|Home | About | Journals | Submit | Contact Us | Français|
Subunit b, the peripheral stalk of bacterial F1Fo ATP synthases, is composed of a membrane-spanning and a soluble part. The soluble part is divided into tether, dimerization, and δ-binding domains. The first solution structure of b30-82, including the tether region and part of the dimerization domain, has been solved by nuclear magnetic resonance, revealing an α-helix between residues 39 and 72. In the solution structure, b30-82 has a length of 48.07 Å. The surface charge distribution of b30-82 shows one side with a hydrophobic surface pattern, formed by alanine residues. Alanine residues 61, 68, 70, and 72 were replaced by single cysteines in the soluble part of subunit b, b22-156. The cysteines at positions 61, 68, and 72 showed disulfide formation. In contrast, no cross-link could be formed for the A70C mutant. The patterns of disulfide bonding, together with the circular dichroism spectroscopy data, are indicative of an adjacent arrangement of residues 61, 68, and 72 in both α-helices in b22-156.
ATP synthesis by oxidative phosphorylation or photophosphorylation is a multistep membrane-located process that provides the bulk of cellular energy in eukaryotes and many prokaryotes. The majority of ATP synthesis is accomplished by the enzyme ATP synthase (EC 188.8.131.52), also called F1Fo ATP synthase, which, in its simplest form, as in bacteria, is composed of eight different subunits (α3, β3, γ, δ, , a, b2, and c9-12). This multisubunit complex is divided into the F1 headpiece, α3:β3, and a membrane-embedded ion-translocating part known as Fo, to which F1 is attached by a central and a peripheral stalk (1, 5, 25). ATP is synthesized or hydrolyzed on the α3:β3 hexamer, and the energy provided for or released during that process is transmitted to the membrane-bound Fo sector, consisting of subunits a and c and part of subunit b (30, 31). The energy coupling between the two active domains occurs via the stalk part(s) (6). The central stalk is made of subunits γ and , and the peripheral stalk is formed by subunits δ and b. The peripheral stalk, which lies at the edge of the multisubunit assembly of the F1Fo ATP synthase, acts as a stator to counter the tendency of the α3:β3 hexamer to follow the rotation of the central stalk and the attached c-ring, and to anchor the membrane-embedded a subunit (17, 36).
In Escherichia coli, subunit b with its 156 residues extends with its soluble part (bsol; b21-156) from the top of the F1 sector down, into, and across the membrane, where it is associated with subunit a (2, 15, 32, 34). The 156-residue b subunit has been divided into four functional domains (28). They are, in order from the N to the C terminus; the membrane domain, the tether region, the dimerization domain, and the δ-binding domain. The structure of the synthesized 33-residue peptide comprising the N-terminal membrane-spanning region has been solved by 1H NMR, showing an α-helical feature (14). The crystallographic structure of the major part of the dimerization domain, b62-122, revealed an α-helix with a length of 9.0 nm (12). Most recently, the NMR solution structure of the very C-terminal segment, b140-156, which interacts with the C terminus of subunit δ (δ91-177), has been determined by NMR spectroscopy (26). This molecule adopts a stable helix formation in solution with a flexible tail between amino acid residues 140 and 145. SAXS (26) and analytical ultracentrifuge studies have indicated that the soluble domain of subunit b (b21-156, b22-156) is dimeric in solution (12). So far, no high-resolution structure of the tether domain, including residues 25 to 52, or the N-terminal segment of the dimerization domain, which is formed by residues 53 to 122, is available (14).
Here, we have turned our attention to the production and purification of residues 30 to 82 of subunit b (b30-82) from E. coli F1Fo ATP synthase, which forms the remaining unsolved structural segment of subunit b. The structural features of this segment have been determined in solution using NMR spectroscopy. The introduction of a cysteine residue into b22-156 at four positions resulted in different intersubunit disulfide patterns, giving insight into the proximity of the residues.
The abbreviations used in this article are as follows: 2D, 2-dimensional; 3D, 3-dimensional; CD, circular dichroism; DSS, 2,2-dimethyl-2-silapentane-5-sulfonate; DTT, dithiothreitol; HSQC, heteronuclear single quantum coherence; IPTG, isopropyl-β-d-thiogalactoside; NMR, nuclear magnetic resonance; NOE, nuclear Overhauser effect; NOESY, NOE spectroscopy; NTA, nitrilotriacetic acid; PAGE, polyacrylamide gel electrophoresis; PFG, pulsed-field gradient; SAXS, small-angle X-ray scattering; R1, longitudinal relaxation time; R2, transverse relaxation time; RMSD, root mean square deviation; SDS, sodium dodecyl sulfate, TOCSY, total correlation spectroscopy; TPPI, time-proportional phase incrementation.
Pfu DNA polymerase and Ni2+-NTA chromatography resins were obtained from Qiagen (Hilden, Germany); restriction enzymes were purchased from MBI Fermentas (St. Leon-Rot, Germany). Chemicals for gel electrophoresis and trypsin used for in-gel digestion were purchased from Serva (Heidelberg, Germany) and Promega (Madison, WI), respectively. All other chemicals were of analytical grade and were obtained from Biomol (Hamburg, Germany), Merck (Darmstadt, Germany), Sigma (Deisenhofen, Germany), or Serva (Heidelberg, Germany). (15NH4)2SO4 and [13C]glucose were purchased from Cambridge Isotope Laboratories (Andover, MA).
In order to produce the truncated b subunit, b30-82, forward primer 5′-GCC ATC CAT GGC AGC AAT C-3′ and reverse primer 5′-CTG GAG CTC TCA CTT GTT CGC CTG-3′ were designed. The atpF-atpH-containing plasmid pPR1, which codes for the soluble domain of subunit b (b22-156) and subunit δ (26), was used as the template for PCR. PCR products incorporating NcoI and SacI restriction sites were digested and ligated to the pET9d1 vector with or without His3 (19). The pET9d-His3 vector, containing the respective gene, was then transformed into E. coli cells [strain BL21(DE3)] and grown on Luria-Bertani agar plates containing 30 μg/ml kanamycin. b30-82 uniformly labeled with 15N alone or with both 15N and 13C was produced in E. coli BL21(DE3) cells using M9 minimal medium containing either 15NH4Cl alone or 15NH4Cl plus [U-13C]glucose, with kanamycin (30 μg ml−1), for about 6 h at 37°C, until an optical density at 600 nm of 0.6 to 0.7 was reached. To induce production of proteins, cultures were supplemented with IPTG to a final concentration of 1 mM. Following incubation for another 4 h at 30°C, the cells were harvested at 10,000 × g for 15 min at 4°C. Subsequently, they were lysed on ice by sonication three times, for 1 min each time, in buffer A (50 mM Tris-HCl [pH 7.5], 250 mM NaCl, and 1 mM phenylmethylsulfonyl fluoride). The lysate was cleared by centrifugation at 10,000 × g for 35 min. The supernatant was filtered (pore size, 0.45 μm; Millipore) and passed over a 2-ml Ni2+-NTA resin column to isolate the proteins. The His-tagged protein was allowed to bind to the matrix for 2 h at 4°C and was eluted with an imidazole gradient (15 to 300 mM) in buffer A. Fractions containing His3-tagged b30-82 were identified by SDS-PAGE (24), pooled, and subsequently applied to an ion-exchange column (Resource Q; 6 ml; GE Healthcare). The proteins were concentrated using Centricon YM-3 (molecular mass cutoff, 3 kDa) spin concentrators (Millipore). The purity of all protein samples was analyzed by SDS-PAGE (24). SDS gels were stained with Coomassie brilliant blue G250. Protein concentrations were determined by the bicinchoninic acid assay (Pierce, Rockford, IL).
The A61C, A68C, A70C, and A72C single mutants of subunit b22-156 were generated by an overlap extension PCR method (21) using the atpF-atpH-containing plasmid pPR1 as a template (26). In two PCRs, mutations were introduced by internal primers: A61C internal forward, 5′-CTG AGC AGA TAA AGT GCG AGG CAA T-3′; A61C internal reverse, 5′-ATT GCC TCG CAC TTT ATC TGC TCA G-3′; A68C internal forward, 5′-CAG ATA AAG GCA GAG TGT ATT GAA GAA GCA-3′; A68C internal reverse, 5′-TGC TTC TTC AAT ACA CTC TGC CTT TAT CTG-3′; A70C internal forward, 5′-AAG AAA GCG AAA TGC GAA GCC CA-3′; A70C internal reverse, 5′-TG GGC TTC GCA TTT CGC TTT CTT-3′; A72C internal forward, 5′-G AAA GCG GAA TGC CAG GTA ATC AT-3′; A72C internal reverse, 5′-AT GAT TAC CTG GCA TTC CGC TTT C-3′. Mutated gene constructs of b22-156 were finally amplified using flanking primers 5′-CCT GTT CAC CAT GGC TTG CA-3′ (forward primer a) and 5′-TTC GAG CTC TTA AGA CTG CAA GAC GTC-3′ (reverse primer d). Following digestion with NcoI and SacI, the PCR product was ligated into the pET9d-His3 vector. The mutations were verified by DNA sequencing. Proteins were produced from these mutants as recently described for b22-156 (26).
Steady-state CD spectra were measured in the far-UV-light range (180 to 260 nm) using a Chirascan spectrometer (Applied Photophysics). Spectra were collected in a 60-μl quartz cell (Hellma) with a path length of 0.1 mm at 20°C and a step resolution of 1 nm. The readings were averages of 2 s at each wavelength, and the recorded millidegree values were averages of three determinations for each sample. CD spectroscopy of b30-82 and the b22-156 mutant proteins (1.0 mg/ml) was performed in a buffer consisting of 50 mM Tris-HCl (pH 7.5) and 250 mM NaCl. The spectrum for the buffer was subtracted from the spectrum for the protein. CD values were converted to mean residue molar ellipticity (θ) in units of degrees times square centimeters per decimole per amino acid using Chirascan software (version 1.2; Applied Photophysics). This baseline corrected spectrum was used as the input for computer methods to obtain predictions of secondary structure. The CD spectra were analyzed as described recently (26).
The b22-156 A61C, A68C, A70C, and A72C mutants were supplemented with 10 μM CuCl2 as a zero-length cross-linker for 20 min on a sample rotator at 4°C. The reaction was stopped by addition of 1 mM EDTA. Samples were dissolved in DTT-free dissociation buffer and were applied to an SDS-polyacrylamide gel as described above.
For the production of uniformly labeled (with 15N alone or with 15N and 13C) b30-82, the expressing bacteria were grown in M9 minimal medium containing 15NH4Cl alone or 15NH4Cl and [13C]glucose. The NMR sample (0.3 mM) was prepared in 90% H2O-10% D2O containing 25 mM NaH2PO4-Na2HPO4 (pH 6.8) and 0.1% NaN3. NMR experiments were performed at 15°C on a Bruker Avance 600-MHz spectrometer using a triple-channel cryoprobe equipped with gradient accessories. The experiments recorded on the 15N-labeled sample were 2D 15N-HSQC and 3D 15N-NOESY-HSQC. The experiments collected on the 15N- and 13C-labeled sample were HNCO, HNCACB, CBCA(CO)NH, HCCCONH, and 3D 15N-NOESY-HSQC. The 3D 15N NOESY-HSQC experiment used a mixing time of 300 ms. All the 3D experiments made use of PFGs for coherence selection and artifact suppression and utilized gradient sensitivity enhancement schemes. Quadrature detection in the indirectly detected dimensions was achieved using either the States-TPPI or the echo-antiecho method. Baseline corrections were applied wherever necessary. The proton chemical shift was referenced to the methyl signal of DSS (Cambridge Isotope Laboratories) as an external reference to 0 ppm. The 13C and 15N chemical shifts were referenced indirectly to DSS. All the NMR spectra were processed either using NMRPipe/NMRDraw (10) or TopSpin, version 2.1, the built-in software of the Bruker Avance spectrometer. Peak picking and data analysis of the Fourier-transformed spectra were performed with the SPARKY program (22).
The structure calculations were performed starting from Met30 at the N terminus, located after Pro29. The His residues of the His tag at the N terminus were ignored. Distance constraints for the structure calculations were collected from 3D 15N-edited NOESY-HSQC by manually and automatically assigned NOEs using CYANA, version 3.0 (20). Dihedral angle restraints were calculated from Cα and Cβ chemical shifts by using TALOS (9). Secondary structure was predicted from the chemical shift index (35) and NOE patterns. Seven cycles of automated NOESY assignment using the CYANA package, version 3.0, were performed. In the final CYANA cycle, NOESY cross peaks were assigned unambiguously, leading to 468 meaningful NOE distance restraints. Ten conformers of the monomeric b30-82 were calculated based on the dihedral angles and NOE restraints obtained. The MOLMOL program was used to visualize the result of the ensemble of minimized conformers (23).
The translational diffusion rates were measured by monitoring 1D 1H signal decay due to molecular diffusion in the z-direction of the sample using PFGs (18, 33) at variable concentrations of b30-82. In each experiment, the PFG strength was linearly increased from 2 to 95 G/cm, with a translational diffusion delay of 200 ms and total encoding and decoding gradient durations of 5 ms. The diffusion rates were estimated using TopSpin (Bruker BioSpin).
SDS-PAGE of the recombinant b30-82 that was produced, including the tether region and part of the dimerization domain of subunit b, revealed a prominent band of 6 kDa, which was found entirely within the soluble fraction. A Ni2+-NTA resin column and an imidazole gradient (15 to 300 mM) in a buffer consisting of 50 mM Tris-HCl (pH 7.5) and 250 mM NaCl was used to separate b30-82 from the main contaminating proteins. The protein eluting at 75 to 125 mM imidazole, b30-82, was collected and subsequently applied to an ion-exchange column (Resource Q). Analysis of the isolated protein by SDS-PAGE revealed the high purity of b30-82 (Fig. (Fig.1A).1A). The secondary structure of this subunit was determined from CD spectra, measured between 185 and 260 nm (Fig. (Fig.1B).1B). The minima at 222 and 208 nm and the maximum at 192 nm indicate the presence of α-helical structures in the protein, and the secondary structure was calculated to be 71% α-helix and 28% random coil. The ratio of the molar ellipticity at 222 nm to that at 208 nm (θ222/θ208) was determined to be 0.81.
To determine the first solution structure of the tether domain, b30-82 was analyzed by NMR spectroscopy. Well-dispersed cross peaks appear in the 2D 1H-15N HSQC spectrum, suggesting that the protein is folded (Fig. (Fig.2A).2A). Except for residues H38 and D82, all residues of b30-82 could be assigned by backbone 15N and 13C signals using 15N NOESY-HSQC as well as triple-resonance backbone experiments [HNCACB, CBCA(CO)NH]. The carbon (Δ13Cα) secondary shifts of b30-82 as a function of the protein sequence show a continuous stretch of positive combined carbon secondary shifts for residues 39 to 72 (Fig. (Fig.2B).2B). The positive carbon secondary shifts of residues 39 to 72 demonstrate the presence of a helical structure in this region. The 3D structure of subunit b30-82 was calculated based on a total of 464 NMR-derived distance restraints, of which 142 were intraresidual, 132 were sequential, and 73 were medium range. Hydrogen bond constraints were determined from the α-helical NOE pattern observed in the 3D 15N-edited NOESY-HSQC spectrum and were introduced at the end of structural calculations. Ribbon diagrams of the NMR solution structure of b30-82 and a representation of the molecular surface with the electrostatic potential of the peptide are shown in Fig. 3A to D. Figure Figure3A3A represents an overlay of the 10 lowest-energy structures of b30-82; the statistics are given in Table Table1.1. b30-82 adopts a regular α-helical conformation extending from residues 39 to 72, giving rise to a rod-like molecule. This might lead to an anisotropic tumbling motion, resulting in a broadening of NOESY peaks. However, since b30-82 is monomeric under the conditions used, more than 95% of backbone assignments were made, and all NOESY cross peaks have been assigned for the helical regions of this protein. The structures have overall RMSDs of 0.359 Å for backbone atoms and 0.964 Å for all heavy atoms in the helical regions (residues 39 to 72) of the 3D structures of the protein. All these structures have energies lower than 4 kcal·mol−1, no NOE violations greater than 0.5 Å, and no dihedral angle violations greater than 5°. Analysis of the Ramachandran plots shows 83.9% of residues in the most favored regions and 16.1% in the additionally allowed regions. Statistics on the structure calculation are presented in Table Table1.1. In the solution structure, b30-82 has a length of 48.07 Å excluding the unstructured regions in the N (residues 30 to 39) and C (residues 73 to 82) termini.
Since b30-82 includes a segment of the proposed dimerization domain (b53-122), the diffusion coefficient of b30-82 was measured as a function of protein concentration. As shown in Fig. Fig.4,4, the decrease in experimentally determined diffusion rates with increasing protein concentrations suggests dimer formation, assuming that the exchange between monomers and dimers is fast on the diffusion time scale. The corresponding dimer dissociation constant for b30-82 was estimated, by direct fit of translation diffusion rates as a function of concentration, to be 0.35 mM. The confidence limit for measurement of the dimer dissociation constant is 90% (±0.035 mM).
The solution structure of b30-82 shows a slightly twisted strip of hydrophobic alanine residues consisting of residues Ala61, Ala68, and Ala72 (Fig. 3B and D). To analyze the proximity of alanine residues of the first and second α-helices of the dimeric b30-82, Ala61, Ala68, and Ala72 were each separately replaced by a cysteine residue, generating the A61C, A68C, and A72C single mutants, respectively, of the soluble and monodisperse subunit b domain b22-156 (26). All three mutant proteins could be isolated in large amounts and at high purity (Fig. (Fig.5A).5A). CD spectroscopy revealed that the mutant proteins of b22-156 are 76% ± 2% α-helical, a structure comparable to that of wild-type b22-156 (26). Figure Figure5B5B shows CD spectra of the b22-156 A61C and A68C mutants, with θ222/θ208 ratios of 0.93 and 0.89, respectively, indicating an increased coiled-coil arrangement relative to that of b30-82 and a slight effect of alanine-to-cysteine mutation in the secondary structures of the mutant proteins. Cross-link formation in the presence of CuCl2 (10 μM) is shown in Fig. Fig.5C5C (lanes 1, 2, and 4), revealing that a significant amount of dimer formation could be generated for all three mutants analyzed. In contrast, when Ala70, located opposite Ala72 (Fig. (Fig.3D)3D) in the helix, was replaced with a cysteine, no significant disulfide formation could be detected under the conditions used (Fig. (Fig.5C,5C, lane 3).
The shape of the hydrated cytoplasmic b22-156 sequence of the stalk subunit b of E. coli F1Fo ATP synthase, determined by solution X-ray scattering, revealed that the protein is dimeric and an elongated particle of 16.2 ± 0.3 nm (26). b22-156 is described by a three-domain model, including the C-terminal δ-binding domain, the dimerization domain, and the tether domain (15). The tether domain, formed by residues 23 to 52, joins the membrane region (residues 1 to 22) to the beginning of the sequence essential for dimerization (residues 53 to 122). The structure of the synthesized 33-residue peptide consisting of the N-terminal membrane-spanning region shows an α-helical feature (14). The crystallographic structure of the major part of the dimerization domain, b62-122, revealed an α-helix with a length of 9.0 nm (12). Most recently, the NMR solution structure of the very C-terminal segment, b140-156, which interacts with the C terminus of subunit δ (δ91-177), has been determined; it adopts a stable helix formation in solution, with a flexible linker between amino acids 140 and 145 (26). The NMR solution structure of b30-82 leads to a more complete understanding of the structural puzzle of the peripheral stalk and provides insight into a segment which was the least defined in subunit b. The crystallographic structure of b62-122 and the NMR structure of b30-82 are superimposed well in the overlapping region, including residues 62 to 72, with an RMSD value of 0.4387 (Fig. (Fig.6A),6A), demonstrating that b30-82 is a structural continuum of b62-122. When positioned inside the sphere of the low-resolution solution structure of the dimeric b22-156, the three monomeric structures of b62-122 (12), b30-82, and b140-156 (26) were accommodated very well (Fig. (Fig.6C).6C). In addition, the structural model of the fitted monomeric NMR and crystallographic structures of b1-33, b30-82, b62-122, and b140-156 (Fig. (Fig.6A)6A) reveals that E. coli subunit b forms a single unbroken curved α-helix (excluding residues 35 to 38, not defined so far), which appears to be a relatively inflexible structure, a phenomenon described for subunit b of the mitochondrial F1Fo ATP synthase (13). This is also in line with mutation studies of the dimerization domain, demonstrating that substitutions affect the enzyme in assembly and/or enzyme function (8, 27). However, deletions and extensions in the C-terminal end of the tether domain and the N-terminal region of the dimerization domain did not alter the assembly or function of the enzyme (4). These data implied a rather flexible, rope-like feature of the bacterial b subunit, giving this subunit the possible role of an elastic peripheral stalk for transient storage of energy between steps in the rotary motion of the complex (7). The structure of the E. coli b subunit that is now available should enable the proposed elastic deformation to be measured.
As demonstrated in Fig. Fig.6B,6B, residues Ala32, Ala45, Ala50, Ala57, Ala61, Ala68, and Ala72 form a hydrophobic surface in the tether domain and the N-terminal segment of the dimerization domain. The cross-link formation of the b22-156 A61C, A68C, and A72C mutants confirm that these residues are located at one side of the helix and that the two cysteines in positions 61 and 61′, 68 and 68′, and 72 and 72′, respectively, of the first and second helices are in close proximity (Fig. (Fig.5C).5C). In contrast, only traces of a homodimer can be seen for the cysteine in position 70, reflecting the location of residue 70 at the opposite sides of the two helices (Fig. (Fig.3D).3D). Since the CD spectrum of b30-82 shows no strong coiled-coil formation, the two helices of the tether domain are predicted to be adjacent to each other, as proposed for the N-terminal domain of b1-34, in which aromatic residues form the hydrophobic interaction surface (14) and are located on the same surface side as residues Ala32, Ala45, Ala50, Ala57, Ala61, Ala68, and Ala72 (Fig. (Fig.6B).6B). Recently, T62C mutation and cross-linking experiments with this mutant have been performed using the entire E. coli F1Fo ATP synthase, and no disulfide formation could be detected in the holoenzyme (29). Figure 6A and D show that residue Thr62 is oriented about 120° away from the site of the hydrophobic surface, formed by residues Ala61, Ala68, and Ala72, supporting the absence of disulfide bond formation in the E. coli F1Fo ATP synthase T62C mutant (29). The structure of b30-82 also indicates that Gln64 is positioned at the exposed side of the helix, which has been used for binding of the donor in single-molecule fluorescence resonance energy transfer during ATP hydrolysis and synthesis of E. coli F1Fo ATP synthase (16, 29).
In summary, the data presented here demonstrate that the tether domain (b30-82) of E. coli subunit b exists mainly as an α-helix in solution. Fitting of the existing high-resolution structures of the N-terminal, tether, dimerization, and very C-terminal domains of E. coli subunit b suggests a rather inflexible, extended α-helical peripheral stalk, similar to the mitochondrial F1Fo ATP synthase b subunit and in line with the static and mechanistic properties of these classes of enzymes. The 3D structure of b30-82 in solution provides a structural basis for the design of biophysical experiments to enhance our understanding of the function of subunit b in E. coli F1Fo ATP synthase.
This research and the fellowships for R. Priya were supported by a grant from the Ministry of Education, Singapore (ARC 6/06 and RG144/06). S. Gayen is a recipient of the Graduate Research Scholarship, Nanyang Technological University, Singapore.
Published ahead of print on 9 October 2009.