Metallo-GTPase HypB from Helicobacter pylori and Its Interaction with Nickel Chaperone Protein HypA*

Background: HypA and HypB are metallochaperones for the activities of [NiFe]-hydrogenase and urease in Helicobacter pylori. Results: Key residues are identified for the GTP-dependent dimerization of HypB. The HypA-HypB interfaces are also identified. Conclusion: Self-dimerization is critical for the regulation of GTPase activity. HypA-HypB interaction facilitates further downstream Ni2+ delivery. Significance: The study is important to understanding [NiFe]-hydrogenase and urease maturation. The maturation of [NiFe]-hydrogenase is highly dependent on a battery of chaperone proteins. Among these, HypA and HypB were proposed to exert nickel delivery functions in the metallocenter assembly process, although the detailed mechanism remains unclear. Herein, we have overexpressed and purified wild-type HypB as well as two mutants, K168A and M186L/F190V, from Helicobacter pylori. We demonstrated that all proteins bind Ni2+ at a stoichiometry of one Ni2+ per monomer of the proteins with dissociation constants at micromolar levels. Ni2+ elevated GTPase activity of WT HypB, which is attributable to a lower affinity of the protein toward GDP as well as Ni2+-induced dimerization. The disruption of GTP-dependent dimerization has led to GTPase activities of both mutants in apo-forms almost completely abolished, compared with the wild-type protein. The GTPase activity is partially restored for HypB(M186L/F190V) mutant but not for HypB(K168A) mutant upon Ni2+ binding. HypB forms a complex with its partner protein HypA with a low affinity (Kd of 52.2 ± 8.8 μM). Such interactions were also observed in vivo both in the absence and presence of nickel using a GFP-fragment reassembly technique. The putative protein-protein interfaces on H. pylori HypA and HypB proteins were identified by NMR chemical shift perturbation and mutagenesis studies, respectively. Intriguingly, the unique N terminus of H. pylori HypB was identified to participate in the interaction with H. pylori HypA. These structural and functional studies provide insight into the molecular mechanism of Ni2+ delivery during maturation of [NiFe]-hydrogenase.

Helicobacter pylori is currently the only discovered bacterial species that can survive under the highly acidic conditions of the human stomach. It is reported that more than 50% of the world population harbor this pathogen in their upper gastroin-testinal tract (1). This pathogenic bacterium was found to be associated with a series of human diseases, such as gastritis, peptic ulcer, and even stomach cancer (2,3). The colonization and survival of H. pylori in human stomach mucosa are heavily dependent on the production of two nickel-containing enzymes, [NiFe]-hydrogenase and urease. The former can be used to provide energy by oxidizing molecular hydrogen (H 2 ) produced by intestinal bacteria (4), whereas the latter hydrolyzes urea into carbamate and ammonia, maintaining the neutral pH of bacterial cytoplasm (5).
Similar to other metalloenzymes, the maturation of [NiFe]hydrogenase and urease are mediated by a battery of chaperone proteins (6 -11). Particularly, the assembly of the Ni 2ϩ -containing metallocenter requires the cooperation of a series of metallochaperones to ensure proper metal incorporation (12). Nickel ions, when present in excess, are lethal; thus, the level of the metal ions must be tightly regulated in H. pylori (13)(14)(15)(16)(17). The Ni 2ϩ incorporation process mediated by metallochaperones is the pivotal step for maturation of the enzyme.
Previous studies have demonstrated that two metallochaperones, HypA and HypB, are critical for the maturation of [NiFe]hydrogenase. The enzyme activity of [NiFe]-hydrogenase is severely impaired in hypA or hypB gene knock-out bacterial strains but can be partially restored by the addition of high concentration of nickel ions in the culture medium (18). These studies indicated that HypA and HypB were probably responsible for the Ni 2ϩ delivery in the [NiFe]-hydrogenase metallocenter assembly process. Intriguingly, a recent study demonstrated that disruption of hypA and hypB genes in H. pylori resulted in reduction of the activity of both urease and [NiFe]hydrogenase, implying a possible "cross-talk" between the maturation pathways of urease and [NiFe]-hydrogenase in H. pylori (19).
The solution structure of HypA from H. pylori was reported recently and exhibits a unique elongated shape with two distinct metal-binding sites. The functional Ni 2ϩ site is located at the N terminus with a typical square planar geometry facilitating further Ni 2ϩ transfer. Zn 2ϩ coordinates to four conserved cysteines tetrahedrally in the zinc site and plays a pivotal role to stabilize the entire protein structure (20). H. pylori HypB was identified as a small GTPase with a highly conserved metalbinding site at its G-domain (21). Although both Ni 2ϩ and Zn 2ϩ could coordinate to this site, only Ni 2ϩ and not Zn 2ϩ induces the dimerization of H. pylori HypB, indicative of the potentially physiological role of Ni 2ϩ (22). It has been demonstrated that the H. pylori HypA-HypB heterodimeric complex could be detected by chemical cross-linking in vitro (21). However, there is still a lack of comprehensive study of H. pylori HypB protein and its interaction with H. pylori HypA.
In the present study, HypB from the H. pylori 26695 bacterial strain was expressed and characterized. A highly conserved Lys residue (Lys-168) and a hydrophobic residue pair (Met-186 and Phe-190) ( Fig. 1) were identified to be critical for the GTPase activity of H. pylori HypB. Ni 2ϩ binds to the conserved site of the G-domain of H. pylori HypB proteins (both WT and mutants), functioning as distinctive regulators to modulate the GTP-dependent dimerization of H. pylori HypB. Importantly, the protein-protein interfaces on H. pylori HypA and HypB were identified for the first time by combined use of biochemical and biophysical techniques. A possible effect exerted by HypA and HypB is proposed for Ni 2ϩ delivery in the maturation of [NiFe]-hydrogenase.

EXPERIMENTAL PROCEDURES
Materials-Restriction endonucleases and T4 polynucleotide kinase were purchased from New England Biolabs. Phusion high fidelity DNA polymerase was obtained from Finnzymes. Pfu DNA polymerase was obtained from Promega. Primers for PCRs synthesized by Invitrogen are listed in supplemental Table S1. Chromatography columns and the FPLC system were from GE Healthcare. All chemical reagents were purchased from Sigma, unless otherwise specified. All solutions were prepared with Milli-Q water (18.2 megaohms).
HypA and HypB Expression Plasmids and Mutants-The generation of pGBHIS-hypA and pET28a-hypB expression plasmids was described previously (20). The hypB gene was amplified by PCR from genomic DNA of H. pylori 26695. The amplified hypB gene fragment contains NdeI and EcoRI restriction sites at the 5Ј-end and 3Ј-end, respectively. The amplified DNA fragment and expression plasmid pET28a were digested with the corresponding restriction enzymes and ligated together with T4 ligase, generating the expression plasmid pET28a-hypB. The construct for expression of HypB⌬24N (residues 25-242) was amplified by PCR using pET28a-hypB vector (full-length) as a template and cloned into pET28a vector. Constructs of HypB(M186L/F190V), HypB(K168A), HypB(K18A), HypB(V20A), HypB(K21A), and HypB(I22A) were generated using Phusion high fidelity DNA polymerase according to the provided protocol. The linear constructs were phosphorylated using T4 polynucleotide kinase and digested with DpnI before ligation. The reaction mixture was subsequently transformed into XL-1 Blue Escherichia coli. All of the constructed plasmids were sequenced (Invitrogen) to verify the cloned sequences.
Protein Expression and Purification-The expression and purification of HypA and 15 N-HypA were carried out as described previously (20). HypB was expressed in the E. coli BL21(DE3) strain. When OD at 600 nm reached 0.6 in LB medium, 0.2 mM IPTG 2 was added to induce protein expression, and cells were harvested by centrifugation after 16 h of incubation at 25°C. The cells were lysed by sonication in buffer A (20 mM Tris-HCl containing 500 mM NaCl, pH 7.4) in the presence of 0.1 mM PMSF. Cell supernatant was loaded into a HisTrap column (GE Healthcare) pre-equilibrated with buffer A supplemented with 50 mM imidazole. Bound proteins were eluted using buffer A containing 300 mM imidazole. His-tagged HypB elution fraction was dialyzed against protease buffer (20 mM Tris-HCl, pH 7.4, 100 mM NaCl, 10 mM MgCl 2 , 1 mM DTT) overnight, followed by the addition of 100 NIH units of throm- bin (Sigma) and 30 units of alkaline phosphatase (calf Intestinal; New England Biolabs) to remove the His tag and bound nucleotides. After incubation at room temperature for 3 h, EDTA was added into the cleaved protein to a final concentration of 20 mM and further incubated at 4°C overnight. The protein was subsequently pooled and applied to a Superdex-75 size exclusion column equilibrated in buffer A in the presence of 1 mM DTT. The yield of purified protein is around 40 mg/liter of LB medium. The metal contents of the purified HypB were determined by inductively coupled plasma MS. DTT was then removed by a HiTrap desalting column (GE Healthcare), and the protein was frozen at Ϫ80°C in 0.5-ml aliquots for further use. All HypB mutants were expressed and purified similarly. The concentrations of all protein samples were determined by a BCA assay.
Electronic Absorption Spectroscopy-All electronic absorption spectra were collected on a Cary 50 UV-visible spectrophotometer with 1-cm quartz cuvette at ambient temperature. Spectra were recorded from 600 to 240 nm with a scan rate of 360 nm/min. For Ni 2ϩ titration experiments, 50 M apo-HypB solutions or its mutants HypB(K168A) and HypB(M186L/ F190V) were prepared in a titration buffer (20 mM Tris-HCl, pH 7.4, containing 100 mM NaCl, 200 M TCEP). Aliquots of Ni 2ϩ stock solution (10 mM in the titration buffer) were added stepwise into protein solutions. Similarly, GTP was titrated into Ni 2ϩ -bound HypB or its mutants except that the titration buffer was supplemented with 1 mM MgSO 4 , which is required for GTP binding. The UV titration curves were fitted to the Ryan-Weber nonlinear equation (23), where I represents UV absorbance intensity; I max is the maximal UV absorbance when all of the ligands are bound; C p and C m are the total concentrations of proteins and ligands, respectively; and K d is the dissociation constant. Size Exclusion Chromatography-Gel filtration analysis of oligomeric states of the proteins was performed on an Ä KTA FPLC system using a Tricorn 30/100 Superdex 75 column at 4°C eluted with a flow rate of 0.5 ml/min. The column was calibrated with a low molecular weight gel filtration calibration kit (GE Healthcare). For Ni 2ϩ -dependent dimerization experiments, a 200 M concentration of either the apo-forms of HypB proteins or their Ni 2ϩ -bound forms, which were prepared by preincubation of 2 mol eq of Ni 2ϩ prior to analysis, were freshly prepared and subjected to chromatography. The proteins were eluted with a gel filtration buffer (20 mM HEPES buffer, 100 mM NaCl, 50 M TCEP, pH 7.0) supplemented with 20 M Ni 2ϩ . GTP-dependent dimerization experiments were carried out similarly except that 5 mM MgSO 4 and 500 M GTP instead of Ni 2ϩ were included in the gel filtration buffer.
Isothermal Titration Calorimetry (ITC)-Metal-bound HypB was prepared as described in the supplemental material. Freshly purified apo-, Zn 2ϩ -, and Ni 2ϩ -HypB were prepared in ITC buffer (20 mM HEPES buffer, 100 mM NaCl, 1 mM MgSO 4 , pH 7.0). GTP (or GTP␥S) and GDP with concentrations of 600 M were used as stock solutions and titrated into 50 M protein samples. Titration of ligands into the buffer was recorded as background. To determine the binding affinity of HypA and HypB, about 1.1 mM HypA in 20 mM HEPES containing 100 mM NaCl, 2 mM TCEP, pH 7.0, was titrated into 0.1 mM apo-HypB in the same buffer. All experiments were carried out on an ITC200 isothermal titration calorimeter (MicroCal) at 25°C. Data were fitted by the Origin 7 software package.
GTPase Activity Assay-The GTPase activity assay for WT HypB and its mutants was performed as described previously (22). The reaction mixture consisting of 200 M GTP, 1 mM MgSO 4 , ϳ10 M proteins in a reaction buffer (20 mM HEPES, 100 mM NaCl, and 1% glycerol, pH 7.0) was incubated at 37°C. Aliquots of the mixture were taken out in a time course for determination of released free phosphate by the Malachite Green phosphate assay kit (Cayman).
GFP-Fragment Reassembly-GFP-fragment reassembly was used to verify the interaction between HypA and HypB inside E. coli cytoplasm (24). Details of plasmid construction are described in the supplemental material, and the primers used are listed in supplemental Table S2.
The constructed plasmids pET32a-NGFP, pBAD33-CGFP, pET32a-NGFP-linker-HypA, and pBAD33-HypB-linker-CGFP were subsequently co-transformed into the BL21(DE3) strain with different combinations. The transformed bacteria were cultured in LB medium at 37°C until A 600 reached 0.6, and 20 M IPTG and 0.2% arabinose were added to induce the protein expression. The bacteria were further cultured overnight at 20°C. The bacteria were then diluted with PBS buffer and examined by an Axiovert 200 M fluorescent microscope (Carl Zeiss) using a 100ϫ oil immersion objective lens. The effects of Ni 2ϩ on intracellular fluorescence were investigated similarly except that the co-transformed BL21(DE3) was cultured in M9 minimal medium with or without a supplement of 10 mM Ni 2ϩ . About 100 M IPTG and 0.015% arabinose were used to induce protein expression, and intracellular fluorescence was examined after culture for 48 h at 25°C.
NMR Spectroscopy-The binding interface on HypA was mapped based on NMR titration experiments. All NMR titration experiments were performed at 25°C on a Bruker Avance 600 spectrometer operating at a 1 H frequency of 600.13 MHz, equipped with a TCI CryoProbe TM . Unlabeled monomeric apo-HypB (ϳ1 mM) was gradually added into 0.3 mM 15 N-labeled HypA in 20 mM Tris-HCl buffer, pH 7.4, containing 100 mM NaCl. Two-dimensional 1 H-15 N HSQC spectra were recorded before and after each addition of HypB. The chemical shift perturbation was quantified in a manner similar to that described previously (25). The intensity of each resonance was plotted against the molar percentage of added HypB, and the resulting slope was used to estimate the extent of line broadening. The chemical shift perturbations (⌬) for each residue were quantified based on ⌬ ϭ ((⌬␦ 2 NH ϩ ⌬␦ 2 N /25)/2) 1 ⁄2 . The dissociation constant (K d ) of HypA-HypB complex was obtained from least squares fitting of the chemical shift changes of residue His-17 of 15 N-HypA (the largest perturbed residue) as a function of total HypB concentration according to Equation 2 (26), where ⌬␦ i represents the observed change of chemical shift, [B] is the HypB concentration at each titration point, [A t ] is the total HypA concentration, and K d is the dissociation constant.
Chemical Cross-linking-Purified HypA (40 M), HypB, or its mutant proteins (20 M) were preincubated in 20 mM HEPES buffer containing 100 mM NaCl at pH 7.4. Cross-linking chemical reagent bis(sulfosuccinimidyl) suberate (BS 3 ) was added into the protein solution to a final concentration of 0.2 mM. The reaction mixtures were further incubated at room temperature for 1 h. The cross-linking reaction was terminated by the addition of 1 M Tris solution to a final concentration of 50 mM. Reaction samples were then subjected to 13.5% SDS-PAGE and stained with Coomassie Blue.

RESULTS
Protein Analysis-All of the purified HypB proteins have more than 95% purity as demonstrated on the SDS-PAGE (Fig. 2). The metal contents of the purified HypB proteins were determined by inductively coupled plasma MS, and the molar ratios of metals to the protein are less than 0.01:1 for Ni 2ϩ , Zn 2ϩ , and Mg 2ϩ , indicating that HypB was purified as an apo-form.
Interaction of Ni 2ϩ /GTP with HypB-The interactions between Ni 2ϩ and apo-forms of WT HypB or two mutants were monitored by UV-visible spectroscopy. Upon the addition of Ni 2ϩ into apoprotein solutions, two new absorption bands appeared at ϳ280 and 340 nm, which represent typical ligandto-metal charge transfer (LMCT) between Cys-S ␥ and Ni 2ϩ (27). The intensities of these bands increased with the further addition of Ni 2ϩ and leveled off at a molar ratio of Ni 2ϩ to the protein of 1.0, indicative of binding of Ni 2ϩ to HypB with a stoichiometry of 1.0 (i.e. one Ni 2ϩ per HypB monomer) ( Fig. 3A and supplemental Fig. S1).
GTP binding and hydrolysis were proposed to be critical in the Ni 2ϩ delivery process (28). Titration of GTP to the solutions of Ni 2ϩ -HypB (WT or mutants) resulted in a strong absorption band at ϳ260 nm, which originated from the self-absorption of GTP molecules. Intriguingly, two relatively weak absorption bands were also observed in the difference UV spectra at around ϳ310 and 370 nm, close to the Ni 2ϩ LMCT band (ϳ340 nm) ( Fig. 3B and supplemental Fig. S1), whereas no such absorption peaks were observed when GTP was titrated into Ni 2ϩ solution (supplemental Fig. S3). The absorbance increased with the addition of GTP and saturated at a molar ratio of GTP to Ni 2ϩ -HypB of 1.0, indicative of the binding of one GTP molecule per HypB monomer (29). The perturbation of the Ni 2ϩ LMCT band upon nucleotide binding was also observed previously in a Ni 2ϩ -binding ATPase (27). Such phenomena are probably due to conformational changes at the Ni 2ϩ -binding site induced by binding of GTP or ATP to proteins.
The Ryan-Weber nonlinear equation was applied to fit both the Ni 2ϩ and GTP titration curves. The dissociation constants (K d ) were calculated to be 1.72 Ϯ 0.38 and 0.57 Ϯ 0.33 M for the binding of Ni 2ϩ to HypB and GTP to Ni 2ϩ -HypB, respectively (Fig. 3).

Conserved Lys Residue and Hydrophobic Residue Pair Are
Important for GTP-dependent Dimerization and GTPase Activity of HypB-Nucleotide-dependent dimerization was proposed to be a common feature to activate G proteins (30). Previously, it has also been demonstrated that GTP binding induced the dimerization of HypB proteins (22,29).
Analysis of the structure of HypB from Methanocaldococcus jannaschii (Protein Data Bank entry 2HF9) revealed two unique structural elements that might contribute to the GTP-dependent dimerization. A highly conserved Lys residue, Lys-153 (corresponding to Lys-168 in H. pylori), formed a hydrogen bond with bound GTP, bridging two HypB monomers together. In addition, a hydrophobic residue pair (i.e. Leu-171 and Val-175, corresponding to Met-186 and Phe-190 in H. pylori), located at an ␣-helical region, may play an important role in stabilizing the dimeric structures by hydrophobic interactions between Val and Leu from each monomer (supplemental Fig.  S3). To investigate whether the corresponding structural elements also contribute to the GTP-dependent dimerization of H. pylori HypB, two HypB mutants (i.e. K168A and M186L/ F190V) were constructed, and their GTP-dependent dimerization properties and GTPase activities were examined.
In the absence of GTP, WT HypB, HypB(K168A), and HypB(M186L/F190V) mutants were eluted as a single peak corresponding to the monomeric form of the protein (Fig. 4A). When GTP was included in the gel filtration buffer, ϳ50% of WT HypB was eluted at a molecular mass of 52 kDa corresponding to dimeric form. In contrast, HypB(K168A) was eluted as a monomeric form. The majority (ϳ80%) of HypB(M186L/F190V) was eluted as a monomer (10.4 ml) with a small portion eluted at an elution volume of 9.0 ml, corresponding to a dimeric form (Fig. 4B). These results indicated that mutations at the conserved Lys residues and the two hydrophobic residues completely or partially abolished GTPdependent dimerization of H. pylori HypB.
The GTPase activities of two mutants were determined by the Malachite Green method and compared with the WT HypB. The rates of GTP hydrolysis of all protein samples were linear over 80 min (supplemental Fig. S4). In contrast to the WT HypB, which has a low GTP hydrolysis rate of 0.30 Ϯ 0.03 nmol⅐min Ϫ1 (mg of protein) Ϫ1 , both apo-HypB(K168A) and apo-HypB(M186L/F190V) exhibited no detectable GTPase activities (Fig. 5A). Upon Ni 2ϩ binding, the GTPase activity increased dramatically for the wild-type protein (1.37 Ϯ 0.03 nmol min Ϫ1 (mg of protein) Ϫ1 ) but partially restored for HypB(M186L/F190V) (0.65 Ϯ 0.03 nmol min Ϫ1 (mg of protein) Ϫ1 ). In contrast, no detectable GTPase activity was found for HypB(K168A) even in the Ni 2ϩ -bound form. Similar to the WT HypB for which Ni 2ϩ binding induces dimerization of the protein (22), both HypB(K168A) and HypB(M186L/F190V) mutants could also form dimer upon Ni 2ϩ binding (Fig. 5B).

Metal Ions Serve as Regulators for Nucleotide Binding to
HypB-Previous studies have demonstrated that Ni 2ϩ instead of Zn 2ϩ could promote the dimerization of H. pylori HypB; binding of Ni 2ϩ enhanced GTPase activity, whereas Zn 2ϩ inhibited it (22).
It is known that GTPase activity is also controlled by the affinity to guanine nucleotide and the nucleotide exchange rate (31). The nucleotide-binding properties of H. pylori HypB in its apo-and metal-bound forms were investigated by ITC. The dissociation constants of GTP␥S, a non-hydrolyzable GTP analog, to H. pylori HypB were determined to be 1.19 Ϯ 0.07 M for apo-HypB, 2.19 Ϯ 0.36 M for Zn 2ϩ -HypB, and 3.15 Ϯ 0.55 M for Ni 2ϩ -HypB (Fig. 6A). Binding of GTP to H. pylori HypB yielded similar results to GTP␥S (supplemental Fig. S5), implying that metal binding did not significantly affect the binding affinities of GTP to H. pylori HypB. Unexpectedly, the binding of GDP to H. pylori HypB in the presence of Zn 2ϩ and Ni 2ϩ revealed different properties. GDP binds apo-and Zn 2ϩ -HypB at least 5 times more tightly than Ni 2ϩ -HypB, with dissociation constants of 0.97 Ϯ 0.17, 0.65 Ϯ 0.07, and 5.62 Ϯ 0.64 M, respectively (Fig. 6B), suggesting that Ni 2ϩ -triggered dimerization affected the GDP binding to the protein.
HypB Binding Interface Mapping on HypA by NMR Chemical Shift Perturbation-Previous cross-linking experiments demonstrated that monomeric HypA and HypB formed a heterodimeric complex (21). To experimentally map the binding interfaces on HypA, NMR spectroscopy was used to examine the residues that experience chemical shift perturbations upon HypB binding. 15 N-HypA was titrated with unlabeled monomeric HypB, and a series of two-dimensional 1 H-15 N HSQC spectra were recorded (Fig. 7A). Certain resonances in the spectra underwent either significant line broadening or chemical shift changes upon the addition of HypB, indicative of an inter-   (Fig. 7D). These residues were mainly located at the cleft between helix 1 and strands 1 and 6 (␤1/␤6).

Interaction between Metallochaperones HypA and HypB
The dissociation constant for HypA and HypB was determined to be 52.2 Ϯ 8.8 M by fitting the changes of chemical shifts (e.g. His-17) as a function of HypB concentration (Fig. 7E), indicative of a relatively weak interaction. The binding constant was also measured by ITC, a technique commonly used for quantification of macromolecule-ligand interactions (33). The dissociation constant (K d of 57.9 Ϯ 8.5 M) is comparable with the affinity by NMR (supplemental Fig. S6).
Identification of HypB Binding Site to HypA-The interaction between HypA and HypB is believed to be important in Ni 2ϩ delivery (21). Unexpectedly, no interaction between HypA and HypB could be detected by GST pull-down experiments (data not shown). The chemical cross-linking method was therefore applied to detect the protein-protein interaction. To investigate whether the N-terminal region of H. pylori HypB is involved in the interaction as previously proposed based on the crystal structure of HypB from M. jannaschii (29), a mutant with N-terminal 24 residues truncated (HypB⌬24N) was expressed and purified. The interactions between HypA and HypB (or HypB⌬24N) were examined by the chemical cross-linking method (Fig. 8A). Incubation of BS 3 , a water-soluble chemical cross-linking agent, with HypA, HypB, and HypB⌬24N resulted in weak bands corresponding to the molecular weight of a homodimer for each protein (ϳ25 kDa for HypA, ϳ50 kDa for  HypB and HypB⌬24N). Incubation of BS 3 with a mixture of HypA and HypB resulted in a strong band with molecular mass around 45 kDa. This band was subjected to MALDI-TOF mass spectrum analysis, which revealed the presence of both HypA and HypB proteins (supplemental Table S3), indicative of formation of a heterodimer between HypA and HypB. In contrast, no such a band was observed between HypB⌬24N and HypA, indicating that the N-terminal 24 residues are required for the formation of a HypA-HypB complex.
Sequence alignments (Fig. 1) of the N-terminal region of HypB from different species revealed three relatively conserved residues (i.e. Lys-18, Val-20, and Ile-22) in the H. pylori HypB sequence. To validate whether these residues involved in HypA-HypB interaction, single-residue mutants of these three conserved residues (K18A, V20A, I22A) and an unconserved residue (K21A) were expressed and purified. The interactions between these HypB mutants and HypA were examined similarly. As shown in Fig. 7B, incubation of HypB(K18A), HypB(V20A), and HypB(I22A) mutants with HypA yielded much weaker heterodimer complex bands compared with the WT HypB, indicative of the involvement of these three conserved residues in HypA-HypB interaction. In contrast, the HypB(K21A) mutant exhibited a strong heterodimer complex band with HypA, which is comparable with that of WT HypB.

Intracellular Interaction of HypA and HypB-Although in
vitro experiments showed the formation of HypA-HypB complex (21), there was a lack of in vivo evidence. GFP-fragment reassembly, a powerful tool to visualize the intracellular interaction between proteins (34), was therefore used. The constructed plasmids were co-transformed into BL21(DE3) bacteria strain with different combinations. The fused proteins were co-expressed, and the phase-contrast fluorescence images of the bacteria are shown in Fig. 9. Only the bacteria containing the pET32a-NGFP-linker-HypA and pBAD33-HypB-linker-CGFP plasmids (NA ϩ BC) exhibited the strongly emitted green fluorescence. In contrast, no fluorescence was detected for the other three control combinations (i.e. E. coli cells that co-expressed CGFP and NGFP-HypA, NGFP and HypB-CGFP, and NGFP and CGFP) (Fig. 9, b-d). The results clearly demonstrated that the interaction between HypA and HypB occurred even in a much more competitive cellular cytoplasm condition. To further investigate the effects of Ni 2ϩ on the intracellular fluorescence, the co-transformed BL21(DE3) bacteria were cultured in M9 minimal medium in the presence or absence of 10 mM Ni 2ϩ . The plasmid combination NA ϩ BC exhibited intracellular fluorescence under both conditions (supplemental Fig.  S7), indicating that the intracellular interaction between HypA and HypB is independent of the availability of Ni 2ϩ . This is consistent with in vitro data indicating that a HypA-HypB complex could be observed in the presence or in the absence of Ni 2ϩ using a cross-linking assay (21).

DISCUSSION
As nickel chaperones, HypA and HypB proteins are known to play an important role in the maturation of both [NiFe]-hydrogenase and urease in H. pylori (19). HypB belongs to the G3E family of GTPase (35) and contains a highly conserved metalbinding site on its G-domain. HypA from both H. pylori and Thermococcus kodakaraensis shows that the protein consists of a zinc domain and a nickel-binding domain (20,36). Zinc coordinates to four conserved cysteines tetrahedrally in the loop region of the domain, whereas nickel binds in the N terminus of the protein. However, there appears to be no detailed study on HypA-HypB interaction.
The homodimeric structure of M. jannaschii HypB revealed that the GTP binding site is composed of residues from both monomers. An invariant lysine, Lys-153 (corresponding to Lys-168 in H. pylori HypB), from monomer B directly contacts the ␥-phosphate group of bound GTP molecule in monomer A, forming a salt bridge to stabilize the dimeric structure (29). We showed that substitution of this lysine by alanine in H. pylori HypB(K168A) completely abrogated the GTP-dependent dimerization of the protein, which was probably due to destruction of the salt bridge between the bound GTP ␥-phosphate and the lysine side chain. Similarly, substitution of the residues on a hydrophobic core on the dimeric interfaces (M186L and F190V) also abolished the GTP-dependent dimerization of H. pylori HypB. Consequently, no GTPase activities could be detected for both mutants (i.e. apo-HypB(K168A) and apo-HypB(M186L/F190V)).
Upon Ni 2ϩ binding, HypB(M186L/F190V) mutant formed dimers (Fig. 5), resulting in the formation of the intact GTP hydrolysis active site at the interface of the dimer of the protein.
As a result, the GTPase activity of Ni 2ϩ -HypB(M186L/F190V) significantly increased compared with its apo-form. Our data clearly indicated that monomeric HypB is in a low activity state. Ni 2ϩ acts as a "metallic bridge" that holds two monomers into an active dimer, in which two GTP hydrolysis active sites are formed by the invariant Lys-168 residue from both monomers and thus fully activated.
In contrast, no GTPase activity of Ni 2ϩ -HypB(K168A) was observed, although the mutant protein underwent Ni 2ϩ -de-  pendent dimerization (Fig. 5B). Because Lys-168 may directly contact with the bound GTP molecule, serving as an important component at the GTP hydrolysis site, mutation of Lys-168 to an alanine led to disruption of the GTP-dependent dimerization of HypB as well as the active site for GTP hydrolysis. Clearly, Lys-168 plays a dual role, whereas Met-186 and Phe-190 only play a role in the dimerization of HypB via hydrophobic contact at the dimeric interface.
Surprisingly, we found that Ni 2ϩ -H. pylori HypB exhibited a much lower affinity toward GDP compared with either apo-or Zn-HypB (K d of 5.6 versus Ͻ1 M, respectively; Fig. 6, A and B). Such lower affinity of GDP may facilitate the GDP/GTP exchange, thus elevating the GTPase activity. It would be of interest to further investigate whether the lower affinity of GDP to Ni 2ϩ -H. pylori HypB is due to a conformational change of the protein upon Ni 2ϩ binding.
The interaction of HypA and HypB has been found to be crucial for Ni 2ϩ delivery to the metalloenzymes. In agreement with a previous report (21), we found a heterodimer of HypA and HypB by the cross-linking method. The binding affinity is ϳ55 M as measured by NMR and ITC, indicative of weak interaction. This may explain why a GST pull-down assay failed to detect the interaction between HypA and HypB. Similarly, an interactomic study by tandem affinity purification also failed to demonstrate the interaction of HypA and HypB (37).
NMR is known to be a particularly powerful technique in characterization of weak macromolecule interactions (38). Some protein-protein interactions that cannot be detected by the pull-down assay are readily monitored by NMR (39). Our 1 H, 15 N HSQC experiments clearly showed that HypA interacts with HypB, as demonstrated by the chemical shift changes of specific resonances on HypA (Fig. 7). Furthermore, we found that the binding interface on HypA is at the cleft between helix 1 (␣1) and ␤1/␤6. It is noteworthy that binding of both Ni 2ϩ and HypB caused similar chemical shift perturbations on residues from HypA (supplemental Fig. S8), implying a potential competition between HypB and Ni 2ϩ . It is likely that binding of HypB could modulate Ni 2ϩ release from Ni 2ϩ -HypA, and our preliminary data indeed implied potential transfer of Ni 2ϩ from Ni 2ϩ -HypA to HypB. 3 The interface of the complex on HypB was examined by mutagenesis and chemical cross-linking. The N terminus of HypB was found to participate in the HypA-HypB interaction, although this region is flexible and absent from the reported M. jannaschii HypB structure (29). Particularly, Lys-18, Val-20, and Ile-22 are involved in the interaction. Our alanine scanning mutagenesis studies on a HypB homologue of Archaeoglobus fulgidus also showed that the N terminus of A. fulgidus HypB participated in the interaction with A. fulgidus HypA, 3 indicative of a conserved function of the N-terminal region. In contrast, UreG, a zinc-binding G-protein that participates in the maturation of urease and shares a high sequence similarity with HypB, lacks of the N terminus (40) (supplemental Fig. S9). This indicates that the N-terminal residues of HypB are unique and could be specifically recognized by HypA.
To validate the HypA-HypB interaction in vivo, we then used a GFP reassembly method. As shown in Fig. 9 and supplemental Fig. S7, the green fluorescence of E. coli cells containing the pET32a-NGFP-linker-HypA and pBAD33-HypB-linker-CGFP plasmids (NA ϩ BC), clearly verified the formation of the HypA-HypB complex in vivo (Fig. 9), implying the biologically relevant function of such an interaction. Interestingly, the interaction between HypA and HypB is not dependent on the availability of Ni 2ϩ because intracellular fluorescence could be observed in M9 medium with or without a supplement of Ni 2ϩ , in agreement with the NMR data.
The GTPase activity of H. pylori HypB can be enhanced by the binding of Ni 2ϩ , the same metal ion that the protein delivers. Similar features have also been found for E. coli ArsA, a soluble ATPase for arsenite and antimonite efflux (41). Intriguingly, a metallochaperone ArsD was identified to transfer trivalent metalloids to ArsA through weak or transient interaction for subsequent metalloid efflux (42)(43)(44). Based on the functional similarity between ArsA and HypB, we anticipate that HypA functions similarly to ArsD (i.e. delivering Ni 2ϩ to HypB for subsequent Ni 2ϩ transfer) (Fig.  10). Considering that both Zn 2ϩ and Ni 2ϩ bind H. pylori HypB, it is reasonable to speculate that H. pylori HypB utilizes an indirect approach, such as through specific proteinprotein interactions, to achieve selective binding to Ni 2ϩ in vivo, Protein-mediated Ni 2ϩ delivery pathway is therefore  postulated. First, Ni 2ϩ is sequestered by HypA via its N-terminal specific Ni 2ϩ binding site (20) and is subsequently transferred to HypB through a specific HypA-HypB interaction, initiating the dimerization of HypB. The active dimeric HypB will then deliver Ni 2ϩ to its downstream receptor (e.g. large subunit of [NiFe]-hydrogenase). Additional chaperone proteins, such as SlyD, which interacts with HypB (45), may also participate in this process. Therefore, further work is warranted to clarify this.