Asymmetric binding of the high-affinity Q(H)(*)(-) ubisemiquinone in quinol oxidase (bo3) from Escherichia coli studied by multifrequency electron paramagnetic resonance spectroscopy.

Abstract

Ubiquinone-2 (UQ-2) selectively labeled with (13)C (I =(1)/(2)) at either the position 1- or the 4-carbonyl carbon is incorporated into the ubiquinol oxidase bo(3) from Escherichia coli in which the native quinone (UQ-8) has been previously removed. The resulting stabilized anion radical in the high-affinity quinone-binding site (Q(H)(*)(-)) is investigated using multifrequency (9, 34, and 94 GHz) electron paramagnetic resonance (EPR) spectroscopy. The corresponding spectra reveal dramatic differences in (13)C hyperfine couplings indicating a strongly asymmetric spin density distribution over the quinone headgroup. By comparison with previous results on labeled ubisemiquinones in proteins as well as in organic solvents, it is concluded that Q(H)(*)(-) is most probably bound to the protein via a one-sided hydrogen bond or a strongly asymmetric hydrogen-bonding network. This observation is discussed with regard to the function of Q(H) in the enzyme and contrasted with the information available on other protein-bound semiquinone radicals.

Full text

Page 1

Asymmetric Binding of the High-Affinity QH•-Ubisemiquinone in Quinol Oxidase
(bo3) from Escherichia coli Studied by Multifrequency Electron Paramagnetic
Resonance Spectroscopy†
S. Grimaldi,‡T. Ostermann,§,|N. Weiden,⊥T. Mogi,#,@H. Miyoshi,∇B. Ludwig,|H. Michel,§T. F. Prisner,‡and
F. MacMillan*,‡
Institut fu ¨r Physikalische und Theoretische Chemie and Institut fu ¨r Biochemie, J. W. Goethe UniVersita ¨t Frankfurt, D-60439
Frankfurt am Main, Germany, Max-Planck Institut fu ¨r Biophysik, D-60528 Frankfurt am Main, Germany, Department of
Biological Sciences, Graduate School of Science, UniVersity of Tokyo, Hongo, Bunkyo-ku, Tokyo 113-0033, Japan, ATP System
Project, Exploratory Research for AdVanced Technology (ERATO), Japan Science and Technology Corporation (JST),
Nagatsuta, Midori-ku, Yokohama 226-0026, Japan, DiVision of Applied Life Sciences, Kyoto UniVersity, Kyoto 606-8502, Japan,
and Institut fu ¨r Physikalische Chemie, Technische UniVersita ¨t Darmstadt, D-64287 Darmstadt, Germany
ReceiVed January 6, 2003; ReVised Manuscript ReceiVed March 7, 2003
ABSTRACT: Ubiquinone-2 (UQ-2) selectively labeled with13C (I )1/2) at either the position 1- or the
4-carbonyl carbon is incorporated into the ubiquinol oxidase bo3from Escherichia coli in which the native
quinone (UQ-8) has been previously removed. The resulting stabilized anion radical in the high-affinity
quinone-binding site (QH•-) is investigated using multifrequency (9, 34, and 94 GHz) electron paramagnetic
resonance (EPR) spectroscopy. The corresponding spectra reveal dramatic differences in13C hyperfine
couplings indicating a strongly asymmetric spin density distribution over the quinone headgroup. By
comparison with previous results on labeled ubisemiquinones in proteins as well as in organic solvents,
it is concluded that QH•-is most probably bound to the protein via a one-sided hydrogen bond or a
strongly asymmetric hydrogen-bonding network. This observation is discussed with regard to the function
of QHin the enzyme and contrasted with the information available on other protein-bound semiquinone
radicals.
Quinone molecules are involved in many of the oxidation-
reduction processes of respiration and photosynthesis in
living cells. Because of their long hydrophobic tail they are
able to freely diffuse in membranes (e.g., quinone pools) or
are able to bind to membrane proteins in well-structured
binding pockets where they act as one- or two-electron gates
and are thus able to couple electron transfer to proton
translocation across the membrane (1). The fully oxidized
quinone form is commonly found in these binding sites,
although sometimes they can be observed in the one-electron
reduced semiquinone form (Q•-), while the two-electron
reduced quinol form usually diffuses away from the binding
pocket. It has been shown that structurally identical quinones
can have very different chemical properties when bound in
different sites within a protein (2). This is probably because
of specific interactions of the quinone molecule with its
immediate protein environment, and which is believed to fine
tune the electronic structure of the quinone for optimum
function. These specific interactions are suggested to involve
hydrogen bonds from the two carbonyl groups of the quinone
to the protein, but may also involve π interactions with the
aromatic ring and the hydrophobic isoprenyl side-chain.
The stable secondary electron acceptors, QAand QB,1of
the purple bacterial photosynthetic reaction center (bRC) are
the most well-studied and best-characterized quinone binding
sites in membrane proteins (2, 3). QA and QB are mainly
ubiquinone derivatives in bacteria but may also be naphtho-
quinones and are plastoquinones in higher plants. Although
QA and QB are of similar or even identical molecular
structure, QAacts only as a one-electron acceptor, while QB
†This work was supported by the Deutsche Forschungsgemeinschaft
(Sfb 472, P15) and the Hermann-Willkomm Stiftung (to F.M.).
* Corresponding author. Fax: (+49 69) 79829404. Telephone: (+49
69) 79829593. E-mail: macmillan@chemie.uni-frankfurt.de.
‡Institut fu ¨r Physikalische und Theoretische Chemie, J. W. Goethe
Universita ¨t Frankfurt and Centre for Biological Magnetic Resonance.
§Max-Planck Institut fu ¨r Biophysik.
|Institut fu ¨r Biochemie, J. W. Goethe Universita ¨t Frankfurt.
⊥Technische Universita ¨t Darmstadt.
#Tokyo Institute of Technology.
∇Kyoto University.
@Japan Science and Technology Corporation.
1Abbreviations: A1, secondary electron acceptor in type-I reaction
centers; aiso, isotropic hyperfine coupling; CIDEP, chemically induced
dynamic electron polarisation; COX, cytochrome c oxidase; CP-MAS
NMR, cross polarization magic angle spinning nuclear magnetic
resonance; Cw, continuous wave; ?-DM, n-dodecyl-?-D-maltoside;
DME, 1,2-dimethoxyethane; DPPH, (R,R′)-diphenyl-?-picrylhydrazyl;
EPR, electron paramagnetic resonance spectroscopy; ENDOR, electron
nuclear double resonance spectroscopy; ESEEM, electron spin-echo
envelope modulation spectroscopy; FTIR, Fourier transform infrared
spectroscopy; G, Gauss; hf, hyperfine; HYSCORE, hyperfine sublevel
correlation spectroscopy; LDAO, N,N-dimethyldodecylamine N-oxide;
mTHF, 2-methyltetrahydrofuran; PSI, photosystem I; PSII, photosystem
II; QA, QB, primary and secondary stable electron acceptors in type II
reaction centres; QH, QL, high- and low-affinity ubiquinone-binding
sites of quinol oxidase; Q0, Qi, quinones of the cytochrome bc1complex;
QOX, quinol oxidase; bRC, bacterial reaction centre; UQ-n, ubiquinone,
n is the number of isoprenyl units.
5632
Biochemistry 2003, 42, 5632-5639
10.1021/bi034010z CCC: $25.00 © 2003 American Chemical Society
Published on Web 04/24/2003

Page 2

accepts two electrons and two protons and diffuses out of
its binding site (2).
A variety of spectroscopic techniques (e.g., FTIR (4),
RAMAN (5), and CP-MAS NMR (6)) have been used to
investigate the function and interaction of such quinones
within their binding sites. Electron paramagnetic resonance
(EPR) spectroscopy is also a very suitable tool for studying
the radical anions of such molecules and has also been
applied extensively to QA•-(for a recent review, see ref 3).
The respiratory chain is another rich source of quinones.
Nearly all the membrane proteins in the respiratory chain
are known to require quinones to catalytically function. The
bo3-type quinol oxidase (QOX) is a transmembrane protein
involved in the aerobic respiratory chain of the Gram-
negative bacterium Escherichia coli and is closely related
to the mitochondrial cytochrome c oxidase (COX) in many
aspects of its structure and function (for a recent review,
see ref 7). Indeed, it acts as a redox-driven proton pump
that couples the vectorial translocation of protons across the
membrane to the reduction of molecular oxygen to water.
The major difference between both proteins, however, is the
nature of the electron donating substrate: cytochrome c
oxidases typically use the water-soluble protein cytochrome
c as the electron donor, whereas quinol oxidase uses a
membrane soluble ubiquinol molecule (in E. coli UQ-8,
Figure 1A). QOX has been proposed to have up to two
ubiquinone binding-sites, one with high (QH) and one with
low (QL) affinity for ubiquinone (8, 9), whereby electrons
are transferred from the QLsite to the next electron acceptor
(heme b) via the QHsite (9).
A crystal structure model of QOX was recently published
at a resolution of 3.5 Å, which indicated that the overall
structure of this complex is quite similar to that of cyto-
chrome c oxidase (10). The crystals used for this structural
model, however, did not contain bound quinones; thus, their
exact location is still unresolved. A functional study of site-
directed mutants was used to propose a model for the possible
QHbinding site, which is located in subunit I close to heme
b (10). In this model, QH is predicted to form up to four
hydrogen bonds with the protein: Asp75 and Arg71 to the
1-carbonyl oxygen and His98 and Gln101 to the 4-carbonyl
oxygen (see Figure 1B).
EPR spectroscopy is a valuable tool for investigating
semiquinone binding in proteins as has been demonstrated
previously (e.g., ref 3). X-band cw EPR has demonstrated
that the QHsite stabilizes a semiquinone radical anion (11,
12). Direct evidence of a specific interaction of QH•-with
the immediate protein environment has been obtained using
pulsed EPR spectroscopy (13). The pulsed EPR technique,
electron spin-echo envelope modulation spectroscopy (ES-
EEM) revealed a strong, specific interaction of the unpaired
electron of QH•-with a peptide nitrogen nucleus that is
consistent with the presence of a hydrogen bond between
QH•-and this nitrogen. Furthermore, based on the magnitude
of the proton hyperfine coupling assigned to the methyl group
at position 5 (Figure 1A) (14), it was suggested that this
hydrogen bond is formed to the 1-carbonyl oxygen (13, 14).
Cw electron nuclear double resonance spectroscopy (EN-
DOR) has demonstrated the presence of such exchangeable
protons consistent with hydrogen bonding to the quinone
oxygens (14, 15).
One way in which such specific interactions can be studied
in greater detail requires the use of exchanged quinoness
either isotopically labeled quinones or chemically different
quinones.
successfully used to study the binding of QAand QBin their
different redox states as well as the distribution of the spin
density on both anion radicals and has greatly extended the
functional and structural information inferred from X-ray
crystallographic studies of the oxidized ground state (for a
review, see ref 16).
Using this approach, both FTIR measurements (17-19)
and Q-band (34 GHz) cw EPR measurements (20, 21)
concluded that the 4-carbonyl oxygen of QA(see Figure 1A)
is strongly hydrogen bonded to the protein. The analysis of
the observed13C-hyperfine tensor elements for quinones
selectively labeled at positions 1 and 4 indicated asymmetric
hydrogen bonding (20, 21), while similar measurements
performed on QB (21) revealed less asymmetry, also in
accordance with FTIR measurements (22, 23). These ob-
served results were suggested to correlate the functional
difference of both quinones. Comparable measurements of
the same quinones in vitro revealed virtually symmetric
hydrogen bonding as is expected in alcoholic solution (24).
It is well-established that the hyperfine (hf) tensor of the
13C nucleus is very sensitive to its local environment and
thus to the local molecular structure (e.g., bond angles and
bond length (25-27)). As has been demonstrated for QAand
QB, EPR spectroscopy of13C-labeled semiquinones recorded
at typical EPR frequencies (around 9 GHz) are difficult to
interpret because of the comparable magnitude of the
g-anisotropy of the radical and1H- and13C-hyperfine tensors
resulting in an unresolved spectrum (e.g., Figure 2D). To
address the question of QH binding in QOX, we have
developed a procedure for incorporating exogenous13C-
13C-selectively labeled ubiquinones have been
FIGURE 1: (A) Atomic numbering of UQ-n. The direction of the
components of the g tensor with respect to the molecular axes is
also indicated. (B) Modeled quinone in the structure of cytochrome
bo3from E. coli as proposed by ref 10.
Asymmetric Binding of QH•-in Ubiquinol Oxidase
Biochemistry, Vol. 42, No. 19, 2003 5633

Page 3

labeled quinones into the QHbinding site. And to overcome
the problem of spectral resolution, we apply multifrequency
EPR to distinguish these different spectral contributions.
In this paper, the semiquinone radical anion that is
selectively
position and reconstituted into the QH binding site of the
bo3quinol oxidase is investigated using EPR spectroscopy
at 9, 34, and 94 GHz. At 94 GHz, the magnetic field range
spanned by the g tensor is sufficiently large that at the high-
field edge practically only centers are observed whose g
tensor z axis is oriented along the external magnetic field,
allowing the direct measurement of the hf splitting along
this direction (AZZ). Determination of13C hyperfine couplings
and resulting assignment to a strongly asymmetrical binding
of QH•-in QOX is discussed.
13C-enriched at either the 1- or 4- carbonyl
MATERIALS AND METHODS
Sample Preparation. Ubiquinone-2 selectively labeled with
13C (I )1/2) at the 1- or the 4-position of the quinone ring
(i.e., [13C1]-UQ2or [13C4]-UQ2) was synthesized as described
previously in ref 28.
The E. coli strain GO105/pJRHisA (29) was grown and
isolated as described in ref 13. Wild-type bo3enzyme with
bound UQ-8 was purified as described previously (30).
Native UQ-8 was removed by purification of QOX using
N,N-dimethyldodecylamine N-oxide (LDAO) as the deter-
gent, followed by detergent exchange with n-dodecyl-?-D-
maltoside (?-DM). Reconstitution with exogenous decyl-
ubiquinone and UQ-2 analogues was performed directly in
the EPR sample tube containing the protein, using quinone
solutions dissolved in a very small amount of 2-propanol
(1 µL). All samples were reduced with an excess of sodium
ascorbate under a strict argon atmosphere.
Instrumentation. X-band EPR spectra were recorded using
a Bruker ESP 300 spectrometer with a standard rectangular
Bruker EPR cavity (ER4102T), and Q-band EPR spectra
were recorded using a Bruker E-500 spectrometer with a
standard Bruker resonator (ER 5106QT-W1). Both instru-
ments were equipped with Oxford helium cryostats (ESR900
and CF935, respectively). W-band EPR spectra were ac-
quired on a Bruker E680 spectrometer equipped with an
Oxford helium-flow cryostat (CF935), a cylindrical Bruker
Teraflex TE110 cavity, and a 6T Magnex superconducting
magnet.
X-Band pulsed EPR measurements were performed on a
Bruker eleXsys E-580 spectrometer using a standard dielec-
tric resonator (MD5EN W1) equipped with an Oxford helium
(CF 935) cryostat. 3 pulse ESEEM and Davies ENDOR
experiments were performed as described previously (13, 31)
except that the microwave pulses (8 ns) were amplified using
a 1 kW microwave amplifier (applied systems engineering).
EPR Data Analysis. In frozen solutions, anisotropic g and
hf interactions are not averaged out, unlike in liquid solutions,
and all molecular orientations relative to the magnetic field
have to be considered. The spin Hamiltonian describing the
magnetic interaction between the electron spin of the
semiquinone radical and the nuclear spins I )1/2(e.g.,13C
or1H nuclei) in an external magnetic field B0is given by
where g ˜ is the electronic g tensor; A ˜I is the hf tensor of
nucleus I with principal values {Aii} (i ) x,y,z); J is the
number of coupled I )1/2nuclei; and S, II, µB, and µIare
the electron and nuclear spin operators and magnetons,
respectively. Signals from molecules with all possible
orientations relative to B0contribute to the EPR spectrum,
which is therefore significantly broadened because of the g
and hf tensor anisotropies. Spectra were simulated using a
home-written simulation and fit program (32), written in
Matlab. The measured g values were corrected for an offset
against a known g standard (DPPH (g ) 2.00351 ( 0.00002)
and N @ C70 (g ) 2.0021 ( 0.0001) for 94 GHz
measurements).
RESULTS
Figure 2 shows the cw X-band EPR spectra of the bo3
quinol oxidase from E. coli under reducing conditions at 120
K. The native ubisemiquinone radical QH•-under nonsat-
urating conditions has a line width of ∼1.0 mT (Figure 2A).
In samples prepared in the presence of LDAO, no EPR signal
from QH•-is observed. Samples reconstituted with exogenous
quinones reveal an EPR signal very similar to that of QH•-
(Figure 2B,C). ESEEM and ENDOR investigations of these
reconstituted samples indicate that the reconstituted quinone
is bound in the QHbinding site (Figure 3, see Discussion).
Unlike the g value, which remains the same, the line width
of the radicals reconstituted with selectively
UQ-2 is strongly affected because of the additional13C-hf
interaction. The observed line width at X-band increases to
1.42 mT and 1.18 mT for [13C1]-UQ-2 and [13C4]-UQ-2,
respectively (Figure 2D,E).
At higher microwave frequencies, the g tensor of QH•-
can be better resolved (Figure 4). At Q-band (34 GHz), it
has been shown that the g tensor anisotropy of QH•-is only
partially resolved (13, 14) (Figure 4B). At 94 GHz, the
resolution of the GZZcomponent is clearly observed, while
13C-labeled
FIGURE 2: X-band EPR spectra of the bo3-QOX of E. coli under
reducing conditions. From top to the bottom: (A) native UQ-8,
(B) after reconstitution with decylubiquinone, (C-E) after recon-
stitution with12C-UQ-2,13C1-UQ-2, and13C4-UQ-2, respectively.
Experimental conditions: MW power ) 0.1 mW; field modulation
frequency ) 100 kHz; field modulation amplitude ) 1 G;
temperature ) 120 K; and microwave frequency ) 9.45 GHz.
Hs) µBB B0g ˜S B - gnIµIB B0I BI+∑
J
S BA ˜II BI
5634 Biochemistry, Vol. 42, No. 19, 2003
Grimaldi et al.

Page 4

the overall line shape remains predominantly axial (Figure
4C). A numerical simulation/fit of this spectrum indicates a
slightly rhombic g tensor whose principal values (error (
0.00005) are
The increase in resolution of the g ˜ tensor at 94 GHz leading
to a separation of the GZZcomponent is of great advantage
when studying the13C-labeled compounds. The effect of13C
labeling at the 1- and 4-carbonyl position is clearly observed
at Q-band (Figure 5), but the magnitude of this13C-hf
interaction (AZZ) along GZZcan be clearly determined from
spectra taken at 94 GHz (Figure 6).
Simulation of these spectra allows a direct determination
of the AZZcomponent of the13C1and13C4hyperfine tensor
(Figure 5), which is assumed to be parallel to the GZZ
component (see Discussion). These simulations have been
carried out including explicitly hyperfine interactions from
the methyl protons at position 5 (CH3:1H-AXX) 7.85 MHz,
1H-AYY) 12.80 MHz, and1H-AZZ) 8.4 MHz, ref 13).
Other smaller couplings (hydrogen bonded-, CH2-, and
methoxy-protons) are taken into account through use of an
anisotropic line width parameter characterized by a Gaussian
line shape. Observed components of13C hyperfine tensors
used for simulations are given in Table 1.
Even at W-band, EPR spectra (solid lines) and simulations
(dashed lines) can only provide an upper limit for the13C-
AXXand13C-AYYcomponents (Table 1) as these couplings
only give rise to a broadening of the low-field feature in the
spectrum. On the other hand, the AZZ components of the
hyperfine coupling tensors are now well-resolved.
DISCUSSION
The primary aim of this paper is to analyze the electronic
structure of QH•-in the bo3quinol oxidase from E. coli as
a consequence of the surrounding protein using a combina-
tion of selective13C isotope labeling and multi-frequency
EPR spectroscopy. It has been previously demonstrated that
the length of the isoprenoid side chain beyond the first
FIGURE 3: Davies1H-ENDOR (top) and Fourier transformed 3
pulse ESEEM (bottom) spectra of the bo3-QOX of E. coli under
reducing conditions. (A) Native UQ-8 and (B) after reconstitution
with
microwave π-pulse length ) 200 ns; radio frequency π-pulse length
) 9 µs; and temperature ) 20 K. Frequency domain 3-pulse
ESEEM spectra: microwave π/2-pulse length ) 8 ns; tau ) 200
ns; and temperature ) 20 K.
12C-UQ-2. Experimental conditions: Davies
1H-ENDOR;
FIGURE 4: Cw-EPR spectra of QH•-taken at (A) X-band (9.47
GHz), (B) Q-band (34.0 GHz), and (C) W-band (94.0 GHz) using
a field modulation amplitude of 0.5G at T ) 80 K. For all other
experimental conditions, see Figures 2, 4, and 5.
GXX) 2.00593, GYY) 2.00543, GZZ) 2.00220, giso)
2.00452
FIGURE 5: Q-band (34 GHz) EPR spectra of bo3-QOX under
reducing conditions. From top to the bottom: (A) reconstituted with
unlabeled UQ-2 (12C), (B) reconstituted with UQ-213C-labeled at
the 1-carbonyl position, and (C) reconstituted with UQ-213C-labeled
at the 4-carbonyl position. Experimental conditions: MW power
) 1.2 × 10-3mW; field modulation frequency ) 100 kHz, field
modulation amplitude ) 3 G; T ) 80 K; and microwave frequency
) 33.98 GHz. Spectral simulations are performed using the g tensor
given in the text, Table 1, ref 13, and a line width of 3.5 G.
Asymmetric Binding of QH•-in Ubiquinol Oxidase
Biochemistry, Vol. 42, No. 19, 2003 5635

Page 5

isoprenyl unit does not influence the spin density of such
semiquinones and the intramolecular hyperfine couplings
appreciably (21, 33, 34); thus, UQ-2 was used as an
exogenous ubiquinone for the reconstitution experiments.
Indeed, UQ-2 is the molecule used in all activity tests for
QOX (9).
Although the multi-frequency EPR spectra shown here are
a direct indication that a semiquinone is formed, it is very
important to check that the species observed is similarly
bound in the same binding pocket as that of the native
quinone. For this purpose, several of the experimentally
observed parameters were compared: (i) the g tensor, which
is very sensitive to the electrostatic environment around the
semiquinone, is identical within experimental error for the
native and the12C-UQ-2•-revealing a very small anisotropy
(GXX ) 2.00593) as compared to other protein-bound
ubiquinone radicals. Such GXXvalues have been shown to
indicate strong hydrogen bonding (24, 33, 35). Indeed, this
value is lower than any other known protein-bound ubiquino-
ne anion radical studied so far; (ii)14N-ESEEM spectroscopy
clearly shows the characteristic quadrupolar interaction,
which is the same for all samples used in this study and
which is assigned to a peptide nitrogen nucleus as observed
in the study of Grimaldi et al. (13) on the native QH•-
ubiquinone radical (Figure 3, bottom); and (iii) pulsed1H
Davies ENDOR spectra of all these samples were very
similar to that obtained from the native QH•-(Figure 3, top).
In particular, the strong proton coupling from the methyl
group at position 5 (indicated in Figure 3, top), which ideally
serves as sensitive monitor of the spin density, is not
significantly altered with respect to values previously
published (13-15). If the reconstituted UQ-2 and the native
UQ-8 would be bound in a different fashion (e.g., sym-
metrically hydrogen-bonded), a dramatic alteration of such
hyperfine couplings would be expected as has been shown
previously using DFT calculations (36, 37). Using CIDEP
spectroscopy, the limiting case of a monoprotonated semi-
quinone has been studied, where such couplings can increase
by a factor of 2 (38).
Taken together, these results indicate that the quinone
signals arise from a specific quinone tightly bound within
the QHbinding pocket of bo3ubiquinol oxidase and not from
quinones in detergent micelles and that the reconstitution
method inserts UQ-2 into the native binding site. In
particular, the electronic structure of the observed radical
species is not dramatically altered upon reconstitution with
UQ-2.
The increase of the line width with respect to the wild
type, which is already clearly observable at the X-band, is
due to the strong hyperfine coupling of the13C nucleus. These
marked differences in this line width alteration indicate that
both labels do not contribute to the EPR spectrum in an
identical fashion. EPR measurements at higher frequencies
are used to resolve the g tensor anisotropy, which then
simplifies the analysis of the
symmetry considerations, the13C hf tensors studied here are
assumed to be collinear with the g tensor axes; thus, their z
direction coincides with the normal of the plane of the
molecule (Figure 1A), leading to the strongest hf coupling
in the z direction. This is confirmed by DFT calculations
(data not shown). Calculations of the magnitude and orienta-
tion of the g tensor have shown that these axes coincide with
the molecular axes within a few degrees even in the case of
asymmetric hydrogen bonding to the carbonyl groups (39).
The x direction of the g tensor is aligned with the axis
connecting the carbonyl oxygens.
Because the spectra of reconstituted13C1-UQ-2 and13C4-
UQ-2 in the QHsite are clearly different, it must be concluded
that the electron spin is asymmetrically distributed over the
respective C-atoms of the quinone in the protein-binding site.
Moreover, these spectra clearly show that13C1-UQ-2 exhibits
a much larger13C splitting of the AZZcomponent than13C4-
UQ-2. The quantitative analysis of this observation has been
obtained through the simulations of the Q- and W-band
spectra (see Table 1). The simulations allow the determina-
tion of the magnitude of the AZZcomponents of the13C1and
13C4hf tensors and to give an estimation of the upper limit
of the two other components AXXand AYY. The hyperfine
tensor is to a good approximation axially symmetric about
the pzorbital (that is about the z direction) as expected for
a carbon atom in a π radical in which the anisotropy in the
coupling is caused by spin density of a p orbital essentially
fixed in space and |Azz| is larger than |Axx| and |Ayy|. Accurate
spin densities can be calculated when the three components
of the13C hyperfine tensors (including signs) are known.
Since we know only |Azz| for the 1- and 4-carbonyl carbons
and can only determine an upper limit for |Axx| and |Ayy|,
we are not yet able to definitively calculate the corresponding
spin densities. From model systems studies, it has been
shown that the sign of the isotropic hyperfine coupling (aiso)
in a protic environment is positive; thus, the large AZZ
component must be positive, and since the trace of the dipolar
part of the hf tensor must be zero, AXX and AYY must be
both negative (21, 24, 40, 41).
However, the present data allow us to draw a few tentative
conclusions. The asymmetry (||13C1-Azz| - |13C4-Azz||) of
13C-hf interactions. From
FIGURE 6: W-band (95 GHz) EPR spectra of bo3-QOX under
reducing conditions. From top to the bottom: (A) native UQ-8;
(B) reconstituted with UQ-2; (C)13C-enriched at the 1-carbonyl
position, and (D)13C-enriched at the 4-carbonyl position. Experi-
mental conditions: MW power ) 5 × 10-5mW; field modulation
frequency ) 100 kHz; field modulation amplitude ) 2 G, frequency
) 100 kHz; T ) 90 K; and microwave frequency ) 93.95 GHz.
Spectral simulations performed using the g and hf tensors given in
the text, Table 1, ref 13, and a line width of 4.5 G.
5636 Biochemistry, Vol. 42, No. 19, 2003
Grimaldi et al.

Page 6

about 4 G contrasts with that of UQ-3•-measured in
2-propanol, where very similar values are observed at both
positions (see Table 1) (21, 24). Although the13C1Azzhf
component lies in the range of those previously determined
for13C-ubiquinone anions in 2-propanol,13C-UQ-3•-in QA
and QBand13C-UQ-10•-in QA(refs 20 and 21, see Table
1), the13C4-Azzhf component reveals a remarkably small
value of about 7 G, which is quite similar to the values
determined for UQ-0•-and UQ-3•-in DME/mTHF where
no hydrogen bonds are formed (24).
Previous studies on quinones in vitro have demonstrated
that the solvent surrounding has a strong effect on the13C
hf couplings at positions 1 and 4. This effect is caused by
the presence or absence of hydrogen bonding to the two
carbonyl positions that directly influences the distribution
of electron spin density over the ring, leading to a change
of the polarization of the 2p orbital of the carbon responsible
for the13C hyperfine coupling (25-27). The shift of spin
density in semiquinone anion radicals can be understood from
a simple valence model (42), recently confirmed by theoreti-
cal calculations on durosemiquinones and asymmetrically
hydrogen-bonded phyllosemiquinones (36, 37). The forma-
tion of a hydrogen bond to one of the carbonyl oxygens of
the semiquinone leads to a shift of spin density and charges
within the semiquinone ring. The bound oxygen will possess
a larger negative charge to stabilize the hydrogen bonding
interaction; thus, the spin density will be partly shifted within
the semiquinone. When a strong hydrogen bond to O1
(respectively, O4) is formed, increase in spin density at carbon
C1, C3, C5, and O4 (respectively, C2, C4, C6, and O1) is
expected. This effect also correlates directly with the
magnitude of the13C-AZZcomponents.
Thus the measurements performed here give direct and
clear evidence that the13C4hf tensor is more isotropic than
the13C1tensor (i.e., that the electron spin density on the C4
is lower than that on the C1 indicating that the stronger
hydrogen bond is formed to O1). This is in agreement with
measurements of the proton methyl hyperfine couplings at
position 5 (13-15), which are an indirect probe of the
unpaired electron density.
In contrast with the situation in a frozen solution of
2-propanol or DME/mTHF, where the electron spin densities
(and consequently the
carbons at positions 1 and 4, the corresponding AZZ hf
couplings determined here differ considerably and especially
the very small value observed for C4 is consistent with a
strongly asymmetric or one-sided binding of QH•-in its
binding pocket. The magnitude of this asymmetry is of
similar magnitude to that observed for QA•-and much
13C hf couplings) are similar for
stronger than that of QB•-(see Table 1). Their differences
were presented as being consistent with QBfunctioning more
like a quinone in protic solution (i.e., acting as a 2e-/2H+
acceptor (3, 21, 24)). Following this reasoning, this suggests
that QHacts more like QAwith regard to its function as an
intermediate electron donor in the protein.
Moreover, although the13C1Azzhf coupling of QH•-lies
in the range of couplings determined previously for the
above-mentioned ubisemiquinones, the value for13C4Azzhas
an unusually small value, very close to the smallest value
determined so far on ubiquinones in aprotic solvents, which
suggests that this carbonyl oxygen is not hydrogen bonded.
This appears rather unusual and is interpreted as strongly
one-sided hydrogen bonding of QH•-to the protein. There
is currently only one other example in the literature where
such an effect could be expected, although no direct
measurements have been performed to date, the secondary
electron acceptor A1in photosystem I (PS I) as revealed in
the recently refined 2.5 Å X-ray structure of PS I from
Synechococcus elongatus (43). Recent density functional
calculations (44) performed on A1•-concluded that one-sided
hydrogen bonding could explain some of the unusual1H-
hfc values that have been observed for A1(45), which are
also similar to those observed here. This would further
support the proposal of a one-sided bonding model for QH•-.
ENDOR studies performed on QH•-have previously shown
the presence of exchangeable protons in the immediate
environment, which were assigned to hydrogen-bonded
protons to the quinone (14, 15).
Two hf couplings have been assigned to the same axial
hyperfine tensor and by comparison with ubisemiquinones
in 2-propanol where two different tensors have been resolved
(34, 46), it was concluded that stronger hydrogen bonds to
the carbonyl oxygens than are present in 2-propanol are
formed to the semiquinone in its binding site (15).
Further, after comparison with ENDOR data from plasto-
quinone-substituted samples, Hastings et al. (15) postulated
a symmetrical hydrogen-bonding environment in contrast to
our findings using ubiquinone derivatives. The model
presented here does not, however, exclude multiple hydrogen
bonding to the 1-carbonyl oxygen as proposed in the modeled
QH binding site. It is important to note that the observed
13C4hf tensor is difficult to reconcile with strong hydrogen
bonding to the 4-carbonyl oxygen.
A recent study of the binding of UQ-2 in the cytochrome
bo3by FTIR spectroscopy using the same labeled quinones
as presented in this work suggested mainly symmetrical and
rather weak hydrogen bonding to the oxidized and fully
reduced states, although the anion radical state was not
Table 1:
13C-hf Tensor Principal Values (in mT) of QH•-in bo3Containing UQ-2 Specifically Labeled at the 1- and 4-Carbonyl Positionsa
13C-UQ-2•-
QH•- b
-0.15 (5)f
-0.45 (5)
+1.10 (3)
-0.25 (5)
-0.37 (5)
+0.72 (3)
0.38
13C-UQ-3•-
13C-UQ-10•-
isopropanolc
-0.43 (6)
-0.37 (6)
+1.09 (2)
-0.40 (6)
-0.35 (6)
+1.15 (2)
0.06
DME/mTHFd
-0.43 (8)
-0.54 (8)
+0.73 (2)
-0.47 (8)
-0.55 (8)
+0.73 (2)
0.00
QB•- c
-0.39 (6)
-0.47 (6)
+0.99 (2)
-0.36 (6)
-0.37 (6)
+1.15 (2)
0.16
QA•- c
-0.45 (6)
-0.52 (6)
+0.81 (2)
-0.33 (6)
-0.35 (6)
+1.25 (2)
0.44
UQ10•-(isoprop)e
n.d.g
n.d.
1.13 (3)
n.d.
n.d.
1.10 (3)
0.03
QA•- e
0.55 (5)
0.65(5)
0.80 (3)
<0.25 (5)
<0.25 (5)
1.27 (3)
0.47
13C1-AXX
13C1-AYY
13C1-AZZ
13C4-AXX
13C4-AYY
13C4-AZZ
|13C1-Azz-13C4-Azz|
aComparison with QA•-and QB•-in bRC and with ubiquinone anion radicals in organic solvents.bThis paper.cRef 21.dRef 24.eRef 20.
fNumbers in parentheses are the errors in the last digit.gn.d.: not determined.
Asymmetric Binding of QH•-in Ubiquinol Oxidase
Biochemistry, Vol. 42, No. 19, 2003 5637

Page 7

investigated (28). Possible reasons for this difference could
be a different binding of QH•-and QHand/or a pH-dependent
binding of QH•-.
The first case has been suggested for QA(47) and shown
for QB(48) in bRC, and the second case has been clearly
demonstrated for QA•-in photosystem II using HYSCORE
spectroscopy (49). Such a stabilization of the semiquinone
state could be a factor for optimum electron transfer and
could depend on the pH, as it has been shown for the
intramolecular electron transfer from ubisemiquinone to heme
b by using the pulse radiolysis technique (50).
In the recent ESEEM study of the native QH•-radical
anion, a strong interaction of the unpaired electron spin with
a nitrogen nucleus was observed (13). The current study
supports the idea that this coupling, suggested to occur via
a hydrogen bond, is to the side of the quinone corresponding
to the 1-carbonyl position. On the basis of the knowledge
of the quadrupolar parameters (K, η) and their comparison
with those obtained for other membrane-protein bound
semiquinone radicals and model systems, it was assigned to
a peptide nitrogen, although an arginine residue cannot be
excluded. Such an arginine has been proposed as hydrogen-
bond donor to the quinone. If this were the case, the proposed
quinone orientation would be supported by this work (i.e.,
the 1-carbonyl oxygen would form a strong hydrogen bond
to Arg71). From both EPR and FTIR measurements on
mutant cytochrome bo3 (Asp75) from E. coli, it has been
proposed that this amino acid could interact with the
semiquinone (51, 52). ESEEM measurements on mutants of
the above-cited amino acids could be useful to study this
hypothesis and are currently being performed in our labora-
tory.
CONCLUSIONS
Using a multifrequency EPR approach, clear evidence is
obtained for the asymmetric electronic structure of the QH•-
radical in the bo3QOX from E. coli. By comparison with
data on ubiquinones measured in vitro, it is suggested to form
strongly asymmetric or one-sided hydrogen bonding to the
protein. In contrast to the binding of QAin bRCs, the strong
hydrogen bond formed here is to the 1-carbonyl position
(which points directly to heme b) and not to the 4-carbonyl
position. In bRCs, QAlinks a one-electron process to a two-
electron process, whereas here QH links a two-electron
process to a one-electron process and is thus a possible
explanation of the observed difference. This may also be an
indication of the direction of electron-transfer to heme b and
thus of the orientation of QHin the protein that now has to
be examined in detail.
ACKNOWLEDGMENT
We would like to thank Dr. Petra Hellwig (University of
Frankfurt, Germany) for providing unpublished data and
useful discussions, Prof. Klaus-Peter Dinse (TU Darmstadt,
Germany) for the use of the W-band EPR spectrometer, and
the EPR division of Bruker Biospin (Rheinstetten, Germany)
for the possibility of performing the Q-band EPR measure-
ments.
REFERENCES
1. Trumpower, B. L. (1982) Function of Quinones in Energy
ConserVing Systems, Academic Press, New York.
2. Shinkarev, M. Y., and Wraight, C. A. (1993) in The Photosynthetic
Reaction Center (Deisenhofer, J., and Norris, J. R. Eds.) Vol. 1,
pp 194-255, Academic Press, San Diego.
3. Lubitz, W., and Feher, G. (1999) Appl. Magn. Reson. 17, 1-48.
4. Breton, J., Burie, J. R., Berthomieu, C., Berger, G., and Nabedryk,
E. (1994) Biochemistry 33, 4953-4965.
5. Zhao, X., Ogura, T., Okamura, M., and Kitagawa, T. (1997) J.
Am. Chem. Soc. 119, 5263-5264.
6. Van Liemt, W. B. S., Boender, G. J., Gast, P., Hoff, A. J.,
Lugtenburg, J., and de Groot, H. J. M. (1995) Biochemistry 34,
10229-10236.
7. Mogi, T., Tsubaki, M., Hori, H., Miyoshi, H., Nakamura, H., and
Anraku, Y. (1998) J. Biochem. Mol. Biol. Biophys. 2, 79-110.
8. Sato-Watanabe, M., Mogi, T., Miyoshi, H., Iwamura, H., Mat-
sushita, K., Adachi, O., and Anraku, Y. (1994) J. Biol. Chem.
269, 28899-29907.
9. Sato-Watanabe, M., Mogi, T., Ogura, T., Kitagawa, T., Miyoshi,
H., Iwamura, H., and Anraku, Y. (1994) J. Biol. Chem. 269,
28908-28912.
10. Abramson, J., Riistama, S., Larsson, G., Jasaitis, A., Svensson-
Ek, M., Laakkonen, L., Puustinen, A., Iwata, S., and Wikstro ¨m,
M. (2000) Nature Struct. Biol. 7, 910-917.
11. Sato-Watanabe, M., Itoh, S., Mogi, T., Matsuura, K., Miyoshi,
H., and Anraku, Y. (1995) FEBS Lett. 374, 265-269.
12. Ingledew, W. J., Ohnishi, T., and Salerno, J. C. (1995) Eur. J.
Biochem. 227, 903-908.
13. Grimaldi, S., MacMillan, F., Ostermann, T., Ludwig, B., Michel,
H., and Prisner, T. (2001) Biochemistry 40, 1037-1043.
14. Veselov, A. V., Osborne, J. P., Gennis, R. B., and Scholes, C. P.
(2000) Biochemistry 39, 3169-3175.
15. Hastings, S. F., Heathcote, P., Ingledew, W. J., and Rigby, S. E.
J. (2000) Eur. J. Biochem. 267, 5638-5645.
16. Michel-Beyerle, M. E., Ed. (1996) in The Reaction Center of
Photosynthetic Bacteria: Structure and Dynamics, Springer-
Verlag, Berlin, Germany.
17. Breton, J., Burie, J. R., Boullais, C., Berger, G., and Nabredyk,
E. (1994) Biochemistry 33, 12405-12415.
18. Brudler, R., de Groot, H. J. M., van Liemt, W. B. S., Steggerda,
W. F., Esmeijer, R., Gast, P., Hoff, A. J., Lugtenburg, J., and
Gerwert, K. (1994) EMBO J. 13, 5523-5530.
19. Breton, J., Boullais, C., Burie, J. R., Nabredyk, E., and Miosk-
owski, C. (1994) Biochemistry 33, 14378-14386.
20. Van den Brink, J. S., Spoyalov, A. P., Gast, P., van Liemt, W. B.
S., Raap, J., Lugtenburg, J., and Hoff, A. J. (1994) FEBS Lett.
353, 273-276.
21. Isaacson, R. A., Abresch, E. C., Lendzian, F., Boullais, C.,
Paddock, M. L., Mioskowski, C., Lubitz, W., and Feher, G. (1996)
in The reaction center of photosynthetic bacteria, structure and
dynamics (Michel-Beyerle, M.-E., Ed.) pp 353-367, Springer-
Verlag, Berlin, Germany.
22. Brudler, R., de Groot, H. J., van Liemt, W. B., Gast, P., Hoff, A.
J., Lugtenburg, J., and Gerwert, K. (1995) FEBS Lett. 370, 88-
92.
23. Breton, J., Boullais, C., Berger, G., Mioskowski, C., and Nabredyk,
E. (1995) Biochemistry 34, 11606-11616.
24. Nimz, O., Lendzian, F., Boullais, C., and Lubitz, W. (1998) Appl.
Magn. Reson. 14, 255-263.
25. Karplus, M. A., and Fraenkel, G. K. (1961) J. Chem. Phys. 35,
1312-1323.
26. Das, M. R., and Fraenkel, G. K. (1965) J. Chem. Phys. 42, 1350-
1360.
27. O’Malley, P. J. (1998) Chem. Phys. Lett. 291, 367-374.
28. Hellwig, P., Mogi, T., Tomson, F. L., Gennis, R. B., Iwata, J.,
Miyoshi, H., and Ma ¨ntele, W. (1999) Biochemistry 38, 14683-
14689.
29. Rumbley, J. N., Furlong Nickels, E. F., and Gennis, R. B. (1997)
Biochim. Biophys. Acta 1340, 131-142.
30. Schro ¨ter, T., Winterstein, C., Ludwig, B., and Richter, O.-M. H.
(1998) FEBS Lett. 432, 109-112.
31. Faller, P., Maly, T., Rutherford, A. W., and MacMillan, F. (2001)
Biochemistry 40, 320-326.
32. Prisner, T. F., Lyubenova, S., Atabay, Y., MacMillan, F., Kro ¨ger,
A., and Klimmek, O. (2003) J. Biol. Inorg. Chem. 8, 419-426.
33. Burghaus, O., Plato, M., Rohrer, M., Mo ¨bius, K., MacMillan, F.,
and Lubitz, W. (1993) J. Phys. Chem. 29, 7639-7647.
34. MacMillan, F., Lendzian, F., and Lubitz, W., (1995) Magn. Reson.
Chem. 33, 81-93.
35. Knu ¨pling, M., To ¨rring, J. T., and Un, S. (1997) Chem. Phys. 219,
291-304.
5638 Biochemistry, Vol. 42, No. 19, 2003
Grimaldi et al.

Page 8

36. O’Malley, P. J. (1998) J. Phys. Chem. A 102, 248-253.
37. O’Malley, P. J. (1999) Biochim. Biophys. Acta 1411, 101-113.
38. Boiden Pedersen, J., Hansen C. E. M., Parbo, H., and Muus, L.
T. (1975) J. Chem. Phys. 63, 2398-2405.
39. Isaacson, R. A., Lendzian, F., Abresch, E., Lubitz, W., and Feher,
G. (1995) Biophys. J. 69, 311-322.
40. Strauss, H. L., and Fraenkel, G. K. (1961) J. Chem. Phys. 35,
1738.
41. Samoilova, R. I., Van Liemt, W. B. S., Steggerda, W. F.,
Lugtenburg, J., and Hoff, A. J. J. (1994) J. Chem. Soc., Perkin
Trans. 2, 609-614.
42. MacMillan, F., Lendzian, F., Renger, G., and Lubitz, W. (1995)
Biochemistry 8144-8156.
43. Jordan, P., Fromme, P., Witt, H. T., Klukas, O., Saenger, W., and
Krauss, N. (2001) Nature 411, 909-917.
44. Kaupp, M. (2002) Biochemistry 41, 2895-2900.
45. Rigby, S. E. J., Heathcote, P., and Evans, M. C. W. (1996)
Biochemistry 35, 6651-6656.
46. Rohrer, M., MacMillan, F., Prisner, T. F., Gardiner, A., Mo ¨bius,
K., and Lubitz, W. (1998) J. Phys. Chem. B 102, 4648-4657.
47. Kropacheva, T. N., van Liemt, W. B. S., Raap, J., Lugtenburg, J.,
and Hoff, A. J. (1996) J. Phys. Chem. 100, 10433-10442.
48. Stowell, M. H., McPhilips, T. M., Rees, D. C., Soltis, S. M.,
Abresch, E., and Feher, G. (1997) Science 276, 812-816.
49. Deligiannakis, Y., Hanley, J., and Rutherford, A. W. (1999) J.
Am. Chem. Soc. 121, 7653-7664.
50. Kobayashi, K., Tagawa, S., and Mogi, T. (2000) Biochemistry 39,
15620-15625.
51. Hellwig, P., Barquera, B., and Gennis, R. B. (2000) Biochemistry
40, 1077-1082.
52. Hellwig, P., Yano, T., Ohnishi, T., and Gennis, R. B. (2002)
Biochemistry 41, 10675-10679.
BI034010Z
Asymmetric Binding of QH•-in Ubiquinol Oxidase
Biochemistry, Vol. 42, No. 19, 2003 5639