The crystal structure of the Helicobacter pylori LlaJI.R1 N-terminal domain provides a model for site-specific DNA binding

Restriction modification systems consist of an endonuclease that cleaves foreign DNA site-specifically and an associated methyltransferase that protects the corresponding target site in the host genome. Modification-dependent restriction systems, in contrast, specifically recognize and cleave methylated and/or glucosylated DNA. The LlaJI restriction system contains two 5-methylcytosine (5mC) methyltransferases (LlaJI.M1 and LlaJI.M2) and two restriction proteins (LlaJI.R1 and LlaJI.R2). LlaJI.R1 and LlaJI.R2 are homologs of McrB and McrC, respectively, which in Escherichia coli function together as a modification-dependent restriction complex specific for 5mC-containing DNA. Lactococcus lactis LlaJI.R1 binds DNA site-specifically, suggesting that the LlaJI system uses a different mode of substrate recognition. Here we present the structure of the N-terminal DNA-binding domain of Helicobacter pylori LlaJI.R1 at 1.97-Å resolution, which adopts a B3 domain fold. Structural comparison to B3 domains in plant transcription factors and other restriction enzymes identifies key recognition motifs responsible for site-specific DNA binding. Moreover, biochemistry and structural modeling provide a rationale for how H. pylori LlaJI.R1 may bind a target site that differs from the 5-bp sequence recognized by other LlaJI homologs and identify residues critical for this recognition activity. These findings underscore the inherent structural plasticity of B3 domains, allowing recognition of a variety of substrates using the same structural core.

predatory bacteriophage viruses (1). These systems consist of a restriction endonuclease and a methyltransferase, which provide the dual function of cleaving exogenous DNA site-specifically, whereas protecting the host genome via methylation of the corresponding recognition sequence (2). Three variants of RM systems, type I, type II, and type III, have been identified and differ in their structural composition and mechanism of restriction. Type I systems are multifunctional complexes containing separate restriction, methylation, and DNA-sequence recognition subunits. These machines require Mg 2ϩ and ATP, catalyze both restriction and methylation, and cut DNA nonspecifically far from their recognition sites (3). Type II systems are the simplest, generally existing as dimers that carry out the recognition and restriction activities. These enzymes do not require ATP and have a separate, associated methyltransferase (4). Homodimeric type II restriction enzymes recognize DNA sequences that are symmetric, whereas those that are heterodimeric can bind asymmetric sequences (5). Type III systems contain separate modification (Mod) and restriction (Res) subunits that form homodimeric Mod2 and heterotetrameric Res2Mod2 complexes and catalyze both restriction and methylation in a Mg 2ϩ and ATP-dependent manner (6). They differ from type I systems, however, in that they require two inversely oriented recognition sites that can vary in their spatial separation (7).
Modification-dependent restriction systems (MDRS), colloquially referred to as type IV systems, recognize and cleave modified DNA (8). McrA and McrBC are prototypical MDRSs that target DNA containing 5-methylcytosine (5mC) and 5-hydroxymethylcytosine (5hmC) (9 -14). McrA is a small, dimeric protein that recognizes the symmetrically methylated sequence Y5mCRG (15). The McrB and McrC proteins together form a conserved, two-component restriction complex capable of long-range DNA translocation similar to type I and type III enzymes. Escherichia coli (Ec) McrB contains an N-terminal DNA-binding domain (12) and a C-terminal AAAϩ motor domain that hydrolyzes GTP and mediates nucleotide-dependent oligomerization into heptameric rings (16). McrB's basal GTPase activity is stimulated via interaction with its partner endonuclease McrC (13), which cannot bind DNA on its own and in vitro only associates with the McrB oligomer (17). Biochemical studies suggest a model for DNA cleavage in which McrB and McrC assemble at two distant R M C sites (where R is a purine, and M C is a methylcytosine) and translocate in a manner that requires stimulated GTP hydrolysis (10,18). Collision of McrBC complexes triggers cleavage of both DNA strands close to one of the R M C sites (14,19). Other MDRS families display a variable spectrum of specificity for different modifications. These include MspJI, which recognizes 5mC and 5hmC (20), the PvuRts1I family, whose members show unique individual specificities for 5hmC and/or 5-glucosylhydroxymethylcytosine (5ghmC) (21), and GmrSD, which recognizes 5ghmC (22). Structural studies of McrB, MspJI, PvuRtsI, and AbaSI suggest type IV systems employ a generalized base-flipping mechanism for recognition of the modified DNA (23)(24)(25)(26)(27).
The LlaJI restriction cassette was first identified in Lactococcus lactis on the naturally occurring plasmid pNP40 and shown to confer resistance against common lactococcal phages (28). It consists of an operon encoding two 5mC-methyltransferases, LlaJI.M1 and LlaJI.M2, and two restriction proteins, LlaJI.R1 and LlaJI.R2, both of which are absolutely required for restriction activity in vivo (29). The M1 and M2 methyltransferase activities modulate expression of LlaJI operon in vivo (30). Although formally classified as a type II R/M system (REBASE enzyme number 10100, New England Biolabs), LlaJI.R1 and LlaJI.R2 share domain homology with McrB and McrC, respectively. R1 contains sequence motifs that identify its C-terminal portion as a GTP-specific AAAϩ domain and R2 contains a conserved C-terminal PD-(D/E)XK endonuclease domain. These features suggest LlaJI enzymes function more like McrBC than other type II systems.
Unlike McrB, however, L. lactis LlaJI.R1 binds DNA sitespecifically, recognizing the asymmetric 5Ј-GACGC-3Ј sequence in one strand and 5Ј-GCGTC-3Ј in the other strand (29). Other LlaJI homologs have been identified in Helicobacter pylori, Streptococcus pyogens, Bacillus cereus, and Clostridium cellulovorans (29,31). Of these, C. cellulovorans LlaJI has also been shown to target the same asymmetric, 5-bp sequence (31). How LlaJI proteins recognize DNA site-specifically is unknown. Here we present the structure of the N-terminal DNA-binding domain of H. pylori LlaJI.R1 (HpR1⌬136) at 1.97-Å resolution, which adopts a B3 domain fold. Structural comparison to B3 domain-containing plant transcription factors and restriction endonucleases identifies the key recognition motifs responsible for site-specific DNA binding. Additional evidence from biochemistry and structural modeling argues that HpLlaJI.R1 binds a target site that differs from the 5-bp sequence recognized by the L. lactis and C. cellulovorans LlaJI homologs. Mutagenesis further identifies residues Arg-17 and Arg-60 as critical determinants of HpLlaJI.R1 DNA binding. Together, these findings underscore the inherent structural plasticity previously noted for B3 domains, which confers specificity to different sequences via the same structural core.

Structure and topology of the HpLlaJI.R1 N-terminal domain
Although previous studies show LlaJI.R1 binds DNA sitespecifically (29,31), the molecular means through which this is achieved remains unknown. Numerous attempts to purify either the full-length L. lactis LlaJI.R1 or its isolated N-terminal DNA-binding domain for structural studies were unsuccessful. Bioinformatics identifies various other species harboring the LlaJI operon, including H. pylori (Hp) (29). Computational analyses of these homologs by fold matching and structural prediction algorithms failed to identify a reliable template for modeling DNA interactions. To understand how the LlaJI.R1 binds DNA site-specifically, we therefore crystallized the Nterminal domain of HpLlaJI.R1 (HpR1⌬136) and determined its structure at 1.97 Å by selenium SAD phasing (32) (Fig. 1).
HpR1⌬136 crystallizes in the space group P1 with four molecules (A-D) in the asymmetric unit organized as two homodimers packed end to end, with molecules A and B and molecules C and D pairing together (Fig. 1A). These dimers superimpose with an r.m.s. deviation of 0.589 Å. Each HpR1⌬136 monomer consists of a core six-stranded ␤ sheet that folds into a pseudo-␤ barrel flanked on four separate edges by ␣ helices (␣1-␣4) (Fig. 1, B and C). An additional ␤-strand (␤7) inserts at the dimer interface and breaks the symmetry, adopting an antiparallel configuration with ␤1 B /␤1 D and a parallel configuration with ␤1 A /␤1 C (Fig. 1, B and C). Clear connectivity between ␤7 and ␣4 can be traced in molecule B (Fig.  1D). Structural superposition of the two asymmetric dimers suggests ␤7 is connected in the same manner in molecule D despite the lack of density for the ␣4 -␤7 loop (Fig. 1E). We observe no density for the corresponding ␤7 strands in either molecule A or molecule C.
␤7 residues Leu-127 and Phe-129 interact with a hydrophobic cluster sandwiched between ␤1 and the amphipathic ␣2 helix in each monomer ( Fig. 2A). Ile-24, His-27, and Phe-28 in ␣2 and Val-115, Leu-116, and Leu-119 in ␣4 provide additional stabilizing contacts across the dimer interface (Fig. 2, A and B). ␤7 insertion helps space these elements and prevent steric clashing that otherwise would occur. A total interaction surface of 800 Å 2 is shared between the monomers. Size exclusion chromatography coupled to multi-angle light scattering (SEC-MALS) indicates HpR1⌬136 dimerizes in solution (Fig. 2C), suggesting the observed molecular organization in the crystal lattice is not simply a packing artifact. Deletion of ␤7 renders HpR1⌬136 insoluble. Point mutations at the dimer interface in ␣2 (I24N, H27E, F28E) and ␣4 (L116E and L119E) displayed similar insolubility phenotypes and could not be purified. Only the V115N mutant retained solubility and, like WT, forms stable dimers in solution when analyzed by SEC-MALS (Fig. 3A). These data support the notion that HpR1⌬136 dimerization is required for stability and biologically relevant.

Structure of H. pylori LlaJI.R1 N-terminal domain
tantly, this fold is structurally distinct from the analogous region in EcMcrB, which preferentially binds DNA containing methylated cytosines.
B3 domains contain two critical regions that confer DNA target site-specificity. These recognition motifs, termed the N-arm and C-arm, reside on opposite edges of the pseudobarrel core and form a wrench-like structure that contacts the major groove (36,37,40). In this arrangement, the N-arm specifically associates with the 5Ј-half of the target site and the C-arm engages the 3Ј-half. Comparison to the DNA-bound AtARF1 structure identifies these key features within the HpR1⌬136 model: the N-arm encompasses the ␤1-␤2 loop and the ␣1 helix and the C-arm localizes to the ␤3-␤4 loop (Figs. 1C and 4B). Both proteins display a comparable electrostatic surface, with an extensive basic patch positioned between the Nand C-arms and coincident with the DNA-binding face of AtARF1 (Fig. 4C). This mode of substrate binding is conserved among DNA-bound B3 domain structures and is consistent with other HpR1⌬136 structural superpositions (Fig. S1). An exception is the B3 domain of NgoAVII, whose orientation on DNA is inverted such that the N-arm associates with the 3Ј-half of the target site and the C-arm with the 5Ј-half (40).
The BfiI contains other unique motifs (termed the N-and C-loops) that provide additional phosphate backbone and minor groove interactions (37). These are shortened in EcoRII and absent in all previously characterized plant B3 domains (36,38,(41)(42)(43). HpR1⌬136 similarly lacks these segments, suggesting it either evolved from a more simplified common ancestor or lost these segments over time due to a lack of selective pressure.
The putative DNA-binding surface of each HpR1⌬136 monomer faces away from the dimer interface, suggesting that The core-fold of each monomer is shown in blue. The relative position and connectivity of the asymmetric ␤7 strand associated with molecules B and D is denoted by dashed outlines and colored in raspberry. The N-and C-arms are colored orange and cyan, respectively. D, 2F o Ϫ F c electron density (blue mesh) of the ␣4 -␤7 region in molecule B contoured to 1. E, superposition of HpR1⌬136 dimers AB and CD. AB dimer is colored green and blue, whereas the CD dimer is colored gray. ␤7 from molecules B and D are colored raspberry and gray, respectively.

Structure of H. pylori LlaJI.R1 N-terminal domain
HpR1⌬136 has the capacity to bind two DNA target sites simultaneously. The asymmetric orientation of the ␤7 strand, however, positions Glu-131 close to one of the binding sites and alters its surface charge potential in a manner that makes it less basic (Fig. 4D). This intrinsic difference would allow one monomer to bind more efficiently and could bias the arrangement of HpLlaJI on DNA.
All previously characterized B3 domains exist as monomers (36 -44). To understand what hinders dimerization in these contexts, we superimposed the coordinates of other B3 domains onto our HpR1⌬136 dimer and examined the orientation of secondary structure features relative to the dimer interface (Fig. 5). Although EcoRII and BfiI have structurally equivalent ␤-strands that partially align with ␤7, they also contain helical segments that sterically prevent two monomers from coming together (Fig. 5, A and B). The ␣1 helices of VRN1 and UbaLAI would similarly clash and block dimerization (Fig. 5, C and D). AtARF1, NgoAVII, RAV1, and At1g16640.1, in contrast, all lack a corresponding ␤7 strand (Fig. 5, E-H), suggesting monomers cannot be properly spaced to avoid collision. The stabilizing hydrophobic interactions provided by ␤7 would also be absent. These observations highlight the importance of secondary structure features in modulating the oligomeric state of B3 domains and will be useful for predicting interactions in other uncharacterized proteins that contain this conserved fold.

HpR1⌬136 structure provides model for site-specific binding
Previous biochemical and genetic studies indicate that L. lactis and C. cellulovorans LlaJI target the 5-bp sequence 5Ј-GACGC-3Ј (29,31). HpR1⌬136 shows weak affinity for DNA containing this sequence (Ll) when assessed by filter binding (Fig. 6A, blue). Scrambling the putative binding sequence (Llscr) has no effect on the affinity (Fig. 6A, green), suggesting this represents the basal level for nonspecific DNA binding by HpR1⌬136. EcMcrB, in contrast, preferentially binds 5mC-containing DNA (5mC) but does not bind a nonmethylated version of the same substrate (nm) under the same assay conditions (Fig. 6A, black versus red). EcMcrB similarly does not bind either the Ll or Llscr substrates (Fig. 6A, yellow and orange), underscoring how its binding depends on the presence of methylated cyotsines.
Unexpectedly, HpR1⌬136 showed enhanced affinity for the E. coli-specific 5mC and nm substrates (Fig. 6A, light blue and purple) relative to Ll and Llscr substrates. We attribute this to subtle sequence differences as the binding is independent of methylation status. The 5mC and nm substrates likely contain sequence fragments that more closely mimic the preferred recognition site of HpR1.LlaJI, which is distinct from both EcMcrB and other LlaJI homologs.
Despite a common fold, B3 domains exhibit divergent sequence preferences. Previous structural and biochemical data show that the C-arm length can influence the length of the recognized target site (Fig. 6B). A longer C-arm is excluded from the major groove (Fig. 6C), decreasing the overall binding footprint and biasing recognition toward a 5-bp site (36,37). A shorter C-arm affords greater access to the DNA bases, which in some instances increases the number of specific contacts and extends the target site to six bases (37,38). The amino acid

Structure of H. pylori LlaJI.R1 N-terminal domain
composition of the N-and C-arms ultimately dictates specificity, however, and thus some B3 domains with shorter C-arms still bind 5-bp sites (40,44). Structural superposition reveals a shorter C-arm in HpR1⌬136 (Fig. 6C).
In the absence of a DNA-bound complex and without explicit knowledge of the HpLlaJI target site, we used the AtARF1-DNA structure as a proxy to identify side chains that might contribute to specificity. The N-arm residue His-136 and C-arm residues Arg-181, Pro-184, and Arg-186 are critical for AtARF1 DNA binding (Fig. 6D). Structural modeling reveals similar residues in HpR1⌬136 (Fig. 6E), with His-14 and Arg-17 in the N-arm and Pro-59 and Arg-60 in the C-arm poised to provide base-specific contacts. Interestingly, Arg-17 is spatially oriented like Arg-181 in AtARF1, hinting that it would contact the 3Ј-half of the target site despite being localized in the N-arm.
To confirm the significance of our structural modeling, we mutated the predicted binding residues in HpR1⌬136 and assessed how each substitution affects interaction with the E. coli-specific nm DNA substrate via filter binding (Fig. 6F). H14A (red), R17A (orange), and R60A (light blue) mutations show a marked decrease in affinity for nm DNA versus wildtype (WT, purple), whereas P59A (green) shows less of an effect. The R17A/R60A double mutant (brown) completely abolishes binding. This finding was corroborated using electrophoretic mobility shift assays (EMSAs) to measure the association of HpR1⌬136 with digested, nonmethylated -phage DNA (Fig.  7). We observed a significant gel shift with WT HpR1⌬136. H14A and P59A show similar shifts in this assay, whereas the individual R17A and R60A substitutions produce a moderate reduction in binding. The R17A/R60A double mutant, however, significantly impairs binding (Fig. 7), similar to its effects on the nm DNA substrate in the filter-binding assay (Fig. 6F). All of these mutants form stable dimers in solution (Fig. 3B), arguing that their effects are not due to global structural perturbations. Together these data implicate
B3 domains contain conserved residues that associate with "clamp" phosphates at the 5Ј ends of each strand in the target site (37). Arg-177 and Lys-126 form these interactions in AtARF1 (Arg-81 and Lys-23 in EcoRII; Arg-272 and Lys-340 in BfiI; Lys-27 and Lys-82 in UbaLAI; Lys-212 and Lys-275 in R.NgoAVII). In HpR1⌬136, Arg-6 is poised to act on one strand, whereas Lys-50 could perform a similar function on the opposing strand. Lys-50 is positioned away from the modeled DNA backbone in the apo state and may be reoriented upon target recognition. Conformational rearrangements in the BfiI and R.NgoAVII B3 domains have previously been observed upon DNA binding (35,37,40).

Discussion
Here we described the structure of the HpLlaJI.R1 DNAbinding domain and demonstrated that it adopts a B3 domain fold. B3 domains are prevalent among bacterial restriction endonucleases and plant transcription factors, where they function as site-specific DNA-binding modules (37,39,45). Previous crystallographic studies revealed that the N-and  Figure 6. Structural modeling of HpR1⌬136 substrate recognition. A, filter binding analysis of HpR1⌬136 (Hp) and full-length EcMcrB (Ec) interactions with different DNA substrates. Substrate abbreviations are as follows: 5mC, methylated EcMcrB-specific substrate; nm, nonmethylated EcMcrB-specific substrate; Ll, site-specific substrate containing the L. lactis LlaJI.R1 5Ј-GACGC-3Ј target site sequence; Llscr, substrate with the L. lactis LlaJI.R1 target site sequence scrambled as a control. Sequences for each substrate can be found under "Experimental procedures." Binding was performed at 30°C for 10 min in a 30-l reaction mixture containing 14.5 nM unlabeled DNA and 0.5 nM labeled DNA. Samples were filtered through KOH-treated nitrocellulose and binding was assessed by scintillation counting. B, relationship between C-arm length and target site length in previously determined B3 domain-DNA complexes. C, orientation of C-arm loops relative to DNA in various B3 domain homologs. DNA from the AtARF1 complex (PDB code 4ldx) is shown. C-arm coloring is labeled below along with corresponding PDB codes. D, key residues in AtARF1 DNA binding. E, residues predicted to be important for HpR1⌬136 DNA binding based on structural comparison. AtARF1 DNA modeled as in D. F, filter binding analysis of HpR1⌬136 mutants. Point mutations of predicted binding residues identified in D (H14A, red; R17A, orange; P59A, green; R60A, light blue; R17A/R60A; brown) were assessed for binding to the nm DNA substrate. Filter binding was carried out as described in A. The WT curve (purple) is the same as shown in A (Hpϩnm).

Structure of H. pylori LlaJI.R1 N-terminal domain
C-arms determine the specificity of each individual B3 domain and confer structural plasticity to the conserved core scaffold (36 -38, 40, 43). Our structural data and modeling identifies the N-and C-arms in HpR1⌬136 along with key residues that likely form direct contacts with the DNA backbone, clamp phosphates, and specific bases. HpR1⌬136 has weak affinity for DNA containing the asymmetric 5-bp site that other LlaJI homologs target (29, 31) and a surprisingly stronger affinity for the EcMcrB-specific DNA substrates, regardless of their methylation status (Fig. 6A). We note that HpR1⌬136 contains a shorter C-arm and thus could potentially bind a 6-bp site. Although further studies are required to pinpoint the target site of HpLlaJI.R1, our findings offer a general model for site-specific binding and provide a structural explanation for why LlaJI homologs do not target modifications despite sharing a similar domain organization with McrBC. Importantly, we identify the N-arm Arg-17 and C-arm Arg-60 residues as critical determinants of DNA binding and specificity in HpR1⌬136. Individual point mutations at these positions display moderate defects, whereas a combined double mutant completely abolishes DNA binding in all assays tested (Figs. 6F and 7). The H14A and P59A mutations show varying effects depending on the specific substrate used and the sensitivity of the assay. Although His-14 and Pro-59 may also impart specific binding interactions, their contribution is likely context dependent.
HpLlaJI.R1 is unique in that its isolated B3 domain dimerizes, both in solution (Fig. 2C) and in crystallo (Fig. 1). ␤7 is absolutely essential for HpR1⌬136 dimerization and structural stability, as a truncation of this motif renders the protein insoluble. Our structure shows that direct dimerization of other B3 domains is hindered by either (i) the intrinsic lack of a structur-

Structure of H. pylori LlaJI.R1 N-terminal domain
ally equivalent ␤7 strand or (ii) the presence of additional helical motifs at the N or C terminus that sterically clash with ␤7 or ␣4 at the dimer interface (Fig. 5). Dimerization of other B3 domain-containing proteins instead occurs through additional structural elements. For instance, AtARF1 monomers associate through a separate dimerization domain, which facilitates cooperative binding of the B3 domains to two anti-parallel 5Ј-TGTCTC-3Ј sites on opposing strands (38). BfiI, EcoRII, and R.NgoAVII also dimerize but through their respective nuclease domains (34,35,40). These observations will help in classifying uncharacterized B3 domains and predicting their architectural organization.
Our structural data show that the ␤7 strand from one HpR1⌬136 monomer is asymmetrically stabilized at the dimer interface, whereas the corresponding region in the other monomer remains disordered. The orientation of this strand dictates the electrostatic landscape on the dimer surface, making the DNA-binding site in one monomer more basic than the other. Although we cannot completely rule out that this is an artifact of crystallization, an analogous phenomenon was noted in the rotavirus A nonstructural protein 3 (NSP3) homodimer (46). There the asymmetric stabilization of a helix from one monomer creates a single positively charged binding site that ultimately leads to a stoichiometry of 2:1 NSP3:viral mRNA (46). We speculate that HpLlaJI.R1 may bind DNA with a similar 2:1 stoichiometry, but that only one B3 domain will directly contact the target site.
Asymmetric binding could have important implications for the assembly of a cleavage-competent LlaJI restriction complex. Like McrB, LlaJI.R1 contains a conserved GTP-specific AAAϩ domain at its C terminus (29). EcMcrB forms heptameric rings in the presence of GTP and this oligomerization is critical for recruiting its partner endonuclease McrC (16), which cannot bind DNA on its own and preferentially associates with the assembled AAAϩ domain (17). Although the exact organization of McrBC on DNA has yet to be elucidated, biochemical and structural studies have shown the EcMcrB N-terminal domain binds a single methylated cytosine via base flipping (23). The intrinsic asymmetry of the McrBC complex therefore imposes constraints on how the individual subunits interact with the R M C site and suggests that some monomers are directly engaged whereas others are not. The asymmetric HpR1⌬136 dimer could reflect a similar structural constraint in the LlaJI system wherein the alternative positioning of the ␤7 strand dictates which monomers bind the target sequence. Further structural characterization of both systems will be necessary to parse out how substrate binding, GTP-dependent assembly, and nuclease recruitment are coordinated in each case. Although LlaJI and McrBC differ in their specificity and targeting, we predict the general molecular mechanisms governing the function of LlaJI and McrBC will be conserved.

Cloning, expression, and purification of HpLlaJI.R1 constructs
DNA encoding the H. pylori LlaJI.R1 protein (DOE IMG/M ID 637022177) was codon optimized for E. coli expression and synthesized commercially by Bio Basic Inc. DNA encoding the N-terminal domain (HpR1⌬136; residues 1-136) was amplified by PCR and cloned into pET21b, introducing a His 6 tag at the C terminus. Selenomethionine-labeled (SeMet) HpR1⌬136 was expressed in minimal media using methionine auxotrophs (T7 Express Crystal Competent E. coli, New England Biolabs) according to manufacturer protocols. Native HpR1⌬136 was transformed into BL21(DE3) cells, grown at 37°C in Terrific Broth to an A 600 of 1.0, and then induced with 0.3 mM isopropyl 1-thio-␤-D-galactopyranoside overnight at 19°C. All cells were harvested, washed with nickel load buffer (20 mM HEPES, pH 7.5, 500 mM NaCl, 30 mM imidazole, 5% glycerol (v/v), and 5 mM ␤-mercaptoethanol), and pelleted a second time. Pellets were typically flash frozen in liquid nitrogen and stored at Ϫ80°C.
Thawed pellets from 500-ml cultures were resuspended in 30 ml of nickel load buffer supplemented with 10 mM phenylmethylsulfonyl fluoride, 5 mg of DNase (Roche Applied Science), 5 mM MgCl 2 , and a Roche complete protease inhibitor mixture tablet (Roche). Lysozyme was added to 1 mg/ml and the mixture was incubated for 15 min rocking at 4°C. Cells were disrupted by sonication and the lysate was cleared of debris by centrifugation at 13,000 rpm (19,685 ϫ g) for 30 min at 4°C. For native and SeMet HpR1⌬136, the supernatant was filtered, loaded onto a 5-ml HiTrap chelating column charged with NiSO 4 and then washed with nickel load buffer. Hp⌬136 was eluted with an imidazole gradient from 30 mM to 1 M. Pooled fractions were dialyzed overnight at 4°C into SP loading buffer (20 mM HEPES, pH 7.5, 50 mM NaCl, 1 mM EDTA, 5% glycerol (v/v), and 5 mM DTT). The sample was applied to a 5-ml HiTrap SP HP column equilibrated with SP loading buffer and then washed with SP loading buffer. HpR1⌬136 was eluted with a NaCl gradient from 50 mM to 1 M. Pooled fractions were concentrated and further purified by SEC using a Superdex 200 10/300 column. All proteins were exchanged into a final buffer of 20 mM HEPES, pH 7.5, 150 mM KCl, 5 mM MgCl 2 , and 1 mM DTT (5 mM for SeMet labeled) during SEC and concentrated to 5-40 mg/ml. Concentrations of purified proteins were determined by SDS-PAGE and densitometry was compared against BSA standards. All amino acid substitutions were introduced into HpR1⌬136 in pET21b by QuikChange PCR and mutant proteins were purified as described for WT.

Cloning, expression, and purification of EcMcrB
DNA encoding the E.coli McrB protein (Uniprot P15005) was codon optimized for E. coli expression and synthesized commercially by GENEART. DNA encoding the full-length protein (EcMcrB FL) was amplified by PCR and cloned into c2xP, a modified pMAL c2x vector with an HRV3C protease site replacing the Factor Xa site directly upstream of the mcrB gene. Native EcMcrB FL was expressed as N-terminal maltosebinding protein fusion in BL21(DE3) cells. Transformed cells were grown at 37°C in Terrific Broth to an A 600 of 0.8 -1.0, and then induced with 0.3 mM isopropyl 1-thio-␤-D-galactopyranoside overnight at 19°C. All cells were harvested, washed with amylose loading buffer (20 mM Tris-HCl, pH 8.0, 500 mM NaCl, 5% glycerol (v/v), 1 mM EDTA, and 1 mM DTT), and pelleted a second time. Pellets were typically flash frozen in liquid nitrogen and stored at Ϫ80°C.

Structure of H. pylori LlaJI.R1 N-terminal domain
Thawed pellets from 500-ml cultures were resuspended in 30 ml of amylose loading buffer supplemented with 10 mM phenylmethylsulfonyl fluoride, 5 mg of DNase (Roche), 5 mM MgCl 2 , and a Roche complete protease inhibitor mixture tablet (Roche). Lysozyme was added to 1 mg/ml and the mixture was incubated for 15 min rocking at 4°C. Cells were disrupted by sonication and the lysate was cleared of debris by centrifugation at 13,000 rpm (19,685 ϫ g) for 30 min at 4°C. The supernatant was filtered, loaded onto 40 ml of packed amylose resin (New England Biolabs), and then washed with amylose loading buffer. EcMcrB FL was eluted with amylose loading buffer supplemented with 10 mM D-maltose. Hrv3C protease was added to pooled fractions and dialyzed overnight at 4°C into Q loading buffer (20 mM Tris-HCl, pH 8.0, 50 mM NaCl, 1 mM EDTA, 5% glycerol (v/v), and 1 mM DTT). The sample was applied to a 5-ml HiTrap Q HP column equilibrated with Q loading buffer and then washed with Q loading buffer. EcMcrB FL was eluted with a NaCl gradient from 50 mM to 1 M. Pooled fractions were concentrated and further purified by SEC using a Superdex 75 10/30pg column. All proteins were exchanged into a final buffer of 20 mM HEPES, pH 7.5, 150 mM KCl, 5 mM MgCl 2 , and 1 mM DTT during SEC and concentrated to 5-40 mg/ml.  (Table 1). Crystal 1 was of the space group P1 with unit cell dimensions a ϭ 37.47, b ϭ 44.39, c ϭ 84.78 Å and ␣ ϭ 98.04°, ␤ ϭ 94.37°, ␥ ϭ 98.52°and showed strong anomalous signal. Crystal 2 was of the space group P1 with unit cell dimensions a ϭ 37.53, b ϭ 43.77, c ϭ 85.09 Å and ␣ ϭ 97.87°, ␤ ϭ 93.86°, ␥ ϭ 97.77°. Both crystals were prepared in the same condition but exhibited mosaicities of 0.20871°and 0.11033°, respectively. Data were integrated and scaled using XDS (47) and AIMLESS (48) via the NE-CAT RAPD pipeline. Se-SAD phasing with the data from crystal 1 yielded an initial model that was incomplete and contained a few regions of ambiguity. Heavy atom sites were located using SHELX (49) and phasing, density modification, and initial model building was carried out using the Autobuild routines of the PHENIX package (50). Further cycles of model building and refinement were carried out manually in COOT (51) and PHENIX, respectively (50), but failed to improve significantly the density and refinement statistics. A more complete model was obtained using the diffraction data from crystal 2. The structure was solved by molecular replacement with PHASER (52) using the SAD-derived structure from crystal 1 as the search model. This vastly improved the resulting maps and statistics following subsequent rounds of manual model building and refinement. The final model of crystal 2 was refined to 1.97-Å resolution with R work /R free ϭ 0.1927/0.2233 (Table 1) and contained four molecules in the asymmetric unit: molecule A, residues 1-121; molecule B, residues 1-131; molecule C, residues 1-121; molecule D, residues 1-131. Threonine 57 exists as a Ramachandran outlier with relatively weak density in the ␤3-␤4 loop of molecules B and D, respectively, and could not be refined further. All structural models were rendered with PyMOL (Schrödinger, Inc.) and surface electrostatics were calculated with APBS (53).

SEC-MALS
Purified HpR1⌬136 at 4 mg/ml was subjected to size-exclusion chromatography using a Superdex 200 10/300 column (GE Healthcare) equilibrated in SEC buffer (20 mM HEPES, pH 7.5, 150 mM KCl, 5 mM MgCl 2 , and 1 mM DTT). The column was coupled to a static 18-angle light scattering detector (DAWN HELEOS-II) and a refractive index detector (Optilab T-rEX) (Wyatt Technology). Data were collected continuously at a flow rate of 0.5 ml/min. Data analysis was carried out using the program Astra VI. Monomeric BSA at 4 mg/ml (Sigma) was used for normalization of the light scattering detectors and data quality control.
Lyophilized single-stranded oligonucleotides were resuspended to 1 mM in 10 mM Tris-HCl and 1 mM EDTA and stored at Ϫ20°C until needed. Single-stranded oligonucleotides were 5Ј end-labeled with [␥-32 P]ATP using polynucleotide kinase (New England Biolabs) and then purified on a P-30 spin column (Bio-Rad) to remove unincorporated label. Duplex substrates were prepared by heating equimolar concentrations of complementary strands (denoted with suffixes "us" and "ls" indicating upper and lower strands) to 95°C for 15 min followed by cooling to room temperature overnight and then purification on an S-300 spin column (GE Healthcare) to remove ssDNA. Four duplex DNA substrates were prepared: a methylated EcMcrBspecific substrate, 5mC (5mC_us and 5mC_ls), a nonmethylated EcMcrB-specific substrate, nm (nm_us and nm_ls), a sitespecific substrate containing the L. lactis LlaJI.R1-binding site Ll (Ll_us and Ll_ls), and a substrate with the L. lactis LlaJI.R1binding site sequence scrambled as a control Llscr (Llscr_us and Llscr_ls).

Filter-binding assays
The standard buffer for the DNA-binding assays contained 25 mM MES, pH 6.5, 2.0 mM MgCl 2 , 0.1 mM DTT, 0.01 mM EDTA, and 40 g/ml BSA. Binding was performed with purified HpR1⌬136 (WT or mutants) or EcMcrB FL at 30°C for 10 min in a 30-l reaction mixture containing 14.5 nM unlabeled DNA and 0.5 nM labeled DNA. Samples were filtered through KOH-treated nitrocellulose filters (Whatman Protran BA 85, 0.45 M) using a Hoefer FH225V filtration device for ϳ1 min. Filters were subsequently analyzed by scintillation counting on a 2910TR digital, liquid scintillation counter (PerkinElmer Life Sciences). All measured values represent the average of at least two independent experiments and were compared with a negative control to determine fraction bound.

EMSA
The standard buffer for the EMSAs contained 10 mM Tris-HCl, pH 8.0, 100 mM NaCl, 1 mM MgCl 2 , and 1 mM DTT. Binding was performed with purified HpR1⌬136 (WT or mutants) at 25°C for 30 min in a 16-l reaction mixture containing 10 ng/l of N 6 -methyladenine-free -phage DNA (New England Biolabs) digested with BamHI and NdeI (New England Biolabs) and purified via a NucleoSpin Gel and PCR Clean-Up kit (Machery-Nagel). Following incubation, samples were analyzed by 0.7% agarose gel in 1ϫ TAE at 4°C and 80 V for 90 min. All gels were stained with SYBR Green in 1ϫ TAE overnight at 25°C (Thermo-Fisher Scientific) and visualized using a Bio-Rad Gel Doc TM EZ imager system.