The N-terminal Arm of the Helicobacter pylori Ni2+-dependent Transcription Factor NikR Is Required for Specific DNA Binding*

The Ni2+-dependent transcription factor NikR is widespread among microbes. The two experimentally characterized NikR orthologs, from Helicobacter pylori and Escherichia coli, display vastly different regulatory capabilities in response to increased intracellular Ni2+. Here, we demonstrate that the nine-residue N-terminal arm present in H. pylori NikR plays a critical role in the expanded regulatory capabilities of this NikR family member. Specifically, the N-terminal arm is required to inhibit NikR binding to low affinity and nonspecific DNA sequences and is also linked to a cation requirement for NikR binding to the nixA promoter. Site-directed mutagenesis and arm-truncation variants of NikR indicate that two residues, Asp-7 and Asp-8, are linked to the cation requirement for binding. Pro-4 and Lys-6 are required for maximal DNA binding affinity of the full-length protein to both the nixA and ureA promoters. The N-terminal arm is highly variable among NikR family members, and these results suggest that it is an adaptable structural feature that can tune the regulatory capabilities of NikR to the nickel physiology of the microbe in which it is found.

Nickel is an essential cofactor in several metalloenzymes (1,2) that are expressed primarily in microorganisms. Many microbes are capable of expressing at least one Ni 2ϩ -enzyme; however no microbial genome encodes more than four known Ni 2ϩ -enzymes. The cellular Ni 2ϩ requirement is often directly proportional to Ni 2ϩ -enzyme expression levels. Facultative anaerobes, such as Escherichia coli, require Ni-Fe hydrogenases for growth under specific anaerobic conditions (3)(4)(5). The induction of hydrogenase expression results in a concomitant increase in Ni 2ϩ transporter expression and intracellular Ni 2ϩ levels (6,7). In some cases microbes maintain high Ni 2ϩ -enzyme levels in the apo form and only increase intracellular Ni 2ϩ under conditions in which enzyme activity is required. For example, Helicobacter pylori urease can compose up to 10% of the total protein of cells grown at neutral pH (8), where the acid buffering activity of urease is not necessary for cell viability (9,10), and only a small proportion of Ni 2ϩ -bound urease exists (11,12). When H. pylori is grown in acidic pH conditions, urease activity is required (8,(13)(14)(15), and the levels of intracellular Ni 2ϩ (16) and active Ni 2ϩ -bound urease (14,17,18) significantly increase. In both examples the coordinated regulation of Ni 2ϩ import with the nickel requirements of each organism is critical for optimizing Ni 2ϩ -enzyme activity to specific growth conditions while also preventing the accumulation of excess nickel ions, which are potentially toxic.
NikR is a member of the ribbon-helix-helix (RHH) family of DNA-binding proteins (19,32,33). DNA recognition by this family of proteins occurs through specific contacts with DNA bases via the surface residues of the antiparallel ␤-sheet (ribbon) at the N terminus of this domain (Fig. 1a) (33). Important DNA contacts that affect DNA binding affinity and specificity have also been identified in residues that precede the ␤-sheet (33)(34)(35)(36), although the sequence and length of these N-terminal extensions are variable. The NikR RHH domain is linked to a C-terminal domain that is homologous to an ACT ligand binding domain and contains a high affinity Ni 2ϩ -binding site (32,37). This domain is tetrameric (38) and results in a dimer of dimers quaternary structure for NikR. These domains and their topological arrangement are unique compared with other known metal-responsive transcriptional regulators (39).
Here, we show that the nine-residue N-terminal arm of HpNikR ( Fig. 1), not visible in the recent crystal structures of HpNikR (45), plays a critical role in NikR DNA binding. The arm inhibited low affinity and nonspecific NikR-DNA interactions. Additionally, removal of the arm relieved a cation requirement for DNA binding specifically at the nixA promoter, providing evidence that the arm plays distinct structural roles at different promoters. Mutagenesis of individual arm residues identified amino acids responsible for the P nixA cation requirement as well as those necessary for high affinity DNA binding. These results suggest that the N-terminal arm of HpNikR is critical for the ability of this transcription factor to recognize degenerate DNA-binding sites and, thus, regulate many genes to integrate the complex Ni 2ϩ physiology of this organism in response to changes in intracellular Ni 2ϩ levels. The N-terminal arms of different NikR family members vary widely in sequence and length, suggesting that this structural feature may be important for evolving regulatory specificity in accordance with cell physiology.

EXPERIMENTAL PROCEDURES
HpNikR, Cloning and Mutagenesis-H. pylori strain 26695 nikR (HP1338) was PCR-amplified from genomic DNA using the primers PC121 and PC122 (Table 1; Integrated DNA Technologies, Coralville, IA) and cloned into pET22-b using the NdeI and XhoI restriction sites (Novagen, Madison, WI) to create pEB116. nt9-and nt5-HpNikR were created using primers EB058 or EB190, respectively, and PC122 to amplify a 5Ј-truncated nikR from pEB116. The resulting products were digested with NdeI and XhoI and ligated into pET22-b digested with the same enzymes to create pEB149 (nt9-HpNikR) and pEB202 (nt5-HpNikR). Site-directed mutagenesis of individual HpNikR residues (described under "Results") was carried out using the QuikChange site-directed mutagenesis protocol (Stratagene, La Jolla, CA) using complementary oligonucleotides with the mutated codon and Pfu DNA polymerase. The DNA sequence of each construct was verified by sequencing (SeqWright, Houston, TX).
Protein Purification and Expression-Native H. pylori NikR and variants were expressed and purified as described previously for E. coli NikR (21,38) except that gel filtration was used as a second purification step instead of ion exchange. Protein concentration was determined in 6 M guanidine hydrochloride using ⑀ 276 ϭ 9895 M Ϫ1 cm Ϫ1 , as predicted by primary sequence analysis. To remove Ni 2ϩ from purified protein, the nickelnitrilotriacetic acid eluate was incubated with 50 mM EDTA for 48 h at 4°C followed by gel filtration (the second purification step). The removal of Ni 2ϩ ions was confirmed using UV-visible spectroscopy at 302 nm.
Promoter Fragments, Cloning and Labeling-DNA fragments for promoter regions were amplified by PCR using the oligonucleotide pairs described in Table 1. A subset of fragments (Table  1) was cloned into pBluescriptII SK (Stratagene) using standard molecular biology techniques. The cloned promoter sequences were confirmed by DNA sequencing.
Promoter fragments for DNA binding assays were generated by PCR as follows. 0.5 M forward (5Ј) primers (listed first in Table 1) for the nixA, ureA, fur, nikR, hpn, and rpoD fragments were 5Ј-end-labeled with [␥-32 P]ATP (GE Biosciences) and T4 polynucleotide kinase (New England Biolabs, Beverly, MA) in a The H. pylori NikR N-terminal Arm Modulates DNA Binding total volume of 40 l. Excess [␥-32 P]ATP was removed by desalting, and the purified primers were used in a PCR reaction with the corresponding reverse primers (listed second in Table  1) using plasmids pEB106, pEB131, or pEB104 as templates for the nixA, ureA, and nikR promoters or H. pylori 26695 genomic DNA for the fur, hpn, and rpoD fragments. The 143-bp fragment of the H. pylori rpoD gene had comparable length and GC content to the other fragments and was used as a negative control. The resulting labeled fragments were purified using a Qiagen PCR purification kit (Qiagen, Valencia, CA). DNA Binding Assays-DNase I footprinting was performed as described previously (19,38) in a binding buffer containing 10 mM Tris-Cl (pH 8.0), 100 mM KCl, 3 mM MgCl 2 , 10 g/ml oxidized E. coli thioredoxin (used to prevent nonspecific NikR accumulation in the gel wells) (46), and 4 ng/l salmon sperm DNA (Fisher). Labeled DNA fragments were incubated with protein at 22°C for 1 h before DNase I (Sigma) addition (final concentration 300 ng/ml). NiCl 2 was added to the binding reactions as described under "Results" and Figs. 2-6.
Hydroxyl radical footprinting was performed as described previously (47) with the following modifications. Binding reactions were performed in 50 l of DNase I footprinting buffer with 10 mM MgCl 2 . Fe(II)-EDTA (2 mM stock) was added to a final concentration of 167 M followed by the addition of 833 M sodium ascorbate (20 mM stock) and 0.05% H 2 O 2 (30% v/v stock). The reaction was carried out for 1 min at 22°C and was quenched with 10 l of 0.1 M thiourea and 1 l of 0.5 M EDTA (pH 8.0). Dimethyl sulfate (DMS) protection experiments were carried out as previously described (48), except binding reactions were in a total volume of 100 l, and the buffer contained 10 mM sodium cacodylate (pH 8.0) instead of Tris-Cl.
Electrophoretic mobility shift assays were performed using 7% polyacrylamide gels and electrophoresis buffer containing 50 mM Tris (pH 8.8), 25 mM boric acid, and NiCl 2 , MgCl 2 , NaCl, KCl, NiSO 4 , or MgSO 4 as described in the under "Results." The binding buffer was identical to that used for DNase I footprinting. The same end-labeled DNA fragments as those used for footprinting were incubated with NikR or mutant proteins at 22°C for 30 min, and 20 l of the 50 l total volume was loaded directly onto a running gel (120 V). Apparent affinities measured by mobility shift assays were calculated from binding curves determined by the ratio of bound (all shifted species) versus free counts as quantified using a GE Healthcare Storm 840 PhosphorImager and ImageQuant Version 5.2 software. Apparent affinities measured by DNase I footprinting were calculated from binding curves determined by the ratio of the protected DNA region normalized to a region of DNA not protected from the same lane versus the same ratio from identical regions of a protein-free lane on the same gel. The data were fit using Micromath Scientist Version 2.01 and the equation y ϭ 1/(1 ϩ (K d /x) n ), where y is the fraction of DNA bound (ratios described above), K d is the protein concentration required for half-maximal DNA binding, x is the protein concentration, and n is the Hill coefficient. All reported affinities are the average of at least two independent experiments using a dilution series of at least 15 protein concentrations. The reported error is the S.D. between the calculated affinities of at least two independent experiments.
UV-visible and CD Spectroscopy-UV-visible spectra were collected on a Shimadzu UV-2401PC spectrophotometer using a 100-l sample volume. CD spectra were collected on a Jasco J-715 spectrapolarimeter using a 900-l sample volume in a cylindrical cuvette with a 1-cm path length. All spectra were collected at 22°C.

RESULTS
Before examining the role of the N-terminal arm, we first tested the role of conserved Ni 2ϩ and DNA binding residues in HpNikR to verify their importance in protein function. Gel mobility shift assays demonstrated that mutations in the surface ␤-sheet residues (R12M or R12A, S14A, S16A) resulted in proteins that were unable to bind DNA (supplemental Fig. 1), confirming the expected essential requirement of these residues in DNA recognition by HpNikR (19,49). Mutation of high affinity Ni 2ϩ -binding site residues (H99A or C107A) completely abrogated Ni 2ϩ -dependent DNA binding (data not shown), demonstrating that HpNikR is activated by metal binding in a similar fashion as EcNikR (41).
The N-terminal Arm of HpNikR Is Not Required for Binding to P nixA or P ureA -The well characterized RHH proteins Arc, Mnt, and MetJ have N-terminal extensions immediately adjacent to the RHH motif that play important roles in DNA recognition (34 -36). HpNikR has an extension of nine amino acids at its N terminus (Fig. 1b). To examine the role of this arm in DNA binding, a truncation mutant of HpNikR (nt9-HpNikR) lacking  (52). The locations of the nt9 arms are indicated by arrows. b, N-terminal amino acid sequence alignment of Hp-and EcNikR. The additional nine residues of HpNikR are referred to as the N-terminal arm (nt9). The gray-shaded residues are located in the ␤-sheet motif and are responsible for making specific DNA contacts in other RHH proteins. The relative location of the high affinity Ni 2ϩ -binding sites in each sequence are indicated as dark gray boxes.

The H. pylori NikR N-terminal Arm Modulates DNA Binding
the first nine residues was constructed. This mutant also contained an Ile to Met substitution at the new N terminus to ensure protein translation. UV-visible spectroscopy indicated that full-length HpNikR and nt9-HpNikR bound Ni 2ϩ upon elution from a nickel-nitrilotriacetic acid column (data not shown). Nickel ions were removed from the protein as described under "Experimental Procedures." The stability of nt9-HpNikR in the presence or absence of stoichiometric Ni 2ϩ was similar to full-length HpNikR (supplemental Fig. 2), suggesting that loss of the arm does not affect the overall secondary structure of the protein.
DNA binding to the nixA and ureA promoters (P nixA and P ureA ) was measured for nt9-and full-length HpNikR (at 1:1 Ni 2ϩ :NikR) using different footprinting techniques in the presence or absence of excess NiCl 2 (50 M). DNase I, hydroxyl radical (Fe-EDTA), and DMS protection of P nixA and P ureA by nt9-HpNikR was identical to that observed for full-length HpNikR ( Fig. 2; data not shown for P nixA bottom strand or for P ureA ), suggesting that the N-terminal arm does not directly interact with either DNA sequence.
DNase I footprinting indicated that both full-length and nt9-HpNikR protected a 36-bp region of P nixA (Ϫ14 to ϩ22 relative to the start of transcription) (28) and a 38-bp region of P ureA (Ϫ91 to Ϫ54 relative to the start of transcription), consistent with published results for fulllength NikR (P nixA , Ϫ13 to ϩ21 (28); P ureA , Ϫ91 to Ϫ56 (28), Ϫ86 to Ϫ53 (26), and Ϫ87 to Ϫ56 (27)) ( Fig.  2d). Hydroxyl radical footprinting revealed that both proteins protect four regions of four bases each that are separated by four to six bases at each promoter, demonstrating that full-length and nt9-HpNikR bind to one face of P nixA and P ureA that spans approximately two turns of the double helix. DMS protection at positions Ϫ4 of the template strand and ϩ11 of the non-template strand of P nixA and positions Ϫ81 and Ϫ79 of the template strand of P ureA demonstrates that base-specific contacts made by HpNikR include these guanines. Furthermore, the DMS protection helped to define the likely half-sites recognized by HpNikR at P nixA and P ureA (boxed sequences in Fig. 2d).
The footprinting analyses of nt9-HpNikR suggest that the N-terminal arm does not make direct contacts with P nixA and P ureA . The contribution of the N-terminal arm to DNA binding affinity was tested using DNase I footprinting titrations with full-length HpNikR or nt9-HpNikR (1:1 Ni 2ϩ :NikR). The affinities of full-length HpNikR for P nixA and P ureA were 49 and 45 nM, respectively ( Fig. 3; Table 2). The affinity of NikR for P ureA is similar to previous reports (51 nM (26) and 57 nM (31)). nt9-HpNikR bound to both P nixA and P ureA with affinities of ϳ5 nM (Fig. 3, Table 2), which are 10-fold tighter than full-length HpNikR, suggesting that the N-terminal arm may actually decrease DNA binding affinity.
The N-terminal Arm Inhibits Nonspecific DNA Binding-The importance of the N-terminal arm in DNA binding was further tested using mobility shift assays (containing 50 M NiCl 2 ; Fig. 4) with promoters that have either been shown (nixA, ureA, fur, and nikR) (25)(26)(27)(28) or predicted (hpn) (25) to be directly bound by HpNikR. Full-length HpNikR displayed a hierarchy of DNA binding affinities to these fragments, exhibiting the highest and equal affinity for P nixA and P ureA and progressively weaker affinities for P fur , P nikR , and P hpn ( Table 2). No binding was detected to a 143-bp internal fragment of H. pylori rpoD that was used as a negative control. nt9-HpNikR bound P nixA , P ureA , and P fur with affinities similar to full-length HpNikR (Fig. 4, Table 2). Interestingly, nt9-HpNikR displayed significantly increased affinity for P nikR and P hpn and formed distinct shifted complexes with each fragment, in contrast to the more diffuse shifts displayed by full-length HpNikR. In addition, nt9-HpNikR was able to bind to the rpoD fragment  (28,54), and the boxed sequences are predicted half-sites recognized by HpNikR. (Fig. 4), although the lack of any detectable binding by fulllength HpNikR to rpoD prohibited a quantitative comparison between the calculated affinities of each protein. These data suggest a role for the N-terminal arm in the inhibition of nonspecific DNA binding.
The Low Affinity Ni 2ϩ Binding Site Is Not Conserved between Hp-and EcNikR-Two independent studies (26,31) that focused on HpNikR binding to P ureA suggested that HpNikR lacks the low affinity Ni 2ϩ -binding site observed for EcNikR (38,44). In one study the addition of NiCl 2 (75 or 800 M) to the mobility shift assay increased the DNA binding affinity of HpNikR for P ureA only 5-fold relative to a DNase I footprinting experiment with stoichiometric Ni 2ϩ (26). But if HpNikR lacks a low affinity Ni 2ϩ -binding site, then binding to specific promoters in a gel shift assay should not require additional Ni 2ϩ in the gel and running buffer. However, P nixA and P ureA binding by HpNikR was not detected in the absence of 50 M Ni 2ϩ in a mobility shift assay, suggesting that this protein may indeed possess a second metal-binding site (Fig. 5a, Table 3).
To explore the specificity of this metal requirement in the mobility shift assay, full-length HpNikR binding to P nixA was measured in the presence of 1 mM MgCl 2 , KCl, and NaCl in the gel and running buffer (Fig. 5a, supplemental Fig. 3). Interestingly, each of these salts resulted in a mobility shift, supporting the idea that HpNikR does not contain a low affinity Ni 2ϩbinding site but, rather, displays a nonspecific cation requirement at least in this assay. DNA binding by HpNikR also occurred when 50 M NiSO 4 or 1 mM MgSO 4 was added to the gel and running buffer, demonstrating that HpNikR required added cations and not Cl Ϫ ions for activity (data not shown). Furthermore, mutation of Glu-39 or Asp-43 to Ala, residues implicated in low affinity Ni 2ϩ binding by EcNikR (49,50), had no measurable effect on the DNA binding activity of HpNikR in the presence or absence of added cations in the mobility shift assay (supplemental Fig. 3). Also, the addition of NiCl 2 and MgCl 2 did not produce an additive effect on DNA binding affinity (data not shown).
Mg 2ϩ and K ϩ are normally present at 10 and 100 mM, respectively, in E. coli, whereas Ni 2ϩ is maintained at significantly lower levels (51). To begin to address which cation is physiologically relevant for HpNikR activity, an apparent affinity of HpNikR for Mg 2ϩ was estimated. DNase I footprinting of stoichiometric Ni 2ϩ :NikR-P nixA titrations was performed at different constant Mg 2ϩ concentrations, since it was not possible to titrate MgCl 2 with a constant HpNikR concentration due to the Mg 2ϩ dependence of DNase I (supplemental Fig. 4). Without MgCl 2 the 1:1 Ni 2ϩ :NikR-P nixA affinity was ϳ400 nM in a buffer containing 100 mM KCl. The highest HpNikR-DNA binding affinity (10 nM) was observed at MgCl 2 concentrations Ն3 mM, and this value was constant up to the highest concentration tested (50 mM). An estimate from a plot of the calculated K d values of HpNikR for P nixA as a function of MgCl 2 concentration indicates an affinity of the HpNikR-DNA complex for  Mg 2ϩ in the M to mM range (supplemental Fig. 4). The estimated affinity of HpNikR for Mg 2ϩ suggests that it may act as the cation required for HpNikR activity in vivo, assuming H. pylori maintains intracellular metal concentrations at levels similar to E. coli. Together these data demonstrate that the low affinity Ni 2ϩ -binding site is not conserved between Hp-and EcNikR but do not reveal a structural basis for the altered HpNikR cation requirement. The N-terminal Arm Plays a Distinct Role in HpNikR Binding to P nixA -nt9-HpNikR (1:1 Ni 2ϩ :NikR) was able to bind to P nixA with a K d of 3.3 nM in the mobility shift assay in the absence of additional cations in the gel and running buffer (Fig. 6, Table 3), indicating that the N-terminal arm is partly responsible for the cation requirement of full-length HpNikR for binding to P nixA . nt9-HpNikR still required stoichiometric Ni 2ϩ to bind to DNA, indicating that the protein is activated by high affinity Ni 2ϩ binding similar to full-length HpNikR (data not shown). nt9-HpNikR was also tested for binding to P ureA , P fur , and P nikR in the absence of added cations in the gel shift assay; however, no binding to these promoters was observed except at the highest protein concentrations tested (Ͼ300 nM for P fur and 1 M for P ureA and P nikR ) (supplemental Fig. 5). These data suggest that the N-terminal arm is playing a distinct structural role at P nixA as compared with P ureA , P fur , and P nikR .
Individual Residues of the N-terminal Arm Play Specific Roles in DNA Binding by HpNikR-To determine the contribution of individual N-terminal arm residues to specific DNA binding and the P nixA cation requirement, amino acids were individually mutated to either alanine (residues 1-5 and 7-9), valine (Asp-2), or methionine (Lys-6). Each mutant described below showed no change in stability relative to native HpNikR in the presence or absence of Ni 2ϩ (supplemental Fig. 2). To qualitatively test for contributions to cation binding in the HpNikR-P nixA complex, the mutant proteins (200 nM; 1:1 Ni 2ϩ : NikR) were examined in mobility shift assays with or without 50 M NiCl 2 (Fig. 5b). P nixA binding of the D7A and D8A mutants was detected in the absence of 50 M NiCl 2 , suggesting that Asp-7 and Asp-8 are linked to the cation requirement of full-length HpNikR. All of the mutants showed at least some binding to P nixA in the presence of 50 M NiCl 2 , demonstrating that none of the proteins was completely nonfunctional in this assay. However, the P4A and K6M mutants did not quantitatively shift the P nixA fragment, suggesting that Pro-4 and Lys-6 are required for maximal DNA binding affinity.
None of the mutant proteins showed cation-independent binding to P ureA in the absence of 50 M NiCl 2 (supplemental Fig. 6). However, the P4A and K6M mutants were again decreased in DNA binding affinity in the presence of 50 M NiCl 2 , suggesting that Pro-4 and Lys-6 are required for maximal DNA binding affinity to all HpNikR-regulated promoters. The mutants were also screened for the ability to bind to a non-promoter fragment (rpoD) under conditions in which nt9-HpNikR is able to bind to nonspecific DNA (250 nM protein, 50 M NiCl 2 ); however, no mutants were capable of shifting this fragment under the conditions tested (data not  Table 2. Highest to lowest protein concentrations are indicated by the black arrowhead of decreasing height above each gel. shown), suggesting that removal of the entire nt9 arm is necessary for increased nonspecific DNA binding by HpNikR.
DNA binding of individual mutant proteins to P nixA and P ureA was examined in further detail because of the differential effect of the nt9 truncation on binding to each promoter. Protein titrations of P4A, K6M, D7A, or D8A HpNikR (1:1 Ni 2ϩ : NikR) with P nixA and P ureA were performed in the presence or absence of 50 M NiCl 2 or 1 mM MgCl 2 in a mobility shift assay ( Table 3, Fig. 6; data are not shown for P4A and K6M). The P4A and K6M mutants had decreased affinity with added NiCl 2 in the gel and running buffer for both P nixA (P4A, 38-fold; K6M, 2-fold) and P ureA (P4A, 259-fold; K6M, 8-fold) ( Table 3). K6M HpNikR also had a 2.7-and 2.2-fold decrease in affinity for P nixA and P ureA , respectively, with added MgCl 2 . P4A HpNikR DNA binding could not be measured with 1 mM MgCl 2 due to protein aggregation. The cause of this aggregation was not explored further. Both P4A and K6M were unable to bind to either DNA fragment in the absence of additional NiCl 2 or MgCl 2 , similar to native HpNikR (data not shown). The D7A and D8A mutants bound to P nixA in the presence or absence of excess cations and to P ureA in the presence of excess cations with similar affinities of ϳ1-6 nM, although the affinity of D7A for P ureA with 1 mM MgCl 2 was not measured due to protein aggregation under this condition (Fig. 6, Table 3). The D7A and D8A affinities for these promoters are similar to those measured for full-length HpNikR and nt9-HpNikR under all the conditions tested, including a modest decrease in affinity for all proteins binding to either promoter in the presence of 1 mM MgCl 2 .

Asp-7 and Asp-8 Are Necessary and Sufficient for the Cation Dependence
of HpNikR-P nixA Binding-The effects of the D7A or D8A mutations on HpNikR P nixA binding suggested that these residues are responsible for the cation requirement for binding to this promoter, but neither mutant yielded a reproducible discrete mobilityshifted species, in contrast to that observed for nt9-HpNikR (Fig. 6). A D7A/D8A double mutant HpNikR was constructed to determine whether the absence of both aspartic acid residues better mimics nt9-HpNikR binding to P nixA . The presence of a discrete shifted species in the mobility shift assay with D7A/ D8A recapitulated the nt9-HpNikR binding result for P nixA (Fig. 6) in the absence of additional cations, indicating that the cation dependence of DNA binding to the nixA promoter by full-length HpNikR is mediated by both Asp-7 and Asp-8. Similar to nt9-HpNikR, D7A/D8A HpNikR required added cations in the gel and running buffer to bind to P ureA (supplemental Fig.  7). Protein titrations of D7A/D8A HpNikR to P nixA or P ureA in the presence (or absence for P nixA ) of additional NiCl 2 or MgCl 2 indicated this mutant was unaltered in binding affinity to both promoters as compared with full-length, native HpNikR (Fig. 6, Table 3).
A truncation mutant lacking the first five amino acids of HpNikR (nt5-HpNikR) was constructed to test if Asp-7 and Asp-8 are sufficient to impose a cation requirement on HpNikR for binding to P nixA or if the presence of other N-terminal arm residues is required. The resulting protein also contained a K6M mutation necessary for protein expression. Purified nt5-HpNikR was unable to bind to P nixA in the absence of excess FIGURE 6. Asp-7 and Asp-8 are necessary and sufficient for the HpNikR-P nixA cation requirement. Titrations of HpNikR mutants serially diluted 1.7-fold from concentrations of 500 nM to 1.7 pM with no added cations, 50 M NiCl 2 , or 1 mM MgCl 2 . Each full titration was run on two separate gels in parallel, and a vertical line separates each pair of gel images. Each titration represents one of at least two replicates. Affinities were calculated as described under "Experimental Procedures" and are listed in Table 3. Highest to lowest protein concentrations are indicated by the black arrowhead of decreasing height above each gel.

TABLE 3 Apparent DNA binding affinities of N-terminal arm mutants under different mobility-shift conditions
Apparent affinities of mutant HpNikR proteins measured by mobility shift assays were calculated as described under "Experimental Procedures." Affinities are reported in nM, and the averages of two independent experiments are shown.

The H. pylori NikR N-terminal Arm Modulates DNA Binding
cations, consistent with the cation requirement imposed by Asp-7 and Asp-8 (Fig. 6). The K6M mutation in the nt5-HpNikR construct did not influence DNA binding affinity in the presence of excess cations ( Table 3), suggesting that Lys-6 is required for maximum DNA binding affinity specifically in the context of the full-length N-terminal arm.

DISCUSSION
HpNikR is predicted to regulate 39 genes arranged in 31 operons, including both activation (28) and repression (25,(27)(28)(29)(30) of gene expression, in response to increased Ni 2ϩ (25). Full-length HpNikR bound multiple promoters with a range of affinities (Fig. 4, Table 2), consistent with the known nixA (28), ureA (26 -28, 31), fur (27), and nikR (25)(26)(27) and predicted hpn (25) direct regulation by HpNikR. The ability of HpNikR to recognize weakly conserved DNA sequences poses an intriguing question regarding HpNikR function in response to increasing intracellular Ni 2ϩ . These data suggest that there may be a hierarchy of Ni 2ϩ -and HpNikR-dependent gene regulation in H. pylori or different degrees to which Ni 2ϩ and HpNikR are able to regulate the expression of different genes. Alternatively, the range of DNA binding affinities may reflect differences in the promoter structure or a requirement for additional factors at some promoters. Additional experiments that closely examine Ni 2ϩ -and HpNikR-dependent regulation in H. pylori are required to test these hypotheses. These results also imply that defining a meaningful HpNikR consensus sequence will depend on further biochemical characterization of sequence-specific HpNikR-DNA interactions.
Site-directed mutation or truncation of residues in the N-terminal arm revealed important features of HpNikR-DNA interactions. These experiments demonstrated differential contributions from several residues and are consistent with previous studies of other RHH proteins, such as Arc and Mnt (34 -36, 55, 56). The different types of HpNikR mutations examined here and the potential ramifications for biological function are discussed in detail below. The inherently asymmetric nature of the tetrameric HpNikR-DNA complex (49) precludes a detailed molecular interpretation of the data. For example, of the four N-terminal arms present in the HpNikR-DNA complex, there will be at least two distinct conformations (Fig. 1a). Discerning the individual contributions of these different conformations to DNA binding interactions is complex, as demonstrated by detailed experiments with the Mnt repressor (57).
Our results show a cation requirement for HpNikR binding to P nixA that is linked to Asp-7 and Asp-8 of the N-terminal arm (Fig. 6). This observation combined with knowledge of several RHH-DNA complexes suggests that the cation may be necessary to prevent repulsive electrostatic interactions between the negatively charged Asp residues and the negatively charged phosphodiester backbone. Adjacent Asp residues that coordinate Mg 2ϩ in the presence of nucleic acids have been observed both functionally and structurally (58 -60). Notably, a structure of human immunodeficiency virus reverse transcriptase reveals a Mg 2ϩ coordinated in part by vicinal Asp residues and an ␣-phosphate oxygen of bound dTTP (60). Alternatively, cations may be required to prevent repulsive interactions between Asp-7 and Asp-8 and other negatively charged amino acids, such as Glu-39 and Asp-43, that may be in close proximity when the protein is bound to DNA.
DNA binding by the nt9-HpNikR to five promoters of HpNikR-regulated genes indicated that the arm is necessary for maintaining a hierarchy of binding affinities (Fig. 4, Table 2) and decreases DNA binding to nonspecific sites (Fig. 4). Several RHH family members, including Arc, Mnt, and MetJ (33)(34)(35)(36), contain N-terminal arms that play important roles in DNA recognition. Arc repressor contains seven amino acids N-terminal to the ␤-sheet motif that are disordered in the absence of DNA but assume fixed conformations in the presence of operator sites (34) with individual residues making hydrophobic, phosphate (33), or electrostatic (34) contacts with the DNA. N-terminal arm mutants of Arc exhibit significant DNA binding defects, although they still display wild-type hydroxyl radical footprints (33), similar to the nt9-HpNikR behavior observed in this study (Fig. 2). The tetrameric Mnt repressor depends on Arg-2 of its four-amino acid arm for DNA binding (35), and analysis of hybrid Arc and Mnt proteins identified two residues (one arm and one ␤-sheet position) that determine DNA binding specificity (55). One hybrid protein containing the Mnt arm residue and the Arc ␤-sheet residue bound both Mnt and Arc operators as well as nonspecific DNA (55), an effect similar to that observed for nt9-HpNikR (Fig. 4). It is clear that HpNikR belongs in the subset of RHH proteins that utilize N-terminal amino acids to modulate their DNA binding activity.
Mobility shift assays with different cations showed that HpNikR does not contain a low affinity Ni 2ϩ -binding site capable of significantly increasing DNA binding affinity at P nixA and P ureA (Fig. 5a, supplemental Fig. 3). These results are consistent with a previous study that showed only a 5-fold increase in DNA binding affinity to P ureA when excess NiCl 2 was present in a mobility shift assay (26). Our data are partially consistent with a recent report suggesting that HpNikR contains a low affinity metal-binding site that is specific for Mg 2ϩ , Ca 2ϩ , or Mn 2ϩ (31). In that study, fluorescence anisotropy was used to detect changes in HpNikR-P ureA binding (31). In striking contrast to the gel-based assays (this study and Abraham et al. (26)), the researchers were unable to detect DNA binding in the presence of 3 mM NiCl 2 . Importantly, the anisotropy experiments contained no added cations, whereas the footprinting experiments contained significant concentrations of cations (100 mM KCl, 1 mM MgCl 2 , 1 mM CaCl 2 in Abraham et al. (26) and 100 mM KCl, 3 mM MgCl 2 in our study). This difference does not explain the cation dependence of the mobility shift assays, which are very sensitive to added cations, whereas the footprinting reaction was sensitive to MgCl 2 even in the presence of 100 mM KCl. Each DNA binding assay (footprinting, mobility shift, and fluorescence anisotropy) likely provides different information regarding HpNikR-DNA interactions that will require further detailed studies to reconcile.
Interestingly, mutation of Glu-39 or Asp-43 to Ala (corresponding to Glu-30 and Asp-34 in EcNikR, which are implicated in low affinity Ni 2ϩ binding (49,50)) had no significant effect on DNA binding in mobility shift assays (supplemental Fig. 3). This result supports the hypothesis that HpNikR does not contain the low affinity Ni 2ϩ -binding site present on EcNikR. Furthermore, the lack of an observable DNA binding effect from mutation of Glu-39 or Asp-43 suggests that HpNikR exists in a significantly different conformation on DNA as compared with EcNikR. The decrease in half-site spacing observed in HpNikR-regulated promoters (Fig. 2d) relative to EcNikR operator spacing likely places different constraints on the HpNikR DNA binding domains relative to the C-terminal Ni 2ϩ binding domains. The nt9 arm of HpNikR may also contribute by sterically occluding cation binding to Glu-39 and Asp-43.
Biologically, the presence or absence of a low affinity Ni 2ϩ site could reflect the different regulatory functions and physiological contexts of the Hp-and EcNikR proteins. H. pylori expresses the hydrogenase and urease nickel enzymes, several Ni 2ϩ binding chaperones important in enzyme assembly, and Ni 2ϩ storage proteins (62). The increased complexity of Ni 2ϩ physiology in H. pylori may have imposed selective pressure on HpNikR, resulting in the augmentation of Ni 2ϩ -dependent regulation and the observed biochemical differences in HpNikR-DNA interactions compared with EcNikR. Experiments examining the ability of HpNikR truncation and site-specific mutants to regulate gene expression in H. pylori will be required to test the role of the N-terminal arm and cation binding in HpNikR function in vivo.
An intracellular Mg 2ϩ concentration in H. pylori similar to the 10 mM measured for E. coli (51) together with the estimated affinity of Ͻ3 mM of the N-terminal arm cation binding site of HpNikR suggest that this site probably exists in a Mg 2ϩ -bound state in the cell. Mg 2ϩ binding to the arm in the absence of Ni 2ϩ likely does not activate HpNikR in vivo, consistent with the known Ni 2ϩ -responsive HpNikR regulation in H. pylori (24 -30). An alternative possibility is that the arm cation-binding site exists only when HpNikR is bound to Ni 2ϩ or Ni 2ϩ and DNA, in which case HpNikR would remain in a fully apo state in the cell until Ni 2ϩ levels increase. Only upon Ni 2ϩ binding to the C-domain high affinity site or Ni-HpNikR DNA binding would Mg 2ϩ bind to the N-terminal arm, possibly stabilizing the protein-DNA complex. Direct measurements of Mg 2ϩ binding to HpNikR are required to distinguish between these scenarios. D7A/D8A HpNikR was unaffected in cation-dependent DNA binding to P ureA , P fur , or P nikR (supplemental Fig. 5), indicating that the nt9 arm is conformationally distinct when HpNikR is bound to P nixA compared with the other three promoters. The differential roles of the N-terminal arm in HpNikR binding to P nixA , an HpNikR-repressed gene, and P ureA , an HpNikR-activated gene, might also suggest that the arm is important for HpNikR-DNA interactions at one promoter and HpNikR-protein interactions at the other. For example, the arm might be involved in HpNikR-RNA polymerase interactions specifically at P ureA , where HpNikR binds upstream of the Ϫ10 and Ϫ35 sequences (Fig. 2b) (26 -28), allowing for the induction of ureAB expression.
Mutation of Pro-4 and Lys-6 resulted in proteins with decreased affinities for both P nixA and P ureA . Interestingly, the loss of the arm and specific changes in amino acid composition of the arm had opposing effects on DNA binding, since nt9-HpNikR has higher affinity for DNA (Table 2) and P4A and K6M HpNikR have decreased affinity for DNA ( Table 3). The decrease in affinity displayed by P4A HpNikR most likely reflects the propensity of proline residues to display unusual peptide bond angles. Our results suggest that Ala substitution for Pro4 may result in greater conformational flexibility in the first five residues of the arm that may adversely impact DNA binding affinity.
The decrease in DNA binding affinity exhibited by K6M HpNikR, but not nt5-HpNikR, suggests that Lys-6 may interact with one or more of the N-terminal five amino acids present in K6M HpNikR. Replacement of Lys with Met in this case might destabilize the protein-DNA complex by imposing non-favorable interactions between other arm residues and Met-6, whereas the N-terminal Met, in the case of nt5-HpNikR, would be relieved from these constraints. It is also possible that the positively charged N terminus of nt5-HpNikR is able to substitute for Lys-6, making additional electrostatic interactions with the DNA or other arm residues.
Previous studies examining HpNikR DNA binding in vitro have used versions of purified HpNikR with significantly altered N termini, including a protein containing a 15-amino acid N-terminal Strep tag (MASWSHPQFEKIEGR) (28,29), a protein with three extra N-terminal amino acids after cleavage of a His tag (GSH) (27) and a protein with a single added alanine remaining after cleavage of N-terminal His and S tags (30). Strep-HpNikR and GSH-HpNikR protected a similar region of P nixA (Strep-HpNikR (28)) and P ureA (Strep-HpNikR and GSH-HpNikR (27,28)) as compared with native HpNikR (Fig. 2d  (26)), a result not surprising since nt9-HpNikR displayed identical protection of P nixA and P ureA as compared with full-length HpNikR (Fig. 2). The dramatic changes in DNA binding affinity and specificity that result from loss of the HpNikR N-terminal arm (Figs. 3 and 4, Table 2) suggest that Strep-HpNikR and GSH-HpNikR might exhibit altered affinity and specificity for DNA; however, the limited experimental data available for each protein (a single, high protein concentration with a high Ni 2ϩ concentration for Strep-HpNikR binding to P nixA (28), P ureA (28), P fecA3 (29), and P frpB4 (29) and four-or five-point protein titrations with a high Ni 2ϩ concentration that are not quantitated for GSH-HpNikR binding to the exbB-nikR intergenic region, P ureA and P fur (27)) precludes a more quantitative assessment of the effects of each non-native N-terminal extension. The importance of the N terminus of HpNikR in DNA binding, as shown by the current study, further indicates that experiments using HpNikR variants with altered N termini should be interpreted with caution.
A comparison of the NikR family members predicted from genome sequence annotations indicates arm lengths up to 32 amino acids N-terminal to the ␤-sheet (supplemental Fig. 7). Significant variability exists in arm sequences of NikR from different H. pylori strains as well as different Helicobacter species. Notably, H. pylori isolate HPAG1 NikR contains a P4H change and Helicobacter acinonychis strain Sheeba NikR contains a P4N change. Helicobacter mustelae NikR contains a completely unique 10 amino acid N-terminal arm: MRT-MEKEKNS. Interestingly, the H. mustelae arm lacks Asp residues and instead contains two alternating Glu and Lys residues at positions 5-8 as well as an Arg at position 2. The rodent pathogen Helicobacter hepaticus contains a NikR with five N-terminal amino acids, only one of which is charged (Lys-2). Detailed biological data characterizing the Ni 2ϩ physiologies of The H. pylori NikR N-terminal Arm Modulates DNA Binding these bacteria are currently lacking, although one study has begun to address the role of the Ni 2ϩ -enzyme urease in H. hepaticus metabolism (63). It is likely that significant differences exist, particularly between H. pylori, H. mustelae, and H. hepaticus, given that they colonize the gastric mucosa of humans (64), ferrets (65), and the rodent liver and intestine (66,67), respectively. The structural consequences of these arm sequence changes are difficult to predict since the target genes of each NikR ortholog are unknown; however, the sequence disparity suggests that the amino acid changes may result in altered NikR function in these different bacteria.
HpNikR has likely adopted its N-terminal arm to respond differently than EcNikR to increased intracellular Ni 2ϩ . High affinity Ni 2ϩ binding is structurally conserved between the two NikR orthologs (32,45), as is its effect on DNA binding affinity. Additional metal binding by each protein and the related DNA binding responses are distinct. It is likely that additional structural differences in HpNikR relative to EcNikR are necessary to fully modulate the activity of the former. Nevertheless, the unique properties of HpNikR attributable to the N-terminal arm indicate that regulatory function can be tuned through localized changes in protein structure. Alignments of predicted bacterial transcription factors that have distinct DNA binding motifs (53,61,68) indicate that the addition of extra amino acids adjacent to DNA binding domains is a common occurrence, suggesting that this may represent a widespread mechanism of regulator evolution.