Structure and function of type II restriction endonucleases

The power of two: protein dimerization in biology

The self-association of proteins to form dimers and higher-order oligomers is a very common phenomenon. Recent structural and biophysical studies show that protein dimerization or oligomerization is a key factor in the regulation of proteins such as enzymes, ion channels, receptors and transcription factors. In addition, self-association can help to minimize genome size, while maintaining the advantages of modular complex formation. Oligomerization, however, can also have deleterious consequences when nonnative oligomers associated with pathogenic states are generated. Specific protein dimerization is integral to biological function, structure and control, and must be under substantial selection pressure to be maintained with such frequency throughout biology.

Isolation and characterization of new homing endonuclease specificities at individual target site positions

Homing endonucleases are highly specific DNA endonucleases, encoded within mobile introns or inteins, that induce targeted recombination, double-strand repair and gene conversion of their cognate target sites. Due to their biological function and high level of target specificity, these enzymes are under intense investigation as tools for gene targeting. These studies require that naturally occurring enzymes be redesigned to recognize novel target sites. Here, we report studies in which the homodimeric LAGLIDADG homing endonuclease I-CreI is altered at individual side-chains corresponding to contact points to distinct base-pairs in its target site. The resulting enzyme constructs drive specific elimination of selected DNA targets in vivo and display shifted specificities of DNA binding and cleavage in vitro. Crystal structures of two of these constructs demonstrate that substitution of individual side-chain/DNA contact patterns can occur with almost no structural deformation or rearrangement of the surrounding complex, facilitating an isolated, modular redesign strategy for homing endonuclease activity and specificity.

Substrate recognition and catalysis by the Holliday junction resolving enzyme Hje

Two archaeal Holliday junction resolving enzymes, Holliday junction cleavage (Hjc) and Holliday junction endonuclease (Hje), have been characterized. Both are members of a nuclease superfamily that includes the type II restriction enzymes, although their DNA cleaving activity is highly specific for four-way junction structure and not nucleic acid sequence. Despite 28% sequence identity, Hje and Hjc cleave junctions with distinct cutting patterns--they cut different strands of a four-way junction, at different distances from the junction centre. We report the high-resolution crystal structure of Hje from Sulfolobus solfataricus. The structure provides a basis to explain the differences in substrate specificity of Hje and Hjc, which result from changes in dimer organization, and suggests a viral origin for the Hje gene. Structural and biochemical data support the modelling of an Hje:DNA junction complex, highlighting a flexible loop that interacts intimately with the junction centre. A highly conserved serine residue on this loop is shown to be essential for the enzyme's activity, suggesting a novel variation of the nuclease active site. The loop may act as a conformational switch, ensuring that the active site is completed only on binding a four-way junction, thus explaining the exquisite specificity of these enzymes.

DNA Binding Provides a Molecular Strap Activating the Adenovirus Proteinase

Human adenovirus proteinase (AVP) requires two cofactors for maximal activity: pVIc, a peptide derived from the C terminus of adenovirus precursor protein pVI, and the viral DNA. Synchrotron protein footprinting was used to map the solvent accessible cofactor binding sites and to identify conformational changes associated with the binding of cofactors to AVP. The binding of pVIc alone or pVIc and DNA together to AVP triggered significant conformational changes adjacent to the active site cleft sandwiched between the two AVP subdomains. In addition, upon binding of DNA to AVP, it was observed that specific residues on each of the two major subdomains were significantly protected from hydroxyl radicals. Based on the locations of these protected side-chain residues and conserved aromatic and positively charged residues within AVP, a three-dimensional model of DNA binding was constructed. The model indicated that DNA binding can alter the relative orientation of the two AVP domains leading to the partial activation of AVP by DNA. In addition, both pVIc and DNA may independently alter the active site conformation as well as drive it cooperatively to fully activate AVP.

Thermodynamic, kinetic and structural basis for recognition and repair of abasic sites in DNA by apurinic/apyrimidinic endonuclease from human placenta

X-ray analysis of enzyme-DNA interactions is very informative in revealing molecular contacts, but provides neither quantitative estimates of the relative importance of these contacts nor information on the relative contributions of specific and nonspecific interactions to the total affinity of enzymes for specific DNA. A stepwise increase in the ligand complexity approach is used to estimate the relative contributions of virtually every nucleotide unit of synthetic DNA containing abasic sites to its affinity for apurinic/apyrimidinic endonuclease (APE1) from human placenta. It was found that APE1 interacts with 9-10 nt units or base pairs of single-stranded and double-stranded ribooligonucleotides and deoxyribooligonucleotides of different lengths and sequences, mainly through weak additive contacts with internucleotide phosphate groups. Such nonspecific interactions of APE1 with nearly every nucleotide within its DNA-binding cleft provides up to seven orders of magnitude (DeltaG degrees approximately -8.7 to -9.0 kcal/mol) of the enzyme affinity for any DNA substrate. In contrast, interactions with the abasic site together with other specific APE1-DNA interactions provide only one order of magnitude (DeltaG degrees approximately -1.1 to -1.5 kcal/mol) of the total affinity of APE1 for specific DNA. We conclude that the enzyme's specificity for abasic sites in DNA is mostly due to a great increase (six to seven orders of magnitude) in the reaction rate with specific DNA, with formation of the Michaelis complex contributing to the substrate preference only marginally.

An Asymmetric Complex of Restriction Endonuclease MspI on Its Palindromic DNA Recognition Site

Most well-known restriction endonucleases recognize palindromic DNA sequences and are classified as Type IIP. Due to the recognition and cleavage symmetry, Type IIP enzymes are usually found to act as homodimers in forming 2-fold symmetric enzyme-DNA complexes. Here we report an asymmetric complex of the Type IIP restriction enzyme MspI in complex with its cognate recognition sequence. Unlike any other Type IIP enzyme reported to date, an MspI monomer and not a dimer binds to a palindromic DNA sequence. The enzyme makes specific contacts with all 4 base pairs in the recognition sequence, by six direct and five water-mediated hydrogen bonds and numerous van der Waal contacts. This MspI-DNA structure represents the first example of asymmetric recognition of a palindromic DNA sequence by two different structural motifs in one polypeptide. A few possible pathways are discussed for MspI to cut both strands of DNA, either as a monomer or dimer.

DNA communications by Type III restriction endonucleases--confirmation of 1D translocation over 3D looping

DNA cleavage by Type III restriction enzymes is governed strictly by the relative arrangement of recognition sites on a DNA substrate--endonuclease activity is usually only triggered by sequences in head-to-head orientation. Tens to thousands of base pairs can separate these sites. Long distance communication over such distances could occur by either one-dimensional (1D) DNA translocation or 3D DNA looping. To distinguish between these alternatives, we analysed the activity of EcoPI and EcoP15I on DNA catenanes in which the recognition sites were either on the same or separate rings. While substrates with a pair of sites located on the same ring were cleaved efficiently, catenanes with sites on separate rings were not cleaved. These results exclude a simple 3D DNA-looping activity. To characterize the interactions further, EcoPI was incubated with plasmids carrying two recognition sites interspersed with two 21res sites for site-specific recombination by Tn21 resolvase; inhibition of recombination would indicate the formation of stable DNA loops. No inhibition was observed, even under conditions where EcoPI translocation could also occur.

Bacteriophage T4 endonuclease II: concerted single-strand nicks yield double-strand cleavage

In vivo, endonuclease II (EndoII) of coliphage T4 cleaves sites with conserved sequence elements (CSEs) to both the left and the right of the cleaved bonds, 16 bp altogether with some variability tolerated. In vitro, however, single-strand nicks in the lower strand predominate at sites containing only the left-side CSE that determines the precise position of lower strand nicks. Upper strand nick positions vary both in vivo and in vitro. A 24 bp substrate was nicked with the same precision as in longer substrates, showing that the conserved sequence suffices for precise nicking by EndoII. Using DNA ligase in vitro, we found that EndoII nicked both strands simultaneously at an in vivo-favoured site but not at an in vitro-favoured site. This indicates that the right-side CSE at in vivo-favoured sites is important for simultaneous nicking of both strands, generating double-strand cleavage. Separate analysis of the two strands following in vitro digestion at two in vitro-favoured sites showed that EndoII nicked the lower strand about 1.5-fold faster than the upper strand. In addition, the upper and lower strands were nicked independently of each other, seldom resulting in double-strand cleavage. Thus, cleavage by EndoII is the fortuitous outcome of two separate nicking events.

Enzyme-mediated DNA looping

Most reactions on DNA are carried out by multimeric protein complexes that interact with two or more sites in the DNA and thus loop out the DNA between the sites. The enzymes that catalyze these reactions usually have no activity until they interact with both sites. This review examines the mechanisms for the assembly of protein complexes spanning two DNA sites and the resultant triggering of enzyme activity. There are two main routes for bringing together distant DNA sites in an enzyme complex: either the proteins bind concurrently to both sites and capture the intervening DNA in a loop, or they translocate the DNA between one site and another into an expanding loop, by an energy-dependent translocation mechanism. Both capture and translocation mechanisms are discussed here, with reference to the various types of restriction endonuclease that interact with two recognition sites before cleaving DNA.

Evolution of genomic islands by deletion and tandem accretion by site-specific recombination: ICESt1-related elements from Streptococcus thermophilus

The 34 734-bp integrative and potentially conjugative element (putative ICE) ICESt1 has been previously found to be site-specifically integrated in the 3' end of the fda locus of Streptococcus thermophilus CNRZ368. Four types of genomic islands related to ICESt1 are integrated in the same position in seven other strains of S. thermophilus. One of these elements, ICESt3, harbours conjugation and recombination modules closely related to those of ICESt1 and excises by site-specific recombination. Two other types of elements, CIME19258 and CIME302, are flanked by site-specific attachment sites closely related to attL and attR of ICESt1 and ICESt3, whereas Delta CIME308 only possesses a putative attR site; none of these three elements carry complete conjugation and recombination modules. ICESt1 contains a functional internal recombination site, attL', that is almost identical to attL of CIME19258. The recombination between attL' and attR of ICESt1 leads to the excision of the expected circular molecule (putative ICE); a cis-mobilizable element (CIME) flanked by an attL site and an attB' site remains integrated into the 3' end of fda. Furthermore, sequences that could be truncated att sites were found within ICESt1, ICESt3 and CIME302. All together, these data suggest that these genomic islands evolved by deletion and tandem accretion of ICEs and CIMEs resulting from site-specific recombination. A model for this evolution is proposed and its application to other genomic islands is discussed.

The isolation of strand-specific nicking endonucleases from a randomized SapI expression library

The Type IIS restriction endonuclease SapI recognizes the DNA sequence 5'-GCTCTTC-3' (top strand by convention) and cleaves downstream (N1/N4) indicating top- and bottom-strand spacing, respectively. The asymmetric nature of DNA recognition presented the possibility that one, if not two, nicking variants might be created from SapI. To explore this possibility, two parallel selection procedures were designed to isolate either top-strand nicking or bottom-strand nicking variants from a randomly mutated SapI expression library. These procedures take advantage of a SapI substrate site designed into the expression plasmid, which allows for in vitro selection of plasmid clones possessing a site-specific and strand-specific nick. A procedure designed to isolate bottom-strand nicking enzymes yielded Nb.SapI-1 containing a critical R420I substitution near the end of the protein. The top-strand procedure yielded several SapI variants with a distinct preference for top-strand cleavage. Mutations present within the selected clones were segregated to confirm a top-strand nicking phenotype for single variants Q240R, E250K, G271R or K273R. The nature of the amino acid substitutions found in the selected variants provides evidence that SapI may possess two active sites per monomer. This work presents a framework for establishing the mechanism of SapI DNA cleavage.

One recognition sequence, seven restriction enzymes, five reaction mechanisms

The diversity of reaction mechanisms employed by Type II restriction enzymes was investigated by analysing the reactions of seven endonucleases at the same DNA sequence. NarI, KasI, Mly113I, SfoI, EgeI, EheI and BbeI cleave DNA at several different positions in the sequence 5'-GGCGCC-3'. Their reactions on plasmids with one or two copies of this sequence revealed five distinct mechanisms. These differ in terms of the number of sites the enzyme binds, and the number of phosphodiester bonds cleaved per turnover. NarI binds two sites, but cleaves only one bond per DNA-binding event. KasI also cuts only one bond per turnover but acts at individual sites, preferring intact to nicked sites. Mly113I cuts both strands of its recognition sites, but shows full activity only when bound to two sites, which are then cleaved concertedly. SfoI, EgeI and EheI cut both strands at individual sites, in the manner historically considered as normal for Type II enzymes. Finally, BbeI displays an absolute requirement for two sites in close physical proximity, which are cleaved concertedly. The range of reaction mechanisms for restriction enzymes is thus larger than commonly imagined, as is the number of enzymes needing two recognition sites.

Crystallization and preliminary crystallographic studies of a bifunctional restriction endonuclease Eco57I

Restriction endonuclease Eco57I from Escherichia coli recognizes asymmetric DNA sequence 5'-CTGAAG and has both restriction (DNA cleavage a short distance away from the recognition site) and modification (methylation) activities residing in a single polypeptide chain. Single crystals of wild-type Eco57I ternary complexes with double-stranded DNA and sinefungin, a stimulator of endonuclease activity, were obtained by the vapor diffusion technique and characterized crystallographically for different variants of the DNA component. The best data for the complex with 25-mer DNA were collected to 4.2-A resolution at 100 K using synchrotron radiation. The crystals are orthorhombic, space group P2(1)2(1)2, with a=164.3, b=293.0, c=71.1 A, and contain two to four copies of the protein in the asymmetric unit.

Bme585 I [5′-CCCGC(4/6)-3′], a new isoschizomer of restriction endonuclease Fau I, isolated from a strain of Bacillus mesentericus

Bme585 I is a new member of the restriction endonuclease type IIS family. It was partially purified from the heterothrophic, mesophilic bacterial strain Bacillus mesentericus 585 by ammonium sulphate precipitation and phosphocellulose column chromatography. Bme585 I is a monomeric protein with a molecular mass of 62 kD. The enzyme is active over a broad pH range from 7.0 to 8.8, has a temperature optimum of 37 degrees C and tolerance of NaCl in reaction buffer from 0 to 400 mM. Bme585 I recognizes the asymmetric sequence 5'-CCCGC(4/6)-3' and is therefore an isoschizomer of restriction endonuclease Fau I.

Crystal structure of BstYI at 1.85A resolution: a thermophilic restriction endonuclease with overlapping specificities to BamHI and BglII

We report here the structure of BstYI, an "intermediate" type II restriction endonuclease with overlapping sequence specificities to BamHI and BglII. BstYI, a thermophilic endonuclease, recognizes and cleaves the degenerate hexanucleotide sequence 5'-RGATCY-3' (where R=A or G and Y=C or T), cleaving DNA after the 5'-R on each strand to produce four-base (5') staggered ends. The crystal structure of free BstYI was solved at 1.85A resolution by multi-wavelength anomalous dispersion (MAD) phasing. Comparison with BamHI and BglII reveals a strong structural consensus between all three enzymes mapping to the alpha/beta core domain and residues involved in catalysis. Unexpectedly, BstYI also contains an additional "arm" substructure outside of the core protein, which enables the enzyme to adopt a more compact, intertwined dimer structure compared with BamHI and BglII. This arm substructure may underlie the thermostability of BstYI. We identify putative DNA recognition residues and speculate as to how this enzyme achieves a "relaxed" DNA specificity.

Molecular requirements for duplex recognition and cleavage by eukaryotic RNase III: discovery of an RNA-dependent DNA cleavage activity of yeast Rnt1p

Members of the double-stranded RNA (dsRNA) specific RNase III family are known to use a conserved dsRNA-binding domain (dsRBD) to distinguish RNA A-form helices from DNA B-form ones, however, the basis of this selectivity and its effect on cleavage specificity remain unknown. Here, we directly examine the molecular requirements for dsRNA recognition and cleavage by the budding yeast RNase III (Rnt1p), and compare it to both bacterial RNase III and fission yeast RNase III (Pac1). We synthesized substrates with either chemically modified nucleotides near the cleavage sites, or with different DNA/RNA combinations, and investigated their binding and cleavage by Rnt1p. Substitution for the ribonucleotide vicinal to the scissile phosphodiester linkage with 2'-deoxy-2'-fluoro-beta-d-ribose (2' F-RNA), a deoxyribonucleotide, or a 2'-O-methylribonucleotide permitted cleavage by Rnt1p, while the introduction of a 2', 5'-phosphodiester linkage permitted binding, but not cleavage. This indicates that the position of the phosphodiester link with respect to the nuclease domain, and not the 2'-OH group, is critical for cleavage by Rnt1p. Surprisingly, Rnt1p bound to a DNA helix capped with an NGNN tetraribonucleotide loop indicating that the binding of at least one member of the RNase III family is not restricted to RNA. The results also suggest that the dsRBD may accommodate B-form DNA duplexes. Interestingly, Rnt1p, but not Pac1 nor bacterial RNase III, cleaved the DNA strand of a DNA/RNA hybrid, indicating that A-form RNA helix is not essential for cleavage by Rnt1p. In contrast, RNA/DNA hybrids bound to, but were not cleaved by Rnt1p, underscoring the critical role for the nucleotide located at 3' end of the tetraloop and suggesting an asymmetrical mode of substrate recognition. In cell extracts, the native enzyme effectively cleaved the DNA/RNA hybrid, indicating much broader Rnt1p substrate specificity than previously thought. The discovery of this novel RNA-dependent deoxyribonuclease activity has potential implications in devising new antiviral strategies that target actively transcribed DNA.

Footprint of the retrotransposon R2Bm protein on its target site before and after cleavage

R2 elements are non-long terminal repeat (non-LTR) retrotransposons that specifically integrate into the 28 S rRNA genes of their host. These elements encode a single open reading frame with a genome-specific endonuclease and a reverse transcriptase that uses the cleaved chromosomal target site to prime reverse transcription. Cleavage of the DNA strand that is used to prime reverse transcription is an efficient process that occurs in the presence or absence of RNA. Cleavage of the second DNA strand is much less efficient and requires RNA. Reverse transcription occurs before second strand cleavage and only if the RNA bound to the protein contains the 3' untranslated region of the R2 element. Thus a complex series of protein interactions with the DNA and conformational changes in the protein are likely to occur during this retrotransposition reaction. Here, we conduct electrophoretic mobility-shift assays and DNase I footprint studies on the binding of the R2 protein to the DNA target in the presence and absence of RNA both before and after first strand cleavage. While the total expanse of the protein footprint on the DNA eventually covers five helical turns, before cleavage the footprint only extends from 17 bp to 40 bp upstream of the cleavage site. This footprint is the same in the presence and absence of RNA. We hypothesize that the active site of the endonuclease domain is analogous to type IIS restriction enzymes in that it is located on a flexible domain that is not tightly bound to the cleavage site. After first strand cleavage the protein footprint extends beyond the cleavage site. We suggest that this increased protection after cleavage is the RT domain that is positioned over the free DNA end to begin reverse transcription on the nicked DNA substrate.

RISC is a 5' phosphomonoester-producing RNA endonuclease

Gene silencing in the process of RNA interference is mediated by a ribonucleoprotein complex referred to as RNA-induced silencing complex (RISC). Here we describe the molecular mechanism of target RNA cleavage using affinity-purified minimal RISC from human cells. Cleavage proceeds via hydrolysis and the release of a 3'-hydroxyl and a 5'-phosphate terminus. Substitution of the 2'-hydroxyl group at the cleavage site by 2'-deoxy had no significant effect, suggesting that product release and/or a conformational transition rather than a chemical step is rate-limiting. Substitution by 2'-O-methyl at the cleavage site substantially reduced cleavage, which is presumably due to steric interference. Mutational analysis of the target RNA revealed that mismatches across from the 5' or the 3' end of the siRNA had little effect and that substrate RNAs as short as 15 nucleotides were cleaved by RISC.

Comparison of the cleavage profiles of oligonucleotide duplexes with or without phosphorothioate linkages by using a chemical nuclease probe

A manganese porphyrin complex, Mn-TMPyP, associated with KHSO(5) is a chemical nuclease able to selectively recognize the minor groove of three consecutive AT base pairs of DNA and to mediate very precise cleavage chemistry at that particular site. This specific recognition and cleavage were used to probe the accessibility of the minor groove of DNA duplexes composed of one phosphodiester strand and one phosphorothioate strand. The cleavage of 5'-GCAAAAGC/5'-GCTTTTGC duplexes by Mn-TMPyP/KHSO(5) was monitored by HPLC coupled to electrospray mass analysis. Each single strand was synthesized with all-phosphate, all- Rp-phosphorothioate and all- Sp-phosphorothioate internucleotide bonds. We found that the manganese porphyrin was able to recognize its favorite (AT)(3)-box binding site within the heteroduplexes, as in the case of natural DNA. Molecular modeling studies on the interactions of the reactive porphyrin manganese-oxo species with both types of duplexes confirmed the experimental data.

Generation of the BfiI restriction endonuclease from the fusion of a DNA recognition domain to a non-specific nuclease from the phospholipase D superfamily

The BfiI endonuclease cleaves DNA at fixed positions downstream of an asymmetric sequence. Unlike other restriction enzymes, it functions without metal ions. The N-terminal half of BfiI is similar to Nuc, an EDTA-resistant nuclease from Salmonella typhimurium that belongs to the phosphoplipase D superfamily. Nuc is a dimer with one active site at its subunit interface, as is BfiI, but it cuts DNA non-specifically. BfiI was cleaved by thermolysin into an N-terminal domain, which forms a dimer with non-specific nuclease activity, and a C-terminal domain, which lacks catalytic activity but binds specifically to the recognition sequence as a monomer. On denaturation with guanidinium, BfiI underwent two unfolding transitions: one at a relatively low concentration of guanidinium, to a dimeric non-specific nuclease; a second at a higher concentration, to an inactive monomer. The isolated C-terminal domain unfolded at the first (relatively low) concentration, the isolated N-terminal at the second. Hence, BfiI consists of two physically separate domains, with catalytic and dimerisation functions in the N terminus and DNA recognition functions in the C terminus. It is the first example of a restriction enzyme generated by the evolutionary fusion of a DNA recognition domain to a phosphodiesterase from the phospholipase D superfamily. BfiI may consist of three structural units: a stable central core with the active site, made from two copies of the N-terminal domain, flanked by relatively unstable C-terminal domains, that each bind a copy of the recognition sequence.

Tetra-amino-acid tandem repeats are involved in HsdS complementation in type IC restriction-modification systems

All known type I restriction and modification (R-M) systems of Escherichia coli and Salmonella enterica belong to one of four discrete families: type IA, IB, IC or ID. The classification of type I systems from a wide range of other genera is mainly based on complementation and molecular evidence derived from the comparison of the amino acid similarity of the corresponding subunits. This affiliation was seldom based on the strictest requirement for membership of a family, which depends on relatedness as demonstrated by complementation tests. This paper presents data indicating that the type I NgoAV R-M system from Neisseria gonorrhoeae, despite the very high identity of HsdM and HsdR subunits with members of the type IC family, does not show complementation with E. coli type IC R-M systems. Sequence analysis of the HsdS subunit of several different potential type IC R-M systems shows that the presence of different tetra-amino-acid sequence repeats, e.g. TAEL, LEAT, SEAL, TSEL, is characteristic for type IC R-M systems encoded by distantly related bacteria. The other regions of the HsdS subunits potentially responsible for subunit interaction are also different between a group of distantly related bacteria, but show high similarity within these bacteria. Complementation between the NgoAV R-M system and members of the EcoR124 R-M family can be restored by changing the tetra-amino-acid repeat within the HsdS subunit. The authors propose that the type IC family of R-M systems could consist of several complementation subgroups whose specificity would depend on differences in the conserved regions of the HsdS polypeptide.

Maintenance of species identity and controlling speciation of bacteria: a new function for restriction/modification systems?

Bacteria frequently exchange DNA among each other by horizontal gene transfer. However, maintenance of species identity and in particular speciation requires a certain barrier against an unregulated uptake of foreign DNA. Here it is suggested that formation of such a barrier is one important biological function of restriction/modification systems, in addition to the classical function of protection of bacteria against bacteriophage infection. This model explains the extreme variability and wide distribution of restriction/modification systems among prokaryotes, the prevalence of RM-systems in pathogenic bacteria and the existence of several RM-systems in single bacterial strains.

Tracking EcoKI and DNA fifty years on: a golden story full of surprises

1953 was a historical year for biology, as it marked the birth of the DNA helix, but also a report by Bertani and Weigle on 'a barrier to infection' of bacteriophage lambda in its natural host, Escherichia coli K-12, that could be lifted by 'host-controlled variation' of the virus. This paper lay dormant till Nobel laureate Arber and PhD student Dussoix showed that the lambda DNA was rejected and degraded upon infection of different bacterial hosts, unless it carried host-specific modification of that DNA, thus laying the foundations for the phenomenon of restriction and modification (R-M). The restriction enzyme of E.coli K-12, EcoKI, was purified in 1968 and required S-adenosylmethionine (AdoMet) and ATP as cofactors. By the end of the decade there was substantial evidence for a chromosomal locus hsdK with three genes encoding restriction (R), modification (M) and specificity (S) subunits that assembled into a large complex of >400 kDa. The 1970s brought the message that EcoKI cut away from its DNA recognition target, to which site the enzyme remained bound while translocating the DNA past itself, with concomitant ATP hydrolysis and subsequent double-strand nicks. This translocation event created clearly visible DNA loops in the electron microscope. EcoKI became the archetypal Type I R-M enzyme with curious DNA translocating properties reminiscent of helicases, recognizing the bipartite asymmetric site AAC(N6)GTGC. Cloning of the hsdK locus in 1976 facilitated molecular understanding of this sophisticated R-M complex and in an elegant 'pas de deux' Murray and Dryden constructed the present model based on a large body of experimental data plus bioinformatics. This review celebrates the golden anniversary of EcoKI and ends with the exciting progress on the vital issue of restriction alleviation after DNA damage, also first reported in 1953, which involves intricate control of R subunit activity by the bacterial proteasome ClpXP, important results that will keep scientists on the EcoKI track for another 50 years to come.

Characterization of cytosine methylated regions and 5-cytosine DNA methyltransferase (Ehmeth) in the protozoan parasite Entamoeba histolytica

The DNA methylation status of the protozoan parasite Entamoeba histolytica was heretofore unknown. In the present study, we developed a new technique, based on the affinity of methylated DNA to 5-methylcytosine antibodies, to identify methylated DNA in this parasite. Ribosomal DNA and ribosomal DNA circles were isolated by this method and we confirmed the validity of our approach by sodium bisulfite sequencing. We also report the identification and the characterization of a gene, Ehmeth, encoding a DNA methyltransferase strongly homologous to the human DNA methyltransferase 2 (Dnmt2). Immunofluorescence microscopy using an antibody raised against a recombinant Ehmeth showed that Ehmeth is concentrated in the nuclei of trophozoites. The recombinant Ehmeth has a weak but significant methyltransferase activity when E.histolytica genomic DNA is used as substrate. 5-Azacytidine (5-AzaC), an inhibitor of DNA methyltransferase, was used to study in vivo the role of DNA methylation in E.histolytica. Genomic DNA of trophozoites grown with 5-AzaC (23 microM) was undermethylated and the ability of 5-AzaC-treated trophozoites to kill mammalian cells or to cause liver abscess in hamsters was strongly impaired.

Single-stranded DNA binding and methylation by EcoP1I DNA methyltransferase

EcoP1I methyltransferase (M.EcoP1I) belongs to the type III restriction-modification system encoded by prophage P1 that infects Escherichia coli. Binding of M.EcoP1I to double-stranded DNA and single-stranded DNA has been characterized. Binding to both single- and double-stranded DNA could be competed out by unlabeled single-stranded DNA. Metal ions did not influence DNA binding. Interestingly, M.EcoP1I was able to methylate single-stranded DNA. Kinetic parameters were determined for single- and double-stranded DNA methylation. This feature of the enzyme probably functions in protecting the phage genome from restriction by type III restriction enzymes and thus could be considered as an anti-restriction system. This study describing in vitro methylation of single-stranded DNA by the type III methyltransferase EcoP1I allows understanding of the mechanism of action of these enzymes and also their role in the biology of single-stranded phages.

Crystal Structure of Type IIE Restriction Endonuclease EcoRII Reveals an Autoinhibition Mechanism by a Novel Effector-binding Fold

EcoRII is a type IIE restriction endonuclease that interacts with two copies of the DNA recognition sequence 5'CCWGG, one being the actual target of cleavage, the other serving as the allosteric effector. The mode of enzyme activation by effector binding is unknown. To investigate the molecular basis of activation and cleavage mechanisms by EcoRII, the crystal structure of EcoRII mutant R88A has been solved at 2.1A resolution. The EcoRII monomer has two domains linked through a hinge loop. The N-terminal effector-binding domain has a novel DNA recognition fold with a prominent cleft. The C-terminal catalytic domain has a restriction endonuclease-like fold. Structure-based sequence alignment identified the putative catalytic site of EcoRII that is spatially blocked by the N-terminal domain. The structure together with the earlier characterized EcoRII enzyme activity enhancement in the absence of its N-terminal domain reveal an autoinhibition/activation mechanism of enzyme activity mediated by a novel effector-binding fold. This is the first case of autoinhibition, a mechanism described for many transcription factors and signal transducing proteins, of a restriction endonuclease.

Diversity of type II restriction endonucleases that require two DNA recognition sites

Orthodox Type IIP restriction endonucleases, which are commonly used in molecular biological work, recognize a single palindromic DNA recognition sequence and cleave within or near this sequence. Several new studies have reported on structural and biochemical peculiarities of restriction endonucleases that differ from the orthodox in that they require two copies of a particular DNA recognition sequence to cleave the DNA. These two sites requiring restriction endonucleases belong to different subtypes of Type II restriction endonucleases, namely Types IIE, IIF and IIS. We compare enzymes of these three types with regard to their DNA recognition and cleavage properties. The simultaneous recognition of two identical DNA sites by these restriction endonucleases ensures that single unmethylated recognition sites do not lead to chromosomal DNA cleavage, and might reflect evolutionary connections to other DNA processing proteins that specifically function with two sites.

DNA recognition by the homing endonuclease PI-SceI involves a divalent metal ion cofactor-induced conformational change

PI-SceI, a homing endonuclease of the LAGLIDADG family, consists of two domains involved in DNA cleavage and protein splicing, respectively. Both domains cooperate in binding the recognition sequence. Comparison of the structures of PI-SceI in the absence and presence of substrate reveals major conformational changes in both the protein and DNA. Notably, in the protein-splicing domain the loop comprising residues 53-70 and adopts a "closed" conformation, thus enabling it to interact with the DNA. We have studied the dynamics of DNA binding and subsequent loop movement by fluorescence techniques. Six amino acids in loop53-70 were individually replaced by cysteine and modified by fluorescein. The interaction of the modified PI-SceI variants with the substrate, unlabeled or labeled with tetramethylrhodamine, was analyzed in equilibrium and stopped-flow experiments. A kinetic scheme was established describing the interaction between PI-SceI and DNA. It is noteworthy that the apparent hinge-flap motion of loop53-70 is only observed in the presence of a divalent metal ion cofactor. Substitution of the major Mg2+-binding ligands in PI-SceI, Asp-218 and Asp-326, by Asn or "nicking" PI-SceI with trypsin at Arg-277, which interferes with formation of an active enzyme.substrate complex, both prevent the conformational change of loop53-70. Deletion of the loop inactivates the enzyme. We conclude that loop53-70 is an important structural element that couples DNA recognition by the splicing domain with DNA cleavage by the catalytic domain and as such "communicates" with the Mg2+ binding sites at the catalytic centers.

A Specificity Switch in Selected Cre Recombinase Variants Is Mediated by Macromolecular Plasticity and Water

The basis for the altered DNA specificities of two Cre recombinase variants, obtained by mutation and selection, was revealed by their cocrystal structures. The proteins share similar substitutions but differ in their preferences for the natural LoxP substrate and an engineered substrate that is inactive with wild-type Cre, LoxM7. One variant preferentially recombines LoxM7 and contacts the substituted bases through a hydrated network of novel interlocking protein-DNA contacts. The other variant recognizes both LoxP and LoxM7 utilizing the same DNA backbone contact but different base contacts, facilitated by an unexpected DNA shift. Assisted by water, novel interaction networks can arise from few protein substitutions, suggesting how new DNA binding specificities might evolve. The contributions of macromolecular plasticity and water networks in specific DNA recognition observed here present a challenge for predictive schemes.

Glucocorticoid Receptor-like Zn(Cys)4 Motifs in BslI Restriction Endonuclease

BslI restriction endonuclease cleaves the symmetric sequence CCN(7)GG (where N=A, C, G or T). The enzyme is composed of two subunits, alpha and beta, that form a heterotetramer (alpha(2)beta(2)) in solution. The alpha subunit is believed to be responsible for DNA recognition, while the beta subunit is thought to mediate cleavage. Here, for the first time, we provide experimental evidence that BslI binds Zn(II). Specifically, using X-ray absorption spectroscopic analysis we show that the alpha subunit of BslI contains two Zn(Cys)(4)-type zinc motifs similar to those in the DNA-binding domain of the glucocorticoid receptor. This conclusion is supported by genetic analysis of the zinc-binding motifs, whereby amino acid substitutions in the zinc finger motifs are demonstrated to abolish or impair cleavage activity. An additional putative zinc-binding motif was identified in the beta subunit, consistent with the X-ray absorption data. These data were corroborated by proton induced X-ray emission measurements showing that full BslI contains at least three fully occupied Zn sites per alpha/beta heterodimer. On the basis of these data, we propose a role for the BslI Zn motifs in protein-DNA as well as protein-protein interactions.

Identification and mutational analysis of Mg2+ binding site in EcoP15I DNA methyltransferase: involvement in target base eversion

EcoP15I DNA methyltransferase catalyzes the transfer of the methyl group of S-adenosyl-l-methionine to the N6 position of the second adenine within the double-stranded DNA sequence 5'-CAGCAG-3'. To achieve catalysis, the enzyme requires a magnesium ion. Binding of magnesium to the enzyme induces significant conformational changes as monitored by circular dichroism spectroscopy. EcoP15I DNA methyltransferase was rapidly inactivated by micromolar concentrations of ferrous sulfate in the presence of ascorbate at pH 8.0. The inactivated enzyme was cleaved into two fragments with molecular masses of 36 and 35 kDa. Using this affinity cleavage assay, we have located the magnesium binding-like motif to amino acids 355-377 of EcoP15I DNA methyltransferase. Sequence homology comparisons between EcoP15I DNA methyltransferase and other restriction endonucleases allowed us to identify a PD(X)n(D/E)XK-like sequence as the putative magnesium ion binding site. Point mutations generated in this region were analyzed for their role in methyltransferase activity, metal coordination, and substrate binding. Although the mutant methyltransferases bind DNA and S-adenosyl-l-methionine as well as the wild-type enzyme does, they are inactive primarily because of their inability to flip the target base. Collectively, these data are consistent with the fact that acidic amino acid residues of the region 355-377 in EcoP15I DNA methyltransferase are important for the critical positioning of magnesium ions for catalysis. This is the first example of metal-dependent function of a DNA methyltransferase. These findings provide impetus for exploring the role(s) of metal ions in the structure and function of DNA methyltransferases.

Kinetic analysis of the coordinated interaction of SgrAI restriction endonuclease with different DNA targets

SgrAI restriction endonuclease cooperatively interacts and cleaves two target sites that include both the canonical sites, CPuCCGGPyG, and the secondary sites, CPuCCGGPy(A/T/C). It has been observed that the cleaved canonical sites stimulate SgrAI cleavage at the secondary sites. Equilibrium binding studies show that SgrAI binds to its canonical sites with a high affinity (Ka = 4-8 x 10(10) M-1) and that it has a 15-fold lower affinity for the cleaved canonical sites and a 30-fold lower affinity for the secondary sites. Steady-state kinetics reveals substrate cooperativity for SgrAI cleavage on both canonical and secondary sites. The specificity of SgrAI for the secondary site CACCGGCT, as measured by kcat/K is about 500-fold lower than that for the canonical site CACCGGCG, but this difference is reduced to 10-fold in the presence of the cleaved canonical sites. The efficiency of canonical site cleavage also increases by 3-fold when the cleaved canonical sites are present in the reaction. Furthermore, the substrate cooperativity for SgrAI cleavage is abolished for both types of sites in the presence of cleaved canonical sites. These results indicate that target site cleavage occurs via a coordinated interaction of two SgrAI protein subunits, where the subunit bound to the cleaved site stimulates the cleavage of the uncut site bound by the other subunit. The free subunits of SgrAI have the flexibility to bind different target sites and, consequently, assemble into various catalytically active complexes, which differ in their catalytic efficiencies.

Metal Ion Binding in the Active Site of the Junction-resolving Enzyme T7 Endonuclease I in the Presence and in the Absence of DNA

Endonuclease I of bacteriophage T7 is a DNA junction-resolving enzyme. We have previously used crystallography to demonstrate the binding of two manganese ions into the active site that is formed by three carboxylate (Glu 20, Asp 55 and Glu 65) and a lysine residue (Lys 67). Endonuclease I is active in the presence of magnesium, manganese, iron (II) and cobalt (II) ions, weakly active in the presence of nickel, copper (II) and zinc ions, and completely inactive in the presence of calcium ions. However, using calorimetry, we have observed the binding of two calcium ions to the free enzyme in a manner very similar to the binding of manganese ions. In the presence of iron (II) ions, we have obtained a cleavage of the continuous strands of a junction bound by endonuclease I, at sites close to (but not identical with) enzyme-induced hydrolysis. The results suggest that this arises from attack by locally generated hydroxyl radicals, arising from iron (II) ions bound into the active site. This therefore provides an indirect way of examining metal ion binding in the enzyme-junction complex. Ion binding in free protein (by calorimetry) and the enzyme-junction complex (iron-induced cleavage) have been studied in series of active-site mutants. Both confirm the importance of the three carboxylate ligands, and the lack of a requirement for Lys67 for the ion binding. Calorimetry points to particularly critical role of Asp55, as mutation completely abolishes all binding of both manganese and calcium ions.

DNA recognition by the EcoRV restriction endonuclease probed using base analogues

The EcoRV restriction endonuclease recognises palindromic GATATC sequences and cuts between the central T and dA bases in a reaction that has an absolute requirement for a divalent metal ion, physiologically Mg(2+). Use has been made of base analogues, which delete hydrogen bonds between the protein and DNA (or hydrophobic interactions in the case of the 5-CH(3) group of thymine), to evaluate the roles of the outer two base-pairs (GATATC) in DNA recognition. Selectivity arises at both the binding steps leading to the formation of the enzyme-DNA-metal ion ternary complex (assayed by measuring the dissociation constant in the presence of the non-reactive metal Ca(2+)) and the catalytic step (evaluated using single-turnover hydrolysis in the presence of Mg(2+)), with each protein-DNA contact contributing to recognition. With the A:T base-pair, binding was reduced by the amount expected for the simple loss of a single contact; much more severe effects were observed with the G:C base-pair, suggesting additional conformational perturbation. Most of the modified bases lowered the rate of hydrolysis; furthermore, the presence of an analogue in one strand of the duplex diminished cutting at the second, unmodified strand, indicative of communication between DNA binding and the active site. The essential metal ion Mg(2+) plays a key role in mediating interactions between the DNA binding site and active centre and in many instances rescue of hydrolysis was seen with Mn(2+). It is suggested that contacts between the GATATC site are required for tight binding and for the correct assembly of metal ions and bound water at the catalytic site, functions important in providing acid/base catalysis and transition state stabilisation.

DNA binding and cleavage by the periplasmic nuclease Vvn: a novel structure with a known active site

The Vibrio vulnificus nuclease, Vvn, is a non-specific periplasmic nuclease capable of digesting DNA and RNA. The crystal structure of Vvn and that of Vvn mutant H80A in complex with DNA were resolved at 2.3 A resolution. Vvn has a novel mixed alpha/beta topology containing four disulfide bridges, suggesting that Vvn is not active under reducing conditions in the cytoplasm. The overall structure of Vvn shows no similarity to other endonucleases; however, a known 'betabetaalpha-metal' motif is identified in the central cleft region. The crystal structure of the mutant Vvn-DNA complex demonstrates that Vvn binds mainly at the minor groove of DNA, resulting in duplex bending towards the major groove by approximately 20 degrees. Only the DNA phosphate backbones make hydrogen bonds with Vvn, suggesting a structural basis for its sequence-independent recognition of DNA and RNA. Based on the enzyme-substrate and enzyme-product structures observed in the mutant Vvn-DNA crystals, a catalytic mechanism is proposed. This structural study suggests that Vvn hydrolyzes DNA by a general single-metal ion mechanism, and indicates how non-specific DNA-binding proteins may recognize DNA.

Pseudocomplementary PNAs as selective modifiers of protein activity on duplex DNA: the case of type IIs restriction enzymes

This study evaluates the potential of pseudocomplementary peptide nucleic acids (pcPNAs) for sequence-specific modification of enzyme activity towards double-stranded DNA (dsDNA). To this end, we analyze the ability of pcPNA-dsDNA complexes to site-selectively interfere with the action of four type IIs restriction enzymes. We have found that pcPNA-dsDNA complexes exhibit a different degree of DNA protection against cleaving/nicking activity of various isoschizomeric endonucleases under investigation (PleI, MlyI and N.BstNBI) depending on their type and mutual arrangement of PNA-binding and enzyme recognition/cleavage sites. We have also found that the pcPNA targeting to closely located PleI or BbsI recognition sites on dsDNA generates in some cases the nicking activity of these DNA cutters. At the same time, MlyI endonuclease, a PleI isoschizomer, does not exhibit any DNA nicking/cleavage activity, being completely blocked by the nearby pcPNA binding. Our results have general implications for effective pcPNA interference with the performance of DNA-processing proteins, thus being important for prospective applications of pcPNAs.

Isolation of BsoBI restriction endonuclease variants with altered substrate specificity

BsoBI is a thermophilic restriction endonuclease that cleaves the degenerate DNA sequence C/PyCGPuG (where/=the cleavage site and Py=C or T, Pu=A or G). In the BsoBI-DNA co-crystal structure the D246 residue makes a water-mediated hydrogen bond to N6 of the degenerate base adenine and was proposed to make a complementary bond to O6 of the alternative guanine residue. To investigate the substrate specificity conferred by D246 and to potentially alter BsoBI specificity, the D246 residue was changed to the other 19 amino acids. Variants D246A, D246C, D246E, D246R, D246S, D246T, and D246Y were purified and their cleavage activity determined. Variants D246A, D246S, and D246T display 0.2% to 0.7% of the wild-type cleavage activity. However, the substrate specificity of the three variants is altered significantly. D246A, D246S, and D246T cleave CTCGAG sites poorly. In filter binding assays using oligonucleotides, wild-type BsoBI shows almost equal affinity for CTCGAG and CCCGGG sites. In contrast, the D246A variant shows 70-fold greater binding affinity for the CCCGGG substrate. Recycled mutagenesis was carried out on the D246A variant, and revertants with enhanced activity were isolated by their dark blue phenotype on a dinD Colon, two colons lacZ DNA damage indicator strain. Most of the amino acid substitutions present within the revertants were located outside the DNA-protein interface. This study demonstrates that endonuclease mutants with altered specificity and non-lethal activity can be evolved towards more active variants using a laboratory evolution strategy.

NEBcutter: A program to cleave DNA with restriction enzymes

NEBcutter, version 1.0, is a program available via a web server (http://tools.neb.com/NEBcutter) that will accept an input DNA sequence and produce a comprehensive report of the restriction enzymes that will cleave the sequence. It produces a variety of outputs including restriction enzyme maps, theoretical digests and links into the restriction enzyme database, REBASE (http://www.neb.com/rebase). Importantly, its table of recognition sites is updated daily from REBASE and it marks all sites that are potentially affected by DNA methylation (Dam, Dcm, etc.). Many options exist to choose the enzymes used for digestion, including all known specificities, subsets of those that are commercially available or sets of enzymes that produce compatible termini.

PspGI, a type II restriction endonuclease from the extreme thermophile Pyrococcus sp.: structural and functional studies to investigate an evolutionary relationship with several mesophilic restriction enzymes

We present here the first detailed biochemical analysis of an archaeal restriction enzyme. PspGI shows sequence similarity to SsoII, EcoRII, NgoMIV and Cfr10I, which recognize related DNA sequences. We demonstrate here that PspGI, like SsoII and unlike EcoRII or NgoMIV and Cfr10I, interacts with and cleaves DNA as a homodimer and is not stimulated by simultaneous binding to two recognition sites. PspGI and SsoII differ in their basic biochemical properties, viz. stability against chemical denaturation and proteolytic digestion, DNA binding and the pH, MgCl(2) and salt-dependence of their DNA cleavage activity. In contrast, the results of mutational analyses and cross-link experiments show that PspGI and SsoII have a very similar DNA binding site and catalytic center as NgoMIV and Cfr10I (whose crystal structures are known), and presumably also as EcoRII, in spite of the fact that these enzymes, which all recognize variants of the sequence -/CC-GG- (/ denotes the site of cleavage), are representatives of different subgroups of type II restriction endonucleases. A sequence comparison of all known restriction endonuclease sequences, furthermore, suggests that several enzymes recognizing other DNA sequences also share amino acid sequence similarities with PspGI, SsoII and EcoRII in the region of the presumptive active site. These results are discussed in an evolutionary context.

Modulation of the active complex assembly and turnover rate by protein-DNA interactions in Cre-LoxP recombination

Cre promotes recombination at the 34 bp LoxP sequence. Substitution of a critical C-G base pair in LoxP with an A-T base pair, to give LoxAT, reduced Cre binding in vitro and abolished recombination in vivo [Hartung, M., and Kisters-Woike, B. (1998) J. Biol. Chem. 273, 22884-22891].We demonstrated that LoxAT can be recombined in vitro. However, Cre discriminates against this substrate both before and after DNA binding. The preference for LoxP over LoxAT is the result of reduced binding and a slower turnover rate, amplified by changes in cooperativity of complex assembly. With LoxAT, similar levels of substrate turnover required 2-2.5-fold higher protein-DNA concentrations compared to LoxP, but the sigmoidal behavior of the concentration dependence was more pronounced. Further, the Cre-LoxAT complexes reacted 4-5-fold more slowly. In the 2.3 A resolution Cre-LoxAT complex structure, the major groove Arg259-guanine interaction was disrupted, explaining the reduced binding. Overall structural shifts and mobility changes indicate more favorable interactions between subunits, providing a hypothesis for the reduced turnover rate. Concomitant with the displacement of Arg259 from the DNA, adjacent charged residues Glu262 and Glu266 shifted to form salt bridges with the Arg259 guanidinium moiety. Substitution of Glu262 and Glu266 with glutamine increased Cre complex assembly efficiency and reaction rates with both LoxAT and LoxP, but diminished Cre's ability to distinguish them. The increased rate of this variant suggests that DNA substrate binding and turnover are coupled. The improved efficiency, made at some expense of sequence discrimination, may be useful for enhancing recombination in vivo.

Continuous Monitoring of a Restriction Enzyme Digest of DNA on a Microchip with Automated Capillary Sample Introduction

Continuous analysis of a DNA restriction enzyme digest on a microfabricated device is demonstrated with minimal intervention and enhanced time resolution. A 62-base-pair fragment of dsDNA containing a KpnI site was used to demonstrate this process. A capillary was used to transfer sample from a single reaction mix to a microfabricated chip with parallel separation lanes. The 6-carboxyfluorescein-labeled DNA fragments were detected with a CCD camera as they separated in the lanes, which were filled with linear polyacrylamide. The products of the restriction enzyme digest were monitored for up to 60 min at an average sampling rate of 1 injection/46 s, with consecutive injections as short as 1 injection/14 s. The digest was injected directly into the chip, eliminating the need for any sample-handling steps after addition of the enzyme to the reaction mix. The effects of temperature and restriction enzyme concentration were briefly examined, as well. This work shows the potential of this method to provide valuable information about the process of restriction enzyme cleavage.

Flexible DNA target site recognition by divergent homing endonuclease isoschizomers I-CreI and I-MsoI

Homing endonucleases are highly specific catalysts of DNA strand breaks that induce the transposition of mobile intervening sequences containing the endonuclease open reading frame. These enzymes recognize long DNA targets while tolerating individual sequence polymorphisms within those sites. Sequences of the homing endonucleases themselves diversify to a great extent after founding intron invasion events, generating highly divergent enzymes that recognize similar target sequences. Here, we visualize the mechanism of flexible DNA recognition and the pattern of structural divergence displayed by two homing endonuclease isoschizomers. We determined structures of I-CreI bound to two DNA target sites that differ at eight of 22 base-pairs, and the structure of an isoschizomer, I-MsoI, bound to a nearly identical DNA target site. This study illustrates several principles governing promiscuous base-pair recognition by DNA-binding proteins, and demonstrates that the isoschizomers display strikingly different protein/DNA contacts. The structures allow us to determine the information content at individual positions in the binding site as a function of the distribution of direct and water-mediated contacts to nucleotide bases, and provide an evolutionary snapshot of endonucleases at an early stage of divergence in their target specificity.

How the BfiI restriction enzyme uses one active site to cut two DNA strands

Unlike other restriction enzymes, BfiI functions without metal ions. It recognizes an asymmetric DNA sequence, 5'-ACTGGG-3', and cuts top and bottom strands at fixed positions downstream of this sequence. Many restriction enzymes are dimers of identical subunits, with one active site for each DNA strand. Others, like FokI, dimerize transiently during catalysis. BfiI is also a dimer but it has only one active site, at the dimer interface. We show here that BfiI remains a dimer as it makes double-strand breaks in DNA and that its single active site acts sequentially, first on the bottom and then the top strand. Hence, after cutting the bottom strand, a rearrangement of either the protein and/or the DNA in the BfiI-DNA complex must switch the active site to the top strand. Low pH values selectively block top-strand cleavage, converting BfiI into a nicking enzyme that cleaves only the bottom strand. The switch to the top strand may depend on the ionization of the cleaved 5' phosphate in the bottom strand. BfiI thus uses a mechanism for making double-strand breaks that is novel among restriction enzymes.