Electron Spin Resonance Spectroscopy Labeling in Peptide and

Protein Analysis

Peter G. Fajer

in
Encyclopedia of Analytical Chemistry
R.A. Meyers (Ed.)
pp. 5725–5761
 John Wiley & Sons Ltd, Chichester, 2000
ESR LABELING IN PEPTIDE AND PROTEIN ANALYSIS 1

Electron Spin Resonance Spin labels are nitroxide derivatives with a stable unpaired
electron and a functional group for specific attachment to
Spectroscopy Labeling in the protein (covalent or as a ligand). The most popular
covalent sites are cysteine residues, which, if necessary, can
Peptide and Protein Analysis be introduced into the protein structure using molecular
biology techniques.
Peter G. Fajer The physical basis for nearly all ESR applications is
the anisotropy of the nitroxide signal and the sensitivity
Florida State University, Tallahassee, USA
of the ESR spectra to various relaxation pathways. The
interaction between an electron of a spin label and
an external magnetic field depends on their relative
1 Introduction 2 orientations. The splitting and the center of ESR spectra of
2 Historical Perspective 2 an oriented sample are used to determine the orientation
of labeled domains. For samples with little disorder the
3 Sample Preparation 3 orientational sensitivity is better than 1° . The width of
3.1 Nitroxide Spin Labels 3 the signal is proportional to the orientational disorder,
3.2 Labeled Sites 3 which is used to measure conformational heterogeneity of
3.3 Attachment Rigidity 4 proteins.
3.4 Impairment of Function/Structure of If the spin label reorientates itself on the ESR timescale
Labeled Proteins 4 (nanoseconds) then the spectral anisotropy is averaged.
4 Techniques and Instrumentation 4 The extent of averaging defines the ESR line shape which
4.1 Continuous Wave Electron Spin is used to determine the rotational rate and anisotropy
Resonance 4 of motion. The dynamic range of ESR is very broad,
4.2 Saturation Transfer Electron Spin rotational correlation times range from 10 12 to 10 7 s for
Resonance 8 conventional ESR and the sensitivity can be extended to
4.3 Time Domain Methods 8 slower motions (10 3 s) with nonlinear saturation transfer
5 Applications of Spin Labeling 10 electron spin resonance (STESR). Protein (spin label)
5.1 Protein Orientation 10 mobility is used to follow conformational changes, steric
5.2 Protein Dynamics 13 restrictions on the spin label and the formation of large
5.3 Kinetic Experiments 20 complexes.
5.4 Protein Folding 21 Spin labels are also sensitive to the presence of other
5.5 Ligand Binding 22 paramagnetic species. Collisions with water and lipid-
5.6 Distance Measurements 22 soluble relaxing agents provide additional relaxation
5.7 Structural Biology 26 pathways measured by changes in relaxation times. The
probability of these collisions reflects the accessibility of a
6 Conclusion 30
spin label to the relaxant. The periodic patterns along the
Acknowledgments 30 polypeptide chain of this accessibility are used to deter-
List of Symbols 30 mine the secondary and tertiary structure of proteins. In
Abbreviations and Acronyms 31 the presence of another bound spin label or a param-
agnetic metal complexed by histidine residues, spectra
Related Articles 32 become broadened by dipolar or exchange interactions.
References 32 Both mechanisms depend on the distance between the
paramagnetic centers. Thus ESR can be used to determine
intra- and intermolecular distances. The range of sensi-
tivity is 5 – 25 Å and there are intensive efforts to increase
Electron spin resonance (ESR) is a powerful analytical the upper range to >50 Å. ESR as a spectroscopic ruler
tool used in protein and peptide biochemistry. It is used is used in protein structure determination and the investi-
in the determination of secondary, tertiary and quaternary gation of macromolecular assembly processes and protein
protein structure and associated conformational changes. folding.
Protein dynamics and the relative orientation of protein The foremost limitation of spin labeling ESR is the
components in ordered systems can also be measured. The necessity to modify a protein with a spin probe. In some
majority of proteins do not contain unpaired electrons cases, the spin labels may perturb protein function and
whose spin transitions give rise to an ESR signal, hence therefore cannot be used for spectroscopy. However, even
necessitating the use of extrinsic probes called spin labels. an unsuccessful modification that results in functional

Encyclopedia of Analytical Chemistry

R.A. Meyers (Ed.) Copyright  John Wiley & Sons Ltd
2 PEPTIDES AND PROTEINS
loss identifies functional regions of proteins and as such
represents successful ‘‘mutational analysis’’ experiments.
1 INTRODUCTION
A spinning electron orbiting around a nucleus is a
magnetic dipole. When placed in an external magnetic
field, the dipole aligns parallel or antiparallel with the
external field. These two orientations of the magnet
represent two energy levels, with the difference in energy
levels of the electron spin proportional to the strength
of the magnetic field. The electron can be excited from
one level (i.e. parallel dipole orientation) to another
(antiparallel orientation) by an oscillating magnetic field.
The energy of the oscillating field has to match the energy
difference between the two levels. For a free electron
in a magnetic field with a strength of a few hundred
gauss, the frequency range of the exciting field is in the
microwave region of the electromagnetic wave spectrum.
The resonance between the orbiting electron and the
microwave field forms the basis of ESR, also known as
electron paramagnetic resonance or electron magnetic
resonance.
ESR is commonly used to investigate protein and
peptide structure, particularly studies of molecular orien-
tation, protein dynamics and ligand binding. Observation
of a resonance requires samples containing an unpaired
electron, e.g. transition metals or organic radicals. Pro-
teins and peptides are generally not paramagnetic and
therefore require the use of extrinsic probes called spin
labels. Spin labels are derivatives of nitroxides, small
stable organic radicals, which are covalently attached to
protein side chains or to metabolic substrates. In the last
decade, the development of site-directed spin labeling
(SDSL), which utilizes molecular biology to introduce
new labeling sites, has established ESR as a protein
structural determination technique. Patterns of side-chain
mobility, accessibility to quenchers and the measurement
of distances between spin labels have allowed the determi-
nation of the secondary, tertiary and quaternary structure
of proteins.
This article is focused exclusively on spin labeling
applications in protein and peptide biochemistry. The
vast literature on metalloproteins, photosynthesis and
reactive radicals in biology is not discussed here, and
interested readers are directed to the many excellent
reviews on these topics..1 – 6/
2 HISTORICAL PERSPECTIVE
The first ESR experiments were performed by Zavoisky
at the University of Kazan (Russia) during the Second
World War..7/ Inspired by the experiments of Gorter.8/
and Rabbi et al..9/ on paramagnetic relaxation and atomic
beams, Zavoisky demonstrated resonance between
microwaves and the precession of Cu2C ions in a mag-
netic field. Resonance was observed as an absorption
of microwaves whenever the frequency of the oscil-
lating microwave field was equal to the ion precession
frequency.
In the decade following the Second World War, ESR
was the domain of physical chemists and physicists,
with the first biological applications appearing in the
mid-1950s. This early work included structural studies
of metalloproteins,.10/ measurement of free radicals in
biological tissues,.11/ carbonized carbohydrates,.12/ and
X-ray irradiated silk and hair..13/ Assenheim provides an
excellent review of this early work with intrinsic ESR
signals..14/
In 1965, McConnell introduced extrinsic spin labels
designed to label proteins. Using nitroxide derivatives
first synthesized in Russia,.15,16/ McConnell et al. demon-
strated a helix – coil transition of a polylysine peptide..17/
Since then, ESR spin labeling has been used to study con-
formational changes in a number of proteins modified by
maleimide nitroxides, which specifically target cysteine
residues. However, reliance on the naturally occurring
cysteine residue was a severe limitation. The SDSL strat-
egy developed by Hubbell in 1989 employs molecular
biology to introduce new cysteines for spin label attach-
ment. The use of SDSL to scan the protein sequence
with cysteines has stimulated the resurgence of ESR as a
structural biology method.
The methodology of ESR was also undergoing an evo-
lution. In 1957, Feher invented electron – nuclear double
resonance (ENDOR) spectroscopy, a combination of
both ESR and nuclear magnetic resonance (NMR),.18/
in which nuclear spin transitions are observed indirectly
by monitoring electron spin transitions. A few years later,
electron – electron double resonance (ELDOR) spec-
troscopy was developed by Hyde et al..19/ and Benderskii
et al..20/ which allowed the measurement of spectral diffu-
sion between distinct spin populations. The development
of spin-echo instruments by Mims et al..21/ introduced
time-domain ESR in the 1960s. This was followed by
Fourier transform electron spin resonance (FTESR),
developed independently in the 1980s by Eliav and
Freed,.22/ Dinse et al..23/ and Bowman..24/ The first spin
label applications appeared in 1986 when Gorester and
Freed performed two-dimensional (2-D) FTESR experi-
ments to measure spin dynamics..25/
ESR moved towards high field (high frequency) with
Lebedev et al.’s construction of a 150-GHz spectro-
meter,.26/ followed by Freed et al.’s 250-GHz spec-
trometer, which was based on quasi-optics. The latter
instrument was used extensively to investigate spin labels
ESR LABELING IN PEPTIDE AND PROTEIN ANALYSIS
in biological systems..27/ Ultra-high-field spectrometers
operating at 550 GHz now exist and their application to
nitroxide labels is only a question of time..28/
Readers interested in the history of ESR are directed
to a series of historic recollections of the ESR pioneers
assembled by Eaton et al..29/
3 SAMPLE PREPARATION
3.1 Nitroxide Spin Labels
Proteins are ESR silent, with the exception of metallopro-
teins, and must therefore be ‘‘labeled’’ with paramagnetic
probes. These probes, or spin labels, are nitroxide deriva-
tives containing an unpaired electron in the pp orbital
of the N O bond (Figure 1a – c). The nitroxide radical
is stable owing to the presence of methyl groups on
neighboring carbon atoms. To limit flexibility, the NO
group is enclosed in either a six-membered piperidine
or a five-membered pyrrole ring. Pyrrole rings with an
unsaturated bond are the least flexible.
The unpaired electron in the pp orbital also interacts
with the spin of the nitrogen nucleus, splitting the ESR
signal into resonances corresponding to different nitrogen
nuclear manifolds. Thus, the number of resonant peaks
depends on the nitrogen isotope, three for 14 N and two
for 15 N. 15 N labels have the advantage of less spectral
dispersion which increases the signal amplitude 1.5-fold
in conventional ESR and allows for full spectral coverage
in FTESR. Reduction of the nuclear manifolds also
simplifies the interpretation of nuclear relaxation and
accelerates computer simulations of ESR line shapes.
15
N labels, however, are considerably more expensive
than 14 N and only a handful of them are available
commercially.
A weaker interaction occurs between the electron spin
and the hydrogen nuclei of the ring and methyl groups.
Each resonance peak is split by the nuclear spin, but
the splittings are unresolved, resulting in a broad peak.
The broadening can be removed by the substitution
of hydrogen with deuterium which increases the peak
height 1.5-fold for Gaussian and 5-fold for Lorentzian
lines.
N N N (c)
(a) O (b) O (c) O
Figure 1 Commonly used nitroxides: (a) six-membered piperi-
dine ring; (b) saturated five-membered pyrroline ring; (c) un-
saturated pyrrolidine ring.
3
3.2 Labeled Sites
Nitroxide spin labels are used either covalently as modi-
fiers of selected amino acids or noncovalently as analogs
of substrates or enzymatic cofactors. The specificity of
the label is conferred by the functional group attached
to the nitroxide. For example, maleimide, iodoacetamide,
indanedione and a-ketone groups attached to the nitrox-
ide moiety target cysteine residues, while lysines are
modified by activated esters in Figure 2(a – d). Attach-
ment of the nitroxides by disulfide bonds allows for
reversible modification. Reduction of the disulfide bonds
with a mild reducing agent yields the unmodified pro-
tein. Bifunctional spin labels with two linker groups
facilitate attachment to two sites on a protein, reduc-
ing probe mobility with respect to the protein. The ability
to engineer neighboring attachment sites in a protein
using molecular biology is likely to increase the use of
bifunctional labels.
The molecular biology revolution has had a profound
impact on spin label ESR. The limitations of using
naturally occurring binding sites are circumvented by the
site-directed spin-labeling method pioneered by Hubbell
et al..30/ In SDSL, native cysteines are mutated out
and new cysteines are introduced at desired vantage
points. The power of this method is best illustrated
by cysteine scanning where each residue along the
polypeptide chain is changed to a cysteine and labeled
with nitroxide.
Noncovalent labels are used in the investigation of
active sites, e.g. substrate or cofactor analogs, adenosine
triphosphate (ATP) or nicotinamide adenine dinucleotide
(NAD) nitroxide adducts (Figure 3a – c). The binding
and function of these substrates are often not compro-
mised by the presence of the nitroxide. Both approaches
H3C
O
O N O S
O
S
N N
(a) O (b) O
O
O
N
H C O O
N CH2
I C O
N N
O (d) O
Figure 2 Various spin labels used in covalent modification of
proteins: (a) maleimide spin label; (b) methyl thiosulfonate spin
label; (c) iodoacetamide spin label; (d) hydroxysuccinamide
(lysines).
4 PEPTIDES AND PROTEINS

NH2 The extent of probe motion is estimated by immobilizing

N N
the protein on either glass or ion-exchange beads and
O O O comparing spectral parameters such as effective splitting
−O P −O
N N
P O P O (in conventional ESR) or line-height ratios (in STESR) to
O
O− O− O− H H their rigid limit values. Alternatively, the protein mobility
H H can be reduced by increasing the medium viscosity, h. The
O H observed spectral parameters can then be plotted against
h (Perrin plots) and extrapolated to infinite viscosity. If
N
the extrapolated values are lower than the rigid limit
(a) O of the nitroxide, or if discontinuities exist in the Perrin
plots, then it can be concluded that the probe moves
O independently of the protein.
N

3.4 Impairment of Function/Structure of Labeled

NH Proteins
N N
Covalent modification of proteins with extrinsic probes
O O O
−O P −O P O P O
N N carries the danger of damaging the function of the
O
O− O− O− H H molecule. Certain labels are innocuous at certain sites
H H while others are not. No generalizations can be made. For
(b) OH H example, out of 32 spin-labeled cysteine mutants of T4
lysozyme, 11 displayed intact activity and 11 had activity
O
N
½50%. Modification of buried residues and residues in
tertiary contacts decreased appreciably the enzymatic
NH2
activity..31/ In KC channels, ion pumping was affected
HN N N by <50% in 46 out of 66 spin-labeled mutants and very
O O O
N N
few were completely inactivated..32/ Therefore, a careful
O P −O P O P O O functional characterization is necessary if the objective of
O− O− O− H H the study is the correlation of structure and function.
H H
OH H The overall global structure of a protein is generally
less affected than its function by labeling. Nitroxide labels
N
O N C
N
are relatively small and a labeled cysteine is not much
H H larger than a tryptophan residue. Surface labeling has
H H been shown to have little effect on protein folding or
(c) OH H
stability. However, when buried residues are labeled, side
Figure 3 (a, b) ATP spin labels and (c) NAD spin label. chains and backbone can shift to accommodate the labels
and restore packing of the protein core..33/

are combined in photoactivated labels with an addi-

tional azido or nitrene groups. The label is guided
by a substrate analog moiety to an active site and 4 TECHNIQUES AND INSTRUMENTATION
photoactivation attaches the label covalently to the
protein. 4.1 Continuous Wave Electron Spin Resonance

3.3 Attachment Rigidity 4.1.1 Theory

Spin labels are attached to a protein via one or more single
bonds about which the nitroxides can rotate on a sub-
nanosecond timescale. In studies of protein orientation
and dynamics, such an independent (librational) motion
is a major hindrance since it averages the orientational
dependence of magnetic tensors – anisotropy. Anisotropy
of magnetic tensors is the basis for ESR orientational and
motional sensitivity as discussed in sections 5.1 and 5.2.
The magnetic moment, µ, of an electron interacts with
an external magnetic field just like a compass needle
interacts with the earth’s magnetic field. The interaction
of electron spin with the field is often referred to as the
Zeeman interaction. The energy (E) of a magnetic dipole
(µ) in a static magnetic field is given by Equation (1):
ED µH .1/
ESR LABELING IN PEPTIDE AND PROTEIN ANALYSIS 5

where H is the magnetic field strength. The magnetic where g is a magnetogyric ratio (ratio of magnetic
moment of an electron is generated by its spin (S) and inertia moments) characteristic of a given electron
(Equation 2): (Equation 5)

µD gbS .2/ gb
gD .5/
h̄
with b denoting the Bohr magneton (intrinsic unit
where h̄ is Planck’s constant. The torque will force the
of electron magnetic moment) and g denoting the
magnetic dipole (µ) to precess around the static field at
spectroscopic splitting factor (relates contribution of spin
a defined frequency, the Larmor frequency, w, given by
and orbital motion of the electron to its total angular
Equation (6):
momentum).
Unlike a compass needle, electron spin is quantized. w D gH .6/
For a single electron, the projection of S on the magnetic
field axis Sz can only take values of š1/2. Thus the energy Substitution of g from Equation (5) into Equation (6)
levels from Equations (1) and (2) are E D š.1/2/gbH, yields the resonant condition hn D gbH stated in Equa-
resulting in an energy gap which increases linearly with tion (3). Applying an oscillating microwave field of the
the magnetic field, E D gbH (Figure 4). An oscillating same frequency as the Larmor frequency cancels the
magnetic field can flip the electrons from one energy level orienting effect of the static magnetic field. The spin
to the other if its own energy, defined by the oscillating then rotates about an axis perpendicular to the static
frequency n, equals the energy gap. Hence, for resonance field direction, periodically aligning itself with or against
between the oscillating field (microwave) and the electron the static field. This is equivalent to dipole (µ) flipping
spin, the condition in Equation (3) has to be satisfied: between the two energy levels.
The extent of microwave absorption, which defines
hn D gbH .3/ the intensity of the ESR signal, is proportional to the
difference in spin populations, N, between the upper and
The resonance condition can also be obtained by lower energy states. The ratio of the two populations is
considering a spinning electron moving in an orbit around determined by the Boltzmann distribution (Equation 7):
a nucleus placed in a magnetic field. From classical  
mechanics, the rate of change of the magnetic moment NC1/2 E
D exp .7/
is proportional to the torque produced by the interaction N 1/2 kT
of the moment and the magnetic field and given by their
The difference in spin populations is increased by either
vector product (Equation 4):
increasing the magnetic field H or reducing the temper-
dµ ature T. For example, at 0.35 T and room temperature
D µ
gH .4/ the population difference is 0.1% but it can be increased
dt
to 13% by reducing the temperature to 3 K. The differ-
ence between the levels decreases with absorption and
S I
+1
an efficient relaxation pathway has to exist to restore the
0 Boltzmann equilibrium. Relaxation pathways include the
−1 dipolar spin – spin relaxation – sharing of energy between
+½ electrons or nuclei – and spin – lattice relaxation – sharing
E hν hν hν vibrational modes with the lattice. They are characterized
−½ by relaxation times T2 and T1 , respectively. Relaxation
−1 times are defined as the time interval between initial
0 perturbation and when the deviation from equilibrium
+1
decays to 1/e of its initial value. The relaxation rates are
additive and their sum defines the width (at half-height)
H0 of the resonance,  (Equation 8):
 
1 1 1
D C .8/
g T2 T1
Faster relaxation (shorter T1 or T2 ) results in broader
Figure 4 Energy level diagram of Zeeman and hyperfine
14
interactions for a [ N]nitroxide (S D 1/2, I D 1). The vertical line widths. For paramagnetic ions, the strong coupling of
arrows denote ESR transitions with the resulting first-derivative spins to lattice (short T1 ) produces broad lines. Lowering
spectrum below. the temperature weakens lattice coupling (increases
6 PEPTIDES AND PROTEINS

T1 ) and is commonly used to observe resonance of and thus the g-value can also be described by a tensor
transition metal ions. The coupling of free radicals (Equation 11):
(including spin labels) to the lattice is weak, therefore  
 gxx 0 0 
spin– spin relaxation is more efficient and the line width  
is determined by T2 . g D  0 gyy 0  .11/
Electrons orbiting around a nucleus experience a small
local field produced by the nuclear magnetic moment.
This field enhances or counteracts the external field
depending on the orientation of the nuclear dipole.
This interaction between the nucleus and the electron
is known as hyperfine interaction. Similar to electron spin,
nuclear spin (I) is also quantized to 2I C 1 levels. Since
the selection rule for spin transitions dictates that the total
spin quantum number can only change by 1, the hyperfine
interactions lead to 2I C 1 transitions (Figure 4). In the
case of nitroxide labels, it is the nitrogen nucleus which
interacts with the unpaired electron. For 15 N the nuclear
spin number mI is 1/2 so that two electron transitions are
observed; for 14 N, I D 1 and therefore there are three
electron transitions.
The resonance condition of Equation (3) is thus modi-
fied to include hyperfine interactions, A (Equation 9):
hn D gbH C mI A .9/
The hyperfine interaction between an electron and the
nucleus has both an isotropic and dipolar component. The
magnitude of the isotropic splitting, a0 , is proportional
to the electron spin density on the nucleus. Since the
unpaired electron is located between the oxygen and
nitrogen, increasing the polarity of the medium decreases
the oxygen’s attraction and increases the electron density
on the nitrogen. Thus, a0 is a sensitive measure of the spin
environment.
The pp orbital of an unpaired electron is asymmet-
ric, making the dipolar interactions of the electron and
nucleus orientation dependent. For example, hyperfine
interactions are stronger when the z-axis of the orbital
is aligned with the magnetic field and weaker when
the field is aligned perpendicular. The hyperfine inter-
actions are best described by a second rank tensor,
A (Equation 10):
 
0 0 gzz
In contrast to the hyperfine tensor, the g tensor is rhombic
with typical values gxx ³ 2.0085, gyy ³ 2.0065 and gzz ³
2.0027. The asymmetry of the Zeeman and hyperfine
interactions defines ESR sensitivity to orientation and to
rotational motion.
4.1.2 Electron Spin Resonance Spectrometer
A modern ESR instrument consists of three basic units:
(a) a microwave bridge and resonator, (b) a variable field
magnet and (c) signal amplification circuitry (Figure 5).
Microwaves of the desired frequency are generated by
either a klystron or Gunn diode. Their intensity is adjusted
by an attenuator and transmitted via a waveguide to the
sample chamber/resonator. During resonance, a small
amount of microwaves is reflected from the resonator
and detected by a Shottky diode. To separate the
reflected and incident microwaves, a circulator is placed
between the attenuator and resonator. The circulator
channels the microwaves in a forward direction: incident
microwaves to the resonator and reflected microwaves
to the detector. The bridge often contains an additional
pathway – a reference arm which taps off a small fraction
of the microwaves from the source – which bypasses the
resonator and falls on to the detector to ensure its bias for
the optimal detection of small intensity changes during
resonance.
A static magnetic field is provided by an electromagnet
stabilized by a Hall probe. The field is slowly swept
by varying the amount of current passing through the
electromagnet. In order to decrease microwave noise,
Lock-in amplifier
Reference arm

  Klystron
 Axx 0 0  Circulator
 
A D  0 Ayy 0  .10/
  Attenuator Detector diode
0 0 Azz
Typical values for nitroxide spin labels are Axx ³ Ayy ³ Resonator
,,,,,,,,,,,,,
,,,,,
,,,,,,,,,,,,,

7 G and Azz ³ 35 G. The difference between the x- and

,,,

y-components is small and often the hyperfine tensor is

assumed to be axially symmetric. Field controller Magnet
The Zeeman interaction of the electron spin with the
static magnetic field (Equation 1), is also anisotropic. Hall probe Modulation coils
Asymmetry of orbital motion in the pp orbital results
in different contributions of spin and orbital momenta Figure 5 Block diagram of a typical ESR spectrometer.
ESR LABELING IN PEPTIDE AND PROTEIN ANALYSIS
(a)
(b)
C before a relatively weak spin echo or free induction decay
L These resonators condense the magnetic component of
L′′ H
H ′′
(c) C′
Figure 6 Conventional ESR signals: (a) absorption, V0 ;
(b) first derivative V1 ; (c) STESR spectrum, second derivative,
90° out-of-phase display, V20 .
the resonance signal is encoded by modulating the static
field with a small magnetic field generated by modulation
coils. The modulation field sweeps periodically through
the nitroxide resonance field. Therefore, the changes
in microwave absorption due to resonance occur with
the modulation frequency. A lock-in amplifier selects
and amplifies only the signal which is in phase and in
frequency with the modulation field and rejects all other
microwave fluctuations as noise. The signal detected using
field modulation is proportional to the changes in the
microwave intensity during one cycle of modulation,
i.e. the signal is the first derivative of the absorption
(Figure 6a – b).
The microwave field produced by the klystron is
too weak to induce any detectable absorption by the
sample. Resonant cavities, loop gap resonators (LGRs)
or, more recently, dielectric resonators (DRs) are used
to increase the microwave magnetic field at the sample.
The cavities rely on the generation of a standing wave
pattern of microwaves whose intensity builds up during
cavity resonance. The main drawback of cavities is the
presence of an electric component of the microwave.
The electric component is absorbed by ‘‘lossy’’, aqueous
samples (common in biology) causing sample heating and
loss of cavity resonance. To avoid this problem, sample
volume is restricted to the nodal planes of the electric field,
limiting the usable volume of the cavity and thus resulting
in a low filling factor, h. Cavities are high-Q structures
(Q D Estored /Edissipated ), storing thousands of times more
7
energy than is dissipated on the walls. High Q can only be
achieved within a narrow frequency bandwidth of stored
microwaves as Q D n/n. Small changes in the sample,
cavity geometry or temperature can all cause frequency
shifts and mismatching of the incident microwave with
the cavity. Automatic frequency control (AFC) circuitry
is employed to track the frequency of the klystron to
that of the cavity. However, the AFC feedback response
time limits the deadtime of signal changes in transient
experiments such as stop-flow. In pulse experiments it is
necessary to wait until the energy of the perturbing pulse
is fully dissipated. This ring-downtime is proportional to
Q; thus in the high-Q structures a longer time has to elapse
(FID) signal can be collected.
Most of these problems with cavities have been
overcome by low-Q resonators such as an LGR or DR.
the microwave, separating it from the electric component.
Lossy samples are no longer heated by the electric
component. Small sample volumes and large filling
factors offer an additional advantage especially when
dealing with genetically engineered proteins which are
often purified in picomolar quantities. Furthermore, fast
dissipation of energy and the large bandwidths of LGRs
and DRs make them suitable for pulsed and transient
experiments.
4.1.3 Instrumental Variables Affecting the Electron Spin
Resonance Spectrum
Two instrumental parameters influence the line shape
of experimental spectra: modulation amplitude and
microwave power. The amplitude of the ESR signal
initially increases with the modulation amplitude (Hm ) as
it approaches the intrinsic line width (Hpp ). Maximum
amplitude is attained at Hm D 3.5Hpp for Lorentzian
and at Hm D 1.8Hpp for Gaussian line shapes. Any
further increases in Hm result in a decrease of the signal.
The broadening of Hpp is observed well before the
maximum amplitude of the signal. When Hm D Hpp , the
observed Lorentzian and Gaussian widths are 25% and
15% larger, respectively. As a rule of thumb, modulation
should be kept at one-fifth of the intrinsic line width when
resolution or line width is of importance. The errors in
line width are then <1%.
The ESR signal is proportional to the static microwave
field, H1 , and hence to the square root of the microwave
power (P1/2 ), in the absence of saturation. When the rate
of relaxation lags behind the rate of excitation and when
the spin populations in the ground and excited states are
equalized, the signal saturates and decreases to nil. The
amplitude of the ESR signal (y) for a Lorentzian line
shape as a function of the microwave field is given by
8 PEPTIDES AND PROTEINS
Equation (12):
y0 H1
yD q .12/
.1 C H12 g2 T1 T2 /3
where y0 is a field-independent parameter. Hence it
is important to keep power levels well below the
maximum amplitude whenever spectra are used to
quantify the number of spins. Line-width distortions are
less pronounced than those due to the modulation field,
but at powers giving maximum amplitude, the observed
line width increases 1.2 times over the intrinsic line width.
4.2 Saturation Transfer Electron Spin Resonance
4.2.1 Qualitative Theory
STESR was developed to study slower molecular dynam-
ics with rotational correlation times (tr ) of >200 ns..34/
The timescale of conventional ESR is determined by
the spin – spin relaxation time T2 (nanoseconds). The
ESR timescale can be extended to a longer spin– lattice
relaxation time T1 (microseconds) if a signal sensitive to
spin saturation is observed. This can be done by saturating
the signal with intense microwaves, creating a ‘‘hole’’
in the absorption spectrum, and subsequently observing
signal recovery. When the saturating microwave is
switched off (or decreased to nonsaturating levels)
the signal recovers with the rate determined by the
spin– lattice relaxation time, T1 . The onset of motion
provides an additional relaxation mechanism: spectral
diffusion. The saturation is relieved as the resonating-
saturated spins rotate away from the resonance field and
the unsaturated spins come into resonance. The ‘‘hole’’
broadens out across the spectrum and the intensity of
the signal increases. The second harmonic, out-of-phase
ESR signal (V20 ), collected at moderate saturation, is
particularly sensitive to spectral diffusion. The line shape
of V20 , in the presence of saturation, bears a strong
resemblance to the absorption spectrum with the intensity
lowered in the spectral regions most sensitive to the
spectral diffusion (see Figure 6c).
4.2.2 Instrumental Parameters
The STESR signal is influenced by nitroxide relaxation
times, spectral diffusion, spin saturation level and the
modulation frequency with which the ‘‘hole’’ is observed.
Hence the instrumental parameters which affect any
of these must be precisely controlled. The saturating
microwave field averaged over the sample volume is set
to 0.25 G. The microwave power is adjusted to this level
using the microwave field conversion factor (c), corrected
for a filling factor (h) and dielectric losses, which lower
the Q factor (Equation 13):
hH12 is D chPQ .13/
where P is incident microwave power. The power-to-
field conversion factor is determined experimentally by
the saturation of Fremy salt [peroxylamine disulfonate
(PADS)] for which the half-saturation field is 0.1067 G.
The modulation frequency and amplitude, which deter-
mine the frequency of the stepping on- and off-resonance,
i.e. the interval between burning and observing the
‘‘hole’’, must also be calibrated. Modulation broaden-
ing of a narrow line-width sample, e.g. Fremy salt, is used
for this purpose. The observed line width (Hpp ) is dom-
inated by modulation broadening when the modulation
amplitude is ¾10 times the intrinsic line width [Hpp .0/],
i.e. Hpp D Hm Hpp .0/. Commonly used values for
the modulation field are 5 G and 50 kHz.
Finally, since the V20 STESR signal is 90° out of phase
(phase quadrature) with modulation, the precise phase
nil must be found. An error of 1° in setting the phase
quadrature can result in significant line-shape changes
due to leakage of a more intense in-phase signal. The
most popular phase nulling method is by interpolation
of the unsaturated in-phase signal: two or three readings
are taken within 15° on each side of the putative nil
and the phase at which the signal is zero is found
by linear interpolation. A general description of the
experimental procedures and calibration can be found
in Fajer and Marsh.35/ and Squier and Thomas..36/ Digital
post-acquisition methods have also been proposed but
are not widely used.
As a footnote, protein mobilities measured by con-
ventional ESR and STESR were independently verified
by optical methods – fluorescence and phosphorescence
anisotropy. Bovine serum albumin labeled with a dual
probe bearing a spin label moiety and the optical
probe eosin was measured using optical methods and
ESR/STESR..37/ The agreement between the fluores-
cence/phosphorescence and ESR was excellent.
4.3 Time Domain Methods
Time domain ESR relies on the perturbation of the equi-
librium magnetization by an intense microwave pulse
which is then followed by one of the following: (a) con-
ventional ESR to observe the return of magnetization
to equilibrium – saturation recovery ESR; (b) refocusing
of the magnetization in the xy-plane – spin-echo ESR; or
(c) free induction decay (FID) of the magnetization in the
xy-plane which is then Fourier transformed (FTESR).
The development of time domain ESR posed a
formidable technical challenge. The microwave pulse
must be short (a few nanoseconds) and strong enough to
ESR LABELING IN PEPTIDE AND PROTEIN ANALYSIS
cover a 70-G wide spectrum of nitroxides. The resonators
must dissipate the pulse energy within tens of nanosec-
onds before the loss of magnetization coherence and the
signal must be digitized with a subnanosecond dwell time
owing to the short nitroxide relaxation times. Fortunately,
technological advances in microwave sources, resonator
design and data acquisition electronics in the last decade
have facilitated development of commercial Fourier
transform spectrometers and the time domain method has
become increasingly popular. The various time domain
ESR techniques are illustrated in Figure 7(a – c).
Saturation recovery electrospin resonance (SRESR) is
a hybrid of continuous wave and pulse methods in which
the pulse saturates a spin population at a desired field
thereby creating a ‘‘hole’’ in the absorption spectrum
(Figure 7a). The kinetics of recovery are determined
by various relaxation pathways: spin– lattice relaxation,
nuclear relaxation, Heisenberg spin exchange (HSE) or
spectral diffusion. These competing pathways can be
resolved by varying the pulse duration. SRESR has been
used successfully in the determination of spin– lattice
correlation times and spin exchange.
Spin-echo electron spin resonance (SEESR) uses a
sequence of pulses; in Hahn echo a 90° pulse is followed
by a 180° pulse t1 seconds later (Figure 7b). The first
pulse tips the magnetization into the xy-plane where
individual spins rotate with their respective Larmor
frequency, w. The difference in Larmor frequencies,
Microwave
Pulse c.w. observation
Signal intensity
Signal recovery
(a) Saturation
t1
(b) 90° 180°
90° 90° 90°
t1 Tm
Preparation Mixing
(c)
Figure 7 Time domain ESR methods: (a) saturation recovery;
(b) spin echo; (c) 2-D FTESR (ELDOR).
9
which arises from different resonant fields, leads to
dephasing of the magnetization in the xy-plane which
is then refocused by the 180° pulse. Spins lagging by
wt1 before the refocusing pulse are now wt1 ahead.
At time 2t1 , spins are brought into coherence and an
echo is formed. In this way static differences in Larmor
frequency due to different resonant fields or different
local fields (inhomogeneous broadening) are annihilated.
The dependence of the echo amplitude on time t1 reveals
Larmor frequency fluctuations that cannot be refocused
by the 180° pulse. These fluctuations contain information
about molecular dynamics, spin exchange and dipolar
interactions. The decrease of the spin echo as a function of
t1 is a measure of the T2 relaxation time. The decay of the
echo amplitude is often recorded as a function of spectral
position by stepping the magnetic field, resulting in a
2-D spectrum: an inhomogeneously broadened spectrum
along the field axis and a homogeneous line shape on
the t1 axis. Since the inhomogeneous broadening often
obscures a multitude of phenomena affecting ESR line
width, then the ability to obtain a pure, homogeneously
broadened spectrum is of considerable value.
2-D FTESR is the most versatile technique of time
domain ESR (Figure 7c). All the spins are excited
simultaneously with a strong, short microwave pulse
which tips the magnetization into the xy-plane. The length
and strength of the signal determine the spectral range
covered, e.g. for nitroxides with a 200-MHz spectral range,
2-kW pulses 5 ns in duration are needed.
Coherently excited spins precess about a magnetic
field at their Larmor frequency. This precession can
be detected as an oscillating signal in the xy-plane
which decays in time as the spins lose coherence. This
FID signal is Fourier transformed into the frequency
domain to yield an absorption spectrum. Application of
two or more pulses spaced by varying intervals allows
sampling of spin coherences in multiple dimensions. In
the simplest of these experiments, two-pulse spin-echo
correlation spectroscopy (SECSY), the first pulse tips
the magnetization into the xy-plane where the spins
become frequency labeled during the evolution time t1 .
A second pulse reverses the magnetization during the
collection time t2 , canceling inhomogeneous broadening.
Fourier transformation with respect to t1 and t2 yields an
Echo t2 absorption spectrum along the f2 axis and a homogeneous
line width along the f1 axis. Thus SECSY is an FTESR
equivalent of field-stepped SEESR. The isolation of the
homogeneous line shapes out of an inhomogeneously
broadened spectrum is used to study molecular dynamics.
FID
Three pulse sequences are used in 2-D ELDOR
t2
experiments (e.g. Figure 7c). The first pulse creates a
transverse magnetization in the xy-plane which evolves
for time t1 . The second pulse stores this frequency
encoded magnetization along the z-axis allowing for
10
ude
Magnit
−2
4
ω −1
1 /2
π 22
−4
(M
Hz 45
0
ω2
−3
)
68
Figure 8 2-D ELDOR of PD-TEMPONE. In addition to auto-
peaks occurring at f1 D f2 , there are off-diagonal cross-peaks
due to HSE..25/ (Reproduced with permission from J. Gorcester,
J.H. Freed, J. Phys. Chem., 88, 4678 – 4693 (1986).)
magnetization transfer to take place during mixing time
TM . A third pulse transforms the magnetization back
into the xy-plane where it is observed during time t2 .
Magnetization transfer changes the resonant frequency
of a spin from fa to fb , creating an off-diagonal cross peak
at (fa , fb ) (Figure 8). The time evolution of the amplitude
of the cross peak, measured by varying TM , is used to
determine the time course of magnetization transfer. The
biophysically relevant phenomena causing magnetization
transfer include HSE, modulation of dipolar interactions,
nuclear flips and, most importantly, spectral diffusion due
to rotational motion.
5 APPLICATIONS OF SPIN LABELING
The orientational difference of magnetic interactions
(referred to anisotropy) forms the basis of spin label-
ing techniques in biological research. In the absence of
motion, each field position corresponds to a defined ori-
entation of the label with respect to the field. The intensity
of the signal at a particular field position is directly pro-
portional to the population of molecules with that given
orientation. Hence an ESR spectrum can be used to
determine the range of orientations present in a sample.
Partially or fully averaging the g- and hyperfine tensor
anisotropy results in spectral line shapes determined by
the frequency and amplitude of molecular motion. ESR
can also be used to measure intra- and intermolecular
distances. The presence of paramagnetic centers in the
vicinity of spin labels modulates spin relaxation pathways
PEPTIDES AND PROTEINS
in a distance-dependent manner. In this section we shall
discuss how ESR is used in the investigation of molecu-
lar orientation, molecular dynamics, ligand binding, intra-
and intermolecular distance measurements and the deter-
mination of various levels of proteins structure.
For each of these applications a qualitative description
of the physical principles allowing for these measurements
will be given, followed by examples. More extensive
reviews of these topics can be found in a mono-
graph by Lichtenstein,.38/ a series edited by Berliner
74
and Reuben.39,40/ and separate reviews by Hubbell
48 et al.,.41 – 43/ Marsh and Horvath,.44/ Millhauser et al..45,46/
)
Hz and, most recently, Hustedt and Beth..47/
22
(M
π
/2 5.1 Protein Orientation
5.1.1 Orientation of a Single Molecule
The anisotropy of the Zeeman and hyperfine interactions
confers orientational sensitivity to ESR spectra. Nitroxide
spin labels with a z-axis parallel to the magnetic field
generate a spectrum with a splitting of 70 G. Spins
oriented perpendicular to the field display a splitting
of 14 G (Figure 9).
The center position of the spectrum, determined by the
g-tensor, is sensitive not only to the position of the z-axis
but also to the orientation of the x- and y-axes. At 9 GHz
the center is shifted 5 G downfield (left) for a spin with its
y-axis aligned with the magnetic field and another 4 G for
spins with x-axis parallel to H0 (Figure 9).
The effective g- and hyperfine splitting tensors for a spin
placed at an arbitrary polar angle (q, f) with respect to the
field are given by Equations (14) and (15), respectively:
g.q, f/ D gxx sin2 q cos2 f C gyy sin2 q sin2 f
C gzz cos2 q .14/
z
H0 || x
,, ,,,,,
,, ,,
x O N
y
H0 || y
H0 || z
Figure 9 Orientational sensitivity of ESR spectra. Splitting of
the spectrum changes when the z-axis of nitroxide rotates with
respect to the magnetic field. The center of the spectrum changes
when nitroxide rotates about any axis.
ESR LABELING IN PEPTIDE AND PROTEIN ANALYSIS 11

A2 .q, f/ D A2xx sin2 q cos2 f C A2yy sin2 q sin2 f

C A2zz cos2 q .15/

It follows, then, that the resonance field for a given

orientation (q, f) is given by Equation (16):
hn
Hres .q, f, mI / D C mI A.q, f/ .16/
bg.q, f/
Equation (16) describes the orientational resolution
of ESR line shapes: a spin with a specific orientation (a)
can be found at a defined position along the field
axis. The ESR intensity at any field position is directly
proportional to the number of spins at the orientation
defining Hres . An ESR spectrum can also be considered
as an orientational distribution function, N.q/. N.q/
is approximated by an orthonormal set of spherical
harmonics and has been developed and applied to
samples with cylindrical and planar symmetry..48,49/
Alternatively, the orientation can be modeled in terms
of a Gaussian distribution with a width q and center q0
(Equation 17):
 
.q q0 /2 (b)
r.q/ D exp ln 2 .17/
q2
The ESR spectrum, Y.H/, is created by calculating
a resonance field Hres for every q within the Gaussian
distribution of orientations and placing a Lorentzian first
derivative line width at Hres with the intensity weighted
by r.q/ (Equation 18):
.H Hres /Hpp
Y.H/ D r.q/ .18/
[.H Hres /2 C Hpp
2 ]2

where the peak-to-peak width of the Lorentzian (Hpp )

is defined by the spin– spin relaxation time, T2 (Equa-
tion 19):
2
Hpp D p
3gT2
ESR is one of the very few biophysical techniques
directly sensitive to orientational disorder. The spectra in
Figure 10(a – c) illustrate this sensitivity. As the width of
the Gaussian distribution increases, the ESR resonances
broaden to a powder pattern limit which is characteristic
of isotropically disordered spins.
In summary, the parameters describing orientational
distribution (axial, azimuthal angles and their disor-
der) can be obtained from spectral parameters: spectral
splitting, center of the spectrum and the line width
respectively. This can be achieved either by graphical
methods or from the automated fitting of full spec-
tral line-shape parameters. Graphical methods compare
the effective splitting and width of the resonance to
(c)
.19/
Figure 10 Broadening of the spectra with increasing disorder:
Gaussian disorder of (a) š1° and (b) š5° and (c) completely
disordered (isotropic) distribution.
graphs obtained from computer simulations. The auto-
mated method allows modeling of more complex bimodal
distributions, in both q and f, and can be linked to stan-
dard fitting routines such as Levenberg – Marquardt.50/ or
Simplex..51/
5.1.2 Macromolecular Assemblies
Of considerable interest is the orientation of molecules
within macromolecular assemblies, e.g. proteins within
lipid membranes or contractile proteins in the muscle
12 PEPTIDES AND PROTEINS

z
Z Euler angles

x
x′ y′
α y
x ′′ X
(a) γ

O N

H0
(b) Spin label Myosin head Fiber Magnet

Figure 11 Definition of (a) Eulerian angles a, b and g; (b) Eulerian transformations of the spin label orientation through the
protein (myosin head); sample (fiber) frame of reference into the laboratory frame (magnet).

fibers. If the assembly of proteins is ordered and ori- and Lt is their transpose:
ented at a specific angle with respect to the magnetic  
 cos b cos a cos g sin b sin a cos g sin b cos g 
field, the orientational distribution of the labeled compo-  
 sin a sin g C cos a sin g 
nents of the assembly can be easily determined. The  
 
spectra of such samples oriented with the symmetry L D  cos b cos a sin g cos b sin a sin g sin b sin g 
 
axis parallel to the field are the same as for a sin-  sin a cos g C cos a cos g 
 
gle spin (Equation 18). When the orientation of the  
sin b cos a sin b sin a cos b
spin label with respect to the protein is known, the .21/
ESR spectra are interpreted in terms of the orienta- The resonant field for each spin packet is calculated as
tion of the labeled domain with respect to the assembly shown in Equation (22):
which is of biological interest. This is achieved using
hn q
Eulerian transformations between three frames:.52,53/ Hres D C mI A2xz C A2yz C A2zz .22/
molecular frame (defining orientation of the label within bgzz
the protein), sample frame (orientation of proteins
Note that the subscripts of the g- and hyperfine tensors in
within the assembly) and laboratory frame (orienta-
Equation (22) denote elements in the laboratory frame.
tion of the sample in the magnetic field) (Figure 11a
and b). 5.1.2.1 Protein Orientation in Membranes The Eule-
The ESR spectra are simulated taking into account rian transformation approach was introduced by Griffith
the orientational distribution in each of the frames. The et al. to determine lipid orientation in membrane bilayers.
magnetic tensors are rotated from molecular to laboratory The orientation of the spin label nitroxide with respect
axes using directional cosine matrices (L) according to to the lipid molecule is well defined. A stack of lipid
Equation (20): membranes is tilted with respect to the magnetic field
at a known angle, and the spectra can be defined solely
Alab D Ltmol Ltsam Ltlab ANO Llab Lsam Lmol .20/ by the orientational distribution in the sample frame of
reference (sample )..52,54/
where Lmol , Lsam and Llab are cosine matrices defined in Membrane-bound proteins are investigated in a similar
Equation (21) for each of the Eulerian transformations manner. The orientation of a spin-labeled ligand of the
ESR LABELING IN PEPTIDE AND PROTEIN ANALYSIS
erythrocyte anion transporter, Band 3, was determined
by flowing red blood cells into thin, flat, sample cells. The
flow shear oriented the red blood cells parallel to the
flow and the sample cell was rotated both parallel and
perpendicular to the field in order to vary lab ..55/ A global
analysis of the tilt series resulted in a full description
of the label orientation with respect to the normal
membrane axis. The derived spin label orientation was
found to be consistent with orientation results determined
independently by analyzing the anisotropy of motion.
5.1.2.2 Muscle Proteins Similar approaches have
been used to describe the orientation of various muscle
proteins. Force generation in muscle is believed to be
brought about by the reorientation of the myosin cross
bridges. Since ESR is sensitive to the orientation of the
proteins, muscle field proved to be a fertile ground for
ESR applications. Muscle fibers form a naturally ordered
assembly with the fiber axis defining the cylindrical
symmetry required for ESR. Proteins can be labeled
directly in muscle cells or labeled as isolated components
and exchanged for corresponding unlabeled proteins
in the muscle sample. These include most of the thin
filament proteins – actin, troponin C (TnC), troponin
I (TnI) and tropomyosin – and also the thick filament
components – myosin heavy chain, regulatory light chain
and essential light chain. The orientation of many of these
components has been extensively studied as a function
of the intermediate states of the acto-myosin cycle and
also during muscle activation. Thomas and Cooke have
established that in the absence of ATP, myosin heads
attach themselves strongly and stereospecifically to actin.
Muscle relaxation, in the presence of ATP, produced
disorder consistent with head detachment and Brownian
Force
sin
sin
.P
TP
i
DP
yo
yo
.A
-m
M
sin
.A
sin
to
yo
Ac
yo
-m
-m
to
Ac
to
Ac
Figure 12 Orientation of catalytic domain of myosin head
in the intermediate stages of the contractile cycle. As the
head progresses through the cycle, the dynamic disorder is
continually diminished, culminating in the disorder-to-order
transition associated with force generation.
13
motion..56/ Subsequent studies using various nucleotide
analogs to trap the intermediate adenosine triphosphatase
(ATPase) states have revealed a sequence of orientational
changes of the catalytic domain: a nonstereospecific
attachment of transient, weakly bound heads followed
by an equally large orientational disorder of the strongly
attached heads in the prepower stroke state..75,76/ Force
generation was associated with the disorder-to-order
transition. The postpower stroke state, with the hydrolysis
product adenosine diphosphate (ADP) in the active site,
showed a local domain heterogeneity, but overall the
catalytic domain was well oriented, q š 8° (Figure 12).
Release of ADP (rigor state) resulted in a slight change
in the twist and the tilt angle of the heads..77/ During
isometric contraction, when most of the myosin heads
should be in the prepower stroke state (immediately prior
to the rate-limiting step of the cycle), no species were
observed at a different angle to that of the postpower
stroke heads..102/ These findings excluded a simple model
in which the catalytic domain (accounting for most of the
myosin head mass) generates a force while rotating by
45° from one well-defined angle to another.
A different story emerged when the labels were placed
in the regulatory domain of the myosin head. Two
distinct populations, centered 36° apart, were observed
in contraction (predominantly prepower stroke heads),
whereas in rigor, only one population was observed..57/
Clearly, the rotation of the head is limited to the
regulatory domain with the catalytic domain shifting from
a disordered to ordered structure.
The disorder-to-order transition was also found for
other muscle proteins. For example, spin-labeled TnC is
well ordered prior to Ca2C activation but becomes dis-
ordered in the presence of Ca2C or activating myosin
heads..58/ The loss of stereospecific, protein – protein
interactions is reflected by changes to the conformational
homogeneity and is the basis of many molecular mech-
anisms. ESR, with its capacity to see directly both the
ordered and the disordered populations, complements
more popular methods such as X-ray crystallography or
electron microscopy image reconstructions which ignore
disorder.
5.2 Protein Dynamics
5.2.1 Sensitivity of Conventional Electron Spin
Resonance
Conventional ESR is used to study molecular dynamics
on the nanosecond timescale. This timescale corresponds
to motions of peptides and small proteins, or the mobility
of labels on the surface of large proteins. Sensitivity of
the ESR signal to motion arises from rotational modu-
lation of the magnetic tensor anisotropy. The anisotropy
14
1/τex >> ∆ωAB
1/τex << ∆ωAB
ωA ωav ωB
∆ωAB
Figure 13 Two-site exchange between environments giving
rise to two resonances separated by w. Increased rate of
exchange initially broadens the resonances then averages their
resonant frequencies.
of Zeeman or hyperfine interactions results in different
resonant fields for different spin orientations. Therefore,
rotational motions which change the spin label orientation
will modulate the ESR line shape. This can be explained
by using a two-site exchange example: two spins, A and
B, resonate with frequencies wA and wB and exchange
their positions at a rate 1/tex . When the exchange rate is
significantly slower than the difference in resonant fre-
quencies (1/tex − w0AB D wA wB , slow exchange), the
spectrum consists of resonances A and B centered at wA
and wB , respectively, and the line widths are determined
by T2 (Figure 13). The effective relaxation rate 1/T2eff is the
sum of the intrinsic 1/T2 rate and the exchange rate 1/tex .
When the exchange frequency increases, the line widths
start to broaden. Further increase of the exchange rate
causes partial averaging of the resonance positions. The
observed difference in resonant frequencies is reduced
according to Equation (23):
s and coefficients B and C assure differential broadening
8 of lines belonging to different nuclear manifolds. These
wAB D w0AB 1 .23/
t2ex w0AB
For exchange rates faster than the difference in
resonant frequencies (1/tex × wAB , fast exchange), the
two resonant peaks coalesce into one peak at the average
frequency (assuming equal populations of A and B)
(Figure 13). The line width is determined by the effective
relaxation time, T2eff , with contributions from T2 of species
A and B and the exchange broadening (Equation 24):
1 1 1 tex
eff
D C C w2AB .24/
T2 T2A T2B 8 width of the central line.
PEPTIDES AND PROTEINS
In other words, at the fast exchange limit (1/tex ×
wAB ), the contribution of the exchange rate to line
width disappears and the spectrum consists of a single
peak at 1/2.wA C wB / with an average line width of
1/T2A C 1/T2B (Figure 13).
The above considerations generally hold true for any
spins exchanging between different environments such
as different local magnetic fields, dipolar interactions
or association with different macromolecular assemblies.
Spin label reorientation with respect to the magnetic
field is also a form of exchange where the rotational
correlation time tr is the exchange rate and the anisotropy
of the g- and hyperfine interactions defines the frequency
difference. At X-band, w D .Azz Axx //h̄ D 500 MHz
or (gxx gzz /bH/h̄ D 185 MHz.
As for the two-site exchange discussed above, various
motional regimes of ESR can be defined: fast (tr ³
10 11 – 10 9 s), slow (tr ³ 10 9 – 2 ð 10 7 s) and very slow
(tr > 2 ð 10 7 s). Fast and slow motion are of the order
of T2 (15 – 30 ns) for electron spin and have visible
effects on conventional ESR spectra which measure
the transverse component of magnetization. The very
slow motions do not affect transverse magnetization, but
they do affect longitudinal magnetization which decays
with T1 (1 – 15 µs). These slow motions can be detected
(indirectly) using saturation transfer, pulsed ELDOR or
saturation recovery ESR.
5.2.1.1 Fast Motion (tr ³ 10 11 – 10 9 s) In the fast
motional regime, the motion completely averages the
anisotropy of the g- and hyperfine tensors. The rotational
rate is obtained from the line width broadening using
Redfield’s perturbation theory..59/ The broadening itself
is a function of the nuclear quantum spin number
as different nuclear manifolds have varying anisotropy
values (Equation 25):
H.mI / D A C BmI C Cm2I .25/
The coefficient A is equal to homogeneous broadening
coefficients are obtained from the line widths of the
Lorentzian lines according to Equations (26) and (27):
p (s s )
3 V.0/ V.0/
BD H.0/ .26/
4 V.C1/ V. 1/
p (s s )
3 V.0/ V.0/
CD H.0/ C 2 .27/
4 V.C1/ V. 1/
where V.mI / is the peak-to-peak height of a given nuclear
manifold resonance and H.0/ is the peak-to-peak line
ESR LABELING IN PEPTIDE AND PROTEIN ANALYSIS 15

Factors B and C are equal for isotropic motion and tiso

r
local director axis. For an isotropic diffusion within the
is calculated directly from Equations (28) and (29): cone angle qc , components of an ordering tensor are given
by Equations (36) and (37):
tiso
B D 1.22 ð 10 9 B .28/
tiso D 1.19 ð 10 C 9
.29/ Szz D 12 .cos2 qc C cos qc / .36/
C
1
Sxx D Syy D S
2 zz
.37/
where B and C are expressed in gauss and t in seconds..61/
In the case of anisotropic motion, B 6D C, and the rates
Szz and Sxx are used to define the motionally averaged
of rotation about the nitroxide z-, x- and y-axes are
magnetic g-tensors gjj and g? in terms of its average value
different. The ratio of C and B can be used to define
g0 and the anisotropy g and dg (Equations 38 and 39):
the anisotropy as the coefficients are independent of the
rate of motion. Various models of anisotropic motion are 2 2
considered in excellent reviews by Marsh.61/ and Beth and gjj D g0 C gSzz C dg.Sxx Syy / .38/
3 3
Robinson..67/ If the molecule is diffusing in an isotropic
1 1
medium, then the rotational correlation times about the g? D g0 gSzz dg.Sxx Syy / .39/
nitroxide z-axis (tjj ) and about an axis perpendicular to 3 3
z.t? / are given by Equations (30) and (31): with equivalent expressions for the effective hyperfine
2t20 t22 splitting Ajj and A? . The latter two are resolved in
tjj D .30/ the experimental line shapes; see Figure 14. Thus Szz
3t20 t22
of the nitroxide z-axis can be easily obtained from
t? D t20 .31/ Equations (38) and (39) and the corresponding cone angle
where t20 and t22 describe spin relaxation and are from Equation (36).
related to the anisotropy of the magnetic interactions
(Equations 32 and 33): 5.2.1.2 Slow Motion (tr ³ 10 9 – 2 ð 10 7 s) For tr >
7
2 ns, Redfield’s theory does not hold. The equation of
1.11 ð 10 5.dA/B 8.dg/HC motion for an electron spin is solved using a stochastic
t20 D .32/
HA gdA dgA Liouville equation (SLE), developed by Schneider and
3.69 ð 10 8
8gHC 5AB Freed..60/ Although the description of the SLE approach
t22 D .33/ is beyond the scope of this article, one can rationalize the
HdA gdA dgA
effect of slow motion on ESR spectra in terms of the two-
where A and dA are given by hyperfine anisotropy site exchange. The low- and high-field extremes of the
(Equations 34 and 35): powder spectra correspond to nitroxides lying with the z-
1 axis parallel to the magnetic field. Rotation (i.e. exchange
A D Azz 2
.Axx C Ayy / .34/ with any other orientation) results first in an exchange
dA D 12 .Axx Ayy / .35/ broadening of the line width and then partial averaging
of the anisotropy. Line width and effective splitting are
with equivalent equations for g-anisotropy. The indices
in Equations (34) and (35) are permutated to calculate
the values of tjj and t? for the rotation about the x- and A||
y-axes of the nitroxide.
The above equations hold for anisotropic motion about
a specific nitroxide axis in an isotropic medium. An
additional complication arises when diffusion takes place
within a strongly orientating potential such as in a lipid
membrane, or within the steric confines of a protein.
The field position of resonances now depends on the
amplitude of motion, which defines the time average of
available angular space; e.g. if the nitroxide can move
only within an angular cone, then only the resonances
corresponding to the orientations within the cone are A⊥
averaged. Motionally averaged spectra are described in
terms of order parameters (S) – time averages of the Figure 14 Definition of the parallel and perpendicular hyper-
direction cosines of the diffusion axis with respect to the fine splitting for calculation of order parameters.
16 PEPTIDES AND PROTEINS

used to determine tr (Equations 40 and 41): strongly on the chosen rigid limit values. A user-friendly
 b0m simulation and optimization program based on the SLE
Hm was developed by Budil et al..50/ Sensitivities of the
tr D a0m R
.40/
Hm 1 conventional ESR and STESR spectra are illustrated
  in Figure 15(a) and (b).
A0zz b
tr D a 1 .41/
ARzz Examples. Side-chain and polypeptide backbone
dynamics are determined using the above formalism.
where Hm and A0zz
are the line widths at half-
Spin labels attached to the surface of small a-helical
height and hyperfine splitting, respectively, and the
superscripts denote their rigid limit values. Coefficients peptides exhibit subnanosecond motions observed by
a0m , b0m , a and b are calculated from SLE simulations. ESR which compare well with motions predicted by
Their precise values depend on the motional model molecular dynamics simulation programs..45,62/ Scanning
used for the simulations. For a Lorentzian line width of the label position along a peptide length reveals a
d D 3.0 G and isotropic Brownian diffusion, a0mD1 D V-shaped gradient of the label mobility. The cone angle
11.5 ns, b0mD1 D 0.943, a0mD 1 D 21.2 ns, b0mD1 D 0.778, for random motion in the middle of the peptide was
a D 0.54 ns and b D 1.36. Values for different line half the value found at either terminus. Interestingly,
widths or motional models can be found in Marsh..61/ the C-terminus was found to be more flexible than the
It should be noted that the calculated tr values depend N-terminus, which explains the decreased stability of
the C-terminus as compared with the N-terminus in a-
helices..45,46/ Backbone dynamics observed in isolated
peptides are further modulated by tertiary interactions.
τr = 20 ns
A survey of 30 cysteine mutants of T4 lysozyme with
spin labels at various structural sites (on the surface of
helices, within the helix termini, interhelical loops, buried
sites and sites involved in tertiary contacts) revealed a
τr = 10 ns
characteristic pattern of spin label mobility in relation
to the secondary structure of the protein..31/ When the
second moment of the spectrum (defined as the reciprocal

,,,,,,,, ,,,,,,,,
,,,
τr = 0.001ns 5.5
, ,
,,, Loops
,
(a)
,
5.0

,,,,,
〈H 2〉−1 × 1000 (G−2)

@@
€€
ÀÀ
,,
,

,,,,,
@@@
€€€
ÀÀÀ
,,,
@@ ,,,
€€
ÀÀ
,,
,

@@@
€€€
ÀÀÀ
,,, @@@
€€€
ÀÀÀ
τr = 200 µs
,,,,,
@@@
€€€
ÀÀÀ
,,, @@@
€€€
ÀÀÀ
,,,
@@@
€€€
ÀÀÀ
,,,
,,,,,
@@@
€€€
ÀÀÀ
,,,
4.5
@@@
€€€
ÀÀÀ
,,, @@
€€
ÀÀ
,, Helix surface
,,,,,
,,,,,,,,,

Buried
,,,,,
@@@
€€€
ÀÀÀ
,,,
@@
€€
ÀÀ
,,
@@@
€€€
ÀÀÀ
,,,
Tertiary contact
,,,,
,,,,

4.0
,

τr = 23µs

0.1 0.2 0.3 0.4 0.5

∆H0−1 (G−1)

τr = 4µs Figure 16 Reciprocal of the square of the splitting ver-

sus reciprocal of the central resonance line width. The
spectral parameters cluster according to the labeled pro-
(b) tein structural elements..31/ [Reprinted with permission from
H.S. McHaourab, M.A. Lietzow, K. Hideg, W.L. Hubbell, Bio-
Figure 15 Sensitivities of (a) the conventional ESR spectra chemistry, 35, 7692 – 7704 (1996). Copyright 1996 American
and (b) STESR spectra. Chemical Society.]
ESR LABELING IN PEPTIDE AND PROTEIN ANALYSIS 17

of maximum splitting squared) is plotted against the
1
reciprocal of the central field line width (HmD0 ), sites in
similar environments are clustered together (Figure 16).
The clustering reflects the degree of motional restrictions,
with the second moment related to the averaging of the
hyperfine anisotropy and the central line width to the
averaging of the g-tensor. When motional restrictions
increase, the averaging decreases and both the second
moment and the line width of the resonances increase.
Hence the second moment and line width can be used as
semiempirical diagnostic tools to evaluate the secondary
and tertiary structure of a labeled site.
5.2.1.3 Very Slow Motion (tr > 10 9 s) When tr >
100 ns, conventional ESR line shapes are no longer
sensitive to motion. The rate of angular exchange is
too small to affect the hyperfine or g-anisotropy and the
line shapes become insensitive to very slow motions. To
study these biologically important motions, a related ESR
technique was developed, STESR.
Isotropic Motion. In the presence of power satura-
tion, the second harmonic out-of-phase (V20 ) line shape
resembles an absorption spectrum (Figure 6c), with the
intensity reflecting the effective relaxation at that point.
Since effective relaxation is related to spectral diffu-
sion and spectral diffusion is a function of the rotational
correlation time, the V20 line shape reflects rotational
mobility. The rate of spectral diffusion (tsd ) is a function
of the resonant field Hres . Some field positions are more
sensitive to angular rotation than others and @Hres /@q
varies across the spectral line shape. For instance, the
rate of spectral diffusion is zero at the turning point
H Ł D Hres .q D 0° ), but increases in the intermediate fields
height at precise positions in the spectrum: L00 , C0 or H 00
at q D 35° (two-thirds of the way between resonant field
corresponding to q D 90° and q D 0° ) and normalize it
to the intensity at H Ł (L, C and H positions) for which
spectral diffusion is zero.
By substituting Equation (44) for the effective relax-
ation rate in Equation (43) a semiempirical expression for
the P0 /P ratio dependence on tr is obtained (Equation 45):
P0 I0 .Hres / 1 C a/tr
D .45/
P.Hres / I0 .H Ł / 1 C b/tr
The parameters a, b and I0 .Hres )/I.H Ł ) can be esti-
mated from Equations (42) and (44) by numerically
evaluating sensitivity @Hres /@q at each spectral position.
In practice, these values are obtained from fits to the
experimental curves of line-height ratios from spectra of
molecules undergoing Brownian diffusion with a known
tr . Spin-labeled hemoglobin or bovine serum albumin
tumbling in media of a known viscosity (water – glycerol
mixtures) is used for this purpose..63/ The rotational
rate of hemoglobin (the abscissa in Figure 17) is calcu-
lated from the Stokes – Einstein equation for a sphere
of radius r, tumbling in a medium with viscosity h
(Equation 46):
4phr3
tr D .46/
3kT
2.50
H ′′/H
2.00
L′′/L
1.50
P ′/P

(Equation 42):
  
1 8 @Hres 2 2 1 1.00
C ′/C
tsd .Hres / D T 2 tr .42/
3p2 @q 0.50
To the first approximation,.63/ the change of the signal
0.00
intensity (I) at any field position is proportional to the 1 10 100 1000 10 000
change of the spin – lattice relaxation time due to spectral
diffusion (Equation 43):
Correlation time (µs)

T1eff .Hres / Figure 17 Dependence of V20 diagnostic ratios on the rotational

I.Hres / D I0 .Hres / .43/ correlation time. The curves are simulated with Equation (63)
T1 using the parameter values from Table 1.
where I0 is the rigid limit intensity in absence of motion
and T1eff is the intrinsic T1 modified by spectral diffusion Table 1 STESR parameters from maleimide spin
according to Equation (44): label – hemoglobin calibration curves.63/

1 C .I0 .Hres //T2 /T10 tsd .Hres / 1

Parameter I0 .Hres //I.H Ł ) a (µs) b (µs)
T1eff .Hres / D T10 .44/
1 C T10 tsd .Hres / 1 L00 /L 1.88 6.18 67.9
C0 /C 1.01 0 21.1
Since T1eff
is a function of the field position (spin angle H 00 /H 2.17 21.7 210
with respect to field), it is customary to define the line
18 PEPTIDES AND PROTEINS

Table 2 Useful constants where d? and djj are as given by Beth and Robinson
(Equations 51 and 52):.67/
Constant Symbol Value Units
   2
b b
Planck’s constant h 6.63 ð 10 34 Js djj D 0.688 0.202 .51/
h̄ 1.06 ð 10 34 Js a a
Bohr magneton b 9.27 ð 10 24 JT 1  
b
Free electron g factor g 2.00232 d? D 0.661 C 0.891 .52/
Electron magnetic moment µ 9.29 ð 10 21 JT 1 a
Magnetogyric ratio g 1.76 ð 10C11 s 1T 1
The anisotropic diffusion tensor (D) creates an addi-
Boltzmann constant k 1.38 ð 10 23 JK 1
tional complication. The effect of the molecular rotation
on the spectral line shape is a function of the label orien-
tation with respect to the diffusion axis. If the principal
The parameter values listed in Table 1 were obtained axis of diffusion is parallel to the z-axis of the spin label,
for maleimide spin-labeled hemoglobin tumbling in the motion interconverts the x- and y-components only.
different water – glycerol mixtures. In principle, these If it is parallel to the x-axis, then the y- and z-components
values should be transferable from one laboratory to will be mixed. To describe fully anisotropic diffusion of
another. In practice, each cavity is sufficiently different the anisotropic tensor, six parameters are needed: three
in its microwave distribution and modulation fields diffusion coefficients about the x-, y- and z- axes and three
that separate calibrations are often constructed. New Eulerian angles describing the orientation of the diffusion
calibrations are also necessary for different spin labels. tensor with respect to the magnetic tensor.
Changes to the magnetic tensors and relaxation times The problem is simplified if either the diffusion tensor
alter STESR line shapes. In rare cases, full numerical (D) and/or the magnetic tensors (g or A) are axially
simulation of the V20 line shape is used to determine the symmetric: the elements of the diffusion tensor are
correlation time, but the computational time required is related to the correlation times by t? D 1/.6D? / with
still prohibitive..68/ a corresponding expression for tjj . It has been shown
that the effective correlation time obtained from the
Anisotropic Motion. The effective rotational corre- L00 /L and H 00 /H line-height ratios of STESR spectra
lation times (teff ) obtained from such calibration curves (mI D š1) can be described in terms of D? , Djj and
reflect rates for isotropic rotation. However, isotropic the angle q between the diffusion and magnetic tensor
motion is not very common in biological systems. axis (Equation 53):.68/
For example, the nonspherical shape of the diffus-
1
ing molecules or the restoring potential of the media teff
R .š1/ D 2
.53/
results in anisotropic motion. Intuitively, rotation about 3[Djj sin q C D? .1 C cos2 q/]
the long axis of a cylinder is faster than the tumbling When q D 0° the outer manifolds reflect D? which
motion around its short axis. Assigning an isotropic tr to defines the z- and x-(y-) element conversion (Djj leaves
an anisotropic motion is obviously in error. For elon- the nitroxide z-direction unchanged). If q D 90° the
gated molecules correlation times for rotation about intensity of L00 and H 00 is determined by Djj , which now
the major and minor axes are given by Equations (47) interconverts the z- and x-, y-axes.
and (48): In some cases, anisotropic rotation is about a single
fjj axis and the motion can be described by a uniaxial model.
tjj D .47/ The mobility of transmembrane peptides or proteins in
4kT.1 C fjj /2f? /
lipid membranes is a good example. A uniaxial model,
f? with a single diffusion tensor element Djj and an angle
t? D .48/
6kT q defining the relative orientation of the magnetic and
diffusion axes, is sufficient to simulate STESR spectra
where T is absolute temperature and the frictional
of membrane-bound proteins..69/ If q is not known, then
coefficients fjj and f? are a function of the shape of
teff .š1/ gives an upper estimate of 0.5tjj (q D 90° ). It
the molecule..64,65/ For a cylinder of length 2a and radius is important to realize that the changes of the q angle,
b the frictional coefficients are given by Equations (49) brought about by conformational changes, might result
and (50):.66/ in STESR line-shape changes which can be mistakenly
interpreted as changes in protein dynamics. A quick
fjj D 8phab2 [0.96.1 C djj /] .49/
diagnostic for the presence of anisotropic motion is the
8pha3 comparison of the effective correlation times estimated
f? D .50/
3[ln.a/b/ C d? ] from the C0 /C ratio and from L00 /L (H 00 /H). If they agree,
ESR LABELING IN PEPTIDE AND PROTEIN ANALYSIS
then the motion is likely to be isotropic. If they are
different, then either the overlap of the nuclear manifolds
is different in hemogiobin calibration spectra (unlikely) or
the motion is anisotropic..70/ If the change in the STESR
spectra is to be interpreted in terms of changed motional
rate and not changed anisotropy of motion, then at least
the ratio of the correlation times teff .š1//teff .0/ should
stay constant.
Another motional model commonly encountered in
biology is restricted diffusion. In such a model, the motion
is isotropic but constrained in amplitude. The smaller the
amplitude of motion, the slower is the apparent mobility
derived from isotropic calibration. When the amplitude
is <30° , the effective correlation time can be a factor
of 10 larger than the actual tr and for amplitudes >60° ,
teff approaches tr ..71/ For small amplitudes, there is no
isotropic line shape which will match the STESR spectrum
of restricted motion. The amplitude effect is not just
reflected in one or two places in the spectrum, but rather
it is distributed across the whole line width.
An extensive review by Beth and Robinson.67/ deals
with the effects of anisotropic motion on STESR spectra
and the theoretical simulations of line shapes. Numerical
simulations are based on the transition rate matrix which
couples neighboring angular zones with the rate of angular
reorientation. The SLE approach and spin density matrix
method have also been used. Both approaches have been
applied successfully to isotropic and anisotropic motional
models. The continuous increase in computational speed
bodes well for the routine application of STESR simula-
tions to analyze experimental data.
5.2.1.4 Examples in Muscle Proteins Microsecond
motions are common for large macromolecular com-
plexes (1 MDa) such as are present in muscle. The
timescales of force generation, the actomyosin ATPase
cycle and muscle activation coincide with the micro-to
millisecond timescale of STESR, thereby making it the
method of choice. The first application of the method
established the dynamics of myosin, its subfragments and
actin..72/ Thomas et al. showed that the myosin head is
capable of moving independently of the large myosin
filament. Such motion was a prerequisite for force pro-
duction. When bound to actin, in the rigor state (no ATP)
the myosin heads were immobilized but when ATP was
added the heads detached and were free to move..73,74/
Subsequent studies in muscle fibers at various inter-
mediate states of the acto-myosin ATPase cycle have
established a progressive decrease of catalytic domain
mobility during the contractile cycle: the 10-µs motion
of relaxed and weakly attached heads.75/ became 80 µs
just before force was generated.76/ and was completely
‘‘frozen out’’ in the postpower stroke states of ADP and
rigor. In the ADP state the head, although globally rigid,
19
retained ‘‘breathing motions’’, which were suppressed
on the release of nucleotide..77/ It is believed that this
gradient in protein mobility reflects tighter and more
stereospecific binding as myosin progresses through the
contractile cycle (Figure 12).
The dynamics of the myosin head are complicated by
the fact that this elongated protein does not behave like a
rigid body. A comparison of the dynamics of the catalytic
and regulatory domains revealed a three-fold difference in
the rate of motion for the two domains..78/ Moreover, the
two domains were found to have dramatically different
orientational distributions..57/ These results highlight
the complexity of the conformational changes in the
actomyosin system: force generation is not synonymous
with force transmission and both events involve changes
of dynamics and orientation.
This complex behavior of myosin is in contrast to that
of actin. Neither the orientation nor the dynamics of actin
monomers, as probed by labels attached near the myosin
binding site, were affected by head attachment..79 – 82/
The absence of any orientational changes in contracting
muscle fibers was also observed using spin-labeled toxin
phalloidin bound rigidly to the interface between the
actin monomers..83/ This agrees with the current model
of actin’s passive role in force production in providing
‘‘tracks’’ for myosin motor protein to ‘‘walk’’ on.
Force activation involves a complex pathway with
subtle changes in protein – protein interactions. It is
mediated by the conformations and dynamics of the
participating molecules. Smooth muscle is activated via
phosphorylation of myosin light chain 2, whereas skeletal
muscle is regulated by a thin filament based system
involving Ca2C binding to TnC. STESR spectra of
phosphorylated myosin with a probe bound to myosin
light chain 1 have implied increased motional freedom of
the head. This finding supports a model.86/ in which
unphosphorylated heads are tied to the surface of
myosin filaments and inhibited from binding to actin..87/
Phosphorylation abolishes the electrostatic attraction to
the filament surface allowing the heads to interact with
actin.
In skeletal muscle, binding of Ca2C to TnC initiates a
signaling pathway from the thin to thick filament which
ultimately activates muscle contraction. Biochemical
changes in the affinity of myosin for actin and of TnC
for Ca2C have a structural basis that is readily observed
by both STESR and conventional ESR. The mobility
and orientation of TnC (and TnI) has been found to
be similarly affected by the binding of myosin heads to
actin or by Ca2C binding to TnC..84/ Interestingly, TnC
was capable of sensing not only the binding of the myosin
heads to actin but also the intermediate ATPase states..85/
5.2.1.5 Examples in Membranes Rotational diffusion
of membrane-bound proteins is often the best way of
20
determining the oligomerization state in their native
environment without the possible dissociative effect
of detergents. STESR of various integral proteins has
revealed monomers such as rhodopsin,.88/ dimers such
as cytochrome oxidase.89/ and ADP– ATP carrier.90/ or
even higher oligomers such as Na, K-ATPase..91/
The biological significance of dynamic structural
changes is best illustrated by the Ca-ATPase,.92/ whose
molecular dynamics correlate with transport activity..93,94/
As shown by ST-EPR, allosteric interactions between
Ca-ATPase polypeptide chains and catalytically impor-
tant domain interactions involved in the transport
cycle are regulated by both alterations in membrane
lipid composition, anesthetics, and the regulatory pro-
tein phospholamban..95 – 97/ Thus, physiological regulators
of calcium transport modulate catalytically important
motions and provide a structural basis for b-adrenergic
stimulation in the heart.
Protein dynamics measured by STESR and conven-
tional ESR have differentiated between two models
of steroid biosynthesis in mitochondria: the shuttle
mechanism and the ternary complex of adrenodoxin,
P450 and adrenodoxin reductase. Adrenoxin was found
to form binary complexes (but not ternary complexes)
with either P450 or adrenodoxin reductase, supporting a
shuttle mechanism..98/
An excellent example of the potential of STESR in
describing complex anisotropic motions is in the study
of the transmembrane anion transporter Band 3 by
Hustedt and Beth..69/ The STESR spectra were simulated
using a uniaxial model for protein rotation. The diffusion
rate and the angle between the magnetic and diffusion
tensor were freely floated in the least-squares fits to
experimental spectra. The uniqueness of the solution was
corroborated by the orientational study of Band 3 in
oriented erythrocytes..55/
5.2.2 Mobility and Time Domain Methods
The measurement of the molecular dynamics by time-
resolved ESR methods is still in its infancy. Specialized
hardware is necessary to perform such experiments.
Spectral diffusion due to the reorientation of spins
can be observed either by recovery from saturation
at the resonant field (saturation recovery ESR) or by
arrival of saturation originally induced at some other
nonresonant field (pulsed ELDOR). The initial promise
of these methods was not fulfilled when it was shown that
the nuclear relaxation, which couples different nuclear
manifolds, contributed significantly to spectral diffusion.
Combining pulsed ELDOR and saturation recovery
differentiates between nuclear relaxation and rotational
spectral diffusion and can be used to measure the true
rotational correlation time..99/
PEPTIDES AND PROTEINS
2-D FTESR methods appear to be more promising.
Nuclear relaxation is seen as cross peaks between the
manifolds and can easily be distinguished from homoge-
neous broadening and spectral diffusion broadening..100/
In the limits of fast motion, tr is obtained directly from
the homogeneous line width and is defined by the pure
T2 (similar information is obtained from the spin-echo
experiments). For slower motions, mixing time between
the pulses is varied (2-D ELDOR) and the dependence
of spectral broadening on mixing time is used to deter-
mine tr . Correlation times in the range 1 – 30 µs have
been measured for small peptides tumbling in viscous
media..101/
5.3 Kinetic Experiments
Elucidation of molecular mechanisms involves primarily
two approaches: (a) entrapment of reaction intermediates
with a subsequent reconstruction of the sequence of
events and (b) transient kinetics in which the reactions
are synchronized with the observed spectral changes.
Each of these approaches have potential problems. In
the ‘‘trapping’’ approach, the states have to be related
to the kinetic intermediates. There are cases in which
states trapped with substrate or product analogs are not
lying on the kinetic pathway. On the other hand, transient
experiments are easier to interpret, but technically more
challenging owing to lower signal levels, fast acquisition
times and difficulties in spectral assignment. The two
approaches should be considered complementary. In
an ideal world, ‘‘trapping’’ approaches should be used
to identify and assign signals collected during transient
experiments.
Historically, optical spectroscopy was used for tran-
sient kinetics owing to inherently higher sensitivity, but
ESR is making substantial inroads..102/ Recent advances
in resonator design allow for millisecond resolution on
microliter samples in the submillimolar concentration
range..103/ The DR developed for this purpose is capa-
ble of measuring millisecond kinetics in a single shot on
100 µL of a 40 µM sample with an 8-ms deadtime..104/
The further development of this DR/stop-flow config-
uration allows the recording of a full spectrum within
100 ms..105/
Transient ESR was used to resolve the stages of
channel formation in lipid membranes. Phospholipid
vesicles and membrane channel collicin were mixed
rapidly and the time course of the protein absorption
to the membrane surface was clearly resolved from the
insertion of the channel into the membrane..106/ For
collicin the process was fairly slow, with a timescale
of seconds, but the formation of another channel
annexin was followed on the millisecond timescale..107/
The millisecond time resolution makes ESR a viable
ESR LABELING IN PEPTIDE AND PROTEIN ANALYSIS
alternative to optical methods in investigations of kinetic
processes.
The photolysis of caged compounds, cATP and
cCa2C , to study conformational transients has been
used primarily in the muscle field and in the study of
Ca2C -ATPase. A single ultraviolet (UV) pulse (10 ns in
duration with an energy flux of 150 mJ cm 2 at 351 nm)
from an excimer laser is capable of liberating 0.5 mM
ATP (Figure 18a and b). The magnetic field is locked
into a position where the initial and final states dis-
play a large spectral difference and the intensity at that
position is followed in time. In myosin, the pre- and
post-ATP hydrolysis states have different mobilities at
a labeled residue near the catalytic site, with correla-
tion times of tr > 100 ns and tr ³ 80 ns. This mobility
difference and the associated line-shape difference was
utilized in measuring the rate of transition between the
two states (43 s 1 ) and was found to correspond to the
hydrolysis rate of ATP.108/ (Figure 18a and b). Simi-
lar experiments in muscle fibers, measuring both the
orientation and the mobility, established that the rapid
disorder of myosin heads follows nucleotide binding but
precedes hydrolysis. These experiments also determined
that the rate of hydrolysis is the same in fibers as in
solution..102/
Transient ESR of the Ca2C -ATPase following cATP
photolysis revealed local domain changes around the
labeled site which correlate well with the formation of
the phosphoenzyme intermediate..109/ A larger and more
+ ATP
No nucleotide
1s
hν
(a)
hν 50 ms
(b)
Figure 18 (a) Transient ESR of myosin head labeled with
iodoacetamide spin label after the photolysis of cATP
and (b) expanded transient..108/ (Reprinted with permission
from E.M. Ostap, H.D. White, D.D. Thomas, Biochemistry, 32,
6712 – 6720 (1993). Copyright 1993 American Chemical Society.)
21
motionally restricted label was also used to observe global
changes, e.g. shape or oligomerization state. No such
changes were observed by transient STESR during the
ATPase cycle..110/
Biological photocycles encountered in rhodopsin and
bacteriorhodopsin are special cases of cycles that are
easily synchronized. The photoisomerization of retinal
in bacteriorhodopsin initiates a series of proton transfer
reactions via short-lived intermediates culminating in the
loss of H C at the extracellular surface. Some 50 µs after
photoactivation, an intermediate M decays to N when
Asp96 transfers a proton to the Schiff base. During the
decay of the N state, Asp96 regains a proton from the
cytoplasmic site and bacteriorhodopsin reverts to the
ground state, thereby completing the cycle. Labels have
been attached to a number of cytoplasmic, interhelical
and extracellular loops in the vicinity of Asp96 and their
mobility was followed after irradiation with light..111 – 113/
Cytoplasmic sites and those near Asp96 all showed
significant changes which coincided with the decay of
the M state and recovered with the decay of the N
state..112/ The efforts of Hubbell’s and Steinhoff’s groups
to describe the molecular mechanism of rhodopsin and
bacteriorhodopsin are an excellent example of the power
of ESR methods to evaluate both static and transient
molecular structures. In summary, the combination of
high sensitivity, short mixing deadtimes, and temporal
resolution makes ESR an increasingly popular method to
study transient kinetics.
5.4 Protein Folding
In the last few years, site-specific spin labeling has
been applied to protein folding problems..114,115/ The
advantage of the ESR approach to protein folding lies
in site specificity as the denaturation of local domains
can be followed independently of global denaturation.
This approach relies on differences in the mobility of
spin labels in folded and denatured proteins. The folded
protein provides steric restrictions due to secondary
structure and tertiary contacts whereas the denatured
one does not. The ESR spectra for the denatured fraction
are a composite of sharp, motionally averaged line shapes
in contrast to broader, immobilized spectra observed for
the folded protein. Fractions of protein in each form
are easily calculated by spectral subtractions and by line
shape integrations..115/ Cooperativity and stability of the
given region are determined from spectral titration with a
denaturing agent, e.g. GdnCl, urea or heat. Differences in
the melting of hydrophobic and aqueous surfaces of the b-
strand pore of FepA receptor were observed by cysteine
scanning of the polypeptide chain lining the channel.
The hydrophilic surface was more stable and cooperative
in the transmembrane portion of the strand than the
22 PEPTIDES AND PROTEINS

extramembraneous strand ends. The residues exposed to 5.6 Distance Measurements

the lipid exhibited noncooperative melting and did not
ESR is capable of measuring short distances (2 – 25 Å)
denature completely even at the highest concentrations
between selected sites. The method relies on the distance-
of denaturants..116/
dependent interactions between two spin labels (spin
ESR is also capable of sensing multiphasic folding
label – spin label method), or an interaction between a spin
intermediates. Carbonic anhydrase was found to denature
label and a paramagnetic metal (spin label – spin probe
via an intermediate characterized by a compact and stable
method). The physical basis for the coupling between
molecular core with a more dynamic periphery..117/ In the
the two spins is the exchange interaction (J) arising from
presence of the chaperonin GroEL, the intermediate core
the overlap of the orbitals of unpaired electrons and
was destabilized and partially melted, explaining how
the dipolar interaction between magnetic moments of
GroEL allows for the refolding of misfolded proteins..118/
the two spin labels. The spin label – spin probe method
The development of new resonators, as mentioned
relies on the enhancement of the relaxation of the spin
previously, has facilitated detailed analysis of folding
labels by paramagnetic metals which have considerably
kinetics. The initial phases (<20 ms), ascribed to helix
faster relaxation rates and provide an efficient relaxation
formation, were recently resolved by stop-flow ESR..104/
pathway for nitroxides.

5.5 Ligand Binding 5.6.1 Spin Label – Spin Label Method

Two less well known applications of spin labeling
are the determination of the binding of small ligands
and the aggregation of large (the latter reviewed in
section 5.7.4). The binding of small ligands is followed
by changes in their mobility. Spin label analogs of ligands
have sharp, motionally narrowed spectra when free in
solution. Binding to a larger target slows the motion of
the label and the spectra become broadened. Spectral
resolution between the broad/bound species and the
narrow/free ligands allows quantification of bound species
and hence calculation of binding isotherms. The binding
of mellitin to a-crystallin.119/ and the binding of nucleotide
analogs to ATPases were determined by this method.
Binding studies are not limited to ligands carrying a
spin label. Competition of unlabeled and labeled analogs
is used to determine the Kd of unlabeled ligands.
The binding of ATP, ADP, adenosine thiotriphosphate
(ATPgS), adenosine imidotriphosphate (AMPPNP) and
adenosine methylenetriphosphate (AMPPCP) to myosin
was determined by the displacement of spin-labeled
ATP..120/
A good example of ESR applications in ligand binding
is the association of lipid spin labels with intrinsic
membrane proteins. The differences in mobility are
small and the spectra of bound and free labels are not
resolved. Small broadening due to exchange between free
and restricted environment was simulated to extract the
equilibrium constant and the number of binding sites..61/
Finally, the binding of metals to proteins can be
established by ESR. Mn(II) has a characteristic six-line
spectrum when free in solution, but no signal when bound
to protein. The decrease of free Mn(II) signal upon
addition of a protein identifies the fraction of bound
metal..121/ As shown in the example with ATP analogs,
displacement of bound Mn(II) by other metals can be
used to determine their binding affinity.
5.6.1.1 Exchange Interaction, Distances <8 Å At
distances shorter than 8 Å, the s- or p-orbitals of the
neighboring unpaired electrons can overlap, creating
a single spin state: singlet (spins are antiparallel) or
triplet (spins are parallel) state. Unlike nuclear spin,
electron spin coupling propagates both through space
and through covalent bonds. Through-bond coupling
drops off very quickly and is insignificant at a distance
greater than a few bonds. The through-space coupling
strength (J) is a function of the interspin distance rdd
(Equation 54). J diminishes exponentially, with a 1-Å
interaction distance from the initial value of J0 D 300 G.
In the strong exchange regime, when J is much larger
than hyperfine interactions, the combined nuclear spin
is 2 and the spectrum displays a five-line pattern. For
weak coupling (J < A0 ), the interaction asymmetrically
broadens the low- and high-field resonances. The shift
of the downfield edge of the low-field resonance in the
presence of the second label (Hdd ) is proportional to J
and hence to the interspin distance.
Hdd ³ J D J0 exp. brdd / .54/
For example, a 0.1-Gauss broadening implies an interspin
distance of ¾8 Å, defining the upper limit of the
sensitivity.
Spin exchange in peptides labeled with two probes has
been used to determine helical folds..122/ Some proteins
contain 310 -helices between b-strands or at the end of
a-helices and their detection in solution is difficult. Fiori
and Millhauser utilized a difference of distance between
the i and i C 4 residues along the a- and 310 -helices to
distinguish between the two helical forms. The i to i C 3
distances are similar in both helices (6 – 8 Å), but the i
to i C 4 distance for the 310 -helix is outside the range
of the exchange interaction (10 – 16 Å) and within the
ESR LABELING IN PEPTIDE AND PROTEIN ANALYSIS
range for an a-helix (7 – 11 Å). Placing pairs of labels
in the i, i C 3 and i C 4 positions along the synthetic
helices identified the predominantly a-helical regions in
the N-terminus (equal broadening of i, i C 3 and i, i C 4
pairs) with the 310 -helix at the C-end (no broadening
of the i, i C 4 pair). A comparison of 16- and 21-
residue peptides revealed a length-dependent equilibrium
between the a- and 310 -helices. Shorter peptides favored
the 310 -helix whereas the longer peptides favored the
a-helix.
Transition between the two helical forms might provide
a mechanical pathway for allosteric mechanisms. The
a-helix has more residues per turn and is significantly
shorter, hence the 310 -helix to a-helix transition will
mechanically pull on the neighboring domains.
5.6.1.2 Dipolar Coupling Distances: 8 – 25 Å In addi-
tion to the exchange interaction, neighboring spins
experience local fields induced by their respective mag-
netic dipoles. The local fields can add to or subtract from
the external field splitting each of the resonances. This
dipolar interaction has an rdd3 dependence which results
in discernible line-shape changes for distances up to 25 Å.
The contribution of the exchange interaction for distances
over 10 Å is negligible.
The strength of the dipolar interaction is a function
of eight parameters: the interspin distance rdd , the angle
between the static magnetic field and the interspin vector
and the six angles describing the orientation of each
of the spin labels with respect to the interspin vector.
Protein dynamics also affects these parameters on the
timescale of the experiment: the global motion of the
molecule modulates the angle with respect to the field,
intradomain motions modulate the interspin distance and
the interspin angles. The full solution of dipolar ESR
line shapes in the presence of motion poses a formidable
challenge.
The geometry of the NAD coenzyme bound to
glyceraldehyde dehydrogenase (GAPDH) is the best
example of the approaches taken to tackle this problem.
Since GAPDH is a tetramer it can bind four molecules of
coenzyme NAD. The relative orientation of the coenzyme
was determined from the dipolar splitting of a spin label
analog of NAD. Hustedt et al. used different microwave
frequencies (9, 35 and 94 GHz) coupled with sophisticated
fitting procedures to solve independently for most of
the parameters listed above..123/ The ESR structure was
consistent with the geometry derived from molecular
modeling using a crystal structure of the apo-enzyme.
Another example is the open form of a-helices in
protic solvents..124/ Alanine-rich peptides, incorporat-
ing the unnatural spin-labeled amino acid TOAC (2,2,
6,6-Tetramethylpiperidine-1-oxyl-4-amino-4-carboxylic
acid), showed a strong dipolar interaction for the i, i C 4
23
labeled peptides with less interaction between the i, i C 3
sites. Molecular modeling revealed a shorter interspin
distance and larger backbone torsional angles, consis-
tent with 3.8 – 3.9 residues per turn compared with 3.6
residues for a standard a-helix. The open configuration
allows for the formation of additional hydrogen bonds
with amide carbonyls while preserving helical hydrogen
bonds. ESR proves to be an important tool in identify-
ing subtle changes in helical folds induced by the local
environment.
Short of solving the general case as above, two simpler
cases are often encountered: (1) the static dipolar case
in which all motions are frozen out, the labels are
disordered on the surface of the protein and the protein
itself is isotropically disordered in the magnetic field, and
(2) the motionally averaged case in which the anisotropy
of the dipolar interactions is averaged (tr < 6 ns) and
the spectrum is homogeneously broadened by spin– spin
interactions.
In the first case, the nitroxides and the sample are rigid
and each of the resonance peaks are broadened by the
Pake function given in Equation (55):
3gb.3 cos2 q 1/
Hdd D š .55/
4r3
where q is the angle of the interspin vector in the magnetic
field.
Convolution of the resonances with the Pake pattern
yields a characteristic line shape with dipolar wings in
the low- and high-field regions of the spectra. The Pake
function is obtained from the spectrum by a Fourier
transform,.125/ line shape fitting.126/ or by empirical
calibrations..38/ Empirical calibrations relate the line
heights of the low- and high-field resonances to the line
height of the central peak (Equation 56):
0.77
rdd D 9.3 C .56/
.d1 /d/ C 0.36Azz 1.76
where rdd is in ångstroms and Azz is in gauss, d is the
line height of central resonance, and d1 is the difference
between low-field peak and high-field trough.
Alternatively, the Van Vleck relation between the
second moment of the central resonance M2 and rdd
is used in obtaining the distance (Equation 57):
3 g2 b2
M2 D .57/
6
20 rdd
where rdd is in ångstroms. The second moment is given
by the Gaussian part of the peak-to-peak line width
(Equation 58):
!2
G
Hpp
M2 D .58/
2
24 PEPTIDES AND PROTEINS

The Pake broadening approach was verified for

interspin distances of doubly labeled rigid synthetic
polypeptides in which distances in the range 8 – 24 Å
were predicted from the structure..125/ The method
was also verified for larger molecules by comparison x 20
with an X-ray determined structure of spin-labeled K65R1/Q69R1
insulin. The interspin distances in a crystal were in 16G
excellent agreement with those found in solution by
ESR. Distribution of distances due to protein flexibility in
solution was found to be twice as large in solution as in the (a)
crystal..126/ It is important to note that ESR is one of the
very few techniques which can estimate conformational 65/69
heterogeneity of a protein in solution.
Static dipolar broadening is rapidly gaining popu-
larity. It was used in the analysis of helix packing 10 35/109
61/68
of lactose permease in the membrane..127/ Spin labels
placed on surfaces which faced each other displayed 65/72 35/137

dipolar broadening. Analysis of the extent of broaden-
ing as the label is moved around the helix identified
the relative rotation of neighboring helices. Dipolar
broadening was also used in the elucidation of the open-
ing/closure of K-channels.128/ as described in detail in
section 5.7.4.
The static dipolar approach (case 1) fails when the
proteins are not rigid on the nanosecond timescale
of an ESR experiment. If the spin mobility is high
enough to modulate the anisotropy of dipolar interactions
(Equation 59)
    1
gbHdd 1 3pg2 b2
tr  D 3
.59/
h rdd h
the rotational modulation of the interspin vector results
in the broadening related to both the distance and the
correlation time (Equation 60):
 
3 g4 3 tr 5 2
Hdd D h̄ 6 3 C C .60/
10 gb rdd 1 C n2 t2r 1 C 4n2 t2r
Line-shape comparisons of singly and doubly labeled
samples reveal the extent of dipolar broadening from
which the rdd distance is calculated assuming (or mea-
suring) an appropriate correlation time for the molecule.
McHaourab et al..129/ tested this approach on a series
of labeled sites in T4 lysozyme (8 Å < rdd < 23 Å) and
found them to be in excellent agreement with the
structure of the protein (Figure 19a and b). Note that
the motional modulation regime is a function of both
the interspin distance and the rotational correlation
time, e.g. dipolar interactions of labels 15 Å apart
are averaged for motions with correlation times of
6 ns.
The above examples illustrate the static and motionally
averaged cases of spin– spin interactions. However, often
the precise mechanism of spin– spin interactions is not
∆ H (G)
4/60
65/76
1
22/109 61/76
0
5 10 20 30
(b) R model (Å)
Figure 19 (a) An ESR spectrum broadened by the presence
of another nitroxide and (b) the line width as a function of
the interspin distance..129/ (Reproduced with permission from
H.S. McHaourab, K.J. Oh, C.J. Fang, W.L. Hubbell, Biochem-
istry, 36, 307 – 316 (1997).)
known as there are contributions from static dipolar
interactions, from the rotational modulation of the
interspin vector and from the modulation of the interspin
distance. Although there is currently no theory fully
describing the changes in the ESR line shape, the presence
of line-width broadening is always indicative of two
labels being in the range 10 – 20 Å from each other.
Even the simplest, qualitative statement of spin– spin
distance can yield important structural information. The
presence of spin – spin interactions has helped to elucidate
changes in the topology of the cytoplasmic portion of
rhodopsin following light activation. It has explained
how rhodopsin initiates the phosphorylation cascade by
rhodopsin kinase..130/
Interested readers are directed to an excellent review
by Hustedt and Beth..47/
5.6.1.3 Very Long Distances: >25 Å The upper limit
of 25 Å for the detection of dipolar interactions is defined
by an observable broadening: ¾3 G for inhomogenously
ESR LABELING IN PEPTIDE AND PROTEIN ANALYSIS
broadened samples of rigid samples and 0.2 G for
the sharp, motionally narrowed line shapes. This limit
can be extended by time-resolved techniques, which
unlike the continuous-wave methods do not rely on line
broadening. A variety of pulsed ESR methods have been
developed for this purpose. Double electron– electron
resonance (DEER) is one method in which the spin
echo is observed as a function of time of an intervening
pulse applied at a second frequency,.131,132/ or at the
same frequency as in ‘‘2 C 1’’ resonance..133/ In both
cases, an amplitude of the spin echo is modulated
by Pake dipolar interactions which are extracted by
Fourier transformation of the echo modulation. In
model compounds, the distances recovered by DEER
ranged between 20 and 33 Å,.132/ whereas the ‘‘2 C
1’’ scheme elucidated an interspin distance of 35 Å
between nitroxides attached to b-93 cysteine in tetrameric
hemoglobin..133/ Double quantum 2-D FTESR is the
most recent technique that looks directly at spin – spin
interactions i.e. the filtering out of a ‘‘normal’’ ESR
spectrum originating from isolated spins. This method
has been tested on solid polyproline peptides with an
interspin distance of 18 Å..101/
5.6.2 Spin Label – Paramagnetic Metal Method
An entirely different way of determining distances is
by coupling an unpaired spin of a nitroxide radical
to a fast-relaxing spin of a paramagnetic metal, e.g.
Cu(II) or Fe(III). The spin– lattice relaxation times of
many metals are three orders of magnitude faster than
those of nitroxide spin labels and thus, when coupled
to nitroxides, they provide an efficient relaxation path.
The relaxation enhancement can be detected by either
increased Lorentzian broadening or directly by the
shortening of T2 or T1 of the spin label in the presence of
the metal (Equation 61):
   
1 µ2 g2 4 24 12
 D 6 C C
T1s 6r T1f 5.w1 w2 /2 5.w1 C w2 /2 5w21
.61/
All three methods have been tested in spin labeled
hemoglobin in which a nitroxide at residue 93 of the
b-chain was coupled to Fe(III) bound to the heme. The
distance of 15 Å between the label and iron was in excel-
lent agreement with molecular modeling..134/
The metal relaxation is not limited to naturally
occurring metal binding sites. Site-specific spin labeling
has recently been extended to the engineering of metal
binding sites.
Voss et al. engineered Cu(II) binding sites to T4
lysozyme and lactose permease by introducing histidine
residues in consecutive turns of the a-helix..135,136/ Spin
labels were placed between 8 and 18 Å away from the
25
metal site by cysteine mutagenesis. The ESR line shapes
were broadened by paramagnetic Cu(II) chelated by the
histidine residues. Binding of diamagnetic metals did not
affect the line shape. The spectra were analyzed in terms
of the dipolar model of Leigh,.137/ which relates line-
shape broadening to the interspin distances rdd and metal
relaxation time T1m (Equation 62):
gbµ2 T1m
Hdd D .1 3 cos2 q/2 .62/
h̄r6
where µ is the magnetic moment of the metal.
A convenient experimental parameter is the amplitude
of the central line. The amplitude decreases with decreas-
ing spin – spin separation. Calibration curves derived from
computer simulations were used to estimate the inter-
spin distances in two model systems, lactose permease
and T4 lysozyme. Excellent agreement was found, not
only for the rigid samples for which Leigh’s model was
originally developed, but also for motionally narrowed
spectra due to the mobility of the nitroxides with respect
to the protein. For Leigh’s model to hold, the inter-
spin vector must not move on the same timescale as
dipolar interactions, which is T1 of the metal. This is
satisfied for molecules larger than 15 kDa (tr > 6 ns) and
for metals such as Cu(II) where T1 is 1 – 3 ns. The dis-
tances obtained for T4 lysozyme in solution at room
temperature were approximately 1 Å shorter than those
obtained from frozen proteins..136/ This small underesti-
mation of the distance is compensated by the biological
advantage of performing experiments at room temper-
ature and by the increased fidelity in measurements of
small-amplitude changes in sharper, motionally narrowed
spectra. Another elegant application of this method
involved the determination of helix packing in lactose per-
mease. Interspin distances between three helices labeled
with nitroxide labels and a metal site containing Cu(II)
determined the relative orientation of the helices and
their relative tilt..138/
Currently this method is limited to distances between 10
and 20 Å for Cu(II) with the X-band. Inspection of Leigh’s
Equation (61) suggests that the distance range might
be extended to ¾50 Å by using lower ESR frequencies
(S-band), metals with a larger magnetic moment (Gd3C )
or shorter T1 (Ni2C ) and also by direct measurement of
relaxation times..139/
5.6.3 Collision Exchange
A variation of spin – spin interactions is the relaxation
of spin labels by collisions with soluble paramagnetic
agents such as metals or O2 . Collisions lead to the
HSE, enhancing spin – lattice relaxation according to
Equation (63):
T1 1 D kWx .63/
26
4 where  is an instrumental factor which depends on
P1/2 = 60 mW
Amplitude
2 collision frequency is determined from Equations (63)
P1/2 = 20 mW
0 5 10 15
(Power)½ [(mW)½]
Figure 20 Power saturation of two samples with different P1/2
values.
where k is a factor accounting for the efficiency
of collisions and statistics of diffusion in two or
three dimensions and Wx is the bimolecular colli-
sion frequency. The collision frequency is of interest
because it reflects the accessibility of the labeled site
to the relaxant. Comparison of Wx for various sites
reveals which residues are exposed or hidden and
their secondary structure content and identifies tertiary
interactions.
Relaxation enhancement is measured directly by pulse
methods (saturation recovery or spin echo ESR) or by
continuous-wave power saturation. The amplitude of the
ESR signal increases linearly with the microwave mag-
netic field (H1 / P1/2 ) until the Boltzmann equilibrium
population difference is perturbed and the signal between
excited and ground states decreases. Samples with a long
T1 saturate easily and addition of relaxing agents relieves
this saturation (Figure 20).
The peak-to-peak amplitude (A) of the first-derivative
spectrum is given by Equation (64):.140/
p
A0 P
AD h p
ie .64/
e
1 C 2 1 .P/P1/2 /
where e depends on the resonance line shape and varies
between 0.5 for purely Lorentzian and 1.5 for Gaussian
line shapes; A0 is an instrument scaling factor and P1/2
is the half-saturation power (the power at which the
signal is half of what it would be in the absence of
saturation). P1/2 is determined either graphically or by
the fitting of experimental curves to Equation (64). The
P1/2 value is then used in calculating T1 according to
Equation (65):
22/3 1 sites. Most of the examples discussed below are from
T1 D
g2 2 P 1/2 T2
PEPTIDES AND PROTEINS
the power-to-magnetic field conversion of the resonator.
Since T2 for nitroxides is 2 – 3 orders of magnitude smaller
than T1 and because T2 is proportional to the peak-to-
peak line width of the central line (T2 / 1/H0 ), the
and (65) (Equation 66):
P1/2
Wx / P1/2 T2 D .66/
H0
In order to account for differences in resonators and
spectrometers between various laboratories, a dimen-
sionless accessibility parameter p was defined; p nor-
malizes Wx to the half-saturation power and line
width of a diphenylpicrylhydrazyl (DPPH) standard..141/
(Equation 67):
P1/2 H0DPPH
pD DPPH
.67/
H0 P1/2
Trends in accessibility to various relaxing agents
are used to determine the local environment of spin
labels. The relaxants can be nonpolar such as O2
partitioning into lipid bilayers, or polar with pref-
erence for the aqueous phase. The latter includes
neutral relaxants such as NiAA [nickel(II) acetylace-
tonate] and NiEDDA [nickel(II) ethylenediaminediac-
etate] and charged relaxants such as CROX [potassium
tris(oxalatochromate)]..142/
An important application of collisional relaxation is
the determination of the secondary structure of peptides
and proteins. Patterns of collisional accessibility along
the polypeptide chain can reveal a-helical folds, b-sheet
strands, immersion in membranes and chain tilt within the
membranes. These applications are described at length in
section 5.7.2.
5.7 Structural Biology
The advent of site-specific spin labeling established
ESR as a structural technique..41 – 43,46,142/ In the first
study,.30/ comparison of the relaxations enhancement
of four labeled cysteine mutants of bacteriorhodopsin
identified membrane embedded and surface exposed
residues. Since then, these approaches have been refined
and extended to establish (a) the topology of membrane
bound proteins, (b) the secondary structure of proteins
by cysteine scanning and following trends in accessibility
and mobility of residues and (c) the tertiary folding of
proteins by distance measurements between engineered
.65/
work of Hubbell et al.
ESR LABELING IN PEPTIDE AND PROTEIN ANALYSIS 27

5.7.1 Side-chain Environment – Immersion in

Membranes 0.09

Π (O2)
Differential effects of nonpolar (O2 ) and polar (NiAA, 0.07
CROX) relaxants are used to measure the immersion
depth of membrane proteins. This technique relies on
0.05
opposite concentration gradients for polar and nonpolar
relaxants within lipid membranes. The concentration of 0.7
nonpolar reagents increases with the immersion depth

Π (NiEDDA)
and the concentration of polar reagents decreases. The 0.1
further the nitroxide is from the aqueous interface, the
stronger is the relaxation enhancement by nonpolar
reagents and the weaker is the effect of polar relaxants.
The difference () between the polar and nonpolar
0.0
reagents as defined in Equation (68) is thus a function 245 247 249 251 253
of the immersion depth:.140/
(a) Residue no.
nonpolar
P1/2 4.0
 D ln polar
.68/
P1/2

Calibration curves of  are constructed using lipid spin 3.0

labels with nitroxides at defined positions along the
acyl chains and these curves are used to determine the
immersion depth of labels attached to membrane-bound
proteins (Figure 21a and b). 2.0
Φ

This relatively simple method, when used in con-

junction with cysteine scanning, differentiates between
membrane-bound and solvent-accessible surfaces of
membrane-associated proteins or peptides. As the
nitroxide is moved along the length of a polypeptide
chain, P1/2 for the polar agent shows minima and max-
ima for the residues interacting with the lipid bilayer and
the surface-exposed residues, respectively. The nonpolar
agent has a similar pattern of minima and maxima,
but it is offset by 180° with respect to the polar relax-
ant, i.e. maximum relaxation will be observed for the
residues interacting with the membrane and minimum
relaxation for water-exposed residues. A similar phase
shift is observed for the residues of helices lining the
aqueous pores of channels. Residues facing the lumen
of the pore show maximum relaxation enhancement for
polar agents (minima for the nonpolar oxygen), while
the residues facing the membrane environment have
maxima for oxygen and minima for CROX, NiAA and
NiEDDA.
Such is the case for the ferric enterobactin receptor
FepA, the transmembrane b-strand of which was found
to line an aqueous channel. The maxima of accessibil-
ity to NiEDDA was alternating with maxima to O2 ,
identifying the b-strand face as lining the channel and
the side of the b-strand facing the lipid bilayer..143/
Similar results have identified residues lining the aque-
ous channels in collicin,.144/ diphtheria toxin.145/ and
annexin..107/
1.0
0.0
245 247 249 251 253
(b) Residue no.
Figure 21 Immersion of a polypeptide chain in the lipid
bilayer. Differential accessibility to polar (N) and nonpolar (ž)
relaxants, , identifies the distance from the aqueous surface.
Calibration curves of  are constructed using lipid spin labels
with nitroxides at defined positions along the acyl chains..143/
(Reproduced with permission from C.S. Klug, W. Su, J.B. Feix,
Biochemistry, 36, 13027 – 13033 (1997).)
5.7.2 Secondary Structure Determination
Cysteine scanning also allows for secondary structure
determination. One method, based on the exchange
interactions between nitroxides attached to the i, i C 3
and i, i C 4 residues, identifies a- and 310 -helices and
was discussed in section 5.5.1.1. Other methods rely on
changes in nitroxide mobility and accessibility to relaxing
agents. The periodicity of steric interactions varies
along the polypeptide chain, which in turn determines
the nitroxide mobility and/or periodicity of relaxation
effects..146/ For example, an a-helix in an unevenly
28 PEPTIDES AND PROTEINS

80

Corrected solvent
60

accessibility
40

P (ω)
20
0
−20
−40
−60

274

276

278

280

282

284

286

288

0
20
40
60
80
100
120
140
160
180
(a) Residue no. Angular frequency (ω)

60
Corrected solvent

40
accessibility

20

P (ω)
0
−20
−40
−60
92
94
96
98
100
102
104
106
108
110

0
20
40
60
80
100
120
140
160
180
(b) Residue no. Angular frequency (ω)
Figure 22 Patterns of solvent accessibility for (a) b-strands of a-hemolysin (two-residue periodicity) and (b) an a-helix of
Streptomyces KC channel (3.6-residue periodicity). The Fourier transform identifying angular periodicity characteristic of a-helix
and b-strands is illustrated. (Courtesy of E. Perozo, unpublished.)

solvated environment (owing to the interaction with a annexin has been solved by X-ray crystallography. A
membrane surface or another polypeptide chain) shows mobility and accessibility profile of 26 single cysteine
a pattern of flexibility and solvent accessibility with a mutants in the helix – loop – helix motif has revealed a
3.6-residue periodicity. b-Strands, on the other hand, will dramatic structural transition when annexin is inserted
display a two-residue periodicity (Figure 22a and b). into the membrane to form a continuous, transmembrane
This characteristic periodicity of 3.6 residues was a-helix. As was expected for a membrane pore made of
observed for a number of helices of transmembrane the annexin trimer, one side of the helix was found to be
proteins: rhodopsin,.146 – 149/ collicin,.143/ K-channel.32/ highly solvated..152/
and the soluble protein T4 lysozyme..31/ Periodicity of
b-strands was observed for the transmembrane protein 5.7.3 Tertiary Structure: Conformational Changes
FepA receptor.143/ and water-soluble a-crystallin..150,151/
In some cases ESR has extended the structural informa- The greatest potential for the above methodologies
tion obtained by other methods, for example interhelical is the determination of tertiary structure. The current
loops in rhodopsin..147/ However, in other cases, the rate of structure determination by X-ray crystallography
secondary structure determined by ESR was the only or NMR (¾1000 per year) is too slow to solve for
available source for example FepA receptor.143/ and a- all 120 000 gene products. Fortunately, most proteins
crystallin..150/ are built from well defined, common structural motifs,
An interesting use of SDSL ESR is to extend but are packed in different ways to give proteins
monomeric (subunit) structures determined by NMR and their unique three-dimensional structure. It seems that
X-ray crystallography to the structures of functioning instead of solving ab initio the atomic structure of
macromolecular complexes. The monomeric structure of each protein it will be simpler to determine the relative
the soluble (nonfunctional) form of the membrane pore arrangement of common structural motifs. For instance,
ESR LABELING IN PEPTIDE AND PROTEIN ANALYSIS
a few chosen mutations can quickly establish whether
given helices or b-strands are in a parallel or antiparallel
arrangement..143,153/
Qualitative information about tertiary structure is
obtained from mobility and solvent accessibility values,
both of which are limited to sites of tertiary contacts,
e.g. p (O2 ) values are 0 – 0.05 for buried sites and 0.3
for solvent-accessible surface sites..119/ Additionally, in-
phase tracking of accessibility to polar and nonpolar
reagents and tracking of mobility patterns identifies
surface residues and residues buried within a protein core.
The surfaces involved in helix packing in rhodopsin.147,148/
and collicin.144/ and b-strand packing in a-crystallin.151/
have been identified by this in-phase behavior.
The tilt angle of polypeptide chains within lipid mem-
branes is easily determined from the immersion depth of
selected residues (see section 5.7.1). The immersion depth
is calculated from the  parameter (ratio of nonpolar to
polar accessibility), which is calibrated in terms of the
distance from the membrane surface of lipid spin labels at
defined positions. The average depth (d) of consecutive
residues is compared with the distance along the chain
(d0 ). The tilt of the chain with respect to the bilayer
normal is given by Equation (69):
 
d
a D cos 1 .69/
d0
For b-barrels, the tilt information, combined with number
of b-strands, can be used to estimate the diameter of the
barrel..144/
Most of the examples identifying conformational
changes are from membrane-bound proteins. In rhodop-
sin, helices flanking the ionone ring of retinal have
been labeled with nitroxides and the interspin distance
tracked upon photoisomerization of the retinal..130,154/
The observed rigid body rotation, with an associated
change of the tilt angle in one of the flanking helices,
resulted in increased accessibility of the cytoplasmic
loop. Increased exposure of the loop facilitates binding
of transducin to rhodopsin, which is the first step
in the phosphorylation cascade of signal transduction
pathway.
Conformational changes accompanying insertion into
a membrane and pore formation were observed by ESR
for the small cytosolic protein annexin. A water-soluble
monomer with a helix – loop– helix motif was rebuilt to
form a continuous transmembrane helix in the pres-
ence of Ca2C . The formation of long helix induced
membrane insertion of annexin..107/ In another example,
smaller conformational changes were observed by vary-
ing the lipid environment of transmembrane proteins.
Reconstitution of lactose permease into proteolipo-
somes induced a small 2-Å movement of neighboring
helices..127/
29
A particularly rewarding example is that of conforma-
tional changes in T4 lysozyme: two structures solved by
X-ray crystallography implied a hinge movement which
opened the active site by 8 Å. Using strategically placed
cysteines near the active site, this predicted opening of
the active site was verified in solution..129/ In addition to
corroborating the presence of the two conformers, ESR
was able to measure an equilibrium of closed and open
structures, yielding a unique estimate of activation energy
associated with catalysis.
The most spectacular application of ESR to the ter-
tiary/quaternary structure of proteins was that of the
bacterial potassium channel by Perozo..32,128/ Nearly a
third of the entire protein including two transmembrane
helices and the interhelical region flanking a selectiv-
ity filter have been scanned with spin labels. This is a
total of 62 mutants for the 160 amino acid polypeptide
chain. The channel is formed by the tetrameric assem-
bly of the two helices, with one helix (TM2) forming
an aqueous pore and the other helix (TM1) located on
the periphery. The structure has been solved indepen-
dently by X-ray crystallography.156/ and by ESR from
accessibility and mobility profiles..32/ The ESR protein
structure determination was further extended to the
structural description of the channel opening. The chan-
nel is activated by lowering the pH. Sequence profiles
of mobility and interspin distances were compared for
the open and closed forms, revealing a physical opening
of the central pore. The open form was brought upon
by a rigid body rotation and tilting of the TM2 helices
with an accompanying movement of the peripheral TM1
helices..128/
5.7.4 Assembly of Polypeptide Chains: Quaternary
Structure
Differences in mobility between monomers and oligomers
can be used to identify the oligomerization state of pro-
teins. For small proteins, ESR line shapes are motionally
narrowed, whereas the spectra of aggregates are consid-
erably broader. Spectral resolution of monomeric and
oligomeric forms in a composite spectrum allows for the
determination of their respective concentrations in solu-
tion and hence thermodynamic parameters of oligomer
formation. Oligomerization in a variety of solvents of
cecropin AD, a small ion channel, was studied in this
way..157/
The kinetics of the formation of amyloid plaques were
monitored by the disappearance of the sharp central
line of the spin-labeled amyloid protein monomer..158/
Monomers were found to aggregate initially into an
amorphous plaque precursor in which the protein
was in equilibrium between soluble monomers and
the aggregated protein. The precursor was an initi-
ation site for fibril formation of which the amyloid
30 PEPTIDES AND PROTEINS

plaques are subsequently formed. Characterization of dynamics, accessibility and distances at consecutive
of various assemblies and the equilibria between positions along the polypeptide chain is used in the
monomers and aggregates are of direct interest in determination of the secondary, tertiary and quaternary
understanding the molecular basis for diseases such as structure of proteins, which is of enormous importance
Creutzfeldt – Jacob (‘‘mad cow disease’’) and Alzheimer’s in the post-genomic era. It is likely that the ease of
disease. determination of relative orientation of known domain
An alternative way of following the formation of motifs will make ESR a method of choice in high-
aggregates is to utilize spin– spin interactions. Interacting throughput structural biology.
monomers lead to a broadening of the ESR spectra, Technical advances in ESR, which include new
provided that the labels are within the range of spin probes, FTESR, higher magnetic fields, increase in
exchange or dipolar interactions. Spin-labeled insulin absolute sensitivity, spectral dispersion and diversity of
B chain was found to aggregate on reduction of the applications bode well for the continued development of
interchain disulfide bonds, but the presence of a-crystallin ESR spectroscopy. Lastly, the development of powerful
was found to prevent aggregation. In the absence of computational simulations makes ESR user-friendly and
crystallin, the spectra of B insulin displayed a broad increases the number of ESR practitioners outside the
Lorentzian pattern, characteristic of closely placed spins. die-hard community of spectroscopists.
This turned into a normal powder pattern upon binding
to crystallin..119/
Titration of spin – spin interactions with unlabeled pro- ACKNOWLEDGMENTS
teins can be used to estimate the number of monomers
forming an oligomer. As the concentration of unla- This work was sponsored by the National Science Founda-
beled monomers increases, the probability of spin – spin tion (NSF-IBN-9808708), NHMFL (in-house grant) and
interactions decreases and dipolar broadening is relieved the American Heart Association (GIA-995024N).
leading to an increase in signal amplitude. For small
oligomers, dilution with small amounts of unlabeled pro-
tein results in a greater increase of signal amplitude LIST OF SYMBOLS
than for larger oligomers. The titration follows a bino-
mial expansion and has been used to establish that the A hyperfine interactions
membrane-bound form of annexin is a trimer..107/ A peak-to-peak amplitude
This qualitative approach has been used for annexin A hyperfine interaction tensor
pores. While the crystal structure of the soluble form of Amax , A0zz maximum hyperfine splitting
annexin suggested a hexameric assembly, nothing was a0 , A0 isotropic hyperfine splitting
known about annexin in membranes. Among the pos- c microwave field conversion factor
sibilities were a trimeric ring and a hexamer consisting C0 /C central manifold line-height ratio
of stacked trimers. To distinguish between these alterna- of the V20 spectrum
tives, cysteines were introduced at the interface between D diffusion tensor
the monomers forming a trimer and on the interface D? , Djj elements of diffusion tensor for motion
between the trimers forming a putative hexamer. In the parallel and perpendicular to the z-axis
soluble form, no dipolar interactions between any of the of a nitroxide
sites were observed, consistent with the monomeric form E energy
in solution. Addition of Ca2C , which triggers membrane f resonant frequency of a spin
binding, resulted in the broadening of the spectra within fjj , f? frictional coefficients
the trimer but not between the trimers, proving that a g Zeeman interaction tensor
trimer and not a hexamer was forming the pore..107/ geff effective g-value
H, H0 magnetic field strength
h, h̄ Planck’s constant
6 CONCLUSION H1 microwave field
Hc center field of a spectrum
ESR of protein is currently enjoying a renaissance of Hm modulation amplitude
sorts. In addition to its contributions in studies of protein Hres resonant field
dynamics and orientation, ESR is being increasingly used I0 .Hres / amplitude of the resonance line
as a structural technique. The advances of molecular J coupling strength
biology facilitate targeting of chosen domains or scanning L directional cosine matrices
of the whole structure with spin labels. Comparison L, C, H turning points of the spectra, q D 0°
ESR LABELING IN PEPTIDE AND PROTEIN ANALYSIS 31

L00 /L; H 00 /H line-height ratios of STESR line shape t? correlation time for rotation about
M2 second moment of the central axis perpendicular to nitroxide z-axis
resonance tr rotational correlation time
mI nuclear spin quantum number, tiso
r isotropic rotational correlation
nuclear manifold time
P microwave power tjj correlation time for rotation about the
P0 /P line-height ratios of V20 STESR spectra nitroxide z-axis
P1/2 half-saturation power qc cone angle
Q resonator quality factor q0 center of Gaussian angular
rdd interspin distance distribution
S electron spin q, f axial and azimuthal polar angles
S order parameter
T1 spin– lattice relaxation time
T1eff effective spin– lattice relaxation time
T2 spin– spin relaxation time
ABBREVIATIONS AND ACRONYMS
T2eff effective spin– spin relaxation time
V.mI / peak-to-peak height of a given ADP Adenosine Diphosphate
nuclear manifold resonance AFC Automatic Frequency Control
V0 absorption spectrum AMPPCP Adenosine Methylenetriphosphate
V1 first-derivative spectrum AMPPNP Adenosine Imidotriphosphate
V20 second-derivative, ATP Adenosine Triphosphate
90° out-of-phase display ATPase Adenosine Triphosphatase
Wx rate of bimolecular collisions ATPgS Adenosine Thiotriphosphate
Y.H/ electron spin resonance spectrum CROX Potassium Tris(oxalatochromate)
b Bohr magneton DEER Double Electron – Electron
Hm line width at half-height of a given Resonance
nuclear manifold resonance line DPPH Diphenylpicrylhydrazyl
Hpp peak-to-peak resonance line width DR Dielectric Resonator
q width of Gaussian angular distribution ELDOR Electron – Electron Double
w0AB difference in resonant frequencies Resonance
between spins A and B ENDOR Electron – Nuclear Double
h resonator filling factor Resonance
h viscosity ESR Electron Spin Resonance
g magnetogyric ratio FID Free Induction Decay
 width (at half-height) of the FTESR Fourier Transform Electron
resonance Spin Resonance
µ magnetic moment of an electron GAPDH Glyceraldehyde Dehydrogenase
n Microwave frequency HSE Heisenberg Spin Exchange
w Larmor frequency LGR Loop Gap Resonator
 orientational distribution NAD Nicotinamide Adenine
p dimensionless accessibility parameter Dinucleotide
to relaxants NiAA Nickel(II) Acetylacetonate
p normalized solvent accessibility to NiEDDA Nickel(II) Ethylenediaminediacetate
various quenchers NMR Nuclear Magnetic Resonance
 differential accessibility to polar PADS Peroxylamine Disulfonate
and nonpolar relaxants SDSL Site-directed Spin Labeling
r.q/ probability of the spins being orientated SECSY Spin-echo Correlation
at angle q with respect to the Spectroscopy
magnetic field SEESR Spin-echo Electron Spin
teff effective rotational correlation time Resonance
teff .mI / effective correlation time obtained SLE Stochastic Liouville Equation
from the STESR calibration curves of STESR Saturation Transfer Electron Spin
P0 /P ratios Resonance
tex exchange time TnC Troponin C
32
TnI Troponin I
TOAC 2,2,6,6-Tetramethylpiperidine-
1-oxyl-4-amino-4-carboxylic
Acid
UV Ultraviolet
2-D Two-dimensional
RELATED ARTICLES
Peptides and Proteins (Volume 7)
Separation and Analysis of Peptides and Proteins:
Introduction
Nuclear Magnetic Resonance and Electron Spin Reso-
nance Spectroscopy (Volume 13)
Nuclear Magnetic Resonance and Electron Spin Reso-
nance Spectroscopy: Introduction ž Electron Spin Reso-
nance Spectroscopy
REFERENCES
1. H. Thomann, M. Bernardo, ‘Pulsed Electron Nuclear
Multiple Resonance Spectroscopic Methods for Met-
alloproteins and Metalloenzymes’, Methods Enzymol.,
227, 118 – 189 (1993).
2. R. Cammack, C.E. Cooper, ‘Electron Spin
Resonance Spectroscopy of Iron Complexes and
Iron-containing Proteins’, Methods Enzymol., 227,
353 – 384 (1993).
3. C.J. Bender, P. Aisen, ‘Continuous Wave Electron
Nuclear Double Resonance Spectroscopy’, Methods
Enzymol., 227, 190 – 231 (1993).
4. G.W. Brudvig, ‘Electron Spin Resonance Spectroscopy’,
Methods Enzymol., 246, 536 – 554 (1995).
5. V.J. DeRose, B.M. Hoffman, ‘Protein Structure and
Mechanism Studied by Electron Nuclear Double Reso-
nance Spectroscopy’, Methods Enzymol., 246, 554 – 589
(1995).
6. M.P. Klein, ‘Perspectives on Magnetic Resonance and
X-ray Absorption Spectroscopy in Biochemistry’, Meth-
ods Enzymol., 246, 529 – 536 (1995).
7. E. Zavoisky, ‘Spin-magnetic Resonance in Paramagnet-
ics’, J. Phys. USSR, 9, 211 (1945).
8. C. Gorter, ‘Paramagnetic Resonance’, Physica, 3,
503 – 513 (1936).
9. I. Rabi, S. Millman, P. Kush, J. Zacharias, ‘The Molec-
ular Beam Resonance Method for Measuring Nuclear
Magnetic Moments’, Phys. Rev., 55, 526 – 535 (1939).
10. D. Ingram, J. Bennett, ‘Paramagnetic Resonance in
Phthalocyanine, Haemoglobin, and Other Organic
Derivatives’, Discuss Faraday Soc., 19, 140 – 146
(1955).
PEPTIDES AND PROTEINS
11. B. Commoner, J. Townsend, G. Pake, ‘Free Radicals in
Biological Materials’, Nature (London), 174, 689 – 691
(1954).
12. D. Ingram, J. Tapley, R. Jackson, R. Bond, A. Murna-
ghan, ‘Paramagnetic Resonance in Carbonaceous Solids’,
Nature (London), 174, 797 – 798 (1954).
13. W. Gordy, W.B. Ard, H. Shields, ‘Microwave Spec-
troscopy of Biological Substances – I: Paramagnetic
Resonance in X-irradiated Amino Acids and Proteins’,
Proc. Natl. Acad. Sci. USA, 41, 983 – 1004 (1955).
14. H.M. Assenheim, Introduction to Electron Spin Reso-
nance, Plenum Press, New York, 1967.
15. E.G. Rozantsev, in Free Nitroxyl Radicals, ed. H. Ulrich,
Plenum Press, New York, 1970.
16. O. Lebedev, S. Kazarnovskii, Tr. Khim. Khim. Tekhnol.
(Gorkii), 8, 649 (1959).
17. T.J. Stone, T. Buckman, P.L. Nordio, H.M. McConnell,
‘Spin-labeled Biomolecules’, Proc. Natl. Acad. Sci. USA,
54, 1010 – 1017 (1965).
18. G. Feher, ‘Method of Polarizing Nuclei in Paramagnetic
Substances’, Phys. Rev., 103, 500 – 501 (1956).
19. J.S. Hyde, J.C.W. Chien, J.H. Freed, ‘Electron – Electron
Double Resonance of Free Radicals in Solution’, J.
Chem. Phys., 48, 4211 (1968).
20. V. Benderskii, L. Blumenfeld, P. Stunzas, E. Sokolov,
‘Double Electron – Electron Resonance of Triplet Exci-
tons in Ion – Radicals Salts’, Nature (London), 220,
365 – 367 (1968).
21. W.B. Mims, K. Nassau, J.D. McGee, ‘Spectral Diffu-
sion in Electron Resonance Lines’, Phys. Rev., 123,
2059 – 2069 (1961).
22. U. Eliav, J.H. Freed, ‘The Oscillatory Nature of Polariza-
tion Evolution in CIDEP’, J. Phys. Chem., 88, 1277 – 1280
(1984).
23. O. Dobbert, T. Prisner, K.P. Dinse, ‘Single-channel Qua-
drature FT ESR’, J. Magn. Reson., 70, 173 – 175 (1986).
24. M.K. Bowman, ‘Fourier-transform Electron Spin Reso-
nance’, Bull. Am. Phys. Soc., 31, 524 (1986).
25. J. Gorcester, J.H. Freed, ‘Two-dimensional Fourier
Transform ESR Spectroscopy’, J. Chem. Phys., 85,
5375 – 5377 (1986).
26. O.Y. Grinberg, A.A. Dubinski, V.F. Shuvalov, L.G.
Oranskii, V.I. Kurochkin, Y.S. Lebedev, ‘Submilimeter
EPR Spectroscopy of Free Radicals’, Dokl. Phys. Chem.
(Engl. Transl.), 230, 884 – 886 (1976).
27. W.B. Lynch, K.A. Earle, J.H. Freed, ‘1-mm Wave ESR
Spectrometer’, Rev. Sci. Instrum., 59, 1345 – 1351 (1985).
28. F. Muller, M.A. Hopkins, N. Coron, M. Grynberg, L.C.
Brunel, G. Martinez, ‘A High Magnetic Field EPR
Spectrometer’, Rev. Sci. Instrum., 60, 3681 – 3684 (1989).
29. G. Eaton, S. Eaton, K. Salikhov, Foundations of Modern
ESR, World Scientific, Singapore, 1998.
30. C. Altenbach, S.L. Flitsch, H.G. Khorana, W.L. Hub-
bell, ‘Structural Studies on Transmembrane Proteins. 2.
Spin Labeling of Bacteriorhodopsin Mutants at Unique
Cysteines’, Biochemistry, 28, 7806 – 7812 (1989).
ESR LABELING IN PEPTIDE AND PROTEIN ANALYSIS
31. H.S. McHaourab, M.A. Lietzow, K. Hideg, W.L. Hub-
bell, ‘Motion of Spin-labeled Side Chains in T4
Lysozyme. Correlation with Protein Structure and
Dynamics’, Biochemistry, 35, 7692 – 7704 (1996).
32. E. Perozo, D.M. Cortes, L.G. Cuello, ‘Three-dimensio-
nal Architecture and Gating Mechanism of a KC Channel
Studied by EPR Spectroscopy’, Nature Struct. Biol., 5,
459 – 469 (1998).
33. B.W. Matthews, ‘Studies on Protein Stability with T4
Lysozyme’, Adv. Protein Chem., 46, 249 – 278 (1995).
34. D.D. Thomas, L.R. Dalton, J.S. Hyde, ‘Rotational Dif-
fusion Studied by Passage Saturation Transfer Elec-
tron Spin Resonance’, J. Chem. Phys., 65, 3006 – 3024
(1976).
35. P.G. Fajer, D. Marsh, ‘Microwave and Modulation Field
Inhomogeneities and the Effect of Cavity Q in Saturation
Transfer ESR Spectra. Dependence on Sample Size’, J.
Magn. Reson., 49, 212 – 224 (1982).
36. T.C. Squier, D.D. Thomas, ‘Methodology for Increased
Precision in Saturation Transfer Electron Spin Reso-
nance Studies of Rotational Dynamics’, Biophys. J., 49,
921 – 935 (1986).
37. E.J. Hustedt, C.E. Cobb, A.H. Beth, J.M. Beechem,
‘Measurement of Rotational Dynamics by the Simul-
taneous Nonlinear Analysis of Optical and EPR Data’,
Biophys. J., 64, 614 – 621 (1993).
38. G.I. Likhtenshtein, Biophysical Labeling Methods in
Molecular Biology, Cambridge University Press, Cam-
bridge, 1993.
39. L. Berliner (ed.), ‘Spin Labeling Theory and Applica-
tions’, in Biological Magnetic Resonance, Plenum Press,
New York, 1998.
40. L. Berliner, J. Reuben (eds.), Biological Magnetic Reso-
nance, Spin Labeling and the Next Millennium, Plenum
Press, New York, 1989.
41. W.L. Hubbell, C. Altenbach, ‘Investigation of Structure
and Dynamics in Membrane Proteins Using Site-directed
Spin Labelling’, Curr. Opin. Struct. Biol., 4, 566 – 573
(1994).
42. W.L. Hubbell, H.S. McHaourab, C. Altenbach, M.A.
Lietzow, ‘Watching Proteins Move Using Site-directed
Spin Labeling’, Structure, 4, 779 – 783 (1996).
43. W.L. Hubbell, A. Gross, R. Langen, M.A. Lietzow, ‘Re-
cent Advances in Site-directed Spin Labeling of Pro-
teins’, Curr. Opin. Struct. Biol., 8, 649 – 656 (1998).
44. D. Marsh, L.I. Horvath, in Advanced ESR: Applications
in Biology and Biochemistry, ed. A.J. Hoff, Elsevier,
Amsterdam, 707 – 752, 1989.
45. G.L. Millhauser, W.R. Fiori, S.M. Miick, ‘Electron Spin
Labels’, Methods Enzymol., 246, 589 – 610 (1995).
46. G.L. Millhauser, ‘Selective Placement of Electron Spin
Resonance Spin Labels: New Structural Methods for
Peptides and Proteins’, Trends Biochem. Sci., 17,
448 – 452 (1992).
47. E.J. Hustedt, A.H. Beth, ‘Nitroxide Spin – Spin Interac-
tions: Applications to Protein Structure and Dynamics’,
33
Annu. Rev. Biophys. Biomol. Struct., 28, 129 – 153
(1999).
48. T.P. Burghardt, N.L. Thompson, ‘Model-independent
Electron Spin Resonance for Measuring Order of Immo-
bile Components in a Biological Assembly’, Biophys. J.,
48, 401 – 409 (1985).
49. T.P. Burghardt, A.R. French, ‘Reconstruction of the
Probe Angular Distribution from a Series of Electron
Spin Resonance Spectra of Tilted Oriented Samples’,
Biophys. J., 56, 525 – 534 (1989).
50. D.E. Budil, S. Lee, S. Saxena, J.H. Freed, ‘Nonlinear-
least-squares Analysis of Slow-motion EPR Spectra
in One and Two Dimensions Using a Modified Lev-
enberg – Marquardt Algorithm’, J. Magn. Reson., 120,
155 – 189 (1996).
51. P.G. Fajer, B. Bennett, C. Polnaszek, E.A. Fajer,
D. Thomas, ‘General Method for Multiparameter
Fitting of High-resolution EPR Spectra Using a
Simplex Algorithm’, J. Magn. Reson., 88, 111 – 125
(1990).
52. P. Jost, O.K. Griffith, in Spin Labeling Theory and
Applications, ed. L.J. Berliner, Academic Press, New
York, 453 – 523, 1976.
53. P.G. Fajer, ‘Determination of Spin-label Orientation
Within the Myosin Head’, Proc. Natl. Acad. Sci. USA,
91, 937 – 941 (1994).
54. J. Seelig, in Spin Labeling Theory and Applications,
ed. L.J. Berliner, Academic Press, New York, 373 – 409,
1976.
55. E.J. Hustedt, A.H. Beth, ‘Determination of the Orien-
tation of a Band 3 Affinity Spin-label Relative to the
Membrane Normal Axis of the Human Erythrocyte’,
Biochemistry, 35, 6944 – 6954 (1996).
56. D.D. Thomas, R. Cooke, ‘Orientation of Spin-labeled
Myosin Heads in Glycerinated Muscle Fibers’, Biophys.
J., 32, 891 – 906 (1980).
57. J.E. Baker, I. Brust-Mascher, S. Ramachandran, L.E.
LaConte, D.D. Thomas, ‘A Large and Distinct Rotation
of the Myosin Light Chain Domain Occurs Upon Muscle
Contraction’, Proc. Natl. Acad. Sci. USA, 95, 2944 – 2949
(1998).
58. H.C. Li, P.G. Fajer, ‘Orientation Changes of Troponin
C Associated with Thin Filament Activation’, Biochem-
istry, 33, 14 324 – 14 332 (1994).
59. A. Redfield, ‘The Theory of Relaxation Processes’, Adv.
Magn. Reson., 1, 1 (1965).
60. D.J. Schneider, J.H. Freed, in Spin Labeling Theory
and Applications, eds. L.J. Berliner, J. Reuben, Plenum
Press, New York, 1 – 76, 1989.
61. D. Marsh, in Spin Labeling Theory and Applications,
eds. L.J. Berliner, J. Reuben, Plenum Press, New York,
255 – 303, 1989.
62. H.J. Steinhoff, W.L. Hubbell, ‘Calculation of Electron
Spin Resonance Spectra from Brownian Dynamics
Trajectories: Application to Nitroxide Side Chains in
Proteins’, Biophys. J., 71, 2201 – 2212 (1996).
34
63. D. Marsh, L. Horvath, ‘A Simple Theoretical Treatment
of the Sensitivity of Saturation-transfer EPR Spectra to
Slow Rotational Diffusion’, J. Magn. Reson., 99, 323 – 331
(1992).
64. S. Koening, ‘Brownian Motion of an Ellipsoid. A Cor-
rection to Perrin’s Result’, Biopolymers, 14, 2421 – 2423
(1975).
65. S. Broersma, ‘Rotational Diffusion Constant of a Cylin-
drical Particle’, J. Chem. Phys., 32, 1626 – 1631 (1960).
66. M.M. Tirado, J.G.d.I. Torre, ‘Translational Friction
Coefficients of Rigid Symmetric Top Macromolecules.
Application to Circular Cylinders’, J. Chem. Phys., 71,
2581 – 2587 (1979).
67. A. Beth, B.H. Robinson, in Spin Labeling Theory and
Applications, eds. L.J. Berliner, J. Reuben, Plenum
Press, New York, 179 – 253, 1989.
68. B.H. Robinson, L.R. Dalton, ‘Anisotropic Rotational
Diffusion Studied by Passage Saturation Transfer Elec-
tron Spin Resonance’, J. Chem. Phys., 72, 1312 – 1324
(1980).
69. E.J. Hustedt, A.H. Beth, ‘Analysis of Saturation Trans-
fer Electron Spin Resonance Spectra of a Spin-labeled
Integral Membrane Protein, Band 3, in Terms of the
Uniaxial Rotational Diffusion Model’, Biophys. J., 69,
1409 – 1423 (1995).
70. P.G. Fajer, D. Marsh, ‘Sensitivity of Saturation Transfer
ESR Spectra to Anisotropic Rotation. Application to
Membrane Systems’, J. Magn. Reson., 51, 446 – 459
(1983).
71. E.C. Howard, K.M. Lindahl, C.F. Polnaszek, D.D. Tho-
mas, ‘Simulation of Saturation Transfer Electron
Spin Resonance Spectra for Rotational Motion with
Restricted Angular Amplitude’, Biophys. J., 64, 581 – 593
(1993).
72. D.D. Thomas, J.C. Seidel, J.S. Hyde, J. Gergely, ‘Motion
of Subfragment-1 in Myosin and Its Supramolecular
Complexes: Saturation Transfer Electron Spin Res-
onance’, Proc. Natl. Acad. Sci. USA, 72, 1729 – 1733
(1975).
73. D.D. Thomas, S. Ishiwata, J.C. Seidel, J. Gergely, ‘Sub-
millisecond Rotational Dynamics of Spin-labeled Myosin
Heads in Myofibrils’, Biophys. J., 32, 873 – 889 (1980).
74. V.A. Barnett, D.D. Thomas, ‘Saturation Transfer Elec-
tron Spin Resonance of Spin-labeled Muscle Fibers.
Dependence of Myosin Head Rotational Motion on
Sarcomere Length’, J. Mol. Biol., 179, 83 – 102 (1984).
75. P.G. Fajer, E.A. Fajer, M. Schoenberg, D.D. Thomas,
‘Orientational Disorder and Motion of Weakly Attached
Cross-Bridges’, Biophys. J., 60, 642 – 649 (1991).
76. D. Raucher, P.G. Fajer, ‘Orientation and Dynamics of
Myosin Heads in Aluminum Fluoride Induced Pre-
power Stroke States: an EPR Study’, Biochemistry, 33,
11 993 – 11 999 (1994).
77. D. Raucher, C.P. Sar, K. Hideg, P.G. Fajer, ‘Myosin
Catalytic Domain Flexibility in MgADP’, Biochemistry,
33, 14 317 – 14 323 (1994).
PEPTIDES AND PROTEINS
78. B. Adhikari, K. Hideg, P.G. Fajer, ‘Independent Mobil-
ity of Catalytic and Regulatory Domains of Myosin
Heads’, Proc. Natl. Acad. Sci. USA, 94, 9643 – 9647
(1997).
79. E.M. Ostap, D.D. Thomas, ‘Rotational Dynamics of
Spin-labeled F-actin During Activation of Myosin S1
ATPase Using Caged ATP’, Biophys. J., 59, 1235 – 1241
(1991).
80. E.M. Ostap, T. Yanagida, D.D. Thomas, ‘Orientation
Distribution of Spin-labeled Actin Oriented by Flow’,
Biophys. J., 63, 966 – 975 (1992).
81. N. Naber, M. Lorenz, R. Cooke, ‘The Orientation of
Spin-probes Attached to Cys374 on Actin in Oriented
Gels’, J. Mol. Biol., 236, 703 – 709 (1994).
82. M. Mossakowska, J. Belagyi, H. Strzelecka-Golaszewska,
‘An EPR Study of the Rotational Dynamics of Actins
from Striated and Smooth Muscle and Their Complexes
with Heavy Meromyosin’, Eur. J. Biochem., 175, 557 – 564
(1988).
83. N. Naber, E.M. Ostap, D.D. Thomas, R. Cooke, ‘Ori-
entation and Rotational Dynamics of Spin-labeled
Phalloidin Bound to Actin in Muscle Fibers’, Proteins,
17, 347 – 354 (1993).
84. H.C. Li, K. Hideg, P.G. Fajer, ‘The Mobility of Troponin
C and Troponin I in Muscle’, J. Mol. Recognit., 10,
194 – 201 (1997).
85. H.C. Li, P.G. Fajer, ‘Structural Coupling of Troponin
C and Actomyosin in Muscle Fibers’, Biochemistry, 37,
6628 – 6635 (1998).
86. H.L. Sweeney, B.F. Bowman, J.T. Stull, ‘Myosin Light
Chain Phosphorylation in Vertebrate Striated Muscle:
Regulation and Function’, Am. J. Physiol., 264,
C1085 – C1095 (1993).
87. B.B. Adhikari, J. Somerset, J.T. Stull, P.G. Fajer, ‘Dy-
namic Modulation of the Regulatory Domain of Myosin
Heads by pH, Ionic Strength, and RLC Phosphoryla-
tion in Synthetic Myosin Filaments’, Biochemistry, 38,
3127 – 3132 (1999).
88. N.J. Ryba, D. Marsh, ‘Protein Rotational Diffusion and
Lipid/Protein Interactions in Recombinants of Bovine
Rhodopsin with Saturated Diacylphosphatidylcholines
of Different Chain Lengths Studied by Conventional
and Saturation-transfer Electron Spin Resonance’, Bio-
chemistry, 31, 7511 – 7518 (1992).
89. P. Fajer, P.F. Knowles, D. Marsh, ‘Rotational Motion
of Yeast Cytochrome Oxidase in Phosphatidylcholine
Complexes Studied by Saturation-transfer Electron Spin
Resonance’, Biochemistry, 28, 5634 – 5643 (1989).
90. L.I. Horvath, A. Munding, K. Beyer, M. Klingenberg,
D. Marsh, ‘Rotational Diffusion of Mitochondrial
ADP/ATP Carrier Studied by Saturation-transfer Elec-
tron Spin Resonance’, Biochemistry, 28, 407 – 414
(1989).
91. M. Esmann, H.O. Hankovszky, K. Hideg, D. Marsh, ‘A
Novel Spin-label for Study of Membrane Protein
Rotational Diffusion Using Saturation Transfer Electron
ESR LABELING IN PEPTIDE AND PROTEIN ANALYSIS
Spin Resonance. Application to Selectively Labeled
Class I and Class II-SH Groups of the Shark Rectal
Gland NaC /KC -ATPase’, Biochim. Biophys. Acta, 978,
209 – 215 (1989).
92. T.C. Squier, S.E. Hughes, D.D. Thomas, ‘Rotational
Dynamics and Protein – Protein Interactions in the Ca-
ATPase Mechanism’, J. Biol. Chem., 263, 9162 – 9170
(1988).
93. J.E. Mahaney, D.D. Thomas, ‘Effects of Melittin on
Molecular Dynamics and Ca-ATPase Activity in Sar-
coplasmic Reticulum Membranes: Electron Spin Reso-
nance’, Biochemistry, 30, 7171 – 7180 (1991).
94. D.A. Ferrington, T.E. Jones, Z. Qin, M. Miller-Schlyer,
T.C. Squier, D.J. Bigelow, ‘Decreased Conformational
Stability of the Sarcoplasmic Reticulum Ca-ATPase in
Aged Skeletal Muscle’, Biochim. Biophys. Acta, 1330,
233 – 247 (1997).
95. R.L. Cornea, L.R. Jones, J.M. Autry, D.D. Thomas,
‘Mutation and Phosphorylation Change the Oligomeric
Structure of Phospholamban in Lipid Bilayers’, Bio-
chemistry, 36, 2960 – 2967 (1997).
96. S. Negash, L.T. Chen, D.J. Bigelow, T.C. Squier, ‘Phos-
phorylation of Phospholamban by cAMP-dependent
Protein Kinase Enhances Interactions Between Ca-
ATPase Polypeptide Chains in Cardiac Sarcoplasmic
Reticulum Membranes’, Biochemistry, 35, 11 247 – 11 259
(1996).
97. D. Thomas, L. Reddy, C. Karim, M. Li, R. Cornea,
J. Autry, L. Jones, J. Stamm, ‘Direct Spectroscopic
Detection of Molecular Dynamics and Interactions of
the Calcium Pump and Phospholamban’, Ann. NY Acad.
Sci., 853, 186 – 194 (1998).
98. D. Schwarz, A. Chernogolov, P. Kisselev, ‘Complex For-
mation in Vesicle-reconstituted Mitochondrial Cytoch-
rome P450 Systems (CYP11A1 and CYP11B1) as
Evidenced by Rotational Diffusion Experiments Using
ESR and STESR’, Biochemistry, 38, 9456 – 9464
(1999).
99. B.H. Robinson, D.A. Haas, C. Mailer, ‘Molecular Dyn-
amics in Liquids: Spin – Lattice Relaxation of Nitroxide
Spin Labels’, Science, 263, 490 – 493 (1994).
100. S. Saxena, J.H. Freed, ‘Absorption Line shapes in Two-
dimensional Electron Spin Resonance and the Effects of
Slow Motions in Complex Fluids’, J. Magn. Reson., 124,
439 – 454 (1997).
101. S. Saxena, J.H. Freed, ‘Theory of Double Quantum
Two-dimensional Electron Spin Resonance with Appli-
cation to Distance Measurements’, J. Chem. Phys., 107,
1317 – 1340 (1997).
102. D.T. Thomas, E.M. Ostap, C.L. Berger, P.G. Fajer, J.E.
Mahaney, in Biological Magnetic Resonance – EMR of
Paramagnetic Molecules, eds. L. Berliner, J. Reuben,
Plenum Press, New York, 323 – 351, 1993.
103. A. Sienkiewicz, K. Qu, C. Scholes, ‘Dielectric Resonator-
based Stopped-flow ESR’, Rev. Sci. Instrum., 65, 68 – 74
(1994).
35
104. K. Qu, J.L. Vaughn, A. Sienkiewicz, C.P. Scholes, J.S.
Fetrow, ‘Kinetics and Motional Dynamics of Spin-
labeled Yeast Iso-1-cytochrome c: 1. Stopped-flow
Electron Spin Resonance as a Probe for Protein
Folding/Unfolding of the C-terminal Helix Spin-labeled
at Cysteine 102’, Biochemistry, 36, 2884 – 2897 (1997).
105. A. Sienkiewicz, A.M. da Costa Ferreira, B. Danner,
C.P. Scholes, ‘Dielectric Resonator-based Flow and
Stopped-flow ESR with Rapid Field Scanning: a Method-
ology for Increasing Kinetic Information’, J. Magn.
Reson., 136, 137 – 142 (1999).
106. Y.K. Shin, C. Levinthal, F. Levinthal, W.L. Hubbell,
‘Colicin E1 Binding to Membranes: Time-resolved Stud-
ies of Spin-labeled Mutants’, Science, 259, 960 – 963
(1993).
107. R. Langen, J.M. Isas, H. Luecke, H.T. Haigler, W.L.
Hubbell, ‘Membrane-mediated Assembly of Annexins
Studied by Site-directed Spin Labeling’, J. Biol. Chem.,
273, 22 453 – 22 457 (1998).
108. E.M. Ostap, H.D. White, D.D. Thomas, ‘Transient Det-
ection of Spin-labeled Myosin Subfragment 1 Confor-
mational States During ATP Hydrolysis’, Biochemistry,
32, 6712 – 6720 (1993).
109. S.M. Lewis, D.D. Thomas, ‘Resolved Conformational
States of Spin-labeled Ca-ATPase During the Enzymatic
Cycle’, Biochemistry, 31, 7381 – 7389 (1992).
110. S.M. Lewis, D.D. Thomas, ‘Microsecond Rotational
Dynamics of Spin-labeled Ca-ATPase During Enzy-
matic Cycling Initiated by Photolysis of Caged ATP’,
Biochemistry, 30, 8331 – 8339 (1991).
111. Z.T. Farahbakhsh, K. Hideg, W.L. Hubbell, ‘Photoacti-
vated Conformational Changes in Rhodopsin: a Time-
resolved Spin Label Study’, Science, 262, 1416 – 1419
(1993).
112. H.J. Steinhoff, R. Mollaaghababa, C. Altenbach, K. Hi-
deg, M. Krebs, H.G. Khorana, W.L. Hubbell, ‘Time-
resolved Detection of Structural Changes During the
Photocycle of Spin-labeled Bacteriorhodopsin’, Science,
266, 105 – 107 (1994).
113. T. Rink, J. Riesle, D. Oesterhelt, K. Gerwert, H.J. Stein-
hoff, ‘Spin-labeling Studies of the Conformational
Changes in the Vicinity of D36, D38, T46, and E161
of Bacteriorhodopsin During the Photocycle’, Biophys.
J., 73, 983 – 993 (1997).
114. M. Lindgren, M. Svensson, P.O. Freskgard, U. Carlsson,
P. Jonasson, L.G. Martensson, B.H. Jonsson, ‘Charac-
terization of a Folding Intermediate of Human Carbonic
Anhydrase II: Probing Local Mobility by Electron Spin
Resonance’, Biophys. J., 69, 202 – 213 (1995).
115. C.S. Klug, W. Su, J. Liu, P.E. Klebba, J.B. Feix, ‘Denat-
urant Unfolding of the Ferric Enterobactin Receptor
and Ligand-induced Stabilization Studied by Site-
directed Spin Labeling’, Biochemistry, 34, 14 230 – 14 236
(1995).
116. C.S. Klug, J.B. Feix, ‘Guanidine Hydrochloride Unfold-
ing of a Transmembrane Beta-strand in FepA Using
36
Site-directed Spin Labeling’, Protein Sci., 7, 1469 – 1476
(1998).
117. M. Svensson, P. Jonasson, P.O. Freskgard, B.H. Jonsson,
M. Lindgren, L.G. Martensson, M. Gentile, K. Boren,
U. Carlsson, ‘Mapping the Folding Intermediate of
Human Carbonic Anhydrase II. Probing Substructure by
Chemical Reactivity and Spin and Fluorescence Label-
ing of Engineered Cysteine Residues’, Biochemistry, 34,
8606 – 8620 (1995).
118. M. Persson, P. Hammarstrom, M. Lindgren, B.H. Jons-
son, M. Svensson, U. Carlsson, ‘EPR Mapping of Inter-
actions Between Spin-labeled Variants of Human Car-
bonic Anhydrase II and GroEL: Evidence for Increased
Flexibility of the Hydrophobic Core by the Interaction’,
Biochemistry, 38, 432 – 441 (1999).
119. Z.T. Farahbakhsh, Q.L. Huang, L.L. Ding, C. Altenbach,
H.J. Steinhoff, J. Horwitz, W.L. Hubbell, ‘Interaction of
Alpha-crystallin with Spin-labeled Peptides’, Biochem-
istry, 34, 509 – 516 (1995).
120. P.G. Fajer, E.A. Fajer, N.J. Brunsvold, D.D. Thomas,
‘Effects of AMPPNP on the Orientation and Rota-
tional Dynamics of Spin-labeled Muscle Cross-bridges’,
Biophys. J., 53, 513 – 524 (1988).
121. M. Zhao, K.C. Zen, W.L. Hubbell, H.R. Kaback, ‘Prox-
imity Between Glu126 and Arg144 in the Lactose Per-
mease of Escherichia coli’, Biochemistry, 38, 7407 – 7412
(1999).
122. W.R. Fiori, G.L. Millhauser, ‘Exploring the Peptide 310 -
helix Reversible Alpha-helix Equilibrium with Double
Label Electron Spin Resonance’, Biopolymers, 37,
243 – 250 (1995).
123. E.J. Hustedt, A.I. Smirnov, C.F. Laub, C.E. Cobb, A.H.
Beth, ‘Molecular Distances from Dipolar Coupled
Spin-labels: the Global Analysis of Multifrequency
Continuous Wave Electron Spin Resonance Data’,
Biophys. J., 72, 1861 – 1877 (1997).
124. P. Hanson, D. Anderson, G. Martinez, G. Millhauser,
F. Formaggio, M. Crisma, C. Toniolo, C. Vita, ‘Electron
Spin Resonance and Structural Analysis of Water
Soluble Alanine-rich Peptides Incorporating TOAC’,
Mol. Phys., 95, 957 – 966 (1998).
125. M.D. Rabenstein, Y.K. Shin, ‘Determination of the Dis-
tance Between Two Spin Labels Attached to a Macro-
molecule’, Proc. Natl. Acad. Sci. USA, 92, 8239 – 8243
(1995).
126. H.J. Steinhoff, N. Radzwil, W. Thevis, V. Lenz, D. Bra-
ndenburg, A. Antson, G. Dodson, A. Wollmer, ‘Deter-
mination of Interspin Distances Between Spin Labels
Attached to Insulin: Comparison of Electron Spin Res-
onance Data with the X-ray Structure’, Biophys. J., 73,
3287 – 3298 (1997).
127. J. Wu, J. Voss, W.L. Hubbell, H.R. Kaback, ‘Site-direc-
ted Spin Labeling and Chemical Crosslinking Demon-
strate that Helix V is Close to Helices VII and VIII in
the Lactose Permease of Escherichia coli’, Proc. Natl.
Acad. Sci. USA, 93, 10 123 – 10 127 (1996).
PEPTIDES AND PROTEINS
128. E. Perozo, D.M. Cortes, L.G. Cuello, ‘Structural Rear-
rangements Underlying KC -channel Activation Gating’,
Science, 285, 73 – 78 (1999).
129. H.S. McHaourab, K.J. Oh, C.J. Fang, W.L. Hubbell,
‘Conformational of T4 Lysozyme in Solution. Hinge-
bending Motion and the Substrate-induced Conforma-
tional Transition Studied by Site-directed Spin Labeling’,
Biochemistry, 36, 307 – 316 (1997).
130. K. Cai, R. Langen, W.L. Hubbell, H.G. Khorana, ‘Struc-
ture and Function in Rhodopsin: Topology of the C-
terminal Polypeptide Chain in Relation to the Cytoplas-
mic Loops’, Proc. Natl. Acad. Sci. USA, 94, 14 267 – 14 272
(1997).
131. D.J. Singel, R.G. Larsen, ‘Double Electron – Electron
Resonance Spin-echo Modulation: Spectroscopic Mea-
surement of Electron Spin Pair Separations in Ori-
entationally Disordered Solids’, J. Chem. Phys., 98,
5134 – 5146 (1993).
132. V. Pfannebecker, H. Klos, M. Hubrich, T. Volkmer,
A. Heuer, U. Wiesner, H.W. Spiess, ‘Determination of
End-to-end Distance in Oligomers by Pulsed EPR’, J.
Phys. Chem., 100, 13 428 – 13 432 (1996).
133. A. Raitsimring, J. Peisach, H.C. Lee, ‘Measurement of
Distance Distribution Between Spin Labels in Spin-
labeled Hemoglobin Using an Electron Spin Echo
Method’, J. Phys. Chem., 96, 3526 – 3531 (1992).
134. V. Budker, J.L. Du, M. Seiter, G.R. Eaton, S.S. Eaton,
‘Electron – Electron Spin – Spin Interaction in Spin-
labeled Low-spin Methemoglobin’, Biophys. J., 68,
2531 – 2542 (1995).
135. J. Voss, W.L. Hubbell, H.R. Kaback, ‘Distance Deter-
mination in Proteins Using Designed Metal Ion Binding
Sites and Site-directed Spin Labeling: Application to the
Lactose Permease of Escherichia coli’, Proc. Natl. Acad.
Sci. USA, 92, 12 300 – 12 303 (1995).
136. J. Voss, L. Salwinski, H.R. Kaback, W.L. Hubbell, ‘A
Method for Distance Determination in Proteins Using a
Designed Metal Ion Binding Site and Site-directed Spin
Labeling: Evaluation with T4 Lysozyme’, Proc. Natl.
Acad. Sci. USA, 92, 12 295 – 12 299 (1995).
137. J.S. Leigh, ‘ESR Rigid-lattice Line Shape in a System of
Two Interacting Spins’, J. Chem. Phys., 52, 2608 – 2612
(1970).
138. J. Voss, W.L. Hubbell, H.R. Kaback, ‘Helix Packing in
the Lactose Permease Determined by Metal – Nitroxide
Interaction’, Biochemistry, 37, 211 – 216 (1998).
139. H.R. Kaback, J. Voss, J. Wu, ‘Helix Packing in Poly-
topic Membrane Proteins: the Lactose Permease of
Escherichia coli’, Curr. Opin. Struct. Biol., 7, 537 – 542
(1997).
140. C. Altenbach, D.A. Greenhalgh, H.G. Khorana, W.L.
Hubbell, ‘A Collision Gradient Method to Deter-
mine the Immersion Depth of Nitroxides in Lipid
Bilayers: Application to Spin-labeled Mutants of
bacteriorhodopsin’, Proc. Natl. Acad. Sci. USA, 91,
1667 – 1671 (1994).
ESR LABELING IN PEPTIDE AND PROTEIN ANALYSIS
141. Z.T. Farahbakhsh, C. Altenbach, W.L. Hubbell, ‘Spin
Labeled Cysteines as Sensors for Protein – Lipid Inter-
action and Conformation in Rhodopsin’, Photochem.
Photobiol., 56, 1019 – 1033 (1992).
142. J.B. Feix, C.S. Klug, in Biological Magnetic Resonance:
Spin Labeling the Next Millenium, ed. L.J. Berliner,
Plenum Press, New York, 251 – 281, 1998.
143. C.S. Klug, W. Su, J.B. Feix, ‘Mapping of the Residues
Involved in a Proposed Beta-strand Located in the Ferric
Enterobactin Receptor FepA Using Site-directed Spin-
labeling’, Biochemistry, 36, 13 027 – 13 033 (1997).
144. L. Salwinski, W.L. Hubbell, ‘Structure in the Channel
Forming Domain of Colicin E1 Bound to Membranes:
the 402 – 424 Sequence’, Protein Sci., 8, 562 – 572 (1999).
145. K.J. Oh, H. Zhan, C. Cui, K. Hideg, R.J. Collier, W.L.
Hubbell, ‘Organized of Diphtheria Toxin T Domain in
Bilayers: a Site-directed Spin Labeling Study’, Science,
273, 810 – 812 (1996).
146. C. Altenbach, T. Marti, H.G. Khorana, W.L. Hubbell,
‘Transmembrane Protein Structure: Spin Labeling of
Bacteriorhodopsin Mutants’, Science, 248, 1088 – 1092
(1990).
147. C. Altenbach, K. Yang, D.L. Farrens, Z.T. Farahbakhsh,
H.G. Khorana, W.L. Hubbell, ‘Structural Features and
Light-dependent Changes in the Cytoplasmic Interhe-
lical E – F Loop Region of Rhodopsin: a Site-directed
Spin-labeling Study’, Biochemistry, 35, 12 470 – 12 478
(1996).
148. C. Altenbach, K. Cai, H.G. Khorana, W.L. Hubbell,
‘Structural Features and Light-dependent Changes in
the Sequence 306 – 322 Extending from Helix VII to
the Palmitoylation Sites in Rhodopsin: a Site-directed
Spin-labeling Study’, Biochemistry, 38, 7931 – 7937
(1999).
149. C. Altenbach, J. Klein-Seetharaman, J. Hwa, H.G. Kho-
rana, W.L. Hubbell, ‘Structural Features and Light-
dependent Changes in the Sequence 59 – 75 Connecting
Helices I and II in Rhodopsin: a Site-directed Spin-
labeling Study’, Biochemistry, 38, 7945 – 7949 (1999).
150. H.S. McHaourab, A.R. Berengian, H.A. Koteiche, ‘Site-
directed Spin-labeling Study of the Structure and Subunit
Interactions Along a Conserved Sequence in the Alpha-
crystallin Domain of Heat-shock Protein 27. Evidence
37
of a conserved Subunit Interface’, Biochemistry, 36,
14 627 – 14 634 (1997).
151. A.R. Berengian, M.P. Bova, H.S. McHaourab, ‘Struc-
ture and Function of the Conserved Domain in
AlphaA-crystallin. Site-directed Spin Labeling Identi-
fies a Beta-strand Located Near a Subunit Interface’,
Biochemistry, 36, 9951 – 9957 (1997).
152. R. Langen, J.M. Isas, W.L. Hubbell, H.T. Haigler, ‘A
Transmembrane Form of Annexin XII Detected by Site-
directed Spin Labeling’, Proc. Natl. Acad. Sci. USA, 95,
14 060 – 14 065 (1998).
153. S.J. Anthony-Cahill, P.A. Benfield, R. Fairman, Z.R.
Wasserman, S.L. Brenner, W.F.D. Stafford, C. Alten-
bach, W.L. Hubbell, W.F. DeGrado, ‘Molecular Char-
acterization of Helix – Loop – Helix Peptides’, Science,
255, 979 – 983 (1992).
154. D.L. Farrens, C. Altenbach, K. Yang, W.L. Hubbell,
H.G. Khorana, ‘Requirement of Rigid-body Motion
of Transmembrane Helices for Light Activation of
Rhodopsin’, Science, 274, 768 – 770 (1996).
155. K. Yang, D.L. Farrens, C. Altenbach, Z.T. Farahbakhsh,
W.L. Hubbell, H.G. Khorana, ‘Structure and Function in
Rhodopsin. Cysteines 65 and 316 are in Proximity in a
Rhodopsin Mutant as Indicated by Disulfide Forma-
tion and Interactions Between Attached Spin Labels’,
Biochemistry, 35, 14 040 – 14 046 (1996).
156. D.A. Doyle, J. Morais Cabral, R.A. Pfuetzner, A. Kuo,
J.M. Gulbis, S.L. Cohen, B.T. Chait, R. MacKinnon,
‘The Structure of the Potassium Channel: Molecular
Basis of KC Conduction and Selectivity’, Science, 280,
69 – 77 (1998).
157. H.S. McHaourab, J.S. Hyde, J.B. Feix, ‘Aggregation
State of Spin-labeled Cecropin AD in Solution’, Bio-
chemistry, 32, 11 895 – 11 902 (1993).
158. K.M. Lundberg, C.J. Stenland, F.E. Cohen, S.B. Prusi-
ner, G.L. Millhauser, ‘Kinetics and Mechanism of Amy-
loid Formation by the Prion Protein H1 Peptide as
Determined by Time-dependent ESR’, Chem. Biol., 4,
345 – 355 (1997).
159. J. Sun, J. Voss, W.L. Hubbell, H.R. Kaback, ‘Proximity
Between Periplasmic Loops in the Lactose Permease
of Escherichia coli as Determined by Site-directed Spin
labeling’, Biochemistry, 38, 3100 – 3105 (1999).