HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

J. Biol. Chem., Vol. 277, Issue 2, 887-895, January 11, 2002

This Article Abstract Full Text (PDF) All Versions of this Article: 277/2/887    most recent M109100200v1 Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Soundararajan, M. Articles by Connolly, B. A. Search for Related Content PubMed PubMed Citation Articles by Soundararajan, M. Articles by Connolly, B. A. Social Bookmarking                      What's this?

DNA Binding and Recognition by the IIs Restriction Endonuclease MboII^*

Meera Soundararajan^§, Zhiyuh Chang^¶, Richard D. Morgan^¶, Pauline Heslop, and Bernard A. Connolly

From the  Department of Biochemistry and Molecular Genetics, The University of Newcastle, Newcastle upon Tyne, NE2 4HH, United Kingdom and ^¶ New England Biolabs, Beverly, Massachusetts 01915

Received for publication, September 20, 2001, and in revised form, October 16, 2001

	ABSTRACT

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

The type IIs restriction endonuclease MboII recognizes nonsymmetrical GAAGA sites, cutting 8 (top strand) and 7 (bottom strand) bases to the right. Gel retardation showed that MboII bound specifically to GAAGA sequences, producing two distinct complexes each containing one MboII and one DNA molecule. Interference analysis indicated that the initial species formed, named complex 1, comprised an interaction between the enzyme and the GAAGA target. Complex 2 involved interaction of the protein with both the GAAGA and the cutting sites. Only in the presence of divalent metal ions such as Ca²⁺ is the conversion of complex 1 to 2 rapid. Additionally, a very retarded complex was seen with Ca²⁺, possibly a (MboII)₂-(DNA)₂ complex. Plasmids containing a single GAAGA site were hydrolyzed slowly by MboII. Plasmids containing two sites were cut far more rapidly, suggesting that the enzyme requires two recognition sites in the same DNA molecule for efficient hydrolysis. MboII appears to have a mechanism similar to the best characterized type IIs enzyme, FokI. Both enzymes initially bind DNA as monomers, followed by dimerization to give an (enzyme)₂-(DNA)₂ complex. Dimerization is efficient only when the two target sites are located in the same DNA molecule and requires divalent metal ions.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

The best characterized type II restriction endonucleases, exemplified by EcoRI, EcoRV, BamHI, and PvuII, are homodimers that cut DNA within palindromic target sites, from 4 to 8 base pairs in size (1-3). Generally one protein subunit recognizes one-half of the palindromic sequence and, in an identical arrangement, the second subunit interacts with the other half, resulting in a symmetric protein-DNA complex. Each of the protein subunits contains a catalytic site, enabling both strands of the DNA to be cut, often in a highly concerted reaction. However, many type II restriction endonucleases do not correspond to these simple paradigms. One category, classified as type IIs systems (4), recognizes nonpalindromic DNA sequences, between 4 and 7 base pairs in length and cut up to 20 bases outside their target sites. A nonsymmetrical DNA target site does not allow for simple recognition using a protein with a homodimeric subunit arrangement, and indeed, most type IIs restriction enzymes appear to be monomeric (5-7). As a monomeric enzyme will contain only a single active site, there is no straightforward way to achieve the hydrolysis of both DNA strands. FokI, the best studied type IIs enzyme, recognizes GGATG [9/13] sites (the numbers in brackets refer to the distance to the cutting site on the top and bottom strands, respectively) and contains separate DNA recognition and cutting domains (8). A crystal structure shows FokI bound to DNA as a monomer with the recognition domain interacting with the GGATG bases; the cutting domain is tightly associated with the recognition domain and too far from the scissile phosphates to cause hydrolysis (9). A second structure, of the free enzyme, shows a dimer in which the dimerization interface is composed of elements of the cutting domains (10). Structural data, together with site-directed mutagenesis, and kinetic and binding studies (11, 12), have led to a model of how FokI cuts DNA (Fig. 1). FokI binds to GGATG sites as a monomer, but this initial complex is not competent for cleavage as the cutting domain is too far from the scissile phosphate. DNA binding releases the cutting domain, allowing dimerization and the assembly of a FokI dimer bound to two GGATG sites (Fig. 1). Cleavage occurs only at one of the GGATG sites, with the presence of two catalytic domains (one associated with the GGATG site being cut and the second "borrowed" from the other GGATG) leading to the hydrolysis of the individual DNA strands at this one site. As cutting does not destroy the GGATG sequence, the enzymes can recycle to cut the second GGATG site.

A key feature of the mechanism shown in Fig. 1 is the requirement of two DNA target sites for efficient cutting. Restriction enzymes with this property are surprisingly widespread, constituting several distinct classes. Type IIe endonucleases, such as NaeI (13) and EcoRII (14), require two sites but only cut at one. One of the DNA targets acts as an allosteric activator, allowing cutting at the second site. Type IIf enzymes, which include SfiI (15), SgrAI (16), and Cfr10I (17), also require two sites but cut both in a concerted manner. These enzymes are tetrameric and so contain four active sites, sufficient for the cutting of the four scissile phosphodiester bonds that occur at two double-stranded target sites. A simple method for assessing the requirement for two sites is to use plasmids containing either one or two copies of the target (16, 18, 19). Orthodox type II enzymes such as EcoRI, EcoRV, BamHI, and PvuII handle both plasmids equally well, as they do not need two sites. However, endonucleases in the sub-classes IIe and IIf show very low activity with plasmids containing only a single target site, confirming a requirement for two copies of their recognition sequence. The plasmid approach has recently been applied to a number of type IIs restriction endonucleases including FokI (19). Despite the enzymes being classified into the same IIs category, a wide variation in properties was observed. Some (BsaI, BsmBI, BsmI, and SapI) behaved like simple type II endonucleases, cutting one- and two-site plasmids with equal facility. Others (BsgI, BpmI, FokI, and BspMI) clearly required two sites, but even the behavior of this group could be further subdivided. With BsgI and BpmI the two target sites were cut in a sequential manner at about the same rate. FokI was slightly different; sequential cutting of the two sites was observed, but the first was cut more rapid than the second. These three enzymes are similar to type IIe systems, where, although two target sites are required, only one is cut in a single round of activity. BspMI showed concerted cleavage of all four DNA strands in the two target sites, reminiscent of type IIf enzymes. Thus, type IIs enzymes may be mechanistically diverse.

To further elucidate the type IIs systems, this article reports on binding and kinetic studies with MboII. This endonuclease has the recognition site GAAGA [8/7] (20) and has been cloned from Moraxella bovis (21). A purification scheme has been described, and the free enzyme was found to be monomeric (6). Previous studies have shown that the enzyme forms a specific complex (K_D = 0.34 nM) with its recognition site with a DNase I footprint from just before the GAAGA sequence to just after the cutting site. This complex contains a monomer of MboII bound to the GAAGA target (22).

	EXPERIMENTAL PROCEDURES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

The MboII restriction endonuclease was purified from Escherichia coli MS100. This strain was derived from ER2538 (F E. coli B fhuA2 [lon] [dcm] ompT gal sulA11 (mcrC-mrr) 114::IS10 R(mcr-73::miniTn10; Tet^s)2 endA1 R(zgb210::Tn10; Tet^s) (DE3)) by the addition of three plasmids: pLysS (chloramphenicol resistance), pSYX20MboIIM (a plasmid that constitutively expresses both MboII methyltransferases and confers kanamycin resistance), and pSYX22MboIIR (a plasmid that has the MboII restriction endonuclease under the control of an inducible T7 (23-25) promoter and confers ampicillin resistance). Constitutive expression of the methylases protects the cell from any premature endonuclease synthesis. Cells were plated out on Luria-Bertani agar containing 100 µg/ml ampicillin, 50 µg/ml kanamycin, and 30 µg/ml chloramphenicol. Material derived from a single colony was transferred to 2 × 10 ml of Luria-Bertani broth (containing the same antibiotics) and grown at 37 °C for 6 h. These cultures were diluted into 2 × 50 ml of Luria-Bertani broth/antibiotics and grown at 37 °C overnight. Approximately 15 ml of the overnight cultures were transferred to 6 × 500 ml of Luria-Bertani broth/antibiotics and grown until an absorbance at 600 nm of 0.6 was achieved. Expression of the endonuclease was induced by the addition of isopropyl-1-thio--D-galactopyranoside to a final concentration of 0.4 mM, followed by 4 h of further incubation at 37 °C. The cells (~1 g/500 ml of culture) were collected by centrifugation and stored frozen at 20 °C until needed.

Enzyme purification was carried out using a buffer comprising 20 mM KH₂PO₄, pH 7.4, containing 1 mM EDTA, 2 mM dithiothreitol, 0.1 mM benzamidine, 0.1 mM phenylmethylsulfonyl fluoride, and 10% (v/v) glycerol supplemented with either 0.2 M NaCl (low salt buffer) or 1 M NaCl (high salt buffer). All steps were carried out on ice or at cold room temperature. The purification was monitored using SDS-PAGE with Coomassie Blue staining and digestion of unmethylated phage  DNA. About 6 g of E. coli ER2538 cell paste was suspended in 50 ml of low salt buffer, and the cells were lysed by sonication. Insoluble material was removed by centrifugation, and the supernatant was applied to a 20 × 3 cm phosphocellulose column (phosphocellulose P11, Amersham Biosciences, Inc.) equilibrated with low salt buffer. The column was washed, at 1.0 ml/min, with 500 ml of low salt buffer and developed with a gradient composed of 250 ml each of low and high salt buffer. Fractions containing MboII (which eluted just before halfway through the gradient) were pooled. (NH₄)₂SO₄ was added to 50% (w/v), and any precipitated proteins were collected by centrifugation and discarded. The remaining supernatant was made 80% (w/v) in (NH₄)₂SO₄, and the precipitated MboII endonuclease was collected by centrifugation. The precipitate was dissolved in a minimal volume of low salt buffer, and final purification was achieved by fast protein liquid chromatography-based gel filtration using a 30 × 1-cm Superdex-75 column (Amersham Biosciences, Inc.) run in low salt buffer. Fractions that contained pure MboII were pooled, made up to 30% (v/v) glycerol, and stored at 20 °C. Typically 1-2 mg of protein was obtained. SDS-PAGE stained with either Coomassie Blue or silver showed only one band at the expected molecular mass of 49 kDa. The concentration of MboII endonuclease was determined by UV absorbance at 280 nm using an extinction coefficient of 4.047 × 10⁴ M¹ cm¹ (26).

The plasmid pUC18 is reported to contain 7 MboII sites at positions 291, 685, 1456, 1547, 2302, 2380, and 2489 (27) (there is no easily available plasmid with a smaller number of sites). During the course of this work, it was discovered that the pUC18 we used contained an additional MboII site at position 1304 (this is the result of a single base change; G at position 1308 to A, converting a GAAGG sequence to GAAGA; it is not clear whether the original pUC sequence contained an error or the plasmid used in this study has picked up a subsequent mutation). Two derivatives were prepared that contained a single MboII site at position 2302 (pMS1) and two MboII sites at positions 685 and 2302 (pMS2). Superfluous MboII sites were removed one at a time by back-to-back PCR-based site-directed mutagenesis (28). The PCR protocol amplified a fragment of pUC18 that contained the mutated MboII site flanked by two unique restriction endonuclease sites. In a subsequent step, the original plasmid was cut using the two restriction endonucleases, allowing the removal of the MboII site. The PCR fragment was inserted at this site, resulting in the addition of the mutated MboII site. E. coli XL-1 blue was used for these manipulations, and in each case the success of the operation was confirmed by DNA sequencing. A single base substitution was made within each GAAGA recognition sequence (changes were selected so as not to change amino acids in coding regions) resulting in GGAGA (position 291), GAGGA (position 685), GAAGG (position 1304), GAAGG (position 1456), GAAAA (position 1547), GAGGA (position 2380), and GAAAA (position 2489). The immediate environment of the MboII site at position 2302 (present in both pMS1 and pMS2) is GCCCCCGAAGAACGTTTTCCAATGATGAGCACT. With the site at position 685 (present in only pMS2), the corresponding sequence is GAAGCGGAAGAGCGCCCAATACGCAAACCGCC. Thus, although pMS2 contains two GAAGA MboII sites, the bases flanking and between the recognition and cutting sites vary. Therefore, a third plasmid was prepared by insertion of the oligodeoxynucleotide GCCCCGAAGAACGTTTTCCAATGATGATGAGCACT into the unique SmaI restriction endonuclease site (position 412) of pMS1. This plasmid contains two MboII sites, one at the original SmaI site and the second at position 2302 with identical sequences flanking both the GAAGA and cutting sites. DNA sequencing confirmed the success of the insertion and revealed two variants, pMS3a and pMS3b, which had the two MboII sites in a head-to-head and head-to-tail arrangement, respectively. The plasmids pMS1, -2, and -3 were isolated from 100 ml of E. coli XL-1 blue grown to an absorbance (600 nm) of about 0.6 in Luria-Bertani broth containing 100 µg/ml ampicillin. Purification used a Midiprep kit (Qiagen, Crawley, West Sussex, UK) following the manufacturer's instructions. The concentration of the purified plasmids was determined by UV absorbance at 260 nm (29).

Most of the studies with MboII were done using a double-stranded oligodeoxynucleotide prepared by hybridizing GCCATTGCGGTAACGTAACTTGCGTCTTCAAGTTGGAGCCTAGC (44-mer) with TTTTTGCTAGGCTCCAACTTGAAGACGCAAGTTACGTTACCGCAATGGC (49-mer) (both strands written in the 5'  3' direction; the MboII recognition site and two bases flanking the cutting site shown in bold). These oligodeoxynucleotides are complementary apart from a short 5' extension in the 49-mer, and duplexes (44/49-mer) were produced by mixing equimolar amounts of the single strands in 10 mM Hepes-NaOH, pH 7.5, containing 100 mM NaCl, 1 mM EDTA, heating to 95 °C, and cooling slowly to room temperature. Nondenaturing gel electrophoresis showed >95% duplex formation. A control had the GAAGA element in the 49-mer strand replaced by AGAAG, with a complementary change in the 44-mer strand. For dimethyl sulfate interference, a variant of the 44/49-mer having the T at position 33 of the 49-mer strand replaced by dG (with a corresponding change in the 44-mer) was used. Some experiments were also carried out with a 25/28-mer duplex produced from TTCCGGAAGACTGTTACGTTACCGC (25-mer) and CGAGCGGTAACGTAACAGTCTTCCGGAA (28-mer). For gel shift analysis, one strand in the duplex was radiolabeled with a 5'-[³²P]phosphate (polynucleotide kinase and -[³²P]ATP). The duplex oligodeoxynucleotide (20 pM) was incubated with increasing amounts of MboII (from 10 pM to 0.45 nM) in 20 µl of 10 mM Hepes-NaOH, pH 7.5, containing 100 mM NaCl, 2 mM dithiothreitol, 1 mM EDTA, and 0.1 mg/ml acetylated bovine serum albumin (MboII analysis buffer). Incubation was for 45 min at 37 °C (occasionally, as mentioned under "Results," the incubation time and temperatures were varied), and analysis used a 15% nondenaturing polyacrylamide gel made up and run in 89 mM Tris borate, pH 7.5, containing 1 mM EDTA. Gels (18 × 16 × 0.3 cm) were pre-run at 30 W¹ for 1 h with cooling. The samples (to which had been added 3 µl of 40% (w/v) sucrose) were loaded and the gels run for 90 min at 30 W with cooling. The progress of the gel was monitored using the dyes bromphenol blue and xylene cyanol FF in spare lanes. The counts in each band were detected using a phosphorimaging device (Fuji BAS-1500). Data were fitted using the Scientist software program (30) as outlined under "Results." For gel retardation analysis, using a mixture of the 44/49-mer and a 200-mer, the long oligodeoxynucleotide was prepared by PCR of pMS1. Primers were selected to amplify a fragment 200 bases in length containing a centrally located GAAGA sequence (this fragment was labeled at its 5' terminus with [³²P]phosphate using polynucleotide kinase and -[³²P]ATP). Incubation mixtures contained ~50 pM levels of both the 44/49-mer and the 200-mer together with 0.7 nM MboII. An 8% polyacrylamide gel was found to work best in this case. For gel retardation analysis in the presence of Ca²⁺, the 44/49-mer and MboII (amounts given under "Results") were mixed in 10 mM Hepes-NaOH, pH 7.5, containing 100 mM NaCl, 2 mM dithiothreitol, 1 mM EDTA, 6 mM CaCl₂ and 0.1 mg/ml acetylated bovine serum albumin. After incubation for the times and at the temperatures given under "Results," gel shift analysis was carried out using either a 12 or 15% nondenaturing polyacrylamide gel made up using 89 mM Tris borate, pH 7.5, and run in 89 mM Tris borate, pH 7.5, containing 5 mM CaCl₂ (the inclusion of CaCl₂ in the buffer to make up the gel gave very poor results with badly smeared bands).

Oligodeoxynucleotide hydrolysis was carried out in 200-µl volumes of MboII analysis buffer (lacking EDTA and containing 10 mM MgCl₂) at 37 °C. The concentration of oligodeoxynucleotide (labeled in both strands with 5'-[³²P]) was 0.24 nM. The different lengths of the two substrate strands and the two labeled product strands allowed the cutting of both strands in the duplex to be individually monitored. The reaction was initiated by adding MboII endonuclease to a final concentration of 4 nM. 10-µl aliquots were withdrawn at the times given under "Results" and mixed with 5 µl of "stop" solution (89 mM Tris borate, pH 8, containing 2.5 M urea, 0.1 M EDTA, 10% (w/v) sucrose and 125 µg/ml each of bromphenol blue and xylene cyanol FF) to quench the reaction. Analysis was by denaturing gel electrophoresis (16% polyacrylamide gel containing 8 M urea and run in 89 mM Tris borate, pH 8, containing 1 mM EDTA). Gels (18 × 16 × 0.3 cm) were run at 30 W for 90 min (progress monitored using the dyes), and band intensities were determined by phosphorimaging as described above. Data were fitted to smooth curves (see "Results") to obtain the half-lives of reactions.

Methylation interference, using dimethyl sulfate, was based on standard protocols (31). Only the variant duplex 44/49-mer was used, and two parallel experiments were carried out; one with only the 44-mer labeled and the other with only the 49-mer labeled (in both cases 5'-[³²P] labeling was used). The oligodeoxynucleotide duplex, in 40 µl of 10 mM Hepes-NaOH containing 100 mM NaCl and 1 mM EDTA, was incubated with 5 µl of a dimethyl sulfate solution (5% v/v in ethanol). Four such reactions were processed simultaneously. After 12-15 min at 37 °C, the reaction was stopped by adding 10 µl of 1 M 2-mercaptoethanol and 10 µl of a 1 mg/ml solution of tRNA (from bakers' yeast, Sigma). The DNA was purified by ethanol precipitation as described (31) and dissolved in 10 µl of MboII analysis buffer. To each of the four solutions obtained was added MboII restriction endonuclease to give a final concentration of 100 nM. After a 45-min incubation at 37 °C, the free and protein-associated DNA bands were separated by nondenaturing gel electrophoresis as described above. Bands were visualized by phosphorimaging and the appropriate gel slices cut out. Oligodeoxynucleotides were extracted from the gel (the four extracts corresponding to the same band were pooled at this stage), cleaved with piperidine, and analyzed by denaturing gel electrophoresis as described (31).

Plasmid hydrolysis was carried out in 300-µl volumes of MboII analysis buffer (lacking EDTA and containing 10 mM MgCl₂) at 37 °C. Reactions were carried out with either an excess of enzyme (80 nM plasmid, 2 µM enzyme) or an excess of DNA (90 nM plasmid, 8 nM enzyme). In all cases the reaction was initiated by enzyme addition, and 20-µl aliquots were removed (times given under "Results") and added to 10 µl of stop mix (0.1 M Tris-HCl, pH 8, containing 0.1 M EDTA, 10% (w/v) sucrose and 125 µg/ml bromphenol blue). The samples were analyzed using a 1% agarose gel, run in 89 mM Tris borate, pH 8.3, containing 1 mM EDTA and 0.5 µg/ml ethidium bromide. Gels were 14 × 11 × 0.5 cm and run at 120 W for 90-120 min. The bands were visualized using a Gel Documentation System 1000 (Bio-Rad ), and the intensity of each band was determined using Bio-Rad Molecular Analyst^TM software. The bands produced with pMS1, pMS2, and pMS3 were supercoiled (SC; starting substrate), open circle (OC; resulting from nicking of one strand), and full-length linear (FLL; resulting from cutting of both strands at one MboII site). All had a size of 2686 base pairs, but the supercoiled ran with an apparent size of 1700 base pairs and the open circle with an apparent size of >3000 base pairs (32). With pMS2 two product bands, resulting from cleavage at both MboII sites, of 1617 and 1069 base pairs were produced. With pMS3 these bands were 1842 and 844 base pairs. A correction factor was applied to the measured intensity to account for differential ethidium bromide binding. This factor was 0.7 for supercoiled DNA and 1 (i.e. no correction) for open circle and full-length linear forms (33). Intensity corresponding to products arising from cutting at two MboII sites were corrected for the fractional length of the product. Plasmid hydrolysis was also carried out with pMS1 (under conditions in which enzyme was in excess) in the presence of an oligodeoxynucleotide containing a single MboII site. Conditions were as described above, but the 44/49-mer was added to a final concentration of 400 nM.

	RESULTS

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Gel retardation analysis was carried out using MboII and oligodeoxynucleotides. Initially all experiments were carried out in the presence of EDTA and lacked divalent metals such as Mg²⁺ or Ca²⁺. As shown in Fig. 2A, for the 44/49-mer, two shifted bands were seen using a 15% gel. The same pattern was found with the 25/28-mer (not shown). The results shown in Fig. 2A were obtained after incubation at 37 °C for 45 min. The band with the lower mobility (labeled Complex 1) was produced essentially instantaneously (i.e. mixing MboII and oligodeoxynucleotide followed by loading onto the gel as quickly as possible). However, the production of slightly faster running band (labeled Complex 2) was both time- and temperature-dependent. At 37 °C, 45 min was required for the maximal appearance of complex 2, with much less of this species being produced at shorter times (Fig. 3A). Between 20 and 32 °C hardly any of complex 2 was formed, even after 1 h of incubation (not shown); this suggests that MboII binding to DNA initially results in complex 1, which then slowly converts to complex 2. To investigate the nature of complexes 1 and 2, interference analysis with dimethyl sulfate has been carried out using a slightly modified version of the 44/49-mer. This contained GA rather than TA at the cutting site on the 49-mer strand, to provide a dimethyl sulfate target at this location. The modified oligodeoxynucleotide behaved in an identical manner to its parent in gel shift and cutting assays (not shown). With the radioactive label present in the 49-mer strand of the duplex, Fig. 2B shows that both complex 1 and complex 2 exhibit very clear interference at the GAAGA recognition site (Footprinted region 1). Dimethyl sulfate reacts with the N7 of dG (major groove location) and N3 of dA (minor groove location) to give N-methylated bases. Most restriction endonucleases make direct contacts through the major groove with the bases that constitute their recognition site (1-3). The strong interference seen with the two dG bases in the GAAGA sequence probably arises directly from the disruption of such contacts. With the three dA bases, disruption of direct protein-DNA contacts, made via the minor groove, may account for the interference. A more likely explanation is that methylation of these dA bases alters their position within the double helix and therefore, indirectly, perturbs protein-DNA interaction. Fig. 2B also illustrates that complex 2 shows additional interference bands (Footprinted region 2) near the cutting site, which are absent from complex 1. Cutting on this strand takes place between G33 and A34. Methylation of G33 itself does not interfere with binding, and the data with A34 are inconclusive. However, bands corresponding to G36 and A39 are clearly absent in the complex 2 ladder, and therefore modification of bases 3' to the cutting site prevents formation of this complex. MboII has no selectivity for bases at this site; interference must arise from indirect effects, e.g. base modification changing the overall DNA conformation. Experiments with the 44-mer strand showed very weak interference patterns, probably because of the lack of dG and dA bases (the target for dimethyl sulfate) at critical positions.

The model for FokI (Fig. 1) shows the formation of (FokI)₂-(DNA)₂ complexes. To determine whether either complex 1 or 2 produced with MboII contained two DNA molecules, a gel shift was carried out with a mixture of the 44/49-mer and a 200-mer, each carrying a single GAAGA sequence. When MboII was incubated separately with either the 44/49-mer or the 200-mer, the shifted bands resulting from the protein-DNA complexes were well separated on an 8% gel (not shown). Mixing MboII with the 44/49-mer and the 200-mer simultaneously produced shifted bands that were simply the sum of the bands observed in the separate incubations, with no additional species being observed (not shown). This approach was developed in a study of R.SfiI (34), an enzyme that can bind two DNA molecules. If MboII was capable of binding two DNA molecules, then the species formed in the mixed incubation would comprise MboII-(44/49-mer)₂, MboII-(200-mer)₂, and MboII-(44/49-mer)(200-mer), the latter appearing as a new band not seen in the individual incubations and running between the complexes containing two DNA molecules of the same length (34). For this explanation we have simply assumed MboII remains monomeric. However, exactly the same arguments apply if the enzyme acts as a dimer. Therefore, we believe that the gel shift and interference data are best accommodated by the following scheme,

where complex 1 is the first species formed between the two macromolecules, with the enzyme interacting with the GAAGA target site. Complex 2 is a time- and temperature-dependent conformational variant, where the protein interacts with both the GAAGA and the cutting site. Complex 1 is probably equivalent to the initial species produced between FokI and DNA (Fig. 1). Complex 2 has similarities with the FokI-DNA species produced in the isomerization step (Fig. 1). Fitting the data shown in Fig. 2A to the above model gave the binding isotherms shown in Fig. 2C (the Scientist software (30) used to produce the fit does not rely on any simplifying assumptions e.g. [MboII] > [DNA]). The fit is reasonable and supports the model. However, the tight binding and distribution of the bound complexes between two species made accurate quantitation difficult. Fig. 2C gives a K_D1 (representing the equilibrium between the protein and the DNA) of 0.1 nM and a K_D2 (describing the protein-DNA isomerization step) of 1. No shifted bands were observed with control oligodeoxynucleotides in which the GAAGA target was replaced by AGAAG at enzyme levels up to 500 nM (not shown). As <1 nM amounts were sufficient to fully shift GAAGA containing sequences, we conclude that specific sequences are favored over nonspecific by a factor of at least 1000-fold. However, the inability to obtain a K_D value for the MboII interaction with nonspecific sites means an accurate measure of binding discrimination cannot be given.

View larger version (12K): [in this window] [in a new window]   Fig. 1.   Proposed mechanism for the type IIs restriction endonuclease FokI (figure adapted from Ref. 10). The enzyme, which recognizes GGATG (9/13) sequences, has a DNA binding domain (rectangle) and a DNA cleavage domain (oval). In the free enzyme the cleavage domain is sequestered by the DNA binding domain and is unavailable for DNA hydrolysis. Initially the enzyme binds as a monomer to its GGATG recognition site using the DNA binding domain. A subsequent isomerization step releases the cleavage domain, and in a key step, two FokI-DNA complexes associate. Dimerization, mediated through the cleavage domains, positions two active sites on one DNA duplex and results in the efficient cutting of both strands of this duplex. As the FokI recognition site is not destroyed in the hydrolysis step, the products can recycle resulting in the cleavage of all FokI target sites. A similar mechanism probably applies to MboII.

View larger version (14K): [in this window] [in a new window]   Fig. 2.   Interaction of the MboII restriction endonuclease with oligodeoxynucleotides. A, gel retardation analysis (no divalent metal ions are present). The 44/49-mer (0.02 nM) was incubated with the increasing amounts of MboII shown and the mixture analyzed by nondenaturing PAGE. Two retarded species, complex 1 and complex 2, are produced. B, methylation interference analysis with dimethyl sulfate. A derivative of the 44/49-mer (containing a T  G change at position 33 in the 49-mer strand; see "Experimental Procedures") was treated with dimethyl sulfate, and gel retardation analysis was carried out with MboII. The bands corresponding to free DNA, complex 1, and complex 2 were excised and treated with piperidine. The ladders obtained from these three species using a 5'-[32P]phosphate-labeled 49-mer reveal two footprints. Region 1 (centered on the GAAGA recognition site from bases 21-25) was found for both complexes 1 and 2. Region 2 (near the cutting site, which occurs between G33 and A34) was only seen with complex 2. C, analysis of the gel retardation data shown in panel A. Data were fitted to a model (see Scheme 1) that assumes an initial association of the two macromolecules to form MboII-DNA (complex 1) followed by an isomerization to give complex 2. Data were fitted using Scientist (30) to give a KD1 of 0.1 nM and a KD2 of 1. The solid line and circles represent complex 1; the hatched line and squares, complex 2. Note that each concentration of R.MboII has a data point for both complexes; however, in many cases the circles are obscured by the squares.

Gel retardation experiments have also been carried out in the presence of Ca²⁺, which often acts as a nonreactive analogue of the essential co-factor, Mg²⁺, and strengthens the interaction of specific sequences with restriction endonucleases e.g. with R.EcoRV (35, 36), R.MunI (37), R.PvuII (38), R.Cfr10I (39), and R.BamHI (40). Recently, Ca²⁺ has been found to be essential for the formation of (FokI)₂-(DNA)₂ complexes (12). Critically, with the 44/49-mer, the conversion of complex 1 to complex 2 was very rapid in the presence of Ca²⁺ (Fig. 3A). At 37 °C the appearance of complex 2 was complete within 30 s; without the metal ion, this isomerization step required 45 min (Fig. 3A). The addition of Ca²⁺ also allowed the production of complex 2 at temperatures between 20 and 30 °C; when Ca²⁺ was absent complex 2 was not formed at these temperatures (not shown). Use of a 12% polyacrylamide gel (rather than the usual 15%) revealed a very retarded species, in addition to complexes 1 and 2, (Fig. 3B) when Ca²⁺ was present. This very slow moving species was not seen in the absence of Ca²⁺ and may represent a (MboII)₂-(DNA)₂ complex. Although the nature of this species has yet to be fully elucidated, its low mobility is at least consistent with the high molecular weight of (MboII)₂-(DNA)₂. In our hands the inclusion of Ca²⁺ in the incubation and gel buffers resulted in smeared gels, which made quantitative analysis and K_D determination difficult. Nevertheless, analysis indicated a slight improvement in the binding leading to complex 1 (K_D1 values reduced 5-10-fold) when Ca²⁺ was present; K_D2 values remained unchanged (not shown).

View larger version (18K): [in this window] [in a new window]   Fig. 3.   Influence of Ca2+ on oligonucleotide binding by MboII. A, MboII (0.5 nM) and the 44/49-mer (0.1 nM) were mixed in the presence of buffer with or without 5 mM CaCl2 and analyzed by nondenaturing PAGE after incubating for the times (in minutes) shown above the lanes. B, gel retardation analysis in the presence of 5 mM Ca2+ analyzed using a 12% gel. MboII (100 nM) and the 44/49-mer (5 nM) were mixed in the presence of buffer containing 5 mM CaCl2 and analyzed by nondenaturing PAGE. In addition to complexes 1 and 2, formed in the absence of Ca2+, a new very retarded species is also seen.

When the 44/49-mer oligodeoxynucleotide was hydrolyzed using an excess of MboII, the results shown in Fig. 4A were obtained. Both strands of the duplex substrate were hydrolyzed to give the products of the expected size. If MboII was behaving as a type II restriction endonuclease requiring only a single site for activity, the conditions shown in Fig. 4A would correspond to a first-order single turnover reaction, as the enzyme is both in excess of the substrate and at a higher concentration than the K_D value. However, as explained under "Discussion," the requirement for two DNA target sites for efficient cleavage by MboII means that reactions with single-site oligodeoxynucleotides are probably not first order. Therefore, fitting to a single exponential to give a rate constant with units of time¹ is unlikely to be appropriate. To obtain a measure of the rate of the reaction shown in Fig. 4A, the data were fitted to a smooth curve allowing an estimation of half-life (Fig. 4B). As can be seen, under the conditions used ([oligodeoxynucleotide] = 0.24 nM, [MboII] = 4 nM), both strands were converted to products with a similar half-life of about 5 min. This half-life can be compared with values obtained for enzymes such as R.EcoRI (41, 42), R.EcoRV (43, 44), R.MunI (45), R.TaqI (46), and R.BamHI (47). Using the conditions of [enzyme] > [DNA], half-lives are generally < 10 s, with values of 1-2 s being typical. Clearly, using an assay consisting of hydrolysis of single-site oligodeoxynucleotides at [enzyme] > [DNA], the type IIs enzyme MboII is an inefficient catalyst when compared with conventional type II endonucleases.

View larger version (59K): [in this window] [in a new window]   Fig. 4.   Hydrolysis of the 44/49-mer oligodeoxynucleotide duplex with an excess of MboII restriction endonuclease. The 44/49-mer (0.24 nM) was incubated with MboII (4 nM), and aliquots were removed at the times shown. Both the 44- and 49-mer were labeled at their 5' ends with [32P]phosphate and, as a consequence, gave rise to labeled products 17 and 33 bases in length, respectively. As shown in panel A, both substrate strands and their associated product strands can be separated by denaturing PAGE. The intensity of each band was quantitated using a phosphorimaging device and used to determine the amount of product produced with time. Panel B shows a plot of the percentage of the 17-mer (squares and solid line) and 33-mer (circles and hashed line) products produced with time. The points are joined with a smooth curve (as explained under "Results") to give half-lives of about 5 min for the production of both products.

To further elucidate the reasons underlying the low reactivity of MboII, plasmids containing one and two GAAGA sites have been used. Using a supercoiled plasmid, pMS1, containing a single MboII site and an excess of enzyme, relatively slow hydrolysis was seen (Fig. 5A). The supercoiled plasmid was converted to a full-length linear product with very little open circle intermediate (which would have resulted from nicking in one strand). About 10 min was needed to convert all of the substrate to product. The half-life of the reaction was estimated at about 5 min (the time at which [substrate] ~ [product] ~ 50%) by simple inspection of Fig. 5A, a value very similar to that seen with the 44/49-mer oligodeoxynucleotide. This experiment was repeated (under identical conditions) using pMS1, which had been linearized (using the unique SacI site at position 402) to produce a nonsupercoiled substrate. In this case (data not shown) the two expected linear products were produced, but the half-life for the reaction was about 10 min. Therefore, a MboII site in a supercoiled plasmid is cut twice as fast as the same site in a relaxed plasmid. When R.MboII was tested with pMS2, a plasmid containing two MboII sites, much more rapid hydrolysis was observed (Fig. 5B). All of the circular substrate was converted into the two expected linear products in about 8 s, as opposed to the 10 min needed for the one-site plasmid. At the very shortest times some open circular product (resulting from nicking a single strand) and full-length linear product (resulting from cutting both strands at one of the two sites) were visible. The results shown in Fig. 5B were obtained with the two-site plasmid, pMS2, containing MboII sites with differing flanking sequences. Identical results were seen using pMS3, which possessed two MboII sites with identical flanking sequences (not shown). No difference in behavior between pMS3a and pMS3b, which have the two MboII sites in head-to-head and head-to-tail orientations, respectively, was noticed (not shown). A comparison of Fig. 5, A and B, clearly shows that MboII works far more efficiently when two target sites are present in the same plasmid. Although the two-site plasmids were cut too rapidly for accurate half-life determination, an estimate of <2 s is similar to the values found with type II endonucleases under single turnover conditions. Furthermore, when the one-site plasmid, pMS1, was cut by MboII in the presence of an oligonucleotide containing a GAAGA sequence ([MboII] = 2 µM; [pMS1] = 80 nM; [44/49-mer] = 400 nM), hydrolysis was speeded up. In this case the reaction was complete in about 4 min with a half-life of about 1 min (not shown), a 5-fold increase in the rate over that seen in the absence of added oligodeoxynucleotide. Further information was obtained using the plasmids under multiple turnover conditions, i.e. with the DNA in excess over the MboII. As shown in Fig. 5C the single-site plasmid, pMS1, was almost refractory to cleavage under these conditions, with hardly any reaction seen even after 240 min. In contrast the two-site plasmid, pMS2 (pMS3 behaved identically), was fully converted to products in about 120 s (Fig. 5D). These experiments confirm the requirement of two DNA sites for the efficient action of MboII. An analysis of the data obtained in Fig. 5D is given in Fig. 5E, which shows an accumulation of the full-length linear intermediate, resulting from cutting at one of the two sites, to levels of about 50%. Several experiments (not shown) have indicated amounts of this product between 50 and 60%. If two sites on a plasmid are cut independently in sequential steps and at the same rate, the maximum amount of full-length linear intermediate will be 40% (18). We believe that the levels of full-length intermediate observed with MboII, 50-60%, are best interpreted as a simple two-step sequential mechanism, with the first cut (on a supercoiled plasmid) proceeding twice as fast as the second (on a relaxed plasmid). Under these conditions the intermediate is expected to accumulate to levels of 50%. The preference of MboII for supercoiled DNA has also been seen with FokI (19).

View larger version (29K): [in this window] [in a new window]   Fig. 5.   Hydrolysis of plasmids by the MboII restriction endonuclease. For panels AD, the plasmid and MboII were incubated and aliquots withdrawn at the times shown. The aliquots were analyzed by agarose gel electrophoresis using ethidium bromide staining (see "Experimental Procedures"). SC, supercoiled plasmid (starting material); OC, open circle plasmid (resulting from nicking of one DNA strand); FLL, full-length linear plasmid (resulting from cutting of both DNA strands at a single MboII site); PI and PII, products resulting from cutting both DNA strands at two MboII sites (when these symbols appear above a panel they represent standards). Panels A and B have enzyme in excess of plasmid: A, [pMS1] (one MboII site) = 80 nM, [MboII] = 2 µM; B, [pMS2] (two MboII sites) = 80 nM, [MboII] = 2 µM. Panels C and D have plasmid in excess of enzyme: C, [pMS1] (one MboII site) = 90 nM, [MboII] = 8 nM; D, [pMS2] (two MboII sites) = 90 nM, [MboII] = 8 nM. Panels B and D show that separation of the SC plasmid and the product PI was not complete on this gel system. These two species could be resolved at other concentrations of agarose (we have not found conditions that resolve all of the bands; the system that separated SC and PI caused other bands to merge). A separate analysis (not shown) indicated that all of the SC plasmid had been removed after the first time point (2 s) for panel B, and thus the band marked PI/SC is solely due to PI. For panel D, the SC plasmid was not removed fully until 60 s, and thus the band marked PI/SC is a mixture of these two species until this time. A second gel (not shown) enabled quantitation of PI and SC, and this gel together with panel D was used for the analysis shown in panel E. In panel E the percentage of the SC (filled circles), FLL (empty circles), PI (empty squares), and PII (filled squares) present at various times are plotted. Points were joined with a smooth line.

	DISCUSSION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Previous studies have shown that the type IIs restriction endonuclease MboII is a monomer in solution (6). Furthermore, the protein contains a single PD(X)_n(E/D)XK (n is a variable number of amino acids, often between 10 and 40) motif (Pro²³⁷, Asp²³⁸, Asp²⁴³, Lys²⁴⁵) (21), which is characteristic of the active site of restriction endonucleases (1), suggesting that each monomer only contains a single DNA cutting site. Other type IIs enzymes, FokI (5, 12) and MmeI (7), have been demonstrated to be monomeric in solution, and crystal structures have clearly demonstrated that FokI contains only one active site per monomer (9, 10). This article addresses how MboII is able to cut both strands of duplex DNA containing GAAGA target sites, despite being a monomer possessing only a single active site.

Gel retardation in the absence of Ca²⁺ has revealed two MboII-DNA complexes. At present the domain structure of MboII is unknown. We have assumed that this enzyme is similar to FokI and contains, minimally, a DNA binding domain and a catalytic domain. Complex 1 is rapidly formed and arises from the initial encounter between the two macromolecules. This complex, which footprints to the GAAGA target site, almost certainly corresponds the structure of monomeric FokI bound to DNA (9). Complex 1 represents the first step in the catalytic cycle of type IIs restriction endonucleases (Fig. 1) and comprises an interaction between the DNA binding domain of the protein and the DNA target site. Structural data with FokI (9) have shown that complex 1 has the catalytic domain tightly associated with the DNA binding domain and unavailable for either protein-DNA or protein-protein interactions. Subsequent to the formation of the initial complex between MboII and DNA, a conformational change results in the production of a second species, complex 2 (Fig. 1). Footprinting shows that with complex 2 the DNA binding domain of the protein is associated with the GAAGA target site, and the catalytic domain is located near the scissile phosphate. Thus complex 2 corresponds to the FokI-DNA species presumed to arise from the isomerization step (Fig. 1) in which the catalytic domain is released from its interaction with the DNA binding domain. Further support for the identification of these two complexes comes from gel retardation analysis of MboII in the presence of two oligodeoxynucleotides of different sizes. Such experiments clearly show that complexes 1 and 2 contain only a single DNA molecule. Previous studies with MboII have used a 171-base pair fragment containing one GAAGA site (22). Gel shift analysis gave a single band that footprinted (using DNase I) to both the GAAGA and the cutting sites. This shifted band could actually represent two species (corresponding to our complexes 1 and 2) that had not been resolved because of the large size of the DNA fragment used; such a mixture would be expected to footprint at both the binding and cutting sites. The K_D of 0.34 nM determined in this earlier publication (22) agrees with the values we observed for K_D1, 0.17 and 0.1 nM for the 25/28- and 44/49-mers, respectively. Analysis of both the gel shift data and glycerol gradient sedimentation (both of which were carried out in the absence of Ca²⁺) also concluded that only monomeric MboII-DNA complexes were produced.

We believe that with MboII both complex 1 and complex 2 are on the kinetic pathway leading to the formation of product. However, in the absence of divalent metal ions, the conformational change leading to the formation of complex 2 from complex 1 is slower than the overall turnover rate. For instance at 37 °C and in the absence of divalent metals, very little complex 2 is visible after 5 and 15 min (Fig. 3A). Under identical (not shown) or similar (Fig. 4, A and B) conditions, save for the presence of 10 mM MgCl₂, considerable product formation is observed. Similarly none of complex 2 was produced at 20 and 30 °C when Ca²⁺ was omitted, but MboII shows activity in the presence of Mg²⁺ at these temperatures (not shown). With the exception of R.BfiI (48) all restriction endonucleases require a divalent metal ion, normally Mg²⁺, which is essential for the cutting of DNA. As the presence of Mg²⁺ results in hydrolysis, it cannot be used in experiments aimed at investigating protein-DNA complexes. However, Ca²⁺, which does not promote DNA hydrolysis, has been extensively used as a Mg²⁺ surrogate. With orthodox type II restriction endonucleases, Ca²⁺ is commonly observed to increase the affinity for cognate DNA sequences often by considerable factors of up to 10⁴ (35-40). Recently it has been shown that the production of dimeric FokI complexes requires the presence of a divalent metal ion. (FokI)₂-(DNA)₂ complexes (Fig. 1) stable enough to be detected by analytical ultracentrifugation and gel filtration chromatography could be produced only in the presence of Ca²⁺ (12). With MboII the conversion of complex 1 to complex 2 is greatly speeded by the addition of Ca²⁺ (Fig. 3A), and this conformational transition is now fast enough to lie on the reaction pathway. In addition, when Ca²⁺ was present there was some evidence of higher assemblies, possibly representing a (MboII)₂-(DNA)₂ species (Fig. 3B). This very retarded complex is currently under further investigation using footprinting methods. Thus, divalent metal ions are required at two stages in the reaction scheme for the type IIs restriction endonucleases shown in scheme 1, that is in the isomerization and dimerization steps. The divalent metal ion binding site for restriction endonucleases is composed of spatially conserved acidic amino acids, normally found in the PD(X)_n(E/D)XK motif (1, 2). With FokI (9, 10), and presumably with other type IIs restriction endonucleases, this motif is found in the catalytic domain. In the absence of divalent metals, type IIs enzymes can specifically bind to their DNA target site, but the complex is unable to proceed along the reaction pathway. Metal ion binding to the catalytic domain greatly speeds up the conformational change, which releases the catalytic domain for dimer formation and subsequent DNA hydrolysis. Thus divalent ions seem to play a slightly different role with orthodox and type IIs restriction endonucleases. With orthodox enzymes the main role is to strengthen the binding of specific sequences. With type IIs systems divalent cations increase the rates of conformational changes required to rearrange DNA binding and catalytic domains along the reaction pathway.

Further evidence that MboII requires two target sites for efficient hydrolysis comes from studies in which the conversion of DNA substrates to products was measured. The use of plasmids containing either one or two GAAGA sites clearly shows that turnover is much faster when two target sites are present. With a plasmid containing one GAAGA site and an excess of MboII over DNA, the hydrolysis could be conveniently measured over a span of about 30 min. Similar results were seen with oligodeoxynucleotides containing a single GAAGA sequence. In fact, the half-lives for the conversion of both plasmids and oligodeoxynucletides to products were alike, between 5 and 10 min. However, with the two-site plasmid and excess enzyme, the half-life was less than 2 s and could not be measured accurately with the manual quenching procedure used in this study. Given the relative concentration of MboII and DNA used in these experiments (0.24 nM oligodeoxynucleotide with 2.4 nM enzyme; 80 nM plasmid with 2 µM enzyme) and the K_D1 value of about 0.1 nM for the initial protein-DNA interaction, every GAAGA site should be bound by the endonuclease. We believe that the best explanation for the much slower hydrolysis of DNA containing one, as opposed to two, GAAGA sites is that MboII has a mechanism similar to that of FokI (10-12) (Fig. 1). The key step is the dimerization of two enzyme-DNA complexes, allowing two cutting domains to be assembled at one of the duplex DNA targets. Using an excess of MboII with single-site plasmids or oligodeoxynucleotides, the dimerization step will be a second order reaction in which two separate protein-DNA complexes must associate. This second order assembly is likely to be the slowest step in the pathway and the reason why DNA containing only a single GAAGA site is poorly cut. The reaction cannot, as is possible with orthodox type II restriction endonucleases (41-47), be fitted to a single exponential to give a rate constant with units of time¹. However, the simple comparison of the half-life seen with one- and two-site plasmids clearly indicates the strong preference of MboII for the latter. The rates of second order reactions depend on the concentration of reactants. Assuming full saturation of GAAGA sites, the MboII-DNA concentrations are 0.24 nM with oligodeoxynucleotides and 80 nM with plasmid. All things being equal, the plasmid should therefore be cut about 333-fold faster than the oligodeoxynucleotide, and yet approximately similar rates are observed. Presumably two large supercoiled MboII-plasmid complexes have difficulty in colliding in a productive manner. This may arise either because of severe electrostatic repulsion or the improbability of an encounter being correctly oriented for reaction. With an excess of enzyme, the rate at which pMS1 was hydrolyzed was increased 5-fold when an oligodeoxynucleotide containing a GAAGA site was added; this arises partially from an increase in MboII-DNA concentration, which increases the probability of dimer formation. Additionally a productive encounter between a MboII-plasmid complex and a small MboII-oligonucleotide complex is more probable than collision between two large MboII-plasmid complexes. With plasmids containing two GAAGA sites, the critical dimerization will be a first order reaction, arising from the assembly of two MboII molecules bound to the same DNA strand. Although the results found with an excess of enzyme can therefore be analyzed as a single exponential to give a k_st, the reaction was too fast to be analyzed accurately using manual quenching. However, it is clear that much of the advantage of having two MboII sites on a single DNA strand arises from the conversion of a slow bimolecular reaction into a much faster unimolecular one.

Additional mechanistic information was obtained by studying the hydrolysis of the plasmids with a limiting amount of enzyme in which any intermediates in the reaction accumulate in solution. Under such conditions the single-site plasmid was almost refractory to cleavage, emphasizing the need for two sites on the same molecule. With the two-site plasmid, the full-length linear intermediate accumulated to levels of between 50 and 60%. Clearly MboII does not cut the two sites in a concerted mechanism (in which little of the intermediate accumulates) like type IIf enzymes or the type IIs enzyme BspMI (19). The behavior of MboII with two-site plasmids is most reminiscent to BpmI and BsgI (19), both of which are type IIs enzymes cutting two sites in a sequential manner. The 50-60% levels of intermediate seen with R.MboII using two-site plasmids are best explained by a sequential mechanism in which the first site is cut at twice the rate of the second. More rapid cutting at the first site arises from a preference of the enzyme for supercoiled DNA rather than flanking sequence effects. We believe that our data with MboII can be explained by the mechanism originally proposed for FokI (10-12) (Fig. 1). However, recent studies with FokI using two-site plasmids have shown that the full-length linear intermediate accumulated to levels of 80% (19). It was suggested that supercoiled DNA might be preferred over relaxed; however, a 10-fold faster rate is necessary for the intermediate to accumulate at 80% levels. The steady state experiments with FokI used a 10-fold excess of plasmid, which means that the chance of two FokI monomers occupying both sites on a single plasmid is 1 in 400. It was suggested that this would provide little catalytic advantage. Therefore, an alternative mechanism in which a free second FokI monomer bound to the FokI-plasmid complex with the dimer being stabilized by capture of the second target site on the plasmid was suggested (19). This explanation was also proposed and rejected by the Aggarwal group (11). With MboII, just over an 11-fold excess of plasmid has been used, and so the same consideration raised with FokI applies. We still prefer the original mechanism (shown in Fig. 1). First DNA binding provides a simple mechanism for release of the protein catalytic domain such that it becomes available for dimerization. If a free protein can interact with the bound protein, one assumes that the catalytic domain of the free protein is available for dimer formation. This being the case it is hard to see why the free protein itself does not dimerize (5, 6, 12). Second, at an excess of enzyme, where all of the target sites will have a high occupancy, the probability of an (enzyme)₂-plasmid complex capturing the second strand, which will need to be free of enzyme, seems remote.

In conclusion, this article demonstrates that the type IIs restriction endonuclease MboII forms a tight and specific initial complex comprising a protein monomer and a GAAGA target site. However, subsequent reaction depends on the assembly of a dimeric protein complex bound to two DNA strands, as demonstrated for FokI and shown in Fig. 1. This process is a much more efficient unimolecular reaction when both DNA sites are on the same molecule, as opposed to the bimolecular reaction in which the sites are separated. However, the full mechanistic details of type IIs enzymes remain largely unexplored. Questions still to be answered include the following. Why are the type IIs class so diverse (19), and how does this relate to the subunit structure of the free protein? When two sites are required exactly how much advantage does having the targets on the same DNA molecule provide? Is all of the advantage simply conferred by conversion of a bimolecular to a unimolecular reaction? What is the exact order of assembly of the key (protein)₂-(DNA)₂ complex (at present the two possibilities mentioned above have not been rigorously distinguished)? What role does the essential divalent metal ion play in the production of the complexes shown in Fig. 1? What are the rate constants that define the steps shown in Fig. 1? Investigations are currently underway to try to elucidate some of these points.

	ACKNOWLEDGEMENTS

We thank Elisabeth Raleigh (New England Biolabs) for information about the properties of the MboII overproducing strain and ideas concerning the purification of R.MboII. Prof. Steve Halford (Bristol) is thanked for supplying preprints and helpful discussion.

	FOOTNOTES

^* The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

^§ Recipient of an Overseas Research Students Ph.D. scholarship.

Recipient of grants from the United Kingdom Biotechnology and Biological Sciences Research Council, Medical Research Council, and Wellcome Trust. To whom correspondence should be addressed. Tel.: 44-191-222-7371; Fax: 44-191-222-7424; E-mail: b.a.connolly@ncl.ac. uk.

Published, JBC Papers in Press, October 17, 2001, DOI 10.1074/jbc.M109100200

	ABBREVIATIONS

The abbreviation used is: W, watt(s).

	REFERENCES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

CiteULike

Complore