibits a clear change in slope at 20°C, while Arrhenius plots of
both lipid hydrocarbon chain and protein rotational mobility are
linear. Thus, neither an abrupt thermotropic change in the lipid
hydrocarbon phase nor a change in the activation energy for protein
mobility is responsible for the Arrhenius breakinCa-ATPase activity.
Lewis and Thomas [130] similarly employed ST-ESR to study the
effects of vanadate on the rotational motion of the same enzyme.
It is shown that decavanadate substantially immobilizes the spin-
labeled protein indicating protein-protein association whereas
monovanadate inhibits the enzyme without a change in protein
mobility.
Fajer et al. [131] use saturation-recovery electron paramagnetic
resonance (SR-EPR), a time-resolved saturation transfer ESR tech
nique, to measure directly the microsecond rotational diffusion of
spin-labeled proteins. This method uses an intense microwave pulse
to saturate a spin population having a narrow distribution of orienta
tions with respect to the magnetic field. The recovery phase is then
analyzed. The saturation recoveries of spin-labeled hemoglobin
tumbling in medias of known viscosities were measured as a func
tion of rotational correlation time and pulse duration and rotational
correlation values estimated from the initial phase of the recovery
were in good agreement with theory. The results demonstrate that
ESR OF BIOLOGICAL MEMBRANES 169
SR-EPR is applicable to the study of the motion of spin-labeled
proteins, and, in principle, has advantages compared to the estab
lished optical techniques.
Hornblow et al. [132] discuss the use of changes in the ratio of
weakly to strongly immobilized ESR signals of erythrocyte mem
branes nonspecifically labeled with the maleimide spin label to
monitor drug-erythrocyte interactions. Factors such as label/protein
ratio, labeling time, temperature, and time after labeling affect the
ratio of weakly to strongly immobilized signals. Thus it is necessary
that all of these factors be rigorously kept constant when this
method is used to monitor drug binding to erythrocytes. Bartosz and
Gaczynska [133] observe that the ratio of low-field amplitudes of
weakly and strongly immobilized signals in ESR spectra of a malei
mide spin label bound to erythrocyte membranes increases pro
gressively during incubation at 37°C. The authors believe this is due to
self-digestion of membrane proteins by endogenous proteases and
this suggests a need for careful interpretation of data from spin-
labeled membrane proteins, especially when longer incubations are
involved.
Freeman et al. [134] describe covalent labeling of cultured thoracic
aorta endothelial cells with a maleimide spin label under conditions
where cell viability is not affected. The ratio of the peak amplitudes
of the weakly and more strongly immobilized species in the ESR
spectrum was irreversibly elevated following incubation with
a superoxide-generating system, indicating increased membrane
fluidity. The endothelial cell membranes appear useful as a model
for assessing radical-mediated cell damage.
Effects of sodium valproate (VPA), a major antiepileptic drug
used in the treatment of generalized epileptic seizures, on mito
chondrial membranes have been studied by Rumbach et al. [135].
VPA-treated mitochondria spin-labeled with a maleimide spin label
show a reduction in the ratio of weakly to more strongly immo
bilized signal compared to control mitochondria. Spectra of mito
chondrial lipids spin-labeled with 5-doxyl stearic methyl ester show,
however, that VPA has no significant effect on the lipid order
parameters. These observations relate to the therapeutic action of
VPA and its effect on membranes. Fujimura et al. [136] study the
effect of glutaraldehyde on the protein and lipid environment of
mitochondria using both covalent (maleimides) and non-covalent
170 JOHANNES J. VOLWERK and O. HAYES GRIFFITH
(stearic acid) spin labels. The data indicate that glutaraldehyde
reduces the motion of the protein-bound maleimide spin label.
Some restriction of lipid mobility near the headgroup region is also
observed.
Molecular properties of cetiedil, an antisickling agent and a vas
cular smooth muscle relaxant, and its interactions with erythrocyte
membranes have been studied by Narasimhan and Fung [137].
Large amounts of cetiedil associate with erythrocyte ghosts and ESR
experiments utilizing maleimide and fatty acid spin labels show that
the motion of both protein and lipid is affected. Binding of cetiedil
is not affected by removal of the spectrin and actin network. Band 3
molecules are implicated in the interaction of cetiedil with the
erythrocyte membrane.
Gwak et al. [138] have studied the interaction between mito
chondrial succinate-ubiquinone and ubiquinol-cytochrom c reduc
tases by differential scanning calorimetry (DSC) and ST-ESR. The
DSC results suggest a specific interaction between the reductases in
the membrane. This idea is supported by ST-ESR data showing that
the rotational correlation time of maleimide spin-labeled ubiquinol-
cytochrome c reductase is increased when mixed with succinate-
ubiquinone reductase prior to embedding in phospholipid vesicles.
Dalton et al. [139] estimate the distance of separation of the
reactive sulfhydryl of D-(3-hydroxybutyrate dehydrogenase from the
bilayer surface. The enzyme, derivatized with the maleimide spin
label in the presence of the cofactor NAD+, was inserted into
phospholipid vesicles and titrated with the spin probes Mn2+ or
Gd3+ until the ESR spectrum was reduced in amplitude to its
limiting value. From this limiting amplitude the radial distance of
closest approach of the paramagnetic ions to the spin label on the
enzyme was estimated to be 17 A. Since the nitroxide moiety of the
maleimide spin label is 8 A from the sulfhydryl group, the two
limiting distances of immersion of the reactive sulfhydryl within the
bilayer are 9 and 25 A, where the shorter distance is considered
more compatible with facile access of the coenzyme to the active site
of the enzyme.
Sharom and Ross [140] describe covalent spin labeling of sialic
acid and galactose residues on the lymphocyte plasma membrane.
Employing specific activation of sugars with periodate or galactose
oxidase, followed by reductive amination with TEMPAMINE, spin
label probes are coupled to glycoproteins or glycolipids at the
ESR OF BIOLOGICAL MEMBRANES 171
lymphocyte cell surface. Binding of several lectins to these prepara
tions produces significant immobilization of cell surface oligosac
charides while others have no effect, dependent on the sugar speci
ficity of the lectin. Binding of lectins to the lymphocyte cell surface
thus seems to have differential effects on the dynamic states of
glycohpids and glycoproteins within the glycocalyx. By a similar
procedure, Farmer and Butterfield [141] use perdeuterated, l5N-
TEMPAMINE spin label to selectively label cell surface sialic acid
of human erythrocyte membranes. This covalent spin label shows
increased sensitivity and decreased data scatter compared to its
protonated 14N-analogue.
Kunicki et al. [142] describe a general method for the production
of nitroxide-labeled antibodies. Fab' fragments are generated from
murine monoclonal antibodies and a free sulfhydryl group is intro
duced in the carboxyl-terminal region of the molecule through
reduction of a disulfide bridge with cysteine. The sulfhydryl group is
then alkylated with maleimide spin label, thereby generating spin-
labeled Fab' fragments. Two monoclonal antibodies, each specific
for a different membrane glycoprotein of human platelets were
tested. The spin-labeled Fab' fragments retain their ability to bind
to these glycoproteins in membranes of intact platelets, indicating
that the Fab' spin probes can be used to monitor integral membrane
protein mobility. However, the ESR spectra indicate considerable
motion of the spin label coupled to the Fab' fragment. Modification
of this approach may be needed for studies where the motion of the
Fab'-antigen complex is of interest.
V. SUMMARY
Spin labeling is now an established spectroscopic technique for
studies of membrane structure and function. What has occurred
during the past five years is a broadening range of applications, the
addition of some new spin labels, and the refinement of some
methods of analysis. Several different strategies have evolved for
answering specific biological questions. One of the most popular
approaches is to use a freely-diffusing spin label that mimics a
172 JOHANNES J. VOLWERK and O. HAYES GRIFFITH
naturally-occuring component of the membrane. For example, a
lipid analog containing the spin label can be introduced into the
system of interest. Changes in segmental motion of lipid chainswith
temperature, change of composition of the membrane, or upon
addition of other lipid soluble molecules are detected through line
shape changes in the ESR spectra and correlated where possible
with membrane function (membrane receptor, ion channel, or
enzymatic activity). An important variation on this approach is the
study of lipid-protein interactions usingfreely diffusable spin labeled
lipids. Quite fortuitously, the time scale of the ESR experiment is
much better that that of NMR and most other spectroscopic tech
niques for the study of lipid-protein interactions. Lipid in contact
with protein exhibits a motion-restricted spectral component readily
distinguishable from the bilayer line shape. In most cases the lipid
exchange on and off the protein surface does not average this
information. From the ratios of the two components, the equi
librium constant for protein binding by one lipid can be determined
relative to a reference lipid. The presence of the spin label has been
shown not to perturb the equilibrium. Considerable progress has
been made in characterizing the relative binding or occupancy of
sites around the hydrophobic protein surface using this approach.
Many questions remain including two central ones: Why do mem
branes contain a diversity of lipids and what is the significance of the
asymmetry in lipid composition? The challenge experimentally is to
extend the range of lipid/protein ratios that can be examined.
A very different approach is to covalently attach a spin label to
proteins. New labels have been added in recent years and the tech
nique of saturation transfer ESR has extended the time scale to the
submillisecond region. In these studies the biochemistry is assuming
an increasingly important role. The results are only as good as the
specificity and site of labeling, so that in many cases the protein of
interest is isolated, labeled, and then reintroduced into the bio
logical system.
Other types of information used in studies of biological mem
branes include exchange interactions between spin labels (to detect
clusteringof labels) and dipolar effects (to determine the orientation
of probes). Not all spin labels are inert. The susceptability of some
spin labels to chemical reduction continues to be useful in deter
mining the kinetics of transbilayer movement. Another application,
outside of the scope of this review, is the use of spin labels as
ESR OF BIOLOGICAL MEMBRANES 173
contrast enhancers in magnetic resonance imaging. With the in
creasing sensitivity of NMR, it is also likely that applications in
membrane research will utilize the effects of spin labels on the
relaxation times of nuclei in NMR.
The future appears to be bright for spin labeling as well as other
spectroscopic labeling methods. With gene cloning and expression,
more proteins, present in minute quantities under physiological
conditions, will become available for biophysical studies. Site
directed mutagenesis, combined with advances in organic chemistry
to produce new spin labels, and advances in magnetic resonance
spectroscopy will make it possible to answer increasingly sophi
sticated questions about the function and regulation of cellular
processes.
ACKNOWLEDGEMENTS
We are grateful to Polly Habliston and Dr. Randall J. Mrsny for
help in performing the literature searches. We thank our colleagues
for providing us with preprints of their most recent work prior to
publication. This work was supported by National Institute of
Health Grant GM 25698.
REFERENCES
1. B.H. Robinson, H. Thomann, A.H. Beth, P. Fajer & L.R. Dalton (1985). in:
EPR and Advanced EPR Studies of Biological Systems, L.R. Dalton Ed.,
CRC, Boca Raton, Fla.
2. G.I. Likhtenstein, A.V. Kulikov, A.I. Kotelnikov & L.A. Levchenko (1986). J.
Biochem. Biophys. Methods 12. 1—28.
3. A. Watts (1985), Biochem. Soc. Transactions 13. 588-593.
4. D. Marsh (1986), in: Supramolecular Structure and Function, G. Pifat-Mrzljak
Ed., Springer, Berlin, FRG, 48-62.
5. D. Marsh (1985), in: Spectroscopy and Dynamics of Molecular and Biological
Systems, P.M. Bayley & R. E. Dale Eds., Academic. London. UK, 209-238.
6. D. Marsh (1985), in: Progress in Protein-Lipid Interactions Vol. 1. A. Watts &
E.H.H.J.M. de Pont Eds. Elsevier, Amsterdam. Ncth.. 143-172.
7. J.H. Davis (1986), Chem. Phys. Lipids 40(2-4). 223-258.
8. P.F. Devaux (1985), in: The Enzymes of Biological Membranes (2nd Ed.) Vol.
1, A.N. Martonosi Ed., Plenum, New York, N.Y., 259-285.
9. P.F. Devaux & M. Seigneuret (1985). Biochim. Biophys. Acta 822(1). 63-125.
174 JOHANNES J. VOLWERK and O. HAYES GRIFFITH
10. M. Seigneuret, J. Davoust & P.F. Devaux (1985), in: Structure and Dynamics of
Molecular Systems, R. Daudel Ed., Reidel, Dordrecht, Ncth., 171-185.
J. Leaver & D. Chapman (1985), in: The Enzymes of Biological Membranes
(2nd Ed.) Vol. 1, A.N. Martonosi Ed.. Plenum, New York, N.Y., 205-227.
M.B. Sankaram & K.R.K. Easwaran (1985), Proc. Indian Acad. Sci. (Chem.
Sci.) 95(1-2). 103-115.
H.M. Swartz (1985), in: Physical Methods on Biological Membranes and their
Model Systems, F. Conti, W.E. Blumberg, J. de Gier & F. Pocchiari Eds.,
NATO ASI Ser., Ser. A: Life Sci., Vol. 71, 39-53.
14. P.D. Morse (1985), in: Struct. Prop. Cell Membr. Vol. 3, G. Benga Ed., CRC,
Boca Raton, FL, 195-236.
I.C.P. Smith & K.W. Butler (1985), in: Effects of Anesthesia, 1-11.
S. Fleischer & J.O. Mclntyre (1985), in: Achievements and Perspectives of
Mitochondrial Research Vol. I, E. Quagliariello et al., Eds.. 347-356.
M.A. Hemminga (1987), J. Chem. Soc. Faraday Trans. /, 83, 203-210.
J.H. Freed (1987), in: Rotational Dynamics of Small and Macromolecules in
Liquids, T. Dorfmuller & R. Pecora Eds., Springer-Verlag, F.R.G., in press.
19. D.D. Thomas (1985), in: The Enzymes of Biological Membranes (2nd Ed.) Vol.
1, A.N. Martonosi Ed., Plenum, New York, N.Y., 287-312.
D.D. Thomas (1986), in: Techniques for the Analysis of Membrane Proteins,
R.J. Cherry & I. Ragan Eds., Chapman & Hall, London, UK, 371-431.
D.D. Thomas (1986), Nature (London) 321(6069). 539-540.
D.D. Thomas, T.M. Eads. V.A. Barnett, KM. Lindahl, DA. Momont & T.C.
Squier, in: Spectroscopy and Dynamics of Molecular and Biological Systems,
P.M. Bayley & R.E. Dale Eds., Academic, London, UK. 239-257.
M. Seigneuret, E. Favre, G. Morrot & P.F. Devaux (1985), Biochim. Biophys.
Acta 813(2), 174-182.
A. Zachowski, C.T. Constantin, F. Galacteros & P.F. Devaux (1985), J. Clin.
Invest. 75(5), 1713-1717.
A. Zachowski, P. Fellman & P.F. Devaux (1985), Biochim. Biophys. Acta
815(3), 510-514.
A. Zachowski, E. Favre, S. Cribier, P. Herve & P.F. Devaux (1986), Bio
chemistry 25(9), 2585-2590.
A. Zachowski, A. Hermann, A. Paraf & P.F. Devaux (1987). Biochim. Biophys.
Acta 897(1), 197-200.
A. Sune, P. Bette-Bobillo, A. Bienvenue, P. Fellmann & P.F. Devaux (1987),
Biochemistry 26, in press.
29. T.M. Fong & M.G. McNamee (1986), Biochemistry 25(4), 830-840.
30. H. Tanaka & J.F. Freed (1985), J. Phys. Chem. 89(2). 350-360.
R.D. Pates & D. Marsh (1987), Biochemistry 26. 29-39.
P.C. Jost, O.H. Griffith, R.A. Capaldi & G. Vanderkooi (1973), Biochim.
Biophys. Acta 311, 141-152.
P.C. Jost & O.H. Griffith (1980), Ann. N.Y. Acad. Sci. 348. 391-407.
B.C. Kunz, M. Rehorek, H. Hauser, K.H. Winterhalter & C. Richter (1985),
Biochemistry 24(12). 2889-2895.
I.D. Volotovski, N.J.P. Ryba & A. Watts (1985), Biochem. Biophys. Res.
Comm. 129(2), 517-521.
J. Nishiyama & T. Kuninori (1985), Agric. Biol. Chem. 49(9). 2557-2561.
11
12
13.
15.
16.
17.
18.
23
24
25.
26.
27
28
31.
32.
33.
34.
35.
36.
ESR OF BIOLOGICAL MEMBRANES 175
37. A. Rietveld, G.A.E. Ponjee, P. Schiffers. W. Jordi, P.J.F.M. van de Coolwijk.
R.A. Demel, D. Marsh & B. de Kruiff (1985), Biochim. Biophys. Acta 818(3).
398-409.
38. A. Rietveld, T.A. Berkhout, A. Roenhorst, D. Marsh & B. de Kruiff (1986),
Biochim. Biophys Acta 858(1). 34-46.
39. H. Goerrissen, D. Marsh, A. Rietveld & B. de Kruiff (1986), Biochemistry
25(10), 2904-2910.
40. J.J. Volwerk, P.C. Jost, G.H. de Haas & O.H. Griffith (1986). Biochemistry
25(7), 1726-1733.
41. W.K. Surewicz & R.M. Epand (1985), Biochemistry 24(13). 3135-3144.
42. T.Y. Ahmad, J.R. Guyton, J.T. Sparrow & J.D. Morrisett (1986), Bio
chemistry 25, 4407-4414.
43. M.P. Mims, M.V. Chari & J.D. Morrisett (1986), Biochemistry 25. 7494-7501.
44. V. Nothig-Laslo & G. Knipping (1985). Chem. Phys. Lipids 36. 373-386.
45. P. Baglioni, M. Carla, L. Dei & E. Martini (1987), J. Phys. Chem. 91.
1460-1465.
46. R.D. Pates, A. Watts, R. Uhl & D. Marsh (1985), Biochim. Biophys. Acta
814(2), 389-397.
47. M. Esmann & D. Marsh (1985), Biochemistry 24(14), 3572-3578.
48. M. Esmann, A. Watts & D. Marsh (1985), Biochemistry 24(6). 1386-1393.
49. G.L. Powell, P.F. Knowles & D. Marsh (1985), Biochim. Biophys. Acta 816(1).
191-194.
50. J.J. Volwerk, R.J. Mrsny, T.W. Patapoff, P.C. Jost & O.H. Griffith (1987),
Biochemistry 26(2), 466-475.
51. J.R. Brotherus, O.H. Griffith. MO. Brotherus. P.C. Jost. J.R. Silvius & L.E.
Hokin (1981), Biochemistry 20. 5261-5267.
52. H.L. Scott (1986), Biochemistry 25(20). 6122-6126.
53. D.A. Pink & A.L. MacDonald (1986), Biochim, Biophys. Acta 863. 243-252.
54. D.J. Laidlaw & D.A. Pink (1985), Eur. Biophys. J. 12(3). 143-151.
55. R.D. Mountain, R.M. Mazo & J.J. Volwerk (1986), Chem. Phys. Lipids 40.
36-45.
56. P. Koole & M.A. Hemminga (1985), J. Magn. Reson. 61(1). 1-10.
57. J.M. Boggs & G. Rangaraj (1985), Biochim. Biophys. Acta 816(2). 221-233.
58. J.M. Boggs & J.T. Mason (1986), Biochim. Biophys. Acta 863. 231-242.
59. A. Kusumi, W.K. Subczynski, M. Pasenkiewicz-Gierula, J.S. Hyde & H.
Merkle (1986), Biochim. Biophys. Acta 854(2). 307-317.
60. A. Lange, D. Marsh, K.H. Wassmer, P. Meier & G. Kothe (1985), Bio
chemistry 24(16), 4383-4392.
61. J.-H. Sachse & D. Marsh (1986), J. Magn. Reson. 68, 540-543.
62. M.D. King & D. Marsh (1986), Biochim. Biophys. Acta 863. 341-344.
63. M.D. King, J.-H. Sachse & D. Marsh (1987), J. Magn. Reson. 72. in press.
64. K.I. Florine & G.W. Feigenson (1987), Biochemistry 26. in press.
65. G. Morrot, J.-F. Bureau, M. Roux, L. Maurin, E. Favre & P.F. Devaux (1987).
Biochim, Biophys. Acta, in press.
66. L. Kar, E. Ney-Igner & J.H. Freed (1985), Biophys. J. 48. 569-595.
67. B.G. McFarland & H.M. McConnell (1971). Proc. Nat. Acad. Sci. USA 68(6).
1274-1278.
68. G.B. Birrell & O.H. Griffith (1976), Arch. Biochem. Biophys. 172. 455-462.
176 JOHANNES J. VOLWERK and O. HAYES GRIFFITH
69. S. Bruno, S. Cannistraro, A. Gliozzi, M. De Rosa & A. Gambacorta (1985),
Eur. Biophys. J. 13. 67-76.
70. F. Berleur. V. Roman, D. Jaskierowicz, M. Fatome, F. Leterrier, L. Ter-
Minassian-Saraga & G. Madelmont (1985), Biochem. Pharmacol. 34(17),
3071-3080.
71. J.K. Raison & G.R. Orr (1986), Plant Physiol. SO, 638-645.
72. W.A. Frezzatti Jr., W.R,. Toselli & S. Schreier (1986), Biochim. Biophys. Acta
860, 531-538.
73. M. Mazur, S. Stole & L.I. Horvath (1986). Drugs Exp. Clin. Res. 12(9-10).
841-844.
74. C.-S. Lai & J.S. Schutzbach (1986), FEBS Lett. 203(2). 153-156.
75. S.D. Venkataramu, D.E. Pearson, A.H. Beth, C.R. Park & J.H. Park (1985),
Tetrahedron Lett. 26(11), 1403-1406.
76. D. Acquotti, S. Sonnino, M. Masserini, L. Casella, G. Fronza & G. Tettamanti
(1986), Chem. Phys. Lipids 40. 71-86.
77. J.S. Hyde, J.J. Yin, W. Fronziss & J.B. Feix (1985), J. Magn. Reson. 63(1).
142-150.
78. C.-S. Lai, M.D. Wirt, J.-J. Yin, W. Froncisz, J.B. Feix. T.J. Kunicki & J.S.
Hyde (1986), Biophys J. 50, 503-506.
79. K. Kuroda, K. Kawasaki & S. Ohnishi (1985). Biochemistry 24(17), 4624-4629.
80. S. Yamada & S. Ohnishi (1986), Biochemistry 25(12). 3703-3708.
81. J.F.W. Keana & S. Pou (1985), Physiol. Chem. Phys. Med. NMR 17, 235-240.
82. M.D. King & D. Marsh (1987), Biochemistry 26. in press.
83. C. Coan (1985), J. Biol. Chem. 260(H). 8134-8144.
84. H. Hauser, G.P. Hazlewood & R.M.C. Dawson (1985), Biochemistry 24(19).
5247-5253.
85. F.G. Herring, A. Krisman, E.G. Sedgwick & P.D. Bragg (1985). Biochim.
Biophys. Acta 819. 231-240.
86. A Cooke, D. Collison, F.E. Mabbs & M.J. Earnshaw (1985), J. Exp. Bot.
36(172), 1799-1808.
87. PH. Rosenberg & A. Alila (1985), Acta Pharmacol. Toxicol. 57. 154-159.
88. H. Utsumi, J.-I. Murayama & A. Hamada (1985), Biochem. Pharmacol. 34,
57-63.
89. J.L. James, D.S. Friend, J.R. MacDonald & E.A. Smuckler (1986), Lab.
Invest. 54(3), 268-74.
90. G.R. Munske, J.A. Magnuson & A.F. Barbet (1986). Exp. Parasitol. 62,
423-429.
91. T. Hongo, M. Setaka & T. Kwan (1985). Japan. J. Pharmacol. 39. 59-66.
92. A. Chattopadhyay & E. London (1987), Biochemistry 26(1). 39-45.
93. T. F. Taraschi, A. Wu & E. Rubin (1985), Biochemistry 24(25). 7096-7101.
94. C.C. Curtain, J.L. Atwell, F.D. Looney & A. Zajac-Cade (1985), in: Proc. 18th
Conv. —Inst. Brew. (Aust. N. Z. Sect.), B. J. Clarke, J.V. Harvey, S. Itcovitz
& G.W. Wheatland Eds., 236-243.
95. C.C. Curtain (1986), Trends Biotechnol. 4, 110.
96. G. Hidalgo (1985), Biophys. J. 47, 757-764.
97. G. Scheel, G. Schwarzmann, P. Hoffmann-Bleihauer & K. Sandhoff (1985),
Eur. J. Biochem. 153. 29-35.
98. F.G. Herring, E.W.N. Lam & H.M. Burt (1986), J. Rheumatol. 13. 623-630.
ESR OF BIOLOGICAL MEMBRANES 177
99. H.M. Burt, E. Evans, E.W.N.
Rheumatol. 13. 778-783.
J.P. Earnest, H.P. Limbacher Jr.,
Biochemistry 25(19), 5809-5818.
R.F. Flewelling & W.L. Hubbell (1986), Biophys. J. 49(2). 531-540.
S.A. Sundberg & W.L. Hubbell (1986), Biophys. J. 49(2). 553-562.
G. Hartzog, R.J. Mehlhorn & L. Packer (1986), in: Ion Interact. Energy
Transfer Biomembr., G.C. Papageorgiou et al. Eds., Plenum N.Y., 39-63.
104. A.P. Winiski, A.C. McLaughlin, R.V. McDaniel. M. Eisenberg & S. Mc
Laughlin (1986), Biochemistry 25(25). 8206-8214.
S.C. Hartsel & D.S. Cafiso (1986), Biochemistry 25(25). 8214-8219.
J.F.W. Keana, V.S. Prabhu, S. Ohmiya & C.E. Klopfenstein (1985), J. Am.
Chem. Soc. 107, 5020-5021.
J.F.W. Keana, V.S. Prabhu. S. Ohmiya & C.E. Klopfenstein (1986), J. Org.
Chem. 51, 3456-3462.
W.K. Subczynski, W.E. Antholine. J.S. Hyde & D.H. Petering (1987). J. Am.
Chem. Soc. 109, 46-52.
G. Erriu, S. Onnis & P. Randaccio (1985), Lett. Nuovo Cimento Soc. Ital. Fis.
44(1), 51-57.
S. Cheifetz, J.M. Boggs & M.A. Moscarello (1985), Biochemistry 24(19).
5170-5175.
S.J. Crook, J.M. Boggs, A.I. Vistnes & K.M. Koshy (1986). Biochemistry
25(23), 7488-7494.
J. Kurantsin-Mills & J.M. Boggs (1986). Biotechnol. Appl. Biochem. 8. 69-74.
113. C. Lagercrantz, T. Larsson & L. Tollsten (1985), Biochem. Pharmacol. 34(1).
31-38.
114. K. Gwozdzinski (1985), Acta Biochim. Pol. 32(2). 127-130.
115. K. Gwozdzinski (1985), Stud. Biophys. 106(1). 43-45.
116. K. Gwozdzinski (1986), Radiat. Environ. Biophys. 25. 107-111.
117. M.C. Ball, R.J. Mehlhorn, N. Terry & L. Packer (1985), Plant Physiol. 78.
1-3.
118. M. Anzur, B. Tome, M. Sentjurc & M. Schara (1986), Biotechnol. Bioeng. 28,
1879-1883..
119. T.L. Lomax, R.J. Mehlhorn & W.R. Briggs (1985). Proc. Natl. Acad. Sci. USA
82, 6541-6545.
120. R.J. Mehlhorn, L. Packer, R. Macey, A. Balaban & I. Dragutan (1986), Meth.
Enzymol. 127, 738-745.
121. O.H. Griffith, D.A, McMillen, J.F.W. Keana & P.C. Jost (1986). Biochemistry
25(3), 574-584.
122. D.A. McMillen, J.J. Volwerk, J. Ohishi, M. Erion, J.F.W. keana, P. C. Jost &
O.H. Griffith (1986), Biochemistry 25(1). 182-193.
123. A. Kuppe, R.J. Mrsny, M. Shimizu, S.J. Firsan, J.F.W. Keana & O.H. Griffith
(1987), Biochemistry, in press.
124. P.A. Baldwin & W.L. Hubbell (1985), Biochemistry 24(11), 2624-2632.
125. A.H. Beth, T.E. Conturo, S.D. Venkataramu & J.V. Staros (1986), Bioche
mistry 25(13), 3824-3832.
126. A.H. Beth, T.E. Conturo, N.A. Abumrad & J.V. Staros (1987). in: Membrane
Proteins, S.C. Goheen Ed., Bio-Rad, Richmond. CA, pp. 371-382.
100.
101.
102.
103.
105.
106.
107.
108.
109.
110.
111.
112.
Lam. P.F. Gehrs & F.G. Herring (1986), J.
M.G. McNamee & H.H. Wang (1986),
178 JOHANNES J. VOLWERK and O. HAYES GRIFFITH
49(4), 921-935.
49(4), 937-942.
(1986), Biochemistry
127 T.C. Squier & D.D. Thomas (1986), Biophys. J.
128. T.C. Squier & D.D. Thomas (1986), Biophys. J.
129. D.J. Bigelow, T.C. Squier & D.D. Thomas 25(1),
194-202.
130. S.M. Lewis & D.D. Thomas (1986), Biochemistry 25(16). 4615-4621.
131. P. Fajer, D.D. Thomas, J.B. Feix & J.S. Hyde (1986), Biophys. J. 50,
1195-1202.
132. H.M. Hornblow, R. Laverty, B.J. Logan & B.M. Peake (1985), J. Pharmacol.
Methods 14, 229-241.
133. G. Bartosz & M. Gaczynska (1985), Biochim Biophys. Acta 821, 175-178.
134. B.A. Freeman, G.M. Rosen & M.J. Barber (1986), J. Biol. Chem. 261(14),
6590-6593.
135. L. Rumbach, C. Mutet, G. Cremel, C.A. Marescaux, G. Micheletti, J.M.
Warter & A. Waksman (1986), Mol. Pharmacol. 30, 270-273.
136. H. Fujimura, Y. Nonaka, T. Tsujimoto & J.-L. Hiraoka (1986), Wakayama
Med. Rep. 28, 53-65.
137. C. Narasimhan & L.W.-M Fung (1986), J. Pharm. Sci. 75(7), 654-659.
138. S. Gwak, L. Yu & C. Yu (1986), Biochemistry 25, 7675-7682.
139. LA. Dalton, J.O. Mclntyre & S. Fleischer (1987), Biochemistry 26, in press.
140. F.J. Sharom & T.E. Ross (1985), Mol. Immunol. 22(5), 521-530.
141. B.T. Farmer & D.A. Butterfield (1985), Anal. Lett. 18(B5). 555-561.
142. T.J. Kunicki, D.J. Nugent, R.S. Piotrowicz & C.-S. Lai (1986), Biochemistry
25(18), 4979-4983.