The persistence of Helicobacter pylori in the hostile environment of the human stomach is ensured by the activity of urease. The essentiality of Ni2+ for this enzyme demands proper intracellular trafficking of this metal ion. The metallo-chaperone UreE promotes Ni2+ insertion into the apo-enzyme in the last step of urease maturation while facilitating concomitant GTP hydrolysis. The present study focuses on the metal-binding properties of HpUreE (Helicobacter pylori UreE) and its interaction with the related accessory protein HpUreG, a GTPase involved in the assembly of the urease active site. ITC (isothermal titration calorimetry) showed that HpUreE binds one equivalent of Ni2+ (Kd=0.15 μM) or Zn2+ (Kd=0.49 μM) per dimer, without modification of the protein oligomeric state, as indicated by light scattering. Different ligand environments for Zn2+ and Ni2+, which involve crucial histidine residues, were revealed by site-directed mutagenesis, suggesting a mechanism for discriminating metal-ion-specific binding. The formation of a HpUreE–HpUreG protein complex was revealed by NMR spectroscopy, and the thermodynamics of this interaction were established using ITC. A role for Zn2+, and not for Ni2+, in the stabilization of this complex was demonstrated using size-exclusion chromatography, light scattering, and ITC experiments. A calculated viable structure for the complex suggested the presence of a novel binding site for Zn2+, actually detected using ITC and site-directed mutagenesis. The results are discussed in relation to available evidence of a UreE–UreG functional interaction in vivo. A possible role for Zn2+ in the Ni2+-dependent urease system is envisaged.
Helicobacter pylori is a widespread human pathogen that colonizes the gastric mucosa of approx. 50% of the world human population  and is responsible for severe diseases such as chronic gastritis, peptic and duodenal ulcers that eventually may lead to cancer . The bacterium is able to survive in the hostile environment of the human stomach through the activity of urease. This enzyme catalyses the hydrolysis of urea to ultimately yield ammonium and bicarbonate ions, thus causing an increase of the acidic local pH of the mucosa to values compatible with the survival of this pathogen. The remarkable enhancement of the rate of the reaction catalysed by urease as compared with the uncatalysed reaction (3×1015-fold)  is determined by the unique presence, among hydrolases, of Ni2+ ions in the active site of the enzyme . Several crystal structures of urease from bacteria [5–7] and plants  are available. The active site of the enzyme contains two essential Ni2+ ions bridged by a fully conserved, post-translationally carbamylated, lysine residue, and by a hydroxide ion, which acts as the nucleophile in the catalytic mechanism .
In general, the essentiality of Ni2+ for the activity of urease, together with the intrinsic cellular toxicity of this metal ion, requires a proper intracellular Ni2+ handling. In particular, urease is assembled in vivo as an inactive apo-enzyme, and undergoes a maturation process that involves Ni2+ incorporation and lysine carbamylation to produce a fully active holo-enzyme . This assembly process requires the involvement of dedicated accessory proteins, namely UreD (UreH in H. pylori), UreF, UreG and UreE. The current model for this process  entails the formation of a protein complex between the apo-enzyme and these protein chaperones, followed by the delivery of Ni2+ concomitantly with the GTP-dependent transfer of a CO2 molecule necessary for lysine carbamylation. Among all urease accessory proteins, the structural properties and role of UreD remain largely obscure; this protein has been proposed to bind to apo-urease, thus inducing a conformational change required for the subsequent steps of the activation process , a hypothesis recently supported by small angle X-ray scattering results . The latter study also appears to indicate that UreF contributes to this conformational change, which ultimately allows CO2 and Ni2+ ions to gain access to the nascent active site. UreG is responsible for the GTP hydrolysis associated to the transfer of CO2 to the active site lysine; the protein is intrinsically disordered, specifically binds Zn2+, and features very low or absent hydrolysing activity in vitro, probably needing the interaction with other protein partners to achieve its fully folded and functional structure [13–16]. UreF has been proposed to act as an activator of the GTPase activity of UreG on the basis of structural bio-modelling studies .
UreE is believed to act as a Ni2+ carrier based on the evidence that the levels of Ni2+ necessary for the in vitro assembly of the urease active site are significantly reduced, and a much larger portion of enzyme molecules is activated, if UreE is present [11,18,19]. The crystal structures of BpUreE (Bacillus pasteurii UreE ) and of a truncated form of KaUreE (Klebsiella aerogenes UreE), H144*KaUreE , indicate that UreE are symmetric homodimers, with each monomer made of an N-terminal domain and a C-terminal domain, the latter involved in the head-to-head dimerization. A metal-ion-binding site, found at the protein dimerization interface, involves two conserved histidines, one from each monomer (His100 in BpUreE and His96 in KaUreE, Figure 1A). This site contains Zn2+ in the structure of BpUreE and Cu2+ in that of H144*KaUreE, but is generally assumed to be occupied by Ni2+ in the protein functional form in vivo, a hypothesis supported by anomalous difference X-ray diffraction maps of BpUreE crystals soaked in a Ni2+ solution . The structure of H144*KaUreE also contains two additional Cu2+ ions bound to a pair of histidines on the surface of each monomer (His110 and His112), but these residues are not conserved in BpUreE (Tyr114 and Lys116), HpUreE (H. pylori UreE) (Phe118 and Lys120) (Figure 1) or other sources .
Structural comparison and sequence alignment of HpUreE, BpUreE and KaUreE
An alternative possible physiological role for UreE is suggested by the observation that the GTP concentration needed for optimal activation of urease in vitro is greatly reduced in the presence of UreE as compared with that required in its absence . This implies that UreE must play an important direct or indirect role in the functional activation of UreG. In this respect, in vivo studies using yeast two-hybrid analysis [23,24] as well as co-immunoprecipitation assays  indicated a direct interaction between UreE and UreG from H. pylori.
On the basis of the available evidence, the question of which functional role is played by UreE in the urease active site assembly is still awaiting a definitive answer. The present study represents an attempt to clarify some details of the reactivity of UreE towards Ni2+ and Zn2+, as well as with HpUreG.
Recombinant wild-type HpUreE was prepared using a protocol adapted from a previous study . Purity was checked using SDS/PAGE (see Supplementary Figure S1 at http://www.BiochemJ.org/bj/422/bj4220091add.htm). Site-directed mutagenesis, aimed at the production of the H102K and H152A mutants, was carried out using standard procedures. The full experimental details of gene cloning, protein expression and purification are provided in the Supplementary Online Data (at http://www.BiochemJ.org/bj/422/bj4220091add.htm). Recombinant HpUreG and its C66A/H68A mutant were obtained as previously described in .
The secondary structure of HpUreE was evaluated by CD spectroscopy, performed on the protein (5 μM) diluted in 20 mM phosphate buffer, pH 8.0, using a JASCO 810 spectropolarimeter flushed with N2, and a cuvette with 0.1 cm path-length. Ten spectra were accumulated from 190 to 240 nm at 0.2 nm intervals, and averaged to achieve an appropriate signal-to-noise ratio. The spectrum of the buffer was subtracted. The secondary structure composition of HpUreE was evaluated using the CDSSTR tool available on the Dichroweb server, with the reference sets 3, 4, 6 and 7 (http://dichroweb.cryst.bbk.ac.uk).
NMR spectroscopy experiments
Uniformly 15N-labelled HpUreE and HpUreG were produced with the same purification procedure used for the unlabelled proteins, using M9 minimal medium containing 15NH4Cl as sole nitrogen source. Solutions containing 15N-labelled HpUreE (200 μM) and unlabelled HpUreG, or 15N-labelled HpUreG (300 μM) and unlabelled HpUreE, were prepared by mixing one equivalent of the HpUreE dimer with two equivalents of the HpUreG monomer. 1H-15N HSQC (heteronuclear single quantum correlation) experiments  and TROSY (transverse relaxation optimized spectroscopy)–HSQC experiments  were carried out at 298 K on a Bruker 800 MHz spectrometer equipped with a TXI cryoprobe. The acquisition parameters for these NMR spectra are provided in Supplementary Table S1 (at http://www.BiochemJ.org/bj/422/bj4220091add.htm). Spectra were processed and analysed using TopSpin (Bruker) and iNMR (http://www.inmr.net/).
Light scattering measurements
In a typical experiment, a protein sample (100 μl, 50 μM) was loaded on to a size-exclusion Superdex 200 HR 10/30 column (GE Healthcare), pre-equilibrated using 20 mM Tris/HCl, pH 8.0, and 150 mM NaCl, and eluted at a flow rate of 0.6 ml/min. The column was connected downstream to a multi-angle laser light (690.0 nm) scattering DAWN EOS photometer (Wyatt Technology) and to a quasi-elastic light scattering WyattQELS device. The concentration of the eluted protein was determined using a refractive index detector (Optilab DSP, Wyatt). The specific refractive index increment (dn/dc) for the proteins was taken as 0.185 ml/g . The value of 1.321 was used for the solvent refractive index. Molecular masses were determined from a Zimm plot. Data were recorded and processed using the Astra 5.1.9 software (Wyatt Technology), following the manufacturer's indications. When the measurements were carried out in the presence of metal ions, stoichiometric amounts of ZnSO4 or NiSO4 were added to the protein samples before loading it on to the size-exclusion column, and the protein was eluted using the same buffer containing 20 μM ZnSO4 or NiSO4. In order to explore the formation of a HpUreE–HpUreG interaction, a solution containing HpUreE (50 μM dimer) and HpUreG (100 μM monomer) was analysed in the absence and in the presence of 200 μM NiSO4 or 200 μM ZnSO4 under the same experimental conditions.
ITC (isothermal titration calorimetry) experiments
Titration experiments were performed at 25 °C using a high-sensitivity VP-ITC microcalorimeter (MicroCal LLC, Northampton, MA, U.S.A.). The proteins and the metal ions (from 100 mM stock solutions) were diluted using the same buffer (20 mM Tris/HCl, pH 7.0, and 150 mM NaCl) and eluted from a size-exclusion column, utilized immediately before the ITC measurement to freshly purify the protein. The measuring cell contained 1.4093 ml of protein solution, and the reference cell was filled with deionized water. Before starting the experiments, the baseline stability was verified. A spacing of 400–600 s between injections was applied in order to allow the system to reach thermal equilibrium after each addition. For each titration, a control experiment was carried out by adding the titrating solution into the buffer alone, under identical conditions. Heats of dilution were negligible. In a set of experiments aimed at determining the metal-binding properties of HpUreE, the protein (10 μM) or its H102K and H152A mutants (10 μM) were titrated with 30 injections (10 μl each) of a solution containing 100 μM NiSO4 or ZnSO4. In order to determine the binding parameters of HpUreE to HpUreG, the latter protein (50 μM monomer) was titrated with 30 injections (10 μl each) of a solution containing 160 μM HpUreE dimer in the same buffer. In the case of Zn2+ titration on to the HpUreE–HpUreG complex (5 μM), generated in situ by mixing 5 μM HpUreE and 10 μM HpUreG monomer, 30 aliquots (10 μl each) of a solution containing 70 μM ZnSO4 were injected into the protein solution. Identical set up was used for the related mutants. The details of data analysis are given in the Supplementary Online Data. The dissociation constants and thermodynamic parameters provided in the present study do not take into account possible events of proton transfer linked to metal binding, or the presence, in solution, of complexes between the metal ions and the buffer. This treatment is beyond the scope of the present study. However, the values of the measured equilibrium constants compare well with those reported in the literature and determined using ITC or other methodologies such as equilibrium dialysis coupled to metal analysis, which, in principle, should also take into account similar effects. These values are therefore only used for comparison purposes.
Structural modelling of the HpUreE–HpUreG complex
The alignment of BpUreE and HpUreE  was used to calculate, using the MODELLER9v5 software , 50 structural models of the HpUreE dimer, imposing the structural identity of the two monomers. The best model was selected on the basis of the lowest value of the MODELLER objective function. The ROSETTADOCK software  was used to calculate an initial complex between the model structure of dimeric HpUreG , and the central C-terminal domains of dimeric HpUreE. The complex with the best ROSETTADOCK score was selected among all generated models for the subsequent refining run, carried out by applying 1000 times a perturbation to the starting structure. The Cα trace of this complex was used, together with the crystal structures of MjHypB (Methanocaldococcus jannaschii HypB , PDB code 1HF8), BpUreE  (PDB code 1EAR), and KaUreE  (PDB code 1GMW) as templates to build 200 structural models of the HpUreE–HpUreG complex using the MODELLER9v5 software . The best model was selected on the basis of the lowest value of the MODELLER objective function. The full details of the model calculation are provided as Supplementary Online Data.
RESULTS AND DISCUSSION
HpUreE structural properties
A prediction of the secondary structure elements of HpUreE (Figure 1B) based on the JPRED algorithm  indicates 21% α-helix and 23% β-strand content, with the remaining 56% constituting turns or random coil conformations. Figure 1(B) also suggests that a similar fold is attained by several different UreE proteins, as previously proposed on the basis of modelling studies . The CD spectrum of HpUreE (Figure 2A) shows the presence of both α-helices and β-strands, with negative deflections around 218 nm and 208 nm and a positive peak at 190 nm. The CD spectrum was quantitatively analysed and the best fit [NRMSD (normalized root mean square deviation)=0.029] estimated a secondary structure composition of 13% α-helix, 33% β-strand, 23% turns and 30% random coil for HpUreE. This composition is similar to that calculated using the DSSP program  for the crystallographic structures of Zn2+-bound BpUreE  (18% α-helix, 37% β-sheet) and of the Cu2+-bound H144*KaUreE  (13% α-helix, 25% β-sheet). The 1H chemical shift spreading observed in the TROSY–HSQC NMR spectrum (Figure 2B) ranges from 6.5 to 10 p.p.m., as observed in the 1H-15N HSQC NMR spectrum of BpUreE , suggesting similar extent of fold for the two proteins.
CD, NMR, and MALS/QELS of HpUreE
The molecular mass and the hydrodynamic radius of HpUreE in solution were determined using a combination of SEC (size-exclusion chromatography) and light scattering [MALS (multiple angle light scattering)/QELS (quasi-elastic light scattering)] (Figure 2C). The elution profile and the light scattering data show that HpUreE is a dimer in solution with M=43.1±4.8 kDa and Rh=3.0±1.4 nm (theoretical mass=39 kDa). This is consistent with all available crystallographic structural information on UreE proteins (Figure 1B) [20,21] as well as with previous evidence collected on HpUreE based on SEC criteria . The light scattering measurements exclude the possibility that oligomers of the apo-protein are formed in solution for concentrations lower than 50 μM, as instead previously proposed for H144*KaUreE . The better quality of the 1H-15N TROSY–HSQC (Figure 2B) with respect to the simple 1H-15N HSQC experiment (results not shown) further supports the presence of a dimeric form at 0.1–0.3 mM concentration. The symmetric architectural arrangement of the two monomers is revealed by the number of observed peptide NH peaks in the NMR spectrum: about 120 unique peaks, out of the expected 170 residues per monomer, can be observed, with missing signals probably including the C-terminal 30 residues predicted to be unstructured using JPRED (Figure 1B). In particular, in the case of glycine residues, seven glycines are present in the sequence of HpUreE (Figure 1B) and the same number of peaks is observed in the 15N 100–110 p.p.m. range typical for glycine NH signals (Figure 2B).
HpUreE metal-binding properties
Ni2+ is generally considered to be the physiological cofactor of UreE, and the understanding of the structural features of Ni2+ binding is therefore important to clarify the role of this chaperone in vivo. Moreover, in several recent instances, interplay between Ni2+ and Zn2+ has been observed and proposed to be functionally important in regulating cellular trafficking of metal ions [13–16,31,36–40]. In particular, Zn2+ is involved in the dimerization of HpUreG, a process that plays a potential regulatory role in the urease active site assembly . Previous equilibrium dialysis experiments carried out on HpUreE established a 1:1 stoichiometry for the Ni2+ binding to the homodimeric protein, with Kd approx. 1 μM . However, the experiments were carried out at pH 8.25 in an apparently non-buffered solution containing only NaCl, no thermodynamic parameters for the metal binding event were determined, and the binding affinity for Zn2+ was not measured . In the present study, the Ni2+ binding to HpUreE was investigated using ITC, and a comparison between Ni2+ and Zn2+ binding was performed.
The ITC measurements were carried out by adding Ni2+ or Zn2+ to the apo-protein in a buffered solution at pH 7.0, and the occurrence of a binding event was revealed by the presence of exothermic peaks that followed each addition (Supplementary Figures S2A and S2B at http://www.BiochemJ.org/bj/422/bj4220091add.htm for Ni2+ and Zn2+ respectively). Fits of the integrated heat data (Figure 3A and 3B for Ni2+ and Zn2+ respectively) were carried out using the simplest model, which entails a single binding event, and yielded a stoichiometry of one equivalent of Ni2+ or Zn2+ bound to the HpUreE dimer. Dissociation constants Kd(Ni)=0.15±0.01 μM and Kd(Zn)=0.49±0.01 μM were calculated for Ni2+ and Zn2+ binding respectively. In both cases, these processes are driven by favourable enthalpic factors [ΔH(Ni)=−13±1 kcal/mol (1 cal≈4.184 J), ΔH(Zn)=−10±1 kcal/mol] that compensate the negative entropic values [ΔS(Ni)= −13 cal·mol−1·K−1, ΔS(Zn)=−4 cal·mol−1·K−1] calculated from the fit. The values of the Kd(Ni) and Kd(Zn) measured for HpUreE are comparable with those established by ITC for the binding of Ni2+ and Zn2+ to BpUreE and H144*KaUreE , and by equilibrium dialysis for the binding of Ni2+ to BpUreE , of Ni2+ and Zn2+ to H144*KaUreE [41,42] and of Ni2+ to HpUreE . All these values are consistent with a role of intracellular metal ion transport associated with UreE proteins .
ITC data of NiSO4 and ZnSO4 binding to wild-type HpUreE and its H102K and H152A mutants
In the structure of Zn2+-BpUreE  and Cu2+-H144*KaUreE , the metal ions are bound to the surface of the protein using the conserved His100 and His96 residues respectively (Figure 1A). In order to firmly establish the role of the corresponding His102 in the binding of Ni2+ and Zn2+ to HpUreE, the H102K mutant was obtained by site-directed mutagenesis. ITC titrations of H102K HpUreE with Ni2+ and Zn2+, performed under identical conditions as for the wild-type protein, proved the absence of a binding event (Figures 3A and 3B), confirming the key role of this residue in metal binding to HpUreE.
In the crystal structure of Zn2+-bound BpUreE, the protein is present as a dimer of dimers, with the metal ion in a bridging position, bound to four conserved His100 residues, one from each monomer . However, this oligomerization has been observed for BpUreE only in the solid state  or in concentrated (mM) solutions , while dynamic light scattering of the metalbound protein in the 50–250 μM range excluded this effect . [The molecular mass and hydrodynamic radius of BpUreE (50 μM dimer) in the absence and in the presence of two equivalents of Ni2+ and Zn2+ per dimer were determined in the present study using SEC on-line with MALS and QELS. The results indicate that, like HpUreE, BpUreE is a dimer in solution both in the absence and in the presence of these metal ions, with M=41.0 kDa and Rh=3.0 nm.] In the case of HpUreE, the influence of metal binding on the quaternary structure of the protein in the range of concentrations used in the microcalorimetric metal binding studies (10–20 μM) was investigated using a combination of light scattering methods (MALS and QELS). The values measured for Ni2+–HpUreE (M=45.7±5.1 kDa, Rh=3.4±1.5 nm) and Zn2+–HpUreE (M=46.4±5.2 kDa, Rh=3.3±1.5 nm) are similar to those established for the apo-protein (Figure 2C), demonstrating that the metal-bound protein is a dimer, and not a dimer of dimers, independently of the presence of bound metal ions. This is consistent also with the similar linewidths in the TROSY–HSQC NMR spectra of apo-, Ni2+-bound and Zn2+-bound HpUreE, which indicates similar protein size in the various metal-bound states (Supplementary Figure S3 at http://www.BiochemJ.org/bj/422/bj4220091add.htm).
While the dissociation constants observed for the metal ion complexes of HpUreE are in the μM range previously determined for UreE proteins from different sources, the 1:1 metal ion binding stoichiometry established for HpUreE differs from previous data obtained for KaUreE and BpUreE, which indicated a 2:1 stoichiometry. KaUreE additionally binds three Ni2+ ions to a histidine-rich tail containing ten histidines among the last 15 residues, absent both in HpUreE and in BpUreE . In the case of BpUreE, the presence of a binuclear [Ni(OH)Ni]3+ centre was proposed on the basis of EXAFS spectra, rendering this protein not only a Ni2+ transporter but also a potential scaffold for the assembly of the dinuclear active site of urease . In BpUreE and KaUreE, a conserved HXH motif is present in this protein region: in particular, in BpUreE the HQH motif is located at the end of the sequence, whereas in KaUreE several possible HXH concatenated motifs constitute the His-rich tail (HGHHHAHHDHHAHSH). On the other hand, in HpUreE a single histidine (His152) is observed in the C-terminal tail (Figure 1B). On the basis of these considerations, a possible reason for the different stoichiometry of the Ni2+ binding to HpUreE on one side (1:1), and to BpUreE and KaUreE on the other (2:1), might reside in the different sequence motifs of histidine residues found at the C-terminal tails of these proteins. These observations suggest a specialized role, in metal ion storage and/or delivery, for the C-terminal portion of the protein, depending on its length and composition, and prompted us to investigate whether His152 is involved in metal ion binding to HpUreE using the H152A mutant.
ITC titrations of H152A HpUreE with Ni2+, performed under identical conditions as for the wild-type protein, proved the occurrence of a binding event (Supplementary Figure S2C). The integrated heat data, fitted using a single site model (Figure 3C), yielded values for the dissociation constant [Kd(Ni)=0.87±0.01 μM], reaction enthalpy [ΔH(Ni)=−12±1 kcal/mol] and reaction entropy [ΔS(Ni)=−11 cal·mol−1·K−1], which are consistent with those obtained for the wild-type protein (Figure 3A). These data indicate that the His152 residue is not involved in binding the Ni2+ ion in HpUreE, suggesting that this residue is not essential for the nickel-delivery function. In turn, this also supports the concept that both histidine residues found in the HXH motif at the C-terminal end of BpUreE and KaUreE are necessary to build up the dinuclear Ni2+ centre observed in those cases.
On the other hand, the titration of H152A HpUreE with Zn2+ showed clear differences in the binding mode as compared with the wild-type protein (Figure 3D and Supplementary Figure S2D). Best fits of the integrated heat data could be obtained using a model involving not one, as in the case of wild-type HpUreE, but two independent binding events, yielding dissociation constants Kd1(Zn)=0.13±0.02 μM and Kd2(Zn)=0.82±0.01 μM. Both events are driven by favourable enthalpic [ΔH1(Zn)=−4±1 kcal/mol, ΔH2(Zn)= −7±1 kcal/mol) and entropic [ΔS1(Zn)=19 cal·mol−1·K−1, ΔS2(Zn)=6 cal·mol−1·K−1] factors. The observation of an additional Zn2+-binding site upon replacement of a histidine with a non-co-ordinating residue like alanine, could be, at first sight, bewildering. This apparent incongruity, resulting from our experimental data, can be explained by taking into consideration the peculiarity of the protein region where the mutation is carried out. This is a long extended and flexible stretch whose conformation or relative orientation with respect to the rest of the protein could change as a consequence of point mutations. Therefore, it is possible that, while His152 is involved in Zn2+ binding by isolated wild-type HpUreE, the resulting conformation of the flexible C-terminal arm masks an additional binding site, which becomes accessible upon mutation of this residue.
The conclusions that can be drawn from these results are that Ni2+ and Zn2+ bind to wild-type HpUreE using different modes. Although Ni2+ is bound to the conserved His102 on the surface of the dimer, without any involvement of His152, Zn2+ binding not only requires the His102 pair, but is also modulated by the two His152 residues at the C-terminal position. A different binding mode for the two metal ions to the wild-type protein is supported by a comparison of the TROSY–HSQC spectra of HpUreE in the apo-form with the same spectra of the Zn2+- and Ni2+-bound forms (see Supplementary Figure S3 at http://www.BiochemJ.org/bj/422/bj4220091add.htm). Residues changing their chemical shifts upon nickel addition are also affected (and at the same extent) by the presence of Zn2+. In addition, a few more peaks change their position in the presence of Zn2+, consistently with the involvement of a larger number of residues in the Zn2+-binding event. The higher availability of intracellular Zn2+ as compared with Ni2+, together with the similar affinity of HpUreE for these two metal ions described above, suggests that the specificity of binding different metals must rely on changes in ligand environment, as indeed observed experimentally.
At least one histidine residue is always present near the C-termini of all UreE sequences , although this feature is not maintained in the H152A HpUreE mutant, or in H144*KaUreE (Figure 1B). Consistent with this observation, the calorimetric Zn2+ titration curve obtained for H152A HpUreE resembles the one reported for the binding of Ni2+ and Cu2+ to H144*KaUreE at protein concentrations ≥25 μM . The data for the latter protein were interpreted as indicating an initial binding of two Ni2+ or Cu2+ ions, one after the other, to the interface of the H144*KaUreE dimer of dimers, assumed to be the most abundant species in solution. This binding site was suggested to give rise to a tetrameric structure and to involve the four conserved His96 residues (one per each monomer) . In the case of HpUreE, however, no oligomerization events occur for the H152A mutant in the presence of Zn2+ or Ni2+, as demonstrated using light scattering experiments (see Supplementary Figure S4 at http://www.BiochemJ.org/bj/422/bj4220091add.htm).
The available experimental evidence indicate that UreE and UreG form a functional complex in vivo [19,23,24]. In order to observe and characterize this interaction in vitro, a solution containing equimolar amounts of the two purified apo-proteins was analysed by SEC and light scattering (Figure 4A). The result of this experiment indicates that the dimeric HpUreE elutes as a species separated from HpUreG, the latter being present in the monomeric state, as recently reported . Considering that the experimental setup used for the SEC–MALS–QELS measurement represents non-equilibrium conditions, we monitored this interaction more quantitatively using ITC. When a solution of HpUreG was titrated with a solution of HpUreE in the same buffer, clear exothermic peaks were observed (Supplementary Figure S5A at http://www.BiochemJ.org/bj/422/bj4220091add.htm) which, after integration, revealed a curve (Figure 4B) that could be fitted using a single binding event model. The stoichiometry of the interaction suggests that two monomers of HpUreG bind to a single dimer of HpUreE, forming a HpUreE–HpUreG complex having a dissociation constant Kd=4.0±0.3 μM, ΔH=−12.5±0.9 kcal/mol, and ΔS=−17.5 cal·mol−1·K−1.
HpUreE–HpUreG interaction analysis
The formation of the HpUreE–HpUreG complex was also monitored using NMR spectroscopy. The TROSY–HSQC spectrum of the solution containing one equivalent of 15N-HpUreE dimer and two equivalents of unlabelled HpUreG monomer differs from that of HpUreE in the absence of HpUreG (Supplementary Figure S6A at http://www.BiochemJ.org/bj/422/bj4220091add.htm). A general broadening of the peaks is observed, and a number of them show small chemical shift changes (Δδav≤0.1 p.p.m.), consistent with the formation of a complex between the two proteins, as observed by ITC. The number of glycine resonances does not increase upon complex formation, suggesting maintenance of the UreE homodimeric symmetry. Addition of one equivalent of unlabelled HpUreE dimer to a solution of 15N-HpUreG causes broadening beyond detection for most of the backbone amide signals observed in a regular 1H-15N HSQC spectrum (results not shown) as compared with the same spectrum of HpUreG . Resonances could be recovered by recording a 1H-15N TROSY–HSQC spectrum (Supplementary Figure S6B), an effect that can be explained with the large molecular mass (approx. 80 kDa) of the HpUreE-HpUreG complex. Small chemical shift changes on selected resonances of backbone amides are observed (Δδav≤0.05 p.p.m.) upon complex formation. An assignment of the resonances, beyond the scope of the present study, would provide information on the surface contact areas, and is currently underway in our laboratories.
The observed 1:2 stoichiometry of the HpUreE–HpUreG complex, coupled to the previously reported dimerization of HpUreG selectively induced by the binding of one equivalent of Zn2+ per protein dimer, and not by Ni2+-binding , prompted us to explore the role of these two metal ions in the stabilization of the protein complex. The SEC elution profile and molar masses of a mixture of HpUreE and HpUreG in a 1:2 ratio, as measured by MALS, were not affected by the presence of Ni2+, indicating that this metal ion is not capable of significantly stabilizing the protein complex. This is consistent with the absence of a role for Ni2+ in the dimerization of HpUreG . On the other hand, when the same experiment was carried out using Zn2+, the elution peaks corresponding to the separate HpUreE and HpUreG proteins completely disappeared, and a new unique peak, with M=79.4±2 kDa and Rh=5.8±0.1 nm, was concomitantly observed (Figure 4A). This result indicates that the interaction between HpUreE and HpUreG, leading to the establishment of a complex, is specifically stabilized by Zn2+. On the basis of the theoretical masses of the HpUreE dimer (39 kDa) and of the HpUreG monomer (22 kDa), the mass of the new species formed in the presence of Zn2+ is fully consistent with the 1:2 stoichiometry established by ITC.
Molecular model of the HpUreE–HpUreG complex
The viability of the HpUreE–HpUreG complex formation was investigated from a structural modelling point of view. The model structure of the HpUreE dimer was docked on to the model of the dimeric form of HpUreG , with optimization of protein backbone and side chains at the interface between the two homodimers. In the resulting structure (Figures 5A and 5B), the two proteins face each other along their extended axes, and only limited modifications of the proteins backbone, restricted both in extent and in topology distribution, were necessary in order to optimize the docking procedure (Figure 5A). The central pocket formed on the HpUreG surface around the conserved Cys66 and His68 residues matches the shape and volume of the protruding region around the pair of conserved His102 residues on the surface of HpUreE. The shallow crevice formed between the central C-terminal domain and the peripheral N-terminal domain of HpUreE is filled with the bulge found on the surface of HpUreG around the rim of the protein dimerization interface (Figures 5A and 5B). Overall, a full size, shape, and electric charge complementarity between the surfaces of the two proteins is observed (Figures 5C and 5D), with the formation of the complex resulting in a large total area [6378 Å2 (1 Å=0.1 nm), 40.4% of HpUreE and 36.3% of HpUreG] that is buried by the two interacting homodimers. The details of the interaction are given in the Supplementary Online Data.
Model of the HpUreE–HpUreG interaction
Role of Zn2+ in the stabilization of the HpUreE–HpUreG complex
A recent article has established the key importance of the surface exposed conserved Cys66 and His68 residues in HpUreG for the binding of Zn2+ . Moreover, the present study indicates a role for His102 and His152 in metal ion binding to HpUreE. The structure of the HpUreE–HpUreG complex features Cys66, His68 of HpUreG, and His102 of HpUreE, in neighbouring positions in the central region of the complex. This suggests the building up of a novel metal-binding site at the interface between the two protein partners (see close-up in Figure 5A). The position of His152 cannot be predicted by the model because of the absence of structural data for the terminal disordered region. In order to experimentally verify the presence of this site, we carried out calorimetric titrations of Zn2+ onto a solution containing a preformed complex obtained in situ by mixing the two proteins with a 1:2 stoichiometry HpUreE dimer:HpUreG monomer (Figure 4C and Supplementary Figure S5). The curve obtained using the wild-type proteins indeed reveals an event of binding, characterized by Kd=1.5±0.3 nM, which is approx. 2–3 orders of magnitude tighter than those observed for isolated HpUreE (the present study) or HpUreG . This event is distinct from an additional following binding step with Kd=0.67±0.05 μM. Therefore, the HpUreE–HpUreG complex binds two Zn2+ ions, in a high-affinity and a low-affinity site.
The identity of the residues involved in these two binding events was investigated by repeating the same experiment using, instead of the wild-type proteins, the mutants H102K HpUreE, H152A HpUreE or C66A/H68A HpUreG (Figure 4C) . In the case of each of the HpUreE mutants, the tight binding event is not observed, whereas it is maintained when the mutant of HpUreG is used. On the other hand, the low-affinity site is still present when the two mutants of HpUreE are utilized, but is disrupted in the case of C66A/H68A HpUreG. The residual binding of two Zn2+ ions in the latter case reproduces what was previously observed for the HpUreG double mutant alone . Overall, these results suggest that His102 and His152 contribute to the building of the high-affinity site in the complex, whereas HpUreG Cys66 and His68 residues are responsible for the low-affinity binding event. The exact topology of these two metal-binding sites cannot be determined at the present stage of the study given the limited structural information on the flexible C-terminal pendant arms containing His152. However, the calculated model structure of the protein complex allows us to speculate that these protein regions could adapt their conformation to fill the cleft that is formed between the proteins' interaction surfaces (Figure 5B), bringing His152 close to the metal-binding site that also involves His102. On the other hand, the lower affinity binding event could involve a nearby site situated close to Cys66 and His68 on HpUreG. It is worth mentioning here that crystallographic evidence indicates the presence of a dinuclear Zn2+ binding site on the surface of MjHypB (a close homologue to HpUreG), which involves a cysteine and a histidine residue corresponding to Cys66 and His68 on HpUreG .
The stabilization of a HpUreE-HpUreG complex in the presence of Zn2+ is significant within the framework of the known role of HpUreG in vivo: this protein is an enzyme that catalyses GTP hydrolysis necessary to the urease activation process . HpUreG belongs to a class of homo-dimeric GTPases (or ATPases) that use GTP (or ATP) hydrolysis as a conserved molecular switch to regulate a large number of cellular processes . The activity of these hydrolases is, in general, tightly controlled by different factors, such as protein dimerization and subsequent interaction with GAP (GTPase activator protein) and GEF (guanine-nucleotide-exchange factor) proteins. These GTPase regulators are stable functional dimers, as observed for UreE. In order to test a possible role for HpUreE in modulating the enzymatic cycle of HpUreG, we measured the GTPase activity of the HpUreE–HpUreG complex formed in the presence of Zn2+ using both a colorimetric and an enzymatic method, as previously described . We found that the interaction with HpUreE stabilized by Zn2+ is not sufficient to promote any detectable GTPase activity. This result is not surprising: it is known that the GTP-dependent process of nickel incorporation into apo-urease occurs only in the presence of a UreDFG complex, implying that UreD and UreF must also play an essential role in UreG activation.
The results presented here allow us to envisage a mechanism for the urease assembly that entails a specific role for both Ni2+ and Zn2+ ions. An exchange of Zn2+ for Ni2+ binding to HpUreE could be the initial switch that modulates the interaction between HpUreE and HpUreG. Ni2+ released from HpUreE could be incorporated into the apo-urease active site, concomitantly with the Zn2+-induced HpUreE-HpUreG complex formation and consequent stimulation of GTPase activity catalysed by HpUreG. This step would lead to the carbamylation of the lysine residue in the urease active site, thus finalizing the activation of the enzyme.
The chaperones involved in the maturation of [NiFe]-hydrogenase in Escherichia coli, such as HypA (and its homologue HybF) [37,38] and HypB , also display a specific Zn2+-binding capability. On the other hand, HypA from H. pylori, responsible for the activation of both urease and hydrogenase enzymes , shows Ni2+-binding capability in vitro . The occurrence of a specific interaction between dimeric HypA and UreE from H. pylori was proved using cross-linking and immuno-blotting experiments, suggesting that this interaction is functional to mediate Ni2+ transfer from HypA to UreE in vivo . Specific protein–protein interactions were observed between HypA and HypB during cross-linking experiments, as well as between HypB and UreG from tandem affinity purification . These results, coupled to the observation of specific UreE–UreG interactions, indicate the possible presence of cross-talk mechanisms in vivo, involving HypA, HypB, UreG and UreE. It is interesting to notice that all these proteins possess Ni2+- and/or Zn2+-binding capability. Therefore, the Zn2+-dependent interaction between HpUreE and HpUreG, as well as the interdependence between Ni2+ and Zn2+, emerging in this study for the H. pylori urease system, suggests a functional role for metal binding to these accessory proteins, modulating the formation of the protein–protein complexes necessary for enzyme maturation. These observations represent a paradigmatic general point to understand the role of Zn2+ in the process of Ni2+ delivery and incorporation in different enzymes.
heteronuclear single quantum correlation
isothermal titration calorimetry
multiple angle light scattering
quasi-elastic light scattering
transverse relaxation optimized spectroscopy
Matteo Bellucci optimized and carried out the preparation of wild-type and mutated HpUreE, as well as the calorimetric titrations. Barbara Zambelli was responsible for gene cloning, the preliminary setup of the protein purification protocol, as well as collection and analysis of CD spectra and light scattering measurements. Francesco Musiani carried out the bio-modeling calculations. Paola Turano planned and performed the NMR experiments. Stefano Ciurli and all co-authors equally contributed to the design of the experiments, analysis of the data, and preparation of the manuscript.
This study was supported by MIUR-PRIN2007. The VP-ITC instrument is property of CIRB-UniBO. The fellowships for M. B. and F. M. are provided by UniBO (University of Bologna). B. Z. was supported by Consorzio Interuniversitario di Risonanze Magnetiche di Metalloproteine Paramagnetiche and by UniBO.
The nucleotide sequence data reported in this study will appear in the DDBJ, EMBL, GenBank® and GSDB Nucleotide Sequence Databases under the accession number ABM16833.