Overview of EPR Methods to

Measure Interspin Distances

Sandra S. Eaton, University of Denver

ACERT Workshop, August 7, 2004

Funding: EB002807
 Outline

• Dipolar Interaction
• Two slowly relaxing spins
• Half-field transition
• Lineshape changes
• Fourier deconvolution
• Pulse methods
• Rapidly relaxing spin + slowly relaxing
• T1e
• T2
• Fluid Solution
 Dipolar Interaction

The energy of interaction of a magnetic dipole µ1 with

magnetic dipole µ2 at distance r is
µ1µ 2
E=U=
r3

More generally, considering the vector properties of the

magnetic dipoles
r r r r r r
µ 0  µ1 ⋅ µ 2 3(µ1 ⋅ r )(µ 2 ⋅ r )
U dipolar =  − 
4π  r 3
r5 

Which is proportional to (1-3 cos2θ) where θ is the angle

between the interspin vector and the external magnetic field.
 Dipolar Splittings

Magnitude of dipolar coupling as a function of interspin distance

coupling: strong strong no exchange

r (Å) D (gauss) D (MHz) 2/3 D (MHz)
5 222 622 414
10 28 78 52
15 8.2 23 15
20 3.5 9.8 6.5
40 0.43 1.2 0.80
60 0.13 0.36 0.24
 Weak and Strong Exchange Cases

∆Ω = ω1 − ω2

Figure prepared by Gunnar Jeschke

 Interaction between two slowly relaxing spins

Typically two spin labels.

Could apply to metal – radical distance if measurement

is done at sufficiently low temperature that the metal is
relaxing slowly relative to the dipolar coupling.
 Human Carbonic Anhydrase II

Selected distances in
HCA II

67-206
Zn
8
9 C206 121-206
7 10 67-121
5 6
4
59-174
1 2
3
A174

I59 V121
N67
 Half-Field Transitions
Dipolar interaction between two spins shifts the triplet state ms =
± 1 energy levels relative to the ms = 0 level, and causes the
normally forbidden transition probability between the ms = -1
and ms = +1 levels to become non-zero. This transition occurs at
half the magnetic field required for the allowed transitions (at
constant microwave frequency), and hence is called the “half-
field” transition.

integrated intensity of half - field signal

relative intensity =
integrated intensity of allowed transitions

(19.5 ± 0.5)(9.1)2
relative intensity =
r 6ν 2

r is the interspin distance in Å and ν is the microwave

frequency, in GHz, at which the experiment was performed.
 Resolved Splittings of CW Spectra

• Analysis by computer simulation of lineshapes

• For shorter distances may need to include exchange as
well as dipolar interaction
• In favorable cases may be able to define the relative
orientations of the interspin vector and hyperfine axes
for two labels.
• Usually assumes that relative orientations of magnetic
axes for two centers are well defined
• Analysis of data at two microwave frequencies may be
required to obtain definitive results.
 Half-field Transitions and
CW Simulations

Interspin distance is 7 – 8 Å

Doubly-labeled
The relative intensity of the half-
field transition is 1.7x10-4

Sum of spectra of
Singly-labeled

After subtraction of singly-labeled,

with simulation
 Fourier Convolution/Deconvolution

• Assume ~ random distribution of relative orientations

or interspin vector and hyperfine axes.
• Fourier convolve spectrum of singly-labeled sample
with broadening function to match spectrum of doubly-
labeled samples
OR
• Divide Fourier transform of doubly-labeled spectrum
by Fourier transform of singly-labeled spectrum to
obtain broadening function

• Calculate the interspin distance from the "average"

broadening.

M. D. Rabenstein and Y.-K. Shin, Proc. Natl. Acad. Sci (US) 92, 8329 (1995).
H.-J. Steinhoff et al., Biophys J. 73, 3287 (1997).
 Fourier Deconvolution
HCAII59-174

Doubly-labeled 0.60

A
0.40

0.20
Sum of singly-labeled C
0.00

20 40
0.50

x4 0.00
After subtraction D

3200 3300 3400 20 40

Magnetic Field (gauss) Data point

Note that the baseline for the

r=8–9Å
deconvoluted function is close to
zero for the subtracted spectrum.
 Simulation and Fourier Deconvolution

HCAII121-206

First integral
A

x1.6 C

3200 3250 3300 3350

Magnetic Field (gauss)

0.00

3200 3250 3300 3350 0 20 40

Magnetic Field (gauss) Data point

r = 16 – 18 Å
The echo intensity is
recorded as a function of t.
In the absence of dipolar
interaction, a pulse at
frequency 2 has no impact on
echo intensity at frequency 1.
Dipolar interaction causes
oscillation in echo intensity
with a period that is
characteristic of the interspin
distance.

M. Pannier, S. Veit, G. Jeschke, and H.

W. Speiss, J. Magn. Reson. 142, 331
(2000).
DEER measurement of distance between
spin labels in carbonic anhydrase

r = 18 Å (70%)
HCAII121-206
24 Å 30%)

HCAII67-206
r = 20 ± 1.8 Å

0.0 0.2 0.4 0.6 0.8

time (µs)
 Distances (Å) Between Spin Labels on Carbonic Anhydrase
Determined from EPR Spectra
Doubly spin- Distance Half-field Fourier Lineshape DEER
labeled variant between β- transition Deconvo- Simulation
carbonsa lution
HCAII67-121 8.8 7 7-8 -

HCAII59-174 5.4 8 8.5-9 9-10 -

HCAII121-206 10.9 - 16-18 14-15b 18 (70%)

17-19c 24 (30%)

HCAII67-206 17.9 - 17-20 20 ± 1.8

aDistance between β−carbons of native amino acids at the sites where substitution with
cysteine was performed, calculated from the X-ray crystal structure, b Including unconstrained
contribution from singly-labeled protein, cAssuming 100% doubly-labeled protein.
Persson et al., Biophys. J. 80, 2886 (2001).
Other pulsed techniques for measuring distances between
two slowly relaxing spins

Double-Quantum Coherence – will be covered in subsequent lectures.

Out-of-phase echo – Specific to spin-correlated radical pairs.

Very powerful in photosynthesis.
Hoff et al., Spectrochim. Acta 54, 2283 (1998).
Measuring Distance Between a Rapidly Relaxing Metal Ion
and a Slowly Relaxing Spin

Types of systems
• Heme iron and a spin label
• Iron-sulfur cluster and a semiquinone radical

Approaches

• Changes in spin lattice relaxation times (T1e)

measured by saturation recovery
• Changes in spin-spin relaxation times (T2) measured
by two-pulse spin echo
Spin-labeled metmyoglobin
 Saturation and Relaxation
After a brief exposure to the microwave field, the populations of the spin
energy levels would become equal, if there were no way for the electrons to
relax back to the ground state. In fact, there are many mechanisms for spins to
relax back to the ground state.

P.F. Knowles, D. Marsh, H.W. E. Rattle, Magnetic Resonance of Biomolecules, Wiley, 1976.
 Relaxation Times
• Two relaxation times are fundamental to the narrowest EPR
line, which we call spin packets.
• T1 characterizes the relaxation of the spin from the excited state
to the ground state = longitudinal relaxation = spin-lattice
relaxation.
• T2 is called the transverse relaxation time or the spin-spin
relaxation time. In the simplest case, it is due to the variation
in resonant fields that result from other spins in the vicinity.
• Usually, T1 is much longer than T2.
• If T1 is short enough, it may determine T2, and then the line
width is characterized by T1 = T2.
 Distance Determination by Saturation Recovery
Measurement of Changes in T1e

Dipolar interaction between a rapidly relaxing center and a

more slowly relaxing center enhances the spin lattice
relaxation rate for the more slowly relaxing center.

Measurements are made of the relaxation times for the two
centers in the absence of interaction.

The saturation recovery curve for the more slowly relaxing
center is measured in the presence of the interaction.

Studies have been done with spin-labeled hemoglobin and
variants of myoglobin prepared by site-directed mutagenesis.
 Measure Metal Relaxation Rates:
Low Spin Metmyoglobin

Temperature (K) Temperature dependence of X-band

4 6 8 10 2 4 6 8 100 2 spin-lattice relaxation rates for 1
mM imidazole adducts of
9.0 metmyoglobin variants: R-Mb-
V66C-Im and R-Mb-K98C-Im (+)
metmyoglobin-cyanide
and 1 mM cyanide adducts: R-Mb-
7.0
V66C-CN, R-Mb-K98C-CN, and
log(1/T1 s-1 )

Raman
horse heart myoglobin cyanide ($)
in 1:1 water:glycerol. The solid
metmyoglobin-imidazole lines through the data are the fits to
5.0 the experimental data. The
contributions from individual
thermal direct processes to the relaxation for the
3.0 imidazole adducts are:
direct ( _ . . _ . . _),
Raman (- - - ), and
0.8 1.2 1.6 2.0
log(temperature)
thermal mode (_ . _ . _ ).
(Figure reproduced from Zhou et
al., 1999).
 Saturation Recovery

T1 and spectral T1
a diffusion

short microwave pulse long microwave pulse or

picket fence

2-pulse echo repetition rate

wait > 5 T1
obtain same echo amplitude
b
Increasing pulse repetition rate

echo

T1 and spectral diffusion

pulse repetition rate

Inversion Recovery

c T

T1 and spectral diffusion

echo

T
SR curves for Spin-Labeled Cyano-methemoglobin
SR signal intensity

120 K
Saturation recovery curves for
the nitroxyl signal in spin-
time(µsec)
0 40 80 120 160 200
labeled methemoglobin cyanide
in 1:1 buffer:glycerol at 9.2 GHz
and 15, 59, and 120 K. The
59 K
dashed lines were calculated for
SR signal intensity

r = 15.5 Å (Reproduced from

time(µsec) Seiter et al., 1995).
0 200 400 600 800 1000
SR signal intensity

15 K

time (µsec)
0 10000 20000 30000 40000 50000
 Bloembergen Equation
b 2T2 f c 2T1f e 2T2 f
1 = 1 + S(S + 1)[ + + ]
T1s T1s o 1+ ( ωf − ωs ) 2 T2 f 2 +b 2T1f T2f 1+ ωs 2T1f 2 1+ (ωf + ωs ) 2 T2 f 2
B term C term E term
2
2
b = 8 [− J − 1 g g β 2 (1−3 cos θ) ]2
3 2 4 s f hr 3
2 2
c 2 = 3g s 2 g f 2β 4 sin θ2cos
6
θ
h r
4
e 2 = 32 g s 2 g f β 4 sin2 6θ
2
h r

where "f" and "s" denote the fast- and slow-relaxing spins, respectively,
T1so is T1 for the slowly-relaxing spin in the absence of spin-spin interaction,
T1s is T1 for the slowly-relaxing spin perturbed by the fast-relaxing spin,
S is the electron spin on the faster-relaxing center,
ωf and ωs are the resonant frequencies for the fast- and slow-relaxing spins,
r is the interspin distance,
J is the electron-electron exchange interaction for the Hamiltonian written as
-JS1.S2, and
θ is the angle between the interspin vector and the external magnetic field.
For metals with S > ½ and large ZFS additional terms including ZFS in the
denominator are required.
Characteristics of metal ion that maximize its impact on
spin label relaxation

• Rapid relaxation at lower temperature where probe

relaxation is slower.
• Values of g near 2.00 maximize the B term.
• Contribution from C term is maximized when metal
relaxation rate is comparable to the EPR frequency.

b 2T2 f c 2T1f e 2T2 f

1 = 1 + S(S + 1)[ + + ]
T1s T1s o 1+ ( ωf − ωs ) 2 T2 f 2 +b 2T1f T2f 1+ ωs 2T1f 2 1+ (ωf + ωs ) 2 T2 f 2
B term C term E term
2
2
b = 8 [− J − 1 g g β 2 (1−3 cos θ) ]2
3 2 4 s f hr 3
2 2
c 2 = 3g s 2 g f 2β 4 sin θ2cos
6
θ
h r
2
e = 3 g 2 g 2β 4 sin 4 θ
2 s f h 2r 6
 Impact of Rapidly Relaxing Metal on Nitroxyl
CW Spectra and on Tm

Temperature/ low/slow intermediate high/fast

relaxation rate
impact of dipolar potentially broadened splittings averaged
interaction on CW resolvable by the rapid
spectra splitting relaxation
impact of dipolar Tm determined by Tm decreases, Tm determined by
interaction on echo other factors goes through a other factors
dephasing minimum and
then increases as
T increases
Two-pulse Spin Echo Decays for MbA15C-CN

6.9 K

19.1 K

70.4 K

0 2 4 6 8
time (µsec)

* maximum amplitudes of the curves have been normalized to 1.0

 Spin-Labeled Low-Spin Methemoglobin

6.0E+6 1/Tm at the center of the

nitroxyl signal in spin-labeled
oxy-hemoglobin (1), two
different spin labels attached to
4.0E+6 methemoglobin cyanide
(+, 5), and spin-labeled
1/Tm (Hz)

methemoglobin imidazole (h).

Although the ESE curves are
2.0E+6
not single exponentials, a fit to
a single exponential was used
to obtain a qualitative
description of the temperature
0.0E+0
dependence of 1/Tm. The lines
0 40 80 120 connect the data points
temperature (K)
(reproduced from Budker et al.,
1995).
 Analysis of electron spin echo decay for nitroxyl
interacting with a rapidly relaxing metal ion.

E(2τ) = R-2{exp(-2τ/τC)[tC-2sinh2(Rτ)+R2cosh2(Rτ)+
RτC-1sinh(2Rτ)+∆2sinh2(Rτ)]}
where
E(2τ) = intensity of echo as a function of interpulse spacing, τ

τC = correlation time for the dynamic process (T1fT2f)1/2

∆= ½ the angular frequency different between the two sites
that are averaged by the dynamic process

R2 = tC-2 – ∆2

Zhidomirov, G. M., Salikhov, K. M., Sov. Phys. JETP 29, 1037 (1969).
 Echo Decay Curves with Simulations

Mb-A15C-CN
6.9 K
r = 28.3 Å

Simulated based on
19.1 K
dynamic averaging
of dipolar splitting
due to rapid Fe(III)
70.4 K
relaxation.
0 2 4 6 8
time (µsec)

* maximum amplitudes of the curves have been normalized to 1.0

 Relative Echo Intensity – Calculated for Low-spin Fe(III)

1.00
Relative echo intensity
calculated for a nitroxyl
0.80 spin label interacting with
low-spin met myoglobin
Fe(III)-imidazole at
relative echo intensity

τ =500 ns for a range

0.60

interspin distances.
0.40

rvgim05.grf
r = 15
r = 20
r = 25
0.20
r = 35
r = 50

0.00

4.00 6.00 8.00 10.00

log(1/TauC)
 Relative Echo Intensities for
Spin-Labeled Metmyoglobin

Interspin distance for low spin metmyoglobin variants

Two-pulse spin echo intensity vs. saturation recovery

0.7 The intensity of a two-pulse

12
spin echo obtained with 40
0.6 and 80 ns pulses and
minimum echo intensity at 200ns

19
interpulse spacing of 200 ns
0.5 was measured as a function
15
149 53
57
of temperature. The
0.4 minimum value correlates
56
150
148 with the interspin distance
0.3 determined by saturation
87 recovery.
91
0.2
98 66

0.1
16 20 24 28 32
Distance from SR, l.s. Fe (Å)
Relative Echo Intensity for Spin-Labeled Mb A19C-CN
at τ= 200 ns
1

0.8

Simulation includes
echo intensity at 200ns

0.6 a distribution in iron

relaxation rates.
0.4

experimental data
0.2
simulated, r=28.0 A

0 20 40 60 80
temperature, K
Relative Echo Intensities for Spin-Labeled
MbA57C-CN at τ =500 ns

0.8
echo intensity at 500ns

0.6

Simulation includes
a contribution from
0.4 hemichromes

0.2
MbA57C-CN
experimental data
simulated, r=26.7A

0 20 40 60 80
temperature, K
Interspin distances (Å) measured by saturation recovery and spin echo EPR compared to those
calculated using Insight II software.

Myoglobin Fe-NO distance Fe-NO distance % Average Fe-NO distance

mutant Measured by SR for used for simulation hemichrome calculated using Insight II
low spin Fe(III) of ESE decay detected by CW
Distance distribution curves EPR
width = 0.1 Å
K98C 17.9 ± 0.3 19.7 0 16.9 ± 0.7
V66C 18.7 ± 0.7 19.7 0 19.0 ± 1.2
K87C 19.3 ± 0.8 21.0 11 18.8 ± 2.7
Q91C 19.7 ± 1.1 22.5 0 17.7 ± 1.1
L149C 22.4 ± 1.4 25.7 14 15.0 ± 2.0
G150C 22.8 ± 0.6 26.3 0 21.8 ± 1.6
K56C 23.0 ± 0.9 25.3 0 22.8 ± 1.8
E148C 23.3 ± 0.6 26.5 6 23.8 ± 3.2
A57C 24.9 ± 0.8 26.3 11 28.5 ± 1.3
A53C 25.1 ± 1.1 26.8 0 26.6 ± 1.0
A19C 26.1 ± 0.8 28.0 0 26.9 ± 1.7
A15C 26.2 ± 1.4 28.3 0 23.9 ± 1.3
H12C 29.8 ± 0.7 31.0 20 30.0 ± 1.2
 Comparison of Distances Obtained by ESE and
SR
32

Data for 13 spin-labeled

28
myoglobin variants. The distances
r from ESE (Angstroms)

obtained by ESE are

systematically longer than from
24 SR. This is the direction that
would be observed if there were a
distribution of distances. The
impact of the dipolar interaction
20
on the ESE minimum intensity
varies as r-3 but the impact on T1e
varies as r-6.
16
16 20 24 28 32
r from SR (Angstroms)
Summary of Methods in Rigid Lattice
Approaches to Measuring Distances

•Two slowly relaxing spins

• Half-field transition
• Lineshape changes
• Fourier deconvolution
• Pulse methods
•Rapidly relaxing spin + slowly relaxing
• Saturation Recovery Measurements of Changes
in T1e
• Two-pulse spin echo measurements on echo
decay shapes and relative echo intensities
These techniques all measure the electron-electron
dipolar interaction.
 Line Broadening in Fluid Solution

Berengian et al., J. Biol.

Chem. 274, 6305-6314
(1999).
 Fourier Deconvolution
in Fluid Solution
Analysis of spectra of doubly-labeled
T4 lysozyme in 40% sucrose
solution at ambient temperature.
a. Comparison of spectrum of double
mutant (D) with sum of spectra of
single mutants (S).
b. Comparison of deconvolution of D
with S and the sum of Pake
patterns obtained by fitting
procedure.
c. Comparison of spectrum (D) with
simulated spectrum (S convoluted
with Pake functions)
d. Distance distribution corresponding
to sum of Pake functions.

Altenbach et al., Biochemistry 40,

15471 (2001)
 Pulse Methods in Fluid Solution
• Best for small dipolar couplings
• To avoid averaging small dipolar couplings molecular tumbling
would have to be very slow.
• Long T2 is needed for pulse methods, which is difficult to achieve in
fluid solution.
 Relaxation Enhancement in Fluid Solution
T1

• Accessible distance range will be smaller than in a rigid lattice

because T1 for organic radicals is shorter in fluid solution.
• Requires metal ion with relaxation times in the correct range, i.e.
about 10-10 to 10-11. Two possibilities may be Cu2+ and Gd3+.

T2

• T2 at ambient temperatures typically is quite short.

• Not likely to have a metal with relaxation rate comparable to dipolar
couplings at ambient temperature.
 Further Information

Biological Magnetic Resonance , vol. 19, 2001

Distance Measurements by EPR

Biological Magnetic Resonance, vol. 23, 2004

Biomedical ESR - Part A: Free Radicals, Metals, Medicine, and Physiology

Biological Magnetic Resonance, vol. 24, 2004

Biomedical ESR - Part B: Methodology, Instrumentation and Dynamics