A Bimodal Nickel-Binding Site on Escherichia coli [NiFe]-Hydrogenase Metallochaperone HypA

Abstract

[NiFe]-hydrogenase enzymes catalyze the reversible oxidation of hydrogen at a bimetallic active site and they are used by bacteria and archaea for anaerobic growth and pathogenesis. Maturation of the [NiFe]-hydrogenase requires several accessory proteins to assemble and insert the components of the active site. The penultimate maturation step is the delivery of nickel to a primed hydrogenase enzyme precursor protein, a process that is accomplished by two nickel metallochaperones, the accessory protein HypA and the GTPase HypB. Recent work demonstrated that nickel is rapidly transferred to HypA from GDP-loaded HypB within the context of a protein complex in a nickel selective and unidirectional process. To investigate the mechanism of metal transfer, we examined the allosteric effect of nucleotide cofactors and partner-proteins on the nickel environments of HypA and HypB by using a combination of biochemical, microbiological, computational, and spectroscopic techniques. We observed that loading HypB with either GDP or a non-hydrolyzable GTP analogue resulted in a similar nickel environment. In addition, interaction with a mutant version of HypA with disrupted nickel binding, H2Q-HypA, does not induce substantial changes to the HypB G-domain nickel site. Instead, the results demonstrate that HypB modifies the acceptor site of HypA. Analysis of a peptide maquette derived from the N-terminus of HypA revealed that nickel is predominately coordinated by atoms from the N-terminal Met-His-motif. Furthermore, HypA is capable of two nickel-binding modes at the N-terminus, a HypB-induced mode and a binding mode that mirrors the peptide maquette. Collectively, these results reveal that HypB brings about changes in the nickel coordination of HypA, providing a mechanism for the HypB-dependent control of the acquisition and release of nickel by HypA.

Graphical Abstract

Introduction

Metallochaperones

Many types of metal ions are essential nutrients because, among other roles, they serve as catalytic or structural cofactors for metalloenzymes.^1–4 To ensure sufficient metal ions to supply the enzymes, while preventing unfettered accumulation of these potentially toxic elements,^5–7 organisms use a coordinated network of metal homeostasis systems, including transporters, storage proteins, metallochaperones, and genetic regulators.^{5, 8–11} Typically, there are separate systems dedicated to the handling of each type of metal ion used by an organism. Such specialization ensures that each enzyme is loaded with the correct metal ion, but it also imposes the obligation of metal selectivity on the metal homeostasis protein factors. This challenge is often discussed in the context of the Irving-Williams series, which highlights that the relative stabilities of aqueous transition metal ion complexes of various small molecule ligands follow a common trend across the first row of the periodic table, and applies to many larger protein-metal complexes as well.^{9, 12, 13} One critical strategy to facilitate the correct distribution of metal ions is to maintain graded levels of availability, which results in restrained amounts of the metals that are at the ‘top’ of the Irving-Williams series, such as nickel, copper, and zinc, and is achieved through the activity of a series of finely-tuned metalloregulators.^{9, 13, 14} This strategy is complemented by the activities of the metallochaperones,^{10, 14, 15} a class of cytosolic proteins that traffic metal ions and ultimately deliver the required metal cofactors to the nascent enzymes.

Metallochaperones are typically dedicated to the production of one metalloenzyme.^14–16 They funnel the cognate metal ion to the enzyme precursor protein, so they are responsible for metal delivery in the context of the competitive cytosolic environment. How metal ions are piloted around the cell is a key question under investigation for the various systems, but it is likely that one common tactic is metal transfer within dedicated protein complexes. This strategy is highlighted by the activities of metallochaperones involved in copper homeostasis.^{14, 15} These metallochaperones use explicit protein-protein interactions to mediate metal transfer; copper is transferred through an associative ligand substitution reaction with proteinaceous ligands rather than a solvent-mediated exchange or a dissociative substitution reaction.^{14, 15, 17} This scheme enables partner specific transfer (through protein interactions), metal specificity (because copper has stable two-ligand coordination environments), and protected kinetic control of metal release (a solvent-mediated transfer would be slower because it requires an additional association step and risks loss of the metal ion into the cytoplasm). However, it is unclear how broadly this mechanism extends to other transition metals, or if metallochaperones would be expected to have an open or labile coordination site.

Another emerging class of metallochaperones is the G3E family of GTPases, which includes proteins with known or proposed duties as metallochaperones of the transition metals Ni, Co, Fe, and Zn.^{14, 18} It is speculated that the GTPase activity provides a regulatory element or the energy to activate metal transfer.^{14, 18, 19} Many of the G3E metallochaperones have a weak GTPase activity that is marginally influenced by the presence or absence of transition metals, indicating a link between metal binding and enzymatic activity.^{14, 20–22} Furthermore, several members of this family, such as MeaB, HypB, and UreG, display changes in partner protein interactions depending on the loading of GDP or GTP,^23–25 thereby providing a mechanism for the control of metal delivery and a mechanism for the acquisition and/or release of the substrate. However, for most of these systems, how the metal is transferred between the source and destination sites remains unresolved.

The nickel metallochaperones provide excellent systems to study the mechanisms of metal selectivity, transfer, and delivery because of the diversity of possible coordination environments, the position of nickel near the top of the Irving-Williams series, low availability in many niches,^{26, 27} and often, single-destination cellular systems.²⁸ Although there are fewer than a dozen known nickel enzymes, most of them are accompanied by a set of dedicated accessory proteins that are required for metallocentre assembly, including one or more metallochaperones. A case in point is the biosynthesis of the [NiFe]-hydrogenase enzymes.

[NiFe]-hydrogenase maturation

[NiFe]-hydrogenase enzymes catalyze the reversible oxidation of molecular hydrogen to protons at an intricate bimetallic active site.^{29, 30} This reaction and associated metabolic processes are implicated in the pathogenesis of a variety of microorganisms and are distinct from any human host systems, so the enzyme and its biosynthesis have been highlighted as promising antimicrobial targets.^31–35 Furthermore, [NiFe]-hydrogenases have potential applications in the hydrogen economy as sources of biohydrogen because they are oxygen-tolerant and use base metals for catalysis.^{30, 36} Therefore, the study of [NiFe]-hydrogenase maturation has additional applications aside from furthering the fundamental understanding of metallochaperone functions.

Assembly of the [NiFe]-hydrogenase metallocentre occurs in three discrete stages: (i) biosynthesis and delivery of the iron center to the active site on the large hydrogenase subunit, (ii) insertion of nickel, and (iii) cleavage of the C-terminus of the precursor protein to remove a short peptide followed by association with the small subunit.¹⁹ The three metallochaperones that are required to deliver nickel to the hydrogenase precursor proteins in E. coli are SlyD, HypB, and HypA (or the homologous HybF).^{19, 37} Deletion of the metallochaperone genes results in a hydrogenase-deficient phenotype that can be complemented by supplementing the growth media with close to millimolar concentrations of nickel,^38–42 highlighting the role of these proteins in delivering nickel within the context of a nutrient-limited environment. Recent work suggested that the fidelity of the Ni(II) insertion process is not attained by metal binding specificity of the individual proteins, but instead is achieved by metal-selective transfer that is mediated by protein-protein interactions between the metallochaperones.^{23, 43–45}

The nickel insertion step

HypA and HybF are homologous proteins that contribute to the production of the different hydrogenase enzymes in E. coli.¹⁹ The current model is that HypA/HybF serve to relay nickel from HypB to the hydrogenase precursor because they dock onto the hydrogenase enzyme precursor proteins^{46, 47} and also form complexes with HypB.^{43, 44} HypA/HybF have two metal-binding sites: a tetrathiolate zinc site that likely functions in a structural role^48–51 and a nickel-binding site that includes residues from the N-terminus.^{48, 51–53} HypA nickel coordination remains unclear despite the establishment of E. coli (Ec) HypA as a nickel metallochaperone more than two decades ago.⁵⁴ Analysis of several homologs revealed that the conserved histidine in the N-terminal MHE motif (His2) is required for nickel binding.^{23, 48, 51–53} Furthermore, a recent study of Helicobacter pylori (Hp) HypA demonstrated that the histidine must be at the second position in the sequence because insertion of leucine upstream abrogated nickel binding.⁵³ Spectroscopic analysis of HpHypA was interpreted as a paramagnetic octahedral site that included histidine and bidentate glutamate ligation.⁵³ This coordination environment is distinct from the nickel site revealed by the crystal structure of Thermococcus kodakarensis (Tk) HypA in a complex with TkHypBATPase, in which nickel is ligated by His2, His98, the N-terminal amine, and the first peptidyl backbone nitrogen in a pseudo-square planar geometry.⁴⁴

EcHypB is a GTPase with two distinct nickel-binding sites, both required for hydrogenase production in this organism, but it remains unclear if both sites are part of the same nickel pathway or if the binding sites have separate functions.^55–57 The N-terminal site or ‘high-affinity site’ is a seven-residue sequence that binds nickel with sub-picomolar affinity and is not well conserved.^{56, 58–60} The second nickel site is found in the G-domain of HypB and is conserved across bacterial species.^{56, 61, 62} Nickel binds to the G-domain site with low-micromolar affinity when HypB is loaded with GTP and the affinity is weakened by an order of magnitude in the GDP state.^{23, 56} In addition, nickel bound to the G-domain site of HypB moves to HypA and GDP-loading of HypB promotes both a stronger interaction with HypA and faster nickel transfer to HypA.^{23, 43} These observations suggest that following GTP hydrolysis, HypB is primed to deliver nickel to HypA in the context of the protein-protein complex. Zinc bound to the G-domain site of HypB inhibits the interaction with HypA,²³ providing a means of metal ion discrimination. Once nickel is loaded onto HypA, the protein complex is weakened, resulting in unidirectional nickel transfer from HypB to HypA.²³ How metal is transferred between these two proteins is unknown.

To uncover the mechanism of the nickel transfer process, we characterized the Ni(II) coordination environment of the G-domain of EcHypB with and without nucleotide and examined the effect that HypA and HypB have on each other by using mutant proteins with disrupted nickel binding. We found that none of GDP, GTP, or HypA exerted substantial effects on the primary nickel coordination environment of the HypB G-domain. Instead, HypB induced a change in the nickel coordination of HypA. We also demonstrated, by using spectroscopic, computational, biochemical, and microbiological analysis of the full-length protein as well as a peptide maquette of the MHE-motif of HypA, that the nickel coordination of EcHypA exists in a mixture of two binding modes: a mode confined to the N-terminal sequence and a HypB-induced mode. This work suggests that the N-terminus of HypA is an adjustable nickel-binding site that is impacted within the complex with HypB, the formation of which is controlled by the GTPase cycle. Altogether, these experiments provide a mechanism for the acquisition and release of nickel by HypA and shed light on how the nickel delivery process is controlled during the maturation of [NiFe]-hydrogenase in E. coli.

Materials and Methods

All the chemicals and buffers used were obtained from BioShop Canada, Bio-Rad, Invitrogen, or Sigma-Aldrich, and were either biology grade or certified ACS reagents. Electronic absorption spectra were recorded on an Agilent 8453 spectrophotometer unless otherwise noted. HypA_Pep (sequence: MHEITW) was ordered from Biomatik (Cambridge, ON) at 98% purity with a C-terminal methyl ester modification. Tryptophan was included at the end to provide a spectroscopic handle. Bacterial strains used are identified in Table S1. All proteins studied and discussed here pertain to the E. coli variant unless otherwise noted.

Cloning, expression, and purification

Mutations were introduced into the pBAD18-kan-HypA_Str plasmid to create the variants used in this work (Table S2) in one step by using QuikChange mutagenesis with the appropriate primers (Table S3). All mutations were verified by sequencing (ACGT, Toronto). The HypA constructs used in this study include a C-terminal Strep-tag to aid in purification and are denoted with a “Str” subscript. This modification to HypA does not affect hydrogenase production in E. coli, and in vitro HypA_Str is functional at complex formation with HypB and accepts nickel from HypB within the protein complex.^{23, 43, 46} HypA_Str and H2Q-HypA_Str were expressed from a pET24b plasmid and were isolated from cell lysates as previously described by using Strep-Tactin Sepharose affinity purification.⁴³ After initial purification through Strep-Tactin Sepharose resin (IBA Life Sciences), HypA_Str and H2Q-HypA_Str were dialyzed [20 mM Tris, 10% glycerol pH 7.5, and 1 mM TCEP] and either stored at −80 °C until use or purified through a HiTrap Q HP column (GE Healthcare) if the amount of zinc loaded was < 90%. SDS-PAGE with Coomassie blue stain confirmed purity from other proteins and >90% zinc loading was verified by non-denaturing ESI-MS and PAR assays.⁴³ The masses of purified proteins were verified by ESI-MS (AIMS Laboratory, University of Toronto).

Wild-type EcHypB and mutant versions were prepared as previously described.^{43, 56} To prepare NB-HypB, a C166A mutation was introduced into TM-HypB-pET24b plasmid by using Phusion mutagenesis followed by the C198T mutation by using QuikChange mutagenesis. All mutations were verified by sequencing (ACGT, Toronto).

Competition experiments

The affinities of HypA_Pep and NB-HypB for nickel were determined through competition with the metal-sensitive dye mag-fura-2 (MF2). The buffer [25 mM HEPES, pH 7.5, 100 mM KCl, 10% glycerol, and 5 mM MgSO₄] was treated with Chelex 100 resin (Bio-Rad) prior to addition of MgSO₄ and stirred overnight under the anaerobic atmosphere to ensure removal of dissolved oxygen. Protein was buffer exchanged using a 3 kDa Amicon Ultra-4 Centrifugal Filter Device (Millipore, MWCO). Where appropriate the buffer was supplemented with 500 μM nucleotide, either GDP or GppCp. Competition experiments with NB-HypB were performed by incubating 20 μM NB-HypB and 1.0 μM MF2 with 0–1000 μM NiSO₄ overnight under an anaerobic atmosphere (95% N₂ and 5% H₂) at 4°C. After incubation, the samples were removed from the anaerobic atmosphere, transferred to a multiwell plate, and immediately analyzed. The fluorescence (λ_em = 340 nm λ_ex = 510 nm) of MF2 was recorded on a CLARIOstar plate reader (BMG Labtech) and fractional saturation was calculated based on the maximal and minimal fluorescence. The affinity of HypA_Pep for nickel was calculated by using the competition experiments previously used to calculate the apparent affinity of HypA_Str.⁴³ Briefly, lyophilized HypA_Pep was dissolved into working buffer (25 mM HEPES, pH 7.5, 100 mM KCl), and the concentration HypA_Pep was quantified using the extinction coefficient 5500 M⁻¹cm⁻¹ at 280 nm.⁶³ A solution of 20 μM HypA_Pep and 20 μM MF2 was incubated with 0–100 μM NiSO₄ overnight at 4°C. The solutions were warmed to RT and the fractional saturation of MF2 with Ni(II) was calculated from the ratiometric [A₃₂₂ nm/(A₃₂₂ nm + A₃₆₉ nm)] absorption data. The MF2 apparent dissociations constants were calculated from direct nickel titrations with the appropriate buffer conditions, including the presence of nucleotide, and methodologies. The apparent dissociation constants of the peptide and proteins were calculated by fitting the competition fractional saturation data to a 1:1 binding model or, in the case of HypB constructs, a 1:2 binding model, using a custom DynaFit script (Figure S1).

XAS sample preparation

TM-HypB, NB-HypB, HypA_Str, and H2Q-HypA_Str protein stocks were buffer exchanged into 25 mM HEPES, pH 7.5, 100 mM KCl, 10% glycerol, and 5 mM MgSO₄ under anaerobic atmosphere (95% N₂, 5% H₂) by using a 3 kDa Amicon Ultra-4 Centrifugal Filter Device (Millipore, MWCO) and finally concentrated to a volume of less than 100 μL. In the case of the TM-HypB and H2Q-HypA_Str mixture, proteins were mixed in equimolar amounts prior to buffer exchange and concentration. The protein concentrations in each sample ranged from 0.4 – 3 mM. Nucleotide was added to a final concentration of 5 mM. Since the stoichiometry of HypB-Ni(II) complexes with nucleotide is 0.5 Ni(II) per HypB, 0.4 equivalents of Ni(II) were added to each sample that contained HypB and 0.8 equivalents were added to each sample that contained only HypA. Nickel was added in aliquots to prevent protein precipitation or aggregation. Samples were transferred to 2 mm path length (2 mm × 4 mm × 20 mm) cuvettes of either Lucite (University of Saskatchewan, Saskatoon, Canada) or polyoxymethylene copolymer (Vantec, Saskatoon, Canada), flash froze with liquid nitrogen, and stored at −80°C until analysis.

X-ray absorption spectroscopy

X-ray absorption spectroscopy (XAS) measurements were conducted at the Stanford Synchrotron Radiation Lightsource (SSRL) in Menlo Park, California, USA using the data acquisition program, XAS-Collect.⁶⁴ Ni K-edge data were collected on the biological XAS beamline (BL) 7–3 with the SPEAR storage ring containing 500 mA at 3.0 GeV. Beamline 7–3 utilizes a Si(220) double-crystal monochromator, and either a rhodium-coated vertically collimating mirror upstream of the monochromator, which achieves harmonic rejection by adjusting the mirror cut-off angle (e.g. Ni K-edge spectra are collected at the 12 keV cut-off) or harmonic rejection was achieved by detuning one monochromator crystal to 50% peak intensity. Samples were maintained at ~10 K using a liquid helium flow cryostat during data collection. Incident and transmitted X-ray intensities were measured using nitrogen-filled gas ionization chambers. Fluorescence spectra were collected by monitoring the K_α fluorescence using a 30-element germanium detector. The monochromator energy was calibrated by reference to a standard nickel foil measured simultaneously with the sample, assuming a lowestenergy inflection point of 8331.6 eV for Ni(II). Each data set is a collection of a minimum of 6 scans.

XAS data analysis

XAS data reduction and analyses were performed using the EXAFSPAK suite of computer programs (George, G. N. (2001) EXAFSPAK, URL: http://ssrl.slac.stanford.edu/exafspak.html). Data channels and scans were examined for differences and any spurious channels or sweeps were removed. The energy was re-calibrated using the lowest energy inflection of the elemental foil (internal standard), and successive scans were averaged. A polynomial pre-edge fit was used to subtract background fluorescence from scatter and filter fluorescence, and a polynomial spline was used to extract the extended x-ray absorption fine structure (EXAFS) oscillations, χ(k). Fourier transforms of the EXAFS oscillations were phase corrected. Cartesian coordinates for each environment were generated in Gaussian 09 (see below) and used by FEFF8.5L^{65, 66} to calculate multiple scattering paths from the absorbing Ni atom. These ab-initio theoretical phase and amplitude functions were then used in the curvefitting program, OPT (a component of EXAFSPAK). Histidine multiple scattering paths were modelled by assuming coordinated imidazole. Multiple scattering paths for second coordination sphere C and N atoms from the imidazole ring were linked to the coordinating nitrogen atom in the fit but were omitted from the fitting tables for clarity.

Non-denaturing electrospray mass spectrometry analysis

HypA_Pep was resuspended into 10 mM ammonium acetate, pH=7.5, and then diluted to 10 μM in the same buffer. For nickel-binding analysis, the peptide was briefly (<5 min) incubated with 0–100 μM nickel acetate prior to injection. The mass spectrometry data were recorded on an AB Sciex QStar XL mass spectrometer with a hot source-induced desolvation (HSID) interface (Ionics Mass Spectrometry Group Inc.) as previously described.⁴³

Molecular modelling

Density functional theory (DFT) calculations were performed using Gaussian 09⁶⁷ on SHARCNET to examine HypA nickel geometries. Calculations used the Becke three-parameter hybrid functional combined with Lee-Yang-Parr correlation functional (B3LYP)^{68, 69} and the Dunning correlation-consistent polarized double zeta basis set (cc-pVDZ).⁷⁰ All computations were performed using the self-consistent reaction field solvation models (SCRF) to accommodate potential solvent effects. Only singlet states were considered because NMR data indicated HypA_Str and HypA_Pep bound nickel was diamagnetic.

Whole-cell hydrogenase activity assay

Escherichia coli from glycerol stocks were plated on LB-agar and grown overnight at 37°C. A single colony was selected and grown to saturation in LB media overnight at 37°C. TYET media (10 g/L Tryptone, 5 g/L yeast extract, and 50 mM Tris pH=7.5) was supplemented with 0.4% glucose, 30 mM sodium formate, 1 μM sodium selenite, and 1 μM sodium molybdate and sterilized using 0.22 um PES membrane Stericup filtration apparatus (EMD Millipore). Additional media components (including arabinose and antibiotic) were added when required. Supplemented TYET was inoculated with 0.8 % v/v from the overnight culture, and 200 μL of the inoculated media was added to the wells of a Corning Costar Round Bottom Polystyrene 96-well plate. The plates were covered with the supplied lid and incubated for 6 hours at 37°C. Plates were cooled to room temperature and the optical density (OD) at 630 nm was recorded. A developing solution (10 mg/mL benzyl viologen and 250 mM sodium formate in 20 mM Tris buffer at pH=7.5) was added to each well (20 μL) and the change in absorbance at 630 nm, caused by the reduction of benzyl viologen, was recorded on a ELx808 (BioTek) 96-well plate reader every 30 seconds for 5 minutes. Data were converted to ΔAbs·min⁻¹ divided by the OD_{630 nm} to adjust for growth and normalized to the average activity of the wild-type strain.

Magnetic susceptibility

Magnetic susceptibility of nickel was determined using Evans’ method,^71–73 a technique that monitors paramagnetic induced chemical shifts of protons in a ¹H NMR spectrum. Proteins were buffered exchanged into D₂O buffer containing 25 mM HEPES (pD 7.5), 100 mM KCl, and 5 mM MgSO₄, loaded with 500 μM GDP where applicable, and separated into two aliquots. One aliquot was loaded to 80% occupancy with nickel using NiSO₄ stocks in D₂O and loaded into a coaxial insert tube (Wilmad-LabGlass) that was placed in an NMR tube containing the other aliquot (without added nickel). Final protein concentrations ranged from 60 μM – 400 μM. The ¹H NMR spectra were recorded on an Agilent DD2 700 spectrometer. NiSO₄ in the same buffer was used as a paramagnetic control. A paramagnetic centre will shift the HDO peak and produce two separate peaks whereas a diamagnetic centre will result in a single peak.

Pull-down assay

Pull-down assays were performed to confirm the interaction between HypA and HypB constructs. HypA (50 μM) and HypB or NB-HypB (50 μM) were incubated for 30 min at 4°C in buffer containing 20 mM Tris, pH 7.5, 100 mM NaCl, 1 mM TCEP, 1 mM PMSF, 500 μM GDP, and 0.1% BSA. The mixtures were loaded onto Strep-Tactin Sepharose resin (IBA Life Sciences), washed with 60x resin volume, and eluted with 200 μL buffer containing 2.5 mM d-desthiobiotin. Fractions were concentrated using SpeedVac concentrator, resolved on 12.5% SDS-polyacrylamide gels, and transferred to polyvinylidene difluoride membranes (Millipore). The blots were probed with an anti-HypB polyclonal rabbit antibody raised against purified HypB peptide⁴⁵ at a 1:1000 dilution and then secondary goat anti-rabbit (Bio-Rad) antibodies conjugated to horseradish peroxidase at a dilution of 1:30 000. Enhanced chemiluminescence (SuperSignal West Pico Chemiluminescence, Pierce) was used for detection.

Results

Nucleotide and HypA do not substantially affect HypB nickel coordination

Previous studies demonstrated that nickel transfer from the G-domain site of HypB to HypA is faster when HypB is loaded with GDP compared with a non-hydrolysable GTP analog (GppCp).²³ One possible explanation for the acceleration of nickel transfer could be that the type of nucleotide has an allosteric impact on the coordination environment of nickel bound to the HypB G-domain site. This model would allow for control of the nickel-binding environment through the different stages of the HypB GTPase cycle and would be consistent with the UV-vis analysis of nickel bound to HpHypB.⁶² To test this hypothesis, X-ray absorption spectroscopy (XAS) data were collected on solutions of nickel bound to a C2,5,7A mutant of HypB, termed triple mutant or “TM” HypB, which does not bind nickel at the N-terminal high-affinity site of HypB.⁵⁶ This construct allows characterization of the G-domain site in the context of the full-length protein. TM-HypB nickel complexes were prepared in the absence or presence of nucleotide (either GDP or GppCp) – and are denoted TM-HypB, TM-HypB(GDP) and TM-HypB(GppCp).

We found that the nickel K-edges for each of the three nucleotide conditions are similar (Figure 1), and given that near-edge spectra are diagnostic of the coordination geometry and the coordinating ligands,^{74, 75} these results suggest that there are no major changes in the type of ligands or coordination geometries in these samples. The near-edge data of nickel loaded in the G-domain of TM-HypB are consistent with the previously reported XAS of nickel bound to the isolated HypB G-domain.⁵⁷ In addition, examination of the extended X-ray absorption fine structure (EXAFS) (Figure 1), supports the conclusion that all three environments are similar, although subtle differences are noticeable, particularly in the first major oscillation, between the spectrum of nickel bound to TM-HypB in the absence of nucleotide and those of GDP or GppCp loaded protein. This difference is also apparent in the major peak of the Fourier transform (Figure 1), and can be attributed to slight changes in bond-lengths. EXAFS curve fitting indicates that an S₂(N/O)₂(His) coordination environment fit best to all three spectra. These spectra were also fit reasonably well to an S₂(N/O)(His) 4-coordinate model (Table 1). Furthermore, the Ni(II)-TM-HypB EXAFS could also be fit to an S(N/O)₂(His) environment, highlighting the differences between the nucleotide and no-nucleotide conditions. All fits contain one histidine ligand, which is distinguished from other types of nitrogen ligands by the characteristic backscattering due to the imidazole ring. Altogether, the changes upon the addition of nucleotide appear to be subtle, and no difference was detected between the Ni(II)-TM-HypB(GDP) or Ni(II)-TM-HypB(GppCp) samples.

Figure 1.

XAS of Ni(II)-TM-HypB with no nucleotide, GDP, or GppCp. Comparison of (A) the near-edges, (B) the k³-weighted EXAFS, and (C) the Fourier transforms of the Ni(II)-TM-HypB (black), Ni(II)-TM-HypB(GDP) (yellow), and Ni(II)-TM-HypB(GppCp) (blue) XAS data. All three spectra are similar, indicating that the nickel coordination environments are similar. The fits (dashed lines) of the k³-weighted EXAFS data and the Fourier transforms of each nickel species (solid line) correspond to S₂(N/O)₂(His) site (Table 1).

Table 1.

EXAFS curve fitting results of Ni(II)-TM-HypB with no nucleotide, GDP, or GppCp

Sample Description	A-Bsa	N	R (Å)	σ2 (Å2)	ΔE0 (eV)	F-factor
Ni(II)-TMHypB	Ni-S	2	2.211(2)	0.00403(14)	−5.5(5)	0.256
	Ni-N/O	2	2.098(5)	0.00702†
	Ni-His	1
	Ni-S	2	2.213(3)	0.00373(13)	−8.4(6)	0.272
	Ni-N/O	1	2.065(8)	0.0065†
	Ni-His	1
	Ni-S	1	2.224(4)	0.00231(14)	−3.5(5)	0.299
	Ni-N/O	2	2.078(5)	0.0042†
	Ni-His	1
Ni(II)-TMHypB(GDP)	Ni-S	2	2.209(3)	0.00318(14)	−8.8(6)	0.286
	Ni-N/O	2	2.074(6)	0.0055†
	Ni-His	1
	Ni-S	2	2.210(3)	0.00269(14)	−11.4(7)	0.295
	Ni-N/O	1	2.046(9)	0.0047†
	Ni-His	1
Ni(II)-TMHypB(GppCp)	Ni-S	2	2.206(2)	0.00320(11)	−9.3(6)	0.243
	Ni-N/O	2	2.064(5)	0.0055†
	Ni-His	1
	Ni-S	2	2.204(2)	0.00297(12)	−12.2(6)	0.243
	Ni-N/O	1	2.037(8)	0.0048†
	Ni-His	1

^aA-Bs denotes absorber and backscatterer interaction; N denotes coordination number; R is given in Å and represents interatomic distances; σ² given in Ų, are the Debeye-Waller factors (mean-square deviations in interatomic distance); the threshold energy shifts, ΔE₀ are given in eV. The values in parentheses are the estimated standard deviations obtained from the diagonal elements of the covariance matrix. The F-factor or fit-error function is defined as (Σk⁶(χ (k)_calcd - χ (k)exptl)²/ Σk⁶ χ (k) _exptl²)^1/2. The summation is over all data points included in the refinement.

^bThe best fit models are bolded.

^†Denotes that the variable value was numerically linked to the corresponding variable in the interaction listed in the table row above in order to reduce the total number of degrees of freedom in the refinement.

The similar nickel coordination environments of the G-domain of HypB when nucleotide was added was unexpected, given that the stoichiometry of metal bound to the protein in solution decreases from one nickel ion per monomer to one ion per dimer upon binding to either GDP or GppCp.²³ An important consideration is that the XAS samples were prepared with high concentrations of protein (~ 1 mM), a condition that would increase HypB homodimer formation, as well as 0.4 equivalents of nickel to minimize any unbound metal interfering with the analysis, so the nucleotide-dependent changes that are observed with low protein concentrations may not be observable with this method. Furthermore, XAS data are limited because they cannot determine if similar ligands are swapped and the overall coordination environment is held chemically similar, so changes in the nickel site that maintain a similar type of coordination set upon adding nucleotide to TM-HypB cannot be ruled out. Perhaps more important, however, is that the lack of a dramatic change of HypB nickel coordination in samples loaded with GDP versus the GTP analog, states which both bind one nickel ion per dimer, suggests that the faster nickel transfer observed from the GDP-loaded protein to HypA, compared to the GTP analog, is not due to a major change in the HypB coordination site.

Another possible model for the rapid nickel transfer from the G-domain site of HypB(GDP) to HypA is that the HypB nickel-binding site is modulated upon complex formation with HypA, which is stronger in the GDP-loaded state. To test this possibility, nickel XAS data were collected on a solution containing both H2Q-HypA_Str and TM-HypB. The H2Q-HypA_Str construct, which cannot mature [NiFe]-hydrogenase in vivo,²³ allows for the investigation of changes of the HypB G-domain Ni(II) environment induced by HypA because H2Q-HypA_Str still forms a complex with HypB but has significantly weakened nickel-binding affinity and it does not accept nickel from HypB.²³ Since H2Q-HypA_Str can bind nickel, albeit weakly,²³ we analyzed the nickel environment of H2Q-HypA_Str as a control (Figure 2), and it was distinct from that of Ni(II)-HypA_Str, consistent with previous biochemical characterization (vide infra).²³ It is likely that this XAS spectrum represents nickel that is weakly and non-specifically bound to H2Q-HypA_Str.

Figure 2.

XAS analysis of Ni(II)-H2Q-HypA+TM-HypB(GDP). (A) Comparison of Ni(II)-H2Q-HypA_Str (purple), Ni(II)-TM-HypB(GDP) (yellow), and Ni(II)-H2Q-HypA_Str+TM-HypB(GDP) (pink) near-edge X-ray absorption spectra. Reconstruction (dashed lines) of the (B) near-edge and (C) k³-weighted EXAFS data of Ni(II)-H2Q-HypA_Str+TM-HypB(GDP) was achieved with a combination of 80 % Ni(II) TM-HypB(GDP) and 20 % Ni(II)-H2Q-HypA_Str, suggesting that the nickel is bound to either of the individual sites and not to a new species.

The nickel K-edge of the sample containing both H2Q-HypA_Str and TM-HypB(GDP) is different from those of either of the individual proteins alone (Figure 2). However, it was possible to reconstruct the near-edge of the mixture from those of the individual components, suggesting that the solution is composed of approximately 80% Ni(II)-TM-HypB(GDP) and 20% Ni(II)-H2Q-HypA_Str (Figure 2). The same proportions were used to reconstruct the EXAFS curve as well, supporting the presence of the two separate nickel-binding sites in the sample. This analysis suggests that the HypA interaction with HypB does not have an allosteric impact on the HypB G-domain nickel-binding site, and is consistent with previously reported solution experiments with a fluorescent indicator, which indicated that H2Q-HypA_Str does not alter the affinity of Ni(II) binding to the HypB G-domain.²³

HypB alters Ni(II)-HypA coordination

Since no significant change in the Ni(II) coordination of HypB was detected upon the addition of H2Q-HypA, we sought to investigate the converse effect, whether HypB induces a change in HypA. To test this possibility, we collected XAS data of Ni(II) in the presence of wild-type HypA_Str alone and a mixture containing both HypA_Str and TM-HypB(GDP). In a similar vein to the previous analysis, we attempted to reconstruct the near-edge of Ni(II)-HypA_Str+TM-HypB(GDP) from the spectra of the individual components and found that the mixture could not be well constructed (Figure S2). This indicated that a new nickel coordination environment is likely formed when the two proteins are mixed together. In order to probe these changes, we sought to further characterize the coordination environment of Ni(II)-HypA.

Ni(II)-HypA XAS

Analysis of the XAS data from samples containing HypA is confounded by the fact that the Ni(II)-binding site of EcHypA has not yet been characterized. In addition, studies of homologous proteins have provided conflicting models, although it is clear that the histidine of the highly conserved MHE motif at the N-terminus of these proteins serves as a nickel ligand.^{44, 51, 53} To investigate this site we collected XAS of Ni(II) bound to HypA (vide infra). The near-edge spectrum has a 1s→4p_z transition at 8336 eV and white line intensity of 1.34 (normalized to the edge jump). Preliminary analysis of the EXAFS region provided a fit of (N/O)₄ or S(N/O)₃ (data not shown). The possible inclusion of sulfur in the primary coordination sphere led to speculation that side chains from amino acids that are predicted to exist in spatial proximity to the N-terminus might also play a role in coordination.

To investigate additional residues as potential ligands, we first examined reports of the nickel coordination environment of the homologous protein HpHypA. NMR data of HpHypA suggested that HpAsp40 is likely involved with nickel coordination.^{51, 76} Based on this information, along with homology models generated from the HpHypA and TkHypA structures (SWISS-MODEL Repository),⁷⁷ we identified Cys38 and Glu40 of EcHypA as residues that may be in proximity to the N-terminus and serve as possible ligands. We also examined the role of Glu3 because previous analysis of E. coli HybF demonstrated that an E3L mutant was deficient in hydrogenase maturation yet the E3Q-HybF mutant was active.⁴⁸ Finally, an H2Q-HypA mutant was used as a negative control because this residue has been demonstrated to have a role in nickel coordination as well as hydrogenase maturation, and HypA_Str was included as a positive control to confirm that the C-terminal tag did not prohibit the function of HypA in vivo.²³ To determine if these residues are important for HypA function in hydrogenase maturation, the impact of the mutations was tested by measuring hydrogenase activity in a whole-cell assay that monitors the colorimetric reduction of benzyl viologen. The experiments were performed with DPABF E. coli, a strain that is lacking endogenous HypA or HybF and thus does not produce active hydrogenase, and complementation of the hydrogenase deficiency was examined following the introduction of wild-type and mutant versions of the hypA gene on an arabinose-inducible plasmid. All mutant variants except for H2Q-HypA were able to support the production of active [NiFe]-hydrogenase, indicating that the substituted residues are dispensable for hydrogenase maturation (Figure 3). Collectively, the hydrogenase activity data did not reveal any likely candidates for additional residues in the ligand coordination sphere of HypA, leading to the possibility that the EcHypA nickel site is comprised solely of residues at the N-terminus.

Figure 3.

Growth and whole-cell hydrogenase activities of HypA_Str variants. E. coli strains MC4100, DPABF (hypA ATG→TAA, ΔhybF), and DPABF transformed with wild-type and mutant versions of the pBAD18-kan-hypA_str plasmid were grown in supplemented TYET media for 6 hrs at 37 °C. The (A) growth and (B) hydrogenase activities of whole-cells, measured by monitoring the reduction of benzyl viologen and normalized to growth, were recorded. All HypA_Str mutant constructs, except H2Q, were able to produce active hydrogenase, indicating that they are dispensable for the metallochaperone function. The results represent the average values from three biological replicates and the error bars indicate ± one standard deviation. * indicates a significant (p < 0.05) change from the MC4100 result.

Ni(II)-HypA_Pep

To test the possibility that the complete nickel-binding site of HypA is localized to the N-terminal sequence of the protein, we created a peptide maquette of the first five residues of HypA (HypA_Pep) with the sequence MHEITW. Tryptophan was included at the end of this construct as a spectroscopic tag, the C-terminus was capped with a methyl ester to prevent metal coordination, and the N-terminus was left as a free amine. The nickel stoichiometry and affinity of HypA_Pep were determined and compared to those of the full-length protein. First, we confirmed that the peptide binds nickel with a stoichiometry of 1:1 by using ESI-MS (Figure S3). Second, the nickel affinity of HypA_Pep was determined by using competition experiments with the metallochromic indicator mag-fura-2 (MF2). The data and the corresponding fits (Figure S3) indicate that the peptide binds nickel with an apparent dissociation constant of 100 ± 30 nM, a result that is similar to the measured affinity for the full-length HypA_Str of 40 to 80 nM.^{23, 43} We also characterized the spin-state of this system by using Evans’ method, an NMR-based technique to measure the magnetic susceptibility of Ni(II), which revealed that Ni(II)-HypA_Pep and Ni(II)-HypA_Str are both diamagnetic (data not shown). These results indicate that the nickel-binding sites of the peptide and the full-length protein have similar properties.

Given that the HypA_Pep reproduces the nickel-binding activity of the full-length protein, we then computationally investigated the HypA_Pep nickel site by using DFT to find an optimized geometry consistent with the XAS (vide infra) and NMR results. This computational work was performed with a simplified N-terminal site that only included the first 3 amino acids, MHE, to reduce the computational cost, and the C-terminus was capped with N-methyl amide. Inspection of HypA_Pep suggested that possible coordinating atoms include the terminal amine, the deprotonated amide nitrogens from His2 and Glu3, an imidazole nitrogen of His2, and the carboxylate of Glu3. Initial attempts to include the amide nitrogen of Glu3 in nickel coordination resulted in a planar chelate consisting of the terminal amine and both amide nitrogens, forcing His2 ligation to the axial position. We found that under these conditions His2 tended to dissociate during optimization, a result that is not consistent with biochemical and spectroscopic analysis that provide evidence for His2 as a coordinating group. It is therefore likely that the amide of Glu3 is protonated and does not coordinate nickel in this system. Examination of histidine ligation also indicates that the δ-nitrogen of histidine, rather than the ε-nitrogen, is the coordinating nitrogen because of geometrical constraints.

Using these constraints and a variable number of possible coordinating water molecules, the geometry of the peptide was optimized to form a distorted square-planar nickel site. This model includes nickel ligation from the terminal amine, the amide nitrogen from His2, the δ-nitrogen of histidine and either an oxygen from the Glu3 sidechain or an oxygen from water (Figure 4 and Table S4). The first three ligands noted above form a chelate (termed HypA chelate) with 5- and 6-member rings, and these rings force a relatively short Ni(II)-N bond between nickel and the amide nitrogen. This structure is reminiscent of the ATCUN (Amino Terminal Cu(II)- and Ni(II)-Binding) motif, where histidine ligation is on the same plane as other coordinating nitrogen atoms.⁷⁸ A similar structure was also observed in Thermoanaerobacter tengcongensis NikM, which is the substrate-binding component of an ECF-type nickel transporter.⁷⁹ Finally, the fact that the hydrogenase activity screen of the HypA mutants revealed that Glu3 is dispensable for the maturation function of HypA indicates that it is likely not needed for nickel coordination, in contrast to the complete disruption of hydrogenase production observed upon mutating His2 in the H2Q-HypA construct.²³ This result provides support for a water molecule as the fourth ligand instead of the Glu3 sidechain, although further structural characterization is needed to confirm this assignment.

Figure 4.

Optimized Ni(II) structures and XAS of Ni(II)-HypA_Pep, Ni(II)-HypA_Str and Ni(II)-H2Q-HypA_Str. Optimized structures of the HypA N-terminal sequence reveal distorted square planar geometries with coordination from the terminal amine, the first backbone amide nitrogen, the δ-nitrogen of His2, and (A) Glu3 (B) or a water molecule. Comparison of (C) the near-edges and (D) the k³-weighted EXAFS data of Ni(II)-HypA_Str (green), Ni(II)-H2Q-HypA_Str (purple), and Ni(II)-HypA_Pep (red) XAS. All three spectra are different, indicating that nickel is in a different coordination environment in each sample. (E) The k³-weighted EXAFS and (F) the Fourier transform of Ni(II)-HypA_Pep (solid line) and the best fit (dashed line) corresponding to an (N/O)₃His coordination environment.

To compare the nickel site of HypA_Pep to that of HypA_Str at a higher resolution and validate this simplified binding site, we completed XAS on Ni(II)-HypA_Pep (Figure 4). The Ni(II) near edge spectra exhibits a small 1s→4p_z transition as well as white line (WL) with an intensity of ~1.3 (normalized to the edge jump) (Figure 4). These results are not consistent with an idealized square planar geometry which has a prominent 1s→4p_z transition.⁷⁵ EXAFS fitting of Ni(II)-HypA_Pep is consistent with either of the DFT optimized structures (Figure 4, Table 2). In particular, the back-scattering caused by the imidazole nitrogen and the carbons in the HypA_Pep chelate, which were included in the FEFF analysis, is observed in the 3–4 Å range. The bond-lengths from the EXAFS fitting are longer than from the DFT calculations, but show a similar trend including a shorter Ni(II)-N bond that is assigned as the amide nitrogen. However, the bond lengths between the scattering paths are within the resolution from the EXAFS data, so a simplified (N/O)₃His environment, which is consistent with the DFT results, was selected as the best fit. Finally, a comparison of these data to those of nickel bound to the full-length HypA_Str, protein demonstrates differences in both the near-edge and EXAFS regions that indicate that the two binding sites are not identical (Figure 4). The variations between the nickel environments found on HypA_Pep and HypAStr, as well as that of TM-HypB+HypA_Str, lead to the hypothesis that the HypA nickel coordination is bimodal.

Table 2.

EXAFS curve fitting results of Ni(II)-HypA_Pep

Sample Description	A-Bsa	N	R (Å)	σ2 (Å2)	ΔE0 (eV)	F-factor
Ni(II)-HypAPep	Ni-Hisc	1	2.052(3)	0.0025(3)	−1.6(4)	0.355
	Ni-N(term.)c	1	2.064†	0.0025†
	Ni-O(water)c	1	2.096†	0.0023†
	Ni-N(amide)c	1	1.924(6)	0.0025†
	Ni-N/O	3	2.086(4)	0.0047(4)	2.7(4)	0.355d
	Ni-His	1	1.986(10)	0.0067(13)
	Ni-N/O	2	2.078(9)	0.0073(4)	3.3(3)	0.389
	Ni-His	2	2.057(8)	0.0074†

^aA-Bs denotes absorber and backscatterer interaction; N denotes coordination number; R is given in Å and represents interatomic distances; σ² given in Ų, are the Debeye-Waller factors (mean-square deviations in interatomic distance); the threshold energy shifts, ΔE₀ are given in eV. The values in parentheses are the estimated standard deviations obtained from the diagonal elements of the covariance matrix. The F-factor or fit-error function is defined as (Σk⁶(χ (k)_calcd - χ (k)exptl)²/ Σk⁶ χ (k)_exptl²)^1/2. The summation is over all data points included in the refinement.

^bThe best fit models are bolded.

^cModels were based on HypA-DFT structures and include a backscattering histidine in the HypA chelate. Absorbers assignments are based on bond length comparison to the HypA DFT structure.

^†Denotes that the variable value was numerically linked to the corresponding variable in the interaction listed in the table row above in order to reduce the total number of degrees of freedom in the refinement.

^dWe note that the value for ΔE₀ for this fit differs from other best fits in this paper (Table 1) and that a different DFT model yielding more consistent ΔE₀ values might be possible.

Bimodal coordination of Ni(II)-HypA

To investigate further the possibility that HypB influences the nickel coordination environment on HypA, we constructed a variant of HypB that is incapable of binding nickel, referred to as “non-binding” HypB (NB-HypB). The NB-HypB variant bears the C198T and C166A mutations, to abrogate or weaken Ni(II) binding to the G-domain site in the context of the TM-HypB construct.^{56, 57, 62} This mutant HypB still forms a complex with HypA_Str, although the interaction is weaker than with wild-type HypB, as observed by using a pull-down assay (Figure S4), and had a similar circular dichroism spectrum to that of wild-type HypB (Figure S5). As expected, nickel binding to NB-HypB was not observed by ESI-MS (data not shown) and a weak apparent nickel dissociation constant was observed through competition with MF2 (Figure S6 and Table S5).

We collected XAS data of Ni(II)-HypA_Str+NB-HypB(GDP) (Figure 5), and as a control, Ni(II)-NB-HypB(GDP) (Figure S7) to interrogate any changes to the nickel site of HypA. These data revealed that the Ni(II) near-edge of Ni(II)-HypA_Str+NB-HypB(GDP) was distinct from those of the separate proteins and could not be reconstructed from a combination of the two data sets (Figure S8), consistent with the analysis performed with TM-HypB(GDP) (Figure S2), and indicating again that a new site was formed. A comparison of the XAS data of Ni(II)-HypA_Str+NB-HypB(GDP) and those of Ni(II)-HypA_Str+TM-HypB(GDP) reveals that the two HypB constructs produce a very similar coordination environment (Figure 5). This result suggests that in the mixtures of the two proteins, the bulk of the nickel is bound to HypA, which is consistent with the nickel transfer model and the relative affinities of the individual proteins. Furthermore, the coordination environment of nickel bound to HypA in the context of the complexes with the HypB proteins is distinct from that of isolated HypA.

Figure 5.

XAS of Ni(II)-HypA_Str-TM-HypB(GDP) and Ni(II)-HypA_Str-NB-HypB(GDP). Comparison of (A) the near-edges, (B) the k³-weighted EXAFS, and (C) the corresponding Fourier transforms of Ni(II)-HypA_Str-NB-HypB(GDP) (orange) and Ni(II)-HypA_Str-TM-HypB(GDP) (grey) XAS data. The spectra are similar, suggesting that nickel is found in the same environment in both samples. The k³-weighted EXAFS and Fourier transforms of each nickel species (solid lines) are overlaid with the best-fit model (dashed lines) corresponding to an S(N/O)₂His coordination environment (Table 3), with slightly different bond lengths.

EXAFS curve fitting analysis of both mixtures of proteins are best fit to an S(N/O)₂His environment and suggests that the site in the presence of HypB includes a similar chelate as in the HypA_Pep site but with a heavier atom replacing the predicted coordinating water molecule (Figure 5, Table 3). The EXAFS data fit well to a model based on the DFT study that includes a sulfur atom, although this fit has multiple scattering paths that cannot be separated within the resolution of the EXAFS data. One possibility for this fourth ligand is a cysteine sulfur. It is expected that this sulfur atom does not come from HypB, given that a similar environment is also observed in the NB-HypB mixture, a HypB construct without potential G-domain coordinating cysteine residues. It is therefore likely that this atom comes from HypA or is a chloride ion from the buffer solution.

Table 3.

EXAFS fitting of Ni(II)-HypA_Str-TM-HypB(GDP) and Ni(II)-HypA_Str-NB-HypB(GDP)

Sample Description	A-Bsa	N	R (Å)	σ2 (Å2)	ΔE0 (eV)	F-factor
Ni(II)-HypAStr + NB-HypB(GDP)	Ni-Hisc	1	1.925(5)	0.0084(3)	−8.7(5)	0.400
	Ni-N(term.)c	1	1.937†	0.0084†
	Ni-N(amide)c	1	1.877†	0.0082†
	Ni-S/Clc	1	2.216(4)	0.0044†
	Ni-N/O	2	1.890(4)	0.0053(3)	−11.7(6)	0.368
	Ni-His	1	2.119(11)	0.0053†
	Ni-S/Cl	1	2.187(3)	0.0037†
	Ni-N/O	2	1.923(6)	0.0054(4)	−5.7(7)	0.413
	Ni-N/O	2	2.153(8)	0.0054†
	Ni-S/Cl	1	2.206(3)	0.0027†
Ni(II)-HypAStr + TM-HypB(GDP)	Ni-N(amide)c	1	1.926(6)	0.0088(4)	−5.4(5)	0.414
	Ni-Hisc	1	1.976†	0.0090†
	Ni-N(term.)c	1	1.987†	0.0090†
	Ni-S/Clc	1	2.234(4)	0.0053†
	Ni-N/O	2	1.910(8)	0.0067(4)	−8.3(8)	0.427
	Ni-His	1	2.049(13)	0.0055†
	Ni-S/Cl	1	2.223(5)	0.0049†
	Ni-N/O	2	1.911(10)	0.0068(3)	−6(1)	0.425
	Ni-N/O	2	2.067(14)	0.0068†
	Ni-S/Cl	1	2.239(5)	0.0049†

^aA-Bs denotes absorber and backscatterer interaction; N denotes coordination number; R is given in Å and represents interatomic distances; σ² given in Ų, are the Debeye-Waller factors (mean-square deviations in interatomic distance); the threshold energy shifts, ΔE₀ are given in eV. The values in parentheses are the estimated standard deviations obtained from the diagonal elements of the covariance matrix. The F-factor or fit-error function is defined as (Σk⁶(χ (k)_calcd - χ (k)exptl)²/ Σk⁶ χ (k)_exptl²)^1/2. The summation is over all data points included in the refinement.

^bThe best fit models are bolded.

^cModel was based on HypA-DFT structures and include a backscattering histidine and the HypA chelate. Absorbers assignments are based on bond length comparison to the HypA DFT structure.

^†Denotes that the variable value was numerically linked to the corresponding variable in the interaction listed in the table row above in order to reduce the total number of degrees of freedom in the refinement.

Subsequent analysis revealed that the near edge spectrum of Ni(II)-HypA_Str could be reconstructed from a mixture of 51% Ni(II)-HypA_Pep and 49% Ni(II)-HypA_Str+NB-HypB(GDP), consistent with the model that nickel bound to HypA_Str is a mixture of two separate binding modes (Figure 6). One state is comprised of the coordinating environment at the N-terminus and the other state is an NB-HypB-induced state. This analysis was supported by reconstructing the EXAFS region from the same proportions (Figure 6). Furthermore, we repeated the analysis of the data collected with TM-HypB instead of NB-HypB, and were able to reconstruct the HypA_Str spectrum by using similar proportions, providing further evidence that the sites induced on HypA by NB-HypB and TM-HypB are the same (Figure 6), and supporting the model that in the complex the bulk of the metal is bound to HypA. Altogether, this fitting analysis suggests that when nickel is bound to the HypA chelate, the 4^th ligand is variable and can serve as the source for the binding site’s plasticity, and that HypB can influence the preference for this ligand.

Figure 6.

XAS reconstruction of Ni(II)-HypAStr. Reconstruction (dashed) of (A) the near-edge and (B) the EXAFS of Ni(II)-HypA_Str (green) with 61% Ni(II)-HypA_Pep (red) and 39% Ni(II)-HypA_Str+NB-HypB(GDP) (orange). Reconstruction (dashed) of (C) the near-edge and (D) the EXAFS of Ni(II)-HypA_Str (green) with 56% Ni(II)-HypA_Pep (red) and 44% Ni(II)-HypA_Str+TM-HypB(GDP) (grey). Successful reconstruction suggests that nickel bound to HypA_Str is found in two different environments. One environment is defined as Ni(II)-HypA_Pep and the other is the HypB-induced Ni(II)-HypA coordination that can be represented by Ni(II)-HypA_Str+TM-HypB(GDP) or Ni(II)-HypA_Str+NB-HypB(GDP).

Discussion

Biosynthesis of [NiFe]-hydrogenase involves several maturation and accessory proteins including HypA and HypB, which are two metallochaperones required for nickel insertion.¹⁹ These proteins help control the distribution of nickel, a necessary but potentially toxic transition metal, and ensure that the nickel ions are funnelled to the hydrogenase precursor enzyme in the penultimate maturation step.¹⁹ Multiple studies have characterized the biochemical and structural features of the individual proteins.^{22, 23, 43, 56, 57, 80–82} Recent work demonstrated that nickel moves from the G-domain site of HypB to HypA in a unidirectional, nickel-selective process within a complex, and that the interaction between the proteins is regulated by the GTPase cycle of HypB.²³ The results of this work fill in that model by demonstrating that the nucleotide-loaded state of HypB does not impact nickel coordination. Instead, complex formation with HypB alters the nickel-binding site on HypA. Furthermore, nickel bound to HypA can adopt at least two coordination environments, one of which is favoured in the context of the complex with its partner protein, suggesting a mechanism for the acquisition and release of nickel by HypA.

Nickel coordination to HypB

Two crystal structures of HypB homologs, Methanocaldococcus jannaschii (Mj) HypB and HpHypB,^{61, 62} loaded in the G-domain site with zinc or nickel, respectively, revealed distinctive coordination environments and different combinations of ligands. In HpHypB a single nickel ion is coordinated in a square planar, tetrathiolate site,⁶² recruiting two cysteines from each HypB in the homodimer. In MjHypB two zinc atoms filled the same space with the additional recruitment of a histidine ligand.⁶¹ Although previous studies confirmed that nucleotide-loaded EcHypB binds a single nickel ion per protein dimer, similar to HpHypB,²³ the work presented here suggests that the coordination site is different from that of the HpHypB crystal structure, because the EXAFS data contain characteristic backscattering from histidine coordination and were best fit to a model site that only included up to two sulfur ligands. This proposed coordination environment of the E. coli protein is consistent with the observation that the H167A mutation weakens nickel binding at the G-domain site of HypB.⁵⁶ In addition, the nickel affinity of HypB is mildly weaker when GDP is loaded on HypB versus GppCp, a non-hydrolysable analog of GTP.²³ This observation leads to speculation that nucleotide was allosterically controlling the nickel site, but the XAS data presented here suggest that the nickel coordination environments Ni(II)-HypB(GDP) and Ni(II)-HypB(GppCp) are similar, so it is unlikely this is a major factor in driving nickel transfer to HypA. This work supports an insertion model where GTP hydrolysis controls the interaction between HypB and HypA and through that interaction regulates nickel delivery.²³

Nickel coordination to HypA

The proposed nickel sites of various HypA homologs and constructs are not the same, yet all these variants contain the conserved N-terminal MHE motif.^{44, 51, 53, 83} Our initial attempts to fit the nickel EXAFS data of EcHypA to a single site were unsuccessful and resulted in poor fits (data not shown). It was only upon analysis of Ni(II)-HypA_Pep and Ni(II)-HypA_Str+TM-HypB(GDP) did it become apparent that nickel bound to EcHypA exists in a mixed state. This plasticity may explain some of the discrepancies amongst previous reports about HypA homologs, but not all of them. For example, the HpHypA site is predicted to be an octahedral paramagnetic site,⁵³ which is not consistent with either the XAS or the magnetic moment of EcHypA, both of which support a diamagnetic, 4- or 5-coordinate environment. Notably, our computational optimization results revealed that if the amide nitrogen of Glu3 was deprotonated then His2 was sterically forced into an axial position, suggesting a potential mechanism for the HpHypA variant to form an octahedral site with the recruitment of additional ligands. It is possible that the discrepancies between these variants are related to the additional role that HpHypA must fulfill as a maturation factor for a second nickel enzyme in H. pylori, urease.⁸⁴ In support of this model, the HpHypA nickel site was recently found to adopt a tighter nickel site in the presence of the urease accessory protein UreE.⁸⁵

Ni(II)-HypA can be described as a mixture composed of nickel bound to a HypB-induced coordination and one that is captured in the isolated HypA_Pep. Such an adaptable binding site might have physiological relevance, given that a metallochaperone that acts as an intracellular trafficking factor will have to adopt several metal-binding modes to accomplish the acquisition, transport, and release of metal. The model HypA nickel site presented here has three coordinating atoms from the terminal nitrogen, the backbone amide nitrogen of His2, and the δ-nitrogen of the imidazole of His2, with a fourth coordination site that is not a part of the local chelate and is variable between the different HypA binding modes. This construct could allow for rapid nickel binding and/or release, appropriate for a relay stop along the nickel transfer pathway and is similar to other nickel binding sites, determined by crystallography, on proteins that have nickel trafficking functions.^44,79 We further speculate that the distorted square planar nickel site is conducive to an associative ligand exchange mechanism because of possible recruitment of additional ligands to the axial sites without spontaneous ligand dissociation, a mechanism that is reminiscent of Atx1 and Ccc2 copper transfer.^{14, 15} Future work will investigate whether such a mechanism is in play with HypA during nickel delivery to the final destination in the hydrogenase enzymes.

The HypB-induced HypA nickel site may recruit a fourth ligand from a different region of the HypA protein or HypB may modulate solvent exposure, allowing a chlorine ion from the buffer to coordinate. Our search for additional coordinating residues did not provide any evidence for such a ligand. Mutation of Cys38 or Glu40, identified as possible coordinating residues based on analogy with the NMR structure of HpHypA,⁵¹ did not impact the role of HypA in the maturation of active hydrogenase enzyme in E. coli, in contrast to the lack of activity observed with the His2 mutant, suggesting that these two residues do not coordinate nickel. The EcHypA nickel-binding modes presented here are more consistent with the nickel site of TkHypA, an archaeal variant.⁴⁴ This variant bears a square planar site very similar to that of the HypA_Pep chelate, with the inclusion of His98 as the fourth ligand. His98 is not conserved in the E. coli protein, so it does not provide a clue to additional ligands. However, that particular binding site of TkHypA is observed when it is in a complex with TkHypB. A different conformation is observed when TkHypA forms a complex with TkHyhL, a large hydrogenase subunit, and in that case the His98 residue of TkHypA is far from the N-terminus of TkHypA, indicating that the nickel site is adaptable,⁴⁷ as we observe with EcHypA. In addition to the work on EcHypA and TkHypA, this type of metal-binding site at the N-terminus is reminiscent of that of another nickel-binding protein, Thermoanaerobacter tengcongensis NikM,⁷⁹ and may represent a common N-terminal nickel-binding motif.

Characterization of HypA_Pep demonstrated the role of the N-terminus of HypA in nickel binding, but the role of Glu3 in the conserved MHE motif remains elusive. Our geometry optimization calculations indicated that Glu3 might either directly coordinate to nickel in the HypA_Pep or help to stabilize a coordinating water molecule through hydrogen bonding. However, the fact that both the E3D and E3A HypA mutants result in similar levels of hydrogenase production in E. coli as the wild-type protein indicates that this residue is dispensable for hydrogenase maturation. These data are inconsistent with a previous analysis of EcHybF that reported that E3Q-HypA was functional for hydrogenase maturation but E3L-HypA was not.⁴⁸ It is possible that Glu3 contributes to the ability of HypA and HybF to distinguish between the different hydrogenase isoforms expressed in E. coli, given that the whole-cell assay does not report on the activity of all of the enzymes in this organism (manuscript in revision), but it is clear that this residue is not essential for nickel delivery to the hydrogenase precursors by HypA.

Conclusions

A key feature of the HypAB maturation system appears to be the rapid and selective transfer of nickel from the G-domain site of HypB to HypA. A working model of this maturation process (Figure 7), illustrates how nickel is shuttled between the two proteins. The protein interactions in this model are based on a combination of in vitro and in vivo studies.^{23, 46} We have shown that this transfer mechanism does not depend on a nucleotide-induced change in nickel coordination of HypB because of an absence of such a change in the XAS collected when HypB was loaded with GDP or GppCp. Thus, it is likely that the main contribution to the GTPase cycle during this stage of the nickel delivery process is to regulate the interaction between the two accessory proteins. Furthermore, by using the mutant H2Q-HypA, a variant that still interacts with HypB but does not accept nickel, we have shown that HypA does not induce a coordination change around the nickel ion bound to HypB. Instead, experiments with several versions of HypB demonstrated that it influences the nickel-binding site of HypA. These observations, coupled with rapid and selective nickel transfer to HypA,⁴³ suggests that nickel is passed between the sites within a protein-protected process. Furthermore, reconstruction of the nickel XAS data of Ni(II)-HypA_Str with those of Ni(II)-HypA_Pep and either Ni(II)-HypA_Str-NB-HypB(GDP) or Ni(II)-HypA_Str-TM-HypB(GDP) indicated that we are not observing a ternary complex with ligands from both proteins, but instead a complex with nickel ligands donated from HypA and possibly solvent.

Figure 7.

HypA-HypB Ni(II) transfer model. HypB binds GTP and dimerizes. In this state, the G-domain of HypB acquires nickel and GTP hydrolysis occurs, perhaps accelerated by an unidentified activation factor. In the GDP-loaded state, the nickel affinity is weakened but the nickel coordination environment remains similar. GTP hydrolysis promotes the association of HypA with HypB, a HypA conformational change to a “HypB induced” HypA nickel site, and rapid nickel transfer (dashed arrow). The HypB dimer is weakened, because nickel is no longer bridging the two HypB monomers, resulting in fragmentation of the HypA-HypB complex, freeing Ni(II)-HypA for nickel loading into the partially assembled hydrogenase precursor and apo-HypB for continuation within its GTPase cycle. Once nickel is loaded in the hydrogenase, the large hydrogenase subunit undergoes C-terminal cleavage and association with the small hydrogenase subunit to form the mature enzyme. Note: the quaternary structure of HypB during nickel transfer to HypA has not yet been conclusively established.

Finally, we have shown that nickel binds to HypA in two different modes, one of which is induced by HypB. This bimodal coordination may facilitate the acquisition and release of nickel. Two coordination modes on the one metallochaperone support the directed nickel transfer model where nickel passes from HypB to HypA and then from HypA to the hydrogenase enzyme precursor protein. In a catch and release scheme, HypA is optimized to fish nickel from HypB and deliver nickel to the hydrogenase precursor. The next step, and subject of future work, is to probe the nickel transfer from HypA to the hydrogenase precursor protein.

Synopsis.

[NiFe]-hydrogenase maturation requires two nickel metallochaperones HypA and HypB. In this work we show that the nickel site of HypA can adopt two binding modes: a HypB induced mode and a binding mode characterized by the N-terminal binding residues. This work provides a mechanism for HypB controlled nickel acquisition and release from HypA within a protected protein complex.

Abbreviations

ATCUN

amino terminal Cu(II)- and Ni(II)-binding motif

DPABF

E. coli MC4100 with ATG codon of hypA mutated into TAA and ΔhybF

Ec-

E. coli homolog

ECF-type

energy coupling factor-type transporter

EXAFS

extended X-ray absorption fine structure

G3E

family of GTPases containing the Walker B (G3) motif sequence ExxG

GDP

guanosine diphosphate

GppCp

non-hydrolyzable GTP analog β,γ-methyleneguanosine 5′-triphosphate

GTP

guanosine triphosphate

H2Q-HypA_Str

mutant HypA_Str [H2Q] with deficient nickel binding

Hp-

Helicobacter pylori homolog

HypAPep

peptide maquette of HypA N-terminus with the sequence MHEITW

HypA_Str

wild-type E. coli HypA with C-terminal Strep-tag

HypB(GNP)

HypB loaded with a guanosine nucleotide

MF2

mag-fura 2

NB-HypB

No-Binding HypB [C2A, C5A, C7A, C166A, C198T] lacking both nickel-binding sites

TM-HypB

Triple-Mutant HypB [C2A, C5A, C7A] lacking the N-terminal high-affinity nickel site

Tk-

Thermococcus kodakarensis homolog

XAS

X-ray absorption spectroscopy