Cofactor Requirement of HpyAV Restriction Endonuclease

Background Helicobacter pylori is the etiologic agent of common gastritis and a risk factor for gastric cancer. It is also one of the richest sources of Type II restriction-modification (R-M) systems in microorganisms. Principal Findings We have cloned, expressed and purified a new restriction endonuclease HpyAV from H. pylori strain 26695. We determined the HpyAV DNA recognition sequence and cleavage site as CCTTC 6/5. In addition, we found that HpyAV has a unique metal ion requirement: its cleavage activity is higher with transition metal ions than in Mg++. The special metal ion requirement of HpyAV can be attributed to the presence of a HNH catalytic site similar to ColE9 nuclease instead of the canonical PD-X-D/EXK catalytic site found in many other REases. Site-directed mutagenesis was carried out to verify the catalytic residues of HpyAV. Mutation of the conserved metal-binding Asn311 and His320 to alanine eliminated cleavage activity. HpyAV variant H295A displayed approximately 1% of wt activity. Conclusions/Significance Some HNH-type endonucleases have unique metal ion cofactor requirement for optimal activities. Homology modeling and site-directed mutagenesis confirmed that HpyAV is a member of the HNH nuclease family. The identification of catalytic residues in HpyAV paved the way for further engineering of the metal binding site. A survey of sequenced microbial genomes uncovered 10 putative R-M systems that show high sequence similarity to the HpyAV system, suggesting lateral transfer of a prototypic HpyAV-like R-M system among these microorganisms.


Introduction
Restriction-modification (R-M) systems that recognize and cleave DNA in a highly specific manner are ubiquitous in prokaryotic microorganisms (and their viruses) [1]. Helicobacter pylori, the etiologic agent of common gastritis and a risk factor for gastric cancer [2], curiously is one of the richest sources of Type II restriction-modification (R-M) systems in any living organisms [3,4]. The extraordinary number of Type II R-M systems appears to be the result of H. pylori's natural competency of transformation by exogenous DNA [4][5][6]. In addition to defense against invading phages, evidence has suggested that the MTases (within active R-M system [7] or orphan MTases [8,9]) are involved in transcriptional regulation of other genes akin to the epigenetics of mammalian cells.
Genome mining of sequenced microbial genomes has resulted in a wealth of restriction enzymes with new specificities or unique properties (ApeKI (G^CWGC), PhoI (GG^CC), CviKI-1 (RG^CY), NmeAIII (GCCGAG 20-21/18-19) [10,11], Nt.CviPII (^CCD) [12]; NEB catalog 2009/10) [1]. The goal of this work was to clone, express, purify and characterize HpyAV restriction endonuclease (REase), which is one of the putative R-M systems from H. pylori 26695 [3]. During the purification process, we found that Ni ++ has a stimulatory effect on HpyAV activity. Bioinformatics analysis showed that HpyAV contains a HNH catalytic site highly similar to that of colicin E9 (ColE9). Sequence alignment of HpyAV and ColE9 and other HNH nucleases identified four highly conserved catalytic residues. By site-directed mutagenesis we confirmed that these residues are important for DNA cleavage. In addition to Ni ++ , we found that HpyAV is also active in Mn ++ and Co ++ . We therefore surveyed a few other HNH REases and found that KpnI is also active in a multitude of transition metals. Finally, a BLASTP search in sequenced bacterial genomes revealed ten putative HpyAV R-M systems. These microorganisms reside within human bodies or in mammals that are closely associated with humans, suggesting a possible lateral transfer mechanism.

The HpyAV R-M System
Restriction mapping and run-off sequencing results indicated that the native HpyAV REase isolated from Helicobacter pylori strain 26695 recognizes the asymmetric target sequence CCTTC and cleaves 6 nt and 5 nt downstream of the top strand and the bottom strand, respectively (CCTTC 6/5; data not shown). Enzymes that recognize asymmetric sequences frequently require two methyltransferases (MTases) to modify the two strands of DNA. In the case of HpyAV the MTase(s) must modify a C of the top strand and an A of the bottom strand in the target sequence. From the genomic sequence of H. pylori 26695 (Genbank nucleotide accession NC_000915), the R gene (hp_0053) of the HpyAV R-M system is located downstream of the M gene (hp_0054) and runs in the same direction as the M gene (Fig. 1A). In addition, the M gene of the HpyAV R-M system is a fusion of a C5 cytosine MTase and a N6 adenine MTase highly homologous to M1.Hin4II and M2.Hin4II, respectively (Fig. 1A). We re-sequenced the junction of the two MTase domains from the cloned M gene and from a PCR product derived from the genomic DNA and found no stop codon Figure 1. Gene organization of the HpyAV/Hin4II R-M systems and the structural model of the HpyAV catalytic site. A. Organization of the HpyAV and Hin4II R-M systems. The Hin4II R-M system consists of separate M1 and M2 genes for C5 cytosine (dark grey) and N6 adenine (light grey) methylation, respectively, preceding the ORF of Hin4II [80]. M.HpyAV is a fusion of C5 and N6A MTase domains with high sequence similarity to M1.Hin4II and M2.Hin4II, respectively. B. The structural model of HpyAV catalytic site and structural alignment to HNH endonucleases. Modeling of amino acid residues 281 to 360 of HpyAV to the bba-Me motif of ColE9 and the structural alignment to ColE9, I-HmuI, Hpy99I and KpnI were done as described in Materials and Methods. The blue dots underneath the alignment indicate metal-binding residues; the red dots indicate the general base His and the green dot indicates the Asn implied to decrease the pKa of the general base His in ColE9 and I-HmuI. Amino acid residues that are assigned to b-strands and the helix of the bba-Me motif are indicated by red arrows and blue rods, respectively. The conserved residues implicated in catalysis are colored in black or grey. doi:10.1371/journal.pone.0009071.g001 between the two domains, confirming that M.HpyAV is a true fusion of C5 cytosine MTase and N6 adenine MTase, although the size of the translation product has not been confirmed biochemically. An over-expression E. coli strain was constructed by transforming E. coli ER3081 (NEB) with pSYX20-hpyAVM and pAII17-hpyAVR by sequential transformation (See Materials and Methods).

Metal Ion Requirement for DNA Cleavage
The purified recombinant HpyAV exhibited very low cleavage activity on l DNA in the standard reaction condition containing 4 mM MgSO 4 ( Fig. 2A). However, when 2 mM NiSO 4 was added into the cleavage reaction in the presence or absence of 4 mM MgSO 4 , HpyAV exhibited equally high cleavage activity ( Fig. 2A). This shows that HpyAV prefers Ni ++ for cleavage activity. This discovery prompted us to examine HpyAV against other divalent metal ions including Ca ++ , an alkaline earth metal ion that is inhibitory to REases containing the canonical PD-X-(D/E)XK catalytic motif, and ions of other transition metals in the same period (Mn ++ , Co ++ , Cu ++ and Zn ++ ). We found that HpyAV showed a different degree of cleavage activity with divalent metal ions. It was most active with MnCl 2 , NiSO 4 or CoSO 4 -complete cleavage of l DNA was achieved with 0.5 to 4 mM of these three metal ions (Fig. 2B). For Cu(OAc) 2 and Zn(OAc) 2 , concentrations higher than 2 mM were inhibitory to HpyAV endonuclease activity (data not shown), and complete cleavage was not obtained under the assay conditions. HpyAV showed much lower activity in the presence of CaCl 2 or MgSO 4 (Fig. 2B). Table 1 summarizes the specific activity of HpyAV with various metal ions. HpyAV is equally active in MnCl 2 , CoSO 4 and NiSO 4 (specific activities are within a 2-fold margin for a 2-fold dilution series of the enzyme), and complete cleavage of l DNA was not achieved in the presence of MgSO 4 , CaCl 2 , Cu(OAc) 2 or Zn(OAc) 2 at the highest enzyme concentration available (40 mmol of HpyAV on 0.3 pmol (1 mg) of l DNA). By comparing the cleavage patterns, HpyAV is estimated to exhibit less that 0.4% of cleavage activity in buffers with MgSO 4 and CaCl 2 , less than 6% with Cu(OAc) 2 and less than 0.8% with Zn(OAc) 2 .

Homology Modeling of Catalytic Residues
The strong stimulation effect of Ni ++ on endonuclease activity is unique to HpyAV. It led us to examine the HpyAV amino acid sequence in more details. HpyAV is not homologous to any known REases except its isoschizomer Hin4II (data not shown). Manual examination of the amino acid sequence of HpyAV revealed a HNH catalytic motif highly homologous to that of colicin E9. Homology modeling of amino acid (aa) residues 287-325 of HpyAV to the bba-Me motif of ColE9 (aa 95-131) resulted in a model free of clashes and with all the conserved catalytic residues (His102, His103, Asn118 and His127 in ColE9; H294, H295, N311, and H320 in HpyAV) structurally aligned to the HNH endonucleases including I-HmuI and Hpy99I (Fig. 1B).

Site-Directed Mutagenesis of the HNH Catalytic Site
From biochemical and structural studies of colicin E9, His103 acts as the general base to deprotonate a water molecule for the hydrolysis of the scissile phosphodiester bond. His102 and His127 coordinate the single divalent metal ion for transition state stabilization. Asn87 of I-HmuI and Asn118 of ColE9 are proposed to form a hydrogen bond to the general base His and increase its pKa for the activation of the nucleophilic water [13,14]. In this study, the corresponding residues of HpyAV (His294, His295, Asn311 and His320) were mutated to verify their role in catalysis. Mutants H294D, H295A, and H320A were constructed and purified. H294D and H320A did not show any cleavage activity at up to 7.5 mg of protein ( Fig. 3 and data not shown) in the presence of 2 mM NiSO 4 or MgSO 4 , indicating that (i) the removal of the imidazole group at position 320 eliminated cleavage activity; (ii) the negatively charged Asp (as found in I-HmuI, Hpy99I and KpnI at the same aa position; Fig. 1B) cannot replace the histidine residue at position 294 for metal coordination in HpyAV. It is somewhat unexpected that substitution of the general base His295 by Ala did not completely eliminate the cleavage activity (Fig. 3); H295A still retains approximately 1% of wt activity, suggesting that an alternative weaker general base exists in the catalytic site when the general base His295 is absent. To explore the consequence of other amino acid substitutions, we also mutated His295 to Lys, Asn or acidic resides Asp/Glu. IPTGinduced cell extracts expressing these four mutants (H295K, H295N, H295D, and H295E) did not show any cleavage activity (data not shown), indicating that Lys, Asn, Asp, or Glu residues cannot replace His295 in the catalytic site. Cell extract with  N311A variant failed to show any detectable cleavage activity (data not shown). It is concluded that His294, H295, Asn311 and His320 are important residues for HpyAV endonuclease activity.

KpnI Endonuclease Activity with Different Divalent Metal Ions
KpnI is an HNH Type IIP REase that has been shown to be active with Mg ++ or Ca ++ [15,16]. The high activity of HpyAV with transition metal ions prompted us to investigate if it is also true for KpnI. Fig. 4 shows that KpnI is active in all of the transition metals tested. Table 1 shows that the specific activity of KpnI is highest with MgSO 4 , followed by CoSO 4 (15% of MgSO 4 ), NiSO 4 and Zn(OAc) 2 (both 15% of MgSO 4 ) and CaCl 2 (2.5% of MgSO 4 ). The specific activity could not be determined with MnCl 2 because cleavage at non-cognate sites (star activity) was observed before complete cleavage of the cognate sites was achieved (data not shown). The specific activity in Cu(OAc) 2 was also not determined because the same incomplete cleavage pattern was observed over a 120-fold difference in enzyme concentration (data not shown). As a control, 10 U of EcoRI, a canonical PD-(D/E)XK Type IIP REase, was also tested. Under the assay conditions, EcoRI was most active in MgSO 4 , with very low level of activity in MnCl 2 and CoSO 4 and no activity in CaCl 2 , NiSO 4 , Cu(OAc) 2 or Zn(OAc) 2 . For all three enzymes, no cleavage activity was observed without the added divalent metal ions, indicating that all the activities observed were caused by the presence of the metal ion cofactors.

Homologous HpyAV Systems from Sequenced Microbial Genomes
HpyAV is an isoschizomer of Hin4II. M.HpyAV contains a C5 cytosine and a N6 adenine MTase domain highly homologous to M1 and M2.Hin4II (Fig. 1A). Sequence analysis showed that the corresponding regions of M.HpyAV are 57% and 56% identical to M1 and M2.Hin4II, respectively ( Table 2). The REases HpyAV and Hin4II also share 48% sequence identity ( Table 2). A BLASTP search of the GenBank genomes discovered 10 putative R-M systems that are highly homologous to the HpyAV system. The MTases and REases of these homologous systems, along with those of HpyAV and Hin4II, are shown in Table 2. Except for the Yersinia kristensenii and Vibrionales bacterium SWAT-3 systems, the M gene precedes the R gene with both of them oriented in the same direction. Also, like the HpyAV system, all of these R-M systems contain single MTase that are fusions of C5 cytosine and N6 adenine MTases. Their high sequence similarity suggests that these putative R-M systems may share the same recognition sequence (CCTTC). It is also noticeable that these homologous R-M systems are mostly carried by infectious microorganisms of human or mammalian hosts closely associated with humans. However, there are two putative endonucleases (EsaSS23P = 393 aa; EsaSS44P = 385 aa) without companion MTases from shotgun-sequenced environmental samples that share significant amino acid sequence identity with HpyAV and Hin4II (EsaSS23P vs. HpyAV = 34% aa sequence identity; EsaSS44P vs Hin4II = 29% aa sequence identity). They may recognize CCTTC or similar target site with 1-bp difference.

Metal Ion Cofactor Preference of HNH Endonucleases
It has been well-documented that for restriction endonucleases (REases) with the canonical PD-X-(D/E)XK catalytic site, Mg ++ and Mn ++ support catalysis but Ca ++ only supports DNA binding. One to two coordinated metal ions have been observed in the catalytic site in the crystal structures of REases in the presence of substrate DNA. Catalytic mechanisms for one-and two-ioninduced cleavage have been proposed [17][18][19][20]. It has been generally accepted that most Type IIP REases use a two-metal ion mechanism where metal ion A deprotonates the nucleophilic water molecule and metal ion B stabilizes the pentavalent phosphoanion transition state and activates a water molecule for protonation of the leaving 39-phosphate oxygen. Some REases (EcoRI and BglII), however, appear to use a one-metal ion mechanism where the metal ion occupies site 1 and coordinates a nucleophilic water molecule for the attack of the scissile phosphate, although it has been noted that the second metal ions might have escaped detection because the second metal ions tend to have low occupancy in other structures [21][22][23]. Recently, Pingoud and colleagues presented experimental evidence and MD simulation results to support a generalized one-metal ion catalytic mechanism where site 1 has a higher affinity to Mg ++ and site 2 plays a modulating role in the cleavage activity [23].
Endonucleases that contain the bba-Me motif, on the other hand, are only observed with one coordinated divalent metal ion in their catalytic sites. The bba-Me motif adopts a V-shape conformation consisting of two beta-strands connected by a loop in one arm followed by a helix that constitutes the other arm. The bba-Me motif is present in non-specific endonucleases such as Serratia nuclease [24,25], E. coli defense nucleases colicin E7 [26,27] and E9 [28,29], Holliday junction resolving T4 endonuclease VII [30] and homing endonucleases I-PpoI of the His-Cys family [31,32]. HNH enzymes are a sub-group of the bba-Me family where the metal ion is coordinated by two negatively charged amino acid residues (two histidines in ColE9 [29,33,34]; a glutamate and a asparagine in I-HmuI [13,14] and Hpy99I [35]) and the non-bridging oxygen of the scissile phosphodiester bond of the transition state. In His-Cys homing endonucleases and Serratia nuclease, only one asparagine is involved in metal ion coordination. The coordinated metal ion is believed to stabilize the transition state by neutralizing the negatively charged pentavalent phosphoanion transition state. In ColE9, it has been proposed that the water molecule coordinated by the metal ion and His131 acts as the general acid that protonates the leaving group [29]. The conserved asparagine lowers the pKa of the invariable histidine which in turn activates the nucleophilic water molecule for in-line attack of the scissile phosphate. Recently, more Type II restriction endonucleases, namely, KpnI [36], MnlI [37], Hpy99I [35], Eco31I [38,39], HphI [40], SphI [41], PacI and others [42] are identified as containing this HNH motif through X-ray crystallography or sequence alignment/structural prediction. GIY-YIG endonucleases (including homing endonucleases I-TevI [43,44], nucleotide excise repair enzyme UvrC and Type IIP REases Hpy188I [45] Eco29kI [46,47] and Cfr42I [47,48]) is proposed to adapt a similar catalytic mechanisms as HNH/His-Cys endonuclease except for the use of Tyr as the general base based on the structure of UvrC [49].
In this study, we demonstrated the importance of conserved catalytic residues H294, H295, N311, and H320 by site-directed mutagenesis. HpyAV variants H294D, N311A, and H320A lack any detectable catalytic activity (less than 0.1% activity). Substitution of the general base His295 by Ala displays impaired cleavage activity only (H295A retains approximately 1% of wt activity), suggesting that an alternative weaker general base exists in the catalytic site when the general base His295 is mutated. In addition, positively charged aa substitution by Lys in H295K did not rescue the catalytic activity, indicating Lys cannot replace His as the general base in HpyAV. Other charged aa residue substitutions at the 295 position (HpyAV variants H295N,  H295D, and H295E) failed to show any detectable cleavage activity. The observation that deletion of the proposed histidine general base in the HpyAV active site yields an endonuclease with reduced activity, rather than completely inactivating the enzyme, is unusual for the HNH superfamily, but not unprecedented. Mutation of the active histidine general base in the I-PpoI homing endonuclease also results in reduced activity [50]; this result is attributed to the ability of a neighboring histidine to participate in a less-efficient long-range proton transfer reaction and activation of the nucleophilic water with reduced activity as compared to the wild-type enzyme.
MnlI and I-PpoI are the only bba-Me endonuclease whose metal ion preference has been systematically studied. In the presence of 1 mM M ++ , the order of MnlI activity was shown to be Mg ++ . Ni ++ = Co ++ . Mn ++ . Ca ++ . Zn ++ [51]. I-PpoI activity follows the order of Mg ++ . Mn ++ . Ca ++ = Co ++ . Ni ++ . Zn ++ (10 mM) [52]. Hpy99I is active in Mg ++ and Mn ++ but not in Ca ++ or Zn ++ [35]. It has also been reported that ColE9 prefers Mg ++ and Ca ++ for dsDNA and Ni ++ for ssDNA substrates [18,34,53]. Our preliminary results showed that PacI and SphI are active with Ni ++ but less so than in Mg ++ , whereas HphI showed comparable activity with Ni ++ and with Mg ++ (data not shown). Non-specific HNH endonucleases colicin E9 has also been reported to have distinct metal ion preference: Mg ++ and Ca ++ are most efficient cofactors for cleavage of double-strand DNA but Ni ++ is most efficient for cleavage of single-strand DNA; and colicin E9 cleaves RNA in the absence of any divalent metal ions [34]. In addition, although Serratia endonuclease is most active with Mg++, mutants that are more active in Mn++, Co++ and Zn++ have been isolated [54]. Compared to our results reported here, where HpyAV activity follows the order of Mn ++ = Ni ++ = Co ++ . Cu ++ . Zn ++ . Ca ++ . Mg ++ (2 mM for Mg ++ , Ca ++ , Mn ++ , Ni ++ and Co ++ ; 0.1 mM for Cu ++ and Zn ++ ) and KpnI activity follows the order of Mg ++ . Co ++ . Ni ++ = Zn ++ . Ca ++ , it appears that HNH endonucleases in general have a less stringent metal ion requirement than their counterparts with the canonical PD-X-(D/E)XK catalytic motif.
HpyAV is unique in that it is the only HNH-type endonuclease characterized so far to have higher double-strand cleavage activity with transition metals (Mn ++ , Ni ++ and Co ++ ) than with Mg ++ . It has been argued that in bba-Me REases, the metal ion is not involved in the coordination of the nucleophilic water but only interacts with the phosphoanion transition state and the leaving group through a water molecule, therefore allowing for a less stringent metal ion requirement for catalysis [55,56]. This property also raises the caution that transition metal ions should be taken into consideration when the cleavage activity of HNH enzymes is to be optimized.
The loss of fidelity of KpnI in the presence of Mn ++ has also been observed in EcoRI [57], PstI [58], EcoRV [59] and CeqI [60] and other DNA enzymes such as HIV ribonuclease H [61], the translesion DNA polymerase Dpo4 [62] and Tn10 in IS10 transposition [63]. This effect has been attributed to the similar chemical nature of Mn ++ to Mg ++ but a relaxed coordination requirement for Mn ++ [55], suggesting a role of the metal ion for the specificity of these REases. In EcoRV, it has also been demonstrated that the affinity for Mg ++ is lower when the enzyme is bound to non-cognate sites [64]. Interestingly, HpyAV does not show increased non-cognate cleavage activity in Mn ++ under our assay conditions, possibly because unlike Type IIP REases, target site recognition and phosphodiester bond cleavage of Type IIS REases such as HpyAV are uncoupled by separate DNA recognition and cleavage domains. The different metal ion preference of different HNH endonuclease is an interesting phenomenon, given their highly similar set of metal coordinators and general base. Further genetic, biochemical and structural studies of HNH enzymes are needed to understand the catalytic role of different metal ions. For example, it is possible to target the metal binding region by localized saturation mutagenesis of HpyAV by construction of a plasmid mutant library in the presence of methylase protection and then transfer the mutant library DNA into non-modified dinD::lacZ indicator strain and screen for blue colonies on X-gal plates supplemented with high concentration of Mg ++ (active mutants will damage chromosomal DNA and induce SOS-induction in vivo). Such active HpyAV mutants may contain altered metal binding site with preference for Mg ++ as a cofactor. KpnI mutants with altered metal binding and preference have been isolated (SHC and SYX, unpublished results).

Distribution of HpyAV Homologous Systems
In addition to Hin4II, a BLASTP search of GenBank database discovered 10 putative R-M systems highly homologous to the HpyAV system. Interestingly, these R-M systems are mostly carried by infectious microorganisms of human or mammalian hosts closely associated with humans ( Table 2). While Helicobacter pylori strains where HpyAV and the putative HpyPORF28P and HpyGORF49P R-M systems reside, are the etiological agent of common gastritis and a risk factor for gastric cancer, Campylobacter upsaliensis, Yersinia kristensenii and Vibrionales bacterium are mainly zoonotic but opportunistic pathogens of humans. Bacteroides stercoris is a symbiotic bacterium in the human intestines that helps to digest food. Haemophilus influenzae, from which Hin4II is isolated, is found in the upper respiratory tract of humans; it can cause bacteremia, pneumonia and acute bacterial meningitis. Neisseria meningitidis causes meningitis in humans. Other HpyAV homologous systems are found in microbes that infect pigs (enteric Brachyspira hyodysenteriae) [65], cattle, goats (Mycoplasma mycoides through inhalation) [66,67] and horses (Streptococcus equi that causes strangles) [68]. It is possible that these pathogenic microbes acquired a prototypical R-M system through lateral transfer when they were brought in the vicinity of a mammalian host organism, which need not be permissive to the microorganisms concerned [69,70]; or they could be transferred during food intake. It has been shown that the virulent factor iceA1 is a functional isoschizomer of NlaIII in H. pylori CH4 [71,72]. Therefore, the acquired HpyAV homologous systems may provide survival advantage to the receiving microorganisms. It is noted, however, that whole-genome sequencing efforts have largely been focused on mammalian pathogens and their sequences are over-repre-sented in sequence databases. It is possible that HpyAV homologous R-M systems exist in non-mammalian-associated microorganisms. Shotgun sequencing of marine samples has revealed two ORFs EsaSS23P and EsaSS44P that are also homologous to HpyAV and Hin4II aa sequences.

Strains, DNA Sequences and Mutagenesis
The HpyAV R-M system was identified in Helicobacter pylori 26695 (Genbank nucleotide accession NC_000915). ORF hp_0054 is the M gene which was amplified in PCR and inserted into pSYX20 at the EcoRV and SphI sites with a GGAGGT ribosome-binding site and upstream stop codons in all three ORFs (pSYX20 carries pSC101 replication origin, Km R , and Tc R ). Expression of the M gene is under the control of the Tc R promoter. ORF hp_0053 is the R gene which was amplified in PCR and inserted into pAII17 (NEB) at NdeI and BamHI sites, under the control of the T7 promoter. The over-expression strain was constructed by sequential transformation of E. coli ER3081 (NEB) by pSYXS20-hpyAVM and then pAII17-hpyAVR. ER3081 (fhuA2 8 lacZ::T7 gene1 [lon] ompT gal attB::pCD13(Ptet-lysY, lacI q ) [Spec R ] sulA11 R(mcr-73::miniTn10-Tet S )2 [dcm] R(zgb-210::Tn10 -Tet S ) endA1 D(mcrC-mrr)114::IS10) is a derivative of ER2566 (T7 Express, NEB). This strain contains the T7 RNA polymerase gene at the chromosomal lac locus, replacing much of lacZY; the K128Y mutant of T7 lysozyme (lysY) and the lacI q gene are expressed from the chromosomal attB site. Stable integration of the lysY and lacI q genes was accomplished using the pCD13PKS plasmid described by Platt et al. [73,74]. Site-directed mutagenesis was carried out using a modified inverse PCR procedure [75] using pAII17-hpyAVR isolated from the over-expression strain as template. Primers designed to construct mutants (H295A, H320A and N311A) were synthesized by Integrated DNA Technologies. All DNA sequences were verified by DNA sequencing.

Protein Expression and Purification
The over-expression strain of HpyAV was cultured in LB medium containing 100 mg/ml ampicillin and 30 mg/ml kanamycin at 30uC and 200 rpm overnight (,15 h). Ten milliliters of the overnight culture was inoculated into 1 L of LB medium containing the same antibiotics and cultured at 30uC and 200 rpm to log phase. The culture was cooled down to 25uC before IPTG was added to a final concentration of 0.25 mM. Growth was then continued at 25uC for ,15 h and the cultures were harvested by centrifugation. The cell pellet was resuspended in 100 ml of 20 mM Tris-HCl, pH 8.0, 50 mM NaCl, 1 mM EDTA (Buffer A) supplemented with 1% PMSF and sonicated on ice. After centrifugation, the supernatant was loaded onto a Heparin HiTrap column (5 ml; GE Life Sciences). Peak fractions from a linear elution gradient of 0.05-1 M NaCl in Buffer A was diluted 4-fold in Buffer A and loaded onto a HiTrap SP HP column (5 ml; GE Life Sciences). Peak fractions from a 0.05-1 M NaCl gradient were pooled and concentrated by Vivaspin 15 (10 kDa MWCO; Sartorius). An equal volume of 60% glycerol was added to the concentrated protein for storage at 220uC.

DNA Cleavage Activity Assays
The DNA cleavage activity of the crude extract or purified HpyAV was assayed in 50 ml reactions containing 20 mM Tris-HCl, pH 7.9, 200 mM NaCl supplemented with the indicated concentrations of MgSO 4 , CaCl 2 , MnCl 2 , CoSO 4 , NiSO 4 , Cu(OAc) 2 or Zn(OAc) 2 and 1 mg of l DNA at 37uC for 1 h. KpnI was assayed in 20 mM Tris-HCl, 50 mM NaCl, pH 7.9 with the same battery of salts using 1 mg of pXba DNA (a 10 kb XbaI fragment of adenovirus DNA inserted into pUC19; NEB). EcoRI activity assay was carried out in the same buffer using 1 mg of l DNA. The cleavage reactions were then analyzed by 1.2% agarose gel electrophoresis in 1x TBE. One enzyme unit is defined as the amount of enzyme needed to cleave the 1 mg of the designated DNA completely at 37uC in 1 h. Specific activity is defined as the number of units per mg of enzyme. Specific activity was determined in duplicate (KpnI) or triplicate (HpyAV) by titrating the enzymes (in steps of 2-fold dilution) in their respective reaction buffer in the presence of 2 mM MgSO 4 , CaCl 2 , MnCl 2 , CoSO 4 or NiSO 4 , or 0.1 mM Cu(OAc) 2 or Zn(OAc) 2 . KpnI and EcoRI were from NEB. All reaction buffers and metal ion solutions were prepared using MilliQ water.