The HgaI restriction-modification system contains two cytosine methylase genes responsible for modification of different DNA strands.

A DNA fragment of about 3.4 kilobase pairs that expressed the HgaI modification activity was cloned from the chromosomal DNA of Haemophilus gallinarum, and its nucleotide sequence was determined. Two open reading frames (ORF) which could code for structurally similar proteins were identified in the upstream and middle regions and a truncated ORF in the downstream region in the same orientation. When the respective ORFs were separately cloned, the clones carrying the upstream and middle ORFs both expressed the modification activity, indicating that the two genes are involved in modification of the HgaI restriction-modification system. In order to determine the sites of modification precisely, the respective genes were recloned into an expression vector, from which gene products were purified. A short DNA fragment carrying the HgaI recognition site was treated with each of these enzymes, and, after separation of the two strands by duplex formation with M13 viral DNAs carrying the respective strands, the presence or absence of modification was judged from susceptibility to HgaI endonuclease. The results of analysis showed that different strands were modified in an asymmetric way by each gene product. Analysis of the species and positions of modified bases by the Maxam-Gilbert method further demonstrated that the gene products from the upstream and middle ORFs participated in methylation of the internal cytosine residues of the strands carrying 3'-CTGCG-5' and 5'-GACGC-3', respectively. We concluded that the HgaI modification system consisted of two cytosine methylase genes responsible for modification of different strands in the target DNA.

The HguI Restriction-Modification System Contains Two Cytosine Methylase Genes Responsible for Modification of Different DNA Strands* (Received for publication, February 5, 1991) Hiroyuki Sugisaki, Katsuhiko Yamamoto, and Mituru Takanami From the Institute for Chemical Research, Kyoto University, Uji, Kyoto 61 I, Japan A DNA fragment of about 3.4 kilobase pairs that expressed the HgaI modification activity was cloned from the chromosomal DNA of Haemophilus gallinarum, and its nucleotide sequence was determined. Two open reading frames (ORF) which could code for structurally similar proteins were identified in the upstream and middle regions and a truncated ORF in the downstream region in the same orientation. When the respective ORFs were separately cloned, the clones carrying the upstream and middle ORFs both expressed the modification activity, indicating that the two genes are involved in modification of the HgaI restrictionmodification system. In order to determine the sites of modification precisely, the respective genes were recloned into an expression vector, from which gene products were purified. A short DNA fragment carrying the HgaI recognition site was treated with each of these enzymes, and, after separation of the two strands by duplex formation with M13 viral DNAs carrying the respective strands, the presence or absence of modification was judged from susceptibility to HgaI endonuclease. The results of analysis showed that different strands were modified in an asymmetric way by each gene product. Analysis of the species and positions of modified bases by the Maxam-Gilbert method further demonstrated that the gene products from the upstream and middle ORFs participated in methylation of the internal cytosine residues of the strands carrying 3'-CTGCG-5' and 5'-GACGC-3', respectively. We concluded that the HgaI modification system consisted of two cytosine methylase genes responsible for modification of different strands in the target DNA.
Modification enzymes of the type I1 restriction-modification systems that recognize symmetrical sequences contain a single functional domain for methylation within the monomeric protein molecule and methylate the target bases of both strands in the rotationally symmetric manner (Lauster, 1989;Klimasauskas et ul., 1989). In the case of the FokI restrictionmodification system that recognizes an asymmetric sequence, however, its methylase has been shown to contain two functional domains within its molecule, each of which is responsible for modification of different strands in the target DNA ~ ~~ *This work was supported by Grants-in-Aid for Scientific Research from the Ministry of Education, Science and Culture of Japan. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. The nucleotide sequence($ reported in this paper has been submitted D90363. (Sugisaki et ul., 1989;Looney et ul., 1989). It is of interest to know whether or not this is a common feature of the methylases in the restriction-modification systems that belong to this category. The HguI restriction-modification system found in Huemophilus gullinarum recognizes a 5-bp' segment of DNA, consisting of 5'-GACGC-3' in one strand and 3'-CTGCG-5' in the other (HguI recognition site). HguI endonuclease (RHguI) introduces staggered cleavages 5 and 10 nucleotides downstream from the recognition sequence (Brown and Smith, 1977;Sugisaki, 1978), and the HguI modification enzyme (MHguI) renders the DNA resistant to digestion by RHguI. Cloning of a DNA fragment that confers the MHguI activity on Escherichia coli cells has been reported by Nwankwo and Wilson (1987), but nothing has been described about the gene structure. In this paper, we cloned a DNA fragment expressing the HguI modification activity from a plasmid library of H. gullinarum DNA and assigned two coding regions for the HguI modification enzymes on the basis of the sequence analysis and assay of the modification activity in uiuo. We further showed by analysis of the strands and bases modified that the products of these genes, respectively, modified the cytosine residues of different strands.
Enzymes and Biochemicals-The restriction enzymes except for RHgaI, bacterial alkaline phosphatase, T4 DNA ligase, T4 polynucleotide kinase, the M13 sequencing kit, and the deletion kit for kilobase sequencing were purchased from Takara Shuzo Co. RHgaI was a product from New England Biolabs. Buffers and conditions followed suppliers' specifications.
S-Adenosyl-L-methionine was obtained from Sigma. [ ( u -~' P ]~C T P for DNA sequencing and [y-32P]ATP for end-labeling were from Amersham and Du Pont-New England Nuclear, respectively. Cloning and Selection of HgaI Modification Clones-H. gallinarum DNA was purified by the procedure of Thomas et al. (1966). Complete HindIII digests of the DNA were fractionated by electrophoresis on 0.7% agarose gel, and fractions predominantly containing the fragments of 2.3-4.4 kb were collected, based on the previous report that a 3.5-kb HindIII fragment expressed the HgaI modification activity (Nwankwo and Wilson, 1987). The sized HindIII fragments were ligated to a phosphatase-treated HindIII digest of pUC18 and intro-The abbreviations used are: bp, base pair(s); kb, kilobase pair(s); ORF, open reading frame; ds, double-stranded ss, single-stranded. duced into E. coli HBlOl cells. Cells were plated onto TY agar plates containing 50 pg/ml ampicillin. Transductants were scraped together, inoculated into 100 ml of TY medium containing 50 rg/ml ampicillin, and grown to saturation, from which plasmid DNA was prepared by the alkaline method. HgaI modification clones were then selected by the procedure suggested by Mann et al. (1978). One microgram of purified plasmid DNA was digested for 1 h at 37 "C with 4 units of RHgaI. The digest was introduced into E. coli HBlOl competent cells, and cells were plated onto TY agar plates containing 50 pg/ml ampicillin. Plasmids were purified from individual transformants by the alkaline method and subjected to restriction analysis.
Purification of MHgaI Gene Products-About 6 g of induced E. coli MCHlO61 cells harboring pKK223-3:MHgaI genes were thawed and suspended in 30 ml of 50 mM Tris-HC1, pH 7.5, 1 mM EDTA, 2 mM dithiothreitol, 0.1 mM phenylmethylsulfonyl fluoride, and 0.5 mg/ml lysozyme. After holding for 30 min on ice, the mixture was briefly sonicated to complete lysis and centrifuged for 1 h at 100,000 X g to obtain the crude extract. To the extract NaCl was added to 0.1 M and then a 10% stock solution of polyethyleneimine to 1%. After removal of the precipitate by centrifugation, solid ammonium sulfate was added to 70% saturation. The resulting precipitates were collected by centrifugation, dissolved in 10 mM potassium phosphate buffer, pH 7.5,lO mM 2-mercaptoethanol, and 5% glycerol (buffer A) containing 0.1 M KC1 and dialyzed against the same buffer. The solution was applied onto a phosphocellulose column (Whatman P-11, 1.5 X 7.5 cm) and chromatographed with a linear KC1 gradient in buffer A (0.1-1 M, 200 ml). The MHgaI activity was eluted at 0.3-0.4 M KC1. The methylase fraction was dialyzed against buffer A, put on a DEAEcellulose column (Whatman DE52, 1.0 X 10 cm), and eluted with a linear gradient of KC1 in buffer A (0-0.4 M, 100 ml). The active fractions, which were eluted at 0.05-0.15 M, were concentrated by dialysis against 50 mM Tris-HC1, pH 7.5, 1 mM EDTA, 10 mM 2mercaptoethanol, and 50% glycerol and stored at -20 'C.
One unit of the enzyme was defined as the amount sufficient to render 1 pg of X-DNA resistant to RHgaI digestion in 1 h at 37 "C.
Assay of Modification Activity in Viuo-Plasmids were purified by CsC1-ethidium bromide centrifugation from E. coli HBlOl cells to test. Five-tenths microgram each of the plasmids was treated for 1 h at 37 "C with 0.5 unit or 1.5 units of RHgaI in 20 pl of the reaction mixture containing 10 mM Tris-HC1, pH 7.5, 10 mM MgC12, and 1 mM dithiothreitol. The digests were electrophoresed on 1% agarose gel and made visible by staining with ethidium bromide. Identification of Methylated Strands-Two picomoles of the 214bp fragment carrying a single HgaI recognition site was dephosphorylated by bacterial alkaline phosphatase and rephosphorylated by the use of [r-"'P]ATP (6000 Ci/mmol) and T4 polynucleotide kinase. The specific activity of the products was about 1.0 X IO4 cpm/fmol. One hundred femtomoles of the 5'-32P-labeled DNA fragment were incubated with 2 units of purified MHgaI gene products in 50 pl of the reaction mixture containing 50 mM Tris-HC1, pH 7.5, 1 mM EDTA, 60 mM NaC1,5 mM 2-mercaptoethanol, and 80" S-adenosyl-L-methionine. Four ferntomoles of the labeled DNA, treated or not treated, were denatured by holding for 5 min at 100 'C in 15 pl of the reaction mixture containing 13 mM Tris-HC1, pH 7.5, and 0.5 mM EDTA. To the reaction mixture was added 1 ~1 (400 fmol) of M13 viral DNA carrying either strand of the 214-bp fragment, The mixture was kept for 60 min at 65 "C, and, after annealing, 1 p1 of 0.2 M MgCl? and 20 mM dithiothreitol, 1 pl (0.4 pg) of XDNA, and 2 p1 (1 unit) of RHgaI were added. The mixture was incubated for 60 min at 37 "C, and the product size was analyzed by electrophoresis on 8% polyacrylamide gel under the denatured conditions followed by autoradiography.
Identification of Methylated Bases-Ten micrograms of the plasmid, that can generate the 214-bp fragment containing the HgaI site by REcoRI-RHindIII digestion, were incubated with 50 units of purified MHgaI gene products under the same reaction conditions as described above. The plasmid DNA was linearized by digestion with either REcoRI or RHindIII and dephosphorylated. After labeling with "P in the T4 polynucleotide kinase reaction, the 214-bp fragments bearing the terminal label at either end were generated by introducing the secondary cleavages. The labeled fragments were purified by 1% agarose gel electrophoresis and subjected for a set of five base-specific reactions of the chemical sequencing method (G-, (A + G)-, (A > C)-, C-, and T-specific KMn04 reactions) (Maxam and Gilbert, 1980;Rubin and Schmid, 1980). Each set of the reaction mixtures was analyzed by electrophoresis on 8% polyacrylamide gel followed by autoradiography.
Other Procedures-DNA ligation, transformation, and plasmid isolation were carried out as described by Sambrook et al. (1989). The dideoxy chain termination methods (Sanger et al., 1977;Messing, 1983) was used for determination of the entire sequence of the 3.4-kb region. The subfragments generated were cloned into appropriate sites of M13mp18 or M13mp19 to provide templates, and the sequence was deduced from the data for both strands.

Isolation of HgaI Modification
Clones-Recombinants expressing the HgaI modification activity were screened by the survivor selection method from a plasmid library of the H. gallinarum DNA that had been extensively digested with RHgaI. About 1000 colonies carrying plasmids were surveyed, and, finally, 20 colonies randomly picked up were found to carry recombinants resistant to RHgaI digestion. When the inserts were analyzed, these recombinants contained a 3.4-kb HindIII fragment in common. Among the recombinants examined, one carrying a single 3.4-kb HindIII fragment was named pKS318.
Analysis of Nucleotide Sequences-The DNA sequence of the 3.4-kb region in pKS318 is shown in Fig. 1. Analysis of coding capacities indicated that two large ORFs were present in the upstream and middle regions and a truncated ORF in the downstream region in the same orientation (see Fig. 2). In the upstream and middle ORFs (ORF1 and ORF2), the first ATG codons appeared at nucleotide positions 207 and 1294 and the termination codons at nucleotide positions 1302 and 2368, respectively. In the truncated ORF (ORF3), the first ATG codon appeared at nucleotide position 2345 and terminated 150 bp inside the pUC18 moiety beyond the rightmost end.
Assignment of Methylation Genes-To assign these ORFs to the modification gene(s), the respective regions were located under the lac promoter of the pUC vectors. For ORF1, the left 1.2-kb region of the 3.4-kb HindIII fragment (from the Nsp(7524)V site to the BglII site in Fig.  2), which encompasses the entire ORF1, was inserted into the AccI-BarnHI region in the multicloning sites of pUC19. For ORF2, the middle 2.2-kb fragment between the two XbaI sites at nucleotide positions 716 and 2909 of the 3.4-kb HindIII fragment was inserted into the XbaI site in the multicloning sites of pUC19 in the proper orientation. Deletion was introduced from both sides of the insert, and a clone carrying a DNA fragment between nucleotide positions 1258 and 2497 which encompasses the entire ORF2 was isolated. For ORF3, the right 1.2-kb region of the 3.4-kb HindIII fragment (from the ClaI site to the HindIII site at the rightmost end in Fig,  21, which encompasses the truncated ORF3, was inserted into the AccI-Hind111 region in the multicloning sites of pUC18. Thus, the gene product from ORF3 should contain 50 extra amino acid residues on its COOH-terminal. The resulting constructs, carrying ORF1, ORF2, and ORF3, were named pKS412, pKS385, and pKS334, respectively (Fig. 2). These plasmids were introduced into E. coli cells, and, after propagation, plasmids were extracted from cells and their susceptibility to RHgaI was compared. pKS334 coding for ORF3 was susceptible to RHgaI to about the same extent as the control plasmid without methylation (Fig. 3, lanes 2 and 3 ) and digested completely by incubation for 1 h with 0.5 unit of RHguI per 0.5 pg of DNA (lunes 11 and 1 2 ) . In contrast, no digestion of pKS412 and pKS385, respectively carrying ORFl and ORF2, occurred a t increased RHguI concentrations (lunes 5 and 6 for pKS412; lunes 8 and 9 for pKS385). It is, therefore, obvious that the gene products from ORFl and ORF2 retained significant levels of the HguI modification activity. These gene products were named MHguIORFl and MHguIORF2, respectively. When the amino acid sequences of both products 1 2 3 4 5 6 7 8 9 1 0 1 1 1 2   FIG. 3. Assay of the modification activities expressed in vivo from each OHF region. I'lasmids PC'(: 18 as cont.rol (lanes 2-3), pKS412 (lanes &6), pKSIR5 (larzes 7-91, and pKS.734 (Ianes IO-12) were purified from E. coli HRlOl cells that harbored each plasmid. 0 5 pg of plasmids was treated for 1 h with 0.5 unit (lanes 2, 5 , 8, and 1 1 ) or 1.5 units (lanes 3 , 6, 9 were compared with each other, a high degree of similarity was found in the overall regions (see Fig. 7 ) indicating that both genes code for structurally similar proteins.
Overexpression of Gene Products-The coding region for MHgaIORF1 could be just generated as a 1.2-kb fra, oment from pKS318 by RNsp(7524)V-RRglII digestion and inserted into the SmaI-BamHI region in the multicloning sites of pKK223-3 by blunt-end ligation. The coding region for MHgaIORF2 could be generated just as a 1.2-kb fragment from pKS385 by REcoRI-RHindIII digestion. Thus, the 1.2kb fragment was repaired and inserted into the S~Q I site of pKK223-3 in the proper orientation. The resulting constructs carrying the MHgdORFl and MHgaIORF2 genes under control of the tac promoter were named pKS417 and pKS401, respectively. The constructs were introduced into E. coli MCHlO61 cells, and, after induction by isopropyl-P-D-thiogalactopyranoside, proteins expressed from the respective genes were purified by phosphocellulose and DEAE-cellulose columns.
Analysis of Methylated Strands by M H g d Gene Products-The DNA strands modified by MHgaIORFl and MHguIORF2 were analyzed by the strategy described previously (see Fig. 5 in Sugisaki et al., 1989). In principle, both strands of a short double-stranded (ds) DNA fragment carrying a single HguI recognition site were terminally labeled and treated with each of the MHgd gene products. The two strands were then separated by duplex formation with M13 viral DNAs in which the respective strands had been inserted. After treatment with RHgaI, the sizes of terminal labeled fragments were measured. Modification in the hybridized strand can be detected from its susceptibility to RHgaI.
The short dsDNA fragment used here was the one generated from pBR322 by RRanHI-RSphI digestion (nucleotide position 376-565 in the pBR322 map; Sutcliffe, 1979). This fragment was cloned into the BamHI-SphI region in the multicloning sites of pHSG398 (Takeshita et ul., 1987). The fragment was regenerated from the construct (pKS360) by REcoRI-RHindIII digestion as a fragment of 214 bp, and M13 clones carrying the 214-bp fragment in either of two orientations were prepared by the use of M13mp18 and -mp19. Viral DNAs prepared from these M13 clones were named M13KS121 and M13KS123 single-stranded (ss) DNAs.
Both strands of the 214-bp fragment were highly labeled with "'P at the 5' termini and reacted with each of purified MHgdORFl and MHgaIORF2 in the presence of S-adenosyl-L-methionine. After denaturation, they were hybridized with M13KS121 and M13KS123 ssDNAs. The duplexes were intensively digested with RHgd and analyzed by sequencing gels under the denaturing conditions. The HguI recognition site in the single-stranded form was not susceptible to RHguI under the digestion condition used here. The results of analysis are shown in Fig. 4, in which lunes 1-4 are for untreated DNA fragments, lunes 5-8 are for those treated with MHgdORFl, and lanes 9-12 are for those treated with MHgaIORF2. With the untreated DNA fragments, the labeled fragment generated from the strand carrying 5'-GACGC-3' (GACGC strand) was 45 nucleotides long (lune 4 ) , and that generated from the other strand carrying 3'-CTGCG-5' (CTGCG strand) was 168 nucleotides long (lune 3). The fragments generated were the same sizes as those yielded from the unmodified 214-bp dsDNA fragment by RHgd digestion (lane 2). With the DNA fragment modified with MHgaIORF1, the 45-nucleotide fragment was generated only when a duplex was formed with M13KS123 ssDNA (lune 8 ) , and, with the DNA fragment treated with MHgaIORF2, the 168-nucleotide fragment was generated only when a duplex was formed with M13KS121 ssDNA (lune l l ) . The results clearly indicated that MHgaIORFl and MHgaIORF2 methylated different strands of the 214-bp fragment. The Two femtomoles each of the 5'-'"'P-laheled fragment (ahout 2 X 10'' cpm) were treated with MHgnIOItF1 (lanrs 5-8) and MHga-IORF2 (lonrs 9-12). Lanrs 1-4 are for the untreated control. The fragments were heat-denatured and hyhridized with M13KS121 ssDNA carrying the GACGC strand or Ml:IKS123 ssDNA carrying the CTGCG strand. The resultingduplexes were digested with RHgoI and the sizes of laheled strands were analyzed hy gel electrophoresis under the denaturing conditions. In each set of four lanes, the first lane is the fragment alone (/ones 1, 5 , and 9 ) . the srcond Ianr is the fragment digested with RHgaI without denaturation (Inncs 2, 6, and IO), the third lone is the fragment digested with KHgnI after hyhridization with Ml3KS121 ssDNA (/ones 9 , 7, and 11 ), and the fourth lone is the fragment digested with RHgaI after hybridization with MlBKS123 ssDNA (lanes 4,8, and 12).
findings also provided information on the strand specificities of the two methylases.
Identification of Methylated Rases-The MHgd sequences showed characteristic features of C-specific methylases (see "Discussion"). Thus, we attempted to identify the species and exact positions of the modified bases, based on the fact that methylated C is not detected by the chemical sequencing method (Ohmori et ul., 1978;Butkus et al., 1985). pKS360 DNA, that generates the 214-bp fragment carrying a single HguI site by REcoRI-RHindIII digestion, was reacted with each of MHgdORFl and MHgaIORF2. The DNAs were digested with either REcoRI or RHindIII, and, after terminal labeling with "'P, secondary cleavages were introduced and the 214-bp fragments each bearing a single terminal label at either the REcoRI or RHindIII end were prepared. The fragments were then subjected to a set of five base-specific reactions.
The results of analysis are shown in Fig. 5 , in which the upper three sets indicate the sequence ladders for the GACGC strand and the lower three sets those for the CTGCG strand. Sets 1 and 4 are for untreated fragment as control, sets 2 and 5 are for those treated with MHguIORFl, and sets 3 and 6 are for those treated with MHgaIORF2. With the untreated fragments, all base-specific bands corresponding to the sequence around the HguI site were seen for both strands (sets I and  4 ) . With the MHguIORFl-modified fragments, however, the bands corresponding to the internal C of the CTGCG strand were missing in the A > C and C reactions (set 5 ) . With the MHgaIORF2-modified fragments, on the other hand, the internal C bands of the GACGC strand disappeared in the A > C and C lanes (set 3 ) . We carefully examined the sequence ladders in the flanking region of the HgaI site, but no other changes were observed. Thus, we concluded that MHgaIORFl and MHgaIORF2 methylated the internal cytosine residues of the CTGCG and GACGC strands, respectively. The results also coincide with the previous observations for the strand specificities of the two methylases.

DISCUSSION
Our results clearly indicated that the products of two MHgaI genes, located in the upstream and middle regions on the 3.4-kb HindIII fragment, respectively, methylate the internal cytosine residues of the different strands within the HgaI recognition sequence, as shown in Fig. 6.
Cytosine residues modified by methylases of the type I1 restriction-modification systems are known to be either 5methyldeoxycytosine (m"C-cytosine) or N"-methyldeoxycytosine (m'N-cytosine) (Kessler and Manta, 1990). m"N-cytosine is reactive to hydrazine, but m"C-cytosine is less reactive to this reagent and also to sodium hydroxide (Ohmori et al., 1978;Butkus et al., 1985). So, m"C-cytosine can be discriminated from m4N-cytosine using A > C and C specific reactions of the chemical sequencing method. By modification with the two HgaI methylases, the bands corresponding to the internal cytosine residues within the recognition sequences disappeared from the sequence ladders in A > C and C specific reactions. It is therefore likely that modified bases by the two MHgaI gene products are m"C-cytosine.
The amino acid sequences of the two HgaI methylases predicted from the nucleotide sequence are given in Fig. 1. According to this assignment, the reading frames are different by one base, and the coding regions overlap by a few amino acids. Although the NH2-terminal sequences of the gene products have not been analyzed, we assume that this assignment is reasonable, because comparison of the predicted amino acid sequences showed striking similarity in overall regions including the NHp-terminal (Fig. 7). A minor modification of this assignment could be the start point of the MHgaIORFl proteins, for the sequence similarity to MHgaIORF2 starts from the second Met, and, this being the c'ase, a more Shine-Dalgarno-like sequence is seen at an appropriate position.
When the amino acid sequences of the two HgaI methylases were compared by computer with published sequences of other m'C-methylases, both MHgaI sequences contained conserved 10 blocks (block I to X in Fig. 7) that have been identified as characteristics of all known m'C-methylases Posfai et al., 1989). In these m"C-methylases, the conserved blocks are separated by short regions of similar length but with different sequences, except for the interval between blocks VI11 and IX. The sequences and distances in this interval are quite different, so that this variable region has been assigned to the DNA target-recognizing domain Posfai et al., 1989). The sequence organizations of the HgaI methylases were essentially similar to those of other methylases. It was noted, however, that the sequences in the variable region mentioned above have been well conserved between the two HgaI methylases. The most striking difference in the primary structure was rather recognized in the interval between the conserved blocks VI and VII, in I II I1 I I IIIIIIIIII I  IIIII I I I I I  which large occasional spacings were required for maximization of similarities (see Fig. 7). If the variable region between blocks VI11 and IX indeed corresponds to the protein domain responsible for sequence specificity, conservation of the sequences in this area suggests that both the HgaI methylases recognize a similar structural element in the recognition sequence. This being the case, the sequence variation of the region between the conserved blocks VI and VI1 may be correlated to the conformational change between the protein domains involved in recognition of the target sequence and methyl group transfer.

HHgaIORFl U I N~D R L~I U G L S L P S S A G~G E Y F L S R V G I D I I Y M~~~
The presence of two methylases responsible for recognition and methylation of the strand-specific sequences would be advantageous for bacterial cells, because nascent strands generated during DNA replication can be simultaneously protected from the attack of site-specific endonucleases. In the FokI system which also recognizes an asymmetric sequence, the methylase contains two functional domains within a single protein molecule, each of which was responsible for modification of different strands (Sugisaki et al., 1989;Looney et al., 1989). Nevertheless, comparison of the two functional domains showed no sequence similarity except for the tetraamino acid sequences common to adenine-specific methylases. In contrast, the sequences of the two HgaI methylases have been well conserved in overall regions, suggesting that the two methylases were evolved from a common ancestor by gene duplication.