Interactions of intermediate semiquinone with surrounding protein residues at the Q_H site of the wild-type and D75H mutant cytochrome bo₃ from Escherichia coli

Abstract

Selective ¹⁵N isotope labeling of the cytochrome bo₃ ubiquinol oxidase from E. coli with auxotrophs was used to characterize the hyperfine couplings with the side-chain nitrogens from R71, H98, and Q101 residues and peptide nitrogens from R71 and H98 residues around the semiquinone (SQ) at the high-affinity Q_H site. The 2D ESEEM (HYSCORE) data have directly identified the N_ε of R71 as an H-bond donor carrying the largest amount of the unpaired spin density. In addition, weaker hyperfine couplings with the side-chain nitrogens from all residues around the SQ were determined. These hyperfine couplings reflect a distribution of the unpaired spin density over the protein in the SQ state of the Q_H site and strength of interaction with different residues. The approach was extended to the virtually inactive D75H mutant, where the intermediate SQ is also stabilized. We found that the N_ε from a histidine residue, presumably H75, carries most of the unpaired spin density instead of the N_ε of R71, as in the wild-type bo₃. However, the detailed characterization of the weakly coupled ¹⁵Ns from selective labeling of R71 and Q101 in D75H was precluded by overlap of the ¹⁵N lines with the much stronger ~1.6 MHz line from quadrupole triplet of the strongly coupled ¹⁴N_ε from H75. Therefore, a reverse labeling approach, in which the enzyme was uniformly labeled except for selected amino acid types, was applied in order to probe the contribution of R71 and Q101 to the ¹⁵N signals. Such labeling has shown only weak coupling with all nitrogens of R71 and Q101. We utilize density functional theory based calculations to model the available information about ¹H, ¹⁵N and ¹³C hyperfine couplings for the Q_H site and to describe the protein-substrate interactions in both enzymes. In particular, we identify the factors responsible for the asymmetric distribution of the unpaired spin density and ponder the significance of this asymmetry to the quinone’s electron transfer function.

Cytochrome bo₃ ubiquinol oxidase (cyt bo₃) from Escherichia coli catalyzes the reduction of molecular oxygen to water using ubiquinol as the electron donor.¹ The enzyme located in the cytoplasmic membrane also functions as a proton pump, conserving much of the energy available from the redox reaction as the proton motive force.^2,3 Cyt bo₃ is a member of heme-copper superfamily. Of the four subunits of cyt bo₃, the catalytic subunit and two other subunits are analogous to the mitochondrially encoded subunits of aa₃-type cytochrome c oxidase (cyt aa₃).⁴ Despite having different electron donors, the proton pumps of cyt bo₃ and cyt aa₃ likely operate in a similar manner.⁵

Previous work has established that cyt bo₃ isolated in the detergent n-dodecyl β-D-maltoside (DDM) is associated with a tightly bound ubiquinone-8 (UQ₈) at the Q_H site, whereas purification in Triton X-100 yields the enzyme without any bound UQ₈.^6,7 The UQ₈ bound at the Q_H site does not exchange with the substrate ubiquinone pool during turnover. Hence, the Q_H site is distinct from the Q_L site, where the oxidation of substrate ubiquinol takes place.^7,8 The Q_H site has been shown to be able to stabilize the one-electron reduced semiquinone (SQ_H), which can be detected by EPR spectroscopy.^9,10 The comparison of the kinetics between the enzyme preparations with and without the tightly bound UQ₈ or with the Q_H site inhibitors led to the conclusion that the UQ₈ at the Q_H site facilitates the fast electron transfer process from the substrate ubiquinol to the low spin heme b.^6,11–13

A crystal structure of cyt bo₃ without a bound UQ₈ has been reported.¹⁴ Based on the crystal structure and mutagenesis studies, residues R71, D75, H98 and Q101 from subunit I were proposed to interact with the bound UQ₈ at the Q_H site (Fig. 1).^14,15 Mutating any of the four residues led to the loss of catalytic activity. Notably, the D75H mutant, despite its inability to reduce molecular oxygen, was found to stabilize a SQ radical with a midpoint potential similar to that of the wild type (WT) enzyme.¹⁵ Hence, an environment resulting in the stabilization of the SQ radical is necessary but not sufficient for proper function. A precise spatial arrangement of the SQ radical and the surrounding residues at the Q_H site is crucial for the successful electron transfer process.

Figure 1.

The current model of UQ at Q_H site of cyt bo₃. The strong H-bonds are shown in dashed lines and the weak one in a dotted line. The figure was generated according to the model based on the X-ray crystal structure by Abramson et al.¹⁴ Adapted from ref. (20).

Previous X-band 1D and 2D ESEEM experiments have offered insights into the nature of interactions between the SQ_H and the Q_H site residues (Fig. 1).^16–20 Hyperfine couplings with methyl group protons and exchangeable, hydrogen bonded protons, are consistent with a neutral SQ_H species in the WT cyt bo₃, indicating significant asymmetry in the distribution of the unpaired spin density.¹⁷ The ¹⁴N ESEEM spectra reveal one strongly coupled nitrogen participating in a hydrogen bond with the SQ_H.^16–18 In a previous communication, we reported selective ¹⁵N-labeling of cyt bo₃, identifying the N_ε of R71 as the strongly coupled nitrogen in the WT cyt bo₃ and have shown that the other nitrogens in R71, H98 and Q101 possess substantially smaller couplings.¹⁹ In contrast, a strongly coupled nitrogen that is observed in the ¹⁴N ESEEM spectra of the SQ_H radical of the inactive D75H mutant was shown to be from a different amino acid residue.¹⁸ The SQ_H is stabilized by hydrogen bonds different from the ones identified with the WT enzyme and its characteristics are shifted towards an anion-radical.

In the current work we provide the comparative analysis of the 2D ESEEM (HYSCORE) spectra of ¹⁵N uniformly labeled (¹⁵N-U) WT and D75H cyt bo₃. We report the complete results of selective ¹⁵N-labeling of the residues at the Q_H site in these two enzymes. In addition to the strongly coupled N_ε of R71, the weakly coupled N_ε of H98 was also found to carry unpaired spin density in the WT cyt bo₃. It is also shown that in the D75H mutant, the N_ε of H75 is the strongly coupled nitrogen, and the N_ε of R71 displays significantly weaker hyperfine interaction with the SQ_H. The principal values of the rhombic hyperfine tensors for the strongly coupled nitrogens in WT and D75H cyt bo₃ were determined using a newly developed method of HYSCORE cross-peak lineshape analysis. In addition to ¹⁵N couplings from the present work, ¹H and ¹³C couplings from previous studies of the WT and D75H cyt bo₃^17,18,20 were utilized in QM/MM calculations to calculate the spatial environment and electronic structure of the SQ_H. In particular, these calculations identify the factors responsible for the asymmetric distribution of the unpaired spin density and charge at the Q_H site and ponder the significance of this asymmetry to the quinone’s electron transfer function.

Materials and Methods

Materials

Isopropyl-β-D-thiogalactopyranoside (IPTG) was bought from Fisher Scientific (Pittsburgh, PA). n-dodecyl-β-D-maltoside (DDM) was purchased from Anatrace (Maumee, OH). ¹⁵NH₄Cl and ¹⁵N-enriched amino acids for the growth of E. coli to express isotope labeled cyt bo₃ were ordered from Cambridge Isotope Laboratories (Andover, MA). Other chemicals used in the preparation of growth medium and buffers were from Sigma-Aldrich (St. Louis, MO).

Bacterial Strains

E. coli C43(DE3) strains with deletions of genes involved in amino-acid biosynthetic pathways were constructed with the λ-Red recombination system as described previously.^20,21 The C43(DE3) auxotroph strains used in the current work are listed in Table 1.

Table 1.

The genotype of E. coli C43(DE3) auxotrophs used for the preparation of selectively ¹⁵N labeled cyt bo₃ samples and amino acids supplemented to the minimal medium to grow these auxotrophs.

E. coli C43(DE3) strains	Genes knocked out	Amino acids added to the minimal medium	Cyt bo3 samples prepared
C43(DE3)	none	none	Uniformly 15N labeled WT cyo bo3
CLY	cyo	none	Uniformly 15N labeled D75H cyo bo3
ML8	cyo, argH	105 mg/L Arg	WT cyt bo3 with 15Nη Arg WT cyt bo3 with 15Nα Arg WT cyt bo3 with 15N4 Arg D75H cyt bo3 with uniform 15N except for Arg D75H cyt bo3 with uniform 15N except for the Nε and Nα of Arg
ML17	glnA	750 mg/L Gln	WT cyt bo3 with 15Nε Gln
ML21	tyrA, hisG	18 mg/L Tyr, 15 mg/L His	WT cyt bo3 with 15N3 His WT cyt bo3 with 15Nδ His
ML22	cyo, ilvE, avtA, aspC, hisG	40 mg/L Ile, 40 mg/L Leu, 35 mg/L Val, 90 mg/L Tyr, 50 mg/L Phe, 40 mg/L Asp, 16 mg/L His	WT cyt bo3 with ring 15N2 His D75H cyt bo3 with 15N3 His D75H cyt bo3 with 15Nδ His
ML26	cyo, ilvE, avtA, aspC, hisG, argH	40 mg/L Ile, 40 mg/L Leu, 35 mg/L Val, 90 mg/L Tyr, 50 mg/L Phe, 40 mg/L Asp, 16 mg/L His, 105 mg/L Arg	D75H cyt bo3 with 15N4 Arg
ML30	cyo, ilvE, avtA, aspC, hisG, argH, glnA	40 mg/L Ile, 40 mg/L Leu, 35 mg/L Val, 90 mg/L Tyr, 50 mg/L Phe, 40 mg/L Asp, 16 mg/L His, 105 mg/L Arg, 750 mg/L Gln	D75H cyt bo3 with 15Nε Gln
ML32	cyo, argH, glnA	105 mg/L Arg, 750 mg/L Gln	D75H cyt bo3 with uniform 15N except for Arg and Gln
ML34	cyo, argH, glnA, hisG	105 mg/L Arg, 750 mg/L Gln, 16 mg/L His	WT cyt bo3 with uniform 15N except for Arg, Gln and His

Preparation of Amino-acid Selective Isotope Labeled Cyt bo₃

Selectively ¹⁵N labeled cyt bo₃ samples were expressed from the various C43(DE3) auxotroph strains as summarized in Table 1. Each auxotroph to be used as an expression host was transformed with the pET-17b vector (Novagen) encoding the WT or D75H mutant cyo operon engineered to encode cyt bo₃ with a 6xHis tag at the C terminus of subunit II.¹⁸ The enzymes overexpressed from the C43(DE3) cultures were solublized in DDM detergent, purified with Ni-NTA affinity chromatography and reduced with sodium ascorbate under anaerobic condition to generate the SQ_H radical as described previously.²⁰

EPR Measurements

The CW EPR measurements were performed on an X-band Varian EPR-E122 spectrometer. The pulsed EPR experiments were carried out using an X-band Bruker ELEXSYS E580 spectrometer equipped with Oxford CF 935 cryostats. Unless otherwise indicated, all measurements were made at 50 K. Some pulsed EPR measurements were also performed at higher temperatures, up to 120 K, and the results were similar to those obtained at 50 K. Several types of ESE experiments with different pulse sequences were employed with appropriate phase cycling schemes to eliminate unwanted features from the experimental echo envelopes. Among them are two-dimensional three-pulse and four-pulse sequences, which are described in detail elsewhere.¹⁷ Spectral processing of three- and four-pulse ESEEM patterns, including subtraction of relaxation decay (fitting by 3–6 degree polynomials), apodization (Hamming window), zero filling, and fast Fourier transformation, was performed using Bruker WIN-EPR software.

Some HYSCORE spectra are presented as contour plots in Matlab R14. For the 2D data representation, the echo decay was eliminated by low-order polynomial (up to fourth-order) base-line corrections in each dimension and apodizied with a Hamming window subsequently. Before the 2D Fourier transformation, the data was zero-filled up to a 1024 × 1024 matrix. The evaluated values were subsequently used to simulate the 2D HYSCORE spectra applying the Kazan Viewer package²² or home written PC software developed by Dr. A. Tyryshkin (now at Princeton University).²³

HYSCORE Spectra from ¹⁵N Nuclei

The experimental data regarding the ligand environment of the semiquinone were obtained in this work from the two-dimensional (2D) ¹⁵N ESEEM (HYSCORE) spectra.²⁴ The HYSCORE technique creates 2D spectra with off-diagonal cross-peaks. ¹⁵N nucleus with I=1/2 has two hyperfine frequencies ν_α and ν_β from opposite m_s=±1/2 electron spin manifolds. These may produce a pair of cross-features (ν_α, ν_β) and (ν_β, ν_α) in (++) quadrant, as well as another pair (−ν_α, ν_β) and (ν_α, −ν_β) in the (+−) quadrant. The appearance of cross-peaks in (++) or (+−) quadrants is governed by the relative values of hyperfine couplings ¹⁵A and Zeeman frequency ¹⁵ν_N.^25,26 Peaks in the (+−) quadrant come primarily from strong hyperfine interaction, i.e. |¹⁵A|>2(¹⁵ν_N), whereas the peaks in the (++) quadrant appear predominantly from interactions with |¹⁵A|<2(¹⁵ν_N). Peaks may appear in both quadrants simultaneously in the intermediate case when both parts of the inequalities are comparable.

Orientation-disordered (i.e. powder) 2D spectra of I=1/2 nuclei also reveal, in the form of cross-ridges contour projections, the interdependence between ν_α and ν_β values in the same orientation. The two coordinates of the arbitrary point at the cross-ridge, described in the first order by equation

$${\mathrm{\nu }}_{\mathrm{\alpha }(\mathrm{\beta })}=|{}^{15}{\mathrm{\nu }}_{\mathrm{N}}+(-)|{}^{15}A|/2|$$ (1)

can be used for the first-order estimate of the corresponding hyperfine coupling constant A:

$$\begin{array}{c}{\mathrm{\nu }}_{\mathrm{\alpha }}-{\mathrm{\nu }}_{\mathrm{\beta }}=|A|\text{ if }|{}^{15}A|<2({}^{15}{\mathrm{\nu }}_{\mathrm{N}})\hfill \\ {\mathrm{\nu }}_{\mathrm{\alpha }}+{\mathrm{\nu }}_{\mathrm{\beta }}=|A|\text{ if }|{}^{15}A|>2({}^{15}{\mathrm{\nu }}_{\mathrm{N}})\hfill \end{array}$$ (2)

On the other hand, analysis of the cross-ridges in (ν_α)² vs. (ν_β)² coordinates and spectral simulations allows in many cases for simultaneous determination of the isotropic a_iso and anisotropic T components of the hyperfine tensor.^25,26

Computational methods

All density functional calculations were performed using Gaussian 09.²⁷ All calculations including geometry optimization and hyperfine coupling were performed using the B3LYP functional and the EPR-II basis set. Specific details concerning hyperfine coupling calculations are as previously described.^28,29 Details of specific models used are given in the text.

Results and discussion

¹⁴N and ¹⁵N ESEEM spectra of Wild Type and D75H Mutant of Cyt bo₃

The interaction of the SQ_H with the protein environment in WT and D75H cyt bo₃ with natural abundance of nitrogen (¹⁴N isotope - 99.63 %) has been previously studied in detail by pulsed EPR spectroscopy.^16–18 1D and 2D ¹⁴N ESEEM spectra show the contribution from only a single nitrogen in each protein (Figs. S1 and S2). These possess different characteristics, i.e. quadrupole coupling constant (qcc) K=0.93 MHz, asymmetry parameter η=0.51, and hyperfine coupling ¹⁴A=1.8 MHz for WT cyt bo₃^16,17 and K=0.43 MHz, η =0.73, and ¹⁴A=2.7 MHz for D75H.¹⁸ The values of K and η characterize the chemical type and electronic configuration of ¹⁴N atom interacting with the SQ_H. For instance, K=e²qQ/4h=0.93 MHz, most closely corresponds to the nitrogen from an NH or NH₂ group.^17,18 This value is ~10% larger than the qcc for the peptide amide nitrogen and more than two times the qcc of the protonated imidazole nitrogens in histidine. Therefore, it was suggested that the most likely candidates for the H-bond donor in WT cyt bo₃ are the nitrogens from the side chains of R71 or Q101. Likewise, a protonated imidazole nitrogen of a histidine residue H75 or H98 was suggested as the H-bond donor in the D75H mutant.^{18 14}N spectra do not show any lines from other side-chain and peptide nitrogens of the nearby environment. These nitrogens are coupled more weakly and do not produce well defined lines in the ¹⁴N powder-type spectra, because of the nuclear quadrupole interactions (nqi) influence.^30,31 In contrast, the lines from weakly coupled nitrogens (N_wc) are well observed in 2D ¹⁵N ESEEM spectra, which are not complicated by nqi.

Two presentations of ¹⁵N HYSCORE spectrum of the SQ_H in ¹⁵N-U WT cyt bo₃ are shown in Fig. 2A,C (see also Fig. S3). The (++) quadrant exhibits a pair of intensive cross-peaks 1 located symmetrically around the diagonal point (¹⁵ν_N,¹⁵ν_N), along the antidiagonal with maxima at (2.74, 0.34) MHz, which correspond to hyperfine coupling ¹⁵A=2.4 MHz. This coupling is in agreement with the expected ¹⁵A of 2.5 MHz rescaled from the hyperfine coupling ¹⁴A of 1.8 MHz measured from ¹⁴N ESEEM spectra. In addition, the (++) quadrant contains feature N_wc with the maximum near the diagonal point (¹⁵ν _N,¹⁵ν_N) with the decaying shoulders symmetrically extended up to ~0.8 MHz along the antidiagonal.

Figure 2.

Contour (A,B) presentation of the ¹⁵N HYSCORE spectra of the SQ_H in ¹⁵N uniformly labeled WT (A) and D75H (B) cyt bo₃ (magnetic field 345.2 mT (A) and 346.1 mT (B), time between first and second pulses τ=136 ns, microwave frequency 9.702 GHz (A) and 9.704 GHz (B)). Stacked (C,D) presentation of the (++) quadrant of the WT (C) and D75H (D) spectra. Stacked presentation of the full (A) and (B) spectra is shown in Figure S4. Red dashed straight segments are defined by |ν₁ ± ν₂|= 2(¹⁵ν_N).

The ¹⁵N HYSCORE spectrum of the ¹⁵N-U D75H mutant (Fig. 2B,D) shows a pair of intensive cross-peaks 1 in the (+‒) quadrant with maxima at ~(±3.3, ‒/+0.4) MHz defining ¹⁵A=3.7 MHz or ¹⁴A=2.6 MHz, also consistent with the coupling estimated from ¹⁴N spectra. The (++) quadrant of the spectrum (Fig. 2B,D) exhibits also N_wc feature located around the (¹⁵ν_N,¹⁵ν_N) point. The N_wc feature in D75H has a different shape compared to that of the WT enzyme. For D75H, the N_wc feature is a triplet, including a central peak at (¹⁵ν_N,¹⁵ν_N) and two other lines symmetrically located around the antidiagonal with the splitting ~0.6 MHz.

The ¹⁴N and ¹⁵N ESEEM spectra show the presence of one strongly coupled nitrogen in the SQ_H environment, which is different in the WT and D75H cyt bo₃ proteins. In contrast, the N_wc features resolved in ¹⁵N spectra of both the WT and D75H mutant result from multiple non-equivalent contributions of weakly-coupled nitrogen nuclei from in the immediate vicinity of the SQ_H. The shapes of the N_wc features indicate differences in the individual interactions for the SQ_H of the WT and D75H mutant. To further resolve the interactions with nitrogens in the SQ_H environment, selective ¹⁵N labeling in different residues, as well as ¹⁵N uniform labeling was employed.

Selective ¹⁵N labeling of the WT cyt bo₃ protein

Arg, His and Gln were targeted for selective ¹⁵N labeling because the corresponding residues are involved in the current model of Q_H-site (Fig. 1). The molecular structure and atomic numbering of each of the three amino acids are displayed in Fig. S4. The following samples of WT cyt bo₃ were prepared with ¹⁵N labels as follows. 1) Arg: a) uniform labeling; b) the two N_η positions; c) the peptide position N_α; 2) His: a) uniform labeling; b) ring-¹⁵N (N_δ and N_ε); c) the N_δ position only; 3) Gln with ¹⁵N in the N_ε position; 4) uniformly labeled with ¹⁵N except for Arg, Gln and His. It is assumed that only R71, Q101 and H98 are significant in interpreting the interactions with SQ_H.

The following results were obtained.

1. As shown previously, a dramatic change of the ESEEM spectra, accompanied by the complete disappearance of the ¹⁴N peaks, is observed with the WT cyt bo₃ with uniformly ¹⁵N-labeled R71 (ref.19, Fig. S5A). The HYSCORE spectrum of the SQ_H in this protein contains two intense cross-peaks 1 similar to ones in the spectrum of ¹⁵N-U WT cyt bo₃ (Fig. 2A,C). In contrast, the N_wc feature observed with ¹⁵N-U R71 WT cyt bo₃ is very different (Fig. S5A), and resolves only a weak doublet centered around the (¹⁵ν_N, ¹⁵ν_N) diagonal point with the splitting ¹⁵A=0.15 MHz. A similar doublet was observed in the spectrum of the WT cyt bo₃ with the selectively labeled ¹⁵N_η ositions in R71 (ref.19, Fig. S5B), whereas no ¹⁵N resolved peaks were observed in a sample in which the peptide nitrogen N_α of R71 was selectively labeled with ¹⁵N. These observations show a weak interaction in the WT cyt bo₃ between the SQ_H with the ¹⁵N_η of R71, and confirm that the N_ε of R71 possesses the largest hyperfine coupling and is responsible for the ¹⁴N spectral features in the WT cyt bo₃.

2. It was previously shown that the spectrum of cyt bo₃ with uniformly ¹⁵N-labeled H98 shows the N_wc feature with a maximum at the (¹⁵ν_N,¹⁵ν_N) point accompanied by extended shoulders up to 0.6 MHz with very poorly resolved maxima.¹⁹ This line can be produced by the interactions with up to three ¹⁵Ns and additional selective ¹⁵N labeling was performed to clarify the origins of this feature. The spectra of the sample with ring-¹⁵N (N_δ and N_ε) labeled His is identical to that with uniformly ¹⁵N labeled His (Fig. S5C and Fig. S5D), but a very week antidiagonal ¹⁵N line is found in the spectrum of the protein with ¹⁵N_δ His. Thus, it is concluded that the extended ¹⁵N_wc feature is primarily due to the N_ε of H98.

It is noted that a peak of very low intensity located at the diagonal point (¹⁵ν_N,¹⁵ν_N) was previously observed for the WT cyt bo₃ with ¹⁵N labeled N_ε of Q101, indicating very weak dipolar interaction between the unpaired electron and a distant ¹⁵N nucleus.¹⁹ No ¹⁵N signal was found in the spectra of WT cyt bo₃ uniformly labeled with ¹⁵N except for Arg, Gln and His. Based on these observations, it is concluded that the only contributions to the N_wc feature are from R71, Q101 and H98.

Selective ¹⁵N labeling of D75H

Selectively ¹⁵N labeled D75H cyt bo₃ samples were also examined. The ¹⁴N signals completely disappeared in the HYSCORE spectrum of the D75H with uniformly ¹⁵N(3)-labeled His (Fig. S6). This result is consistent with the prediction, based on the value of the qcc, that the ¹⁴N ESEEM spectrum of the D75H mutant arises from the protonated imidazole ring nitrogen of a histidine residue, presumably either H75 or H98.¹⁸ In order to definitively identify whether the N_δ or N_ε of His contribute to the ¹⁴N ESEEM spectra, a D75H with ¹⁵N_δ His was prepared. The ¹⁴N HYSCORE spectrum of this sample is identical to the spectrum of unlabeled D75H in Fig. S2, so it can be concluded that these spectroscopic features originate from the N_ε of a His residue.

The N_wc feature in the HYSCORE spectrum from the enzyme with uniformly labeled ¹⁵N(3) His possesses the maximum intensity at the (¹⁵ν_N,¹⁵ν_N) diagonal point and shoulders extended up to 0.5 MHz in both directions from the diagonal (Fig. 3B). However, the total intensity of the diagonal peak and shoulders is substantially lower than in the ¹⁵N-U enzyme (Fig. 3A), using the cross-peaks 1 as the reference for the intensity comparison. The larger intensity and different shape of the N_wc feature that is only observed in the uniformly labeled ¹⁵N-U D75H mutant, must be due to the additional contributions from nitrogens in R71 and Q101 residues.

Figure 3.

N_wc feature in ¹⁵N HYSCORE spectra of SQ_H in ¹⁵N -U D75H (A), D75H with uniformly ¹⁵N(3) labeled His (B), (C) obtained as a difference between the spectra of ¹⁵N -U D75H and ¹⁵N-U D75H except Arg, (D) obtained as a difference between the spectra of ¹⁵N-U D75H except Arg and D75H with uniformly ¹⁵N(3) labeled His.

Unfortunately, the HYSCORE spectra collected from the D75H selectively labeled with ¹⁵N Arg or ¹⁵N Gln do not resolve the signals from the weakly coupled ¹⁵N because any such signal is overshadowed by much more intense peak ν₊ =1.61 MHz from the strongly coupled ¹⁴N_ε of His (Fig. S7). In order to visualize the N_wc from R71 and Q101 in the D75H mutant, a reverse labeling approach was employed to prepare ¹⁵N-U D75H except for Arg. The contributions of the nitrogen interactions from R71 and Q101 to the ¹⁵N spectrum were then obtained using the spectra of (a) ¹⁵N-U D75H; (b) D75H with ¹⁵N-His(3); and (c) ¹⁵N-U D75H with ¹⁴N Arg. The subtractions a–c and c–b give the ¹⁵N_wc difference spectra from R71 and Q101, respectively (Fig. 3 C,D). The difference spectrum isolating the contribution from R71 shows extended shoulders with the maximum corresponding to the splitting ~0.6 MHz. The difference spectrum showing the contribution from Q101 consists of a weak peak at the diagonal point (¹⁵ν_N,¹⁵ν_N). In addition we have prepared the ¹⁵N-U D75H except for the N_ε and N_α of Arg. The shape of N_wc for this sample is similar with one for D75H with ¹⁵N-His(3). This result indicates that the cross-peaks with the splitting 0.6 MHz are produced by the N_ε of R71 that forms H-bond with SQ_H in D75H as well.

In contrast to the WT cyt bo₃, there are two histidines, i.e. H75 and H98, around the SQ_H in the D75H mutant. Experimental ^14,15N spectra do not give any indication which of them carries largest spin density on the N_ε atom. However, in our previous work we have provided arguments based on the comparison of the hyperfine couplings with methyl protons reflecting asymmetry in spin density distribution with the Q_A site SQ of the reaction center that stronger interaction with the N_ε of H75 is more preferable.¹⁸ Taking into account this assignment one can conclude that the N_wc pattern in D75H with ¹⁵N-His(3) includes contributions from N_δ of H75 and N_δ and N_ε of H98. The N_wc shoulders in this sample (Fig. 3B) show slight increase of intensity corresponding to the splitting ~0.8±0.2 MHz. We suggested that the N_ε from H98 forms the H-bond with the SQ_H similarly with the WT cyt bo₃ and produces this splitting. The N_δ s in H75 and H98 separated by two bonds from H-bonded N_εs would have much smaller couplings and would contribute to the central part of the N_wc line around the diagonal point. This suggestion is supported by about 1/20 ratio of the hyperfine couplings for the remote and coordinated imidazole nitrogens in complexes with metals and clusters.³² We also suggested that N_αs from both His residues produce negligible influence on the spectra similarly as in the WT protein.

Hyperfine tensors of strongly coupled nitrogens

So far in our description of the experimental spectra we have used the hyperfine couplings determined from the position of cross-peak maxima using first-order expressions for two nuclear frequencies (Eqs.1, 2) of ¹⁵N nuclei. However, the lineshape of the cross-peaks from strongly coupled nitrogens in the WT and D75H cyt bo₃ allows one to determine all of the principal values of hyperfine tensors. The cross-peaks from these nuclei possess the horn-like lineshape that indicates a significant rhombicity of the corresponding hyperfine tensors.^25,26 The principal values of the rhombic hyperfine tensor can be defined as follow: A_x = a_iso − T (1 + δ), A_y = a_iso − T (1 − δ), A_z = a_iso + 2T with 0 ≤ δ ≤ 1, where a_iso, T are the isotropic and anisotropic components of hyperfine coupling and δ is a rhombic parameter. The two nuclear frequencies of ¹⁵N (I = 1/2) from opposite m_S = ±½ electron spin manifolds for each principal value i = x, y, z are ν_αi = |¹⁵ν_N + |A_i|/2| and ν_βi = |¹⁵ν_N − |A_i|/2|. An estimate of the principal components can be performed using theoretical predictions of the lineshape of the cross-peaks in powder-type spectra. The borders of the ideal cross-peak horn in such spectra are formed by three arc-type ridges between the pairs of three points (ν_αx, ν_βx), (ν_αy, ν_βy) and (ν_αz, ν_βz) located on the |ν₁±ν₂|=2(¹⁵ν_N) lines. The shape of these ridges is described by the general equation (where Q and G are coefficients which are functions of a_iso,T, δ and ¹⁵ν_N)^25,26:

$${{\mathrm{\nu }}_{\mathrm{\alpha }}}^2=\mathrm{Q}{{\mathrm{\nu }}_{\mathrm{\beta }}}^2+\mathrm{G}$$ (3)

The arc-type ridges transform to straight segments in ((ν₁)² vs. (ν₂)²) plots, producing a triangle lineshape of the cross-peak with triangle vertexes at ((ν_α(β)x)², (ν_β(α)x)²), ((ν_α(β)y)², (ν_β(α)y)²), and ((ν_α(β)z)², (ν_β(α)z)²).²⁵ It should be noted that HYSCORE intensity at (ν_αx, ν_βx), (ν_αy, ν_βy) and (ν_αz, ν_βz) points corresponding to orientations of magnetic field along the principal directions (x, y, z) of the hyperfine tensor is equal to zero and significantly suppressed in the orientations around principal directions.²⁶ Therefore, in HYSCORE spectra only the central part of the border cross-ridges, which correspond to orientations of the magnetic field substantially different from the principal directions, will possess substantial intensity.²⁶ It means that in real spectrum the cross-peak borders should not cross the |ν₁±ν₂|=2(¹⁵ν_N) line(s) and the crossing points (ν_αx, ν_βx), (ν_αy, ν_βy) and (ν_αz, ν_βz) can be obtained through the linear regression of the observed parts of border arcs in ((ν₁)² vs. (ν₂)²) presentation of the spectrum.

Fig. 4 (top) shows a presentation of the (++) quadrant of the ¹⁵N HYSCORE spectrum of the SQ_H in WT cyt bo₃ in ((ν₁)² vs. (ν₂)²) coordinates. The borders of the cross-peaks can be estimated by the area of the sharp increase of the peak intensity, i.e. where the blue background transformed into colored area in Fig. 4. In the spectrum shown in Fig. 4 (bottom), only one border segment with a well-defined linear portion can be defined (pink line segment). Linear regression of this segment (pink line in Fig. 4, bottom) gives two crossing points ((ν_βx)², (ν_αx)²) and ((ν_βz)², (ν_αz)²), corresponding to the minimal and maximal principal values. The larger coordinate of the crossing points is estimated by the values 5.5–5.8 MHz² and 9.7–10.0 MHz² that defines ν_αx =2.3–2.4 MHz or |A_x|=1.7±0.1 MHz and ν_αz=3.1–3.2 MHz or |A_z|=3.3±0.1 MHz, respectively. The intermediate principal value was determined from the simulations of ¹⁵N HYSCORE spectra. Simulations varying the tensor anisotropy from axial (δ=1) to completely rhombic (δ=0) show that best agreement in the cross-peaks 1 location (Fig. S8) was achieved with δ=0.64±0.05, and defines the complete hyperfine tensor as |A_z|=3.3 MHz, |A_y|=2.3 MHz |A_x|=1.7 MHz (±0.1MHz) with a_iso=2.42 MHz and T=(0.88, −0.16, −0.72) MHz (signs a and T components are relative). All principal values A_i should have the same sign. Only this selection correctly describes the location of the cross-peak and gives the isotropic coupling consistent with the values estimated from ¹⁴N and ¹⁵N spectra. This tensor is assigned to the N_ε of R71. A similar analysis was performed for cross-peaks 1 in the (+−) quadrant of the ¹⁵N HYSCORE spectrum of the SQ_H in the D75H mutant (see Figs. S9 and S10). Determined principal values of the hyperfine tensor assigned to the N_ε of H75 are shown in Table 2.

Figure 4.

(top) Contour presentation of the (++) quadrant from the ¹⁵N HYSCORE spectrum of the SQ_H in ¹⁵N uniformly labeled WT cyt bo₃ (Fig.2A) in ((ν₁)² vs. (ν₂)²) coordinates. The red curve is defined by |ν₁ ± ν₂|= 2(¹⁵ν_N).

(bottom) Analysis of the cross-peak 1 contour lineshape. Linear regression of the border segment (pink line) gives two crossing points ((ν_βx)², (ν_αx)²) and ((ν_βz)², (ν_αz)²) with |ν₁ ± ν₂|= 2(¹⁵ν_N) curve corresponding to minimum and maximum principal values of hyperfine tensor.

Table 2.

Hyperfine tensors of ¹⁵N nuclei at the Q_H-site of cyt bo₃ proteins.

Cytochrome bo3	Residue	Nitrogen	Hyperfine tensors (MHz)
Wild-type	R71	Nε Nη Nα	aiso=2.42, T=(0.88, −0.16, −0.72) aiso ~0.15, T<0.05–0.1 ~0
	H98	Nδ Nε Nα	~0 aiso=0.3, T~0.3–0.4 ~0
	Q101	Nε	Weak dipolar coupling <0.05–0.1 MHz
D75H	H75	Nε	aiso=3.5, T= (0.9, −0.2, −0.7)
	H98	Nε Nδ	aiso=0.8, |T| ~ 0.3–0.4
	R71	Nε Nη	aiso=0.6, |T| ~ 0.4–0.5 ~0
	Q101	Nε	Weak dipolar coupling <0.05–0.1 MHz

Hyperfine tensors of other nitrogens

In addition simulations of the ¹⁵N HYSCORE spectra were used to estimate the isotropic and anisotropic components of the hyperfine tensors for weakly coupled nitrogens, i.e. N_η of R71 and N_ε of H98 in the WT cyt bo₃ and N_ε of H98 and N_ε of R71 in the D75H cyt bo₃. The results of simulations are provided in Table 2. The N_wc spectra are extended along antidiagonal of the (++) quadrant. They possess narrow linewidth in the direction normal to the antidiagonal and symmetrical line shapes with the cross-peak maximum corresponding to the undefined orientation of the magnetic field relative to the principal axes. The intensity is suppressed at the cross-peak wings corresponding to field orientations along or near the axes with maximum and minimum principal values of the tensor. Our previous analysis of the ¹⁵N HYSCORE spectra²³ and simulations have shown that the position of the maximum gives accurate estimate of the isotropic coupling (~±0.05 MHz) but the relative signs of the isotropic and anisotropic components and symmetry of the tensor (i.e., axial or rhombic) are uncertain from the N_WC lineshape. This uncertainty influences the accuracy in the anisotropic tensor estimate in larger degree, i.e. for T~0.3–0.4 MHz, and the accuracy in its determination is ~0.1–0.15 MHz.

Density Functional Studies

Previous modeling studies used water molecules and N-methylformamide groups as hydrogen bond donors to the O1 and O4 atoms of the SQ_H and did not take into account the varying strength of hydrogen bond interaction with the SQ_H by different groups.^33–35 In our work, based on the X-ray structure, Fig. 1, and the EPR and mutational data for the WT enzyme, the SQ_H was modeled to have hydrogen bonds to the OH of the carboxylic acid group of D75, the N_εH of the imidazole group of H98 and the N_εH group of the guanidium group of R71. In the absence of accurate data from the X-ray crystal structure, we explored idealized small models with geometry optimization. Therefore, the overall significance and relative strength of each interaction can be assessed by the spin density distribution, but the correct orientation of the hydrogen bonds with the SQ_H will not be well reproduced.

Computed Geometries

Table 3 shows the calculated geometry of the isolated SQ_H and the effect of hydrogen bonding on this geometry in the WT and D75H models. For the WT model, the hydrogen bond lengths of O1 with the R71 N_εH group (1.65 Ǻ) and the D75 carboxylic acid group (1.60 Ǻ) are considerably shorter than those formed at O4 with the imidazole NH representing H98 (1.83 Ǻ). These bond distance trends suggest stronger hydrogen bonding to the O1 atom of the SQ_H caused principally by the strong interaction with the carboxylic acid group and the positively charged guanidium group. The NH group of the H98 imidazole is a weaker hydrogen bond donor. For the D75H model, the hydrogen bond length to the imidazole NH group from H75 is also significantly longer (1.79 Ǻ) than that found for the carboxylic acid group of D75 in the WT model (1.60 Ǻ) suggesting a weaker hydrogen bond formed by the imidazole group in the mutant.

Table 3.

Optimised geometries. All distances in angstroms.

	Isolated	WT	D75H
C1-O1	1.27	1.30	1.30
C4-O4	1.27	1.27	1.27
O1-H (R71)	-	1.65	1.61
O1-H (D75/H75)	-	1.60	1.79
O4-H (H98)	-	1.83	1.82

Spin densities and populations

Fig. 5 shows the spin density distribution changes that occur on going from the isolated SQ to the WT and D75H models. Fig. 6 provides a more quantitative picture using spin populations obtained from a Mulliken population analysis. The spin density distribution of the isolated SQ is symmetric whereas, in contrast, it is highly asymmetric for both the WT and D75H. For simpler semiquinone models, the primary effect of hydrogen bonding to either of the semiquinone carbonyl oxygen atoms is a redistribution of spin density from the oxygen atom position to the adjacent carbon.³⁶ In our WT model (Fig. 6), the spin density redistribution is mainly from O1 to C1 with a much smaller redistribution occurring between C4 and O4. The charged guanidinum and carboxylic acid groups polarize the C1-O1 bond. The increased spin population at C1 leads via spin polarization to a lower spin population at positions C2 and C6 and a higher spin population at positions C3 and C5. This “domino” spin polarization effect should lead to a significantly lower spin population at C4, but this is offset by the presence of a hydrogen bond from the imidazole group of H98 to the O4 atom, which partially balances the spin polarization effect on C4. The spin populations obtained for WT can, therefore, be explained by the presence of strong hydrogen bonding to the O1 atom of the SQ_H and a significantly weaker hydrogen bond to the O4 atom, in accord with EPR studies. For the D75H model, the spin populations are slightly less asymmetric compared to the WT model. As noted above the hydrogen bond made by the imidazole is longer than that calculated for the carboxylic acid group suggesting a weaker hydrogen bond is formed by this group. The predicted spin density distributions are also in accord with the lower 5-CH₃ ¹H and ¹³C hyperfine couplings measured for the D75H mutant, as discussed below.

Figure 5.

Spin density contour plots (0.004e/au) for (a) Isolated, (b) WT and (c) D75H models. Atom numbering is as shown in Fig. 1.

Figure 6.

Mulliken spin populations for (a) Isolated, (b) WT and (c) D75H models. Atom numbering is as shown in Fig. 1.

Comparison of calculated and experimental hyperfine couplings

In addition to the spin density distribution, hyperfine couplings can be calculated and compared with available ¹⁵N, ¹H and ¹³C experimental values. Table 4 shows the ¹⁵N isotropic and anisotropic hyperfine couplings calculated for the WT and D75H models. In addition, the calculated ¹⁴N qcc, K, and asymmetry parameters, η, are compared with experimental values. The experimental ¹⁵N isotropic hyperfine couplings for the R71 N_ε and N_η (2.4 MHz and 0.15 MHz, respectively), as well as K (0.93 MHz) and η (0.51) for the R71 N_ε are excellently reproduced by the WT model. For the H98 imidazole NH, the calculations give a_iso = 0.4 MHz, in reasonable agreement with the experimental value of 0.3 MHz. The value of the isotropic hyperfine couplings for these nitrogens is very sensitive to the angle made by the NH donor with respect to the SQ ring plane.³⁷ The very close correspondence between calculated and experimental value for the WT model suggests that the orientation of these hydrogen bond donors in the optimized model is very similar to that adopted in the actual Q_H binding site.

Table 4.

¹⁵N isotropic, a_iso, and anisotropic, T_ii, hyperfine tensors calculated for the Q_H site models. Calculated values for the ¹⁴N Nuclear Quadrupole Coupling Constant (K) and the asymmetry parameter, η are also given. Experimental values are given in brackets. All values given in MHz.

	WT Model				D75H
Pos.	TzzTyyTxx	aiso	K	η	TzzTyyTxx	aiso	K	η
R71 - Nε	0.6(0.9) −0.3(−0.6) −0.3(0.3)	2.4 (2.4)	0.9 (0.9)	0.7 (0.5)	0.0 0.0 0.0	1.4, 0.5* (0.6)	0.9	0.8
R71 - Nη	0.0 0.0 0.0	0.2 (0.15)	1.1	0.4	0.0 0.0 0.0	0.0 (0.0)	1.1	0.5
R71 - Nη	0.0 0.0 0.0	0.0	1.2	0.2	0.0 0.0 0.0	0.0 (0.0)	1.2	0.3
H98 - N	0.4 −0.1 −0.3	0.4 (0.3)	0.6	0.5	0.5 −0.2 −0.3	0.6 (0.8)	0.6	0.6
H75 - N	-	-	-	-	0.4, 0.7*(0.9) −0.2,−0.4*(−0.7) −0.2,−0.3*(−0.2)	0.8, 2.5* (3.5)	0.5(0.4)	0.6 (0.7)

^*Values using adjusted orientation, see text for details.

For the D75H model, the optimized geometry gives rise to ¹⁵N isotropic hyperfine couplings which deviate somewhat from the experimental determinations. The R71 N_ε has a calculated value of 1.4 MHz whereas the experimental assignment is 0.6 MHz. The calculated value for the N_εH group of H75 is 0.8 MHz whereas the experimental assignment is 3.5 MHz. For the H98 imidazole group, the calculated value, 0.6 MHz, is in good agreement with the experimental assignment of 0.8 MHz. In the optimized D75H model, the imidazole NH group is oriented 42° out of the SQ ring plane. The sensitivity of the ¹⁵N value to this orientation is demonstrated by changing the orientation of the H75 NH group from the optimized value of 42° to 90°. This changes the calculated ¹⁵N isotropic hyperfine coupling from 0.8 to 2.5 MHz, much closer to the experimentally measured 3.5 MHz. Likewise changing the out of plane orientation of the guanidium group N_εH from the optimized value of 41° to 35 ° changes the ¹⁵N isotropic coupling from 1.4 MHz to 0.6 MHz in exact agreement with the experimental assignment. For the H75 imidazole NH group the calculated ¹⁴N qcc K has a value of 0.5 MHz and the asymmetry parameter, η is calculated to be 0.6. These are in excellent agreement with our experimental determinations of 0.4 MHz and 0.7, respectively. Our conclusion, therefore, is that the D75H model is a good representation of the SQ_H in the mutant but the orientation of the hydrogen bond donors is different from the optimized small model calculation.

Calculated hyperfine couplings can also be compared with previous experimental determinations of ¹H hyperfine couplings for the WT and D75H mutant. Rotating methyl groups are observed readily using ENDOR spectroscopy as they give a strong ENDOR response and have been used as a primary marker in gauging the spin density asymmetry within SQs in photosynthetic reaction centers.³⁸ The 5-CH₃ ¹H hyperfine coupling has been measured in numerous studies and indeed is the principal indicator that the spin density of the WT SQ_H is highly asymmetric compared with the radical generated in vitro. For SQ in a non-polar solvent, and, therefore, not involved in hydrogen bonding, a_iso is 6.0 MHz.³⁸ Our calculated value for the isolated model is 5.9 MHz, which is in excellent quantitative agreement with this determination. For our WT model of SQ_H, the calculated ¹H a_iso value is 9.2 MHz, which is in good accord with experimental studies which vary from 9.5–10.0 MHz.³⁴ The elevated hyperfine coupling for the 5-CH₃ position is a result of the elevated spin population calculated for the ring C5 position shown in Fig. 6. Previous DFT studies^33–35 using SQ anion models have been unable to match the experimental value for this position and this can be mainly attributed to the use of water or amide groups as the hydrogen bonding groups in these modeling studies. Because the elevated 5-CH₃ hyperfine coupling was similar to that found for neutral semiquinone free radicals, it has been proposed that the Q_H site SQ is a neutral radical.^17,18 No evidence for formation of the neutral SQ form has been obtained with the current computational model although this cannot be ruled out with a different larger model of the Q_H site. The asymmetry in spin density distribution for both scenarios is very similar. Our current computational studies indicate that the elevated 5-CH₃ coupling arises from strong hydrogen bonding of a semiquione anion radical to a positively charged guanidium group and a carboxylic acid group, i.e. R71 and D75 in the WT SQ_H.

For the D75H model, the calculated 5-CH₃ isotropic hyperfine coupling of 8.2 MHz is lower than that of the WT model and is in excellent agreement with the value found experimentally for the D75H mutant of 8.0 MHz.¹⁸ The lower value for this hyperfine coupling, compared to the WT, can be ascribed to the more symmetric spin density distribution for the D75H model compared with the WT (Fig. 5 and 6). Replacement of the carboxylic acid group of D75 with the imidazole group of H75 results in significantly weaker hydrogen bonding to the O1 atom of SQ_H. The more symmetric spin density distribution leads to a lower spin density at C5 and a lower ¹H hyperfine coupling value for the 5-CH₃ group compared with the WT model. The good reproduction of this well characterized coupling suggests that the spin density distribution of the D75H model is reasonably accurate.

More recently, 2D ESEEM studies have obtained ¹³C isotropic hyperfine coupling values for the 5-CH₃ carbon nucleus for both WT and D75H samples.²⁰ For the WT and D75H models, the calculated isotropic values −5.3 MHz and −4.4 MHz are in good agreement with the experimental values of −6.1 MHz and −4.7 MHz, respectively.

¹H hyperfine couplings have also been resolved from exchangeable, presumably hydrogen bonded, protons in both the WT and D75H mutant. Compared with methyl group hyperfine couplings, hydrogen bonded protons are difficult to interpret in complex systems where there can be significant overlap of spectral lines. For the WT SQ_H, 2D ESEEM identified three protons having isotropic and anisotropic values (a_iso, T) of (−/+ 0.7, ± 6.3) MHz; (−/+ 1.2, ±4.2) MHz; (−/+ 4.2, ±1.7) MHz.¹⁷ The calculated values of (a_iso, T=T_11(max)/2) for the three hydrogen bonded protons from D75, R71 and H98 are, respectively: (0.0, 4.5) MHz, (−0.1, 4.3) MHz and (−0.6, 4.1) MHz (Table 5). The good agreement between the calculated R71N_ε, ¹⁵N a_iso, value and experimental value indicates that the NH orientation relative to the SQ_H ring plane is accurately modeled for the WT. The most likely candidate for the strong H-bond observed experimentally is the COOH group of D75, which exhibits the shortest optimized hydrogen bonding distance. In the optimized model of the WT, this is oriented 40° out of the ring plane. By changing this orientation to 90° we calculate that the anisotropic coupling increases to 5.4 MHz with an isotropic coupling value of 0.8 MHz, which is in better agreement with the experimental values. Therefore, we assign the largest (−0.7, +6.3) MHz proton hyperfine coupling to the COOH proton of D75. The (−1.2, +4.2) MHz coupling could be due to an overlap of the H98 NH and R71 N_εH lines. None of our calculated values match the very unusual coupling of (−/+ 4.2, ±1.7) MHz. The unusually large isotropic coupling together with the relatively small anisotropic component would suggest that this coupling does not arise from a hydrogen bonded proton, and further experimental characterization is required.

Table 5.

5-Methyl and Hydrogen bonded ¹H isotropic, a_iso, and anisotropic, T_ii, hyperfine tensors calculated for the Q_H site model. Experimental values are given in brackets. All values are given in MHz.

	Isolated		WT		D75H
Pos.	TzzTyyTxx	aiso	TzzTyyTxx	aiso	TzzTyyTxx	aiso
5- CH3	2.6 −1.3 −1.3	5.9 (6.0)	3.0 −1.5 −1.5	9.2 (9.5–10.0)	2.7 −1.4 −1.4	8.2 (8.0)
R71 - NHε	-	-	8.6 −4.1 −4.5	−0.1	8.4 −4.1 −4.3	−0.2
R71 - NHη	-	-	1.8 −0.6 −1.2	0.0	1.8 −0.6 −1.2	0.0
R71 - NHη	-	-	0.8 −0.3 −0.5	0.0	0.8 −0.3 −0.5	0.0
H98 - NH	-	-	8.2 −3.9 −4.2	−0.6	8.3 −4.0 −4.3	−0.7
D75/H75 – COOH/NH	-	-	8.9 −4.4 −4.5	0.0	6.5 −3.2 −3.3	−0.4

For the D75H mutant exchangeable protons have been experimentally assigned to ¹H hyperfine couplings (a_iso, T) of (−/+ 1.0, ±4.6) MHz and (−/+ 4.3, ±1.2) MHz. Three hydrogen bond donors are present in our D75H model: R71 N_ε ¹H, H75 N¹H and H98 N¹H (Table 5). The calculated (a_iso, T) values of each of these are in reasonable agreement with the (1.0 MHz, +4.6 MHz) experimental value, so all three may contribute to this experimentally determined coupling. As found with the WT model, the other experimentally observed coupling with a large isotropic and small anisotropic component, (−/+ 4.3, ±1.2) MHz, has no counterpart in the calculated values and is unlikely to be due to a proton(s) hydrogen bonded with carbonyls. One can suggest that exchangeable resonances with weaker hfi couplings observed in the spectra result from the overlap of the signals from the several protons in the SQ environment. Overlap of the cross-peak with different (but weak) hyperfine couplings would give cross-peaks with the shape and length substantially different from contributing signals. As the result the analysis would give some “effective” couplings not related directly to the real values. We suggest that further experiments using fully deuterated protein and cofactor as well as HYSCORE and ENDOR experiments at Q-band will better resolve these couplings, which would identify their source. Q-band will also address specific questions about the orientation of the H-bonds around the SQ_H site based on the hyperfine (and nuclear quadrupole) tensors of exchangeable protons (deuterons) that would lead to more adequate computational models.³⁹

Relevance to quinone one-electron transfer role

For in vitro systems, in protic solvents, quinone reduction or quinol oxidation proceeds via sequential loss or gain of two electrons via the semiquinone free radical form with consequent loss or gain of two protons. In biological systems, quinones can function additionally as one-electron transfer agents. For example the Q_A-site quinone in Type II photosynthetic reaction centers acts as a one-electron transfer agent between a pheophytin molecule and the quinone reductase site Q_B.^28,38 In the Photostsystem I reaction center, a phylloquinone molecule, A₁, also acts as a one-electron transfer agent between a chlorophyll molecule and an Fe₄S₄ center.^40,41 For Q_B, which undergoes two electron reduction and subsequent protonation to the quinol, as in in vitro systems, a more symmetric spin density distribution is found.³⁸ The Q_B quinone is in dynamic equilibrium with a quinone pool and binds to receive two electrons and two protons before unbinding in the reduced quinol form. It is noteworthy that for the quinones functioning as one-electron transfer agents, an extremely asymmetric spin density distribution has been found for the SQ form, whereas in a true substrate binding quinone site, such as Q_B, a more symmetric spin density has been found i.e. similar to that found in protic solvents. To fulfill the role of a one-electron transfer agent it may be a requirement of the quinone binding site to produce a highly asymmetric spin density distribution for the SQ intermediate. The asymmetry in the spin density distribution for the semiquinone reflects a more contracted/localized electron density distribution for the singly occupied molecular orbital, SOMO, of the free radical anion. This is clearly shown by the spin density contours in Fig. 5. This contracted electron density distribution would be expected to lead to a less stable SQ form compared with the symmetrical form but would also imply that further one electron reduction to the two electron reduced dianion or quinol form would be unfavourable as now two electrons have to occupy the contracted asymmetric orbital. While the exact mechanism of electron transfer and reduction at the Q_H site is still unknown, it is believed to shuttle electrons from a true quinone substrate binding site Q_L to heme b for eventual reduction of oxygen at the heme o₃-Cu_B active site.^7,13 It would appear to perform a similar function as the Cu_A site in cytochrome c oxidase. As for the Q_A and A₁ quinones described above, the high asymmetry in spin density distribution observed for SQ_H, would indicate that it performs a similar one-electron transfer role and that high asymmetry in the SOMO electron density distribution may be a crucial factor in determining the quinone role as an effective one-electron transfer agent. In the present study therefore one possibility for the non-activity exhibited by the D75H mutant could be related to the more symmetrical spin density distribution of the Q_H site SQ compared with the WT. The asymmetry in spin density distribution is still quite large however for the D75H mutant, compared to the symmetrical situation. This would argue against the above hypothesis, unless rapid one-electron transfer requires the extreme asymmetry which is present in the WT sample and also present in the Q_A and A₁ sites.

Conclusions

In this study, the pulsed EPR experiments performed with the selectively ¹⁵N labeled cyt bo3 samples have led to the unambiguous assignment of the N_ε of R71 as the nitrogen that is strongly coupled to the Q_H-site SQ in the wild-type enzyme and gives rise to the observed ¹⁴N ESEEM signals. In addition, selective ¹⁵N labeling of cyt bo₃ has enabled the detection of even weak interactions between the SQ_H and individual residues at the ubiquinone binding site as summarized in Table 2. The HYSCORE data suggest that H98 at the Q_H-site is weakly coupled to the semiquinone radical and there is no direct interaction between the SQ_H and the N_ε of Q101. The ¹⁵N selective labeling has also identified the N_ε of H75 as the nitrogen hydrogen bonded with the SQ_H of the D75H mutant enzyme.

Density functional calculations on models of the active site for both the WT and D75H systems show good agreement between for experimentally observed and calculated values of available ^14,15N, ¹H and ¹³C couplings. The model studies indicate that a very strong hydrogen bonding interaction occurs between the O1 atom of the SQ_H and the hydrogen bond donor groups of R71 and D75 with a relatively weaker interaction occurring for O4 and the imidazole NH group of H98. This is mainly responsible for the highly asymmetric spin density distribution observed experimentally. Replacement of D75 by H75 in the D75H mutant model leads to a lower spin density asymmetry for the SQ_H. The HYSCORE results also imply that in the D75H mutant, the N_εof R71 possesses significantly smaller hyperfine coupling than that in the WT protein despite practically the same length of H-O bond in optimized structure. This suggests that the orientation of the hydrogen bond donors is different from the optimized small model calculation for D75H. Based on previous studies of quinone one-electron transfer sites, we further postulate that the highly asymmetric spin density distribution of the WT SQ_H may be a significant factor in its role as a one-electron transfer agent between the substrate binding site and heme b.

Overall, the combined pulsed EPR experiments and electronic structure calculations carried out in this study constitute a major step towards complete characterization of the distribution of the unpaired spin density of the SQ in the Q_H site. The interactions between the SQ and nearby residues unraveled in this study are crucial to understand how the radical is stabilized inside the protein’s binding pocket and establish a foundation for future studies of quinone structure-function relationships in bioenergetics.

Abbreviations

2D

two-dimensional

CW EPR

continuous wave electron paramagnetic resonance

cyt bo₃

cytochrome bo₃ ubiquinol oxidase from E. coli

DDM

n-dodecyl-β-D-maltoside

DFT

Density Functional Theory

ESEEM

electron spin echo envelope modulation

HYSCORE

hyperfine sublevel correlation

nqi

nuclear quadrupole interactions

N_wc

weakly coupled nitrogens

qcc

quadrupole coupling constant

Q_H

the high affinity quinone-binding site

Q_L

the low affinity quinone-binding site

SQ

semiquinone

SQ_H

semiquinone at the Q_H site

UQ₈

ubiquinone-8

WT

wild-type