Crystal structure of the electron transfer complex rubredoxin–rubredoxin reductase of Pseudomonas aeruginosa

Abstract

Crude oil spills represent a major ecological threat because of the chemical inertness of the constituent n-alkanes. The Gram-negative bacterium Pseudomonas aeruginosa is one of the few bacterial species able to metabolize such compounds. Three chromosomal genes, rubB, rubA1, and rubA2 coding for an NAD(P)H:rubredoxin reductase (RdxR) and two rubredoxins (Rdxs) are indispensable for this ability. They constitute an electron transport (ET) pathway that shuttles reducing equivalents from carbon metabolism to the membrane-bound alkane hydroxylases AlkB1 and AlkB2. The RdxR–Rdx system also is crucial as part of the oxidative stress response in archaea or anaerobic bacteria. The redox couple has been analyzed in detail as a model system for ET processes. We have solved the structure of RdxR of P. aeruginosa both alone and in complex with Rdx, without the need for cross-linking, and both structures were refined at 2.40- and 2.45-Å resolution, respectively. RdxR consists of two cofactor-binding domains and a C-terminal domain essential for the specific recognition of Rdx. Only a small number of direct interactions govern mutual recognition of RdxR and Rdx, corroborating the transient nature of the complex. The shortest distance between the redox centers is observed to be 6.2 Å.

The ubiquitous, Gram-negative bacterium Pseudomonas aeruginosa is metabolically highly versatile, allowing it to survive numerous specialized ecological niches in addition to soil and aquatic environments. This versatility allows P. aeruginosa to be both an opportunistic pathogen, chronically colonizing the respiratory tract of cystic fibrosis patients or causing acute infections in open wounds (1, 2), and a valuable ally by degrading ecological pollutants such as detergents (3, 4) or n-alkanes (5, 6). Mineralization of n-alkanes inter alia from crude oil spills involves five genes (7). The membrane-bound alkane hydroxylases AlkB1 and AlkB2 oxidize terminal carbons of chemically inert n-alkanes, allowing further oxidation and degradation. Electrons for this initial reaction derive from carbon metabolism relayed through an electron transport (ET) chain involving FAD-dependent NAD(P)H:rubredoxin reductase (RdxR) RubB and two AlkG2-type (see below) rubredoxins (Rdxs) RubA2 and RubA1 (7–9), encoded by the gene cluster rubB (PA5349), rubA2 (PA5350), and rubA1 (PA5351). Whereas alkB1 and alkB2 expression is strictly n-alkane-dependent, RubB/RubA1/RubA2 are constitutively produced (8), indicating a more general but as yet unidentified role for this ET chain in P. aeruginosa (7).

Rdxs are small (≈6 kDa), redox-active iron–sulfur proteins found in anaerobic or microaerophilic archaea and bacteria (10). A central iron, coordinated by four cysteines, constitutes the redox-active site alternating between +2 and +3 oxidation states. Rdxs are crucial for oxidative stress responses in anaerobic organisms by rapidly transferring metabolic reducing equivalents to superoxide reductases or rubredoxin:oxygen oxidoreductases to reduce oxygen or reactive oxygen species (11, 12). The link to general metabolism is provided by NAD(P)H and NAD(P)H:RdxRs (11, 13, 14). RdxRs are either heterodimeric (class 1, RdxR from Desulfovibrio gigas) and bind the cofactors FAD and FMN (15), or single-chain proteins (class 2, RubB) exclusively binding FAD. Related enzymes occur in aerobic bacteria and eukaryotes, and include both electron shuttling enzymes and redox enzymes such as glutathione reductase (GR) (16). For simplicity, the gene products of PA5349, PA5350, and PA5351 are henceforth referred to as RdxR and Rdx, respectively.

Several RdxR-like enzymes and many Rdxs have been extensively studied both structurally and biophysically (17–30). Here, we present the crystal structure of a dedicated RdxR from P. aeruginosa strain PAO1 at 2.40-Å resolution. The interaction of RdxR and Rdx, which we describe at 2.45-Å resolution, has not been analyzed structurally so far, possibly because of the assumption of a highly transient and hence structurally inaccessible ET complex. These structures of RdxR and Rdx in a functional complex provide a structural basis to understand the ET processes of these much-studied proteins.

Results

RdxR Is Indispensable for n-Alkane Oxidation in P. aeruginosa.

A mutant of P. aeruginosa with an isogenic rdxR transposon insertion is unable to grow on hexadecane [supporting information (SI) Fig. 5], confirming earlier reports that RdxR is essential for the utilization of n-alkanes as sole carbon source by different Pseudomonas strains (7, 31). This effect is completely reverted by supplying the RdxR gene on plasmid pUCP20 (SI Fig. 5). Interestingly, both RdxR and the Rdxs rubA1 and rubA2 from P. aeruginosa can substitute for their counterparts in Pseudomonas putida GPo1 (7).

Structure Determination and Refinement.

Yellow, FAD-bound RdxR crystals (space group P6₁22) diffract x-rays to 2.40 Å. The structure was solved by molecular replacement using nitrite reductase from Pyrococcus furiosus [Protein Data Bank (PDB) ID code 1XHC; 26% sequence identity] as a search model. The final model of RdxR, comprising residues 4–384, refines to a final R factor of 16.6% (R_free, 20.4%) (Table 1).

Table 1.

Crystallographic data

Measurement	RdxR–Rdx	RdxR
Data statistics
Space group	P2221	P6122
Unit cell length, Å
a	61.1	119.6
b	97.1	119.6
c	81.3	158.1
Wavelength, Å	1.741	0.981
Resolution range, Å	50.0–2.3	50.0–2.4
Mosaicity, °	1.4	0.5
Completeness, %	90.6 (86.2)	98.1 (82.9)
Redundancy	3.8 (3.4)	3.6 (2.8)
Unique reflections	19,054	25,303
Wilson B factor, Å2	41.0	38.7
I/σ(I)	12.1 (3.0)	11.8 (2.7)
Rmerge, %	10.1 (40.2)	7.3 (34.0)
Refinement statistics
Resolution range, Å	50.0–2.45	50.0–2.40
R/Rfree, %	18.1/24.6	16.6/20.4
Average B factor of protein atoms, Å2	40.5	30.2
rmsd bonds, Å	0.01	0.02
rmsd angles, °	1.5	1.6
Ramachandran plot, %
Favored	97.4	97.6
Allowed	100	100

Values in parentheses refer to the shell of highest resolution.

Orange-red crystals of the RdxR–Rdx complex, produced by mixing oxidized RdxR and oxidized Rdx in a molar ratio of 1:1.2, belong to the orthorhombic space group P222₁ and diffract x-rays to 2.30-Å resolution. The structure was solved by molecular replacement using the refined structure of RdxR and the Rdx from P. furiosus (32) (PDB ID code 1BQ8; sequence identity, 54%). The complex includes residues 4–384 of RdxR and residues 1–53 (of 55) of Rdx, and refines to an R factor of 18.1% (R_free, 24.6%) (Table 1).

Structure of RdxR.

Structurally, RdxR comprises an NAD(P)H-binding domain, an FAD-binding domain, and a C-terminal Rdx-binding domain (Fig. 1 A and B). The NAD(P)H-binding domain, encompassing residues 111–240, is inserted into the FAD-binding domain through two extended loops, βG/α5 and βN/α8, dividing the latter into an N-terminal (residues 1–110) and a C-terminal (residues 241–313) subdomain. The Rdx-binding domain consists of a five-stranded mixed β-sheet (βR-V), topped by the C-terminal α-helix α10.

Fig. 1.

Structure of RdxR. (A) Cartoon representation of uncomplexed RdxR using a green (N terminus) to orange (C terminus) color gradient. FAD and NADH (modeled) are shown as ball-and-stick models in orange and translucent gray. (B) Schematic representation of the RdxR secondary structure: circles, triangles, and lines indicate α-helices, β-strands, and loops, respectively. (C) Schematic view of the FAD-binding site. Hydrogen bonds and salt bridges are indicated by dotted lines, and hydrophobic interactions are indicated by gray arcs. 2F_o − F_c electron density for FAD (blue mesh) is contoured at 1.0σ. Color coding: C, gray; N, blue; O, red; S, yellow; P, orange; Fe, green. Covalent bonds are yellow for RdxR and red for Rdx.

The fold of RdxR is related to those of GR (33), murine apoptosis inducing factor (mAIF) (20), the ferredoxin reductase component of biphenyl dioxygenase [bacterial ferredoxin reductase (BphA4)] (19, 34), and putidaredoxin reductase (PdxR) (35) (see below). Although a sequence alignment of these proteins reveals several insertions and deletions in peripheral regions (SI Fig. 6), the cofactor-binding domains are structurally conserved resulting in rmsd values of RdxR relative to mAIF, BphA4, and PdxR of 2.2, 2.9, and 3.7 Å, respectively. Compared with the 384 residues of RdxR, BphA4 (408), PdxR (422), and mature mAIF (489) are considerably longer. In BphA4 and PdxR, these additional residues mostly constitute two α-helices forming a C-terminal three-helix bundle. The C terminus of mAIF is even longer because of the insertion of a putative protein–protein interaction loop (20, 36).

FAD- and NAD(P)H-Binding Sites.

FAD is recognized by RdxR through numerous interactions including water-mediated and direct hydrogen bonds, a salt bridge from the AMP-phosphate to Lys-45, and hydrophobic interactions mainly to the aromatic adenosine and isoalloxazine moieties (Fig. 1C). The extended conformation of FAD is similar to that observed in GR, PdxR, BphA4, and mAIF. Whereas the xylene ring of the isoalloxazine moiety is buried in a hydrophobic pocket and the si-side is shielded from solvent, the re-side faces the NAD(P)H-binding cavity (Fig. 1 A and C; SI Fig. 7). The atoms N5 and O4 of FAD are part of an extensive hydrogen bonding network to a salt-bridged glutamate–lysine pair (Glu-159/Lys-320) (Fig. 1C; SI Fig. 7 and below). Strikingly, Lys-320 functionally replaces a lysine located near the N terminus (approximately at position 50) in all other GR-fold enzymes (SI Fig. 6). In addition, RdxR lacks a tryptophan-lid shielding the pteridine-ring of FAD from the solvent in PdxR, BphA4, and mAIF (SI Fig. 6).

Neither of our crystal structures includes NAD(P)H bound to RdxR. However, FAD-based superpositioning of NAD(P)H-bound GR (1GRB; ref. 16) or BphA4 (1F3P; ref. 19) on RdxR, clearly indicates the NAD(P)H-binding pocket to be conserved in RdxR. Minor torsion angle adjustments of residues along the length of the modeled cofactor suffice to allow NADH being placed into the binding pocket. Similar conformational adjustments have been observed to occur on NAD(P)⁺ binding in both GR and BphA4 (19, 37). In the modeled complex, the nicotinamide ring of NAD(P)H stacks on the isoalloxazine ring of FAD at a distance of ≈3 Å, suitable for hydride transfer. The positions of atoms NN7 and NO7 of the modeled NAD(P)H are occupied by two water molecules in the NAD(P)H-free structure of RdxR, further corroborating the inferred model (SI Fig. 7A). Distinct conformations of Tyr-288 observed in RdxR and RdxR–Rdx presumably represent conformations adopted in the absence or presence of NAD(P)H, respectively. The latter conformation places the side chain of Tyr-288 opposite the NO2NC2 bond of the nicotinamide-ribose indicating that bound NAD(P)H would be further stabilized by an O–H–π interaction (SI Fig. 7A).

Dimerization of RdxR.

Despite distinct crystal packing arrangements for RdxR and RdxR–Rdx, RdxR forms identical dimers in both structures. The observed dimer, burying a surface area of 3,260 Ų (38), involves the convex surface of each monomer. Dimers have also been reported for GR, PdxR, BphA4, and mAIF (19, 20, 23, 33). In GR and similar S-S oxidoreductases, dimerization is essential as residues from both monomers create the active-site pocket (16, 37). The functional relevance of dimerization observed in electron transferases (PdxR, BphA4, and RdxR), however, currently remains unclear (23). In RdxR, dimerization restricts access to the NAD(P)H binding pocket and results in a steric clash between the modeled adenine moiety of NAD(P)H and α-helix α8′ of the neighboring molecule (SI Fig. 8). Moreover, numerous water molecules at the dimer interface and an unfavorable Arg-242–Arg-242′ contact all support a weak RdxR–RdxR′ interaction. We thus assume that RdxR dimers form at high protein concentrations used during crystallization, rather than being functionally relevant.

Structure of the RdxR–Rdx Complex.

We have cocrystallized RdxR with RubA2 (PA5350), an AlkG2-type (see below) Rdx from P. aeruginosa. As in other Rdxs, the redox-active Fe³⁺ (confirmed by x-ray anomalous scattering) (Fig. 2A) of RubA2, is tetrahedrally coordinated by four cysteines (Figs. 1C and 2A). Crystal structures of several Rdx have been discussed in detail (17, 27–30, 32, 39). RubA2 most closely resembles the Rdx of P. furiosus (1BQ8; rmsd, 0.88 Å) (32) (SI Fig. 9).

Fig. 2.

Structure of the Rdx–RdxR complex. (A) Cartoon representation. RdxR colors are as in Fig. 1A, and Rdx is in dark red. FAD, modeled NADH (translucent), and cysteine residues of the iron-binding site are shown in ball-and-stick mode, and Fe³⁺ is shown as a green sphere. Anomalous-difference electron density (red) contoured at 6.0σ documents the presence of the Fe³⁺. Amino acid exchanges between RubA1 and RubA2 are indicated as spheres in Rdx (conservative, orange; nonconservative, blue). (B) Interactions surrounding the redox-active site of the complex. Interacting residues are shown as ball-and-stick models (C of RdxR, yellow, and of Rdx, red-brown). FAD is shown in ball-and-stick mode. (C) Binding curves for the interaction of RdxR to Fe³⁺ (red)- and Ni²⁺ (green)-substituted Rdx.

As expected, the electron transfer complex between RdxR and Rdx is a 1:1 complex of the two proteins (Fig. 2A). Rdx binds to the concave side of RdxR, interacting with each of its three domains. The accessible surface area buried on formation of the complex is a mere ≈1,120 Å (38). This comparatively small interface (40) covers 17% and 3% of the total surface areas of Rdx and RdxR, respectively.

The shortest direct distance between redox centers of both molecules (FAD-N3 to Cys-9_Rdx-S⁻) is 6.2 Å, significantly below the 15-Å limit proposed for physiologically relevant ET reactions (41). A well resolved, bridging water molecule subdivides this distance into two shorter segments of 2.8 and 5.7 Å, potentially affecting both likelihood and speed of electron transfer between the redox partners (Fig. 1C) (42–44).

Electrostatic surface potentials for both Rdx and RdxR, calculated by using Adaptive Poisson–Boltzmann Solver (45), reveal significant charge complementarity (see below). The interface is dominated by H-bonded interactions centered about the axis running from FAD-N3 to Rdx-Fe³⁺. These interactions include direct hydrogen bonds between residues Met-290_RdxR–Cys-42_Rdx, Arg-297_RdxR–Gly-43_Rdx, Ser-44_RdxR–Val-8_Rdx, and Ser-44_RdxR–Val-7_Rdx, as well as water-mediated hydrogen bonds between residues Thr-318_RdxR–Asp-41_Rdx and Pro-316_RdxR–Asp-41_Rdx (Fig. 2B). Three salt bridges within the interface connecting residues Glu-21_Rdx–Lys-377_RdxR, Arg-21_RdxR–Asp-48_Rdx, and Asp-41_Rdx–Lys-372_RdxR are solvent exposed and partially disordered, indicating weak contributions to recognition. This matches the observation that the affinity of P. putida GPO1 (or Pseudomonas oleovorans) RdxR–Rdx is independent of ionic strength (27). Finally, the methylene groups of Glu-21_Rdx contribute a van der Waals interaction to Leu-373_RdxR. Overall, the high charge complementarity of the two molecules, coupled to a small interaction surface and relatively few directed interactions, ensure a specific yet transient interaction between RdxR and Rdx.

Binding of Native and Ni²⁺-Substituted Rdx to RdxR.

Purification of Rdx via Ni-nitrilotriacetic acid affinity chromatography results in a ≈1:1 mixture of Ni²⁺- and Fe³⁺-substituted Rdx separable by anion exchange chromatography. To test whether RdxR–Rdx affinity is influenced by the metal content and oxidation state of Rdx, we determined the K_D of Rdx^Fe3+ and Rdx^Ni2+ for Rdx (27). Nonlinear regression of the Rdx^Fe binding data (Fig. 2C) indicates a K_D of 3 ± 1 μM, comparable with the K_D of 1 ± 0.1 μM for P. putida GPO1 RdxR–Rdx (27). Despite a diminished effect on FAD fluorescence by Rdx^Ni2+ (lower equilibrium ΔF₅₁₅) (Fig. 2C), the K_D of 5 ± 2 μM of RdxR–Rdx^Ni is comparable with that of RdxR–Rdx^Fe.

Discussion

Relationship of FAD-Dependent Reductases.

Phylogenetic analysis (46) of several structurally characterized, FAD-dependent NAD(P)H oxidoreductases, indicates GR-fold enzymes to be evolutionarily quite distant both from AdxR- and FNR-type enzymes (Fig. 3A). The insertion of the NAD(P)H-binding domain into the FAD-binding domain is distinct for the three branches, indicating independent gene insertion events (47, 48). The GR-branch itself comprises two phylogenetically and functionally distinct subbranches, namely electron transferases and disulfide oxidoreductases (Fig. 3A). Thioredoxin reductase (TrxR) (22, 49–51) occupies an intermediate position by mechanistically resembling disulfide oxidoreductases but functionally belonging to the electron transferases.

Fig. 3.

Phylogenetic topology of structurally characterized enzymes functionally and/or structurally related to RdxR. (A) FNR, plant-type ferredoxin reductase; AdxR, adrenodoxin reductase; AIF, apoptosis inducing factor; LpdR, lipoamide reductase; TptR, trypanothione reductase. Sequence identity to RdxR is indicated by percentage. (B) Superposition of uncomplexed (gray) and complexed RdxR (colored as in Fig. 1A). Cofactors are shown as ball-and-stick models. For clarity, the position of Rdx is indicated by a brown sphere. (C) Comparison of the Rdx (red) and Trx (black) binding sites of RdxR and TrxR. For clarity, only the molecular surface of RdxR is shown. ΔG⁰ optimized ET rates (42) between FAD and each point of the RdxR surface are indicated by a color gradient (red, high ET rate; green, medium; blue, low). (Inset) Close-up view of the cofactors involved.

Several distinguishing features of RdxR underline its evolutionary distance from other GR-branch electron transferases such as the mitochondrial apoptosis inducing factor, prokaryotic PdxR, and BphA4. Most importantly, C-terminal Lys-320 of RdxR functionally replaces a conserved N-terminal lysine, implicated in FAD binding and hydride transfer (19, 20, 23, 37). The pteridine moiety of FAD, shielded from the solvent by a Trp-lid in other FdxRs, is exposed at the FAD-N3 atom in RdxR. Finally, the C-terminal domains of RdxRs and FdxRs share only marginal sequence and structural similarity (SI Fig. 6), reflecting their divergent substrate specificities.

RdxR Provides a Preformed Redox Scaffold for Rdx.

Interestingly, although TrxR and RdxR share a similar fold, the enzymes employ divergent catalytic strategies. In TrxR, domains rotate relative to each other so that both NADPH and the catalytic disulfide interact with the same side of the flavin cofactor (49, 50). In RdxR, by contrast, substrate- and cofactor-binding sites lie on opposite sides of FAD (as in GR), eliminating the need for major conformational changes and allowing for a simpler electron transport pathway. Correspondingly, the structure of RdxR and, in particular, the FAD-binding domain remains largely unperturbed by Rdx complexation (Fig. 3B). rmsd values for C_α atoms are 0.42 Å for RdxR and 0.22 Å for the FAD-binding domain. Small yet significant conformational changes (52) do, however, occur both in the NAD(P)H-binding domain (presumably because NAD(P)H is lacking) and in the loops connecting both cofactor binding domains. Similarly, complex formation induces the RdxR C-terminal domain to rotate away slightly from the Rdx-binding site resolving an unfavorably close contact between Leu-373_RdxR and Glu-22_Rdx. The uncomplexed conformation of RdxR thus represents a preformed Rdx binding site, avoiding time-consuming conformational changes that gate ET reactions in other complexes (53).

RdxR Discriminates Between Two Types of Rdx.

Sequence alignments indicate that Rdxs involved in alkane oxidation fall into two classes, denoted AlkG1- and AlkG2-type Rdxs (9). RubA2 and RubA1 of P. aeruginosa PAO1 are both AlkG2-type Rdxs, are encoded by neighboring genes, and are 80% identical by amino acid sequence. Several exchanged residues cluster around the Fe^3+/2+-binding site (Fig. 2A) potentially affecting the Fe-center redox potential (26). Some substitutions affect the molecular surface involved in target recognition, implying that RubA2 and RubA1 may interact with distinct electron acceptors.

Other n-alkane using bacteria encode two distinct Rdxs, of AlkG1- and AlkG2-type. Only the latter transfers electrons from RdxR to alkane hydroxylases (9). In AlkG1-type Rdxs, an additional arginine is inserted immediately downstream of the second metal-binding CXXCG motif. Our complex structure indicates that this results in an unfavorably close contact to the positively charged molecular surface of RdxR, explaining why AlkG1-type Rdxs are unable to productively interact with RdxR (9, 27).

Electron Transfer from RdxR to Rdx.

Electron transfer from NAD(P)H to Rdx involves a reductive and an oxidative step with respect to RdxR. NAD(P)H binding initiates the reductive half reaction, involving hydride transfer from NAD(P)H-C4 to FAD-N5. The resulting blue charge transfer complex between FADH⁻ and NAD⁺ is detectable as a broad absorption peak between 500 and 800 nm after bleaching FAD (data not shown). During the oxidative half reaction, two electrons are sequentially transferred from FADH to two Rdx molecules (25), resulting in a flavin semiquinone reaction intermediate that has, however, not been observed spectroscopically for RdxR (25).

The electron transfer reactions alter the protonation state of FAD-N5. First, a hydride is transferred to N5 from NAD(P)H, whereas the loss of the second electron to Rdx necessitates the removal of a proton. A Glu/Lys pair (Glu-159/Lys-320 in RdxR), involved in an intricate interaction network with FAD-N5 (Fig. 1C; SI Fig. 7A) and functionally conserved in ETases or replaced by a histidine-bound “central water” (SI Fig. 7B) in AdxR (48), was thought to be crucial for the hydrogen transfer reactions (19, 20, 23, 37, 48). Replacing these residues in mAIF, however, results in loss of FAD (20), whereas adding FAD increases activity beyond wild-type levels (20). This residue pair thus appears primarily to be required for FAD binding in RdxR-type enzymes rather than for hydride transfer.

Implications for Interprotein ET.

According to the Marcus theory (54–57), ET kinetics depend on the edge-to-edge distance of the redox cofactors involved, the driving force of the reaction (ΔG⁰), and the reorganization energy (λ) (55). At a given distance, ET rates will be highest for activationless reactions where ΔG⁰ = −λ. For this optimal case, the distance of 6.2 Å between the two redox centers in the RdxR–Rdx complex, would allow transfer rates in the picosecond range (10¹¹ s⁻¹) (41). Assuming the redox potentials for RdxR and Rdx of P. aeruginosa to be similar to those of P. putida (25), indicates a ΔG⁰ of around −0.25 eV. Combined with λ = 1 eV, typical for intermolecular ET (41), and a packing density of 0.64 [calculated by using the ET rates package (41)], this reduces the ET rate to 10⁹ s⁻¹ (or 10⁸ s⁻¹ for λ = 1.5 eV) (41). However, note that RdxR successively transfers single electrons to two distinct Rdx molecules. ΔG⁰ is therefore presumably different for both reactions (25), leading to differing ET rates for each of the reactions. Thus, although the crystal structure of RdxR–Rdx allows the maximal theoretical rate of electron transfer to be estimated, parameters such as ΔG⁰ and λ would need to be determined experimentally to reveal the true ET rate between RdxR and Rdx. Nevertheless, the ET rate for RdxR–Rdx is probably faster than typical turnover rates of ET proteins (41), implying that diffusional steps such as Rdx- and NADH-binding may be rate limiting. The binding site is thus under evolutionary pressure to optimize the interaction. ΔG⁰-optimized ET rates, calculated for all points on the surface of RdxR (ref. 41; Fig. 3C), demonstrate that the binding site is optimized to allow docking of Rdx at a pronounced tunneling hot spot of RdxR. Interestingly, structurally unrelated Trx binds to a distinct, second hot spot located on the opposite side of the FAD cofactor in the related structure of thioredoxin reductase (Fig. 3C), although the process of electron transfer in that system is more complex than in RdxR–Rdx (22).

Electrostatics are important for the formation and kinetics of redox complexes (40). Comparing the charge distribution at the molecular surface of RdxR–Rdx and related complexes indicates charges to be complementary for each redox pair (Fig. 4). The respective dipole moments (bioportal.weizmann.ac.il/dipol/) intersect at acute angles (37° for FNR/Fdx, 47° for TrxR/Trx) allowing for favorable steering of the incoming reaction partner (Fig. 4) (58). Thus, despite evolutionary pressure to conserve substrate and cofactor interactions (59), ET couples have diverged with respect to electrostatic forces and guidance of their particular substrate.

Fig. 4.

Electrostatics of interprotein ET reactions. RdxR–Rdx (A) is compared with other ET complexes: TrxR–Trx (B) (50) and FNR–Fdx (C) (34). Complexes are shown in cartoon representation with the reductase in yellow and the substrate in red. FADs are shown in ball-and-stick mode, and dipole moments are shown as dumbbells. Electrostatic surface potentials are mapped onto van der Waals surfaces (red, negative; blue, positive; white, neutral) in open-book mode.

The pronounced electrostatic homing system, the docking of Rdx precisely at the tunneling hot spot of RdxR, and the absence of any appreciable conformational changes during Rdx binding would appear to suggest complex formation to proceed by a “simple-” rather than a “gated-” or “dynamic-docking” mechanism (60).

Concluding Remarks.

Biological one-electron redox reactions generally depend on an electron distribution system consisting of an FAD-dependent NAD(P)H-oxidoreductase combined with a small iron-binding protein such as ferredoxin, cytochrome, or rubredoxin. Small size and differing redox potentials of the iron-binding proteins ensure that a multitude of electron acceptors can be supplied by a comparatively small number of electron carriers. Our analyses indicate the RdxR–Rdx system to be a structurally and functionally distinct electron shuttling system. Features of the complex such as the preformed interaction surface, the pronounced electrostatic homing system, a short tunneling distance, as well as the weak dissociation constant imply the complex to be optimized for rapid transport of reducing equivalents to the actual place of reduction. Clearly, the RdxR–Rdx system is crucial to P. aeruginosa to grow on n-alkanes. However, employing such a sophisticated redox chain for this single purpose appears disproportionate, implying additional but as-yet-unidentified roles of the RdxR–Rdx couple in P. aeruginosa.

Methods

Cloning, Protein Expression, and Purification.

The genes PA5349 (RdxR) and PA5350 (Rdx) of P. aeruginosa PAO1 were amplified from genomic DNA using following PCR primers: RdxR (PA5349), 5′-GCGCTCTAGATAACGAGGGCAAAAAATGAG C G A GCGTGCGCCCCTGGTA-3′, 5′-GCGCA G A T C TAGCCATGAGGCCGGGTAACTCTTTG-3′; and Rdx (PA5350), 5′-GACGGCCATATGCGCAAGTGGCAATGCGTGGTC-3′, 5′-GCGCAGATCTGGCGATCTCGATCATCTCGAA-3′.

The NdeI/BglII digestion products were cloned into pBBR22bII, a derivative of pBBR22b (61), resulting in His₆ fusions of the target proteins. Proteins were produced in Escherichia coli Tuner cells (Novagen, Madison, WI) in LB medium and 37 μg/ml chloramphenicol. At an OD₆₀₀ of 1.0 (37°C), protein expression was induced by 0.5 mM isopropyl-β-d-thiogalactoside and continued overnight at 20°C. Cells were centrifuged, resuspended, and lysed by French press, and cell debris was removed by centrifugation. Purification involved Ni-nitrilotriacetic acid affinity chromatography (Qiagen, Hilden, Germany), anion exchange chromatography (MonoQ; GE Healthcare, Chalfont St. Giles, U.K.), and gel filtration (Superdex 75; GE Healthcare). Ni²⁺- (because of Ni-nitrilotriacetic acid) and Fe³⁺-binding Rdx were separated during ion-exchange chromatography. Proteins were dialyzed against 100 mM NaCl, 50 mM Tris·HCl (pH 8). A 5 mM concentration of β-mercaptoethanol was added to RdxR, to prevent cysteine oxidation. Proteins were stored at 4°C at concentrations of 8.5 mg/ml (RdxR) and 30 mg/ml (Rdx).

Alkane Oxidation.

P. aeruginosa strain TBCF10839 wild type, the isogenic RdxR-transposon mutant (62), and the RdxR-complemented strain were grown on E2 minimal agar. The plates were placed in a sealed container, and hexadecane was supplied as the sole carbon source. Growth was monitored for 3–5 days at 30°C.

Equilibrium Binding Studies.

A Nanodrop ND-3300 fluorospectrometer was used to monitor Rdx-induced changes in FAD fluorescence. The intensity of blue light-emitting diode (λ_max = 470 nm)-induced FAD fluorescence was monitored at 515 nm. A concentration of 12 μM for RdxR was used throughout, whereas that of Rdx varied between 0 and 140 μM. A short optical pathway (<1 mm) largely eliminates inner filter effects. Measurements in triplicate were performed in 50 mM Tris·HCl (pH 8) and 50 mM NaCl. Dissociation constants (K_D) were determined as described (27).

Crystallization.

Hexagonal RdxR crystals were grown at 20°C by sitting-drop vapor diffusion with equal volumes (0.1 μl) of protein (8.5 mg/ml) and reservoir solution [5% PEG 1000, 40% PEG 300, 0.1 M Tris·HCl (pH 7)]. Crystals were flash frozen in liquid nitrogen without further cryoprotection.

RdxR–Rdx complex crystals were also grown at 20°C by sitting-drop vapor diffusion. RdxR at 10 mg/ml was mixed with a 1.2 molar excess of Rdx and diluted to 8.5 mg/ml. Equal volumes (0.1 μl) of protein mixture and reservoir solution (0.2 M KF, 20% PEG 3350) were used. Microseeding with severely intergrown initial crystals yielded plate-shaped orthorhombic crystals of the complex. Mother liquor supplemented with 25% PEG 400 was used for cryoprotection.

Structure Determination.

X-ray diffraction data for RdxR were collected at λ = 0.92 Å at beamline BL1 (Berliner Elektronenspeicherring-Gesellschaft für Synchrotronstrahlung, Berlin, Germany), for RdxR–Rdx at λ = 1.7 Å at beamline BW7A (European Molecular Biology Laboratory, Hamburg, Germany) using MarCCD detectors (MarResearch, Norderstedt, Germany). HKL2000 (63) was used for data processing, PHASER (64) for molecular replacement, CNS (65) and CCP4 suites (66) for structure refinement, and COOT (67) for model building, structural analysis, and validation. Figures were prepared by using PYMOL. Crystallographic statistics are listed in Table 1.

Abbreviations

ET

electron transport

Rdx

rubredoxin

RdxR

rubredoxin reductase

GR

glutathione reductase

mAIF

murine apoptosis inducing factor

BphA4

bacterial ferredoxin reductase

PdxR

putidaredoxin reductase

TrxR

thioredoxin reductase.