ATOMIC RESOLUTION STRUCTURES OF RIESKE IRON-SULFUR PROTEIN: EXPLORING THE ROLE OF HYDROGEN BONDS IN TUNING REDOX POTENTIAL OF IRON-SULFUR CLUSTERS

SUMMARY

The Rieske [2Fe-2S] iron-sulfur protein of cytochrome bc₁ functions as the initial electron acceptor in the rate-limiting step of the catalytic reaction. Prior studies have established roles for a number of conserved residues that hydrogen bond to ligands of the [2Fe-2S] cluster. We have constructed site-specific variants at two of these residues, measured their thermodynamic and functional properties, and determined atomic resolution X-ray crystal structures, for the native protein at 1.2 Å resolution, and for five variants (Ser-154→Ala, Ser-154→Thr, Ser-154→Cys, Tyr-156→Phe and Tyr-156→Trp) to resolutions between 1.5 Å and 1.1 Å. These structures and complementary biophysical data provide a molecular framework for understanding the role of hydrogen bonds to the cluster in tuning thermodynamic properties, and hence the rate of this bioenergetic reaction. These studies provide the first detailed structure-function dissection of the role of hydrogen bonds in tuning the redox potentials of [2Fe-2S] clusters.

Introduction

The ubihydroquinone:cytochrome c oxidoreductase (E.C. 1.10.2.2) (bc₁ complex) plays a fundamental role in the major biological energy conservation pathways, including those of the mitochondrial and bacterial respiratory and photosynthetic chains (Crofts, 2004; Crofts and Berry, 1998). The enzyme catalyzes the transfer of electrons from quinol to a small soluble protein, such as cytochrome c, and couples the redox work to the generation of a proton gradient across the membrane (Crofts and Berry, 1998). The core components of this multi-subunit integral membrane enzyme consist of three catalytic subunits (cytochrome b, cytochrome c₁, and the Rieske iron-sulfur protein, ISP) bearing four redox centers (two cytochrome b hemes, one cytochrome c₁ heme and a high-potential Rieske-type [2Fe-2S] cluster) (Berry et al., 2000). The proton-motive Q-cycle model (Crofts et al., 1983; Mitchell, 1976), accounts for the overall function, and more recent work has shown that the rate-limiting step in the reaction is the oxidation of quinol at the Q_o site (Crofts and Wang, 1989). The rate limitation is in the first electron to the high potential ISP (Hong et al., 1999), from where it is transferred through to cytochrome c₁. The semiquinone species remaining at the Q_o-site is oxidized on transfer of the second electron to the lower-potential b type heme (Brandt and Trumpower, 1994), (Meinhardt and Crofts 1983).

Although initial crystallographic studies of mitochondrial cytochrome bc₁ complex did not resolve the ISP, they showed that the Rieske cluster could occur in different configurations (Iwata et al., 1998; Xia et al., 1997). The first structures in which the ISP was resolved showed that the Rieske ISP extrinsic domain was in several different positions, leading to the suggestion that this domain undergoes movement during catalysis (Zhang et al., 1998). This movement is reflected in at least eight different positions now observed in native structures in different crystal forms of the bc₁ complex (Zhang et al., 1998). Consistent with this hypothesis, mutants of the bc₁ complex in which movement of the Rieske ISP domain is constrained by inter-subunit disulfide bridges showed a decrease in the rate of intra-molecular electron transfer (Xiao et al., 2000). Although the movement of the Rieske ISP has been proposed to be a rate limiting step in electron transfer, kinetic measurements following flash activation of the Rhodobacter sphaeroides bc₁ complex in situ or in the isolated complex demonstrate that rate associated with the movement of the Rieske ISP are much more rapid than the rate of ubiquinol oxidation (Crofts et al., 2003) (Engstrom et al., 2002).

Despite the low overall sequence homology between the mitochondrial and chloroplast bc₁ complexes, the crystal structure of their respective isolated Rieske ISP domains are quite similar and residues that surround the [2Fe-2S] cluster are highly conserved (Carrell et al., 1997; Iwata et al., 1996). Most importantly, the protein ligands that engage the cluster are surrounded by a network of hydrogen bonds mediated by backbone amides, and residue side-chains that are identical between the mitochondrial, bacterial, and chloroplast enzymes of the bc₁ complex family, but vary in lower potential Rieske-like proteins (Bonisch et al., 2002; Carrell et al., 1997; Hunsicker-Wang et al., 2003; Iwata et al., 1996). In particular, the phenolic side chain of Tyr-156 (Rhodobacter numbering) is located within hydrogen bond distance to the sulfhydryl of a cluster ligand residue Cys-129, and the hydroxyl of Ser-154 is proximal to one of the sulfur atoms in the cluster. Mutations of the residues equivalent to Tyr-156 in Rhodobacter sphaeroides (Guergova-Kuras et al., 2000), Saccharomyces cerevisiae (Denke et al., 1998), Bos taurus (Leggate and Hirst, 2005), and Paracoccus denitrificans (Schroter et al., 1998) are shown to affect the redox potential of the Rieske ISP. This hydrogen bond network is thought to increase the potential of the [2Fe-2S] cluster by charge delocalization to preferential stabilization of the reduced state.

Recently, we constructed mutant strains of R. sphaeroides expressing variant bc₁ complexes in which these residues have been replaced by site-directed mutagenesis (Guergova-Kuras et al., 2000). Each of these mutant strains was able to assemble the bc₁ complex and support photosynthetic growth. Redox titration of purified mutant bc₁ complexes demonstrate that the Tyr-156→ Leu, Phe, and Trp substitutions result in an alteration of the midpoint potential of the Rieske ISP and a decrease in the rate of quinol oxidation. Flash induced kinetic measurements of bc₁ complexes in situ demonstrate that changes in the pK value of the oxidized Rieske ISP results in a change in the pH dependence of the rate of quinol oxidation (Guergova-Kuras et al., 2000).

In order to provide a context for further exploring the contribution of hydrogen bonds in tuning the redox potential of iron-sulfur clusters, we have characterized the kinetic properties and undertaken structural studies of the wild-type and site-specific variants of the isolated, water-soluble Rieske ISP domain from R. sphaeroides. The redox potential and pK values of these site specific variants were measured using protein film voltammetry (Zu et al., 2003). In order to provide a molecular basis for understanding these biophysical measurements, we have determined high-resolution X-ray crystal structures of wild-type Rieske ISP and five mutants at cluster proximal residues Ser-154→Ala, Ser-154→Thr, Ser-154→Cys, Tyr-156→Phe, and Tyr-156→Trp to resolutions of better than 1.5 Å. The high resolution of the structural data allow for unambiguous assignments of all atoms and solvent molecules, as well as alternate side chain conformers. These structures allow for atomic level dissection of biophysical results to demonstrate the specific role of protein ligands in tuning the electrochemical properties of the [2Fe-2S] clusters. Our studies demonstrate that observed decreases in midpoint potential that result from the mutation of residues involved in hydrogen bonding with the cluster are a direct consequence of the removal of individual hydrogen bonds.

Results and Discussion

Structure Determination

A highly redundant data set, collected at the Cu Kα wavelength, was used to determine the crystal structure of the water-soluble Tyr156→Phe Rieske ISP by single wavelength anomalous diffraction methods. Despite the modest anomalous signal from Fe at this wavelength (f’”=3.02 at 1.54 Å), the quality of the resultant electron density map was excellent and allowed for the entire trace of the protein main chain using automated methods. Rapid manual building was facilitated by the excellent starting phase set, allowing for a straightforward trace of the entire polypeptide and associated solvent molecules. The structures of wild-type, Ser156→Ala, Ser156→Thr, Ser154→Cys, and Tyr156→Trp Rieske ISP were solved by molecular replacement using the final refined coordinates of Tyr156→Phe Rieske and refined to resolutions between 1.1 Å and 1.5 Å resolution. The crystal structures of fully oxidized and fully reduced wild-type Rieske ISP were each determined to a resolution of 1.05 Å, utilizing crystals soaked in a five-fold molar excess of either sodium ascorbate (reduced state) or potassium hexacyanoferrate (oxidized state). All data collection and refinement statistics may be found in Table 4.

Table 4.

Data Collection and Refinement Statistics

Data collection Statistics
	Wild-type	Tyr-156→Phe	Tyr-156→Trp	Ser-154→Ala	Ser-154→Thr	Ser-154→Cys
Space group	I4	I4	I4	I4	I4	I4
Beam Line	32ID	Home	32ID	22ID	17ID	22ID
Cell lengths (Å)	a=70.57 c=54.77	a=70.63 c=54.78	a=70.64 c=54.60	a=69.94 c=54.58	a=70.65 c=54.72	a=70.88 c=54.93
Resolution (Å)	1.2	1.5	1.1	1.35	1.5	1.5
Total observations	171,725	304,203	289,260	170,255	78,492	84,468
Unique reflections	41,254	21,562	54,430	29,270	20,919	21,968
Completeness (%)	98.0 (99.7)	99.7 (97.0)	99.8 (99.7)	100 (100)	96.7 (87.7)	99.9 (100)
Redundancy	4.2	14.1	5.3	5.8	3.8	3.9
Rmerge (%)	3.0 (6.5)*	5.1 (18.3)*	5.0 (13.0)*	5.8 (47.8)*	5.6 (19.9)*	8.7 (45.3)*
I/σ (I)	24.7 (19.6)*	67 (15)*	16.7 (12.0)*	26 (3.8) *	21 (8)*	14.9 (3.5)
SAD analysis		Tyr-156→Phe
Wavelength (Å)		1.54
Mean FOM		0.58
Density modified FOM		0.75
Refinement Statistics
	Wild-type	Tyr-156→Phe	Tyr-156→Trp	Ser-154→Ala	Ser-154→Thr	Ser-154→Cys
Resolution bins (Å)	12–1.2	18–1.5	13–1.2	18–1.35	18–1.5	18–1.5
Reflections used	38,022	19,979	51,606	27,466	19,380	20,692
Rcrystal	11.66	15.20	12.56	14.20	13.35	13.16
Rfree	13.37	16.72	13.74	15.92	17.37	17.85
Number of Atoms
Protein	1088	1065	1123	1065	1079	1062
Solvent	198	113	232	149	181	179
Glycerol	0	0	12	12	12	12
Average B (Å2)	8.66	19.33	9.25	21	13.14	16.45
RMSD bond length	0.007	0.006	0.009	0.007	0.008	0.009
RMSD bond angle	1.508	1.334	1.476	1.388	1.406	1.462
PDB access code	2NUK	2NUM	2NWF	2NVG	2NVE	2NVF

^*Numbers in parentheses correspond to values for the outermost shell.

R_merge = ∑|I – <I>|/∑I, where I is the integrated intensity of a given reflection.

R_crystal = ∑||F_obs| - |F_calc||/∑| F_obs|. R_free was calculated using 5–7% of data excluded from refinement.

Overall Structure

The overall polypeptide fold is similar to that of isolated Rieske ISP from mitochondria and chloroplast, as well as to those of Rieske-type proteins of bacterial and archaeal origins (Bonisch et al., 2002; Carrell et al., 1997; Hunsicker-Wang et al., 2003; Iwata et al., 1996). As detailed comparison of the high-resolution structures of oxidized and reduced wild-type Rieske ISP did not show any significant differences between the structures (SKN, DK, JB, and ARC, unpublished results), the structural detailed discussed below are applicable to both redox states. It should be noted that although crystals of the oxidized form of the Rieske ISP were maintained in the presence of excess oxidant, exposure to high flux X radiation may have resulted in conversion into the reduced form.

The polypeptide consists of ten anti-parallel β-sheets layered into three stacks (Figure 1). These secondary structure elements are arranged to form two distinct entities, the cluster binding domain and the basal domain. The first layer of β-sheet is formed by strands β1, β11, and β12; the second layer is formed by strands β3 through β5 and β10; and the third layer is formed by strands β6 through β9. A single α-helix adjoins β3 and β4 and connects the basal and cluster-binding domains. A disulfide bridge between two cysteine residues, located in the loop regions between β5-β6 (Cys-134) and β7-β8 (Cys-151) tethers β5-β8 strand together and provides further stability for loops that contain the cluster binding residues. The basal domain of the R. sphaeroides Rieske ISP is structurally similar to equivalent domains of Rieske proteins from bovine, plant, and archaeal sources (Bonisch et al., 2002; Carrell et al., 1997; Colbert et al., 2000; Hunsicker-Wang et al., 2003; Iwata et al., 1996). The cluster-binding domain spans roughly 50 residues within the center of the polypeptide sequence, and bears the cluster ligands and cluster interacting residues.

Figure 1.

Orthogonal views of the overall structure of the Rieske iron-sulfur protein from R. sphaeroides. (A) A ribbon representation of the polypeptide is shown labeled with the corresponding secondary structural elements as defined in the text. Atoms within the [2Fe-2S] cluster are shown as white and yellow spheres, respectively. (B) The structure is shown, rotated 90° about the vertical axis, viewed into the location of the [2Fe-2S] cluster.

Pairwise comparison of the structure of R. sphaeroides ISP with the isolated Rieske domains from other organisms reveals closest structural homology to the mitochondrial Rieske ISP (root mean square deviation of 1.1 Å over 126 aligned residues; Z-score = 22.7) (Iwata et al., 1996). As studies of other Rieske ISPs have demonstrated relative reorientations of the basal and cluster-binding domains (Hunsicker-Wang et al., 2003), a comparison of these individual domains is warranted. The cluster-binding domain of the R. sphaeroides and mitochondrial ISPs show a root mean square deviation of 0.4 Å over 48 aligned residues (75% identity), while the basal domains show a RMSD of 1.2 Å over 78 aligned residues (29% identity). The greatest deviation between the two basal domains occurs as a result of 13 amino acid loop insertion spanning Thr-96 through Ala-109 in the structure of the R. sphaeroides ISP. Structure based comparisons with the isolated Rieske ISP from chloroplast demonstrates weaker homology (root mean square deviation of 2.7 Å over 94 aligned residues; Z-score = 8.7; the cluster-binding domains show an RMSD of 1.0 Å over 45 aligned residues with 49% sequence identity; for the basal domains show an RMSD of 2.6 Å over 47 aligned residues with 13% sequence identity) (Carrell et al., 1997). Hence, despite the overall weak sequence similarity, structural determinants between the R. sphaeroides and chloroplast ISPs are quite similar.

The [2Fe-2S] cluster is engaged by protein ligands Cys-129 (Sγ-Fe distance of 2.33 Å), His-131 (Nδ1-Fe distance of 2.11 Å), Cys-149 (Sγ-Fe distance of 2.29 Å), and His-152 (Nδ1-Fe distance of 2.10 Å). The exceptionally high resolution of 1.2 Å for the wild-type ISP structure provides an accurate limit to the metal-ligand distances. Full-matrix, least-squares refinement yields estimated standard deviations in the range of 0.01 Å for the Fe-S bond lengths and 0.02 Å for the Fe-N bond lengths. A detailed list of the cluster-ligand distances for wild-type and site-specific mutant R. sphaeroides Rieske ISP can be found in Table 1.

Table 1.

Atomic distances (in Å) between [2Fe-2S] cluster residues and protein ligands of interest.

Variant	C129-Fe1	H131-Fe2	C149-Fe1	H152-Fe2	Y156-C129	S154-S1
Wild-type	2.33	2.11	2.29	2.10	3.14	3.17
S154A	2.33	2.06	2.31	2.12	3.11	N/A
S154C	2.35	2.18	2.31	2.16	3.13	3.07
S154T	2.33	2.12	2.30	2.12	3.12	3.16
Y156F	2.31	2.14	2.32	2.14	N/A	3.18
Y156W	2.32	2.12	2.29	2.10	N/A	3.18

Role of cluster proximal residues Tyr-156

In addition to the [2Fe-2S] cluster and its associated ligands, the local environment around the clusters of bc₁ Rieske domains are characterized by a disulfide bridge formed between Cys-134 and Cys-151, and by two polar residues Ser-154 and Tyr-156. Tyr-156 makes a hydrogen bond with the cluster ligand Cys-129 (Oη-Sγ distance of 3.1 Å) (Figure 2A) and Ser-154 is within hydrogen bond distance to atom S₁ of the cluster (S₁-Oγ distance of 3.2 Å) (Figure 3A). Biochemical and biophysical studies establish that these hydrogen bonds can affect the redox potential and pK_ox1 of the Rieske ISP (see Table 2). For example, the Tyr-156→Phe mutation in the R. sphaeroides Rieske ISP results in a 45 mV decrease in the midpoint potential relative to the wild-type (Table 2). Similar results were obtained in S. cerevisiae and P. denitrificans where the equivalent Tyr→Phe substitution results in a decrease in the midpoint potential of the ISP by 65 and 44 mV, respectively. In the absence of structural data, it was unclear if these changes in the midpoint potential were solely the consequence of a loss of a hydrogen bond between Tyr-156 and Cys-129.

Figure 2.

(A–C) Difference Fourier maps shown around residue 156 of wild-type and variant Rieske ISP, calculated with coefficients F_o–F_c using experimental amplitudes and phases calculated from the final refined structures, minus the coordinates of residue 156. The contour levels of the maps are 2.5 σ (blue). Coordinates from the final refined structures are superimposed upon the respective maps, and atoms of the [2Fe-2S] cluster is shown as stick figures in magenta and orange. The structures shown are of (A) wild-type, (B) Tyr-156→Phe and (C) Tyr-156→Trp Rieske ISP. (D) Superposition of the structures of wild-type (cyan), and Tyr-156→Trp (yellow) ISP showing the gross structural movements that accompany the introduction of the Trp-156 side chain.

Figure 3.

(A–D) Difference Fourier maps shown around residue 154 of wild-type and variant Rieske ISP, calculated with coefficients F_o–F_c using experimental amplitudes and phases calculated from the final refined structures, minus the coordinates of residue 154. The contour levels of the maps are 2.5 σ (blue), and 8 σ (red). Coordinates from the final refined structures are superimposed upon the respective maps and atoms of the [2Fe-2S] cluster is shown as stick figures in magenta and orange. The structures shown are of (A) wild-type, (B) Ser-154→Ala and (C) Ser-154→Cys, and (D) Ser-154→Thr Rieske ISP.

Table 2.

Biochemical and Biophysical Parameters for ISP Variants

Variant	ΔpH optimumA	Method	Em, 7.4 noteB	Eacid (mV)	Ealk (mV)	pKox1	pKox2	pKre d1	pKr ed2	Ref.
R. sphaeroides
wild typeF wild typeC	-	PFV CD	297 283	308	−134	7.60	9.60	12.4		(Zu et al., 2003) (Ugulava and Crofts, 1998)
S154TF S154TC	0.3	PFV CD	274 255	283	−168	7.70	9.28	12.37		# #
S154CF S154CC	0.5	PFV CD	295 300	319	−107	7.14	9.73	12.10		# #
S154AF S154AC	−0.3	PFV CD	170 186	173	−307	8.02	9.92	13.10		# #
Y156F	−0.2	CD	238	256	<56	7.5	9.2	>10		(Guergova-Kuras et al., 2000)
Y156W	1.2	CD	198	198	<114	8.5	>10		>10	(Guergova-Kuras et al., 2000)
Bovine
wild type R wild type F		PFV CD		311 315	−152	7.55 7.70	9.10 9.10	11.8 0	12.8 1	(Leggate and Hirst, 2005) (Link, 1999)
S163AR		PFV		164	−297	8.15	9.31	11.8 8	13.5 4	(Leggate and Hirst, 2005)
Y165FR		PFV		252	−225	7.74	9.43	12.0 8	13.3 1	(Leggate and Hirst, 2005)
Y165WR		PFV		214	<−216	8.00	9.40	>12.40		(Leggate and Hirst, 2005)

Notes:

^ApH optimum for turnover of the Q_o-site was at pH 7.6, measured at E_h ~100 mV in ref. (Ugulava and Crofts, 1998), and at pH 7.2, measured at E_h ~70 mV in current work.

^BIn the intact bc₁ complex, formation of the g_x=1.800 complex with ubiquinone shifts the apparent E_m

^FRieske iron-sulfur protein (ISP)

^CMeasured in situ in the bc₁ complex

^RRecombinant ISP protein expressed in E. coli.

^#Present work.

Our 1.5 Å resolution crystal structure of the Tyr-156→Phe R. sphaeroides Rieske ISP reveals that the structure of the mutant is nearly identical to that of the wild-type ISP. Overall, the crystal structures of the two ISPs can be superimposed with a room mean square deviation of 0.105 Å, and there are no local alterations in the polypeptide scaffold that result from the Tyr-156→Phe mutation. Analysis of electron density maps calculated with either experimental phases or phases from the coordinates of the final, refined structure do not show the presence of any additional features near the vicinity of Phe-156 side chain (Figure 2B). Most importantly, there is no evidence of any ordered solvent molecules that may compensate for the loss of the hydroxyl group in the Phe-156 mutant. Hence, the measured 45 mV decrease in the midpoint potential in Tyr-156→Phe R. sphaeroides ISP, relative to the wild-type, is solely the consequence of the loss of the hydrogen bond between Phe-156 and the cluster ligand Cys-129.

In contrast to the modest changes in both structure and biophysical characteristics that accompany the Tyr-156→Phe mutation, significant alterations are observed in our 1.1 Å structure of the Tyr-156→Trp mutant R. sphaeroides ISP (Figure 2C). Although the overall structure of this variant is similar to that of the wild-type ISP (RMSD of main chain atoms of 0.124 Å), accommodation of the indole ring of Trp-156 requires movement of the polypeptide backbone at regions adjacent to this residues. Specifically, the main chain along two loop regions that flank Trp156, spanning residues Gly-127 through Ile-137, and Gly-146 through Gly-153, are pushed apart roughly 0.3–0.4 Å (Figure 2D). As all of the cluster ligands are harbored within this region, this movement results in a slight shift of the entire cluster by roughly 0.4 Å relative to its position in the wild-type ISP. However, despite the structural perturbation, the integrity of the hydrogen bond between Ser-154 and atom S₁ of the cluster is not compromised (S₁-Oγ distance of 3.1 Å).

Prior studies from our laboratory have shown that the Tyr-156→Trp mutation in the R. sphaeroides ISP results in a decrease in the midpoint potential by 100 mV, a change in pK_1,ox by ~0.9 units, and substantial changes in the EPR spectra relative to the wild-type ISP (Guergova-Kuras et al., 2000). However, despite the modification in the EPR spectra, this mutant retained the characteristic Rieske-type signal, and displayed sensitivity to the inhibitor stigmatellin, similar to that of wild-type ISP. Our 1.1 Å resolution crystal structure provides a structural rationale for understanding these biophysical characteristics of the Tyr-156→Trp mutant R. sphaeroides ISP. The structural rearrangements noted above, that are required to accommodate the Tyr-156→Trp mutation, result in perturbations of the EPR spectra of this variant relative to that of the wild-type ISP, and a greater perturbation of the thermodynamic properties than in the Tyr-156→Phe mutation.

Role of cluster proximal residues Ser-154

In the wild-type structure of R. sphaeroides ISP, Ser-154 is within hydrogen bond distance of the bridging sulfur atom S₁ of the cluster (S₁-Oγ distance of 3.2 Å) (Figure 3A). In order to determine the contribution of this residue in modulating redox potential, we have characterized kinetic parameters of site-specific variants at this residue, and determined high-resolution crystal structures of the variant ISPs. Removal of the hydroxyl group in the Ser-154→Ala mutation results in a decrease in the midpoint potential by 97 mV, an increase in pK_ox1 to 8.02, a 10-fold decrease in the association constant for formation of ES-complex with quinol, and a 5-fold decrease in sensitivity to inhibition by stigmatellin (Table 3). The structure of Ser-154→Ala R. sphaeroides ISP has been determined to a resolution of 1.35 Å. A comparison of a least squares superposition of the wild-type and Ser-154→Ala ISP structures (RMS deviation of main chain atoms of 0.134) shows that minimal structural perturbations result as a consequence of the mutation. No additional features, corresponding to ordered solvents, can be observed near the vicinity of the cavity created by the Ser-154→Ala mutation (Figure 3B). Consequently, the measured changes in kinetic parameters in the Ser-154→Ala mutant ISP can be attributed to be a direct consequence of the loss of the hydrogen bond between Ser-154 and a bridging sulfur atom in the [2Fe-2S] cluster.

Table 3.

Kinetic parameters for FS154 mutants

Strain	Em of the ISP (mV)1)	cyt bH reduction rates2)	% of reduction rates	Apparent Em of the cyt bH reduction	Association constant for QH2 (KQH2) 3)	Stigmatellin Inhibition 4)
WT 5)	283 ± 5	1370	100	129.4 ± 2.4	23.5	0.585
FS154T	255 ± 5	669	48.7	134.5 ± 1.9	35.3	2.39
FS154C	300 ± 4	871	63.4	119.2 ± 2.9	10.4	2.30
FS154A	186 ± 5	46.7	3.40	101.5 ± 2.1	2.52	2.57

¹The E_m values are measure at pH 7.4 by CD spectroscopy.

²⁾When the quinone pool is ~30% reduced (mol. cyt b_H^reduced _ (mol. total bc₁)⁻¹ _ sec⁻¹)

³The values are derived as suggested in (Zhang et al., 1998).

⁴The amount of inhibitor that blocked half of the functional bc₁ complex (mol. stigmatellin/mol. bc₁ complex)

⁵⁾Rb. sphaeroides Ga strain

The 1.5 Å crystal structure of the Ser-154→Thr R. sphaeroides ISP is nearly identical to that of the wild-type ISP (RMSD of main chain atoms of 0.141 Å), and reveals no local or global changes in the polypeptide that result from the mutation. The hydroxyl of Thr-154 occupies the same location as that of Ser-154 in the wild-type ISP structure, and donates a hydrogen bond to atom S₁ of the cluster (S₁-Oγ distance of 3.2 Å) (Figure 3C). The Cγ of Thr-154 protrudes intro a hydrophobic cleft defined by Ile-162 and Tyr-156, and is accommodated into this cleft by movement of Cδ of Ile-162, with minimal changes in the position of neighboring atoms. The Ser-154→Thr mutation results in only moderate alteration in kinetic parameters, with a 28 mV decrease in midpoint potential, and a modest increase in association constant for interaction with quinone. As Thr-154 donates a hydrogen bond to atom S₁ of the cluster (S₁-Oγ distance of 3.2 Å), the changes in the kinetic parameters observed in the Ser-154→Thr, relative to wild-type ISP, are relatively modest.

The Ser-154→Cys mutation results in a 17 mV increase in the midpoint potential relative to the wild-type ISP. The 1.5 Å crystal structure of Ser-154→Cys R. sphaeroides ISP shows no significant changes in the polypeptide structure as a consequence of the mutation relative to the wild-type ISP (RMSD of main chain atoms of 0.128 Å). Comparison of the wild-type and Ser-154→Cys mutant ISP reveals that the side chain of Cys-154 is rotated about torsion angle χ₁. As a result, the Sγ of Cys-154 occupies a position closer to atom S₁ of the [2Fe-2S] cluster (S₁-Oγ distance of 2.9 Å). The increase in midpoint potential in this mutant relative to the wild-type enzyme may be a result of the decrease in the inter-atomic distance to the [2Fe-2S] cluster, which would result in a stronger hydrogen bond between Cys-154 and the bridging sulfur. Presumably, formation of a disulfide between Cys-154 and the bridging sulfur is precluded by the cluster microenvironment.

Conclusions

The importance of hydrogen bonds in determining the redox potential of [2Fe-2S] clusters have been long been recognized (Adman et al., 1975; Carter, 1977). Earlier attempts to dissect the contributions of hydrogen bonding to redox potentials utilized spectroscopic and structural studies of [4Fe-4S] clusters from different proteins (Backes et al., 1991). However, electrostatic calculations demonstrate that the observed variations in redox potentials amongst these different proteins are not due solely to differences in hydrogen-bonding patterns as cluster microenvironment contributes significantly to the redox potential (Langen et al., 1992). More recent efforts to analyze the contribution of protein ligands to the tuning of iron-sulfur cluster redox potentials have focused on mutational analyses of Rieske [2Fe-2S] proteins (Denke et al., 1998; Leggate and Hirst, 2005; Schroter et al., 1998). Rieske ISP are ideal candidates for such analysis due to (a) their higher redox potential as compared to other [2Fe-2S] proteins, and (b) the complex hydrogen bonding network surrounding the cluster, which facilitates mutational analyses without severely compromising structural integrity (Carrell et al., 1997; Iwata et al., 1996). However, mutational studies are incomplete in the absence of high-resolution structural information to provide a framework for understanding the structural consequences. Our spectroscopic, kinetic, and crystallographic analyses of site-specific mutants of the R. sphaeroides Rieske ISP represent the first detailed structure-function dissection of the role of hydrogen bonds in tuning the redox potentials of [2Fe-2S] clusters. These studies demonstrate that the observed decreases in midpoint potential that result from the mutation of residues involved in hydrogen bonding with the cluster are a direct consequence of the removal of individual hydrogen bonds. The Rieske ISP is able to tolerate modest changes in the nature of the hydrogen bonding ligand with minimal effects on either the structure or the redox potential of the [2Fe-2S] cluster. Finally, our studies demonstrate that the ISP is sufficiently plastic to accommodate introduction of large changes into the proximity of the cluster with only slight alterations in the protein scaffold.

Experimental Procedures

Rieske Protein Production

Details on the construction of Rhodobacter sphaeroides strains expressing mutant Rieske ISPs have already been described (Guergova-Kuras et al., 2000). Briefly, mutations in the ISP were made using the QuikChange kit (Stratagene) for site-directed mutagenesis, with minor modifications. In order to express the desired mutant proteins, the DNA fragment containing a mutation site was cloned in a broad host plasmid, pRK415, and transferred into host strain, Rb. sphaeroides BC17 from which the chromosomal fbc operon encoding the bc₁ complex had been deleted (Yun et al., 1990). The integrity of each mutation was confirmed at the DNA level by sequencing of the plasmid purified from the cells from which the protein was purified. Wild type and mutant cytochrome bc₁ complexes were purified using nickel affinity resin by virtue of an engineered carboxy-terminal polyhistidine tag (Guergova-Kuras et al., 1999).

Preparation of Rieske soluble fragment

After purification of the multi-subunit complex, the soluble Rieske subunit extrinsic domain was liberated using thermolysin digestion and further purified by hydrophobic interaction chromatography. Mass spectrometric analysis documents that the soluble domain spans residues 47 to 196 of the full-length subunit. Purified samples were supplemented with glycerol and stored at 193K.

Preparation of intact Rieske subunit

For protein film voltammetry, the intact subunit was isolated from the cytochrome bc₁ complex with retention of the hydrophobic N-terminal domain. The cytochrome bc₁ complex was prepared as described previously (Guergova-Kuras et al., 1999). For fractionation, it was exchanged into a solution containing 10 mM CAPS and 0.01% n-decyl-B-D-maltoside at pH 12, incubated on ice for 30 min, and filtered using a Centriplus centrifugal concentrator with a cutoff of 50 kDa (Amicon, Beverly, MA). The Rieske protein was contained in the filtrate. It was exchanged into 20 mM histidine, pH 7.5, 0.5 mM EDTA, 1% glycerol (vol/vol), and 500 mM KCl and concentrated with a 7-kDa cutoff concentrator.

Protein Film Voltammetry

Reduction potentials were measured using protein film voltammetry essentially as described previously (Zu et al., 2002a; Zu et al., 2002b). It was found that the intact Rieske subunit provided better adhesion to the electrode surface, and this preparation was used for most experiments. Briefly, the protein was applied directly to a freshly polished pyrolytic graphite-edge (PGE) electrode surface and then placed into solution in an all-glass cell. The cell was thermostated and encased in a Faraday cage. Measurements were performed aerobically, but this was not a problem as oxygen did not interfere with measurements in the range of 100–500 mV (vs. SHE). Analogue-scan cyclic voltammetry was performed using a Bioanalytical Systems (West Lafayette, IN) CV-27 Voltammograph with an in-house amplifier, and results were recorded via an in-house program. Data were analyzed using Fourier transformation and cubic-spline subtraction in Origin 6.1 (OriginLab Corp., Northampton, MA). Solution pH values were controlled using mixtures of four of the following buffers (total concentration, 40 mM), 10 mM sodium acetate, HEPES, MES, TAPS, CAPS, and sodium phosphate, depending on the pH. Volumetric solutions of NaOH were used above pH 13. All standard reagents were supplied by Fluka (Buchs, Switzerland) or Merck-BDH (Poole, U.K.). The pH of each solution was checked immediately following measurement; 1 M NaOH was used as a pH 14 standard. The effects of high Na⁺ concentration were corrected using standard formulae (Bard and Faulkner, 2001).

Crystallization

Samples of Rieske ISP soluble extrinsic domain were thawed and exchanged into buffer consisting of 100 mM KCl, 20 mM HEPES, pH=7.5 through ultrafiltration. Initial crystallization conditions were established by the sparse-matrix sampling methods using commercial and home-made screens. Refinement of promising conditions yielded large, red-colored crystals suitable for diffraction analysis. Crystals of the Rieske ISP were grown using the hanging drop vapor diffusion method. Briefly, 2 μl of protein sample (8 mg/ml) was added to 2 μl of precipitant (1.5 M (NH₄) ₂SO₄, 100 mM sodium acetate, pH=4.6, and 20% (v/v) anhydrous glycerol) and equilibrated over a well containing the precipitant solution at 20° C. Crystals grew in 24 hours and reached a maximum size of 0.5x0.5x0.8 mm in 3 days. Crystals of mutant Rieske ISP were grown under similar conditions and grew to the same size. Crystals of fully oxidized and fully reduced wild-type Rieske ISP were were produced by soaking preformed crystals in a five-fold molar excess of either sodium ascorbate (reduced state) or potassium hexacyanoferrate (oxidized state). Despite the presence of suitable concentrations of a cryo-preservant in the crystallization buffer, diffraction data from crystals harvested directly from the crystallization drop were hampered by the presence of ice rings. To circumvent this problem, crystals of Rieske ISP were transiently streaked through a solution of Paratone-N immediately prior to flash cooling by direct submersion into liquid nitrogen.

Data collection and structure determination

Crystals of Rieske ISP diffracted to better than 1.5 Å using a Raxis IV++ image plate detector mounted on a CuKα Rigaku R3H home source rotating anode generator. These crystals occupied space group I4 with unit cell parameters a = b = 70.63 Å, c=54.78 Å. A data set encompassing a complete 360-degree sweep (1° oscillations) was collected from a single crystal of the Tyr156→Phe mutant Rieske ISP. This data set had an overall redundancy of 15-fold including all data and 7-fold with anomalous pairs kept separate. Data were integrated and scaled using the HKL2000 package (Otwinowski and Minor, 1997).

The high symmetry space group occupied by these crystals allowed for the simultaneous measurement of a significant number of Bijvoet pairs, and coupled with the high quality of the data (see Table 1), produced a strong anomalous signal from the [2Fe-2S] cluster at the wavelength of Cu Kα. The position of the two Fe peaks were readily identified using SHELXC, and SHELXD/SHELXE were used to extract phases from these heavy atom sites (Schneider and Sheldrick, 2002). The resultant phases were density modified using RESOLVE (Terwilliger, 2003) and the modified phases were imported into ARP/wARP version 5.1 (Morris et al., 2003) for automated chain tracing. Autotracing resulted in a complete trace of the protein main chain and assignment of more than half of the side chains of the protein sequence. The remainder of the model was built using XtalView (McRee, 1999) and further improved through rounds of refinement using REFMAC5 (Murshudov et al., 1997; Vagin et al., 2004) and manual rebuilding. Cross-validation, using 7% of the data for the calculation of the free R factor, was utilized to monitor building bias (Brunger, 1993).

Data from crystals of wild-type, Ser156→Ala, Ser156→Thr, Ser156→Cys and Tyr156→Trp Rieske ISP were collected to limiting resolutions between 1.5 and 1.1 Å at an insertion device synchrotron source (Sector 32ID, Sector 17ID, and Sector 22ID, Advanced Photon Source, Argonne, IL) using charge coupled device detectors. Phases for each structure were determined by molecular replacement with the program MOLREP (Vagin and Teplyakov, 2000) using the Tyr156→Phe structure as a probe. Multiple rounds of manual model building were interspersed with refinement using REFMAC5 (Murshudov et al., 1997; Vagin et al., 2004) to complete structure refinement. Alternate conformers were modeled where appropriate. Individual hydrogen atoms were included as riding atoms in the highest resolution structures (1.2 Å or better) based on features in difference Fourier maps. Cross-validation used 5–7% of the data in the calculation of the free R factor. Graphics were created using Ribbons (Carson, 1996), and Pymol (pymol.sourceforge.net).