Genetic Organization and Molecular Analysis of the EcoVIII Restriction-Modification System of Escherichia coli E1585-68 and Its Comparison with Isospecific Homologs

Abstract

The EcoVIII restriction-modification (R-M) system is carried by the Escherichia coli E1585-68 natural plasmid pEC156 (4,312 bp). The two genes were cloned and characterized. The G+C content of the EcoVIII R-M system is 36.1%, which is significantly lower than the average G+C content of either plasmid pEC156 (43.6%) or E. coli genomic DNA (50.8%). The difference suggests that there is a possibility that the EcoVIII R-M system was recently acquired by the genome. The 921-bp EcoVIII endonuclease (R · EcoVIII) gene (ecoVIIIR) encodes a 307-amino-acid protein with an M_r of 35,554. The convergently oriented EcoVIII methyltransferase (M · EcoVIII) gene (ecoVIIIM) consists of 912 bp that code for a 304-amino-acid protein with an M_r of 33,930. The exact positions of the start codon AUG were determined by protein microsequencing. Both enzymes recognize the specific palindromic sequence 5′-AAGCTT-3′. Preparations of EcoVIII R-M enzymes purified to homogeneity were characterized. R · EcoVIII acts as a dimer and cleaves a specific sequence between two adenine residues, leaving 4-nucleotide 5′ protruding ends. M · EcoVIII functions as a monomer and modifies the first adenine residue at the 5′ end of the specific sequence to N⁶-methyladenine. These enzymes are thus functionally identical to the corresponding enzymes of the HindIII (Haemophilus influenzae Rd) and LlaCI (Lactococcus lactis subsp. cremoris W15) R-M systems. This finding is reflected by the levels of homology of M · EcoVIII with M · HindIII and M · LlaCI at the amino acid sequence level (50 and 62%, respectively) and by the presence of nine sequence motifs conserved among m⁶ N-adenine β-class methyltransferases. The deduced amino acid sequence of R · EcoVIII shows weak homology with its two isoschizomers, R · HindIII (26%) and R · LlaCI (17%). A catalytic sequence motif characteristic of restriction endonucleases was found in the primary structure of R · EcoVIII (D¹⁰⁸X₁₂DXK¹²³), as well as in the primary structures of R · LlaCI and R · HindIII. Polyclonal antibodies raised against R · EcoVIII did not react with R · HindIII, while anti-M · EcoVIII antibodies cross-reacted with M · LlaCI but not with M · HindIII. R · EcoVIII requires Mg(II) ions for phosphodiester bond cleavage. We found that the same ions are strong inhibitors of the M · EcoVIII enzyme. The biological implications of this finding are discussed.

The classic type II restriction-modification (R-M) system is composed of two enzymes, an endonuclease (ENase) that recognizes and cleaves a short DNA-specific sequence (4 to 8 bp) and a methyltransferase (MTase) that modifies the same sequence to protect the host DNA against the action of the cognate restriction enzyme. As these enzymes make up a type II R-M system that recognizes the same specific sequence, it could be expected that this fact would be reflected in extended homology of the two proteins. Surprisingly, there is a clear lack of homology between restriction ENases and their cognate MTases at the amino acid sequence level (14). Apparently, the ENases and their cognate MTases are the products of separate evolutionary processes. A lack of significant homology is also observed between isoschizomers of restriction enzymes that are isolated from different bacteria, have the same specificity, and cut DNA in the same manner (e.g., HhaII and HinfI [13]; FnuDI and BsuRI [39]; and Sau3AI and DpnII [70]). This observation suggests that the isospecific ENases appeared independently and were shaped by convergent evolution. However, in recent studies, analyses of protein structure stability elements have produced evidence that all type II restriction ENases containing a common catalytic motif diverged from a common ancestor (9, 21). In general, restriction ENases form a highly diverse group of proteins. However, there are a few examples of isoschizomers which exhibit significant identity at the amino acid sequence level; these include EcoRI and RsrI (50% identity) (79), AvaI and BsoBI (55%) (66), FnuDI and NgoPII (59%) (82), PvuII and SbaI (68%) (55), TaqI and TthHB8I (77%) (4), Cfr9I and XcyI (80%) (47), NspI and NspHI (89%) (90), and EcoHK31I and EaeI (92%) (44). It is very likely that the observed homology is the result of a common origin. On the other hand, data derived from crystallographic studies suggest that despite their lack of homology restriction enzymes have similar tertiary structures, with a protein core consisting of a central five-strand β-sheet surrounded by two α-helices (86). This structure is required to bring together into spatial proximity amino acid residues of the putative catalytic DNA cleavage-Mg(II) binding motif PD/EX_nD/EXK (57, 89). DNA MTases, unlike restriction enzymes, are highly homologous. Apart from a common core structure which is composed of a mixed seven-strand β-sheet (51), they contain several characteristic conserved sequence motifs whose order reflects the base specificity of a particular enzyme (11, 24, 50, 60, 88).

In our laboratory we started research aimed at investigating the nature of the isospecificity phenomenon in type II R-M systems, and we were especially interested in finding out what the R-M enzymes have in common. As a model in our study we decided to use a group of systems isospecific to HindIII, an R-M system from Haemophilus influenzae Rd (65, 76). This group consists of over 30 R-M systems isolated from different bacteria (61). To date, except for HindIII (56) only two other systems, LlaCI from Lactococcus lactis subsp. cremoris W15 (49) and EcoVIII from Escherichia coli E1585-68 (36), have been cloned and sequenced.

The phenomenon of isospecificity among type II R-M systems is interesting in many ways. First, it poses the intriguing question, how do genes that encode functional homologs evolve in bacteria that are often not phylogenetically related? In addition, it is interesting because structural analysis of these enzymes can help in the localization of particular motifs responsible for catalytic reactions and for target recognition and consequently could be used in designing enzymes with novel specificities. At the moment, the small number of cloned isospecific systems recognizing the same specific sequence makes such studies difficult.

In our studies we addressed the following questions. (i) How similar are the genes that encode isospecific enzymes? (ii) Is it possible to map the functional domains of these enzymes? (iii) Do the enzymes recognize cognate sequences in the same way? (iv) What is the mode of action of the enzymes?

In this paper we describe a molecular analysis of the EcoVIII R-M system from E. coli E1585-68 and provide some answers to the questions mentioned above. We also provide details concerning the genetic structure of the two genes constituting the EcoVIII system, as well as data obtained from an molecular analysis of their products in comparison with isospecific enzymes of the HindIII and LlaCI R-M systems.

MATERIALS AND METHODS

Bacterial strains and plasmids.

The bacterial strains used in this study were E. coli E1585-68 (a producer of EcoVIII enzymes) (53), DH5α, MM294 (67), BL21(DE3) (81), MC1061 (12), and ER2418 [fhuA2 supE44 e14⁻ McrA⁻ rfbD1 thi-1 Δ(mcrC-mrr)114::IS10; New England Biolabs]. Bacteria were cultivated in Luria-Bertani (LB) medium or TY medium (67) supplemented with ampicillin and chloramphenicol at concentrations of 100 and 30 μg/ml, respectively, when required. The following plasmids were used in this work: pGW7HindIII (Ap^r HindIII R⁺/M⁺; obtained from Elisabeth Raleigh, New England Biolabs) (56) for overproduction of the HindIII restriction ENase (R · HindIII); pSX1 (Ap^r LlaCI R⁻/M⁺; provided by Jytte Josephsen, The Royal Veterinary and Agricultural University, Frederiksberg, Denmark) (49) as a source of the llaCIM gene; pGEM3Zf(+) (Ap^r; Promega) as a vector in cloning experiments; pT7-6 (Ap^r) (83) as a vector in the construction of overexpressing plasmids; pLysS (Cm^r) (80) for enzyme overproduction; and pRS415 (Ap^r, promoterless lacZ) (74) as a vector in the construction of the transcriptional fusions.

DNA isolation and manipulation.

Molecular cloning experiments were performed by using standard techniques (67). After cloning, hybrid plasmids were transformed into an appropriate E. coli strain. The structures of recombinant plasmids were verified by restriction analysis and automated dideoxy DNA sequencing. Restriction ENases and DNA-modifying enzymes were purchased from New England Biolabs and MBI Fermentas. The conditions used for enzymatic reactions were those suggested by the suppliers. Analysis of DNA fragments was performed by electrophoresis in 0.8 to 1.5% agarose slab gels supplemented with ethidium bromide (0.25 μg/ml). Electrophoresis was carried out in TBE buffer (89 mM Tris base, 89 mM boric acid, 2 mM EDTA-Na₂; pH 8.0), and DNA was visualized by UV transillumination. PCRs were performed with DyNAzyme II DNA polymerase from Thermus brockianus (Finnzymes Inc.) by using an MJ Research thermal cycler. Oligodeoxynucleotide primers were purchased from Interactiva.

R · EcoVIII assay.

R · EcoVIII activity was assayed by using a 20-μl reaction mixture containing 0.5 μg of λ DNA, 10 mM Tris-HCl (pH 7.9), 50 mM NaCl, 10 mM MgCl₂, 1 mM dithiothreitol, and 2 μl of column fractions (1 h, 37°C). The DNA was analyzed in 1% agarose gels in 1× TBE buffer. One unit of endonucleolytic activity was defined as the minimal amount of enzyme required to complete digestion of 1 μg of λ DNA in 1 h at 37°C under the assay conditions described above.

M · EcoVIII assay.

The EcoVIII MTase (M · EcoVIII) protection assay was performed by using a 20-μl reaction mixture containing 0.5 μg of λ DNA, 80 μM S-adenosyl-l-methionine (AdoMet), 10 mM Tris-HCl (pH 7.0), and 2 μl of column fractions (1 h, 37°C). The reaction was stopped by heating the mixture at 60°C for 10 min. Following cooling, 2 μl of 10× NEB2 reaction buffer (New England Biolabs) and 5 U of R · EcoVIII or R · HindIII were added (1 h, 37°C), and the DNA was analyzed by 0.8% agarose gel electrophoresis. One unit of modification activity was defined as the minimal amount of MTase that in 1 h at 37°C conferred to 1 μg of λ DNA complete resistance to cleavage by the cognate restriction ENase under the assay conditions described above. The second assay for M · EcoVIII activity was based on the enzyme-catalyzed transfer of ³H-labeled methyl groups from [methyl-³H]AdoMet to DNA. Methylation was performed in a 30-μl reaction mixture containing 5 μg of λ DNA (or synthetic oligonucleotide), 10 mM Tris-HCl (pH 7.0), 0.5 μM [methyl-³H]AdoMet (80 Ci/mmol; Amersham), and an aliquot of the enzyme (1 h, 37°C). The reaction was stopped by adding 30 μl of 50% trichloroacetic acid. A sample was centrifuged (10,000 × g, 10 min). The pellet was washed with 1 ml of 70% ethanol, centrifuged, and dried. Scintillation counting was used to estimate the incorporated radioactivity.

Protein purification.

All purification procedures were carried out at 4°C. The buffers in which bacteria were sonicated contained phenylmethylsulfonyl fluoride (0.15 mM). Each chromatographic column after protein loading was washed extensively with an appropriate buffer. The restriction ENases and MTases used in this study were purified to apparent electrophoretic homogeneity. At each step, the progress in protein purification was monitored by sodium dodecyl sulfate (SDS)-10% polyacrylamide gel electrophoresis (PAGE) (43). Gels were stained with Coomassie brilliant blue R-250 and destained in 10% (vol/vol) acetic acid-20% (vol/vol) methanol.

Purification of R · EcoVIII.

E. coli E1585-68 was cultivated in LB medium at 37°C. The culture was harvested by centrifugation after the stationary phase of growth was reached and then stored at −70°C. For standard purification, 80 g of cell paste was suspended in 260 ml of buffer P (10 mM potassium phosphate [K/PO₄] [pH 7.0], 30 mM KCl, 1 mM EDTA-Na₂, 10 mM 2-mercaptoethanol [ME], 5% [vol/vol] glycerol) and disrupted by sonication (60 10-s pulses). After centrifugation (14,000 × g, 40 min), the clarified lysate was passed over a phosphocellulose column (P-11; 2.5 by 12-cm; Whatman) equilibrated with buffer P. After they were washed with 10 column volumes of buffer P, the adsorbed proteins were eluted with a 500-ml linear gradient of KCl (0.03 to 0.43 M) in buffer P. Fractions with R · EcoVIII activity were pooled, dialyzed against buffer P, and then applied to a CM-Sephadex C-50 cation-exchange column (1.8 by 24-cm; Pharmacia) equilibrated with buffer P. Bound proteins were eluted with a 250-ml linear gradient of KCl (0.03 to 0.6 M) in buffer P. Active fractions were dialyzed against buffer DE (10 mM K/PO₄ [pH 7.8], 1 mM EDTA-Na₂, 10 mM ME) and applied to a DEAE-cellulose anion-exchange column (DE52; 2.5 by 6-cm; Whatman) equilibrated with buffer DE. Adsorbed proteins were eluted with a 150-ml linear gradient of KCl (0 to 0.2 M) in buffer DE. Active fractions were directly loaded into a hydroxylapatite Bio-Gel HTP column (1 by 5 cm; Bio-Rad) equilibrated with buffer H (10 mM K/PO₄ [pH 7.0], 0.2 M KCl, 10 mM ME, 5% [vol/vol] glycerol). Proteins bound to the column were eluted with a 150-ml linear gradient of K/PO₄ (pH 7.0) (0.01 to 0.1 M) in buffer H. Active fractions were dialyzed against buffer HA (10 mM K/PO₄ [pH 7.5], 1 mM EDTA-Na₂, 10 mM ME), applied to a heparin-agarose column (1 by 5 cm; Sigma Chemical Co.), and eluted with a 100-ml linear gradient of KCl (0 to 0.8 M) in buffer HA. The enzyme-containing fractions were concentrated by overnight dialysis against buffer E (10 mM K/PO₄ [pH 7.5], 50 mM KCl, 10 mM ME, 0.5 mM EDTA-Na₂, 50% [vol/vol] glycerol) and stored at −20°C.

Purification of M · EcoVIII.

M · EcoVIII was purified from bacteria carrying plasmid pT7-EcoVIIIM, which was constructed by cloning a 988-bp RsaI-SspI fragment from natural plasmid pEC156 (54) encoding the ecoVIIIM gene into the SmaI site of expression plasmid pT7-6, located downstream from the φ10 promoter of phage T7. In this plasmid, the start codon of ecoVIIIM is located 85 nucleotides (nt) downstream from the φ10 promoter. E. coli BL21(DE3)(pLysS) was used as a host for pT7-EcoVIIIM. Bacteria were cultivated at 37°C in 1 liter of TY medium supplemented with ampicillin and chloramphenicol. When the A₆₀₀ reached 0.4, overproduction of M · EcoVIII was induced by adding isopropyl-1-thio-β-d-galactopyranoside (IPTG) to a final concentration of 1 mM. Induction was performed for 2.5 h. Cells were harvested by centrifugation and stored at −70°C until they were used. Frozen cells (2.5 g) were thawed in 10 ml of buffer PM (10 mM K/PO₄ [pH 7.0], 50 mM KCl, 1 mM EDTA-Na₂, 10 mM ME, 5% [vol/vol] glycerol). The bacteria were sonicated (60 10-s pulses), and after centrifugation the clarified lysate was applied to a P-11 column (2.5 by 4 cm). Proteins bound to the column were eluted with a 150-ml linear gradient of KCl (0.05 to 1.0 M) in buffer PM. Active fractions were applied directly to a hydroxylapatite column (1 by 5 cm). The adsorbed proteins were eluted with a 100-ml linear gradient of K/PO₄ (pH 7.0) (0.01 to 0.1 M) in buffer H (see above). After dialysis against buffer HA (see above), active fractions were loaded onto a heparin-agarose column (1 by 5 cm) and eluted with a 100-ml linear gradient of KCl (0 to 1.0 M) in buffer HA. Fractions with the highest M · EcoVIII activity were dialyzed overnight against buffer E (see above) and stored at −20°C.

Purification of M · LlaCI.

M · LlaCI was prepared from E. coli BL21(DE3)(pLysS) transformed with pLlaMet3. This plasmid was constructed by cloning into a pT7-6 vector linearized with BamHI and SmaI a 1.15-kb DNA fragment carrying the M · LlaCI gene that was obtained by a PCR followed by BamHI digestion and T4 polynucleotide kinase phosphorylation. The forward and reverse primers were 5′-GAGGATCCACGACATTTA-3′ and 5′-ATGAATTCCAGTTTTGAT-3′, respectively (the underlining indicates the BamHI site). Plasmid pSX1 (LlaCI R⁻/M⁺) (49) was used as a template in a PCR. In recombinant plasmid pLlaMet3 the start codon of the llaCIM gene is located 164 nt downstream from the φ10 promoter of phage T7. Bacteria carrying the overproducing plasmid were cultivated at 37°C in TY broth (1 liter) supplemented with ampicillin and chloramphenicol until the A₆₀₀ was 0.3. At this time overproduction of M · LlaCI was induced by adding IPTG to a final concentration of 1 mM as described above for M · EcoVIII. Induction was performed for 2.5 h. Cells were harvested by centrifugation and stored at −70°C. For enzyme purification, cells (2.5 g) were resuspended in 10 ml of buffer PM (see above) and disrupted by sonication (60 10-s bursts). Clarified lysate was applied to a P-11 column (2.5 by 5 cm) equilibrated with buffer PM. Proteins bound to the column were eluted with a 150-ml linear gradient of KCl (0.05 to 1.0 M) in buffer PM. Active fractions were applied directly to a blue agarose column (1 by 5 cm; Pharmacia) equilibrated with buffer BA (10 mM K/PO₄ [pH 7.0], 1 mM EDTA-Na₂, 10 mM ME, 5% [vol/vol] glycerol). Protein was eluted with a 100-ml linear gradient of KCl (0 to 1 M) in buffer BA. The final step was chromatography on Superose 12HR (fast-performance liquid chromatography; Pharmacia) equilibrated with buffer SU (10 mM K/PO₄ [pH 7.6], 150 mM KCl, 1 mM EDTA-Na₂, 10 mM ME, 2.5% [vol/vol] glycerol). Fractions with the highest M · LlaCI activity were dialyzed overnight against buffer E (see above) and stored at −20°C. The M · LlaCI activity was assayed essentially as described above for the M · EcoVIII activity.

Purification of R · HindIII.

R · HindIII was prepared from E. coli ER2418 transformed with plasmid pGW7HindIII (56). Bacteria were cultivated at 30°C in LB medium (1 liter) supplemented with ampicillin. When the A₆₀₀ of the culture reached 0.3, overproduction of R · HindIII was induced by raising the temperature to 42°C for 15 min, followed by incubation for 4 h at 37°C. The harvested bacteria were stored at −70°C. Frozen cells (2.2 g) were resuspended in 10 ml of buffer PH (10 mM K/PO₄ [pH 6.5], 50 mM KCl, 1 mM EDTA-Na₂, 10 mM ME, 5% [vol/vol] glycerol) and disrupted by sonication (60 10-s bursts). Clarified lysate was applied to a P-11 column (2.5 by 5 cm) equilibrated with buffer PH. Protein bound to the column was eluted with a 150-ml linear gradient of KCl (0.05 to 1.0 M) in buffer PH. Active fractions were applied to a hydroxylapatite column (1 by 5 cm) equilibrated with buffer H (see above). Protein was eluted with a 100-ml linear gradient of K/PO₄ (pH 7.0) (0.01 to 0.1 M) in buffer H. Active fractions were loaded onto a heparin-agarose column (1 by 5 cm) and eluted with a 100-ml linear gradient of KCl (0 to 1.0 M) in buffer HA (see above). Fractions with the highest R · HindIII activity were dialyzed overnight against buffer E (see above) and stored at −20°C. To test R · HindIII activity, an assay used for the R · EcoVIII enzyme was used.

Purification of M · HindIII.

M · HindIII was prepared from E. coli BL21(DE3)(pLysS) transformed with plasmid pT7-6-M · HindIII. This plasmid was constructed by cloning into a pT7-6 vector linearized with BamHI and EcoRI a 1.22-kb DNA fragment carrying the M · HindIII gene that was obtained by performing a PCR followed by double digestion with EcoRI and BamHI. The forward and reverse primers were 5′-CGGAATTCCAATGCAATAG-3′ and 5′-GAAGAAGGATCCTTGATAG-3′, respectively (the underlining indicates EcoRI and BamHI sites, respectively). Plasmid pGW7HindIII (HindIII R⁺/M⁺) (56) was used as a template in the PCR. In plasmid pT7-6-M · HindIII the start codon is located 181 nt downstream from the φ10 promoter of phage T7. Overproduction of M · HindIII was induced by adding IPTG to a final concentration of 1 mM as described above for M · EcoVIII. Harvested cells (3.0 g) were resuspended in 10 ml of buffer PD (10 mM K/PO₄ [pH 7.1], 0.15 M KCl, 1 mM EDTA-Na₂, 10 mM ME, 5% [vol/vol] glycerol) and disrupted by sonication (60 10-s bursts). Clarified lysate was applied to a P-11 column (2.5 by 5 cm) equilibrated with buffer PD. Protein bound to the column was eluted with a 200-ml linear gradient of KCl (0.15 to 1.5 M) in buffer PD. Active fractions were applied to a blue agarose column (1 by 5 cm) equilibrated with buffer BH (10 mM K/PO₄ [pH 7.5], 50 mM KCl, 1 mM EDTA-Na₂, 10 mM ME, 5% [vol/vol] glycerol). Protein was eluted with a 150-ml linear gradient of KCl (0.05 to 1.2 M) in buffer BH. The final step was chromatography on a column (1 by 5 cm) with heparin-agarose equilibrated with buffer AH (10 mM K/PO₄ [pH 7.1], 1 mM EDTA-Na₂, 10 mM ME, 5% [vol/vol] glycerol). The enzyme was eluted with a 100-ml linear gradient of KCl (0 to 1.0 M) in buffer AH. Fractions with the highest M · HindIII activity were dialyzed overnight against buffer E (see above) and stored at −20°C. The M · HindIII activity was assayed essentially as described above for the M · EcoVIII enzyme activity.

Molecular weight determination.

Purified preparations of R · EcoVIII and M · EcoVIII were analyzed by SDS-PAGE (43) in order to estimate purity and molecular weight under denaturing conditions. After electrophoresis, protein positions were visualized by Coomassie brilliant blue R-250 staining. The relative molecular weights of the EcoVIII enzymes were calculated by using a calibration curve obtained with the following standard proteins: phosphorylase b (M_r, 94,000), bovine serum albumin (67,000), ovalbumin (43,000), carbonic anhydrase (30,000), trypsin inhibitor (20,100), and α-lactalbumin (14,400). The reference proteins were purchased from Pharmacia.

Determination of the R · EcoVIII cleavage site.

The cleavage site of the R · EcoVIII enzyme was determined by the primer extension method (8) by using plasmid pGEM3Zf(+) (Promega) as a template. Primers T7 (5′-TAATACGACTCACTATAG) and M13/pUC reverse (5′-CAGGAAACAGCTATGAC), labeled with [γ-³²P]ATP (>5,000 Ci/mmol; Amersham), were extended with T7 DNA polymerase (Sequenase 2.0; United States Biochemical Corp.) under standard conditions beyond the EcoVIII recognition site. After protein removal (33) the DNA was subjected to R · EcoVIII digestion. To localize the EcoVIII cleavage site, dideoxy DNA sequencing reactions with primer T7 or M13/pUC reverse and plasmid pGEM3Zf(+) as the template were performed on the same sequencing gel (6% acrylamide-bisacrylamide [19:1], 7 M urea) side by side with EcoVIII digestion products of the extension reactions.

Determination of the methylated base.

To determine the base methylated by M · EcoVIII, the method employing type IIS restriction ENases was used (59). In order to obtain proper templates, two synthetic DNA fragments carrying the EcoVIII recognition site overlapped by the MboII cut site (5′-AATTCGAAGAATCGATCA-3′ and 3′-GCTTCTTAGCTAGTTCGA; and 5′-AATTCGAAGATCGATCA-3′ and 3′-GCTTCTAGCTAGTTCGA-5′) were cloned into plasmid pGEM3Zf(+) double-digested with HindIII and EcoRI, resulting in plasmids pIM1 and pIM2, respectively. These plasmids differ by 1 bp in the region between the EcoVIII and MboII sites. Plasmid pIM3 without the EcoVIII site was used as a control. This plasmid was constructed by digestion of pIM1 with R · EcoVIII, followed by filling-in of the 5′ protruding ends with the Klenow fragment in the presence of deoxynucleoside triphosphates. After ligation and transformation into E. coli MM294 cells, recombinants were checked by DNA sequence analysis. To determine the EcoVIII methylation pattern, 346-bp (pIM1), 345-bp (pIM2), and 350-bp (pIM3) DNA fragments were amplified by using pair of primers complementary to the pGEM3Zf(+) vector (ADE1 [5′-TTACGCCAGCTGGCGAAAG] and ADE2 [5′-CATTAATGCAGCTGGCAC]). DNA fragments carrying a synthetic oligomer isolated either from pIM1 (346 bp) or from pIM2 (345 bp) were methylated by M · EcoVIII. In the control experiment a 350-bp DNA fragment without an EcoVIII site, derived from plasmid pIM3, was used. Methylation was performed in a 30-μl reaction mixture containing 0.5 μg of a DNA fragment (345, 346, or 350 bp), 10 mM Tris-HCl (pH 7.0), 10 μCi of [³H]AdoMet (80 Ci/mmol; Amersham), and 50 ng of M · EcoVIII (1 h, 37°C). Cleavage by the MboII enzyme of the [methyl-³H]DNA fragment separated either adenine from adenine (346-bp DNA fragment) or adenine from the guanine residue (345-bp DNA fragment) within the M · EcoVIII recognition site. After electrophoresis, DNA fragments were extracted from a 1.5% agarose gel by using a method involving a DEAE-cellulose membrane (34). Scintillation counting was used to estimate the radioactivities of DNA fragments.

Analysis of the N-terminal amino acid sequences of R · EcoVIII and M · EcoVIII.

In order to determine the amino acid sequence of the N-terminal region of an enzyme, 300 pmol of a homogeneous preparation of R · EcoVIII or M · EcoVIII was loaded onto an Applied Biosystems 491 gas phase protein sequencer. The phenylthiohydantoin (PTH) derivatives of the amino acids were identified with an Applied Biosystems 140C PTH analyzer connected to the sequencer. The first 10 PTH amino acids were unambiguously identified.

Computational analysis of DNA and proteins.

Nucleotide and protein sequences were searched for among the sequences in the GenBank database available on the National Center for Biotechnology Information web page (http://www.ncbi.nlm.nih.gov) by using the BLAST program (2). Protein sequences were aligned by using the CLUSTAL W program (85) accessible through European Bioinformatics Institute server (http://www.ebi.ac.uk). Genes encoding the EcoVIII R-M system were analyzed with DNASIS software (Hitachi Software Engineering). Codon usage data were obtained from the Kazusa codon usage database (Kazusa DNA Research Institute, Chiba, Japan; http://www.kazusa.or.jp).

Western blot analysis.

Homogeneous preparations of the enzymes (R · EcoVIII, R · HindIII, M · EcoVIII, M · LlaCI, and M · HindIII) were separated by SDS-10% PAGE and then electroblotted for 12 h at 20 V onto a nitrocellulose membrane (BA85; Schleicher and Schuell) in electrotransfer buffer (25 mM Tris base, 192 mM glycine; pH 8.3) by using a Biotrans electrophoretic transfer cell (Kucharczyk TE). After blocking with 3% skim milk, the membrane was incubated with a 1:25 dilution of either rabbit anti-R · EcoVIII or anti-M · EcoVIII polyclonal antibodies which were prepared by standard protocols (26). The primary antibodies were tagged successively with a 1:10,000 dilution of goat anti-rabbit polyclonal antibodies conjugated with alkaline phosphatase (Sigma Chemical Co.). All antibodies were diluted in phosphate-buffered saline (67). Reactive bands were visualized by using 5-bromo-4-chloro-3-indolylphosphate (BCIP) as the alkaline phosphatase substrate and nitroblue tetrazolium as the color development reagent.

Effect of divalent metal ions on M · EcoVIII activity.

The activity of M · EcoVIII was measured in a reaction buffer (10 mM Tris-HCl [pH 7.0]) containing divalent cations [Mg(II), Mn(II), Ca(II), and Zn(II)] at different concentrations. Enzymatic activity was assayed by measuring the incorporation of ³H-labeled methyl groups from [methyl-³H]AdoMet into 2.5 pmol of a 29-bp double-stranded oligonucleotide (5′TGCAGTCGCGAAGCTTGGTCACCTTGAGG-3′ and 3′-GTCAGCGCTTCGAACCAGTGAACTCCGT-5′ [the underlining indicates the EcoVIII site]) by using 35 ng of the enzyme. Each experiment was repeated at least twice. For each cation the 50% inhibitory concentration (IC₅₀) was determined.

Construction of transcriptional fusions.

Plasmid pRS415 (74) carrying a promoterless lacZ reporter gene was used to test the strength of the ecoVIIIR and ecoVIIIM promoters. Plasmid pIM4 was constructed by insertion of a 200-bp PCR-amplified DNA fragment carrying the ecoVIIIR promoter region into pRS415 digested with SmaI. To do this, the following two phosphorylated primers were used: 5′-CCAGGGGCGAGCTCAGGT-3′ and 5′-CCTTCCAGTGTTACAAAC-3′. Plasmid pIM5 was obtained by insertion of a 156-bp PCR-amplified DNA fragment carrying the ecoVIIIM promoter region into pRS415 digested with EcoRI and BamHI. The restriction sites used in DNA cloning were provided by a pair of PCR primers (5′-ATCGAATTCCCATTGGTAACGG-3′ and 5′-GAAAGGATCCGATGAAAGG-3′ [the underlining indicates EcoRI and BamHI sites, respectively]). In both PCRs plasmid pEC156 (54) was used as a template. The proper orientation of inserts with respect to the reporter gene was confirmed by DNA sequencing. Cells of E. coli MC1061 transformed with the appropriate plasmid (pIM4 or pIM5) were assayed for β-galactosidase activity as previously described (52).

Protein cross-linking with glutaraldehyde.

In order to determine the state of aggregation in solution, EcoVIII enzymes were subjected to a cross-linking reaction, which was performed by using the previously described protocol (69). The EcoVIII R-M enzymes were incubated at 30°C for 10 min in 30-μl reaction mixtures in standard buffers. Glutaraldehyde (1.9 μl of a 2.5% solution) was added to each reaction mixture (2 min, 30°C), and this was followed by sodium borohydride treatment (2.6 μl of a 1 M solution, 20 min, 4°C). The reaction was terminated by adding Tris-HCl (pH 7.5) (9 μl of a 1 M solution) and incubating the preparation for 5 min at 4°C. Samples were analyzed by SDS-5% PAGE (87), followed by silver staining (28).

Nucleotide sequence accession number.

The nucleotide sequence of plasmid pEC156 coding for the EcoVIII R-M system has been deposited in the GenBank database under accession number AF158026.

RESULTS

Genetic organization of the EcoVIII R-M system.

The genes coding for the EcoVIII R-M system are carried by natural plasmid pEC156 (53, 54). The complete nucleotide sequence of pEC156 was determined in our laboratory. The plasmid is composed of 4,312 bp. A computational search for potential open reading frames (ORF) revealed the presence of two ORFs (ORF1 and ORF2), clustered and convergently oriented, large enough to encode the genes of the EcoVIII R-M system. The two genes are located close to each other. There is no intergenic region between them. The overall guanine-plus-cytosine (G+C) content of these genes is 36.1% (34.5% for ecoVIIIR and 37.7% for ecoVIIIM), which is significantly lower than the average G+C content of either plasmid pEC156 (43.6%) (54) or E. coli genomic DNA (50.8%) (5). The difference suggests that there is a possibility that the EcoVIII R-M system was recently acquired by the genome. Deletion analysis indicated that ORF1, extending from nt 802 to nt 1722, encodes R · EcoVIII (307 amino acids), whereas ORF2, extending from nt 1729 to nt 2640 on the complementary strand, encodes M · EcoVIII (304 amino acids). ORF1 showed homology to the HindIII and LlaCI ENase genes (Fig. 1A), whereas ORF2 showed a high level of identity to the corresponding HindIII and LlaCI MTase genes (Fig. 1B). This information was obtained by performing a BLAST search. In the sequence analyzed, we did not find an ORF coding for a putative C protein involved in some systems in expression control of the R-M genes (84).

FIG. 1.

Comparison of the amino acid sequences of R · EcoVIII, R · HindIII, and R · LlaCI (A) and the amino acid sequences of M · EcoVIII, M · HindIII, and M · LlaCI (B). The numbers on the right indicate the amino acid positions relative to the N terminus. Sequences were aligned by using the CLUSTAL W computer program. The region of pronounced homology in all three restriction ENases is enclosed in a box. The amino acids of the putative catalytic-magnesium binding motif PD/EX_nD/EXK and the putative DNA binding motif RXXR are indicated. The conserved motifs of m⁶N-adenine MTases are enclosed in boxes and labeled with Roman numerals. The position of the putative target recognition domain (TRD) is indicated. The accession numbers for the nucleotide sequences of the EcoVIII, HindIII and LlaCI R-M genes that have been deposited in the GenBank database are AF158026, L15391, and AJ002064, respectively. Identical amino acids are indicated by asterisks; colons and periods indicate very similar and somewhat similar amino acids, respectively; and dashes indicate gaps in the aligned sequences.

The genes of the EcoVIII R-M system are transcribed from separate promoters. The exact positions of the start codon (AUG) of the ecoVIIIR and ecoVIIIM genes were determined by microsequencing the N termini of R · EcoVIII and M · EcoVIII, respectively. The first 12 amino acids of R · EcoVIII were found to be MDLTNIEFINDA, while the first 10 amino acids of M · EcoVIII were found to be MLNXIQLPDE (X = unidentified amino acid). These sequences correspond to those predicted from the nucleotide sequences of the respective genes. We determined that transcription of the genes analyzed begins 92 and 13 nt upstream from the start codon for the ecoVIIIR gene and the ecoVIIIM gene, respectively (54). Based on these data, for each promoter, the −10 and −35 sequences were determined under the conditions necessary to identify consensus sequences (46). The −10 and −35 regions of these promoters are 5′-TTTCAT and 5′-CTGATA for the ecoVIIIR promoter, respectively, and 5′-TTTAAA and 5′-TTAACG for the ecoVIIIM promoter, respectively. These sequences differ from the E. coli consensus promoter sequences (−10, 5′-TATAAT; −35, 5′-TTGACA) (46).

The DNA sequence upstream of the ecoVIIIR gene contains a putative ribosome binding site (AAGG) 10 bp from the start codon (ATG). The ecoVIIIM gene has a ribosome binding site (AGGA) located 6 bp upstream from the start codon. The spacing between the ribosome binding site and the initiation codon varies in natural transcripts, and the average distance is 7 nt. Experimental measurements of protein expression revealed that a spacer that is 5 nt long is optimal. Longer or shorter spacers may be deleterious to efficient initiation of translation (17).

Analysis of the amino acid sequences.

The predicted amino acid sequence of R · EcoVIII, when it was compared in pairs, showed 26% identity to the R · HindIII sequence and 17% identity to the R · LlaCI sequence. The amino acid sequence of M · EcoVIII showed 62% identity to the M · LlaCI sequence and 50% to the M · HindIII sequence. The overall level of identity for all enzymes analyzed was low for ENases (9%) and substantial for MTases (47%). Despite the fact that the two EcoVIII enzymes recognize the same specific sequence, there was no homology between the restriction ENase and the MTase when their deduced amino acid sequences were compared. The data presented here were obtained by using the CLUSTAL W computer program.

The predicted amino acid sequence of R · EcoVIII allowed a putative catalytic-Mg(II) binding sequence motif characteristic of restriction ENases, PE⁶⁶X₅₄DXK¹²³, to be located in the N-terminal portion of the enzyme (Fig. 1A). This motif is also present in HindIII (PE⁵²X₅₆DXK¹¹¹) (18) and the LlaCI restriction enzyme (PD⁴⁹X₅₄DXK¹⁰⁶) (48). Site-directed mutagenesis of the gene encoding the HindIII enzyme produced evidence that any mutation within the P⁵¹X₅₆DXK¹¹¹ catalytic motif eliminated enzyme activity (18). However, analysis of the R · HindIII molecular model constructed by Janusz Bujnicki (Bioinfoformatics Laboratory, International Institute of Molecular and Cell Biology, Warsaw, Poland) revealed that the PE⁵² residues are located a significant spatial distance from the second part of the catalytic motif (DXK¹¹¹) (data not shown). Therefore, we concluded that most probably the catalytic motif of this enzyme is different from that suggested above and comprises charged residues located close together. The spatial architecture of the R · HindIII molecular model prompted us to propose that the D⁹⁴X₁₂DXK¹¹¹ motif is involved in the formation of the enzyme active site. The role of the PE⁵² motif remains obscure. Thus, the suggested catalytic motifs for R · EcoVIII and R · LlaCI are D¹⁰⁸X₁₂DXK¹²³ and D⁹¹X₁₂DXK¹⁰⁶, respectively (Fig. 1A). The catalytic motif complements a region of pronounced homology composed of 20 amino acids, which is present in all three ENases analyzed (Fig. 1A). In the same region we noted the presence of a putative DNA binding motif (RXXR) whose importance has been postulated for several restriction ENases (58). In the amino acid sequences of all three restriction enzymes, in addition to the common catalytic motif, we found 28 other identical amino acid residues (Fig. 1A). Since the locations of these amino acids are identical in all three ENases analyzed, we hypothesized that the functions of these residues might be the same.

The presence and distribution of nine highly conserved amino acid sequence motifs and a putative target recognition domain in the enzyme structure suggest that M · EcoVIII belongs to the m⁶N-adenine β-class of MTases (Fig. 1B). These motifs can be grouped in three clusters which are responsible for the following three principal functions: (i) sequence-specific DNA recognition (target recognition domain); (ii) binding of a methyl group donor, AdoMet (motifs X, I, and II); and (iii) catalysis of methyl group transfer (motifs III, IV, V, VI, VII, and VIII) (10, 24, 50, 88). The same organization of motifs was also observed in M · HindIII and M · LlaCI (Fig. 1B).

Codon usage of the EcoVIII R-M genes.

The codon usage in both EcoVIII genes reflects the low G+C content (36.1%) and a marked A or T bias at the wobble position. In the ecoVIIIR gene, 71.3% of the codons end with either A or T, and in the ecoVIIIM gene 74.3% of the codons have such an ending. The values obtained are significantly higher than those obtained for other E. coli genes (44.1%). In order to determine how the codon usage pattern of the EcoVIII R-M genes differs from that of E. coli, we calculated the value of the codon adaptation index (CAI) for each EcoVIII gene using a standard method (72). The CAI evaluates synonymous codon usage bias according to the codon usage of highly expressed genes. This means that the CAI can also be used as a factor to predict the range of gene expression. Genes with a high CAI (near 1) belong to the family of highly expressed housekeeping genes. On the other hand, genes with low CAI values may be recent horizontal transfer acquisitions that still show the optimal codon usage of their former host (37). The CAI values obtained (0.151 and 0.185 for the ecoVIIIM and ecoVIIIR genes, respectively) indicate that both genes belong to the subset of E. coli genes that were most probably acquired by horizontal gene transfer.

Analysis of the ecoVIIIR and ecoVIIIM promoter strength.

Two promoters are associated with the EcoVIII R-M system. They were identified on the basis of transcriptional analysis (54), as well as by comparison with other E. coli promoter sequences. In front of the R · EcoVIII coding region (17 nt from the transcription start point; coordinates 667 to 692) we found a 12-bp perfect inverted repeat which may form a stable stem-loop structure (ΔG = −22.2 kcal/mol). This structure overlaps postulated −35 and −10 sequences in the ecoVIIIR promoter region. It has been shown that similar secondary structures present in mRNA greatly reduce gene expression in some cases (R · EcoRV [6], R · FokI [40], R · TaqI [3]), which indicates that these structures are involved in some kind of regulation mechanism. No such secondary structure is present in the promoter region of the ecoVIIIM gene. In order to evaluate the strength of both promoters and, indirectly, the effect of the observed stem-loop structure on gene expression, we cloned DNA fragments carrying the ecoVIIIR or ecoVIIIM promoter regions (from nt −117 to nt 83 and from nt −68 to nt 70 with respect to the predicted starting points of transcription of the ecoVIIIR and ecoVIIIM genes, respectively) into the vector pRS415 upstream of a promoterless lacZ gene. The correct orientation of each promoter was checked by DNA sequencing. After introduction of recombinant plasmids into E. coli MC1061, the cells were tested for β-galactosidase activity. As a result, we found that expression of β-galactosidase dependent on the presence of the ecoVIIIM promoter was slightly higher (12%) than expression of β-galactosidase dependent on the presence of the ecoVIIIR promoter (data not shown).

Overproduction and purification of EcoVIII enzymes.

Overproduction of M · EcoVIII is toxic for E. coli BL21(DE3) cells, most probably because background transcription of the lacUV5 promoter controlling expression of the gene coding for T7 RNA polymerase. No toxicity was observed when E. coli MM294 was used as a host for overproducing plasmid pT7-EcoVIIIM. This strain does not contain a copy of the T7 RNA polymerase gene. In E. coli BL21(DE3)(pT7-EcoVIIIM) small amounts of T7 RNA polymerase permitted expression of the ecoVIIIM gene and consequently caused inhibition of E. coli growth. To overcome this, we used plasmid pLysS encoding T7 lysozyme, which is a potent inhibitor of T7 RNA polymerase (80). Therefore, for M · EcoVIII enzyme purification on a large scale, E. coli BL21(DE3)(pLysS, pT7-EcoVIIIM) was used. The standard purification procedure consisted of three chromatographic steps. The final enzyme preparation did not contain nonspecific nucleases. From 2.5 g of bacteria we were able to obtain 1.5 mg of homogeneous enzyme preparation with an overall yield of 18%.

On the other hand, despite many efforts we were not able to construct a plasmid overexpressing R · EcoVIII. Therefore, we were forced to purify this enzyme from a natural producer, E. coli E1585-68. An electrophoretically homogeneous preparation of R · EcoVIII enzyme free of detectable nonspecific nucleases was obtained after four chromatographic steps. From 80 g of cell paste we obtained 0.15 mg of the enzyme.

The purified preparations of EcoVIII R-M enzymes were analyzed by SDS-PAGE in order to estimate the purity and molecular weight (M_r) under denaturing conditions. Both enzymes were found to be at least 95% pure on Coomassie blue-stained gels (Fig. 2, lanes 2 and 3). By using the standards having known M_rs, M_rs of 36,700 ± 1,000 for R · EcoVIII and 36,000 ± 1,000 for M · EcoVIII were calculated, which are close to the M_rs of 35,554 and 33,930 deduced from the nucleotide sequences of the genes encoding R · EcoVIII and M · EcoVIII, respectively.

FIG. 2.

Coomassie blue-stained SDS-10% polyacrylamide gel of purified EcoVIII R-M enzymes. Lanes 1 and 4, molecular weight markers (Pharmacia); lane 2, R · EcoVIII preparation after chromatography on a heparin-agarose column (10 μg of protein); lane 3, M · EcoVIII preparation after chromatography on a heparin-agarose column (3 μg of protein). The molecular masses of reference proteins (Pharmacia) (in kilodaltons) are indicated on the right and left.

Protein cross-linking experiment.

To determine the state of aggregation, both EcoVIII R-M enzymes were tested by using a glutaraldehyde cross-linking reaction under optimal conditions (10 mM K/PO₄ [pH 7.0], 80 μM AdoMet, and 10 mM EDTA-Na₂ for M · EcoVIII; 10 mM K/PO₄ [pH 7.9], 10 mM MgCl₂, 10 mM KCl, and 1 mM dithiothreitol for R · EcoVIII) with M · EcoVIII (2.5 μg) and R · EcoVIII (1.8 μg). As a result, we found that M · EcoVIII exists in solution predominantly as a monomer (Fig. 3, lane 3) and that R · EcoVIII exists as a dimer (Fig. 3, lane 5). A well-known complex-forming protein, E. coli DnaK, was used as a control. Upon cross-linking, the DnaK protein, which is known to produce multimeric forms (45), was found to produce three distinct bands corresponding to DnaK monomers, dimers, and trimers (Fig. 3, lane 2).

FIG. 3.

Cross-linking of EcoVIII enzymes with glutaraldehyde. The SDS-5% PAGE gel was silver stained. The results of the cross-linking reaction for the following proteins are shown: DnaK (3.5 μg) (lane 2); M · EcoVIII (2.5 μg) (lane 3); and R · EcoVIII (1.8 μg) (lane 5). The following proteins that were not treated with glutaraldehyde were used as controls: DnaK (3.5 μg) (lane 1); M · EcoVIII (2.5 μg) (lane 4); and R · EcoVIII (1.8 μg) (lane 6). The positions of the corresponding forms of the proteins after cross-linking are indicated (I, monomer; II, dimer; III, trimer).

Determination of the cleavage and methylation specificity of the EcoVIII enzymes.

When commonly used plasmid DNAs (λ, pBR322, pUC19, etc.) are digested, the lengths of the restriction fragments produced by the R · EcoVIII correspond to those predicted for the HindIII enzyme (data not shown). The cleavage site of R · EcoVIII was determined by using the primer extension method (8) with plasmid pGEM3Zf(+) as the template and primers T7 and M13/pUC reverse (Fig. 4). When information on the EcoVIII restriction pattern was combined with the results of the cleavage site determination analysis, it was clear that R · EcoVIII recognizes the palindromic sequence 5′-AAGCTT-3′ and cleaves double-stranded DNA symmetrically between two adenine residues, leaving 4-nt 5′ protruding ends. These results demonstrate that R · EcoVIII is a homoisoschizomer of R · HindIII, an enzyme of H. influenzae Rd.

FIG. 4.

Identification of the cleavage points of R · EcoVIII. The experiment was performed by using the method described previously (8). Plasmid pGEM3Zf(+) double-stranded DNA was alkaline denatured and then annealed with γ-³²P-labeled primer T7 (5′-TAATACGACTCACTATAG) (A) or M13/pUC reverse (5′-CAGGAAACAGCTATGAC) (B). Both oligonucleotides were extended through the EcoVIII (HindIII) site by using the T7 DNA polymerase (Sequenase 2.0), and the resulting products were digested with the R · EcoVIII enzyme and separated on a 6% polyacrylamide sequencing gel (lanes 1 and 2, respectively). The sizing ladders (lanes G, A, T, and C) were obtained by using either primer T7 (A) or primer M13/pUC reverse (B) and plasmid pGEM3Zf(+) as the template. The sequence targeted by R · EcoVIII is indicated by boldface type. The positions of cleavage points are indicated by arrows.

In the case of the MTase, we first found that λ DNA methylated by the M · EcoVIII enzyme is protected from R · HindIII cleavage (data not shown). Precise information on the base methylated by M · EcoVIII was obtained by using a method which involves class IIS restriction ENases (59). We used two kinds of DNA fragments with the EcoVIII site next to the recognition site of the MboII enzyme. The two sites were located in a way that ensured separation of the two adenine residues or adenine and guanine residues within the EcoVIII recognition site after cleavage of the ³H-labeled methylated DNA fragment by the MboII enzyme (Fig. 5A and B). After methylation of the DNA fragments with M · EcoVIII in the presence of [methyl-³H]AdoMet as the methyl group donor, we found that the products of MboII digestion of either fragment a or b were ³H labeled (Fig. 5D). This suggested that the base modified by M · EcoVIII is the first adenine residue in the sequence 5′-^6mAAGCTT-3′. This was confirmed when a 356-bp fragment derived from pIM2 was used as a substrate in a methylation reaction (Fig. 5B and D). In a control experiment the DNA fragment lacking an EcoVIII site failed to be a substrate for M · EcoVIII modification activity (Fig. 5C and D). The observed difference in scintillation counts between products derived from plasmids pIM1 and pIM2 was a result of the presence of traces of exonucleolytic activity contaminating the commercial preparation of the MboII enzyme. We experienced the same problem during purification of MboII ENase to catalytic homogeneity that was described in a previous study (71). The results obtained demonstrate that M · EcoVIII is a homomethylomer of M · HindIII, an enzyme of H. influenzae Rd.

FIG. 5.

Analysis of the M · EcoVIII methylation pattern. DNA fragments (346 bp [A], 345 bp [B], and 350 bp [C]) were amplified by using primers ADE1 and ADE2. Plasmids pIM1 (A), pIM2 (B), and pIM3 (C) were used as the templates. Specific DNA fragments were methylated by M · EcoVIII by using [³H]AdoMet. After cleavage with the MboII enzyme, DNA was electrophoresed in a 1.5% agarose gel, and restriction fragments (indicated by lowercase letters) were isolated. The labeled methyl group contents of the fragments are shown in panel D. The base modified by M · EcoVIII is indicated by an asterisk. The MboII cleavage sites are indicated by arrows.

Effect of divalent metal ions on the activity of M · EcoVIII.

We confirmed that Mg(II) ions are necessary for R · EcoVIII activity. This activity is retained when Mg(II) ions are replaced by Mn(II) (0.5 mM). Higher concentrations of Mn(II) (>1 mM) inhibit the enzyme activity (data not shown). However, in the case of M · EcoVIII, divalent metal ions are strong inhibitors [IC₅₀ of Mg(II), 2.0 mM]. The IC₅₀ values obtained for other divalent cations were as follows: Ca(II), 1.5 mM; Mn(II), 0.8 mM; and Zn(II), 0.04 mM.

Immunological analysis of EcoVIII enzymes.

Based on the similarities of the amino acid sequences, we attempted to determine if the enzymes shared epitopes with homologous proteins. Thus, the ability of antiserum prepared against M · EcoVIII or R · EcoVIII to cross-react with isospecific enzymes (M · LlaCI, M · HindIII, R · HindIII) was examined by the immunoblot method (Western blotting). Despite the observed homology of all three isospecific MTases, we found cross-reactions of antibodies raised against the M · EcoVIII enzyme with the M · LlaCI protein but not with the M · HindIII protein (Fig. 6B, lanes 2 and 3). A homogeneous preparation of the M · EcoVIII enzyme was used as the control (Fig. 6B, lane 1). This experiment demonstrated that M · EcoVIII possesses substantial antigenic similarity to the M · LlaCI enzyme, suggesting that the two proteins may have identical epitopes. The two MTases, M · EcoVIII and M · LlaCI, are 62% identical at the amino acid sequence level. The level of identity of M · EcoVIII and M · HindIII is much lower (50%).

FIG. 6.

Immunological analysis of the EcoVIII R-M enzymes and their homologs. (A) Purified preparations of M · EcoVIII (lane 1), M · LlaCI (lane 2), and M · HindIII (lane 3) separated by SDS-10% PAGE. (B) Immunological cross-reactivity of polyclonal anti-M · EcoVIII antibodies with preparations of purified MTases blotted onto a nitrocellulose membrane. (C) Purified preparations of R · EcoVIII (lane 1) and R · HindIII (lane 2) separated by SDS-10% PAGE. (D) Immunological cross-reactivity of polyclonal anti-R · EcoVIII antibodies with preparations of purified restriction ENases blotted on a nitrocellulose membrane. Lanes in panels B and D correspond to those of panels A and C, respectively. The same amount (2.0 μg) of each protein was loaded on the gel. Anti-rabbit alkaline phosphatase conjugate, BCIP, and nitroblue tetrazolium were used to detect specific complexes.

When anti-R · EcoVIII serum was used, a positive cross-reaction was clearly visible with the purified R · EcoVIII enzyme (positive control) but not with R · HindIII (Fig. 6D, lanes 1 and 2). This result is consistent with the observed weak identity of the two proteins (26%) at the amino acid sequence level (Fig. 1A).

DISCUSSION

In this paper we present genetic organization and molecular analysis data for the EcoVIII R-M system from E. coli E1585-68. The nucleotide sequence of this system is homologous to the nucleotide sequences of the HindIII system from H. influenzae Rd (56) and the LlaCI system from L. lactis subsp. cremoris W15 (49), which are the only two isospecific R-M systems that have been sequenced. However, a significant level of identity is observed only between isospecific MTases. The homology between ENases is weak (less than 30% identity when the enzymes are compared in pairs) and is limited to the putative catalytic center. Both the EcoVIII and LlaCI R-M systems are carried by naturally occurring plasmids (49, 54), whereas the HindIII system is encoded by cryptic prophage φflu located in the H. influenzae Rd genome (27). Moreover, there are differences in the arrangement of the genetic units. While the genes of the EcoVIII and LlaCI R-M systems converge, the genes of the HindIII system are arranged in tandem, with the gene encoding the ENase preceding the gene encoding the cognate MTase. It has been proposed that localization of genes encoding R-M enzymes on mobile genetic elements (plasmids and prophages) can greatly facilitate spread of the genes among bacteria by means of horizontal gene transfer (29, 31, 41). This idea is in agreement with observed differences in G+C contents between a particular R-M system and its parental genomic DNA. The overall G+C content of the EcoVIII R-M system is 36.1% (54), while the overall G+C content of the LlaCI R-M system is 29.7% (49) and the overall G+C content of the HindIII R-M system is 32.0% (56). All three isospecific R-M systems have G+C contents that are significantly lower than those of the parental bacterial genomic DNA (E. coli, 50.8% [5]; H. influenzae, 38.0% [19]; L. lactis, 34.8 to 35.6% [68]). On the other hand, the observed differences reflect the general tendency that plasmids and phages usually have lower G+C contents than their hosts. Most probably, this results from competition of these genetic elements for bacterial metabolic resources, as proposed recently (63). The low G+C content of EcoVIII genes is reflected in their codon preferences and the presence of codons which are infrequently used in E. coli (e.g., AGA for Arg, AUA for Ile, or CUA for Leu). It was shown previously that in case of regulatory genes, the presence of rare codons can modulate gene expression (16, 23, 42). Calculated CAI values for the ecoVIIIM and ecoVIIIR genes indicate that both of these genes were most probably acquired through horizontal gene transfer. The results obtained are consistent with data on codon usage deviation for other R-M systems isolated from E. coli (EcoRI and EcoRV) and from other members of the family Enterobacteriaceae (KpnI of Klebsiella pneumoniae; SmaI of Serratia marcescens; SinI of Salmonella infantis; PvuII of Proteus vulgaris) (31). This indicates that interspecies genetic exchange contributed to the evolution and spread of these systems.

We noted the presence of a characteristic amino acid motif in the primary structure of R · EcoVIII (PD⁶⁴X₅₄DXK¹²³); however, analysis of the molecular model of the R · HindIII enzyme enabled us to propose, by analogy, that the sequence of the R · EcoVIII catalytic motif is D¹⁰⁸X₁₂DXK¹²³. According to the model proposed by Horton et al. (30) for other type II restriction ENases (e.g., EcoRV), charged residues present in this motif are involved in the process of catalysis. Site-directed mutagenesis of the gene encoding the HindIII enzyme (isoschizomer of R · EcoVIII) produced evidence that any mutation within the P⁵¹X₅₆DXK¹¹¹ motif eliminates enzyme activity (18). However, data derived from the crystallographic structure and site-directed mutagenesis of R · Cfr10I and homing ENase I-PpoI provided evidence that the presence of this motif is not as important as the spatial architecture of the active site. For instance, in the R · Cfr10I enzyme, the second acidic residue of the PD¹³⁴X₅₃(E²⁰⁴)SVK₁₉₀ motif that is engaged for the catalytic site comes from a distal part of the protein (7, 75). In a recent study similar active site architecture was also proposed for other restriction ENases belonging to subfamilies IIE and IIF (58). In the case of the homing ENase I-PpoI the importance of the putative catalytic motif PD⁸⁷X₃₀DXK¹²⁰ was not confirmed by crystallographic studies (22).

In the primary structure of M · EcoVIII we observed the presence of nine conserved motifs characteristic of m⁶N-adenine β-class MTases. In addition, we found that the deduced amino acid sequence of the putative target recognition domain located between conserved motifs VIII and X of M · EcoVIII was 74 and 82% identical to the sequences of the M · HindIII and M · LlaCI enzymes, respectively.

Expression of some R-M genes is controlled by regulatory elements. Upstream of the R · EcoVIII coding region we found a 12-bp perfect inverted repeat able to produce a stable stem-loop structure. Based on experiments in which we investigated the activity of a promoter using transcriptional fusions, it seemed reasonable to conclude that the observed secondary structure has little effect on ecoVIIIR gene expression. On the other hand, it is possible that this structure might serve as a transcriptional terminator whose presence prevents potentially dangerous readthrough into the ecoVIIIR gene.

The genes of the EcoVIII R-M system encode two enzymes. The first is R · EcoVIII (307 amino acids; M_r, 35,554), whose size is similar to the sizes of R · HindIII (300 amino acids; M_r, 34,951) (56) and R · LlaCI (324 amino acids; M_r, 38,327) (49). The second protein is M · EcoVIII (304 amino acids; M_r, 33,930), which is homologous to the isomethylomers M · HindIII (309 amino acids; M_r, 35,549) (56) and M · LlaCI (296 amino acids; M_r, 33,896) (49).

Both EcoVIII enzymes were purified to electrophoretic and catalytic homogeneity. In the case of M · EcoVIII this process was facilitated by using an overproducing clone. The relative molecular weights of purified R · EcoVIII and M · EcoVIII were calculated to be 36,700 and 36,000, respectively, and are in total agreement when they are compared to the M_rs of their isospecific counterparts in the HindIII and LlaCI R-M systems. Despite many attempts we were not able to overexpress the ecoVIIIR gene. Most probably this was due to the large number of arginine codons AGG and AGA, which can affect the translation efficiency of genes cloned in E. coli by causing frameshifts or inhibition of protein synthesis (23, 62, 64, 77). Interestingly, two AGG codons arranged in tandem are present in the N-terminal portion of the R · EcoVIII coding sequence (RR¹⁷). It was shown previously that such localization of AGG codons may have a severe effect on protein synthesis (16).

Most of the type II restriction ENases exist in solution as dimers. The exceptions to this rule are enzymes belonging to type IIS family (20, 32, 71). On the other hand, class II DNA MTases are functional as monomers. The EcoVIII enzymes fit this pattern; the ENase exists in solution in dimeric form, and the MTase is monomeric. Other aggregates were observed in the presence of elevated concentrations of a particular enzyme.

An interesting observation was made concerning the effect of divalent metal ions on M · EcoVIII methylation activity. We confirmed that Mg(II) ions are necessary for EcoVIII ENase activity, but in the case of M · EcoVIII these ions are strong inhibitors. This property is shared with a few other bacterial MTases, including M · BamHI, M · EcoRI, M · AluI, M · FokI, M · RsrI, M · TaqI, and E. coli Dam (25, 35, 38). The intracellular concentration of Mg(II) ions in E. coli cells was estimated to be in the range from 1 to 4 mM (1, 48). This means that inside bacterial cells the activity of M · EcoVIII is strongly inhibited [IC₅₀ of Mg(II), 2.0 mM], while the activity of the cognate restriction ENase is stimulated. This finding underlines the defensive character of type II R-M systems. For bacterial cells, high activity of the restriction ENase seems to be profitable. Clearly, such activity promotes DNA restriction over DNA methylation. It also explains experiments that have demonstrated the recombinogenic role of restriction ENases in vivo (15, 73, 78). Moreover, this specific behavior of some R-M enzymes in the presence of magnesium ions can effectively modulate the flow of genes among bacteria, which seems to be fundamental for the evolution of prokaryotes.

DNA MTases, in contrast to restriction ENases, form a homogeneous group of proteins having a common building plan with easily identifiable conserved amino acid sequence motifs. In the present study we were interested in whether similarities between the homologs analyzed at the amino acid sequence level can also be observed in the tertiary structure. The ability of antiserum prepared against M · EcoVIII or R · EcoVIII to cross-react with isospecific enzymes (M · LlaCI, M · HindIII, R · HindIII) was examined by using the immunoblot method. As expected, the results obtained were negative in the case of R · HindIII (the two proteins analyzed exhibited little homology), but in the case of MTases the results were quite interesting. We observed that despite the fact that all three MTases analyzed exhibited substantial levels of homology, they did not cross-react identically with polyclonal antibodies raised against M · EcoVIII. This means that our experiment revealed that there are substantial differences in the tertiary structures of the MTases analyzed, despite the homology noted at the amino acid sequence level. The data obtained indicate that M · EcoVIII and M · LlaCI may have similar architectures since antigenic closeness is generally correlated with three-dimensional similarity. On the other hand, the lack of cross-reactivity of M · HindIII with antibodies raised against M · EcoVIII suggests that there may be structural dissimilarities between these two enzymes.

In conclusion, our molecular analysis of the EcoVIII R-M system provides solid ground for future experiments aimed at investigating in detail the phenomenon of isospecificity among type II R-M enzymes.