Determination of the Proton Environment of High Stability Menasemiquinone Intermediate in Escherichia coli Nitrate Reductase A by Pulsed EPR

Background: Escherichia coli nitrate reductase A highly stabilizes a semiquinone catalytic intermediate. Results: Three proton hyperfine couplings to this radical with atypical characteristics are characterized. Conclusion: Semiquinone binding is strongly asymmetric and occurs via a single short in-plane H-bond. Significance: Learning how the protein environment tunes the semiquinone properties is crucial for understanding the quinol utilization mechanism by energy-transducing enzymes.

fine coupling together with an almost zero isotropic contribution. It is assigned to a proton involved in a short ϳ1.6 Å inplane hydrogen bond between the quinone O1 oxygen and the N␦ of the His-66 residue, an axial ligand of the distal heme b D . Structural and mechanistic implications of these results for the electron-coupled proton translocation mechanism at the Q D site are discussed, in light of the unusually high thermodynamic stability of MSQ D .
Quinones are small lipophilic organic molecules found in energy-transducing membranes of all living organisms except methanogens (1). Due to their ability to transfer up to two electrons and two protons, they are widely used in photosynthetic and respiratory electron transfer chains. Quinones can freely diffuse in the hydrophobic core of lipid membranes. They can therefore bind into specific quinone-reactive sites (Q-sites) 5 of membrane proteins in which they function as two-electron and proton carriers and are responsible for exchange of reducing equivalents between different electron transport complexes. In this case, the quinones leave the protein after completion of the redox cycle. Typical examples are the Q B site of bacterial reaction center (RC) or photosystem II and the Q-sites (Q o and Q i ) of bc 1 complex. In contrast, non-dissociable quinones can be tightly bound at specific quinone-reactive sites of proteins in which they can be involved in electron transfer processes as prosthetic groups. Well known representatives of this type include quinones in the Q A site in RCs of purple bacteria and in photosystem II or the A 1 sites in photosystem I (2,3).
The different redox states of quinones may also adopt different conformations in the quinone-binding pockets, as evidenced for ubiquinone and ubisemiquinone at the Q B site of bacterial RC (4). The functional diversity of Q-sites arises from a particular tuning of the protein environment. Despite the fact that high-resolution structural data are available for several Q-sites, how protein-cofactor interactions relate to and control the functional properties of the bound quinone is largely unknown. In particular, understanding the molecular mechanism underlying the coupling between electron transfer and proton translocation that occurs at dissociable Q-sites requires obtaining structural information on all three forms, quinone (Q), semiquinone (SQ), and quinol (QH 2 ). For this purpose, high-resolution EPR methods such as ENDOR (electron nuclear double resonance) and ESEEM (electron spin echo envelope modulation) spectroscopies were proven to be valuable by giving detailed structural information on protein-bound semiquinone intermediates, provided that this paramagnetic state can be trapped for spectroscopic studies.
Escherichia coli nitrate reductase A (NarGHI) is a membrane-bound heterotrimeric enzyme induced by anaerobiosis and the presence of nitrate. Involved in the nitrate respiratory pathway, a major alternative to the bacterial oxidative phosphorylation, it couples the oxidation of menaquinols or ubiquinols at a periplasmically oriented Q-site (named Q D ) to the cytoplasmic reduction of nitrate. Thus, both substrate turnovers contribute to the generation of a proton motive force across the cytoplasmic membrane. NarGHI contains eight redox-active metal centers (5-9): a molybdenum cofactor and an Fe 4 S 4 cluster (FS0) in the nitrate-reducing subunit NarG; one Fe 3 S 4 cluster (FS4) and three Fe 4 S 4 clusters (FS1-3) in the electron transfer subunit NarH; and two low spin hemes b in the membrane-anchor subunit NarI, termed b D and b P to indicate their distal and proximal position to the catalytic site. Importantly, NarI stabilizes an EPR-detectable semiquinone intermediate of both natural substrates at its quinol oxidation site Q D close to heme b D (10 -12). Remarkably, the resulting menasemiquinone species herein referred to as MSQ D has the largest thermodynamic stability measured so far in respiratory complexes stabilizing semiquinone intermediates. These peculiar properties render NarGHI ideally suited for investigating the molecular factors responsible for the reactivity of respiratory enzymes toward quinols.
Although no high-resolution structural data revealing the binding mode of the natural quinol/quinone substrate are available, we recently utilized high-resolution EPR techniques on endogenous MSQ D and ubisemiquinone radical (USQ D ) stabilized in NarGHI-enriched inner membrane vesicles (IMVs) of E. coli to explore their environment using the unpaired electron as a probe. The use of ESEEM and HYSCORE (hyperfine sublevel correlation) spectroscopies on either the wild-type enzyme or the enzyme uniformly enriched with 15 N nuclei provided direct evidence for nitrogen ligation to MSQ D and USQ D . On the basis of the direct determination of the quadrupolar parameters of the corresponding interacting 14 N by S-band (ϳ3 GHz) HYSCORE experiments, we assigned the latter to an N ␦ imidazole nitrogen and proposed it to arise from the heme b D axial ligand His-66 (13). The non-zero isotropic hyperfine coupling of this nitrogen suggests that the interaction occurs via a hydrogen bond, allowing electron spin density to be transferred from the radical to the interacting nucleus. Interestingly, these experiments did not support a direct H-bond between MSQ D (or USQ D ) and Lys-86, a residue in the Q D site that was previously shown to be essential for quinol oxidation and menasemiquinone detection (11,14). Indeed, no evidence for the transfer of a measurable spin density on any other nuclei than that mentioned above was found. Thus, we tentatively proposed that a water-mediated interaction is formed between MSQ D (or USQ D ) and Lys-86, consistent with the latter being involved in reactivity toward quinols (13). Moreover, we have recently shown that a cardiolipin molecule specifically bound to the complex is necessary for quinol substrate fixation at the Q D site, probably through the action of one of its acyl chains located in the vicinity of His-66 (15). Clearly, additional information is required to improve our understanding of the semiquinone binding mode in the Q D site and of its functional tuning by the protein environment.
In this work, high-resolution EPR techniques have been used to map the environment and the binding mode of MSQ D via the detection of proton hyperfine couplings to the radical. Using a combination of X-band (ϳ9 GHz) ESEEM/HYSCORE and Q-band (ϳ34 GHz) Mims ENDOR experiments on MSQ D prepared in either a protonated or a deuterated solvent, one exchangeable and two non-exchangeable protons magnetically coupled to the radical were detected. Their detailed characterization allows their assignment to specific protons in the vicinity of the radical. Implications of these results for deciphering the semiquinone binding mode and the catalytic mechanism at the Q D site are discussed.
Purified E. coli NarGHI-enriched IMVs were used for this study, allowing us to maintain an unmodified lipid environment and to study the interaction of NarGHI with its endogenous menaquinol substrate. For this purpose, purified E. coli NarGHI-enriched IMVs were isolated by differential centrifugation and sucrose gradient step as described in Ref. 11 using a buffer containing 100 mM MOPS and 5 mM EDTA at pH 7.5. Deuterium-exchanged samples were prepared using the same membrane extraction protocol with a buffer containing 2 H 2 O (99.9% atom 2 H) instead of 1 H 2 O. The functionality of NarGHI in our samples was confirmed spectrophotometrically by measuring the quinol:nitrate oxidoreductase activity. Stabilization of the semiquinone at the Q D site was achieved through redox titrations under the same conditions as those used in our previous works (10,11,13). Redox potentials are given in the text with respect to the standard hydrogen electrode. The semiquinone concentration in our samples estimated from the double integration of their corresponding EPR spectra and by comparison with a standard (1 mM CuSO 4 ) was estimated in the range of 10 -12 M.
Pulsed EPR/ENDOR Experiments-X-band (ϳ9 GHz) and Q-band (ϳ34 GHz) pulsed EPR/ENDOR experiments were performed using a Bruker EleXsys E580-Q spectrometer equipped with an Oxford Instruments CF 935 cryostat. Spectra were measured at 90 K to avoid contamination from fast relaxing metal centers such as FeS centers in NarGHI (13). For the two-pulse experiments (/2--), the echo intensity was measured as a function of magnetic field at fixed time interval between the two microwave pulses for field sweep ESE or as a function of at a fixed magnetic field value for two-pulse ESEEM.
Two-pulse and four-pulse (/2--/2-T/2--T/2-/2) ESEEM and HYSCORE (/2--/2-t 1 --t 2 -/2) experiments were performed at a magnetic field corresponding to the maximum intensity of the MSQ D two-pulse field sweep ESE spectrum where all orientations of the semiquinone with respect to the external magnetic field contribute, giving rise to powder ESEEM/HYSCORE spectra (see the supplemental material). Spectra were processed using the Bruker Xepr software. Relaxation decays were subtracted (fitting by polynomial functions) followed by zero-filling and tapering with a Hamming window, before Fourier transformation, which finally gives the spectrum in frequency domain. All spectra are shown in absolute value mode. HYSCORE spectra are presented as contour plots.
Q-band pulsed 2 H ENDOR spectra were obtained using the Mims (/2--/2-t-/2--echo) sequence (17). A radio frequency pulse was applied during the time interval t. The radio frequency power was delivered by a 2-kilowatt Dressler solid state radio frequency amplifier. It was optimized for radio frequency pulse lengths of 40 s for deuterium Mims ENDOR experiments. Pulsed ENDOR spectra were recorded at a magnetic field corresponding to the g Ќ position of the nearly axial Q-band EPR signal of MSQ D (10). Pulsed EPR/ENDOR spectra were simulated in the MATLAB environment using the Easyspin software package (release 3.1.0) (18).
Hyperfine and Quadrupole Interactions-A hyperfine coupling between an S ϭ 1 ⁄ 2 radical and a nucleus with nuclear spin value I consists in general of (i) the isotropic contribution A iso ϭ 2 0 g e g n ␤ e ␤ n ͉ 0 (0)͉ 2 /3h, where ͉ 0 (0)͉ 2 is the electron spin density at the nucleus, g e and g n are electron and nuclear g-factors, respectively, ␤ e and ␤ n are Bohr and nuclear magnetons, respectively, h is Planck's constant, and (ii) the anisotropic contribution described by the traceless dipolar coupling tensor T . In most cases, T can be assumed to be axial, with principal values (ϪT, ϪT, 2T).
The hyperfine couplings of different isotopes of the same element are proportional to a very good approximation to the corresponding g n values. In this study, the direct and simultaneous determination of A iso and T of the protons interacting with MSQ D were derived from the analysis of HYSCORE crosspeak contours as detailed in the supplemental material (19).
A 2 H nucleus has a quadrupole moment that interacts with the electric field gradient at the nucleus. The components of the electric field gradient tensor are defined in its principal axis system and ordered according to ͉q ZZ ͉ Ն ͉q YY ͉ Ն ͉q XX ͉. This traceless tensor can then be fully described by only two parameters: (i) the 2 H nuclear quadrupole coupling constant ϭ ͉e 2 q ZZ Q/h͉, where e is the charge of electron, Q is the 2 H nuclear electric quadrupole moment; and (ii) the asymmetry parameter ϭ ͉q YY Ϫ q XX /q ZZ ͉. is a measure of the strength of the interaction between the nuclear quadrupole moment and the electric field gradient at the 2 H nucleus site due to anisotropic charge distribution around the nucleus, whereas is a measure of the deviation of this distribution from axial symmetry. Thus, the electric field gradient is related to the specific binding geometry. Its components can, therefore, be used to obtain detailed information on hydrogen bonds (20 -26). In this study the parameters and of the 2 H interacting with MSQ D were estimated by simulation of the Q-band 2 H Mims ENDOR spectrum.

X-band Pulsed EPR (Field Sweep, Two-pulse ESEEM)-
X-band field sweep ESE spectra of NarGHI-enriched IMVs were recorded at 90 K in samples redox-poised at ϳϪ100 mV prepared in either 1 H 2 O or 2 H 2 O. They show a single line from the MSQ stabilized at the Q D site of NarGHI with g ϳ2.0045 and the width ϳ0.8 mT in 1 H 2 O (13). Replacement of 1 H 2 O by 2 H 2 O decreases the line width by less than 0.1 mT (Fig. 1A). The weakness of this effect is due to the primary contribution to the line shape of the g-tensor anisotropy, which was previously resolved using numerical simulation of the MSQ D Q-band EPR spectrum (10). The two-pulse spin echo decay of the radical measured in 1 H 2 O at 90 K is depicted in Fig. 1B. It mainly shows the modulation associated with weakly coupled protons in the immediate environment, with Zeeman frequencies I ( 1 H) X-band 1 H HYSCORE-The low frequency parts of the X-band HYSCORE spectra of MSQ D were previously shown and analyzed in detail. They revealed cross-peaks arising from a single 14 N hyperfine coupling assigned to the heme b D ligand His-66 residue (12,13). In addition to these 14 N signals, several cross-features from protons symmetrically positioned with respect to the 1 H Zeeman frequency ( I ( 1 H) ϳ14.7 MHz) are clearly resolved in the 10 -20-MHz frequency range in the (ϩ,ϩ) quadrant of these spectra ( Fig. 2A). This indicates that several protons are magnetically coupled to the radical. The appearance of these correlations in the (ϩ,ϩ) quadrant indicates that the corresponding hyperfine couplings for a given proton satisfy the relationships ͉Tϩ2A iso ͉ Ͻ Ͻ 4 I ( 1 H) (27). To further analyze the spectrum and discriminate between exchangeable and non-exchangeable features, HYSCORE experiments were also performed under the same conditions in the sample prepared in 2 H 2 O. Fig. 2 shows the proton region of the corresponding HYSCORE spectra recorded with ϭ 204 ns in 1 H 2 O ( Fig. 2A) or 2 H 2 O (Fig. 2B). In addition to the diagonal peak at I ( 1 H) ϳ14.7 MHz, four pairs of cross-features located symmetrically relative to the diagonal are well resolved in the spectrum shown in Fig. 2A. They are designated 1, 1Ј, 2, 2Ј, 3, 3Ј, 4, and 4Ј. The ridges 2-2Ј exhibit the smallest resolved hyperfine splitting, of the order of ϳ2 MHz, whereas the largest one is observed for cross-peaks 1-1Ј. Cross-ridges 3-3Ј possess the most extended anisotropic contour, with the largest deviation from the diagonal, whereas cross-peaks 4-4Ј deviate significantly from the normal to the diagonal. These two features indicate a significant anisotropic hyperfine component. Contours 1-1Ј and 2-2Ј are approximately normal to the diagonal, suggesting a smaller anisotropy. Cross-peaks 1 and 4 partially overlap.
Cross-peaks 3-3Ј and 4-4Ј completely disappear in the proton HYSCORE spectrum measured in 2 H 2 O, demonstrating that they are produced by at least one exchangeable proton (Fig.  2B). In contrast, cross-peaks 1-1Ј and 2-2Ј still appear in the spectrum measured in 2 H 2 O, showing that they arise from nonexchangeable (i.e. covalently bound) protons.
X-band 1 H Four-pulse ESEEM-Additional information about the interacting protons was obtained from one-dimensional four-pulse ESEEM spectra, which are particularly useful for the observation of proton sum combination lines with improved resolution (27, 28). The four-pulse ESEEM spectrum

TABLE 1 Hyperfine tensors derived from contour line shape analysis of HYSCORE spectra
A Ќ is the component of the hyperfine tensor where the direction of magnetic field is perpendicular to the symmetry axis of the axially symmetric hyperfine tensor, and A , the component where the direction of magnetic field is parallel to it (see the supplemental data). A iso ϭ 1/3(2A Ќ ϩA ). The preferred sets are indicated in bold (see "Discussion").  Fig. 3A. The most intense line appears exactly at the 2 I ( 1 H) frequency and represents the contribution of weakly coupled protons from the protein environment. In addition, the spectrum exhibits a peak of lower intensity shifted from 2 I ( 1 H) to higher frequencies by ϳ1.2 MHz. This shifted peak completely disappears in the spectra of the sample prepared in 2 H 2 O (Fig. 3B). This indicates that the line shifted from 2 I ( 1 H) arises from an exchangeable proton. The shift observed in the four-pulse ESEEM is well described by

Q D Site Menasemiquinone in Nitrate Reductase
from which the anisotropic component T of the hyperfine coupling can be estimated (see supplemental material). The shift of ϳ1.2 MHz corresponds to T ϭ 5.7 MHz, which is in very good agreement with the corresponding value determined for H3 from the analysis of the HYSCORE spectra. The expected shifts from H1 and H2 (0.06 and 0.05 MHz, respectively) are too small to be resolved in a four-pulse ESEEM spectrum (supplemental Table S2). The proton sum combination peak therefore confirms the assignment made in the HYSCORE spectra for the exchangeable proton and the hyperfine coupling determined from these spectra.
Q-band 2 H Mims ENDOR-Further details concerning exchangeable protons coupled to MSQ D were obtained through the use of pulsed 2 H ENDOR spectroscopy. Fig. 4 shows the Q-band 2 H Mims ENDOR spectrum of MSQ D in NarGHI-enriched IMVs prepared in 2 H 2 O. It has been recorded at a magnetic field value corresponding to the maximum intensity of the nearly axially symmetric Q-band EPR spectrum of MSQ D with g-tensor principal values g x ϭ 2.0061, g y ϭ 2.0051, g z ϭ 2.0023 Ϯ 0.0001 (Fig. 4, top) (10). The ENDOR spectrum exhibits two pairs of well resolved intense lines located symmetrically with respect to the 2 H nuclear Larmor frequency ( I ( 2 H) ϳ7.84 MHz) and subject to the nuclear quadrupole interaction. The splitting of their center away from the Larmor frequency (ϳ0.8 MHz) is determined by the hyperfine coupling constant, and the splitting within the pair (ϳ0.13 MHz) is given by the quadrupole interaction. The same holds for the two pairs of less intense lines resolved in the spectrum shown in Fig. 4, separated by ϳ1.5 MHz and split each by ϳ0. 26 MHz. The hyperfine coupling values of ϳ0.8 and ϳ1.5 MHz are first estimates of the A Ќ and A ʈ components of an almost purely dipolar hyperfine tensor. They are very close to those found for H3 and scaled to the 2 H nucleus, i.e. A Ќ ϭ 0.87 and A ʈ ϭ 1.77 MHz. Similarly, a nuclear quadrupole coupling constant ϳ 4 ⁄ 3 ϫ 0.13 ϳ0.173 MHz can be estimated from the 0.13-MHz splitting of the two most intense doublets. A numerical simulation of the spectrum is shown in Fig. 4. It was obtained using the proton hyperfine coupling values of H3 deduced from the analysis of the HYSCORE spectra and rescaled by the factor g n ( 1 H)/g n ( 2 H) ϳ6.5. This best simulation was obtained assuming that the g-, A-, and Q-tensors are collinear, with quadrupole parameters ϭ (0.176 Ϯ 0.004) MHz and ϭ 0.20 Ϯ 0.05. Finally, this procedure allowed us to select unambiguously the right set (͉A iso ͉, ͉T͉) for H3 from the two alternatives (Table 1). Overall, the data show that a single exchangeable proton is coupled to the radical, in agreement with the analysis above.

Single Exchangeable Proton Coupling with Peculiar Hyperfine
Coupling Characteristics-Our data clearly show the presence of a single exchangeable proton in the vicinity of MSQ D characterized by ͉A iso ͉ ϭ 0.06 MHz and ͉T͉ ϭ 5.73 MHz. Such hyperfine coupling constants are in the range of those measured for  exchangeable protons coupled to protein-bound semiquinones and assigned to protons hydrogen-bonded to the quinone carbonyl oxygens. Typical examples include exchangeable proton couplings to the menasemiquinone stabilized at the Q H site of the aa 3 menaquinol oxidase from Bacillus subtilis (͉T͉ ϭ 5.6 MHz and ͉A iso ͉ ϭ 5.4 MHz) or to the photoaccumulated phyllosemiquinone A 1 ⅐ Ϫ in Thermococcus elongatus photosystem I (͉T͉ ϳ3.7 MHz and ͉A iso ͉ ϳ0.1 MHz) (29,30). Thus, we assign H3 to a proton involved in H-bonding to one of the MSQ D carbonyl oxygens. However, the hyperfine coupling tensor to H3 has atypical properties as it combines both an almost zero isotropic hyperfine coupling constant A iso and a large anisotropic part T. The magnitude of the H-bond tensor is determined by the geometry of the H-bond. A small A iso is expected when the H-bond lies in the molecular plane due to the small overlap between the hydrogen 1s orbital and the oxygen 2pz orbital forming part of the semiquinone singly occupied molecular orbital. Almost purely anisotropic hyperfine tensors have thus been observed for in-plane hydrogen-bonded protons to unsubstituted quinones measured in alcoholic solvent, whereas the corresponding ͉T͉ values do not exceed 3 MHz (23,24,31,32). The T value measured for H3 is one of the largest measured so far for a proton hydrogen-bonded to a semiquinone. According to density functional theory calculations, the large T value of H3 is predicted to account for a short hydrogen-bond length, typically in the range of 1.3-1.4 Å (22). In this case, Sinnecker et al. (21,22,24) have shown that the point dipole model does not work due to the increased covalent character of the H-bond that is not covered in the point dipole approximation. A more reliable alternative approach is to evaluate H-bond distances from the nuclear quadrupole coupling constant of 2 H. As empirically proposed by Soda and Chiba (25) and Hunt and Mackay (26), it has been shown that the nuclear quadrupole coupling constant of 2 H nuclei H-bonded to semiquinones follows a r Ϫ3 (O-2 H) dependence of the form where a and b are empirical parameters (22). Using a ϭ 319 kHz and b ϭ 607 kHz Å 3 (20) (20). Assignment of Non-exchangeable Proton Couplings-In addition to H3, two non-exchangeable proton couplings H1 and H2 are clearly resolved in the HYSCORE spectra of MSQ D measured in 2 H 2 O. For their assignment, we rely on previous experimental and theoretical studies on phylloquinone (also called vitamin K 1 ) and menaquinone (vitamin K 2 ) radicals examined in liquid and solid organic solvents or in proteins (23,29,30,(33)(34)(35)(36)(37). Indeed, these quinones share the same naphthoquinone ring structure methylated at the second position but differ in their aliphatic side chain attached at the 3-position (see supplemental Fig. S1). It has been shown both experimentally and theoretically that the aliphatic side chain properties have only a weak influence on the proton hyperfine coupling tensors measured in organic solvents (23,33,34,38,39). From these previous studies, it is evident that the non-exchangeable proton couplings from H1 and H2 Ͼ 2 MHz originate from the ring methyl protons, from the ␤-methylene isoprenyl protons, or from ␣-protons directly attached to the quinone ring. Due to rapid rotation of the methyl group even at low temperature, methyl protons of vitamin K molecules have equal hyperfine tensors and give prominent ENDOR/ESEEM signals. They are characterized by an almost axial hyperfine tensor, a predominant isotropic hyperfine coupling value in the range of 6.8 -12.3 MHz, and a characteristic relative hyperfine anisotropy (A ʈ ϪA Ќ )/A iso in the range of 0.26 -0.45 with A ʈ Ͼ A Ќ Ͼ 0. The latter value increases up to 0.76 for the methyl protons of the asymmetrically bound ubisemiquinone Q A ⅐ Ϫ in the RC from R. sphaeroides (2). Based on these results, we assign H1 to the methyl protons of MSQ D, with A iso ϭ ϩ5.53 MHz and T ϭ ϩ1.25 MHz leading to a (A ʈ Ϫ A Ќ )/A iso value of ϳ0.68 (Fig. 5).
Hyperfine data for ␤-methylene protons and ring ␣ protons in semiquinones are less abundant. Unlike the methyl protons, the methylene protons are not expected to rotate freely (40). They are also characterized by a nearly axial hyperfine tensor with A ʈ Ͼ A Ќ Ͼ 0 (40,41). In contrast, a high degree of anisotropy is expected for a ring ␣ proton hyperfine tensor due to the short distance of the proton and the spin density (30,40). Because of the axial symmetry of its hyperfine coupling tensor, we tentatively assign H2 to one of the ␤-methylene isoprenyl protons with A iso ϭ ϩ0.96 MHz and T ϭ ϩ1.18 MHz (Fig. 5). Overall, the 1 H hyperfine coupling constants determined in this work are consistent with those estimated from our previous continuous wave X-band 1 H ENDOR studies of MSQ D (10,11).
Strongly Asymmetric Spin Distribution in MSQ D -The 1 H hyperfine coupling constants are an excellent probe of the asymmetry of the spin density distribution in the quinone. In particular, the predominant isotropic component, A iso , of hyperfine coupling to the methyl protons is directly proportional to the unpaired spin density in the orbital on the adjacent ␣-carbon ( C ), as described by the McConnell relation, A iso ϭ C B 2 /2, where B 2 has been taken in the range from 120 to 212 MHz (42). The corresponding value for the methyl protons of MSQ D (A iso ϭ 5.53 MHz) is the smallest one ever reported for these protons for vitamin K molecules bound to proteins or in protic solvents (23,29,30,33,34,36,38,39). In particular, the spin density on the C␣ at the 2-position, which is sensed by the methyl protons, is reduced by ϳ30% in MSQ D as compared with the symmetrically hydrogen bonded MSQ-4 prepared in 2-propanol (39). This decrease can be explained by a strong asymmetry of hydrogen bonding to the carbonyl oxygens of MSQ D in NarGHI, which leads to a redistribution of both the spin density and charges within the quinone ring (2,43,44). Indeed, a stronger hydrogen bond to the carbonyl oxygen O1, as compared with oxygen O4, is expected to lead to an increase of spin density on carbon 3 but a decrease of the spin density on carbon 2 as observed here for MSQ D . A similar but less pronounced spin density shift has been proposed for the menasemiquinone-9 in the Q A site of the RC from Rhodopseudomonas viridis for which the methyl protons isotropic hyperfine coupling constant is about 6.8 MHz (39). In contrast, the large isotropic constants for methyl protons of the MSQ-7 in the aa 3 menaquinol oxidase from B. subtilis (A iso ϳ11.0 MHz) (29) or of the radical anion of phylloquinone in the A 1 site of photosystem I from Thermosynechococcus elongatus (A iso ϭ 9.8 MHz) (30) suggest that a stronger hydrogen bond to oxygen O4 is formed as compared with oxygen O1, a strongly asymmetric binding mode reverse to that observed in NarGHI.
Model for MSQ D Binding to NarGHI and Mechanistic Implications-Altogether, the data inferred from the present work allow us to refine the MSQ D binding model previously proposed (13). We conclude that the asymmetrical spin density distribution in MSQ D is primarily due to the strong hydrogen bond formed to the O1 oxygen of MSQ D involving the exchangeable H3 proton coupling (Fig. 5). Hence, the latter appears mostly responsible for the transfer of spin density from the radical to the interacting nitrogen nucleus that was deduced from the measurement of a small 14 N isotropic hyperfine coupling of A iso ϳ0.8 MHz to MSQ D using HYSCORE spectroscopy. Based on the measurement of its nuclear quadrupole parameters by S-band HYSCORE spectroscopy ( ϭ 0.49, ϭ 0.50), this nucleus was assigned to the N ␦ imidazole nitrogen from the heme b D axial ligand His-66 (13). In addition, this model is consistent with the relatively small value found for this imidazole nitrogen that is close to that predicted by density functional theory calculations on imidazoles forming a strong in-plane hydrogen bond to one of the benzosemiquinone oxygen atoms (45). Finally, the presence of the positive charges on the nearby heme Fe 2ϩ ion may also contribute to the observed asymmetrical spin distribution in MSQ D .
Interestingly, a single strong and highly ordered H-bond to MSQ D was detected in the present work by using both 1 H HYSCORE and 2 H Q-band ENDOR spectroscopies. In the most recent studies of protein-bound semiquinones, the radicals appear to be coupled to at least two exchangeable protons assigned to those involved in H-bonds (20, 29, 46 -49). Onesided H-bond was resolved only for the photoaccumulated phyllosemiquinone A 1 ⅐ Ϫ in photosystem I (30,50). Based on their experimental and theoretical results, Niklas et al. (30) have shown that the single detected short ϳ1.64 Å H-bond can fully account for the observed asymmetry in the spin density distribution of the SQ in the A 1 site. Remarkably, both the H-bond length and the ϳ30% variation of the spin density on the C␣ at the 2-position of the phyllosemiquinone in the A 1 site with respect to the symmetrically hydrogen-bonded radical are comparable with the corresponding values measured for MSQ D . Thus, our results indicate that MSQ D most likely binds to the protein via a one-sided H-bond.
Consequently, our present work indicates that the O4 carbonyl oxygen of MSQ D is not protonated in this intermediate state, showing that at least one proton has to be released to the periplasm consecutively to the first electron transfer step to heme b D . This step likely involves Lys-86, a residue that is located at the protein surface, at the entrance of the Q D cavity, and that is required for correct binding of quinol analogues and for semiquinone detection at the Q D site of NarGHI (11,13,14). Thus, we speculate that Lys-86 could be a direct hydrogen bond donor to the quinol molecule, facilitating proton abstraction at the quinol O4 oxygen and, concomitantly, the first electron transfer step. This would be accompanied by a movement of Lys-86 away from the substrate, leading to an asymmetric binding mode of the semiquinone intermediate and allowing for proton release toward the periplasm (Fig. 5). Whether quinol deprotonation at the O1 oxygen is coupled to the first or the second electron transfer step remains unclear. Additional studies aimed at measuring the pH dependence of the redox reactions occurring at the Q D site will be useful to further understanding how quinol deprotonation and electron transfer are synchronized at the Q D site. Such studies are currently being performed in our laboratories.
Role of Protein Environment in Quinol Utilization and Semiquinone Stabilization-The high stabilization of MSQ D in NarGHI is directly related to the redox potentials of the redox transitions MQH 2 /MSQ (E m,7.5 ϭ Ϫ150 mV) and MSQ/MQ (E m,7.5 ϭ Ϫ40 mV), which are both thermodynamically favorable for electron transfer to the b D heme (10). We questioned whether the unusually high stabilization of the semiquinone state at the NarGHI Q D site can be related to the strongly asymmetric binding mode of MSQ D mainly due to the short in-plane H-bond formed to the radical. Remarkably, the two other protein-bound semiquinones with redox properties most strongly affected by the protein environment as compared with the corresponding species in alcoholic solvents are the very low potential (E m ϳϪ750 mV) phyllosemiquinone anion A 1 ⅐ Ϫ in photosystem I and the high affinity ubisemiquinone at the Q H site of cytochrome bo 3 , which has a high stability, albeit 10 times smaller than that measured for MSQ D . In both cases, the semiquinones interact with the protein environment in a very asymmetric manner (46,(51)(52)(53). This contrasts to the much lower stability of the more symmetrically bound semiquinones at the Q B site of photosynthetic bacterial RC (48,54) or at the Q i site in R. sphaeroides bc 1 complex (47,55). In addition, the strong H-bond to MSQ D is expected to withdraw electron density and stabilize the semiquinone form, thereby raising the redox potential of the second oxidation step from semiquinone to quinone, as experimentally measured. Consequently, we hypothesize that the atypical binding mode of MSQ D could strongly contribute to its unusual redox properties. In addition, other effects such as the electrostatic environment of the nearby protein should also be taken into account. Evaluating their respective contribution to the MSQ D redox properties requires further work. Finally, the functional implications of this high stabilization remain to be established.
It has been shown both experimentally and theoretically that the presence of bulky substituents on the quinone ring force hydrogen-bond formation out-of-plane, thereby increasing simultaneously the isotropic and anisotropic coupling of the hydrogen-bonded protons (32,56). Remarkably, measurement of a small A iso and a simultaneous large T value for the proton hydrogen-bonded to MSQ D indicates that the protein environment around the radical strongly constrains the geometry of the hydrogen bond by maintaining a short in-plane H-bond to the radical, thus leading to the observed peculiar hyperfine coupling characteristics. We have recently shown that an endogenous USQ D can also be stabilized at the NarGHI Q D site (12). It binds to the protein via an H-bond to the same nitrogen as menasemiquinone does, i.e. most likely His-66 N ␦ . The similar 14 N HYSCORE pattern observed for both radicals suggests that the H-bond involved in binding USQ D has similar characteristics to that detected in the present study. This provides further support for its involvement in ubisemiquinone stabilization at the Q D site. Finally, our work indicates that the protein environment counteracts the effect of the presence of bulky substituents to impose an atypical binding mode. Further work is in progress in our laboratories to evaluate the importance of the quinone substituents to accommodate and utilize various substrates at this Q-site and to stabilize semiquinone intermediates.
Concluding Remarks-From the experiments reported in this work, we conclude that MSQ D is involved in a single strong in-plane and highly ordered H-bond with a solvent exchangeable proton. This strongly asymmetric binding causes a shift of the electron spin density over the quinone ring consistent with the formation of a strong hydrogen bond to the quinone carbonyl oxygen O1. This peculiar binding mode could strongly contribute to the unusual redox properties of MSQ D .