UreG, a chaperone in the urease assembly process, is an intrinsically unstructured GTPase that specifically binds Zn2+.

Bacillus pasteurii UreG, a chaperone involved in the urease active site assembly, was overexpressed in Escherichia coli BL21(DE3) and purified to homogeneity. The identity of the recombinant protein was confirmed by SDS-PAGE, protein sequencing, and mass spectrometry. A combination of size exclusion chromatography and multiangle and dynamic laser light scattering established that BpUreG is present in solution as a dimer. Analysis of circular dichroism spectra indicated that the protein contains large portions of helices (15%) and strands (29%), whereas NMR spectroscopy indicated the presence of conformational fluxionality of the protein backbone in solution. BpUreG catalyzes the hydrolysis of GTP with a kcat=0.04 min(-1), confirming a role for this class of proteins in coupling energy requirements and nickel incorporation into the urease active site. BpUreG binds two Zn2+ ions per dimer, with a KD=42 +/- 3 microm, and has a 10-fold lower affinity for Ni2+. A structural model for BpUreG was calculated by using threading algorithms. The protein, in the fully folded state, features the typical structural architecture of GTPases, with an open beta-barrel surrounded by alpha-helices and a P-loop at the N terminus. The protein dynamic behavior observed in solution is critically discussed relative to the structural model, using algorithms for disorder predictions. The results suggest that UreG proteins belong to the class of intrinsically unstructured proteins that need the interaction with cofactors or other protein partners to perform their function. It is also proposed that metal ions such as Zn2+ could have important structural roles in the urease activation process.

Urease is a nickel-containing enzyme found in plants, fungi, and bacteria that catalyzes the hydrolysis of urea in the last step of nitrogen mineralization (1, 2) (Scheme 1).
Over the past few years, intensive studies have been carried out to achieve an elucidation of its catalytic mechanism. Structures of the native enzyme isolated from Klebsiella aerogenes (Ka) 1 (3), Bacillus pasteurii (Bp) (4), and Helicobacter pylori (Hp) (5) revealed a dinuclear metallo-center, with two Ni 2ϩ ions bridged by a carbamoylated lysine residue and a hydroxide ion. The enzyme, consisting of a heterotrimeric ␣ 3 ␤ 3 ␥ 3 quaternary structure with the three different subunits encoded by the ureC, ureB, and ureA genes, respectively, is synthesized in the apo-form that is devoid of nickel. The incorporation of Ni 2ϩ into the active site, leading to the activation of the enzyme, is still poorly understood and is thought to occur in vivo as a stepwise assembly process (6).
The assembly of the active site in vitro can be achieved by using high, nonphysiological concentrations of Ni 2ϩ ions and bicarbonate as the source of CO 2 , which is needed for the carbamoylation of the active site lysine residue (7). At physiological concentrations of Ni 2ϩ and CO 2 , the participation of four accessory proteins is required (8 -12). These proteins (UreD, UreF, UreG, and UreE) are encoded by four genes, which are also present in the urease operon together with ureA, ureB, and ureC (12).
Many functional studies have been carried out on the urease accessory proteins from K. aerogenes. KaUreD (ϳ30 kDa) binds to apourease and appears to induce a conformational change required for the next steps of the activation process (13,14). KaUreF (ϳ25 kDa) binds the KaUreD-apourease complex and seems to facilitate carbamoylation of the nickel-bridging lysine residue and to prevent Ni 2ϩ binding to the noncarbamoylated apourease (15). Cross-linking experiments showed an interaction between KaUreD and the ␣ and ␤ subunits of apourease, whereas KaUreF was shown to interact with the ␤ subunit and to induce a conformational change capable of increasing the accessibility of the nickel ions and CO 2 to residues in the active site (16). The interaction of UreD with UreF and with the ␣-subunit of apourease was suggested by immunoprecipitation and two-hybrid studies carried out on these proteins from Proteus mirabilis (17). In H. pylori a similar experiment indicated the interaction between UreF and UreH, the latter corresponding to UreD in other bacteria (18). KaUreG (ϳ22 kDa) can form a quaternary complex with KaUreDF-apourease, suggesting that such large aggregate is the minimum competent species for the in vitro urease activation (7,19) and could be required for the process occurring in vivo (20). Finally, KaUreE (a homodimer of ϳ35 kDa) is thought to bind the KaUreDFGapourease complex, acting as a nickel transporter that delivers Ni 2ϩ to the active site of the enzyme (21)(22)(23)(24)(25).
Among the four accessory proteins, UreG plays an essential role in coupling cellular metabolism and bioenergetics to the assembly of urease. This protein contains a fully conserved P-loop motif, which is also present in many nucleotide-binding proteins and which is probably related to the in vivo GTP requirement for assembly of the urease active site (26). A direct relationship between UreG and GTP requirement is proven by the evidence that GTP is needed for activation of the Ka-UreDFG-apourease complex, although it has an inhibitory effect on the nickel reconstitution of the apourease and of the KaUreD-apourease and KaUreDF-apourease complexes (26). UreG is also involved in delivering CO 2 necessary for the carbamoylation of the nickel-bridging lysine; the curves that correlate the urease activation to different bicarbonate concentrations indicate a higher rate and level of enzymatic activation in the presence of UreDFG-apourease complex (26). In vitro, optimal levels of apourease activation require 0.5 mM GTP. Larger GTP concentrations lead to a decrease of urease activity, probably caused by the chelation of Ni 2ϩ by the nucleotide; this is consistent with the fact that elevated levels of Mg 2ϩ ions partially restore the activation (26). In the presence of UreE, this inhibitory effect is lost, and the urease activation occurs at significantly lower GTP concentrations (27). These observations indicate a correlation between GTP hydrolysis by UreG and Ni 2ϩ transfer to apourease by UreE. Additional evidence of the UreE-UreG interaction in vivo has been obtained by using two-hybrid systems and immunoprecipitation experiments in H. pylori (18). It has been proposed that UreG may induce GTP-dependent structural changes of the apourease, increasing the accessibility of both Ni 2ϩ and CO 2 to the developing active site. Alternatively, UreG may use GTP and CO 2 to synthesize carboxyphosphate, which could serve as an excellent CO 2 donor to the Lys residue (26).
The thorough functional studies described above have paved the way to an understanding of the mechanism of the urease accessory proteins and, in particular, of UreG in Ni 2ϩ trafficking and metabolic regulation. The next echelon in the comprehension of the urease chemistry implies the study of the interaction mechanisms between the accessory proteins and the enzyme at the molecular level, in order to understand how the urease active site is assembled. This goal cannot be achieved without detailed structural information on each accessory protein and its biochemical properties. The only crystal structures available for the urease chaperones are those of UreE from K. aerogenes (28) and B. pasteurii (29). The high degree of similarity between these two structures and between the structures of Ka-and Bp-urease are indications of a conserved molecular mechanism of urease activity, activation, and metalcenter building in different species (30). No structural detail is available for the other three chaperones. A study, published in 1997 (20), reported that KaUreG is monomeric in solution and that it does not, by itself, hydrolyze GTP or ATP. Another report in 2003 (31) and concerning HpUreG essentially confirmed these results.
This paper describes a thorough study performed on the recombinant UreG from B. pasteurii. In particular, the cloning, expression, and purification of the protein in its native and His-tagged forms are described, together with evidence confirming the identity of the isolated protein. The oligomerization state and the hydrodynamic properties of the protein in solution were examined by using size exclusion chromatography coupled with light scattering experiments, and the protein folding was checked using circular dichroism, mass spectrometry, and NMR. The metal-binding capability and enzymatic GTPase activity of BpUreG were established for the first time and discussed. Finally, threading (fold recognition) algorithms were applied to calculate a model for the protein structure that is consistent with its solution properties and enzymatic activity. The results represent a significant contribution to the understanding of the role of this metallochaperone in the urease active site assembly.

EXPERIMENTAL PROCEDURES
BpUreG Cloning-In accordance with the DNA sequence of the B. pasteurii urease operon available from GenBank TM (accession number AF361945), two 24-and 26-bp oligonucleotide primers were designed and synthesized to amplify the ureG gene by the PCR technique using the pUC19 plasmid containing the B. pasteurii ure operon (32) as template. The following forward and reverse primers introduced NdeI and BamHI restriction enzyme recognition sites, respectively (boldface type): 5Ј-CTAGGAGATTGTGCATATGAAAAC-3Ј; 5Ј-CAATATCGAG GGATCCAAACGGTATT-3Ј.
Taq polymerase for the PCR and dNTPs were from Display System Biotech. The oligonucleotide primers were synthesized in the Nucleic Acid Facility at the Pennsylvania State University (University Park, PA). The PCR product obtained by using these primers was digested by a combination of NdeI and BamHI restriction enzymes (New England Biolabs), purified by electrophoresis on a 1% (w/v) agarose gel, extracted, and precipitated. By using T4 DNA ligase (Promega), this DNA fragment was ligated at a 2-fold excess of insert to vector, into plasmid pET3a (Novagen), which had been digested with NdeI and BamHI, treated with alkaline phosphatase, and purified by electrophoresis. For the His-tagged BpUreG, the plasmid pET15b was used. Plasmid DNA was isolated from transformants of Escherichia coli strain DH5␣ (Invitrogen) by the rapid alkaline extraction method, as described (33), digested with appropriate restriction enzymes, and analyzed by agarose gel electrophoresis. The resulting pET3a::ureG and pET15b::ureG plasmids were purified using the StrataPrep™ Plasmid MiniPrep kit (Stratagene). The sequence of the cloned BpUreG gene was confirmed by DNA sequencing. The constructs for both plasmids were inserted by electroporation (Bio-Rad GenePulser II) into the E. coli BL21(DE3) expression host (Novagen) grown in shaking flasks at 37°C in a medium with the Luria-Bertani (LB) composition (Amersham Biosciences) or on agar (1.5%) plates with the same composition.
BpUreG Expression-Based on the T7 system (34), large scale expression of BpUreG and His-tagged BpUreG was achieved in 2.5-liter batches of minimum M9 liquid media (1 liter contained 6 g of Na 2 HPO 4 , 3 g of KH 2 PO 4 , 0.5 g of NaCl, 1.25 g of (NH 4 ) 2 SO 4 , 0.246 g of MgSO 4 ) supplemented with 4 g of glucose per liter of culture. The 15 N-enriched proteins were obtained using a medium containing ( 15 NH 4 ) 2 SO 4 . Transformed E. coli BL21(DE3) cells were grown at 37°C (28°C for the His-tagged protein) with vigorous stirring, until the A 600 reached 0.5-0.8. Expression was induced by addition of isopropyl ␤-thiogalactopyranoside (IPTG) to a final concentration of 0.5 mM. The cells were harvested 4 h after induction by centrifugation at 8,000 ϫ g for 10 min at 4°C. The cells were resuspended in 25 ml of 50 mM Tris-HCl buffer, pH 8, containing 5 mM EDTA and lysozyme (200 g/ml). After incuba-SCHEME 1 UreG Is a Natively Unfolded GTPase That Binds Zn 2ϩ tion at 30°C for 20 min, followed by the addition of DNase I (20 g ml Ϫ1 ) and additional incubation at 37°C for 20 min, the cells were disrupted by two passages through a French pressure cell (SLM, Aminco) at 20,000 pounds/square inch. The cell pellet was separated from the supernatant by centrifugation at 15,000 ϫ g for 15 min at 4°C.
BpUreG Purification-In the case of native BpUreG, the pellet was washed (resuspended in 25 ml of buffer using a mixer homogenizer and centrifuged at 15,000 ϫ g for 15 min at 4°C) three times with 50 mM Tris-HCl buffer, pH 8, containing 5 mM EDTA, 1 mM DTT, and 2% (w/v) Triton X-100 and three times with the same buffer without Triton X-100. The pellet was subsequently resuspended and incubated overnight at 4°C in 50 mM Tris-HCl buffer, pH 8, containing 1 mM DTT and 2 M urea. The soluble fraction, obtained after removal of the precipitated material by centrifugation (15,000 ϫ g, 15 min), was loaded onto a Q-Sepharose XK 26/10 column (Amersham Biosciences) that had been pre-equilibrated with 2 volumes of 50 mM Tris-HCl buffer, pH 8, containing 1 mM DTT and 2 M urea. The column was washed using a flow rate of 3 ml min Ϫ1 with the starting buffer until the base line was stable. The protein was eluted from the column with a 400-ml linear gradient of NaCl (0 -1 M). Fractions containing BpUreG were combined, diluted with the elution buffer to a protein concentration of 0.3 mg/ml, and dialyzed (5-kDa cut-off membrane) overnight at 4°C against 50 mM Tris-HCl buffer, pH 8. The resulting solution of BpUreG was concentrated by using 5-kDa cut-off membrane Amicon and Centricon ultrafiltration units (Millipore), to a final volume of 5 ml, and centrifuged (15 min at 14,000 ϫ g) to remove the precipitated material. The resulting solution was loaded onto a Superdex 75 XK 26/60 column conditioned with 50 mM Tris-HCl buffer, pH 8, containing 0.15 M NaCl and 1 mM DTT. BpUreG was eluted at a flow rate of 2 ml min Ϫ1 , and the purified protein, amounting to ϳ50 mg per liter of culture, was concentrated to 2.5 mg ml Ϫ1 and stored at Ϫ80°C.
In the case of His-tagged BpUreG, the supernatant after pellet separation was loaded onto a column containing 8 ml of the nickel-nitrilotriacetic acid Superflow affinity resin (Qiagen) pre-equilibrated with 40 ml of 50 mM Tris-HCl buffer, pH 8.0, containing 5 mM imidazole, washed with 30 ml of the same buffer containing 20 mM imidazole, and eluted using the same buffer containing 100 mM imidazole.
Protein purity, as well as the molecular mass of BpUreG in denaturing conditions, was estimated by SDS-PAGE according to the method of Laemmli (35), by using a Bio-Rad Mini-Protean II apparatus. Proteins were separated on 15% (w/v) acrylamide-bisacrylamide separating gels that were stained using either Coomassie Brilliant Blue R-250 or silver staining.
Protein concentration was measured by using a Jasco 7800 spectrophotometer and a value for the extinction coefficient (⑀ 280 ϭ 10,810 M Ϫ1 cm Ϫ1 ) calculated from the amino acid sequence using the ProtParam web site (au.expasy.org/tools/protparam.html). This value is in good agreement with that obtained by using the Bio-Rad assay that is based on the Bradford colorimetric method (36).
Protein Sequence Determination-All reagents, solvents and instruments were obtained from Applied Biosystems. N-terminal sequence analysis was performed in the gas-pulsed liquid phase using a model 476A protein sequencer with a micro-reaction chamber and an on-line high pressure liquid chromatography system for phenylthiohydantoin analysis. Absorbance was monitored at 269 nm. C-terminal sequence analysis was performed on a Procise 494C protein sequencer using C-terminal sequencing chemistry (37,38). Prior to this analysis, the sample was adsorbed on a Prosorb sample preparation cartridge and, after subsequent washes with MilliQ-filtered water, treated with 200 mM phenylisocyanate in acetonitrile under basic conditions (124 mM diisopropylethylamine/acetonitrile) in order to modify the ⑀-amino group of the lysine residues into stable phenylureas. The alkylated thiohydantoin (ATH)-amino acids were analyzed on-line using a thermostated (38°C) C18 reverse-phase column (2.1 ϫ 220 mm, 5 M). A linear gradient with a flow rate of 300 l min Ϫ1 was formed using a 140C microgradient system with the following solvents: solvent A, 35 mM sodium acetate buffer, 3.5% tetrahydrofuran/MQ water, and solvent B, 100% acetonitrile. The ATH-amino acid derivatives were monitored using a 785A absorbance detector set at 254 nm, and quantified relative to a 100 pmol of ATH-amino acid standard. The methyl naphthylthiohydantoin amino acid standards were obtained from the supplier.
Mass Spectrometry-All mass spectrometric analyses were performed on a quadrupole TOF mass spectrometer (Micromass), interfaced to a chip-based nano-ESI source (NanoMate100, Advion Biosciences). The mass spectra were processed using MassLynx version 3.1 software of Micromass. Before analysis, the buffer was changed to 50 mM NH 4 acetate, pH 6.5. The denatured protein (1 M) was measured in 50% acetonitrile, 0.1% formic acid. For native measurements, the pro-tein (5 M) was kept in 50 mM NH 4 acetate, pH 6.5. Acquisition conditions were a spraying voltage of 1.5 (denatured protein) or 1.7 kV (native protein), gas pressure of 0.3 pounds/square inch, and an acquisition time of 3 (denatured protein) or 10 min (native protein) across an m/z range of 500 -3000.
Hydrodynamic Properties-The molecular mass and hydrodynamic radius of the native protein were estimated by standard size exclusion chromatography. A small amount (100 l, 2.5 mg/ml) of the purified protein solution was applied to a Superdex-75 HR 10/30 FPLC column that had been equilibrated with 50 mM Tris-HCl, pH 8.0, containing 0.15 M NaCl, at a flow rate of 0.5 ml min Ϫ1 in order to estimate the apparent hydrodynamic volume of the purified protein. The column was calibrated using an Amersham Biosciences low molecular weight gel filtration calibration kit.
Absolute estimates of molecular mass and hydrodynamic radius of BpUreG were determined using a combination of size exclusion chromatography, multiple angle light scattering (MALS), and quasi-elastic light scattering (QELS). BpUreG (100 l, 2.5 mg/ml) in 20 mM Tris-HCl, pH 7.5, 150 mM NaCl was loaded onto an S-200 16/60 column (Amersham Biosciences), pre-equilibrated with the same buffer, and eluted at room temperature at a flow rate of 1 ml/min. The column was connected downstream to a multiangle laser light (690.0 nm) scattering DAWN EOS photometer (Wyatt Technology). Quasi-elastic (dynamic) light scattering data were collected at a 90°angle by using a Wyatt-QELS device. The concentration of the eluted protein was determined using a refractive index detector (Optilab DSP, Wyatt). Values of 0.185 for the refractive index increment (dn/dc) and 1.330 for the solvent refractive index were used. Molecular weights were determined from a Zimm plot. Data were analyzed using the Astra 4.90.07 software (Wyatt Technology), following the manufacturer's indications.
Circular Dichroism Spectroscopy-The CD spectra of BpUreG and its His-tagged analogue were measured at 20°C, using a Jasco 710 spectropolarimeter flushed with N 2 , and a cuvette with 0.01-cm path length. The buffer was 20 mM phosphate, pH 7.5, containing 0.15 M NaCl. The spectra were registered from 190 to 300 nm at 0.2-nm intervals. Ten spectra were accumulated at room temperature and averaged to achieve an appropriate signal-to-noise ratio. The spectrum of the buffer was subtracted. The secondary structure composition of BpUreG was evaluated with the tool available on the Dichroweb server of the Centre for Protein and Membrane Structure and Dynamics, www.cryst. bbk.ac.uk/cdweb/html/home.html (39) using the reference sets 4 and 7.
NMR Spectroscopy-NMR spectra of 15 N-enriched KaUreG and Histagged BpUreG were recorded at pH 8.0 and 298 K on a Bruker Avance 800 spectrometer operating at 800.13 MHz. The KaUreG spectrum was recorded using a 5-mm reverse detection probe on a 1-mM sample, whereas the spectrum of His-tagged BpUreG was obtained using a TXI cryoprobe on a 0.45-mM sample. The spectrum of BpUreG isolated from inclusion bodies was recorded on a 0.45-mm 15 N-enriched sample at pH 8.0 and 298 K using a Bruker DRX Avance 500 spectrometer operating at 500.13 MHz and equipped with a TXO cryoprobe. 1 H, 15 N-HSQC spectra were acquired using sensitivity improvement (40 -42) and consisted of 8 -48 scans, spectral windows of 11-16 ppm in the proton dimension, and 30 -40 ppm in the nitrogen dimension, with the carrier set at the water frequency and 118 ppm, respectively. Relaxation delays (including acquisition time) in the range 0.9 -1.2 s were employed. Matrices of 1024 ϫ 256 points or 2048 ϫ 128 points were acquired and transformed into 1024 ϫ 512 or 2048 ϫ 512 points.
Measurement of GTPase Activity-GTP hydrolyzing activity was measured using a colorimetric method. The reaction mixture, containing 20 mM Tris-HCl, pH 8.0, 0.075 M NaCl, 5 mM MgCl 2 , 2 mM GTP, and 5 M BpUreG, in the absence or presence of 25 M ZnSO 4 , was incubated at 37°C. Aliquots (90 l) were removed at different incubation times and added to 30 l of a 35% trichloroacetic acid/water solution. Phosphate concentration was determined by the malachite green assay (43).
Metal Binding Experiments-In all operations, care was taken to avoid exogenous metal contamination. Ni 2ϩ and Zn 2ϩ nitrate salts solutions were prepared starting from ICP 1000 ppm standard solutions (CPI International) diluted to 1 mM with buffer A, containing NaCl 0.15 M. Equal volumes of BpUreG and metal solutions were mixed in 1:1 ratio to yield a constant concentration of protein (40 and 20 M BpUreG for nickel and zinc, respectively) and an increasing concentration of metal ion. No precipitation was observed during the titration. The resulting mixtures were incubated for 1 h at 37°C and overnight at 4°C and then filtered by centrifugation using 0.5-ml Centricon filter (cut-off membrane 5 kDa). 400 l of the filtered solution was diluted to 8 ml with MilliQ water. Metal analysis was performed using a Spectro Ciros CCD ICP optical emission spectrometer (Spectro Analytical Instru-ments) in combination with a Lichte nebulizer and a peristaltic pump for sample introduction. The ICP emission spectrometer system was calibrated by serial dilutions of appropriate single and multielement standards (CPI International). The standardization curve was made by using standard solutions in the range 0 -500 M of nickel and zinc in 2.5 mM Tris-HCl, pH 8, and 7.5 mM NaCl with a linear fitting. An R f power of 1400 watts, a nebulizer gas flow of 0.8 liter min Ϫ1 , and a plasma gas flow of 14 liters min Ϫ1 were used. The sample uptake was set at 2 ml min Ϫ1 for 24 s and a wash time of 15 s at 4 ml min Ϫ1 plus 45 s at 2 ml min Ϫ1 for each sample. Quality control was established by evaluation of buffer containing standards. In order to estimate the total metal added to every protein sample, 200 l of every metal solution were mixed in a 1:1 ratio with the blank buffer, diluted to 8 ml with MilliQ water, and measured as the filtered samples. The 221.648 and 231.604 nm lines for nickel and the 202.548, 206.191, and 213.604 nm lines for zinc were used for analysis. The measured metal ion concentrations were corrected with the value obtained for the filtered solution of the protein incubated with the buffer blank, without metal ion. The experimental points were fitted using the MacCurveFit software, and the fit was optimized using a Quasi-Newton algorithm.
Calculation of BpUreG Molecular Structure-A similarity search of the protein sequences related to BpUreG was carried out using the program FASTA3 (44,45) applied to the SwissProt data base. Multiple alignment of all sequences was performed using ClustalW (www. ebi.ac.uk/clustalw) (46). The alignment was optimized using information deriving from secondary structure predictions provided by the program JPRED (www.compbio.dundee.ac.uk/ϳwww-jpred) (47). The PONDR VL-XT algorithm (48,49) for the prediction of disordered regions of the UreG sequences was accessed through the web site www.pondr.com/(Molecular Kinetics, Inc.).
The sequence of BpUreG was used to search for templates using the 3D-Jury predictor meta-server (50,51) available at the address bioinfo.pl/Meta/. Only the templates with a 3D-Jury score higher than 80% of the best score were selected. Multiple sequence alignment of the sequence of BpUreG with the selected templates was performed using ClustalW (46). Alignment optimization was carried out comparing information deriving from the secondary structure calculated for the templates using DSSP (52) with BpUreG secondary structure prediction provided by the program JPRED (47).
Model structures were calculated using the program MODELER 6.2 (53) with the model-default options. The program PROSA II (version 3.0, 1994) (54) was used for selecting the best models provided by MODELER and for protein structure analysis to test the coherency and validity of the model structures. The Z score reported in this work is derived through the standard "hide and seek" procedure of the program, by which the score is correlated to the difference in potential energy, calculated using mean field potentials, between the input structure and other randomly assigned folds for its amino acid sequence. A lower Z score corresponds to a more favorable potential energy associated with the structure under examination.
Structure validation was performed using PROCHECK (55) and WHATIF (56). The calculated final structure was deposited in the www.postgenomicnmr.net site. The molecular surface and the electrostatic color coding was generated by the program GRASP (57) using a probe radius of 1.4 Å. The electrostatic potential was calculated using a simple version of a Poisson-Boltzmann solver with the GRASP full charge set. All the histidine residues were considered neutral, and the N-and C-terminal residues were charged. Dielectric constants of 80 and 2 were used for the solvent and protein interior, respectively. The topological diagram was drawn using the program Topology of Protein Structure (TOPS) (58) available at www.tops.leeds.ac.uk.

RESULTS
BpUreG Cloning, Expression, and Purification-The sequence analysis of a 5.3-kbp DNA fragment isolated from B. pasteurii indicated the presence of four open reading frames identified as the genes expressing the urease accessory proteins UreE, UreF, UreG, and UreD (32). The ureG gene was cloned from B. pasteurii chromosomal DNA by PCR amplification. The gene was inserted between the NdeI and BamHI sites of a pET3a plasmid, and this construct was used to overproduce BpUreG in E. coli BL21(DE3) strain by induction with IPTG. This protocol produced an abundant polypeptide with an apparent molecular mass of 25 kDa that was absent from noninduced cells (Fig. 1A, lanes 1 and 2). Fractionation of cells into soluble and insoluble extracts showed that the overproduced protein product accumulated almost exclusively in the insoluble fraction (Fig. 1A, lanes 3 and 4). In order to obtain the protein in a soluble form, the insoluble BpUreG was treated with increasing amounts of urea from 2 to 8 M. These studies showed that 2 M urea was the optimum denaturing agent concentration needed to extract the maximum amount of BpUreG without associated protein degradation or re-aggregation (Fig. 1A, lane 5). The extract was purified by ion exchange chromatography in the presence of 2 M urea, and the isolated fractions were dialyzed overnight to remove the denaturing agent and to refold the protein. The soluble protein obtained was further purified by size exclusion chromatography (yield 50 mg per liter of culture), and its purity was checked by SDS-PAGE (Fig. 1A, lane 6).
The formation of inclusion bodies during heterologous overexpression of recombinant proteins is generally attributed to the high levels of protein production that result from constructs built using plasmids containing the strong T7 promoter. This was a problem encountered in our studies using the pET system for the overexpression of BpUreG. Lowering the growth temperature or decreasing the amount of added inducer failed to reduce the expression level and to increase the protein solubility. Alternative expression systems involving glutathione S-transferase-tagged protein were unsuccessful and yielded lower amounts of protein without solving the solubility problem. Therefore, we relied upon the pET system and a novel protocol for protein purification, which yielded large amounts of pure protein. It is generally believed that inclusion bodies contain proteins in a misfolded state, and several protocols are available to solubilize and purify proteins from inclusion bodies. Proteins are typically solubilized by large concentrations of denaturing agents such as urea or guanidinium chloride and are refolded during an extensive dialysis step. These conditions often cause problems in achieving an efficient and reliable folding in vitro because of the many re-folding pathways potentially available for the random-coiled protein and to the reaggregation processes occurring at intermediate denaturant concentrations. In the case of BpUreG, the amount of urea found to solubilize the inclusion bodies was relatively low (2 M) suggesting that the protein was only partially unfolded in the insoluble aggregate. The low urea concentration also prevents the complete unfolding of the protein, a state from which a native fold is difficult to attain.
Another approach to solve the solubility problem of BpUreG was attempted by cloning and expressing the His-tagged protein using the pET15b plasmid. This method, similar to the pET3a system, produced large amounts of protein in the inclusion bodies but allowed the purification of the protein from the supernatant in a single chromatographic step using a Ni 2ϩaffinity column. The yield of His-tagged BpUreG was significantly lower than in the case of the protein purified from the insoluble pellet (10 mg per liter), but this amount allowed us to compare the fold properties of the protein isolated from the soluble and insoluble fraction, using CD and NMR spectroscopy, as well as the activity assay.
Electrospray ionization, quadrupole TOF mass spectra of BpUreG in denaturing conditions (Fig. 1B) confirmed the high purity of the isolated protein and indicated a mass of 23,084.5 Da, in agreement with the theoretical mass (23,084.1 Da).
As expected from the gene sequence, Edman degradation of the N terminus of BpUreG yielded the sequence MKTIHL. The protein was also subjected to C-terminal amino acid sequence analysis, which provided the expected sequence ESK. In the case of the latter experiment, the sensitivity dropped rather rapidly, yielding only the last three residues, most likely because of the presence of consecutive "problematic amino acids" (59). However, this information, together with the mass of the recombinant protein, is sufficient to demonstrate that the recombinant-purified BpUreG protein is intact and unmodified.
Hydrodynamic Properties and Oligomerization of BpUreG-Gel filtration experiments performed on BpUreG revealed the presence of two peaks corresponding, according to their retention volumes, to hydrodynamic radii (R h ) of 2.8 (minor peak) and 3.5 nm (predominant form), and apparent molecular mass of 39 and 59 kDa. The gel filtration profile for the soluble His-tagged BpUreG analogue was very similar to that of BpUreG isolated from inclusion bodies, with a smaller amount of the minor peak. These masses are not easily related to multimeric forms of the BpUreG monomer (23,084 Da). These data, based on the calibration of the column with globular standard proteins, assuming a globular structure for BpUreG, are strongly dependent on the properties of the protein and the column, as well as on the possible interactions between the protein and the solid phase. Therefore, a better estimate of the molecular masses and hydrodynamic radii of the protein eluted in the two different peaks was obtained by using a combination of SEC, MALS, and QELS (60, 61). The advantage of such a system is that the molecular mass determination is based on the fundamental light scattering properties of the macromolecule. The results obtained are shown in Fig. 2. The elution profile and the MALS data are consistent with the presence of a predominant dimeric form of the protein in solution (Ն95%, mass ϭ 40.0 Ϯ 0.5 kDa), whereas the monomeric BpUreG represents only a minor fraction (Յ5%, mass ϭ 28.6 Ϯ 4.9 kDa).
The MALS data confirm the anomalous chromatographic behavior of the protein and suggest either a nonspherical (prolate or oblate) shape of the protein or the presence of nonspecific interactions between the protein and the column phase. Because the scattering of the applied light (wavelength 690 nm) by such small molecules is isotropic, no information on the protein size can be obtained from the static MALS. However, such information can be derived from dynamic light scattering (QELS) experiments performed simultaneously with the MALS measurements. The hydrodynamic radius (R h ) of the dimeric BpUreG, as determined by QELS, is 2.00 Ϯ 0.02 nm, whereas the value of R h for the monomer is larger (2.70 Ϯ 0.20 nm). The apparent inconsistency between the mass and the volume of the monomer and dimer of BpUreG can be resolved by considering that the monomer is largely unfolded. Considering the larger value of R h for the monomer, the retention volume for the monomeric form in the size exclusion column is expected to be smaller than that for the dimeric protein; the aberrant behavior of the monomer in the gel filtration experiment could be explained with unspecific interactions occurring between the unfolded monomer and the solid phase of the column. The mass spectrum of BpUreG under nondenaturing conditions revealed the presence of the monomer only, an indication that the dimerization does not involve the formation of covalent bonds (not shown). In the ESI-MS experiments, the injected sample must be maintained at low ionic strength, a condition that could destabilize the dimer, especially if hydrophobic forces are the main responsibility for the dimerization of the protein. Moreover, hydrophobic interactions are weak in gas phase, further causing the dimer to dissociate in the TOF stage. The mass spectrum also confirms that the monomer is largely unfolded, with the presence of populations featuring charge states (from 9ϩ to 29ϩ) reflecting high solvent accessibility of protonation sites.
Circular Dichroism and Secondary Structure of BpUreG-In order to evaluate the secondary structure composition of BpUreG in solution, the protein was analyzed by using CD spectroscopy. The obtained spectrum (Fig. 3) shows negative deflections with a minimum at ϳ208 nm and a pronounced shoulder at ϳ220 nm, as well as a maximum positive deflection at ϳ193 nm, typical for the presence of both ␣-helix and ␤-strand regions in the protein. The spectrum was quantitatively analyzed in the range 190 -240 nm, on a per amide basis calculated from protein content and sequence, using all the different fitting programs at the Dichroweb server (39) and all the possible reference sets. The best fit was selected on the basis of the normalized root mean square deviation (0.038) between the experimental and calculated data, obtained by using the variable selection method program CDSSTR (62) and the reference set n ϭ 4. From this analysis, a secondary structure composition of 15% ␣-helix, 29% ␤-strand, 26% turns, and 30% random coil was estimated for BpUreG. The CD spectrum of His-tagged BpUreG confirmed these data, yielding 18% ␣-helix, 25% ␤-strand, 29% turns, and 30% random coil.
NMR Spectroscopy of BpUreG-In order to monitor the conformational properties of the protein in solution, NMR spectroscopy was applied. The 1 H, 15 N-HSQC spectra of BpUreG (Fig. 4A), His-tagged BpUreG (Fig. 4B), and KaUreG (Fig. 4C) are very similar. They are characterized by poorly resolved resonances with little dispersion in the 1 H dimension of the backbone amides, many of them falling in the random coil region (7.6 -8.5 ppm chemical shift range). This observation indicates that large portions of the protein backbone experience exchange between multiple conformations and lack a well defined secondary structure. The signals of the side chain NH 2 groups of the 12 Asn residues present in the protein are also poorly resolved and give rise to two broad envelopes, centered at 112.5 ppm ( 15 N) and 7.4 -6.7 ppm ( 1 H). Lack of differentiation in these resonances suggests lack of specific interactions for the Asn side chains and therefore confirms the fluxional behavior of UreG in solution.
Measurement of GTPase Activity of BpUreG-UreG is involved in the hydrolysis of GTP concomitantly with the carbamoylation of the lysine residue in the urease active site (26). The GTPase enzymatic activity of BpUreG was checked and measured using a colorimetric method that determines the concentration of phosphate released by the hydrolysis of added GTP at various times (Fig. 5). BpUreG and His-tagged BpUreG show a significant and comparable level of GTPase activity. In the assay mixture, the concentration of substrate GTP is 3 orders of magnitude larger than the concentration of enzyme, a condition that allows the use of the time course data of the reaction for the derivation of the value of k cat ϭ 0.04 min Ϫ1 and 0.03 min Ϫ1 for wild-type BpUreG and His-tagged BpUreG, respectively.
Ni 2ϩ and Zn 2ϩ Binding to BpUreG-The isolated BpUreG did not contain bound metal ions, as established by ICP-ES metal analysis. Considering that UreG is involved in the building of the Ni 2ϩ -containing active site in the urease activation process (10,20,26) and UreE, the most known accessory protein in this system, features Ni 2ϩ and Zn 2ϩ binding properties (21)(22)(23)(24)(25), the affinity of BpUreG for Ni 2ϩ and Zn 2ϩ was quantitatively investigated. A fixed concentration of BpUreG was titrated with increasing concentrations of Zn 2ϩ and Ni 2ϩ ions, and the concentration of free ions was measured using ICP-ES.
The binding curves, obtained by reporting the amount of total metal bound (M b ) as a function of the total amount of metal added (M t ), showed saturation with a maximum of four Ni 2ϩ ions and two Zn 2ϩ ions bound to the BpUreG dimer in the presence of excess metal. Fig. 6 reports the amount of bound metal per dimer (M b /P tot ) as a function of the free ion per dimer at equilibrium (M f /P tot ). By assuming a single-site binding model (that is considering that the metal ions show identical affinities for the different sites, i.e. homogeneous binding) the experimental points can be fitted to a curve described by Equation 1, where P tot is the total concentration of BpUreG dimer, K D is the dissociation constant, and n is the number of binding sites. The best fit was obtained with n ϭ 4 for Ni 2ϩ and n ϭ 2 for Zn 2ϩ . The K D estimated for the binding is 360 Ϯ 30 M for Ni 2ϩ and 42 Ϯ 3 M for Zn 2ϩ , revealing a 10-fold higher affinity of BpUreG for Zn 2ϩ than for Ni 2ϩ . Calculation of a Structural Model for BpUreG-The deduced amino acid sequence of the cloned ureG gene confirmed the identity of the encoded protein as belonging to the UreG family. Multiple alignment of the sequence of BpUreG with related sequences found in a similarity search highlights the conservation of the predicted secondary structure elements and the overall sequence profile and motifs (Fig. 7). This family of proteins appears to be characterized by the presence of five long helices and seven short ␤-strands. Two additional helices (H4 and H5) and one strand (S4) are found in some sequences but are not fully conserved. These structural elements are separated by large portions of turns or coils of variable length. The regions predicted to be in a strand conformation are mainly characterized by the presence of hydrophobic residues, suggesting that they constitute the hydrophobic core of the protein. On the other hand, the helices are mostly amphipathic, suggesting their involvement in intermediate regions between the hydro-phobic core and the solvent-exposed surface of the protein. The similarity of the secondary structure elements is paralleled by a high degree of sequence identity among all UreG proteins (between 49 and 62%). Large portions of the sequences are fully conserved, and these regions mostly occur in the loops and coils rather than in the helices and strands. This suggests that these less structured regions are functionally most important, whereas the helices and strands are only needed to confer the necessary overall structure to the protein. The P-loop, needed for GTP binding, is found between strand S1 and helix H1 in the N-terminal region of the protein and consists of a conserved GPVGXGKT motif, where X is usually Ser or, rarely, Ala.
The program PONDR VL-XT (48,49,63) was used to calculate the tendency of each residue in BpUreG and its homologous proteins to be disordered (PONDR score above 0.5, residues underlined in Fig. 7). The program predicts that large portions of all UreG proteins (for example 44 residues in BpUreG), found in the central region of the sequence, are disordered. This type of prediction is consistent with the observation that the monomeric form of BpUreG is largely unfolded as determined by light scattering and that the dimeric form, the predominant oligomer in solution, also undergoes dynamic processes in large portions of the protein, as detected by NMR spectroscopy.
In order to calculate a model structure for the fully folded BpUreG, the first choice involved the use of protein templates having a known structure and high homology with the target protein (homology building). However, a data base search for homologous proteins of known structure using FASTA3 (44,45) and PSI-BLAST (64, 65) did not result in any hit with a sequence identity higher than 30%, preventing the use of standard homology modeling protocols for structure prediction (66). Therefore, algorithms based on fold recognition were attempted. The 3D-Jury approach was singled out for its performance in structure prediction in the most recent CASP5 experiment (67), and the use of this protocol resulted in five template structures with a score higher than 81.3 (i.e. higher than 80% of the best), as reported in Table I. The PDB 1FFH structure has been refined (68) to a higher resolution (1.10 Å) and the PDB code 1LS1 was used in its place. All the found templates fall into the c.37.1 SCOP class (69), indicating that BpUreG is similar to proteins that contain a P-loop and are involved in GTP-based metabolic processes.
The sequence alignment of these proteins with the sequence of BpUreG, optimized using information derived from secondary structure predictions, was used as input to obtain a first set of model structures, which were then analyzed to identify local fold problems using PROSA. Whenever possible, these problems were corrected by modifying the alignment in the interested regions and building new models, following the same procedure used recently by us to model a set of UreE proteins (30). During the optimization of the structural model, misfolding problems were encountered in the region between residues 40 and 80, with the formation of a knotted loop. This is in line with the PONDR prediction that this region is disordered. In order to avoid this artifact, two template structures (PDB codes 1EGA and 1J8Y) had to be removed from the templates ensemble. 1J8Y is a GTPase domain found in a signal recognition particle, and its structure is very similar to 1LS1 (sequence identity, 43%; backbone root mean square deviation,1.26 Å). The use of this structure as template introduces redundancy in the modeling, somehow hampering the calculation of a good model in the cited region of BpUreG. 1EGA was excluded because of its low resolution (2.40 Å) and for the absence of the helix predicted between residues 45 and 51 of BpUreG. The final multiple sequence alignment of BpUreG and the used structural templates is reported in Fig. 8.
The BpUreG model structure obtained from this procedure features a high percentage of residues in the core and allowed region of the Ramachandran plot (88.8 and 8.7% respectively), with only few residues localized in the generously allowed region (2.5%), and no residue found in the disallowed region of the diagram. The low PROSA Z score (Ϫ6.56) also confirms the good quality of the model.
The fold of the BpUreG monomer is characterized by a central open ␤-barrel, formed by seven parallel and two antiparallel strands, surrounded by six ␣-helices connected with loops (Fig. 8). The 36% of the modeled residues of BpUreG are involved in ␣-helices, and the 22% are part of the central ␤-strands, whereas the remaining 42% comprises turns and coils. The P-loop is located between strand S1 and helix H1 and is found on one side of a deep pocket defined, on the other side, by the loop located between strand S5 and helix H4. Fig. 8 also report the positions of the so-called Switch I (between strand S2 and helix H2) and Switch II (between strand S5 and helix H4) regions, found to be important for the binding of the GTP-Mg 2ϩ adduct in G-proteins (70). The binding of Mg 2ϩ is generally needed for GTPase function, as it is involved in the GTP phosphate chain binding to the protein at typical consensus sequence ([DE]XXG) (70). The topological diagram for the model of BpUreG is also shown in Fig. 8. The C-terminal portion of BpUreG, comprising residues 189 -211, was not modeled due to the absence of a structural template for this region. This peptide sequence is predicted to exist in a ␣-helical conformation (H7 in Fig. 7), whose topology and orientation with respect to the rest of the protein could not be determined. The presence in the model of strand S9 at the C terminus, in contrast to the predicted presence of an helix (H7), may be explained by an artifact because of a termination effect of the modeling process.  (46), optimized by considering the prediction of secondary structure performed using JPRED (47). The predicted secondary structural elements are highlighted in yellow (helix) and turquoise (strand). The P-loop motif is colored red, and the residues putatively involved in Zn 2ϩ binding are in green. Fully conserved residues are indicated by *, and conservative substitution are marked with :. Residues predicted to be disordered by the program PONDR VL-XT (48,49,63)   In order to gain more information on BpUreG model structure characteristics, the solid surface representation of the electrostatic potential was calculated (Fig. 9). The molecular surface of the model structure is characterized by large patches of negative charge, with the exception of a large neutral zone situated in the region of helixes H4, H5, and H6. The large pocket observed in the C-terminal region is probably due to the absence, in the model, of the C-terminal helix, which comprises the last 24 residues. This suggests that such helix could fill this pocket in a folded form of the protein.
The deep pocket found near the P-loop, and identified as the putative GTP-binding site, features a negatively charged surface. Consistently with other GTPases co-crystallized with GTP analogues (see Refs. 71-75 for some recent studies), the pocket also contains the Mg 2ϩ ion-binding site, localized in the proximity of the Thr-17 and Thr-42 residues, conserved among UreGs (Fig. 7) and template structures (Fig. 8). DISCUSSION The thorough functional information collected for UreG in the past few years and by establishing this protein as an essential chaperone for the Ni 2ϩ active site assembly of urease have been complemented by the present study on the biochemical and structural properties of this protein in solution, using the recombinant UreG from B. pasteurii. The large amount of protein required for this type of characterization was obtained by cloning and expressing the protein and by establishing a protocol for protein purification.
By using a combination of SEC and MALS (60,61), BpUreG was shown to exist in solution as a dimer. In the past, hydrodynamic studies on KaUreG reported the presence of only the monomer in solution (20), whereas for HpUreG the molecular mass under native conditions was not determined (31). In the case of KaUreG only chromatographic experiments were used to establish its molecular aggregation form in solution. This approach is partially uncorrected and subject to errors, because it is strictly dependent on the shape of the protein and on its interaction properties with the solid chromatographic phase. This possibility was proven for BpUreG, shown to produce an aberrant chromatographic profile that could have lead to wrong conclusions were it not for the use of light scattering techniques, which yielded an incontrovertible value for the molecular mass of the protein in solution. The aggregation state of BpUreG is similar to that found for H. pylori HypB, a GTPase involved in the activation of Ni 2ϩ -dependent hydrogenase and reported to exist as a mixture of the monomeric and dimeric forms (76). The hydrodynamic radii measured by dynamic light scattering (QELS) for the dimer and the monomer BpUreG with the G domain of the signal recognition protein Ffh from Thermus aquaticus (PDB code 1LS1), signal recognition particle receptor from E. coli (PDB code 1FTS), and hypothetical protein Yjia from E. coli (PDB code 1NIJ) as obtained from ClustalW and optimized using JPRED. The predicted secondary structure elements are highlighted in yellow (helices) and turquoise (sheets). The residues involved in the P-loop are shown in red boldface. The residues in italics were not modeled because of the absence of structural data for the homologous proteins. The fully conserved residues are indicated with an *. Bottom panels report BpUreG model structure shown as "schematic" (left panel) and as topological diagram (right panel). The secondary structure elements range from deep blue in the proximity of N-terminal to red at the C terminus. Bottom left panel was made with MOLSCRIPT and RASTER3D (94,95), and the bottom right panel was made with TOPS (58).
latter is largely unfolded, whereas the dimer is present with a compact behavior, therefore representing the actual functional form of the protein.
In solution, BpUreG showed a well defined secondary structure, with both ␣-helices (15%) and ␤-strand regions (29%), as determined using CD spectroscopy (Fig. 3). These data do not contain any information regarding the tertiary structure of BpUreG, which has therefore been evaluated using NMR spectroscopy. 1 H, 15 N-HSQC NMR spectra revealed that BpUreG does not possess a rigid tertiary structure but exists in solution in fast equilibrium among different conformations (Fig. 4) and therefore contains large portions of unfolded backbone. The intrinsically unfolded state of BpUreG in solution is not an artifact due to the purification method, which involved the use of a small amount of urea to solubilize the protein from inclusion bodies. This is proven by the essentially identical spectroscopic and hydrodynamic properties of the His-tagged form of BpUreG, purified in a single step from the soluble cellular extract.
The large similarity of the NMR spectral properties of Ka-UreG and BpUreG suggests that the presence of large portions of the protein backbone undergoing conformational changes, and therefore causing a substantial intrinsic protein unfolding, is a general feature for all UreG chaperones. This evidence indicates that UreG belongs to the ever-growing class of intrinsically unstructured proteins. These are natively unfolded polypeptides that undergo disorder-order transitions among the random coil, pre-molten globule, molten globule, or fully folded states during or prior to their biological function (63,77,78). The observed conformational plasticity of UreG proteins is consistent with their role as chaperones with GTPase activity, assuming a fully active conformation only in the presence of a preformed complex with other urease accessory proteins, as experimentally proven (20). This characteristic could be related to the functional role of UreG in vivo, because it could permit us to minimize the unwanted hydrolysis of GTP unless the protein is ready to perform its role together with its partner chaperones in the preformed UreDFG-apourease complex. The high homology found in the predicted unstructured parts of the UreG sequences could well be related to this need, indicating the evolutionary functional importance of these loop regions, whereas the ␣ and ␤ elements perform a structural role. Large unstructured portions were predicted to be present in the central part of all UreG sequences found in the data base search (Fig. 7). This characteristic could explain the fluxional behavior detected by NMR spectroscopy for both BpUreG and KaUreG, confirming that this characteristic is a general feature of this class of proteins.
The structure of BpUreG in the fully folded state has been modeled by using a fold recognition procedure. Considering that BpUreG is an intrinsically unstructured protein, we can presume that this model predicts the structure of the protein when other cofactors or protein partners force it to assume its functional conformation. The structural prediction shows a fold typical for a GTPase protein (Figs. 8 and 9) with an identifiable negatively charged P-loop region, likely capable of containing a GTP molecule, and the Switch I and Switch II regions, putatively involved in the Mg 2ϩ binding and generally required for the GTPase function. The overall fold comprises an internal open ␤-barrel surrounded by ␣-helices. The hydrophobic and amphipathic composition of the sequences, predicted respectively as ␤-strands and ␣-helices in the multiple sequence alignment (Fig. 7), confirms this architecture. The secondary structure composition derived from the model is calculated as 36% for the helix and 22% for ␤-strands, whereas the remaining 42% is in coils or turns. The partial contrast between this prediction and the calculation of the secondary structure from the CD spectrum (15% of ␣-helices and 29% of ␤-strands) can be explained considering that the contradiction involves mostly the more solvent-exposed ␣-helices, likely more subject to conformational fluctuations, rather than the ␤-strands, situated in the internal and protected hydrophobic protein core.
The observation of the presence, in the UreG sequences, of a fully conserved P-loop motif, has led us to consider the possible GTPase activity for this protein. A study, published in 1997 (20), reported that KaUreG does not, by itself, hydrolyze GTP or ATP. Indeed, no native or added GTP or ATP was found to be associated with isolated KaUreG, and no interaction of Ka-UreG with ATP-or GTP-linked resins was observed. Similar results were obtained for HpUreG, which showed negligible GTPase activity for the isolated protein (31). The present study demonstrates that BpUreG (and its His-tagged analogue) features a clear GTPase activity, even if low (k cat ϭ 0.04 min Ϫ1 ) compared with other GTPases (79) (Fig. 5). This activity is lower than (but comparable with) the one showed by HypB from B. japonicum (k cat ϭ 0.18 min Ϫ1 ) (80) and from E. coli (k cat ϭ 0.17 min Ϫ1 ) (81). These results indicate that BpUreG, although present in solution as a fluxional, partially unfolded molecule, displays a level of enzymatic activity indicating that a significant fraction of UreG molecules is in the correct fold for catalysis. Alternatively, the fold around the catalytic site could well be correct, and the registered activity intrinsic for this protein, with the unfolded conformation possibly involving a different protein region, as the one indicated by PONDR prediction.
BpUreG has been demonstrated to bind two Zn 2ϩ ions for dimer, whereas the affinity is 10-fold lower for Ni 2ϩ ions (Fig.  6). The analysis of the structural model of BpUreG reveals the presence of a putative metal-binding site, rich in residues com- monly found in zinc-binding protein sites and could reasonably be proposed as implicated in the binding of one Zn 2ϩ per monomer. The fully conserved residues likely involved in the metal binding (Glu-64, Cys-68, and His-70) would fall well within the predicted disordered region of UreG (Fig. 7). In this regard, it is interesting to show that this putative metal-binding site is in a region (residues 40-80 in BpUreG numeration) that has been very difficult to model, as described above, probably because of its natively unfolded trait. Therefore, we could speculate that the binding of the Zn 2ϩ ion could induce a conformational change and contribute to the stabilization of the protein backbone in this region. In this case, the metal ion would assume a structural role, stabilizing a certain protein conformation upon binding, rather than a role in the catalytic activity of the protein. Indeed, no change in GTPase activity was observed in the presence of a 5-fold excess Zn 2ϩ ion in the assay mixture.
The specific metal-binding capability of UreG has never been observed before and could be related to its direct role in the assembly of the urease active site (10,20,26). HypB, the counterpart of UreG in the [Ni,Fe]-hydrogenase system, has often shown a nickel-sequestering ability, probably due to the presence of an N-terminal His tag, as in the case of Rhizobium leguminosarum (82) and Bradyrhizobium japonicum (80), whereas HpHypB and the HypB from E. coli, lacking the Nterminal His tag, do not bind nickel (76,81). Our study revealed that BpUreG is able to bind zinc even in the absence of a His tag. In this regard, it is important to consider that UreE, the best characterized accessory protein in this system, is able to bind both Ni 2ϩ and Zn 2ϩ with comparable affinity (22,24). The in vivo effect of zinc on the urease activation is unknown. However, zinc cannot substitute for nickel during urease biosynthesis in the absence of nickel ions (6). A role for HybF, a zinc-binding protein analogue to HypA, in the insertion of nickel in [Ni,Fe]-hydrogenase was recently established (83). The results presented here strongly suggest that metal ions other than nickel play an important role in the activation of urease, in some sort of co-metabolic metal trafficking crossroad.
The high sequence homology among UreG proteins is paralleled by a functional conservation through different species that was demonstrated in vivo by complementing a Ka-ureasedefective mutant with UreG from potatoes, expressed in E. coli (84). UreG has counterparts in the activation of urease from different organisms as well as in the assembly of other nickelcontaining enzymes. Activation of CO dehydrogenase in Rhodospirillum rubrum requires the activity of CooC, a nucleotidebinding protein related in sequence to UreG (85,86). UreG is also homologous to HypB, a GTP-binding accessory protein involved in the GTP-dependent activation of nickel-containing hydrogenase and urease in many organisms; in H. pylori inactivation of hypB leads to a phenotype that is both hydrogenaseand urease-negative (31,87). Eu3, the UreG analogue in Arabidopsis thaliana and soybean (88), has a high degree of sequence similarity to both UreG and HypB, and also exhibits a putative nucleotide-binding site. Given the observed sequence similarities among all UreG proteins from different organisms, as well as the functional parallelism between UreG and the related proteins in other nickel-containing systems, it is reasonable to presume that the structural and functional insights obtained in the present study on BpUreG can be extended to different systems and/or different organisms and that the results described here contribute to a general model for Ni 2ϩ incorporation into metalloproteins.