The strength of EPR and ENDOR techniques in revealing

structure–function relationships in metalloproteins

Sabine Van Doorslaer* and Evi Vinck

Received 1st February 2007, Accepted 11th May 2007

First published as an Advance Article on the web 6th June 2007

DOI: 10.1039/b701568b

Recent technological and methodological advances have strongly increased the potential of

electron paramagnetic resonance (EPR) and electron nuclear double resonance (ENDOR)

techniques to characterize the structure and dynamics of metalloproteins. These developments

include the introduction of powerful pulsed EPR/ENDOR methodologies and the development of

spectrometers operating at very high microwave frequencies and high magnetic fields. This

overview focuses on how valuable information about metalloprotein structure–function relations

can be obtained using a combination of EPR and ENDOR techniques. After an overview of the

historical development and a limited theoretical description of some of the key EPR and ENDOR

techniques, their potential will be highlighted using selected examples of applications to iron-,

nickel-, cobalt-, and copper-containing proteins. We will end with an outlook of future

developments.

1. Introduction

Since the introduction of electron paramagnetic resonance

(EPR) spectroscopy in 1944 by Zavoisky1 and of the related

electron nuclear double resonance (ENDOR) spectroscopy in

1956 by Feher,2 both techniques have been applied extensively

in the fields of biology, medicine, chemistry, physics and

material sciences. The techniques can be used to determine

the structure and dynamics of paramagnetic systems (i.e.

systems containing one or more unpaired electrons), and can

also be applied to diamagnetic systems using spin probes3 or

spin labels.4 Until the end of the 1980s, the vast majority of the

EPR/ENDOR studies were performed using continuous wave

(CW) techniques at X-band microwave (mw) frequencies

(B9.5 GHz). Although the use of these CW-EPR and

ENDOR techniques have proven to be invaluable for many

analyses, the amount of information that can be obtained is

limited by factors inherent to the CW approach. In the 1960s,

it was shown that the complementary use of pulsed EPR and

ENDOR may overcome these problems,5–7 but it was only

with the introduction of fast electronics and the commerciali-

zation of pulsed-EPR/ENDOR spectrometers in the late 1980s

that these techniques started to be applied on a larger scale.

Since then, the field of EPR has been revolutionized by new

developments both in pulsed-EPR methodology and in types

of applications, similar to the evolution observed in the field of

nuclear magnetic resonance (NMR), EPR’s (more famous)

sister technique. Furthermore, the progresses in EPR have

been paralleled by rapid instrumental developments, amongst

which is the construction of so-called ‘high-field EPR spectro-

meters’. These spectrometers operate at mw frequencies of

95 GHz and higher, and their development was initiated by the

seminal work of Lebedev.8 Details on the new methodological

and instrumental developments in EPR and on various appli-

cations are reported in the reference book on pulsed EPR by

Schweiger and Jeschke9 and in recent review papers,10–19

including two recent Invited Articles in this journal on the

applications of high-field EPR18 and high-field ENDOR.19 It

should be noted that EPR and ENDOR are by no means

routinely applicable techniques and the interpretation of EPR

and ENDOR spectra is often very complex.

For decades, EPR has been one of the methods of choice to

study paramagnetic metalloproteins. Indeed, many of the

metal ions binding to proteins are paramagnetic: Fe(III),

Cu(II), Ni(I), Ni(III), Co(II), Mn(II), VO(II), etc., and thus form

ideal intrinsic probes for EPR. The recent instrumental and

methodological developments in EPR and ENDOR have

made these techniques unparalleled tools for unravelling the

electronic and local geometric structure of the metal centre.

In this Invited Article, we will highlight some of the

possibilities of advanced EPR and ENDOR techniques for

the study of metalloproteins. The manuscript is structured as

follows. After an introduction to EPR and ENDOR spectro-

scopy, we show how the recent developments in EPR and

ENDOR techniques have given a new impetus to the elucida-

tion of metalloprotein structure and function. We hereby focus

on examples from our own research work, but extend the

discussion with work of others for those applications we feel

exemplify the field, even though they are not part of our

current research. We will conclude with indicating some of

the challenges for the future. Note that this overview is not

meant to be exhaustive and we apologize to our colleagues

whose work we may have inadvertently failed to quote.

SIBAC Laboratory, University of Antwerp, Universiteitsplein 1,B-2160 Wilrijk-Antwerp, Belgium. E-mail:sabine.vandoorslaer@ua.ac.be; Fax: +32 3 820 24 70; Tel: +32 3820 24 61

4620 | Phys. Chem. Chem. Phys., 2007, 9, 4620–4638 This journal is �c the Owner Societies 2007

INVITED ARTICLE www.rsc.org/pccp | Physical Chemistry Chemical Physics

2. Theoretical background of EPR spectroscopy

2.1. What information can be obtained using EPR

spectroscopy?

The spin Hamiltonian, H, describing the unpaired electron(s)

coupled to nuclear spins in an external magnetic field, forms

the basis for understanding the outcome of EPR experiments.

For an unpaired electron (S = 1/2), coupled to different

nuclear spins (Ik), the spin Hamiltonian is given by9,20–23

H ¼ be ~B0gS=hþXk

ð ~SAkIk � bngn;k ~B0Ik=hÞ

þX

k;Ik41=2

~IkPkIk ð1Þ

where be is the Bohr magneton, h is Planck’s constant, bn is thenuclear magneton and B0 is the external magnetic field. We

make use of the symbol B to indicate the transpose of a

vector.

The g tensor in the field-dependent electron Zeeman term

(first term of eqn (1)) contains information about the local

symmetry of the paramagnetic site and its electronic state. The

hyperfine interactions between the unpaired electron and the

surrounding nuclear spins are described by the second Hamil-

tonian term in eqn (1). The hyperfine matrix A can be written

as the sum of the isotropic, Fermi contact interaction, aiso,

originating in the finite probability for the unpaired electron to

be found at the nucleus, and of the dipole–dipole coupling, T.

For some cases, in which the electron–nucleus distance, r, is

larger than 0.26 nm and the spin delocalization over the ligand

is negligible, the latter term can be approximated by a

point–dipolar interaction that allows for a direct determina-

tion of r,24

Tij ¼m04p

1

r3gnbnbegið3rirj � dijÞ ði; j ¼ x; y; zÞ ð2Þ

with rx, ry and rz the direction cosines defining the orientation

of the magnetic field in the g tensor frame.

For transition-metal centers, knowledge of the metal hyper-

fine tensor (for metal isotopes with nuclear spin) in combina-

tion with the g tensor often provides a fingerprint to identify

the metal ion, its oxidation state and the local symmetry of the

metal site.

The third term in eqn (1) represents the nuclear Zeeman

interactions, whereby the nuclear g factors, gn,k, are identifying

the type of nucleus interacting with the unpaired electron. The

nuclear Zeeman interaction can be considered isotropic for

most EPR experiments, although in some particular cases a

pseudo-nuclear Zeeman interaction needs to be considered

(see also section 3.1.1).

The fourth term in eqn (1) represents the nuclear quadru-

pole interaction found for nuclei with spins I 4 1/2. It reflects

the interaction between the non-spherical charge distribution

of the nucleus with the electric field gradient, eq, caused by the

electrons and nuclei in the environment of the nucleus. The

principal components of the traceless nuclear quadrupole

tensor, P, are usually described as

Pz ¼e2qQ

2Ið2I � 1Þh ; Z ¼Px � Py

Pzð3Þ

where Q is the nuclear electrical quadrupole moment, Z is the

asymmetry parameter, and |Px| o |Py| o |Pz|.

For high-spin systems with S 4 1/2, the spin Hamiltonian

in eqn (1) will be extended with a field-independent fine-

structure term, the zero-field splitting,

HZFS = SDS (4)

The D tensor is the traceless, symmetric zero-field interaction

describing the dipole–dipole coupling between the electron

spins and a contribution from the spin–orbit coupling. It gives

direct information on the ground state and spin state of the

system at hand.

Finally, if the unpaired electrons of two neighbouring

paramagnetic sites are weakly interacting, as can be found in

biradicals or in cases of high concentration of paramagnetic

molecules, the spin Hamiltonian will be the sum of the two

individual spins added with an exchange-coupling term and a

dipole–dipole coupling term, revealing information on the

spin–spin distances. More details on this matter can be found

elsewhere.9

In short, a set of EPR experiments can in principle provide

insight into the following structural characteristics of a

metalloprotein. The g tensor and the zero-field interaction

can reveal important information on the ground state, oxida-

tion state and local symmetry of the metal center. Knowledge

of the hyperfine interaction between the unpaired electron(s)

and the transition-metal nucleus (I Z 1/2) can corroborate

this derivation. Furthermore, analysis of the hyperfine inter-

action of the surrounding nuclei allows one to unravel the spin

distribution in the system (via the Fermi contact term) and to

determine electron–nucleus distances and spatial information

(dipole–dipole contribution). The nuclear quadrupole interac-

tion reflects the electric-field gradient at the different quad-

rupolar nuclei. This gradient depends on the full electronic

structure. Put together, these EPR parameters allow for the

determination of both the local geometric and electronic

structure of the metal site. Finally, since most of the EPR

parameters are tensors (see eqn (1)), a detection of the time

averaging of these tensors will give direct information about

the dynamics of the system under study.

2.2. How to obtain this information

2.2.1. General. In order to determine the set of parameters

described in the previous section, the EPR spectroscopist can

use a multitude of CW and pulsed methods. The applicable

techniques can differ in many ways. They can make use of a

single mw frequency (e.g. X-band CW-EPR), multiple micro-

wave frequencies (e.g. electron double resonance (ELDOR)

spectroscopy), a combination of mw and radio frequencies (rf)

(ENDOR) or a combination of mw frequencies and optical

frequencies (e.g. optically detected magnetic resonance

(ODMR)). In order to enhance the obtainable information,

the spectroscopist may choose to combine results of experi-

ments performed at different mw frequencies. While the

majority of experiments is still performed at X-band mw

frequencies (B9.5 GHz), the impact of so-called high-field

EPR and ENDOR experiments, performed at frequencies

This journal is �c the Owner Societies 2007 Phys. Chem. Chem. Phys., 2007, 9, 4620–4638 | 4621

from 95 GHz (W-band) upwards, has been growing exten-

sively in recent years.10,16,18,19

In practice, at all times the EPR spectroscopist will have to

combine a number of techniques in order to obtain an as-

complete-as-possible set of parameters. Although the selection

of experiments is in general not trivial, we outline a rough

guideline for the study of transition metal complexes by means

of the example of a histidine-bound Cu(II). This scheme will

help the reader to understand the examples given in section 3.

Note that this guideline only includes the major techniques

highlighted in this paper, but does not pretend to represent the

full arsenal of techniques that can be used. In general, field-

swept EPR techniques, such as CW-EPR, can reveal informa-

tion about the g tensor, the zero-field tensor, the strongest

hyperfine interactions and possible electron spin–spin interac-

tions. The type of information that can be retrieved is spin-

system dependent (see section 2.1). In the case of the

Cu(II)–His example, the CW-EPR experiment will provide

information about the g and copper hyperfine tensor. Further-

more, ENDOR and ELDOR-detected NMR techniques can

reveal the hyperfine, nuclear Zeeman and nuclear quadrupole

couplings of nearby nuclei (Cu(II)–His: information about the

directly coordinating 14N of His and the nearby 1H). Finally,

electron spin echo envelope modulation (ESEEM) techniques

allow for an identification of these parameters for the nuclei

that are even further away from the unpaired electron(s)

(Cu(II)–His: remote nitrogen of the His and protein backbone

nitrogens). In short, as we move further away from the

unpaired electron(s), we will shift from field-swept EPR

techniques to ENDOR and ELDOR-detected NMR tools,

and finally to ESEEM techniques. In practice, the application

area of the different techniques will overlap partially and the

choice of the appropriate techniques is also nucleus dependent

(e.g. protons at a distance of 0.4 nm away from the unpaired

electron are perfectly detectable with X-band ENDOR, where-

by nitrogens at the same distance need to be probed with

X-band ESEEM techniques).

In the following sections we outline in more detail the most

important classes of techniques that are mentioned above and

that will recur in section 3.

2.2.2. Field-swept EPR techniques. Any EPR analysis will

start with the measurement of the CW-EPR spectrum of the

sample under study. In CW-EPR the absorption of continu-

ously irradiated microwaves with a constant frequency is

monitored as a function of the magnetic field, B0. For techni-

cal reasons, the derivative of the absorption signal is mea-

sured. For a simple S = 1/2 system, the magnetic field will lift

the degeneracy of the mS manifolds (see electron Zeeman term

in eqn (1)), and the state of the electron spin can be changed by

mw irradiation with an energy equal to the energy difference

between the two spin states. In general, the allowed EPR

transitions follow the selection rules |DmS| = 1, |DmI| = 0.

The field at which the resonance occurs links to the g value as

described by eqn (1). The resonance depends on the orienta-

tion of the magnetic field versus the molecular frame, since the

g tensor in eqn (1) has, in its most general case, non-equivalent

principal values. Since, in a powder or a frozen solution, the

paramagnetic molecule will take on all possible orientations

versus the static magnetic field, its CW-EPR spectrum is the

sum of the resonances of all these orientations. Because of the

magnetic-field dependence of the electron Zeeman interation

(eqn (1)), small differences in the principal values can be

resolved at high mw frequencies, where they may remain

unnoticed at the standard X-band frequency. The increased

g resolution is therefore one of the important advantages of

going to higher mw frequencies.

The CW-EPR spectra become more complex if hyperfine

interactions with the transition-metal nucleus or surrounding

strongly coupled nuclei are (partially) resolved. Furthermore,

the zero-field splitting will additionally complicate the spec-

trum for high-spin systems. If the zero-field interaction is

considerably larger than the electron Zeeman term, it may

become impossible to derive all parameters from the CW-EPR

spectrum. Again, experiments at higher mw frequencies can

help, since the field dependence of the electron Zeeman term

allows for a fine-tuning of the ratio between the two terms.

CW-EPR can be used not only to determine the static

structure of the paramagnetic molecule, but it is also a

valuable tool for measuring dynamical aspects. Molecular

motions that are very fast compared to the applied mw

frequency will cause a total averaging of the orientation-

dependent contributions and result in a so-called ‘isotropic’

spectrum (fast motion limit), whereas motions that are con-

siderably slower than the mw frequency are observed as static

(‘powder-like’ spectrum). All intermediate motions will give

rise to complex spectral features that allow the determination

of the nature and correlation time of the motion. Change of

the mw frequency thus shifts the reference to which the motion

is compared and a multi-frequency approach facilitates largely

the interpretation of the complex spectra. Further details on

CW-EPR can be found in the standard text books on this

topic.20–23

In principle, similar information can be obtained from

electron spin echo (ESE)-detected EPR. In these experiments,

a spin echo, usually a Hahn echo or a primary echo, is

generated using a specific mw pulse sequence and the echo

intensity is monitored as a function of the external magnetic

field. For an S = 1/2 system, this approach generates a

spectrum matching the integrated form of the corresponding

CW-EPR spectrum. In cases where the unpaired electron(s)

are interacting with surrounding nuclear spins, the ESE-de-

tected EPR spectrum may be distorted due to echo modula-

tions (see later) and special precautions should be taken when

interpreting these data.9 ESE-detected EPR may be preferred

over CW-EPR at high mw frequencies at which echo modula-

tions are shallow and the poor phase stability of the CW-EPR

experiments can cause severe spectral deformations. Further

details about the advantages of ESE-detected EPR and alter-

native mw pulse sequences can be found in ref. 9.

2.2.3. ENDOR spectroscopy. Although in some cases the

hyperfine interactions of the surrounding nuclei can partially

be resolved by field-swept EPR experiments, other techniques

need to be applied to map out the hyperfine and nuclear

quadrupole couplings in detail. In particular, the spectro-

scopist will look for methods to directly probe the nuclear

transitions (|DmS| = 0, |DmI| Z 1). Fig. 1a shows the energy

4622 | Phys. Chem. Chem. Phys., 2007, 9, 4620–4638 This journal is �c the Owner Societies 2007

levels for a S = 1/2, I = 1/2 system (e.g. an unpaired electron

coupled to a proton). The nuclear frequencies are in this case

given by9

nmS¼1=2 ¼ na ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiðA=2þ nI Þ2 þ B2=4

q

nmS¼�1=2 ¼ nb ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffið�A=2þ nI Þ2 þ B2=4

q ð5Þ

where A and B are related to the secular (Azz) and pseudo-

secular (Azx and Azy) hyperfine couplings: A = Azz and

B = (Azx2 + Azy

2)1/2. nI = �bngnB0/h is the nuclear Lamor

frequency. For isotropic hyperfine values and cases where |nI|c |A|, B, the nuclear frequencies reduce to

na,b = |nI � A/2| (6)

In the case where |nI| 4 |A/2|, one refers to the nuclei as being

weakly coupled, whereas |nI| o |A/2| is typical for strongly

coupled nuclei. The |nI| E |A/2| case is called the cancellation

condition.

For nuclei with I4 1/2, the nuclear frequencies additionally

depend on the nuclear quadrupole interactions.9,20,21,23

In order to detect the nuclear frequencies, ENDOR experi-

ments can be performed. In CW ENDOR, introduced by

Feher2 in 1956, the CW-EPR signal at a given field is saturated

by an increase in the mw power. The recovery of this signal is

then monitored as a function of the frequency of continuously

irradiated radiofrequency waves. Indeed, as the radio fre-

quency matches a nuclear frequency, the populations of all

energy levels will be changed and a net mw absorption can be

detected. The typical ENDOR spectra for the cases described

by eqn (6) are shown in Fig. 1b. Inspection of Fig. 1b reveals

immediately the advantage of ENDOR at high mw frequen-

cies. Indeed, higher mw frequencies imply the use of higher

magnetic fields and lead, through the magnetic-field depen-

dence of nI, to a better peak separation in the ENDOR

spectra. In many cases, a multi-frequency approach, combin-

ing ENDOR results obtained at different mw frequencies,

proves to be favourable. Note that the ENDOR spectrum will

only reflect the nuclear frequencies corresponding to the

orientations selected by the magnetic-field setting at which

the ENDOR experiment is performed. In order to reconstruct

the full hyperfine and nuclear quadrupole tensors, experiments

at different magnetic-field settings (‘observer positions’)

should be performed.

One of the most often used pulsed-ENDOR techniques is

Davies ENDOR spectroscopy25 (Fig. 2a). The first selective

mw p pulse inverts the population of a particular EPR

transition. Each time the frequency of the subsequent fre-

quency-swept rf p pulse becomes equal to a nuclear frequency,

the population of the EPR transitions will be affected. This

change is monitored via a two-pulse-echo detection scheme

(last two mw pulses). Mims ENDOR6 (Fig. 2b) offers an

alternative scheme for recording the nuclear frequencies

whereby the rf-induced changes in the grated polarization

created by the first two mw p/2 pulses are detected.

Pulsed ENDOR schemes offer the advantage over CW

ENDOR that certain ENDOR signals can be selectively

suppressed. An example of this is commonly used in Davies

ENDOR in order to disentangle overlapping ENDOR signals

of weakly coupled protons and strongly coupled nitrogens.

The former signals can be largely suppressed by the use of

short mw pulses (so-called hyperfine contrast selective

ENDOR9). Furthermore, more complex pulsed-ENDOR

schemes allow for a correlation of the nuclear frequencies

with other parameters in a second dimension. The HYEND

technique, in which ENDOR frequencies are directly corre-

lated to the hyperfine splitting, is a good example of this.26

Since a recent Invited Article in this journal focused on

pulsed ENDOR techniques,19 we refer to this work and other

reviews and standard text books for further details on CW and

pulsed ENDOR.9,11,19,27–29

2.2.4. HYSCORE spectroscopy. The nuclear frequencies

can also be determined using ESEEM techniques. In many

cases these techniques give complementary information to the

ENDOR experiments. Low-frequency signals (o5 MHz) are

best detected using ESEEM. In this way, ESEEM will be the

method of choice for the study of weakly coupled nitrogens,

as are often found for protein-embedded transition-metal

complexes.

Fig. 1 (a) Energy level for an S=1/2, I=1/2 system, whereby nI o 0, A, B4 0 and |nI|4 |A/2|. (b) The ENDOR spectrum described by eqn (6)

for the weak-coupling case |nI| 4 |A/2| (top spectrum) and the strong-coupling case |nI| o |A/2| (bottom spectrum).

This journal is �c the Owner Societies 2007 Phys. Chem. Chem. Phys., 2007, 9, 4620–4638 | 4623

In ESEEM experiments a sequence of mw pulses is applied

that generates an electron spin echo. The modulation of the

intensity of this echo is then monitored as a function of one of

the pulse intervals.9 Fourier transformation of this time signal

results in frequency spectra reflecting the nuclear frequencies

similar to an ENDOR spectrum. The origin of this echo

modulation is explained in detail in the standard text books9,30

and will not be treated here.

Here, we will only focus on the basic interpretation of

hyperfine sublevel correlation (HYSCORE) experiments.31

HYSCORE is a four-pulse ESEEM technique (Fig. 2c) where-

by the two interpulse distances t1 and t2 are varied indepen-

dently. Two-dimensional Fourier transformation leads to

frequency-domain spectra where the nuclear frequencies be-

longing to different mS manifolds are correlated. The

HYSCORE sequence is based on the three-pulse ESEEM

experiment, that consists of three p/2 mw pulses whereby the

primary spin echo is detected as a function of the time between

the last two pulses.

For the S = 1/2, I = 1/2 system described earlier, the

HYSCORE spectrum is characterized by a correlation pattern

as shown in Fig. 3a. For the weak-coupling case, the cross

peaks are found in the (++) quadrant, whereas they appear

in the (�+) quadrant for the strong-coupling case. As can be

seen from Fig. 3a, the hyperfine value can easily be estimated

from the peak positions. Furthermore, the correlation pattern

considerably facilitates spectral assignment for complex sys-

tems where the electron is interacting with many nuclei. For

disordered systems, e.g. frozen solutions of proteins, each

magnetic-field setting corresponds to a selection of molecular

orientations (so-called orientation selection) and the

HYSCORE signals will no longer be sharp cross peaks but

extended ridges representing the sum of the cross peaks

corresponding to each of these orientations. This is shown in

Fig. 3b, where a proton HYSCORE spectrum of a frozen

solution of a Cu(II)-bound prion protein (PrP(23-231)) is

depicted. The intersection of the diagonal and the anti-diag-

onal (dashed line) represents the proton Larmor frequency for

this field. This is a typical example of the weak-coupling case

(signals in (++) quadrant). Analysis of the HYSCORE

spectra of disordered systems requires spectral simulation

and is non-trivial.

For an S = 1/2, I = 1 system (e.g. an unpaired electron

interacting with a 14N nucleus), the correlation pattern be-

comes more complicated. In cases where only one molecular

orientation is selected by the magnetic field setting (single-

crystal case), six nuclear frequencies can be detected: four

single-quantum frequencies (|DmI| = 1) and two double-

quantum (DQ) frequencies (|DmI| = 2) (Fig. 3c). The corre-

sponding HYSCORE spectrum will then reveal information

about the nine correlations linking the nuclear frequencies of

the two mS manifolds. For cases with a considerable nuclear

quadrupole contribution (which is generally the case for 14N

nuclei), the HYSCORE spectra are dominated by the double-

quantum (DQ) cross peaks, linking the frequencies32

nDQa ;b ¼ 2

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffinI �

a

2

� �2þ e2qQ

4h

� �2

ð3þ Z2Þ

sð7Þ

where a is the hyperfine coupling at the particular observer

position. These DQ cross peaks will appear in the (++)

quadrant for a weak-coupling case and in the (�+) quadrant

for a strong-coupling case. The other cross peaks can be found

spread over both quadrants. This is exemplified in Fig. 3d,

showing the 14N HYSCORE spectrum of a VO(II) salen-type

molecule taken at an observer position known to excite all

molecular orientations in the ligand plane. This example was

chosen because it nicely represents some of the essential

characteristics of nitrogen HYSCORE spectra of disordered

systems. Similar spectra have been reported for VO(II) model

systems of the enzyme bromoperoxidase.33 Similar to the case

depicted in Fig. 3b, the HYSCORE features in Fig. 3d are

broad, because several molecular orientations are excited. The

SQ ridges are clearly broader than the DQ cross peaks. This

results from the fact that the SQ nuclear frequencies depend in

first order on the nuclear quadrupole interaction, whereby the

DQ nuclear frequencies only depend in second order on this

coupling (eqn (7)), resulting in less extended and hence more

intense DQ ridges. Using eqn (7) and the position and width of

the DQ cross peaks in Fig. 3d, we can immediately estimate

that the in-plane 14N principal hyperfine values will lie

Fig. 2 Pulsed EPR and ENDOR sequences: (a) Davies ENDOR, (b)

Mims ENDOR, (c) HYSCORE. In matched HYSCORE, the second

and third p/2 pulse are replaced by matched pulses, (d) SMART

HYSCORE, and (e) ELDOR-detected NMR.

4624 | Phys. Chem. Chem. Phys., 2007, 9, 4620–4638 This journal is �c the Owner Societies 2007

between 5.2 and 6.1 MHz and that 1.77 MHz r |e2qQ/h| r2.05 MHz for 1 Z Z Z 0, which will largely facilitate the

further spectral analysis and simulations.

In the exact cancellation case, the effective field experienced

by the nucleus in one of the two mS manifolds is approxi-

mately zero. The ESEEM frequencies within this manifold are

then close to the true nuclear quadrupole frequencies34

n0 ¼e2qQ

2hZ; n� ¼

e2qQ

4hð3� ZÞ; nþ ¼

e2qQ

4hð3þ ZÞ ð8Þ

In the HYSCORE spectra, sharp cross-peaks will be observed

linking the above frequencies with the DQ nuclear frequency

from the other mS manifold.

The nuclear quadrupole couplings of 2H nuclei are in

general very small, so that in this case, the HYSCORE spectra

are dominated by cross peaks between the basic frequencies,

rather than by the DQ frequencies. The typical forms of these

HYSCORE spectra are described in ref. 35.

Significant sensitivity enhancement can be obtained in

ESEEM experiments by the use of matched mw pulses.36

Matched pulses are high turning angle (HTA) pulses that

enhance the efficiency of forbidden transfers and thus drasti-

cally increase the intensity of basic- and combination-fre-

quency transitions. In the case of three-pulse ESEEM and

HYSCORE spectroscopy, the second and third p/2 pulse will

be matched to increase the efficiency of the forbidden transfers

between allowed electron coherence and nuclear coherence. If

one wants to enhance the ESEEM signals of a weakly coupled

nucleus, the field strength of the matching mw pulses, n1match,

should be taken to be equal to the nuclear Zeeman frequency

of this nucleus. For a strongly coupled nucleus, the field is

matched on the hyperfine coupling, which, in practice, usually

means that maximal mw power needs to be applied. The

optimal length of the HTA pulse then has to be determined

experimentally using a 2D three-pulse ESEEM experiment,

where the pulse length of the second and third p/2 pulse is

varied simultaneously in one dimension, and the time between

these two pulses is varied in the second dimension (normal

three-pulse ESEEM dimension).36 Fig. 4 shows a nice example

of the effect of mw pulse matching on the three-pulse ESEEM

spectra of an organic radical in polymethylmetacrylate. The

matching is done on weakly coupled protons (n1match = 15.125

MHz). Fig. 4a shows the echo modulation for a standard

three-pulse ESEEM (all mw pulse lengths equal to 16 ns)

showing shallow proton modulations. Fig. 4b depicts the huge

change in the modulation pattern when optimal matching

pulses are used (tp/2, match = 48 ns). This difference is also

reflected in the corresponding frequency-domain spectrum

(figure inset), showing now intense proton signals. Not only

is a clear signal centered around nH observed, but also the low

frequency component of a proton hyperfine coupling of

B28 MHz is now clearly visible (indicated by an asterisk in

the inset). Similar effects can be observed for the correspond-

ing HYSCORE spectra (not shown).

It is evident that the use of matched mw pulses has largely

increased the potential of the different ESEEM techniques.

Fig. 3 (a) Schematic representation of a HYSCORE spectrum of an S = 1/2, I = 1/2 system, with nI o 0, A, B 4 0 and B oo A, nI. For theweak-coupling case, the cross-peaks between the basic frequencies,na and nb, will appear in the (++) quadrant (squares), in the strong-coupling

case, the cross peaks are found in the (�+) quadrant (circles). (b) Experimental proton HYSCORE spectrum of a frozen solution of Cu(II)-bound

PrP(23–231) taken at an observer position corresponding to g= 2.37. t was 96 ns, and the pulse lengths, tp/2 = 16 ns and, tp = 16 ns, were taken.

(c) Schematic representation of the energy levels for an S = 1/2, I = 1 system indicating the double-quantum (DQ) and single-quantum (SQ)

nuclear frequencies. The example represents a strong coupling case. (d) Experimental 14N HYSCORE spectrum of a frozen toluene solution of

(R,R)-N,N0-bis(3,5-di-tert-butylsalicylidene)-1,2-cyclohexane-diamino vanadyl (II) at g = 2.056 (strong-coupling case). The DQ and SQ cross

peaks are indicated. t was 176 ns, and the pulse lengths, tp/2 = 16 ns and, tp = 16 ns, were taken.

This journal is �c the Owner Societies 2007 Phys. Chem. Chem. Phys., 2007, 9, 4620–4638 | 4625

Where ESEEM analyses would have been abandoned before

because of lack of significant echo modulations, the relevant

nuclear information can now be obtained.

Matched pulses can also be used to overcome one of the

most annoying problems of HYSCORE experiments, namely

the fact that the intensity of the HYSCORE signals depends

strongly on the time t (so-called t-dependent blind spots).

Because of this, HYSCORE spectra with different interpulse

distances, t, should be recorded and added for each observer

position. This considerably increases the measurement time.

The blind spots can be avoided by the use of an alternative

SMART HYSCORE pulse sequence37 (Fig. 2d). In this

scheme, the original p/2–t–p/2 nuclear-coherence generator

is replaced by a one-pulse nuclear-coherence generator

(matched pulse), avoiding the t-dependent blind spots. The

second p pulse in the sequence is needed to refocus the free

induction decay (FID) signal and allow echo detection.

Finally, since the modulation depth is proportional to

I(I + 1), the echo modulation tends to be dominated by the

modulations of the nuclei with the highest nuclear spin. In

complex situations where the unpaired electron is interacting

with several nuclei, this may lead to cross-suppression of the

weaker modulations.38 This can be circumvented by the use of

matching pulses36 or more extended ESEEM schemes.38

Further information on HYSCORE, ESEEM and other

pulsed EPR sequences can be found in ref. 9.

2.2.5. ELDOR-detected NMR. Although ELDOR-de-

tected NMR (Fig. 2e) is by far not as commonly applied as

the different pulsed-ENDOR or ESEEM sequences described

above, its use can sometimes be very advantageous.39,40 The

basics of the method are quite simple. The selective p pulse

with mw frequency 1 inverts the population of an EPR

transition. Whenever the variable microwave frequency of

the mw2 pulse matches one of the allowed or forbidden

EPR transitions, the population in the observed transition

will change. Observation of the FID signal as a function of the

difference between the two mw frequencies again results in a

spectrum reflecting the nuclear frequencies Alternatively, the

mw1 p pulse can be replaced by a p/2–t–p scheme whereby the

spin echo is then monitored as a function of the varying mw

frequency 2. A drawback of the ELDOR-detected NMR

method is the fact that the central hole (around nmw1 =

nmw2) will cover the signals at low frequencies. Fig. 5a shows

an extreme example of a broad central hole, but also in more

favourable cases, the central hole can still be significant

(Fig. 5b). The ELDOR-detected NMR approach is therefore

Fig. 5 (a) Q-Band ELDOR-detected NMR spectrum of a frozen

solution of Cu(II) tetraphenylporphyrin taken at an observer position

corresponding to g = 2.054. The spectrum is detected using a 60 ms(HTA, mw2) - 2.5 ms (delay) – 32 ns (mw1) - t - 64 ns (mw1) - t - echosequence, with t= 820 ns. The experiment was performed at 11 K. (b)

W-Band ELDOR-detected NMR spectrum of a frozen solution of

Cu(II)-bound peptide taken at an observer position corresponding to

g = 2.0465. The spectrum is detected using a 3 ms (HTA, mw2) - 5

ms(delay) – 120 ns (mw1) – FID sequence. The experiment was

performed at 10 K.

Fig. 4 Three-pulse ESEEM modulation pattern for the propagating radical of polymethylmetacrylate polymers. (a) Standard three-pulse

ESEEM, 16–t–16–t–16–t–echo, t = 200 ns. (b) Matched three-pulse ESEEM, 16–t–48(HTA)–t–48(HTA)–t—echo, n1match = 15.125 MHz, t =200 ns. The inset shows the corresponding frequency-domain spectra obtained after Fourier transformation. The signals marked with an asterisk

are discussed in the text.

4626 | Phys. Chem. Chem. Phys., 2007, 9, 4620–4638 This journal is �c the Owner Societies 2007

best suited to the detection of strongly coupled nuclei. Alter-

natively, the experiments can be performed at higher mw

frequencies (making use of the increase in the nuclear Zeeman

frequencies). The experiment depicted in Fig. 5a was per-

formed at Q-band mw frequencies. The proton Zeeman fre-

quency lies here at B50 MHz and the proton frequencies are

clearly separated from the central hole. At X-band mw

frequencies, nH lies around 15 MHz and a similar experiment

would then not allow an analysis of the proton interactions.

Note that the width of the central hole can be reduced by pulse

shaping and/or going to lower temperatures. It depends on the

local concentration of the paramagnetic centres (links to the

phase-memory time). The example in Fig. 5a can thus be

further optimized, but was chosen here to illustrate the

worst-case scenario.

At W-band mw frequencies, the ELDOR-detected NMR

approach can lead to better signal-to-noise ratios than pulsed

ENDOR, especially in the case of strongly coupled nuclei.40

Detection of these interactions implies the use of Davies

ENDOR, and the detection of the ENDOR spectrum of

strongly coupled nitrogen nuclei is notoriously difficult (and

sometimes impossible) at W-band, requiring thousands of scans.

The corresponding W-band ELDOR-detected NMR spectrum

can usually be detected in a few scans. The explanation for this is

probably manifold and factors such as sample heating due to the

rf pulse in the Davies scheme, the different relaxation paths

involved in the two experiments and the smaller amount of

pulses in the ELDOR-detected scheme should be considered.

Fig. 5b shows an example of a single-scan W-band ELDOR-

detected NMR spectrum of a Cu(II)-bound small His-containing

peptide. The spectral contributions of the directly binding

nitrogen of the histidine ligand can be readily identified after

only one scan. The full hyperfine and nuclear quadrupole tensor

can in this way be determined by recording the W-band

ELDOR-detected NMR spectra at different magnetic-field set-

tings (Maria Fittipaldi, work in progress).

2.3. From spectrum to model

In order to extract the EPR parameters from the different

experimental results, adequate simulation and fitting programs

are imperative. Due to the complexity and large diversity of

the EPR experiments and the spin systems that can be

addressed by them, an accurate simulation of the spectra is

not trivial. Although simulation programs targeted at the

simulation of specific EPR problems have been developed

over the years,41–47 it is only recently that attempts to develop

software packages that can tackle CW and pulsed EPR/

ENDOR simulations on a general level have been started.48

The matlab-based EasySpin program48 is at present the most

extended and flexible simulation package for both CW and

pulsed EPR spectra, but it still does not cover everything. It is

evident that the further development and generalization of the

simulation programs will remain one of the most important

aims in the field.

Once the EPR parameters have been determined experimen-

tally, the key question is how to interpret the different spin

Hamiltonian parameters in terms of useful electronic and geo-

metric structural information. In section 2.1. we already outlined

how the individual EPR parameters can in principle be linked to

different structural and electronic parameters. In practice, this

step proves to be very difficult. Let us take the example of the

hyperfine coupling. The Amatrix can be split in an isotropic and

an anisotropic part. While the isotropic part (Fermi contact

part) can still be easily linked to the spin density on the nucleus,

the interpretation of the anisotropic part is often less straight-

forward. This part contains orbital contributions and an elec-

tron–nuclear-distance dependence that are not easily separated.

Although several useful approaches have been explored and

used for many decades,9,20–24 such as the point–dipolar approx-

imation (see eqn (2)), there are many cases reported in which the

full information hidden in the A matrix could not be satisfacto-

rily decoded. This has, in combination with the fast increase in

EPR methodology and its linked increase in experimental data,

triggered the development of different quantum chemical ap-

proaches to compute the different spin Hamiltonian para-

meters.49,50 In this, a symbiotic relation has now been formed

between the theoretical and experimental approaches, as will

also become clear from some of the examples in section 3.

Experimental EPR data provide the ultimate control for the

quality of the quantum chemical computation. Although for

some cases density functional theory (DFT) and other ab initio

methods have proven to generate theoretical spin Hamiltionian

parameters that agree quantitatively with the experiment, a

qualitative agreement is, for the majority of cases, still the best

one can currently achieve, especially for transition metal com-

plexes.49,50 Even with these limitations, the quantum chemical

computations are starting to play an essential role in EPR

analysis. As mentioned in the previous section, the spectral

simulation of EPR/ENDOR spectra can become very complex

and at some points only a fraction of the hyperfine and nuclear

quadrupole information can be extracted in a first analysis. If

these parameters are then used to evaluate different theoretical

models, valuable hints about the principal values and directions

of the A and P tensors of the remaining magnetic nuclei can be

derived from the quantum chemical computations. This infor-

mation can then be used again as starting values to optimize the

spectral simulations in order to derive the full set of spin

Hamiltonian parameters for the system under study.51,52

Furthermore, once a satisfying agreement is found between

the experimental data and the theoretical model, not only

important structural information can be obtained from the

best-matching model, but also the computed ground-state

wavefunction can be linked to the inner working of the

transition-metal site. Indeed, important characteristics, such

as redox functions or ligand binding, are not only governed by

the geometric structure of the site, the electronic state plays an

equally important factor. Knowledge of the electronic struc-

ture is therefore essential if one wants to understand, for

instance, why apparently small changes can induce function

loss in one protein, and why seemingly more drastic changes

still lead to (partial) preservation of the metalloprotein’s

function in another case. Once a good theoretical model is

found, the effect of changes in the structure (e.g. mutagenesis)

can be probed theoretically. Thus, the amount of further

experiments can be reduced to a selective and well-targeted

set. In this way, the final outcome of the study exceeds what

could possibly be reached by a separate approach. One of the

This journal is �c the Owner Societies 2007 Phys. Chem. Chem. Phys., 2007, 9, 4620–4638 | 4627

essential goals for the future is therefore a flawless matching of

the two approaches, whereby the earlier-mentioned experi-

mental and theoretical problems have to be overcome.

2.4. Advantages and drawbacks of using EPR to study

metalloproteins

One of the important advantages of EPR is the large variety of

experiments that can be performed. For many paramagnetic

transition metal complexes, CW-EPR spectra can be obtained

in a large temperature range up to room temperature. For

these complexes, EPR can reveal important information about

the molecular dynamics on timescales between 10 ps and a few

ms.53,54 Information on the static structure of the transition-

metal site can be obtained from the different advanced pulsed-

EPR/ENDOR methods described in the previous sections.

However, in order to be able to use the latter methods, the

samples need to be cooled to very low temperatures (o25 K).

The predominant reason for the latter is the need to slow down

different relaxation pathways.9 Cooling is generally done using

liquid helium and together with the high cost of the EPR

equipment, this makes running an EPR facility expensive.

However, other advanced techniques, such as XRD, neutron

diffraction or NMR, are also very costly. A second drawback

is the fact that the spectral complexity requires specialist’s

input and a full structural analysis using EPR cannot be done

on a high through-put basis.

The amount of (protein) material needed for the analysis

depends on many factors, including the specific spin Hamilto-

nian parameters, the sensitivity of the specific spectrometer at

hand, and even the time (and money) the particular spectro-

scopist is willing to invest (measurement time and liquid-helium

consumption). In the most favourable cases, protein concentra-

tions of 50–100 mM are sufficient. This is the case when the

paramagnetic centre is characterised by a g tensor exhibiting

little anisotropy and when the (metal) hyperfine coupling is

small. An example of this is the NO-ligated ferrous form of a

heme protein. In extreme cases, protein concentrations in the

mM range are needed. This is the case for systems exhibiting

large g anisotropy and/or large hyperfine couplings. The EPR

study of low-spin ferric heme proteins requires concentrations of

1–3 mM (see section 3.1.1.). Note that the spin–spin relaxation

will become faster as the concentration of paramagnetic mole-

cules increases. At a certain concentration, the relaxation will be

so fast that no spin-echo signal will be detectable. Hence, an

increase in the concentration of a protein does not necessarily

lead to a higher echo intensity and the optimum concentration

needs to be established for each system separately. The volume

needed for the experiment depends on the mw frequency. While

X-band EPR measurements require about 100 ml of the sample,

only 1 ml is needed at W-band mw frequencies. In practice, the

available material will thus partially determine the type of

experiments that can be performed.

In many cases, EPR provides valuable (and sometimes

essential) information complementary to other structure char-

acterization methods. In NMR, the presence of the unpaired

electron(s) will usually cause large shifts and line broadenings,

making it difficult if not impossible to gain structural informa-

tion in the neighbourhood of the unpaired electron. EPR can

provide this essential information and thus complete the

experimental data. Single-crystal X-ray diffraction structure

analyses and neutron-diffraction analyses have proven their

extreme value in the determination of (metallo)protein struc-

tures. However, EPR studies can also provide unique com-

plementary information that may put XRD results in another

perspective, as will be exemplified in sections 3.1.1 and 3.2.1.

Other spectroscopic methods, such as extended X-ray absorp-

tion fine structure (EXAFS), optical and laser spectroscopies

are also often used complementarily to EPR spectroscopy.

3. Applications

In this section, we will describe a few examples of EPR and

ENDOR studies of iron-, nickel-, cobalt-, and copper-contain-

ing proteins. The examples are chosen so as to illustrate the

potential of state-of-the-art EPR and are not meant to be

exhaustive.

3.1. Iron-containing proteins

A glance in one of the many standard books on bioinorganic

chemistry reveals that various proteins contain iron-binding

sites or have an iron co-factor, many of them being para-

magnetic. We will focus in this section on examples of ferric

forms of heme proteins from our own research and on

iron–sulfur proteins. The latter proteins are not studied by

us, but their importance in biological processes merits some

attention in this overview.

3.1.1. Ferric form of heme proteins. The most well-known

iron-containing co-factor is probably the heme group (Fig.

6a), which is for instance found in globins. The iron ion of the

heme group is reported to occur in the ferrous, ferric and ferryl

form. The (paramagnetic) ferric (Fe(III)) forms of cyto-

chromes, peroxidases and catalases are biologically active.

The ferric or so-called ‘met-form’ of the globins was long

thought to be biological inactive, because oxygen can only

reversibly bind at the ferrous heme iron to realize the globin’s

most documented functions of oxygen storage and transport.

However, recently discovered new functions of vertebrate

myoglobin55 and the discovery of new globin-type proteins

in all kingdoms of nature with unknown functions56 have

renewed the interest in the ferric form of globins.

Ferric heme proteins can have different spin states: a low-

spin (LS) state (S= 1/2), or a high-spin (HS) state (S= 3/2 or

S = 5/2). The spin states are governed by the strength of the

axial ligands. All globins have a conserved histidine at position

F8 that binds to the heme iron (proximal side), whereby the

distal ligand can vary. A weak axial ligand, such as a distal

water, will lead to a HS state, which is typically observed for

the met form of vertebrate myo- and hemoglobin. Coordina-

tion of a strong axial ligand, such as an endogenous histidine,

then leads to a LS iron(III) state, as found for the recently

discovered vertebrate neuroglobin and cytoglobin.57 The

X-band CW-EPR spectra of frozen solutions of LS and HS

ferric heme proteins differ considerably. The LS ferric heme

proteins are characterized by a highly rhombic g tensor, as

exemplified by the principal g values of ferric mouse neuro-

globin: gx = 1.29, gy = 2.15, gz = 3.12.58 The X-band

4628 | Phys. Chem. Chem. Phys., 2007, 9, 4620–4638 This journal is �c the Owner Societies 2007

CW-EPR of HS globins can be described by the effective g

values (gx,eff E gy,eff E 6 and gz,eff E 2) (see later). X-Band

CW-EPR thus offers a fast and easy method to determine the

spin state of the iron.

Peisach and Blumberg systemized the analysis of the

CW-EPR spectra of several biological molecules and related

model systems. They constructed the so-called ‘‘truth tables’’,

which provide a means to relate the EPR parameters of a

paramagnetic site to structural information. This approach

was applied successfully to LS ferric heme proteins59 whereby

the g values give indications on the type and orientation of the

axial ligands binding to the heme iron. Although these tables

are based on crude assumptions, they can nevertheless give

important initial information on the system. Indeed, using this

simple approach, we could show for the case of ferric neuro-

globin that the iron center is bis-histidine coordinated, and the

histidine planes are not eclipsing the Fe–pyrrole nitrogen

bonds. Furthermore, the EPR parameters indicate that a

conformation in between parallel and near to perpendicular

alignment of the histidine planes is present.60 This was later

confirmed by the X-ray data.61,62

In 1986, Scholes and co-workers used CW-ENDOR to

study frozen solutions of bis(imidazole)-ligated LS ferric heme

complexes.63 Although they could derive interesting informa-

tion on the electronic and geometric structure of the heme

environment from this study, there were several method-

inherent limitations to the work, mainly due to the high degree

of spectral overlap of the contributions of the individual

nuclei. The same problems were observed for X-band Davies

ENDOR64 and three-pulse ESEEM65,66 studies of ferric heme

proteins, since in all cases the spectra are one-dimensional and

the spectral contributions of different nuclei will show a

similar degree of overlap.

Hoffman and co-workers showed that some of the above-

mentioned problems can be solved by using a combination of

CW, Davies- and Mims-ENDOR techniques at Q-band

(35 GHz) mw frequencies.67,68 Due to the magnetic-field

dependence of the Larmor frequency, nI, the spectral contri-

butions of the different nuclei can then be better separated at

higher fields. In this way, the authors established that the

allylbenzene-bound heme of inactivated chloroperoxidase is in

fact an N-alkylhemin metallocycle with a carbon of allylben-

zene bonded to the pyrrole nitrogen.68

Raitsimring, Walker and Astashkin showed in a series of

papers that it is advantageous to perform two- or four-pulse

ESEEM experiments at different settings of the magnetic field

and different mw frequencies.69–71 From the magnetic field

dependence of the proton combination peaks, detailed informa-

tion about the orientation of the proton hyperfine tensor in the g

tensor frame can then be obtained. For the nearby protons of the

axial imidazole ligands, this information can be directly linked to

the orientation of imidazole ligands in the heme pocket.

The first two-dimensional ESEEM experiments on ferric LS

heme proteins were performed by Garcıa-Rubio et al.72 They

compared the HYSCORE spectra of a bis(imidazole) ligated

heme compound, with selective isotopic substitution of the

nitrogens, to those of cytochrome b559. In this way the

HYSCORE peaks of cytochrome b559 could be assigned. The

large spectral similarity between cytochrome b559 and the

model system led to a detailed identification of the heme-

ligand orientations.

In our recent work on a ferric porphyrin model system, we

combined X-band CW-EPR, HYSCORE, and four-pulse

ESEEM experiments to maximize the information that can

be obtained from an EPR study of ferric heme complexes.73

The strategy consists of three steps. In a first step, the

Fig. 6 Schematic representation of (a) Fe(II) protoporphyrin IX (heme), (b) Rieske-type [2Fe–2S] centre, (c) coenzyme F430, (d) vitamin B12

(R = CN) or coenzyme B12 (R = 50-deoxy-50-adenosyl), (e) a copper site in azurin.

This journal is �c the Owner Societies 2007 Phys. Chem. Chem. Phys., 2007, 9, 4620–4638 | 4629

orientation of the hyperfine and nuclear quadrupole tensors of

the pyrrole nitrogens is determined in the g tensor frame by the

simulation of the nitrogen HYSCORE spectra. For all metal-

loporphyrin complexes the highest nuclear quadrupole value

of the pyrrole nitrogen is found to lie in the porphyrin plane

perpendicular to the metal–nitrogen bond. This provides a

means to determine the orientation of the g tensor frame in the

molecular frame. The analysis of the proton HYSCORE and

combination-peak experiments then provides the hyperfine

tensors of the imidazole protons. These parameters can be

decoded into the proton–iron distance and orientation of the

Fe–H vector in the g tensor frame. Using the results of the first

step, this information is then linked to the orientation of the

imidazole planes versus the porphyrin. In a final control step,

the imidazole nitrogen hyperfine and nuclear quadrupole

tensors are determined from the nitrogen HYSCORE spectra.

The in-plane rotation of the principal axes of these tensors is

linked to the orientation of the imidazole plane and should

match the earlier proton data. In a later work, this methodo-

logy was extended with pulsed ENDOR experiments and

applied to ferric mouse neuroglobin.74 Fig. 7a shows a typical

nitrogen HYSCORE spectrum of ferric mouse neuroglobin.

The DQ cross peaks of the heme and histidine nitrogens are

well separated. As outlined in section 2.2.4., initial spin

Hamiltonian parameters can be derived from the positions

of these cross-peaks using eqn (7), and this can then be refined

by spectral simulation. The structural parameters obtained by

this approach for the F8His and E7His of ferric mouse

neuroglobin matched the earlier X-ray data. A similar proce-

dure was also successfully applied to unravel the direct heme

environment of tomato hemoglobin, a globin for which no

prior structural information was available.75 In our recent

study of ferric cytoglobin, the strategy could also be used to

solve an ambiguity between two conflicting X-ray studies

(I. Ioanitescu, to be published). Although the error margins

on the structural parameters are larger than for those obtained

from X-ray diffraction studies, the method does not require

single crystals and can thus be used to study every ferric LS

heme protein without prior crystallization. The methodology

now also opens the possibility of identifying disulfide-bridge-

induced changes in the heme pocket of human neuroglobin. It

was earlier found that most neuroglobins, with the exception

of Rodentia and zebrafish neuroglobins, can form a disulfide

bridge between the cysteines on positions CD7 and D5.76

Formation of this disulfide bridge is linked to a change in

the oxygen affinity of the ferrous heme center. We showed that

the CW-EPR spectrum of ferric human neuroglobin is char-

acterised by two LS components with slightly different princi-

pal g values and that one of these components is clearly linked

to the disulfide bridge formation.58 Up until now, all efforts to

isolate and crystalize the latter form of the protein have failed.

The combined pulsed EPR/ENDOR approach may well pro-

vide the only way to identify the structural change occurring in

the heme pocket upon disulfide bridge formation. A large part

of the ferric forms of heme proteins are in a HS state. The mw

quantum energy at X-band (B9.5 GHz) is much smaller than

the zero field splitting (5–10 cm�1), so that the observed EPR

spectrum only arises from the transitions of the lower Kramers

doublet and can be described as an Seff = 1/2 system whereby

in second order the effective spin Hamiltonian is given by

Heff ¼ be~B0geffSeff=h

þXk

ð~SeffAk;effIk � bngn;k;eff ~B0Ik=hÞ

þX

k;Ik41=2

~IkPk;effIk ð9Þ

whereby the effective spin Hamiltonian parameters link to the

general spin Hamiltonian values of eqn (1) and (4) by

gk;x;eff ¼ gk;x 3� 12E

D

� �;

gk;y;eff ¼ gk;y 3þ 12E

D

� �;

gk;z;eff ¼ gk;z;

ð10Þ

Ak;x;eff ¼ Ak;xð3�12E

DÞ;

Ak;y;eff ¼ Ak;yð3þ12E

DÞ;

Ak;z;eff ¼ Ak;z �2Ak;zAk;y

D;

ð11Þ

Fig. 7 (a) Standard X-band HYSCORE spectrum of a frozen solu-

tion of ferric mouse neuroglobin taken at an observer position

corresponding to g = 1.86. t was taken to be 96 ns. The prominent

DQ cross-peaks stemming from the directly coordinating heme and

histidine nitrogens are indicated. (b) Matched X-band HYSCORE of a

frozen solution of the ferric form of the E7Q mutant of human

neuroglobin taken at an observer position corresponding to g =

3.48. The matching pulse sequence 8–t–24(HTA)–t1–16–t2–24(HTA)–

t echo, with n1match = 31.25 MHz and t = 96 ns, was used.

4630 | Phys. Chem. Chem. Phys., 2007, 9, 4620–4638 This journal is �c the Owner Societies 2007

gn;k;x;eff ¼ gn;k þ2gxbeAk;x

bnD;

gn;k;y;eff ¼ gn;k þ2gybeAk;y

bnD;

gn;k;z;eff ¼ gn;k:

ð12Þ

where D and E are the tetragonal and rhombic zero-field

splitting. The effective nuclear quadrupole tensor, Peff, usually

shows only minor deviations from the P tensor in eqn (1).77

As mentioned before, the X-band CW-EPR spectrum of HS

ferric heme compounds can be simulated with an effective

electron spin of 1/2, and effective g values. Unfortunately, the

X-band CW-EPR spectra of different HS ferric heme proteins

tend to be alike and reveal little information on the system

other than the identification of the spin state. A direct

determination of the zero-field splitting using CW-EPR is only

possible at high mw frequencies.78

In 1982, Scholes et al. published a seminal paper on the

CW-ENDOR analysis of a single crystal of aquometmyoglo-

bin, unravelling the hyperfine and nuclear quadrupole tensors

of the heme and histidine nitrogens.79 Since then, only a few

ESEEM or pulsed ENDOR studies of HS heme compounds

have been reported,80 contrasting with the recent evolution in

the field of EPR. There are two reasons for this. First of all, the

effective hyperfine values of the heme and imidazole nitrogens

reach values of 30–45 MHz for orientations in the heme

plane.79 These interactions cannot be observed using standard

ESEEM techniques. Furthermore, the pseudo-nuclear contri-

butions to the nuclear Zeeman term given in eqn (12) con-

siderably complicate the analysis of ENDOR and ESEEM

spectra, especially when no single crystals are available. In a

recent study, we showed how some of these problems can be

circumvented by using matched HYSCORE spectroscopy.77

Fig. 7b shows a matched X-band nitrogen HYSCORE spec-

trum of the ferric form of the E7Q mutant of neuroglobin

taken at an observer position g = 3.48. Because of the

sensitivity enhancement by the matched mw pulses mentioned

earlier, the nitrogen DQ cross peaks can readily be detected in

this spectrum. These signals are not revealed in the standard

HYSCORE experiment. Furthermore, clear ridges stemming

from nearby protons can be recognized in the (++) quadrant.

The ridges in the (�+) quadrant at frequencies lower than

15 MHz are complex to interpret and agree with the SQ

nitrogen frequencies and different combination frequencies

(nitrogen–nitrogen and nitrogen–proton).

The HYSCORE study of the ferric form of the E7Q-mutant

of human neuroglobin also revealed that this protein lacks the

characteristic distal water at neutral pH and that the lysine at

position E10 coordinates to the iron at high pH.77 Our current

work now focuses on testing the performance of different

advanced pulsed-EPR and ENDOR techniques, e.g.

HYSCORE, ELDOR-detected NMR, and HYEND, at dif-

ferent mw frequencies for the analysis of HS ferric heme

proteins (work in progress). The ELDOR-detected NMR

technique at W-band (95 GHz) mw frequencies in particular

promises to become a valuable tool in the analysis of ferric HS

systems.

3.1.2. Iron–sulfur proteins. Another important class of iron

proteins comprises the so-called iron–sulfur proteins. These

proteins have different functions, but are most frequently

involved in electron transport. Different iron–sulfur clusters

([2Fe–2S], [3Fe–4S], and [4Fe–4S]) are observed. Character-

istic of iron–sulfur proteins is the coordination of the iron ion

by cysteines, or occasionally by other amino acids of the

protein side chains, and, for polynuclear Fe–S centers, the

remarkable ligation with bridging ‘inorganic’ sulfide ions. It is

safe to say that EPR has played a crucial role in iron–sulfur

protein research. Indeed, based on their CW-EPR analysis,

Bertrand et al., could propose the existence of a new class of

[2Fe–2S] proteins, the so-called ‘‘Rieske-type’’ proteins (Fig.

6b), which are characterized by a larger g tensor anisotropy

compared to the classical ferredoxin type (gav B 1.91 instead

of gav B 1.96).81 They pointed out that the unusual g values of

this new type of [2Fe–2S] protein are consistent with the

presence of ligands at the ferrous site that are more electro-

negative than sulfide, which is in contrast to the classical

ferredoxin-type clusters where all ligation is provided by

sulfide. In recent years, more advanced pulsed EPR techniques

have been used successfully to identify the direct environment

of the iron–sulfur clusters as exemplified in ref. 82–84. A

detailed overview of the EPR and ENDOR work done on

iron–sulfur proteins is presented in ref. 84 and 85.

3.2. Nickel-containing proteins

Ni(I) (d9) and Ni(III) (d7) are paramagnetic ions. In many

cases, identification of the oxidation state of the nickel ion is

not straightforward and the nickel centre of many nickel-

containing enzymes is poorly characterized, as will become

apparent from the following examples. We focus here on the

EPR analyses of different forms of methyl-coenzyme M

reductase (MCR) to which one of us (S.V.D) also contributed.

In the second part, we also summarize some of the EPR work

done in field of [FeNi] hydrogenases. This is not our own

research field, but we feel it illustrates the potential and also

future challenges of EPR in the analysis of metal-containing

proteins nicely and it can therefore not been left out of this

overview.

3.2.1. Methyl-coenzyme M reductase (MCR). MCR plays

a crucial role in the methane-forming step of the energy

metabolism of archaea, since it catalyzes the reduction of

methyl-coenzyme M (2-(methylthio)ethane-sulfonate,

CH3SCoM) with coenzyme B (7-thioheptanoyl-threonine-

phospate, HSCoB) to methane and CoMS-SCoB.86 The en-

zyme contains the nickel-centred porphinoid co-factor, F430

(Fig. 6c). Different forms of the enzyme and their interconver-

sion relationships have been reported.87 Since many of these

forms are paramagnetic (including the enzymatically active

(‘ready’) forms), X-band CW-EPR played an important role in

differentiating the MCR forms.

The CW-EPR and ENDOR spectra of the isolated F430 co-

factor and of the active form of the enzyme, MCRred1, are very

alike and typical for a d9 Ni(I) complex with the unpaired

electron residing in the dx2�y2 orbital.88,89 The principal g

values are given in Table 1. Although the hyperfine couplings

of the four pyrrole nitrogens show small differences, they are

This journal is �c the Owner Societies 2007 Phys. Chem. Chem. Phys., 2007, 9, 4620–4638 | 4631

of the same order and typical for strongly coupled nitrogens

(Aiso E 28.5 MHz for MCRred1 and Aiso E 27.8 MHz89 for

Ni(I) F43088). Fig. 8a shows the X-band Davies ENDOR

spectra of the Ni(I) form of F430 pentamethyl ester. The top

spectrum was recorded using selective, soft mw pulses. It is

quasi-symmetrical around the proton Larmor frequency, nH,and is dominated by the spectral contributions of the nearby

protons of the F430 macrocycle. When using hard mw pulses

(bottom spectrum in Fig. 8a), the spectrum changes comple-

tely. The proton signals are now suppressed and the under-

lying signals of the strongly coupled nitrogens become

apparent. Since at this observer position all in-plane orienta-

tions are excited, the nitrogen ENDOR spectrum is complex,

stemming from many orientations. The two groups of four

lines in Fig. 8a indicate the basic nuclear frequencies corre-

sponding to the two in-plane principal directions of the

hyperfine and nuclear quadrupole tensor (Px E �1.85 MHz,

Py E 1.1 MHz88). The extra features stem from the other in-

plane orientations. Fig. 8a illustrates the hyperfine contrast

selective ENDOR effect mentioned in section 2.2.3.

In the presence of coenzyme M (2-mercaptoehanesulfonate,

HSCoM) and HSCoB, MCRred1 is converted into MCRred2

(maximum conversion is 50%). The CW-EPR spectrum of this

form is characterized by an unusual, highly rhombic g tensor,

whereby the broadening of the EPR line widths observed for61Ni-labeled MCRred2 reveals that the paramagnetic centre is

nickel based.87 Combination of X-band and Q-band ENDOR

and HYSCORE spectroscopy allowed us to determine the

hyperfine and nuclear quadrupole parameters of the pyrrole

nitrogens. Contrasting the case of MCRred1, two sets of

nitrogens with hyperfine values differing by about a factor of

two could be identified for MCRred290 (Aiso E 23.9 MHz and

Aiso E 13.8 MHz). The smaller hyperfine coupling could be

detected using Q-band HYSCORE (Fig. 8b). It was assigned

to the nitrogen of pyrrole A. The pyrrole ring A is not

p-conjugated and therefore more flexible than the other pyr-

role rings (Fig. 6c). Reduction of the nitrogen hyperfine values

therefore indicates that pyrrole A is slightly bent out of the

macrocyle plane. The Q-band HYSCORE spectrum of33S-labeled coenzyme M revealed additional features (signals

marked by the arrows in Fig. 8b) that stem from the 33S

interactions. The two cross-peaks in the (�+) quadrant at

(�10.8, 31.8) MHz and (�31.8, 10.8) MHz are assigned to the

triple-quantum transitions. The appearance of the 33S-related

cross-peaks proved unambiguously the coordination of CoM

to the nickel center in MCRred2.91

MCRox1 is an enzymatically inactive but ‘ready’ form of

MCR, which can be readily reduced to the active MCRred1.

The oxidation state of MCRox1 is highly debated, Ni(I) (d9),89,92

Ni(III) (d7),93 a Ni(III)-thiolate and a high-spin Ni(II) ion (S= 1)

antiferromagnetically coupled to a thiyl radical94 have all been

proposed. The fact that MCRox1 is converted to MCRred1 after

addition of Ti(III) citrate at pH 9 and 60 1C indicates that

MCRox1 is a Ni(III) complex.95 On the other hand, MCRox1 can

also be generated by cryoreduction of MCRox1-silent, an enzy-

matically inactive EPR-silent Ni(II) form, which makes a Ni(I)

oxidation state more probable. The g values (Table 1) and

pyrrole nitrogen hyperfine values (Aiso E 25 MHz) are typical

of an S = 1/2 species with an unpaired electron predominantly

residing in the nickel dx2�y2 orbital. Since both Ni(I) and Ni(III)

complexes with the unpaired electron in the dx2�y2 orbital have

been reported, this does not shed light on the oxidation state of

nickel in MCRox1. Furthermore, the X-ray absorption (XAS)

spectroscopy,96 UV-Vis97 and magnetic circular dichroism98

data indicate a close resemblance of MCRox1 to the EPR-silent

Ni(II) states.

Fig. 8 (a) X-Band Davies ENDOR spectra of the Ni(I) form of F430

pentamethyl ester taken at an observer position corresponding to g Egx (B0 = 337.4 mT). The top spectrum was taken using selective mw

pulses (tp/2 = 200 ns, tp = 400 ns) and a long rf p pulse (10 ms), thebottom spectrum was taken using hard mw pulses (tp/2 = 26 ns, tp =

52 ns) and a shorter rf pulse (5.4 ms). The meaning of the two groups of

four lines is explained in the text. (b) Q-Band HYSCORE spectrum

recorded at 25 K of MCR in the MCRred2 state with H33S-CoM,

observer position g = 2.28. The arrows identify the peaks that

originate from 33S interactions. The remaining cross-peaks stem from

an interaction with the nitrogen of pyrrole A. The spectrum was

reproduced with permission from C. Finazzo, J. Harmer, C. Bauer,

B. Jaun, E. C. Duin, F. Mahlert, M. Goenrich, R. K. Thauer, S. Van

Doorslaer and A. Schweiger, J. Am. Chem. Soc., 2003, 125, 4988–4989.

Copyright 2003 American Chemical Society.

Table 1 Principal g values of coenzyme F430 and different MCRforms

Complex gx gy gz Ref.

Ni(I)–F430 2.063 2.063 2.244 88MCRred1 2.060 2.070 2.2485 89MCRred2 2.1753 2.2313 2.2869 90MCRox1 2.1527 2.1678 2.2312 94

4632 | Phys. Chem. Chem. Phys., 2007, 9, 4620–4638 This journal is �c the Owner Societies 2007

Only recently, our multi-frequency EPR, ENDOR and

HYSCORE approach in combination with 61Ni, 33S and 2H

isotope labeling led to an identification of MCRox1 as a Ni(III)

thiolate in resonance with a thiyl radical/high-spin Ni(II)

complex.94 Analysis of the 61Ni and 33S hyperfine values

proved that Ni–S coordination was present in which 7 � 3%

of the spin density is on the sulfur. The spin-density distribu-

tion and Ni–S bonding show the non-innocent electron donat-

ing character of the sulfur ligand on the oxidation state and

can explain the apparent Ni(II)-like spectra observed with

XAS, UV-Vis and MCD. Deuteration of the CoM ligand

revealed that the distance between the nickel centre and b-protons of CoM is very similar to those of Ni(II)-MCRox1-silent

for which an X-ray study exists.99 Exchange experiments in

D2O also suggest that the binding of the SO3� group of CoM

may be stabilized by hydrogen bonding with two nearby

tyrosines (Tyra333 and Tyrb367).94

In the context of the previous descriptions it is worthwhile

to mention that all known X-ray structures of MCR describe

inactive, oxidized (Ni(II)) forms. At present, no crystals of the

important paramagnetic forms have been grown, so the EPR

technique provides the main characterization method for the

active centre of MCR.

3.2.2. [NiFe] hydrogenases. Hydrogenases catalyze the het-

erolytic splitting of molecular hydrogen. The metal-containing

hydrogenases are usually classified by the type of metal in their

active site: [NiFe] and [Fe]-only hydrogenases, of which the

former class is the largest.100 The active site of the [NiFe]

hydrogenases consists of a heterobimetallic cluster, in which

the nickel and iron atoms are bridged by the sulfur atoms of

two cysteines. Furthermore, two additional cysteines bind

terminally to the Ni atom. The iron atom is ligated by three

non-protein di-atomic molecules (two CN� and one CO). At

least seven oxidation states have been reported for [NiFe]

hydrogenases. Advanced EPR techniques have played and are

still performing a crucial role in the elucidation of the structure

and activation mechanisms of hydrogenases.101

In the oxidized form, two EPR-active forms co-exist: the

‘unready’ Ni–A and the ‘ready’ Ni–B form, which can be

activated by reduction under a H2 atmosphere in a few

minutes (Ni–B) or after incubation for hours (Ni–A).102 In

the aerobically ‘as-isolated’ enzyme (Ni–A) an additional

bridging of the Ni and Fe atom is achieved by (presumably)

an oxygen species as could be inferred from 17O ENDOR at

Q-band mw frequencies.103 Combination of the EPR and

ENDOR data with density functional theory (DFT) calcula-

tions lead to two possible models: (i) OH� as bridging ligand

and (ii) the bridging ligand is a diatomic-type ligand.104 A

similar approach for Ni–B allowed for the identification of an

OH� bridging ligand.105

Reduction of the enzyme by two electrons leads to the

catalytically active Ni–C state. Illumination at low tempera-

ture causes the conversion of Ni–C into another paramagnetic

state Ni–L. Using 2H2O exchange experiments in combination

with ENDOR and HYSCORE spectroscopy the presence of a

hydride as a bridging ligand could be proven for a regulatory

hydrogenase106 and for a catalytic hydrogenase.107 After

illumination at 10 K, the strong hyperfine coupling assigned

to the bridging hydride is lost, indicating a cleavage of the

metal–hydride bond.106

The EPR story of [NiFe] hydrogenases can, in our view, be

considered a textbook example of the potential of modern-day

EPR. Not only are the experimental data gathered using a

variety of multi-frequency EPR and ENDOR techniques101

revealing information unattainable by other spectroscopic

methods, the interpretation of the EPR data is also corrobo-

rated by DFT computations.104,108 As stated earlier, one of the

great challenges is the theoretical modelling of the magnetic

parameters extracted from the experimental data in order to

obtain direct structural information about the system at hand.

The example of the study of [NiFe] hydrogenases shows the

potential of an integrated EPR-DFT approach. Note that we

focused in this section on the Ni-related centres in [NiFe]

hydrogenases. However, these proteins contain 11 more iron

atoms besides the one in the Ni–Fe subunit. These additional

Fe atoms are part of three iron–sulfur clusters and can also be

studied by EPR (see section 3.1.2).

3.3. Cobalt-containing proteins. Cobalt(II) is found in nat-

ural proteins, such as some active forms of B12 enzymes, but is

also a good functional mimic of other metal ions. Further-

more, Co(II) is an excellent spectroscopic probe since it is both

optically and EPR active. For these reasons, the unique

oxygen binding properties of Co(II) porphyrins have obtained

considerable attention in biochemical studies, because these

complexes can be used as model systems for the ligand-binding

characteristics of globins.109 Because of the fast autoxidation

rates, the analysis of the oxygenated native heme proteins is

often very difficult. Cobaltous-substituted porphyrins can then

offer a way to study both the oxy and deoxy species using CW

and pulsed EPR and ENDOR.110,111 Similarly, the EPR

analysis of Co(II)-substituted zinc-dependent aminopeptidases

revealed essential structural information that was not obtained

by other techniques.112 We here focus on the study of the

EPR-active forms of B12 proteins.

3.3.1. Cobaltous forms of B12 proteins. B12 derivatives are

involved in unique organometallic biological reactions.113 The

B12 co-factors are essentially cobalt-containing corrin com-

plexes (Fig. 6d). Cobalt(II) corrins such as cob(II)alamine (B12r)

are often relevant intermediates in the reaction cycles of B12

enzymes, but they are only transiently formed and therefore

difficult to observe. Cob(II)alamin complexes in which the

dimethylbenzimidazole (DBI) group is coordinated to the

cobalt atom are named base-on forms, whereas, if such a

coordination is absent or replaced by an exogeneous ligand,

these complexes are referred to as base-off complexes.

For isolated B12 co-factors in solution at neutral pH and in

the solid state, the base-on form is the most stable constitu-

tion. In contrast, at low pH, a base-off form is found. This

transition between the two forms is clearly reflected in the

X-band CW-EPR spectra. In the EPR spectrum of the base-on

form, a clear three-fold splitting of the high-field lines can be

observed, due to the hyperfine interaction with the 14N nucleus

(I = 1) of the coordinating DBI ligand. This disappears in the

EPR spectrum of the base-off form, since DBI is no longer

coordinating. In a recent work, we studied the solvent effects

This journal is �c the Owner Societies 2007 Phys. Chem. Chem. Phys., 2007, 9, 4620–4638 | 4633

on base-off Co(II) corrinates.114 Previous literature addressed

Co(II) corrinates in solution often as four-coordinated Co(II)

corrins.115 Using pulsed EPR and ENDOR in combination

with 2H2O exchange experiments we could clearly prove axial

solvent ligation for Cob(II)ester (heptamethyl cobyrinate per-

chlorate) and the base-off form of cob(II)alamin (B12r) in polar

solvents.114 Fig. 9 shows the difference HYSCORE spectrum

of Cob(II)ester in CD3OD and CH3OH. The 2H cross-peaks

reveal direct information on the hyperfine coupling to the

methanol deuterons proving axial coordination of the solvent.

The additional cross-peaks stem from combination between

the 2H basic frequency and one of the DQ frequencies of the

ligand nitrogens as indicated in Fig. 9. Since only combina-

tions between nuclear frequencies within the same mS mani-

fold can occur, the appearance of these cross-peaks give direct

information on the relative sign of the 2H and 14N hyperfine

couplings.114

In addition, the axial ligation of Cob(II)ester in methanol

was shown to be influenced by the phase transition of the

solvent. Axial ligation of one methanol molecule is observed in

the a-crystalline phase, whereas six-coordination becomes

possible in the glassy state. In a methanol : H2O mixture, only

five-coordination of Cob(II)ester is observed. Addition of

water to methanol changes its phase-transition properties

and leads to frozen solutions with a large degree of crystalline

structure. Differences in the solvent ligation could be related to

changes in the network formed by the solvent molecules and

the ring substituents. Furthermore, we could show that in

apolar solvents Cob(II)ester forms a contact-ion pair with its

counter ion, perchlorate. No evidence of pure four-coordi-

nated CoII corrins in solution is found. These observations

were corroborated by DFT computations, whereby again the

power of a combined EPR-DFT approach could be proven.

In addition, the hyperfine and nuclear quadrupole para-

meters of the magnetic nuclei in the vicinity of the unpaired

electron could be determined. Combined with our earlier study

on the base-on form of cob(II)alamin,116 these results can be

used as reference data to study in detail the electronic structure

of the Co(II) forms of B12 proteins.

The initial EPR studies on B12 proteins focused on the

analysis of the cobalamin form and of possible intermediately

generated radicals using X-band CW-EPR. The base-on bind-

ing mode has been found in diol dehydratase (DD)117 and in

B12-dependent ribonucleotide reductase (RNR).118 For other

B12 proteins, the intramolecular dimethylbenzimidazole base is

replaced by the imidazole side chain of a protein histidine

residue. One such example is glutamate mutase from Closti-

dium cochlearium. The EPR spectrum of this protein shows a

typical base-on pattern with the three-fold splitting of the

high-field line.119 Upon binding of cobalamin to the 15N-

labeled protein, the three-fold splitting changed into a two-

fold splitting (15N, I = 1/2), indicating that the cobalt is

coordinating to a nitrogen of the enzyme. Furthermore, the

EPR spectrum of a sample of the 15N-labelled protein in which

only the histidines were 14N labeled, showed again the char-

acteristic three-fold splitting. This EPR study was the first

unambiguous proof that in the carbon-skeleton rearranging

mutase a histidine residue of the enzyme rather than DBI

coordinates to the cobalt center. Finally, the base-off form of

cob(II)alamin is found to occur as an intermediate state in a

number of proteins. As an example, methanol:5-hydroxy-

benzimidazolylcobamide methlytransferase (MT1) is the first

of two enzymes playing a role in the transmethylation of

methanol to 2-mercaptoethanesulfonic acid inMethanosarcina

barkeri. Binding of MT1 to the methyl group of CH3OH only

occurs when the cobalt corrin is reduced to the Co(I) state.

This reduction requires H2, hydrogenase, methyltransferase

activation protein and ATP. Using CW-EPR, it could be

shown that the as-isolated oxidized Co(III) state is reduced to

a base-on cob(II)alamin (DBI internal coordination) by addi-

tion of H2 and hydrogenase.120 Subsequent addition of

methyltransferase activation protein and ATP leads to base-

off cob(II)alamin.

At present, advanced pulsed EPR studies on B12 proteins

are still very scarce. The Co(II)-product radical pair state of

ethanolamine deaminase is the B12 protein that is by far the

most studied by pulsed-EPR methods,121 but even for this

system only X-band three-pulse ESEEM techniques and no

advanced multi-dimensional pulsed EPR methods were ap-

plied. In our opinion, a large potential to understand the

inner workings of these proteins is in this way left aside and

we share the surprise of Bennet112 that relatively few EPR

studies of Co(II)-containing or Co(II)-substituted enzymes are

reported.

3.4. Copper-containing proteins

Undoubtedly, the majority of the EPR work on metallopro-

teins is performed on copper-containing proteins. Cu(II) (d9,

S = 1/2) is an ideal EPR probe. The ion is detectable by

CW-EPR in a broad range of temperatures, making it a

perfect tool to determine the dynamics and static structure

of the protein via the study of the temperature dependence of

the g and copper hyperfine values (63Cu, I = 3/2, abundance

69.15%, 65Cu, I= 3/2, natural abundance 30.85%). Although

Fig. 9 Difference HYSCORE spectrum of cob(II)ester in CD3OD and

in CH3OH obtained by subtraction of the spectra after normalization

to the DQ cross peak of the corrin nitrogen interaction. The observer

position corresponds to g||, mI = �7/2. From the analysis of the 2H

cross peaks in addition to W-band deuterium Mims ENDOR experi-

ments the axial ligation of the solvent could be proven.114 The other

cross-peaks in the (++) quadrant originate from combination fre-

quencies between the nitrogen DQ and deuterium basic frequencies.

4634 | Phys. Chem. Chem. Phys., 2007, 9, 4620–4638 This journal is �c the Owner Societies 2007

the examples of EPR studies of copper-containing proteins are

legion, including our own work on the Cu(II)-binding of prions

and prion analogues,122–125 we will here focus only on the

study of azurin using multi-frequency EPR techniques,126–136

because we think it offers the reader a clear insight into the

possibilities of what can be obtained by these advanced EPR

techniques.

3.4.1. Unravelling the structure of azurin. Although a quick

scan through the EPR/ENDOR literature shows that the

majority of the studies on copper proteins are performed at

X-band mw frequencies, the use of high-field EPR techniques

has great potential. This is exemplified by the study of wild-

type azurin from Pseudomonas aeruginosa and some of its

mutants. The Cu(II)-binding site in azurin is a typical type-I

copper centre characterized by the strong ligation of two

histidines and a cysteine and a weak ligation of commonly a

methionine residue (Fig. 6e). In order to understand the

electron-accepting and electron-donating processes occurring

in this blue-copper protein, the electronic structure of the

copper sites should be mapped out. EPR is one of the

techniques of choice. However, at X-band frequencies, it is

impossible to accurately determine the principal g values. This

high resolution can be obtained when using 95 GHz EPR

spectroscopy. The EPR lines in the W-band ESE-detected

EPR spectra of single crystals of wild-type azurin could be

related to the 16 protein molecules in the unit cell.126 Similar

analyses for the M121Q and M121H mutants of azurin

revealed clear spectral changes that could be assigned to

admixture of dyz (M121H) or dz2(M121Q) character into the

wave function.127,128 W-Band nitrogen ENDOR and X-band

ESEEM studies allowed for a complete determination of the

hyperfine and nuclear quadrupole tensors of the remote

nitrogens Ne of the histidine ligands.129–133 In a seminal work,

Groenen and co-workers detected for the first time single-

crystal W-band ESEEM spectra arising from the directly

coordinated nitrogens of the histidines that ligate copper in

azurin.134 The set of experimental data was completed by a

recent W-band deuterium ENDOR study of single crystals of

azurin selectively deuterated at the Cb position of the copper-

binding cysteine-112.135 The magnitude of the isotropic hy-

perfine coupling of the b deuterons can directly be related to

the spin density on the cysteine sulfur. All these EPR studies

have not only provided a direct insight into aspects of the

electronic structure of the copper site in azurin, they also form

a critical and necessary test for the different quantum-chemical

studies performed on azurin.136–138 Only a good qualitative

agreement between theory and experiment gives confidence in

the quantum-chemical approach and in the deduced conclu-

sions on the inner working of the protein.

3.5. Other metal-containing proteins

The above paragraphs highlighted the type of metal-contain-

ing proteins we are familiar with from our own research.

However, the list of paramagnetic ions reported to bind in

vivo to proteins is much larger (Mn(II), Mo(V), ...) and so are

the type of EPR studies performed on them. For a detailed

overview of these systems, we refer to a number of recent

reviews.15,84,89,139,140

4. Conclusions and challenges for the future

Recent methodological advances in EPR, amongst which are

the development of different microwave pulse sequences and

the construction of high-performance CW and pulsed EPR

and ENDOR spectrometers operating at high mw frequencies,

have largely increased the potential of applying these techni-

ques to different problems. However, EPR remains a specia-

list’s tool and because of this, the technique is still relatively

unknown to other scientific communities despite the recent

booming of the field. In this overview we have tried to give the

non-specialist reader an insight into the advantages and dis-

advantages of using EPR to study metalloproteins. It should

be noted, however, that the application domains of EPR reach

much further than the field of metalloproteins, comprising

applications in biology, chemistry, material sciences, physics,

medicine and pharmacology.

In our opinion, the most important challenges for the future

include instrumental aspects, such as improvement of sensi-

tivity and spectral resolution through new spectrometer design

at high mw frequencies, the further development and automa-

tion of adequate simulation programs, and the improvement

of the quantum chemical strategies for computation of EPR

parameters (see section 2.3). The ultimate goal is to create an

integrated toolbox that includes the spectral derivation of the

different spin Hamiltonian parameters up to the interpretation

of these magnetic interaction parameters in term of (geometric

and electronic) structure and dynamics of the complex under

study.

Acknowledgements

The majority of the own work cited in this manuscript stems

from the corresponding author’s research work from the last

10 years done at the ETH Zurich (1996–2001) and at the

University of Antwerp (2002 till now). Many students, post-

doctoral researchers and co-workers have contributed or are

contributing to the research on metalloproteins. Many thanks

go to all of them. The high-level technical support of Jorg

Forrer (ETH Zurich) and Paul Casteels (Antwerp) is highly

appreciated and was essential for many experiments. Special

thanks and thoughts go to the recently deceased Arthur

Schweiger, who was the corresponding author’s mentor in

Zurich and initiated and trained her in pulsed EPR. In general,

Arthur Schweiger has played a key role in the development of

advanced pulsed EPR techniques. He is missed as a great

scientist and a good friend.

In addition, S.V.D. has been inspired for the present article

through the many contacts and collaborations with fellow

EPR spectroscopists throughout the world, not in the least

within the COST P15 action ‘Advanced Paramagnetic Reso-

nance Methods in Molecular Biophysics’, chaired by the

author. Furthermore, S.V.D. wants to thank the many (bio)-

chemists she worked with in the course of her work on

metalloproteins. S.V.D. gratefully acknowledges the

This journal is �c the Owner Societies 2007 Phys. Chem. Chem. Phys., 2007, 9, 4620–4638 | 4635

University of Antwerp and the Fund for Scientific Research-

Flanders (FWO) for funding. E.V. is a research assistant of the

FWO.

References

1 E. Zavoisky, J. Phys. USSR, 1945, 9, 211.2 G. Feher, Phys. Rev., 1956, 103, 834.3 L. H. Sutcliffe, Phys. Med. Biol., 1998, 43, 1987.4 W. L. Hubbell, A. Gross, R. Langen and M. A. Lietzow, Curr.Opin. Struct. Biol., 1998, 8, 649.

5 W. B. Mims, Phys. Rev., 1964, 133, A835.6 W. B. Mims, Proc. R. Soc. London, Ser. A, 1965, 283, 452.7 G. M. Zhidomirov, K. M. Salhikov, Y. D. Tsvetkov, V. F.Yudanov and A. M. Raimtsimring, J. Struct. Chem., 1968, 9,704.

8 Ya. S. Lebedev, in Modern Pulsed and Continuous-Wave ElectronSpin Resonance, ed. L. Kevan and M. K. Bowman, Wiley, NewYork, 1990, p. 365.

9 A. Schweiger and G. Jescke, Principles of Pulse Electron Para-magnetic Resonance, University Press Oxford, Oxford, 2001.

10 J. H. Freed, Annu. Rev. Phys. Chem., 2000, 51, 655.11 B. M. Hoffman, Acc. Chem. Res., 2003, 36, 522.12 D. Goldfarb and D. Arieli, Annu. Rev. Phys. Chem., 2004, 33, 441.13 G. Jeschke, Curr. Opin. Solid State Mater. Sci., 2003, 7, 181.14 T. Prisner, M. Rohrer and F. MacMillan, Annu. Rev. Phys.

Chem., 2001, 52, 279.15 M. Ubbink, J. A. R. Worrall, G. W. Canters, E. J. J. Groenen and

M. Huber, Annu. Rev. Biophys. Biomol. Struct., 2002, 31, 393.16 M. Bennati and T. Prisner, Rep. Prog. Phys., 2005, 68, 411.17 W. Lubitz, Pure Appl. Chem., 2003, 75, 1021.18 K. Mobius, A. Savitsky, A. Schnegg, M. Plato and M. Fuchs,

Phys. Chem. Chem. Phys., 2005, 7, 19.19 D. Goldfarb, Phys. Chem. Chem. Phys., 2006, 8, 2325.20 A. Abragam and B. Bleaney, Electron Paramagnetic Resonance of

Transition Ions, Clarendon, Oxford, 1970.21 J. A. Weil, J. R. Bolton and J. E. Wertz, Electron Paramagnetic

Resonance, Wiley, New York, 1994.22 J. R. Pilbrow, Transition Ion Electron Paramagnetic Resonance,

Clarendon, Oxford, 1990.23 N. M. Atherton, Principles of Electron Spin Resonance, Ellis

Horwood, New York, 1993.24 G. C. Hurst, T. A. Henderson and R. W. Kreilick, J. Am. Chem.

Soc., 1985, 107, 7294.25 E. R. Davies, Phys. Lett., 1974, A47, 1.26 G. Jeschke and A. Schweiger, Chem. Phys. Lett., 1995, 246,

431.27 H. Kurreck, B. Kirste and W. Lubitz, Electron Nuclear Double

Resonance Spectroscopy of Radicals in Solution, VCH Publishers,New York, 1988.

28 H. Thomann and M. Bernardo, Methods Enzymol., 1993, 227,118.

29 C. Gemperle and A. Schweiger, Chem. Rev., 1991, 91, 1481.30 S. A. Dikanov and Yu. D. Tsvetkov, Electron Spin Echo Envelope

Modulation (ESEEM) Spectroscopy, CRC Press, Boca Raton,FL, 1992.

31 P. Hofer, A. Grupp, H. Nebenfuhr and M. Mehring, Chem. Phys.Lett., 1986, 132, 279.

32 A. Dikanov, Yu. D. Tsvetkov, M. K. Bowman and A. V.Astashkin, Chem. Phys. Lett., 1982, 90, 149.

33 J. J. Shane, Electron Spin Echo Envelope Modulation Spectroscopyof Disordered Solids, PhD thesis Nijmegen., 1993.

34 H. L. Flanagan and D. J. Singel, J. Chem. Phys., 1987, 87,5606.

35 A. Poppl and R. Bottcher, Chem. Phys., 1997, 221, 53.36 G. Jeschke, R. Rakhmatullin and A. Schweiger, J. Magn. Reson.,

1998, 131, 261.37 L. Liesum and A. Schweiger, J. Chem. Phys., 2001, 114, 9478.38 S. Stoll, C. Calle, G. Mitrikas and A. Schweiger, J. Magn. Reson.,

2005, 177, 93.39 P. Schosseler, Th. Wacker and A. Schweiger, Chem. Phys. Lett.,

1994, 224, 319.40 G. Jeschke and H. W. Spiess, Chem. Phys. Lett., 1998, 293, 9.

41 Z. L. Madi, S. Van Doorslaer and A. Schweiger, J. Magn. Reson.,2002, 154, 181.

42 R. Szosenfogel and D. Goldfarb, Mol. Phys., 1998, 95, 1295.43 A. Adamski, T. Spalek and Z. Sojka, Res. Chem. Intermed., 2003,

29, 793.44 M. J. Mombourquette and J. A. Weil, J. Magn. Reson., 1992, 99,

37.45 G. R. Hanson, K. E. Gates, C. J. Noble, M. Griffin, A. Mitchell

and S. Benson, J. Inorg. Biochem., 2004, 98, 903.46 C. J. Jacobsen, E. Pedersen, J. Villadsen and H. Weihe, Inorg.

Chem., 1993, 32, 1216.47 C. P. Keijzers, E. J. Reijerse, P. stam, M. F. Dumont and M. C.

M. Gribnau, J. Chem. Soc., Faraday Trans. 1, 1987, 83,3493.

48 S. Stoll and A. Schweiger, J. Magn. Reson., 2006, 178, 42.49 Calculation of NMR and EPR parameters, ed. M. Kaupp, M.

Buhl and V. G. Malkin, Wiley-VCH, Weinheim, 2004.50 F. Neese and E. I. Solomon, inMagnetism: molecules to materials

IV, ed. J. S. Miller and M. Drillon, Wiley-VCH, Weinheim, 2003,p. 345.

51 J. Harmer, S. Van Doorslaer, I. Gromov, M. Broring, G. Jeschkeand A. Schweiger, J. Phys. Chem. B, 2002, 106, 2801.

52 S. Kababya, J. Nelson, C. Calle, F. Neese and D. Goldfarb,J. Am. Chem. Soc., 2006, 128, 2017.

53 A. Fragoso, M. C. Baratto, A. Diaz, Y. Rodriguez, R. Pogni, R.Basosi and R. Cao, Dalton Trans., 2004, 1456.

54 V. Noethig-Laslo, N. Paulic, R. Basosi and R. Pogni, Polyhedron,2002, 21, 1643.

55 M. Brunori, Trends Biochem. Sci., 2001, 26, 209.56 S. N. Vinogradov, D. Hoogewijs, X. Bailly, R. Arredondo-Peter,

J. Gough, S. Dewilde, L. Moens and J. R. Vanfleteren, BMCEvolutionary Biology, 2006, 6, 31.

57 A. Pesce, M. Bolognesi, A. Bocedi, P. Ascenzi, S. Dewilde, L.Moens, T. Hankeln and T. Burmester, EMBO Rep., 2002, 3,1146.

58 E. Vinck, S. Van Doorslaer, S. Dewilde and L. Moens, J. Am.Chem. Soc., 2004, 126, 4516.

59 (a) W. E. Blumberg and J. Peisach, in Bioinorganic Chemistry, ed.R. F. Gould, American Chemical Society Publishers, Washing-ton, DC, 1971, p. 271; (b) W. E. Blumberg and J. Peisach, inProbes of Structure and Function of Macromolecules and Mem-branes, II. Probes of Enzymes and Hemoproteins, ed. B. Chance,T. Yonetani and A. S. Mildvan, Academic Press, New York,1971, p. 215.

60 S. V. Nistor, E. Goovaerts, S. Van Doorslaer, S. Dewilde and L.Moens, Chem. Phys. Lett., 2002, 361, 355.

61 A. Pesce, S. Dewilde, M. Nardini, L. Moens, P. Ascenzi, T.Hankeln, T. Burmester and M. Bolognesi, Structure, 2003, 11,1087.

62 B. Vallone, K. Nienhaus, M. Brunori and G. U. Nienhous,Proteins: Struct., Funct., Bioinf., 2004, 56, 85.

63 C. P. Scholes, K. M. Falkowski, S. Chen and J. Bank, J. Am.Chem. Soc., 1986, 108, 1660.

64 M. Fahnenschmidt, R. Bittl, H. K. Rau, W. Haehnel and W.Lubitz, Chem. Phys. Lett., 2000, 323, 329.

65 R. S. Magliozzo and J. Peisach, Biochemistry, 1992, 31, 189.66 R. S. Magliozzo and J. Peisach, Biochemistry, 1993, 32, 8446.67 Y. C. Fann, N. C. Gerber, P. A. Osmulski, L. P. Hager, S. G.

Sligar and B. M. Hoffman, J. Am. Chem. Soc., 2994, 116, 5989.68 H. I. Lee, A. F. Dexter, Y. C. Fann, F. J. Lakner, L. P. Hager and

B. M. Hoffman, J. Am. Chem. Soc., 1997, 119, 4059.69 A. M. Raitsimring and F. A. Walker, J. Am. Chem. Soc., 1998,

120, 991.70 A. M. Raitsimring, P. Borbat, T. Kh. Shokhireva and F. A.

Walker, J. Phys. Chem., 1996, 100, 5235.71 A. V. Astashkin, A. M. Raitsimring and F. A. Walker, Chem.

Phys. Lett., 1999, 306, 9.72 I. Garcıa-Rubio, J. I. Martınez, R. Picorel, I. Yruela and P. J.

Alonso, J. Am. Chem. Soc., 2003, 125, 15846.73 E. Vinck and S. Van Doorslaer, Phys. Chem. Chem. Phys., 2004,

6, 5324.74 E. Vinck, S. Van Doorslaer, S. Dewilde, G. Mitrikas, A. Schwei-

ger and L. Moens, JBIC, J. Biol. Inorg. Chem., 2006, 11, 467.75 A. I. Ioanitescu, S. Dewilde, L. Kiger, M. C. Marden, L. Moens

and S. Van Doorslaer, Biophys. J., 2005, 89, 2628.

4636 | Phys. Chem. Chem. Phys., 2007, 9, 4620–4638 This journal is �c the Owner Societies 2007

76 D. Hamdane, L. Kiger, S. Dewilde, B. Green, A. Pesce, J. Uzan,T. Burmester, T. Hankeln, M. Bolognesi, L. Moens and M. C.Marden, J. Biol. Chem., 2004, 278, 51713.

77 F. Trandafir, P. ter Heerdt, M. Fittipaldi, E. Vinck, S. Dewilde, L.Moens and S. Van Doorslaer, Appl. Magn. Reson., 2006, 31,553.

78 P. J. M. van Kan, E. van der Horst, E. J. Reijerse, P. J. M. vanBentum and W. R. Hagen, J. Chem. Soc., Faraday Trans., 1998,94, 2975.

79 C. P. Scholes, A. Lapidot, R. Mascarenhas, T. Inubushi, R. A.Isaacson and G. Feher, J. Am. Chem. Soc., 1982, 104, 2724.

80 (a) Y.-C. Fann, J.-L. Ong, J. M. Nocek and B. M. Hoffman, J.Am. Chem. Soc., 1995, 117, 6109; (b) H.-I. Lee, Bull. KoreanChem. Soc., 2002, 23, 1769; (c) R. Bachmann, A. Schweiger, H.Aissaoui and W. D. Woggon, in Magnetic Resonance and RelatedPhenomena, Proceedings of the Joint 29th AMPERE—13thISMAR International Conference, Berlin, TU Berlin, 1998, vo-lume II; (d) H. Aissaoui, R. Bachmann, A. Schweiger and W. D.Woggon, Angew. Chem., Int. Ed., 1998, 37, 2998.

81 P. Bertrand, B. Guigliarelli, J.-P. Gayda, P. Beardwood and J. F.Gibson, Biochim. Biophys. Acta, 1985, 831, 261.

82 R. I. Samoilova, D. Kolling, T. Uzawa, T. Iwasaki, A. R. Croftsand S. A. Dikanov, J. Biol. Chem., 2002, 277, 4605.

83 S. A. Dikanov, A. A. Shubin, A. Kounosu, T. Iwasaki and R. I.Samoilova, JBIC, J. Biol. Inorg. Chem., 2004, 9, 753.

84 Y. Deligiannakis, M. Louloudi and N. Hadjiliadis, Coord. Chem.Rev., 2000, 204, 1.

85 R. Cammack, E. Gay and J. K. Shergill, Coord. Chem. Rev., 1999,190–192, 1003.

86 R. K. Thauer, Microbiology, 1998, 144, 2377.87 (a) F. Mahlert, W. Grabarse, J. Kahnt, R. K. Thauer and E. C.

Duin, JBIC, J. Biol. Inorg. Chem., 2002, 7, 101; (b) F. Mahlert, C.Bauer, B. Jaun, R. K. Thauer and E. C. Duin, JBIC, J. Biol.Inorg. Chem., 2002, 7, 500.

88 J. Telser, Y.-C. Fann, M. W. Renner, J. Fajer, S. Wang, H.Zhang, R. A. Scott and B. M. Hoffman, J. Am. Chem. Soc., 1997,119, 733.

89 J. Telser, Y. C. Horng, D. F. Becker, B. M. Hoffman and S. W.Ragsdale, J. Am. Chem. Soc., 2000, 122, 182.

90 C. Finazzo, J. Harmer, B. Jaun, E. C. Duin, F. Mahlert, R. K.Thauer, S. Van Doorslaer and A. Schweiger, JBIC, J. Biol. Inorg.Chem., 2003, 8, 586.

91 C. Finazzo, J. Harmer, C. Bauer, B. Jaun, E. C. Duin, F. Mahlert,M. Groenich, R. T. Thauer, S. Van Doorslaer and A. Schweiger,J. Am. Chem. Soc., 2003, 125, 4988.

92 J. Telser, R. Davydov, Y.-C. Horng, S. W. Ragsdale and B. M.Hoffman, J. Am. Chem. Soc., 2001, 123, 5853.

93 T. Wondimagegen and A. Ghosh, J. Am. Chem. Soc., 2001, 123,1543.

94 J. Harmer, C. Finazzo, R. Piskorski, C. Bauer, B. Jaun, E. C.Duin, M. Goenrich, R. K. Thauer, S. Van Doorslaer and A.Schweiger, J. Am. Chem. Soc., 2005, 127, 17744.

95 M. Goubeaud, G. Schreiner and R. K. Thauer, Eur. J. Biochem.,1997, 243, 110.

96 (a) Q. Tang, P. E. Carrington, Y. C. Horng, M. Maroney, S. W.Ragsdale and D. F. Bocian, J. Am. Chem. Soc., 2002, 124, 13242;(b) E. C. Duin, N. J. Cosper, F. Mahlert, R. K. Thauer and R. A.Scott, JBIC, J. Biol. Inorg. Chem., 2003, 8, 141.

97 B. Jaun, Helv. Chim. Acta, 1990, 73, 2209.98 E. C. Duin, L. Signor, R. Piskorski, F. Mahlert, M. D. Clay, M.

Groenich, R. K. Thauer, B. Jaun and M. K. Johnson, JBIC,J. Biol. Inorg. Chem., 2004, 9, 563.

99 W. Grabarse, F. Mahlert, E. C. Duin, M. Goubeaud, S. Shima,R. K. Thauer, V. Lamzin and U. Ermler, J. Mol. Biol., 2001, 309,315.

100 Hydrogen as a Fuel Learning from Nature, ed. R. Cammack, M.Frey and R. Robson, Taylor and Francis, London, 2001.

101 (a) F. A. Armstrong, Curr. Opin. Chem. Biol., 2004, 8, 133; (b) W.Lubitz, M. Brecht, S. Foerster, M. van Gastel and M. Stein, ACSSymp. Ser., 2003, 858, 128.

102 V. M. Fernandez, E. C. Hatchikian, D. S. Patil and R. Cammack,Biochim. Biophys. Acta, 1986, 883, 145.

103 M. Carepo, D. L. Tierney, C. D. Brondino, T. C. Yang, A.Pamplona, J. Telser, I. Moura, J. J. G. Moura and B. M.Hoffman, J. Am. Chem. Soc., 2002, 124, 281.

104 M. van Gastel, C. Fichtner, F. Neese and W. Lubitz, Biochem.Soc. Trans., 2005, 33, 7.

105 M. van Gastel, S. Matthias, M. Brecht, O. Schroder, F. Lendzian,R. Bittl, H. Ogata, Y. Higuchi and W. Lubitz, JBIC, J. Biol.Inorg. Chem., 2006, 11, 41.

106 M. Brecht, M. van Gastel, T. Buhrke, B. Friedrich andW. Lubitz,J. Am. Chem. Soc., 2003, 125, 13075.

107 S. Foerster, M. van Gastel, M. Brecht and W. Lubitz, JBIC,J. Biol. Inorg. Chem., 2005, 10, 51.

108 (a) M. Stein and W. Lubitz, J. Inorg. Biochem., 2994, 98, 862; (b)M. Stein and W. Lubitz, Curr. Opin. Chem. Biol., 2002, 6, 243.

109 B. Hoffman and D. H. Petering, Proc. Natl. Acad. Sci. U. S. A.,1970, 67, 637.

110 L. C. Dickinson and J. C. W. Chien, Proc. Natl. Acad. Sci. U. S.A., 1980, 77, 1235.

111 S. Van Doorslaer and A. Schweiger, J. Phys. Chem. B, 2000, 104,2919.

112 B. Bennet, Curr. Topics Biophys., 2002, 26, 49.113 R. Banjeree and S. Ragsdale, Annu. Rev. Biochem., 2003, 72,

209.114 S. Van Doorslaer, G. Jeschke, B. Epel, D. Goldfarb, R.-A. Eichel,

B. Krautler and A. Schweiger, J. Am. Chem. Soc., 2003, 125,5915.

115 (a) M. D. Wirt, M. Kumar, S. W. Ragsdale and M. R. Chance, J.Am. Chem. Soc., 1993, 115, 2146; (b) Vitamin B12 and B12-Proteins, ed. B. Krautler, D. Arigoni and B. T. Golding, Wiley-VCH, Weinheim, 1998.

116 J. Harmer, S. Van Doorslaer, I. Gromov and A. Schweiger,Chem. Phys. Lett., 2002, 358, 8.

117 M. Yamanishi, S. Yamada, H. Muguruma, Y. Murakami, T.Tobimatsu, A. Ishida, J. Yamauchi and T. Toraya, Biochemistry,1998, 37, 4799.

118 C. C. Lawrence, G. J. Gerfen, V. Samano, R. Nitsche, M. J.Robins, J. Retey and J. Stubbe, J. Biol. Chem., 1999, 274, 7039.

119 O. Zelder, B. Beatrix, F. Kroll and W. Buckel, FEBS Lett., 1995,369, 252.

120 P. J. H. Daas, W. R. Hagen, J. T. Keltjens, C. van der Drift andG. D. Vogels, J. Biol. Chem., 1996, 271, 22346.

121 (a) K. Warncke, Biochemistry, 2005, 44, 3184; (b) K. Warnckeand A. S. Utada, J. Am. Chem. Soc., 2001, 123, 8564; (c) K.Warncke, J. C. Schmidt and S.-C. Ke, J. Am. Chem. Soc., 1999,121, 10522; (d) S.-C. Ke and K. Warncke, J. Am. Chem. Soc.,1999, 121, 9922.

122 S. Van Doorslaer, G. Cereghetti, R. Glockshuber and A. Schwei-ger, J. Phys. Chem. B, 2001, 105, 1631.

123 G. Cereghetti, A. Schweiger, R. Glockshuber and S. Van Door-slaer, Biophys. J., 2001, 81, 516.

124 G. Cereghetti, A. Schweiger, R. Glockshuber and S. Van Door-slaer, Biophys. J., 2003, 84, 1985.

125 G. M. Cereghetti, A. Negro, E. Vinck, M. L. Massimino, M. C.Sorgato and S. Van Doorslaer, J. Biol. Chem., 2004, 279, 36497.

126 J. W. A. Coremans, O. G. Poluektov, E. J. J. Groenen, G. W.Canters, H. Nar and A. Messerschmidt, J. Am. Chem. Soc., 1994,116, 3097.

127 J. W. A. Coremans, O. G. Poluektov, E. J. J. Groenen, G. C. M.Warmerdam, G. W. Canters, H. Nar and A. Messerschmidt, J.Phys. Chem., 1996, 100, 19706.

128 M. van Gastel, G. W. Canters, H. Krupka, A. Messerschmidt, E.C. de Waal, G. C. M. Warmerdam and E. J. J. Groenen, J. Am.Chem. Soc., 2000, 122, 2322.

129 J. W. A. Coremans, M. van Gastel, O. G. Poluektov, E. J. J.Groenen, T. den Blaauwen, G. van Pouderoyen, G. W. Canters,H. Nar, C. Hammann and A. Messerschmidt, Chem. Phys. Lett.,1995, 235, 202.

130 J. W. A. Coremans, O. G. Poluektov, E. J. J. Groenen, G. W.Canters, H. Nar and A. Messerschmidt, J. Am. Chem. Soc., 1996,118, 12141.

131 V. Kofman, O. Farver, I. Pecht and D. Goldfarb, J. Am.Chem.Soc., 1996, 118, 1201.

132 M. van Gastel, J. W. A. Coremans, L. J. C. Jeuken, G. W.Canters and E. J. J. Groenen, J. Phys. Chem. A, 1998, 102,4462.

133 M. van Gastel, J. W. A. Coremans, J. Mol, L. J. C. Jeuken, G. W.Canters and E. J. J. Groenen, JBIC, J. Biol. Inorg. Chem., 1999, 4,257.

This journal is �c the Owner Societies 2007 Phys. Chem. Chem. Phys., 2007, 9, 4620–4638 | 4637

134 J. W. A. Coremans, O. G. Poluektov, E. J. J. Groenen, G. W.Canters, H. Nar and A. Messerschmidt, J. Am. Chem. Soc., 1997,119, 4726.

135 M. Fittipalid, G. C. M. Warmerdam, E. C. de Waal, G. W.Canters, D. Cavazzini, G. L. Rossi, M. Huber and E. J. J.Groenen, ChemPhysChem, 2006, 7, 1286.

136 M. van Gastel, J. W. A. Coremans, H. Sommerdijk, M. C. vanHemert and E. J. J. Groenen, J. Am. Chem. Soc., 2002, 124,2035.

137 A. R. Jaszewski and J. Jezierska, Chem. Phys. Lett., 2001, 343,571.

138 S. Moons, S. Patchkovskii and D. R. Salahub, J. Mol. Struct.(THEOCHEM), 2003, 632, 287.

139 J. M. Peloquin and R. D. Britt, Biochim. Biophys. Acta, 2001,1503, 96.

140 R. D. Britt, K. A. Campbell, J. M. Peloquin, M. L. Gilchrist, C.P. Aznar, M. M. Dicus, J. Robblee and J. Messinger, Biochim.Biophys. Acta, 2004, 1655, 158.

4638 | Phys. Chem. Chem. Phys., 2007, 9, 4620–4638 This journal is �c the Owner Societies 2007