Mutational Analysis of Two Putative Catalytic Motifs of the Type IV Restriction Endonuclease Eco 57I*

The role of two sequence motifs (SM) as putative cleavage catalytic centers 77 PD X 13 EAK (SM I) and 811 PD X 20 DQK (SM II) of type IV restriction endonucle- ase Eco 57I was studied by site-directed mutational analysis. Substitutions within SM I; D78N, D78A, D78K, and E92Q reduced cleavage activity of Eco 57I to a level undetectable both in vivo and in vitro. Residual endonucleolytic activity of the E92Q mutant was detected only when the Mg 2 1 in the standard reaction mixture was replaced with Mn 2 1 . The mutants D78N and E92Q retained the ability to interact with DNA specifically. The mutants also retained DNA methylation activity of Eco 57I. The properties of the SM I mutants indicate that Asp 78 and Glu 92 residues are essential for cleavage activity of the Eco 57I, suggesting that the sequence motif 77 PD X 13 EAK represents the cleavage active site of this endonuclease. Eco 57I mutants containing single amino acid substitutions within SM II (D812A, D833N, D833A) revealed only a small or moderate decrease of cleavage activity as compared with wild-type Eco 57I, indicating that the SM II motif does not represent the catalytic center of Eco 57I. The results, taken together, allow us to conclude that the Eco 57I restriction endonuclease has one catalytic center for cleavage of DNA. fragment was excised from the phagemid with Eco RI and Hin dIII restriction endonucleases. Both specific and nonspecific fragments were gel purified and radiolabeled using Klenow polymerase to fill in their 5 9 extensions with [ a - 33 P]dATP and the other three dNTPs (16). DNA concentrations were determined spectrophotometrically. Gel Mobility Shift Assay— Binding reactions (20 m l) contained the 33 P-endlabeled- specific or -nonspecific DNA fragment (final concentra- tion 10 p M ) and the wt Eco 57I or mutant protein (final concentration in the range of 0–10 n M ). The incubation was performed in binding buffer (10 m M Tris-HCl, pH 8.0, 50 m M NaCl, 10 m M CaCl 2 , 0.1 mg/ml bovine serum albumin, 10% glycerol) for 20 min at room temperature, and then the samples were applied to a 6% polyacrylamide gel (29:1 acrylamide/ bis). Electrophoresis was carried out at 11 V/cm at room temperature in 40 m M Tris acetate, 10 m M CaCl 2 , pH 8.0 for 2 h. After electrophoresis, the gels were dried, and radioactive bands were visualized using the OptiQuant TM Image Analysis Software (Pacard). Binding of Wt and D78N, E92Q Mutant Proteins to the Eco57I Target Site— The effects of mutations on binding of the mutant Eco 57I proteins to DNA were examined using the gel mobility shift assay. Two DNA fragments were used: a 210-bp DNA fragment containing one Eco 57I site in the middle of the sequence (specific DNA) and a nonspecific DNA fragment, which had the same sequence, except it lacked the Eco 57I site as a result of a 1-bp substitution. The experiments were performed with purified wild-type Eco 57I, D78N, and E92Q mutant proteins. In initial experiments, the DNA binding for the wild-type Eco 57I restriction endonuclease was characterized. Increasing of Eco 57I endonuclease of 33 P-endlabeled specific or -nonspecific DNA free shift

PDX 13 EAK represents the cleavage active site of this endonuclease. Eco57I mutants containing single amino acid substitutions within SM II (D812A, D833N, D833A) revealed only a small or moderate decrease of cleavage activity as compared with wild-type Eco57I, indicating that the SM II motif does not represent the catalytic center of Eco57I. The results, taken together, allow us to conclude that the Eco57I restriction endonuclease has one catalytic center for cleavage of DNA.
Nearly 3000 restriction endonucleases with over 200 different specificities, which together with cognate DNA methyltransferases constitute restriction-modification (R-M) 1 systems, have been identified in bacteria (1). Restrictionmodification enzymes are traditionally divided into three classes designated type I, II, and III on the basis of enzyme subunit composition, cofactor requirements, substrate specificity characteristics, and reaction products (2). An increasing number of restriction endonucleases that do not fit into the conventional classification however have been reported (3)(4)(5)(6)(7). Their differences from the type I and type III enzymes are so substantial that a classification as new kinds of restriction endonucleases; type IIS, type IIT, type IV, and Bcg-like has been suggested (3)(4)(5)(6)(7)(8).
The type IV restriction endonuclease Eco57I has been stud-ied in detail (4). Similar to type IIS endonucleases, it recognizes an asymmetric nucleotide sequence, cleaves both DNA strands outside the target site 5Ј-CTGAAG(N) 16/142 and exists in solution as a monomer. Other features of Eco57I, however, such as stimulation of endonucleolytic reaction with the DNA methyltransferase cofactor S-adenosyl-L-methionine (AdoMet) and methylation of one strand of the recognition duplex, makes it similar to type III enzymes. Both endonucleolytic and methylation activities reside within a single large polypeptide of the enzyme. In addition to the bifunctional restriction endonuclease, the Eco57I R-M system also includes a separate Eco57I methyltransferase, which modifies both DNA strands of the target duplex. The methylation domain has been previously assigned to the carboxyl-half of the Eco57I restriction endonuclease, where conserved amino acid sequence motifs typical for m 6 A DNA methyltransferases involved in AdoMet binding and catalysis of methyl group transfer are located (9). The location and identity of the endonuclease active center though remains to be determined and is addressed here. In contrast to DNA methyltransferases, the amino acid sequences of restriction endonucleases share little similarity. This observation therefore reduces the possibility of identifying catalytic sites of restriction enzymes on the basis of sequence alignment. Structural and mutational analysis of type II restriction endonucleases revealed however the PDX n (D/E)XK motif as a catalytic/Mg 2ϩ binding signature motif (8,10,11). Two putative catalytic/Mg 2ϩ binding motifs (i.e. 77 PDX 13 EAK and 811 PDX 20 DQK, located in the N-terminal and C-terminal parts of Eco57I, respectively) have been described in the amino acid sequence of the enzyme (10). The statistical significance of these motifs however is low, and their presence does not allow unambiguous prediction of the active site, as is evidenced by the following observations. (i) Cfr10I contains the PDX n (D/ E)XK motif, but it is not part of its catalytic center (12) and (ii) EcoRI contains two such motifs, one of which is not involved in catalysis (10).
On the other hand it cannot be excluded that two active centers are necessary for monomeric Eco57I to cleave both DNA strands. The asymmetric nature of the Eco57I target sequence is inconsistent with the use of a symmetric dimer for recognition and DNA cleavage, as in the type II restriction endonucleases. A single molecule containing two endonucleolytic centers could cleave both DNA strands. It has been suggested that a molecule recognizing an asymmetric nucleotide sequence with a single catalytic center must rearrange the catalytic center for sequential cleavage of each DNA strand, or it must form a higher order complex to cleave both strands of DNA (13). This second mechanism is utilized by the restriction endonuclease FokI, the only type IIS enzyme characterized in this respect so far (14). The identification of the Eco57I catalytic center(s) would increase our understanding of the functional organization of a unique enzyme, which shares proper-ties with type IIS and type III enzymes. We therefore constructed single amino acid substitutions in the putative catalytic motifs of Eco57I to determine the role of the two putative active centers, if any, in DNA cleavage. The properties of the mutants suggest that of the two putative catalytic motifs, only the motif 77 PDX 13 EAK is involved in DNA cleavage catalysis.

EXPERIMENTAL PROCEDURES
Bacterial Strains, Plasmids, Phage, and Media-The Escherichia coli strain ER2267 was used as host for cloning procedures and was used to assess the endonuclease activity of the wt Eco57I and mutants in vivo. It was kindly provided by E. Raleigh. The E. coli CJ236 strain of genotype FЈ cat (ϭ pCJ105; M13 s Cm R )/dut ung1 thi-1 relA1 spoT1 mcrA was used to prepare single-stranded DNA for site-directed mutagenesis. The E. coli BL21 strain (Novagen) was used for the expression of the wild-type eco57IR and mutant genes. The pUC19-based phagemid pTZ19R (15) was used as a vector in site-directed mutagenesis experiments and DNA sequencing. The plasmid pEco57IR3.6 (Ap R ) was constructed by subcloning eco57IR from pEco57IRM6.3 (9) into the pET-21b expression vector (Novagen) in the orientation coinciding with that of the T7 promoter. It was used for the construction of mutant eco57IR genes and their expression for purposes of protein purification. Plasmid pEco57IM3.3 (Cm R ) carries the Eco57I methyltransferase gene cloned in the vector pACYC184 (9). Transformations of E. coli were carried out by the CaCl 2 heat shock method (16). All strains were grown in Luria-Bertani medium at 37°C. The following concentrations of antibiotics were used when necessary: ampicillin (Ap), 60 g/ml; kanamycin (Km), 50 g/ml; chloramphenicol (Cm), 30 g/ml. vir was used to test the in vivo function of the wild-type Eco57I and mutants. Stocks of this phage were prepared according to Sambrook et al. (16).
Enzymes, Chemicals, and Oligonucleotides-All enzymes, including a homogeneous preparation of wt Eco57I, kits, and DNA were provided by MBI Fermentas and used according to the manufacturer's recommendations. [␣-33 P]dATP and [␥-33 P]ATP were purchased from Amersham Pharmacia Biotech. Synthetic oligonucleotides were synthesized at the facilities of MBI Fermentas. All other chemicals were reagent grade commercial products.
Analysis of Viability of Strains Containing Wt and Mutant Eco57I-The ability of strains harboring wt or mutant Eco57I to survive in the presence or absence of Eco57I methylase was tested by transforming E. coli strain ER2267 containing or lacking pEco57IM3.3 with plasmids carrying either wt or mutant eco57IR genes. An aliquot of 0.5 g of each plasmid DNA (in a total volume of 10 l) was used to transform a 100-l aliquot of competent cells. Ap R or Ap R Cm R transformants were selected. The transformation efficiency of competent cells was tested by transforming them with the control plasmid pBR322. Two independent transformation experiments were carried out.
DNA Preparation and Manipulation-Plasmids were prepared by the alkaline lysis procedure (17) and purified additionally as described by Marko et al. (18). Recombinant plasmid construction and isolation of DNA fragments from agarose gels were performed according to standard techniques (16). DNA sequencing was carried out by the chain termination method (19).
Site-directed Mutagenesis-To generate the single-stranded DNA needed for mutagenesis, a 180-bp Eco88I-Bst1107I and 680-bp PstI-Eco105I DNA fragments of the eco57IR gene (GenBank TM /EBI accession no. X61122) containing N-terminal (SM I) and C-terminal (SM II) putative catalytic motifs, respectively were subcloned into the pTZ19R phagemid, which was then multiplied in the CJ236 strain. Site-directed mutants were obtained by oligonucleotide mutagenesis using the method of Kunkel et al. (20). Eco57I mutant proteins that differ from the wild type by only a single amino acid were made using the corresponding oligonucleotides: D78N, TAAAAAGCCAAACTACACG; D78A, TAAAAAGCCAGCCTACACGT; D78K, TAAAAAGCCAAAGTACACG-TTT; E92Q, TTTTTCCTTCAAGCCAAA; D812A, TATTAAGCCGGCG-CCAACTGGC; D833N, CTGCGATGTTAACCAGAAGCT; D833A, CTG-CGATGTTGCGCAGAAGCT. After introducing changes verified by DNA sequencing, the fragments of the eco57IR containing desired point mutations were exchanged with the corresponding fragments of the wild-type gene in the pEco57IR3.6. The integrity of subcloning sites was checked by restriction analysis.
Gel Electrophoresis of Proteins-Gel electrophoresis of proteins under denaturing conditions was performed as previously described (21). SDS-PAGE was carried out on a 7.5% separating gel. Protein bands were visualized after Coomassie Blue R250 staining.
Purification of Mutant and Wild-type Eco57I Endonucleases-E. coli BL21, freshly transformed by the plasmid containing one of the mutant eco57IR genes, was used as the source of the mutant enzyme. Expression was induced by adjusting the culture to 1 mM isopropyl-1-thio-␤-D-galactopyranoside at an A 600 of about 0.6. After 3 h, the cells were chilled on ice, harvested by centrifugation and stored at Ϫ20°C. All further steps were carried out at ϩ4°C. SDS-PAGE electrophoresis was used throughout all purification steps to identify chromatographic fractions containing mutant Eco57I proteins. The purification procedure described below was used for isolation of all mutant proteins used in experiments.
Twenty five grams of frozen cells were thawed and suspended in 80 ml of buffer A (10 mM potassium phosphate, pH 7.4, 1 mM EDTA, 7 mM ␤-mercaptoethanol) containing 0.1 M NaCl. Cells were disrupted by sonication and cell debris was removed by centrifugation. The supernatant was applied to a Heparin-Sepharose column (1.5 ϫ 30 cm) equilibrated with buffer A containing 0.1 M NaCl. The column was washed with the same buffer and eluted with a 400-ml linear gradient of 0.1-1.0 M NaCl in buffer A. Fractions containing the mutant Eco57I protein, as determined by SDS-PAGE electrophoresis, eluted at ϳ0.28 -0.44 M NaCl. They were pooled and dialyzed against buffer A containing 0.05 M NaCl and applied to a Sepharose Q column (1.5 ϫ 20 cm). The column was washed with the same buffer and eluted with a 340-ml linear gradient of 0.05-0.3 M NaCl in buffer A. The peak fractions, which eluted at ϳ0.12-0.17 M NaCl were pooled and dialyzed against buffer A containing 0.1 M NaCl and applied to an AH-Sepharose column (1.5 ϫ 9 cm). After the column was washed with the same buffer, sample was eluted with a 200-ml linear gradient of 0.1-1.0 M NaCl in buffer A. The peak fractions (eluted at 0.31-0.38 M NaCl) were pooled, dialyzed against the storage buffer (10 mM potassium phosphate, pH 7.4, 100 mM NaCl, 1 mM EDTA, 7 mM ␤-mercaptoethanol, and 50% glycerol) and stored at Ϫ20°C. Essentially the same procedure was used for purification of wt Eco57I, which was kindly provided by MBI Fermentas. The proteins were homogeneous as judged by polyacrylamide gel electrophoresis. Protein concentrations were determined spectrophotometrically at 280 nm using an extinction coefficient of 120,390 M Ϫ1 cm Ϫ1 for a monomer calculated from the amino acid composition (22). The concentrations of Eco57I are given in terms of the monomeric protein.
DNA Cleavage Assay-The endonuclease activity in vitro was tested by incubation of serial dilutions of purified proteins or cell-free extracts prepared as previously described (23) with 1 g of DNA at 37°C for 1 h in a 50-l reaction volume containing 10 mM Tris-HCl, pH 7.5, 10 mM MgCl 2 , 0.01 mM AdoMet, 0.1 mg/ml bovine serum albumin (standard reaction mixture), followed by electrophoresis on 0.8% agarose gels. The same reaction buffer was used for determination of specific activity of purified proteins. 1 unit of the endonuclease was defined as the amount required to hydrolyze 1 g of DNA in 1 h at 37°C until no change in the cleavage pattern was observed. In some experiments the standard reaction mixture was modified to include MnCl 2 instead of MgCl 2 and Sinefungin instead of AdoMet.
In vivo activity of wt and mutant restriction endonucleases was tested by comparison of plating efficiency of the vir bacteriophage on the strain ER2267 expressing wt Eco57I or mutants, in the presence of Eco57I methyltransferase on the compatible plasmid pEco57IM3.3, to that on ER2267 cells expressing only Eco57I methylase (nonrestricting host). Portions of serially diluted phage stock were spotted on a lawn of bacteria, and the plates were incubated at 37°C. The phage titer was determined (16). The efficiency of plating (e.o.p.) was defined as the phage titer on the host under investigation divided by the phage titer on a nonrestricting host.
DNA Methylation Assay-The modification activity in vitro was tested by the DNA protection assay where 1 g of DNA served as substrate in 50 l of reaction mixture (10 mM Tris-HCl, pH 7.5, 50 mM NaCl, 1 mM dithiothreitol, 0.1 mg/ml bovine serum albumin, 0.1 mM AdoMet). The reaction was initiated by the addition of varying amounts of sample solution and incubated for 1 h at 37°C. The reaction was terminated by heating at 65°C for 20 min. MgCl 2 solution (final concentration of 10 mM), and an excess of Eco57I was then added to the reaction mixture. The incubation was continued for 1 h at 37°C. The reaction products were resolved by agarose gel electrophoresis. 1 unit of the modification activity was defined as the amount of the enzyme that in 1 h at 37°C rendered 1 g of DNA resistant to cleavage by Eco57I.
Preparation of DNA Fragments for Gel Mobility Shift Assay-A 210-bp EcoRI-HindIII DNA fragment excised from pEco57IRM6.3 containing a single Eco57I site in the middle of the sequence was used as the specific DNA fragment. It was cloned into the phagemid pTZ19R, and a single-nucleotide substitution was introduced using a mispaired oligonucleotide by the method of Kunkel et al. (20), generating a PstI site instead of the Eco57I site. The resultant nonspecific 210-bp DNA fragment was excised from the phagemid with EcoRI and HindIII restriction endonucleases. Both specific and nonspecific fragments were gel purified and radiolabeled using Klenow polymerase to fill in their 5Ј extensions with [␣-33 P]dATP and the other three dNTPs (16). DNA concentrations were determined spectrophotometrically.
Gel Mobility Shift Assay-Binding reactions (20 l) contained the 33 P-endlabeled-specific or -nonspecific DNA fragment (final concentration 10 pM) and the wt Eco57I or mutant protein (final concentration in the range of 0 -10 nM). The incubation was performed in binding buffer (10 mM Tris-HCl, pH 8.0, 50 mM NaCl, 10 mM CaCl 2 , 0.1 mg/ml bovine serum albumin, 10% glycerol) for 20 min at room temperature, and then the samples were applied to a 6% polyacrylamide gel (29:1 acrylamide/ bis). Electrophoresis was carried out at 11 V/cm at room temperature in 40 mM Tris acetate, 10 mM CaCl 2 , pH 8.0 for 2 h. After electrophoresis, the gels were dried, and radioactive bands were visualized using the OptiQuant TM Image Analysis Software (Pacard).

RESULTS
Analysis of the Eco57I amino acid sequence revealed two putative catalytic/Mg 2ϩ binding sites: sequence motifs 77 PDX 13 EAK (SM I) and 811 PDX 20 DQK (SM II) (10). To assess their relevance, if any, to cleavage activity of the enzyme, we constructed a range of single amino acid substitution mutants of the acidic residues of the motifs; the residues most conserved in catalytic/Mg 2ϩ binding centers of restriction endonucleases (8,11). The following mutants were constructed by site-directed mutagenesis: D78N, D78A, D78K, E92Q for SM I and D812A, D833N, D833A for SM II. The mutants were analyzed for their cleavage, methylation, and DNA binding activities.
Assay of Endonuclease Activity of Wild-type and Mutant Eco57I-Plasmids carrying either wild-type or mutant Eco57I genes were used to transform E. coli ER2267 strains with or without the Eco57I methylase gene (in trans). As expected the plasmid encoding the wt Eco57I did not transform E. coli unless it harbored the protecting Eco57I methylase (see Table I). Plasmids encoding the SM II mutants D812A, D833A, and D833N expressed sufficient Eco57I endonuclease activity to be lethal in the absence of protecting methylase, but transformation was effective in cells that expressed the Eco57I methylase. Cells expressing the SM I mutants D78N, D78A, D78K, and E92Q did survive without cognate methylase protection however, suggesting that the mutants had no or very low endonucleolytic activity.
Wild-type Eco57I and mutants were also characterized for their ability to protect E. coli cells against vir phage infection. Phage titers were determined on E. coli ER2267 strains expressing wild-type or mutant eco57IR genes in the presence of the Eco57I methyltransferase and were compared with vir phage titers on cells expressing only Eco57I methylase. The e.o.p. of vir on the cells expressing mutants D78N, D78A, D78K, and E92Q of SM I was practically the same as that on the nonrestricting host indicating that the mutants failed to restrict the incoming phage (see Table I). The ability of the D812A mutant to restrict the vir phage was the same as that of the wild-type Eco57I. The e.o.p. of vir phage on the ER2267 strain expressing the D833N and D833A mutants was about three orders and two orders of magnitude higher than on the cells with the wild-type enzyme, respectively, indicating that the restriction activity of these mutants was reduced as compared with the wild-type Eco57I.
To assess the activities of the mutants in vitro, crude cell lysates were used to cleave the DNA substrate. D812A, D833N, and D833A mutants of SM II cleaved DNA specifically, like wild-type Eco57I (see Table I). No cleavage activity was detected in crude cell lysates obtained from cells expressing the D78N, D78A, D78K, and E92Q mutants of SM I. For further studies of mutant protein enzymatic properties, the D78N, E92Q, D812A, D833A, and D833N mutants were purified to apparent homogeneity, as described under "Experimental Procedures." Mutants of the conserved motif SM II-D812A, D833A, and D833N retained specific activities ranging from approximately the same as wt Eco57I (D812A) to ϳ40% (D833N) of wt Eco57I specific activity (see Table I). The D78N and E92Q mutant proteins did not show any DNA cleavage activity under standard reaction conditions, even at the highest protein concentrations tested (52 g/ml), whereas partial DNA digestion was observed with wt Eco57I at 0.013 g/ml (data not shown). Hence, the D78N and E92Q substitutions reduced Eco57I endonuclease activity at least 4000-fold. The activity of E92Q (but not that of D78N) was detected only after Mg 2ϩ in the reaction mixture was replaced with Mn 2ϩ (Fig. 1). The activity of the E92Q mutant in the presence of Mn 2ϩ was lower than the activity of wt Eco57I, as assessed under the same reaction conditions. Both activities were lower than the cleavage activity of the wt Eco57I tested in the presence of Mg 2ϩ . Even at the highest wt Eco57I concentration (26 g/ml) used in our experiments in the presence of Mn 2ϩ , only partial DNA digestion was observed (see Fig. 1), whereas 7.2 g/ml was sufficient in the presence of Mg 2ϩ to yield complete DNA cleavage (data not shown).
In vitro cleavage activity correlated well enough with the restriction activities observed in vivo to suggest that amino acids Asp-78 and Glu-92 are essential for the endonucleolytic activity of the Eco57I restriction endonuclease, whereas Asp-812 and Asp-833 of SM II are not. To test whether this effect was caused by the loss of ability to cleave DNA by the D78N and E92Q mutants or a deficiency in specific DNA binding, an electrophoretic mobility shift assay was used for characterization of the mutants as compared with wild-type Eco57I.

Binding of Wt and D78N, E92Q Mutant Proteins to the Eco57I Target Site-
The effects of mutations on binding of the mutant Eco57I proteins to DNA were examined using the gel mobility shift assay. Two DNA fragments were used: a 210-bp DNA fragment containing one Eco57I site in the middle of the sequence (specific DNA) and a nonspecific DNA fragment, which had the same sequence, except it lacked the Eco57I site as a result of a 1-bp substitution. The experiments were performed with purified wild-type Eco57I, D78N, and E92Q mutant proteins. In initial experiments, the DNA binding for the wild-type Eco57I restriction endonuclease was characterized. Increasing amounts of the Eco57I endonuclease were added to a fixed amount of 33 P-endlabeled specific or -nonspecific DNA fragment, and the free DNA was separated by electrophoresis from the DNA complexed with the protein. Several attempts to visualize specific Eco57I-DNA complexes in the absence of divalent metal ions were unsuccessful. As previously described for EcoRV (24), MunI (25), PvuII (26), and Cfr10I (27), a stable protein-DNA complex was formed in the presence of CaCl 2 . The gel shift assay of wt Eco57I binding with the specific fragment in the presence of 10 mM CaCl 2 revealed a shifted DNA band at low (0.01 nM) protein concentration (Fig. 2). The amount of the initial complex increased with increasing protein concentration in the range of 0.01-0.1 nM and then progressively decreased as protein concentration increased further (0.5-10 nM). Similar binding studies of wt Eco57I with the noncognate DNA revealed no shifted band corresponding to the initial complex. The DNA with retarded electrophoretic mobilities was observed only at high protein concentrations (0.5-10 nM). Further, comparison of the wt Eco57I interaction with cognate and noncognate DNA indicates that the initial complex (Fig. 2) corresponds to the specific enzyme-DNA complex, whereas bands of lower mobility correspond to the complexes that are represented by DNA fragments bound both by specifically and nonspecifically interacting enzyme molecules (28). The binding pattern of the D78N and E92Q mutants to specific DNA fragments indicates that they retain the ability to generate specific complexes with cognate DNA similar to that of wt enzyme (Fig.  2). The same results were obtained with the D812A, D833N, and D833A mutant proteins (data not shown). At the same time, the single amino acid substitutions D78N and E92Q were likely to weaken nonspecific binding as judged by the absence (compared with wt Eco57I), of the clearly defined band with the nonspecific DNA fragment and the band of lowest mobility with the specific fragment at 10 nM protein concentration (see Fig.  2).
Methylation Activity of Wild-type and Mutant Eco57I-The cleavage-deficient mutants D78N and E92Q were also tested for methylation activity. The reactions were performed with purified proteins using the DNA protection assay (see "Experimental Procedures"). Both mutants methylated DNA efficiently, and the specific activity of the D78N mutant was even higher than that of the wt Eco57I (Table I). The reaction mixture used in our experiments for assessing the cleavage activity of the wt and mutant Eco57I included AdoMet, an effective stimulator of endonucleolytic activity of the enzyme (4). It could not be excluded therefore that the observed cleavage deficiency of the D78N and/or E92Q mutants was attributed to the substitution of amino acids affecting the competition between cleavage and methylation activities of the bifunctional enzyme, rather than the amino acids involved in DNA cleavage catalysis per se. Such mutants in contrast to wt Eco57I, would initially methylate but not cleave DNA, rendering the modified DNA resistant to hydrolysis. AdoMet was therefore replaced by its analog Sinefungin, an inhibitor of the methylation reaction, in the reaction mixture to discriminate between these two alternatives: a cleavage deficiency as compared with preferential methylation mutant phenotypes. Sinefungin activated wt Eco57I as effectively as AdoMet (see Fig.  1). Under the same reaction conditions, no cleavage activity was detected in either the D78N or the E92Q mutant (Fig. 1).

DISCUSSION
Eco57I Catalytic Site-The results of mutational analysis in this study strongly suggest that of the two putative catalytic motifs 77 PDX 13 EAK (SM I) and 811 PDX 20 DQK (SM II), only the first one represents a DNA cleavage active site. Substitutions within SM I of negatively charged amino acids, which are most conserved in the catalytic centers of restriction endonucleases (8,11), with functionally unrelated (D78A, D78K) or function- ally similar (D78N, E92Q) amino acids reduced cleavage activity of the Eco57I to a level undetectable both in vivo and in vitro. Two mutants D78N and E92Q, which were selected for more detailed studies, however, retained the ability, as assessed by gel mobility shift assay, to interact with the target site specifically. The mutations also spared the DNA methylation activity of Eco57I restriction endonuclease, an observation which provides additional evidence that specific DNA binding as well as AdoMet binding were not affected by D78N and E92Q substitutions. Hence D78N and E92Q mutations display the properties expected for active site residue mutants; they uncouple the sequence-specific DNA binding and strand scission activities of the enzyme. However, bifunctionality of Eco57I suggested still another explanation for the observed phenotype of the mutant proteins. Namely, mutations that reverse the preferential order of expression of the two wild-type Eco57I enzymatic activities (cleavage then methylation) would be phenotypically indistinguishable from mutations affecting amino acid residues involved in catalysis/metal ion binding. This possibility was excluded by the observation that the endonucleolytic activity of the wt Eco57I was effectively stimulated not only by AdoMet but also by its analog Sinefungin. However, no DNA cleavage was detected regardless of which of the two cofactors was added to the reaction mixture with D78N and E92Q mutants. Some activity of the E92Q mutant (but not that of the D78N mutant) was detected only when the Mg 2ϩ in the standard reaction mixture was replaced with Mn 2ϩ . Similar effects of metal ion replacement on enzymatic activities of the MunI catalytic mutants D83A and E98Q have been previously reported and explained on the basis of differential effects of the mutational replacements on the binding of Mg 2ϩ and Mn 2ϩ (25). This observation provides additional evidence to support the conclusion that the Asp-78 and Glu-92 residues of Eco57I are involved in catalysis and metal ion binding. The fact that the Asp-83 and Glu-98 residues of MunI (25) and the Asp-78 and Glu-92 residues of Eco57I occupy equivalent positions in the PDX n EXK motif of the respective enzymes is noteworthy.
Eco57I mutants containing single amino acid substitutions within SM II (D812A, D833N, and D833A) revealed only a small or moderate decrease of cleavage activity as compared with wt Eco57I to suggest that SM II does not represent the catalytic center.
Circumstantial evidence supporting the conclusion that SM I represents the DNA cleavage center and SM II only mimics it as an amino acid sequence motif comes from an alignment of Eco57I and GsuI restriction endonucleases, whose recognition sequences are related (CTGAAG and CTGGAG, respectively). The homology of the enzymes is significant enough to conclude that they are evolutionary related (29). 2 Of the two putative Eco57I catalytic motifs, only a homolog of SM I is represented in GsuI, whereas that of SM II is absent.
It has been demonstrated that as expected, an asymmetric DNA target is recognized by the FokI monomer (30), whereas cleavage of both strands of DNA is carried out by two FokI molecules dimerized through their endonucleolytic domains (14). The asymmetric nature of the Eco57I target sequence is also inconsistent with the use of a symmetric dimer for recognition and DNA cleavage. Whether a mechanism similar to that of FokI is used by Eco57I to cleave DNA remains to be determined.
Restriction endonucleases catalyze phosphodiester bond cleavage in double-stranded DNA in the presence of Mg 2ϩ , leaving a 5Ј-phosphate and a 3Ј-hydroxyl group (8). The active sites of only a few of the numerous restriction endonucleases have been characterized so far (8). They share a triad of charged Asp, (Glu/Asp), and (Lys/Asp) amino acids. The catalytic relevance of the triad has been demonstrated not only for type II endonucleases but also for the type IIS restriction endonuclease FokI (13), another nontype II endonuclease BcgI (31), and a few type I enzymes (32,33). This study provides evidence that the catalytic/metal binding center of the type IV restriction endonuclease Eco57I is most likely also represented by a triad of charged amino acids Asp, Glu, and Lys located within the sequence motif 77 PDX 13 EAK of Eco57I. Restriction endonucleases, even those that belong to the same type, share little primary sequence similarity. The available data however indicate that the similarity of the chemical reaction catalyzed by these enzymes and their cofactor (Mg 2ϩ ) requirements dictate a similarity in their catalytic/Mg 2ϩ binding sites, independent of any affiliation of restriction endonucleases to the specific types of these enzymes.
Structure-Function Organization of Eco57I-The restriction endonuclease Eco57I combines, in a single polypeptide DNAspecific recognition, cleavage, and methylation activities (4). This study provides evidence that the catalytic DNA cleavage center is located near the N-terminal end of Eco57I (Fig. 3). The central part is occupied by conserved motifs of m 6 A DNA methylases, involved in AdoMet binding and catalysis (9). The methylase part of Eco57I belongs to the ␥ group of DNA aminomethyltransferases (34). It has been suggested that the Cterminal variable part of enzymes in this group of methylases comprise DNA recognition region, which includes the target recognition domain (TRD) (34). We speculate that the same is true for Eco57I restriction endonuclease. On the basis of the above data, the Eco57I primary sequence can be provisionally subdivided into three structure-function regions (Fig. 3).
The SM II mutations are located in the C-terminal part of the enzyme. They did not impair specific DNA binding nor dramatically decrease the enzyme activity, which could be expected for TRD mutants. Because the exact position of TRD in the carboxyl domain of Eco57I is not known, the SM II mutations may be located beyond it. Even if the opposite were true, the SM II mutations are site specific with respect to the putative catalytic motif (SM II) but are random with respect to the putative TRD. Analysis of random mutants of a few m 6 A MTases showed that only a small fraction of them located in the variable region reveals the expected phenotype; a loss of specific DNA binding and activity (35,36). More detailed mutational analysis of the C-terminal part is required to prove or disprove the suggestion that it includes the Eco57I target recognition domain.
Structure-function organization of Eco57I suggests that its evolution involves fusion of a methylase and an endonuclease. It has been speculated that the progenitor of Eco57I was generated by the fusion of Mod and Res subunits of a type III endonuclease (9). This assumption was based on enzymatic properties (incomplete cleavage of DNA, stimulation of cleav-2 R. Vaisvila and A. Janulaitis, unpublished data. age activity by AdoMet, methylation of one DNA strand), substrate specificity (asymmetric recognition sequence), and cleavage mode (outside of the target duplex) of Eco57I, which are similar to the properties of type III restriction endonucleases. However, the following observations make the R-M fusion hypothesis unlikely. (i) In contrast to type III endonucleases, Eco57I is not stimulated by ATP (4); (ii) DEAD box motifs involved in ATP hydrolysis and DNA translocation that have been identified in Res subunit (37) are not present in the Eco57I amino acid sequence; and (iii) the size of the Eco57I endonuclease domain is ϳ350 amino acids (see Fig. 3), whereas that of Res subunits is 873-982 amino acids (38 -40). Therefore generating the progenitor for the endonuclease domain of Eco57I from the Res subunit would require dramatic rearrangements of the latter.
The evolution of type IIS enzymes, which like Eco57I recognize asymmetric nucleotide sequences and cleave DNA at a defined distance from them but are not stimulated by AdoMet (neither contain the conserved motifs of MTases), possibly involves fusion of a DNA-specific-binding protein and an endonuclease (41). In type IIS R-M systems, two separate MTases exist, each specific for a different strand of asymmetric target duplex (42)(43)(44). The fusion of a single-strand-specific MTase with a DNA endonuclease could generate a progenitor of Eco57I, which potentially could reveal both DNA cleavage and methylation (of one strand). The cofactor requirement (Mg 2ϩ and AdoMet), enzymatic activities (endonucleolytic, methylation at one DNA strand), and structure (a methylase fused with endonuclease) of such a hypothetical hybrid protein makes it a more likely candidate for the Eco57I predecessor than the product of Res-Mod fusion.