Specific cleavage of DNA molecules at RecA-mediated triple-strand stru
http://www.100md.com
《核酸研究医学期刊》
Laboratory of Human Gene Research II, Kazusa DNA Research Institute, Kazusakamatari 2-6-7, Kisarazu, Chiba 292-0812, Japan
*To whom correspondence should be addressed. Tel: +81 438 52 3945; Fax: +81 438 52 3946; Email: oishi@kazusa.or.jp
Present address:
Yasushi Shigemori, Biotechnology Division, Aisin Cosmos R & D Co. Ltd, 2-3-9 Kazusa-kamatari, Kisarazu, Chiba 292-0818, Japan
ABSTRACT
A novel procedure to cleave DNA molecules at any desired base sequence is presented. This procedure is based upon our finding that double-stranded DNA molecules at a site where RecA-mediated triple-stranded DNA structure with a complimentary deoxyoligonucleotide is located can be cleaved by a single-strand specific nuclease, such as nuclease S1 or BAL31, between the first base at the 5' termini of the deoxyoligonucleotides and the nearest base proximal to the 5' termini. Accordingly, the sequence as well as the number of the cleavage sites to be cleaved can be custom designed by selecting deoxyoligonucleotides with specific base sequences for triple-stranded DNA formation. The basic characteristics of the cleavage reaction and typical applications of the procedure are presented with actual results, including those which involve cleavage of complex genomic DNA at the very sites one desires.
INTRODUCTION
Although restriction enzymes have been widely used in molecular biological studies, the DNA cleavage sites are limited to specific base sequences, base sequences with palindromic structures (1). Cleavage at any site in DNA molecules, if established, would greatly expand current DNA technologies beyond the boundaries imposed by the limitations in recognizing sequences associated with restriction enzymes. While studying susceptibility of triple-stranded DNA molecules to various nucleases we found that nuclease S1 (2) and BAL31 (3) cleave DNA molecules specifically at the site where a triple-stranded DNA structure (4–9) is located. Nuclease S1 is known to cleave single-stranded DNA (or RNA) endonuclolytically to yield mono- and oligonucleotides with 5' phosphate as the ultimate product (2,10–12). The enzyme also attacks double-stranded DNA molecules at the sites where nicks are present producing blunt-ended double-stranded DNA with 5' phosphate (11,12). BAL31 has similar endonucleolytic activities towards single-stranded DNA but it also attacks double-stranded DNA in a non-discriminate fashion to a much lesser extent than nuclease S1 does (3,13).
Since triple-stranded DNA can be formed readily at any location in double-stranded DNA with a homologous deoxyoligonucleotide in the presence of RecA protein (4–16), we have explored the possibility of applying this finding to specific cleavage (and subsequent cloning) of DNA molecules (fragment between) at the very sites one desires.
MATERIALS AND METHODS
Deoxyoligonucleotides
Deoxyoligonucleotides (HPLC purified grade) were custom synthesized by Sawady Technology, Tokyo, Japan. Unless otherwise specified, no phosphate moiety was present at the 5' termini.
Triple-strand cleavage
The following protocols were employed for triple-strand cleavage except for the cleavage of human genomic DNA (see below). Target DNA (600 ng), homologous 60mer deoxyoligonucleotides and RecA protein (6 μg; Epicenter Technologies, Madison, WI, USA) were mixed and incubated in a reaction mixture (20 μl), which contained ATP-S (2.4 mM, HPLC purified grade; Roche Diagnostics, Mannheim, Germany), magnesium acetate (14 mM) and Tris-acetate (30 mM, pH 7.2), for 30 min at 37°C. For nuclease S1 digestion, the mixture was combined with a reaction mixture (80 μl) containing nuclease S1 (200 units; Takara Shuzo, Shiga, Japan), NaCl (340 mM) and ZnSO4 (1.35 mM), glycerol (6.8%, v/v) and sodium acetate (40 mM, pH 4.6) for 30 min at 55°C. For BAL31 digestion, the mixture was combined with a reaction mixture (80 μl) containing nuclease BAL31 (20 units; Takara Shuzo), NaCl (600 mM), CaCl2 (12 mM), MgCl2 (12 mM), EDTA (1 mM) and Tris–HCl (20 mM, pH 8.0) for 15 min at 37°C. S1 nuclease (or BAL31) and RecA protein were then inactivated by treating the reaction mixture with SDS (0.5%, w/v), Proteinase K (0.1 mg/ml; Roche Diagnostics), EDTA (5 mM) in Tris–HCl buffer (150 mM, pH 8.8) for 30 min at 54°C, followed by phenol–chloroform treatment. After the addition of a one-tenth volume of sodium acetate (3 M), DNA was ethanol precipitated, dissolved in 20 μl of TE buffer (10 mM Tris–HCl, pH 8.0 and 1 mM EDTA) and subjected to agarose (1%) gel electrophoresis or other analyses.
For the cleavage of human genomic DNA, human DNA (50 μg) (Promega, Madison, WI, USA), which had been treated with NotI (New England Biolabs, Beverly, MA, USA), was incubated in a reaction mixture (100 μl) containing deoxyoligonucleotides , RecA protein (50 μg; Epicenter Technologies), ATP-S (4.8 mM; Roche Diagnostics), magnesium acetate (14 mM) and Tris-acetate (30 mM, pH 7.2) for 3 h at 37°C. The sample was diluted with a reaction mixture (300 μl) containing recombinant nuclease S1 (17) (see text; 200 units, kindly provided by Ozeki Shuzo Co., Nishinomiya, Japan), NaCl (340 mM), ZnSO4 (1.35 mM), glycerol (6.8%, v/v) and sodium acetate (40 mM, pH 4.6) incubated for 30 min at 40°C. The reaction was terminated by incubating (54°C for 30 min) in a solution with EDTA (5 mM), SDS (0.5%), Proteinase K (0.1 mg/ml; Roche Diagnostics) and Tris–HCl (150 mM, pH 8.8). After phenol–chloroform treatment, DNA was precipitated by ethanol and dissolved in TE buffer (40 μl). DNA was then electrophoresed on agarose gel (1%).
Determination of the nuclease S1 cleavage site
pBR322 DNA, which had been linearized by NdeI, was incubated with one of the following deoxyoligonucleotides (60mer, HPLC purified grade) terminating their 5' termini with cytosine (oligo C, nucleotides 5–64 of pBR322 DNA), adenine (oligo D, nucleotides 6–65), thymine (oligo E, nucleotides 7–66) or guanine (oligo F, nucleotides 8–67) in the presence of RecA protein. Each sample was subjected to nuclease S1 treatment. After removal of RecA protein and nuclease S1, the samples were treated with either restriction enzyme SspI (New England Biolabs) or EcoRV (New England Biolabs). The 5' terminus of each fragment was labeled with 32P through the exchange reaction by T4 polynucleotide kinase (MEGALABELTM Labeling Kit; Takara Shuzo) using ATP (3000 Ci/mmol; Amersham Pharmacia Biotech, Buckinghamshire, UK), electrophoresed on denaturing polyacrylamide sequence gel (4.5%) and autoradiographed. For determination of the size of each fragment (number of bases), labeled control DNA (M13mp18 using nucleotides 6310–6326 deoxyoligonucleotide as a primer for the dideoxy-mediated chain termination method) was elecrophoresed in the same gel.
RESULTS
Triple-strand cleavage by nuclease S1 and BAL31
As shown in Figure 1A, when pBR322 DNA, which had previously been subjected to triple-stranded DNA formation with a 60mer homologous deoxyoligonucleotide (oligo A, nucleotides 3787–3846 of pBR322 DNA), was incubated with nuclease S1 and electrophoresed, two additional shorter DNA bands with 2.9 and 1.5 kb, which were expected from the cleavage of DNA at or near the triple-strand structure, appeared (lane 1). The shorter DNA bands did not appear without RecA protein for triple-stranded DNA formation (lane 2), without nuclease S1 treatment (lane 3), without both treatments (control, lane 4), or with a non-homologous deoxyoligonucleotide (oligo B, homologous to nucleotides 1210–1269 of M13mp18 DNA) (lane 5). The cleavage efficiency was generally 60–80% and was not increased further with higher nuclease S1 concentrations suggesting that the efficiency of triple-stranded DNA formation (with RecA protein) is a limiting factor for the cleavage efficiency under the condition we employed. Essentially the same results were obtained with BAL31 treatment, but with considerably lesser cleavage efficiency (Fig. 1B, lanes 3 and 4). Because of better efficiency of cleavage, further characterization was carried out with nuclease S1.
Figure 1. Cleavage of DNA molecules carrying a RecA-mediated triple-stranded DNA structure by nuclease S1 and BAL31. (A) Linearized (NdeI) pBR322 was incubated with a 60mer deoxyoligonucleotide (oligo A or oligo B) in the presence of RecA protein, treated with nuclease S1, subjected to agarose (1%) gel electrophoresis and stained with ethidium bromide. (B) The same as (A), but includes the results with BAL31 (lane 4) and control (lane 3). We show molecular size markers (kb) in lane M.
Effect of the chain length of deoxyoligonucleotides
As seen in Figure 2, the cleavage of DNA molecules by nuclease S1 at sites where triple-strand structures are present was observed most prominently for DNA that had been subjected to triple-strand formation using deoxyoligonucleotides with more than 60 nt (lanes 1, 2 and 3 for 100, 80 and 60mers, respectively) and no cleavage was observed with an oligomer of 20 nt (lane 6). It is likely that either a portion of the triple-strand structures with <60mer deoxyoligonucleotides becomes dissociated during incubation with nuclease S1, or triple-strand formation itself is affected by the chain length of the deoxyoligonucleotides, or both. For the experiments described below, we used 60mer deoxyoligonucleotides throughout for the cleavage of triple-strand DNA.
Figure 2. Effect of the chain length of deoxyoligonucleotides. pBR322 DNA, which had been incubated with deoxyoligonucleotides whose chain length ranged from 100 to 20 nt in the presence of RecA protein, was treated with nuclease S1, subjected to agarose gel electrophoresis and stained with ethidium bromide. Lane 1, pBR322 DNA incubated with a 100mer deoxyoligonucleotide (nucleotides 5–104); lane 2, the same as lane 1, but with a 80mer deoxyoligonucleotide (nucleotides 5–84); lane 3, the same as lane 1, but with a 60mer deoxyoligonucleotide (oligo C, nucleotides 5–64); lane 4, the same as lane 1, but with a 40mer deoxyoligonucleotide (nucleotides 5–44); lane 5, the same as lane 1, but with a 30mer deoxyoligonucleotide (nucleotides 5–34); and lane 6, the same as lane 1, but with a 20mer deoxyoligonucleotide (nucleotides 5–24). We show molecular size markers (kb) in lane M.
Determination of cleavage sites
First, pBR322 DNA linearized by the digestion with StyI was subjected to triple-strand formation with one of the four following deoxyoligonucleotides (60mer): oligo C (nucleotides 5–64 with cytosine at nucleotide 5), oligo D (nucleotides 6–65 with adenine at nucleotide 6), oligo E (nucleotides 7–66 with thymine at nucleotide 7) and oligo F (nucleotides 8–67 with guanine at nucleotide 8) (see Fig. 3A) and subsequently to cleavage by nuclease S1. As seen in Figure 3B, in addition to the original DNA (4.4 kb), two shorter DNA fragments (3 and 1.4 kb, respectively), each apparently corresponding to the fragments produced by the cleavage at or near the triple-stranded DNA structures, were produced with almost equal cleavage efficiency with all of these four DNA samples. To determine the precise site of cleavage, pBR322 DNA linearized by NdeI and carrying triple-strand structures with these deoxyoligonucleotides were subjected to similar nuclease S1 cleavage. After removal of RecA protein and nuclease S1, DNA was treated with a restriction enzyme (SspI) whose cutting site (with a blunt end) exists between nucleotides 4170 and 4171 of pBR322 DNA (see Fig. 3A). The products, after labeling the 5' termini of all the products with 32P through the exchange reaction by T4 polynucleotide kinase, were electrophoresed on sequencing gel and autoradiographed. As seen in Figure 3C, in addition to the signals at the top which were derived from labeled larger DNA fragments (presumably the NdeI–SspI fragment, the fragments between the cleaved sites at the triple-strand structures and NdeI, and the uncleaved SspI–NdeI fragment, see Fig. 3A), signals derived from small DNA fragments with 195 (lane 1 for oligo C), 196 (lane 2 for oligo D), 197 (lane 3 for oligo E) and 198 nt (lane 4 for oligo F) in their chain length were observed. Since those chain lengths correspond exactly to those of the fragments between the SspI site and the 5' termini of deoxyoligonucleotides in the triple-strand structure (see Fig. 3A), we have concluded that nuclease S1 cleaves both of the target DNA strands between the first base at the 5' termini of deoxyoligonucleotides in the triple-strand structure and the nearest base proximal to the 5' termini, producing blunt-ended DNA fragments with 5' phosphates. Since the intensity of the signals was not significantly varied among these bands (see also Fig. 3B), the cleavage apparently proceeds regardless of the type of base located at the 5' termini of deoxyoligonucleotides in the triple-strand structure. The presence of a phosphate moiety at the 5' terminus deoxyoligonucleotides had no effect on the cleavage pattern (data not shown). When we electrophoresed the same DNA samples but treated them with EcoRV (instead of SspI), which cleaves the DNA between nucleotides 186 and 187 (see Fig. 3A), we detected minor DNA bands (Fig. 3D) in addition to the major band, suggesting that nuclease S1 gives additional, probably secondary, cuts inside the triple-stranded DNA structure, producing one or a few base shorter cleavage products (see below). A tentative model for the overall triple-strand cleavage reaction by nuclease S1 is presented in Figure 4.
Figure 3. Determination of the nuclease S1 cleavage site. Linearized pBR322 DNA was subjected to triple-strand cleavage with a deoxyoligonucleotide terminating the 5' termini with either cytosine (oligo C), adenine (oligo D), thymine (oligo E) or guanine (oligo F). A portion of the DNA from each sample was electrophoresed on agarose gel and another portion of the DNA, after treating with SspI or EcoRV and labeling the 5' termini with 32P, was electrophoresed on a sequence gel and autoradiographed. (A) Diagrammatic representation of the positions and directions of deoxyoligonucleotides used for triplex cleavage. Nucleotide numbers (nt. numbers) are those registered in GenBank. The distances in the diagram do not necessarily reflect actual distances. (B) Agarose gel electrophoresis patterns of the products. Lane 1, products with oligo C; lane 2, products with oligo D; lane 3, products with oligo E; and lane 4, products with oligo F. In lane M, we show molecular size markers. (C) Autoradigraphic patterns of DNA after digestion with SspI and terminal labeling. (D) Auto radigraphic patterns of DNA after digestion with EcoRV and terminal labeling. (C and D) Lane 1, products with oligo C; lane 2, products with oligo D; lane 3, products with oligo E; and lane 4, products with oligo F. We also show sequence patterns of the control M13mp18 DNA on the left-hand side (lane M) and size of the DNA fragments (nt) on the right-hand side.
Figure 4. A tentative model for the overall triple-strand cleavage reaction by nuclease S1. The triple-strand structure shown in IV is highly speculative and may exist only very transiently.
Use of triple-strand cleavage for isolation (cloning) and generation of deletion of specific DNA sequences
In order to demonstrate the usefulness of our finding (to examine whether this procedure, in fact, can be used for excision and subsequent cloning of a specific DNA fragment between the very sites one desires), we attempted to actually excise and isolate the sequence between nucleotides 1 and 1276 in pBR322 DNA which corresponds to the entire tetracycline resistance gene. pBR322 DNA was subjected to simultaneous cleavages at two triple-strand structures in a single DNA molecule using a pair of 60mer deoxyoligonucleotides with opposite directions (their 5' termini facing each other) to avoid possible minor cleavage in the triple-strand structures (see Fig. 3D). One of the deoxyoligonucleotides (oligo G in Fig. 5A, complimentary to nucleotides 4302–4361) had a 5' terminus corresponding to the nearest nucleotide (nucleotide 4361), just proximal to the first nucleotide (nucleotide 1) of the gene, and the other (oligo H in Fig. 5A, nucleotides 1277–1336) had a 5' terminus corresponding to the nucleotide 1277, just distal to the last base (nucleotide 1276) of the gene (Fig. 5A). Figure 5B shows the cleavage pattern; no deoxyoligonucleotides (lane 1), oligo H alone (lane 2), oligo G alone (lane 3) and oligo G plus H (lane 4). When oligo G and H were simultaneously employed (lane 4), an extra band of 1.3 kb (indicated by an arrow) which corresponds to the entire tetracycline gene (1276 bp) was detected. The DNA band was eluted from the gel and blunt-end-ligated to a vector (pUC19), which is tetracycline sensitive, at a dephosphorylated HincII site. Colonies which presumably carried inserts were first examined for their phenotype (tetracycline resistance). All of the clones with inserts (total 22) were tetracycline resistant. Sequence analysis showed four out of five inserts represented complete sequences of the tetracycline resistance gene (nucleotides 1–1276). In one insert, one nucleotide at nucleotide 1 was missing in the 1276 bp insert. These results demonstrate that by employing a pair of deoxyoligonucleotides in which 5' termini of the deoxyoligonucleotides are facing each other, excision and isolation of DNA sequences from the sites one wants has become possible.
Figure 5. Simultaneous double triple-strand cleavage and cloning of tetracycline resistant gene. Linearized (NdeI) pBR322 DNA was subjected to triple-strand cleavage using oligo G or oligo H, or both. (A) Diagrammatic representation of double triple-strand cleavage in which positions and directions of the deoxyoligonucleotides used are shown. P1, P2, A(TG) and (TG)A indicate promoter 1, promoter 2, the initiation codon and termination codon of the gene, respectively. Distances in the diagram do not necessarily reflect actual distances. (B) Products stained with ethidium bromide after simultaneous double triple-strand cleavage. Arrow indicates the DNA fragment (1.3 kb) corresponding to the excised tetracycline resistant gene.
We were also able to generate deletions (the entire sequence of the tetracycline resistance gene) in open circular pBR322 DNA by employing a pair of deoxyoligonucleotides just inside the 5' and 3' termini of the target sequence (data not shown). In this case, the 5' termini of the deoxyoligonicleotides are directed outwardly and in the opposite direction. Generation of small deletions, shorter than the sum of the deoxynucleotides, was produced by employing pairs of deoxyoligonucleotides in which the 5' termini were facing each other, although the success rate of obtaining deletions exactly as originally intended was slightly <30% among the clones examined (data not shown).
Triple-strand cleavage of complex genomic DNA
We examined whether triple-strand cleavage can be performed directly with highly complex genomic DNA. Human genomic DNA pre-treated with NotI was subjected to simultaneous triple-strand cleavage at a specific gene (the p53 gene). One pair with oligo I (complimentary to nucleotides 553–612) and oligo J (nucleotides 965–1024) was used to excise a block of sequence (351 bp from nucleotides 843 to 949) which includes the exon 1 (107 bp) of the p53 gene, and another pair with oligo K (complimentary to nucleotides 18539–18598) and oligo L (nucleotides 19877–19936) was employed to excise the entire sequence of exon 11 (1278 bp from nucleotides 18599 to 19876) exactly (see Fig. 6A). Since treatment of complex genomic DNA such as human DNA with the nuclease S1 preparation used for the above experiments gave a low level non-specific cut, probably by a small amount of nucleases contaminated in the preparation, we employed nuclease S1 which was derived from a cloned nuclease S1 gene (17) for this experiment. After triple-strand cleavage, the DNA was electrophoresed and Southern hybridized using a radiolabeled p53 probe. As seen in Figure 6B, when the oligo I and J pair was employed, a band (0.3 kb) corresponding to the expected excised sequence (352 bp) was detected (lane 2 indicated by an arrow). No such band was detected in the control experiment in which non-homologous deoxyoligonucleotides (oligo G and H) were employed for the triple-strand cleavage (lane 1). Similar results were obtained with DNA products with another pair of deoxyoligonucleotides (oligo K and L), in which a DNA fragment (1.3 kb) close to the expected size (1278 bp) was observed (lane 4). The results suggest that triple-strand cleavage can be applied to highly complex DNA such as human genomic DNA.
Figure 6. Triple-strand cleavage of sequences specific to human p53 gene. (A) Diagrammatic representation of triple-strand cleavage of the sequences in p53 gene in which positions and directions of the deoxyoligonucleotides used are shown (top, oligo I and J; bottom, oligo K and L). In the diagram, exons and introns are shown as shaded (exons) and bold (introns) lines. Distances in the diagram do not necessarily reflect actual distances. (B) Patterns of Southern blot hybridization. NotI-treated human DNA (50 μg) was incubated with RecA protein and subjected to nuclease S1 treatment as described in Materials and Methods. DNA was electrophoresed on agarose gel (1%) and, after staining with ethidium bromide, subjected to Southern blot hybridization using radiolabeled p53 probes which were randomly primed by dCTP (6000 Ci/mmol; Amersham Pharmacia Biotech) using a kit (BcaBESTTM Labeling Kit; Takara Shuzo) and autoadiographed. Arrows indicate the positions of signals detected by hybridization. (C) Patterns of DNA stained with ethidium bromide after electrophoresis.
DISCUSSION
In this paper, we have presented our findings, which makes it possible to cleave DNA molecules at any known sequence one desires by a relatively simple process. The sequences as well as the number of cleavage sites can be custom designed by selecting specific base sequences. This may be particularly useful when one wants to make a single cut to introduce a specific sequence or to generate a deletion at a desired position in DNA molecules. There have been previous attempts to cleave DNA molecules at targeted sites. DNA molecules can be specifically cleaved by chemical treatment of triple-strand structures formed through Hoogsteen base pairing, although the target base sequences to be cleaved are limited only to polypurines and polypyrimidines (18–20). Another approach was to first protect the sites to be cleaved against methylation either by forming a complex with specific proteins (21,22) or Hoogsteen base pairing (23), and protected (unmethylated) sites were subsequently cleaved by methylation-sensitive restriction enzymes. In these cases, however, the sequences to be cleaved are still limited to those recognized by (methylation-sensitive) restriction enzymes.
Besides the introduction of a single cut or generation of a deletion at desired sites, one of the possible applications of the finding would be for isolation of any sequence of interest from even complex genomic DNA. Nowadays, isolation of a specific sequence can be relatively easily achieved by amplifying the sequence by PCR. On the other hand, there are several cases in which PCR is not satisfactorily applicable. For example, amplification of a long stretch of sequence over several kilobases is still very difficult, though not impossible, by PCR. Introduction of mismatched bases during PCR in these long sequences often poses a problem. Our procedure may offer an alternative way of isolation of specific sequences in such cases.
We have shown that nuclease S1 cleaves both the target DNA strands between the first base at the 5' termini of deoxyoligonucleotides in the triple-strand structure and the nearest base proximal to the 5' termini, thus producing blunt-ended DNA fragments with 5' phosphate. Triple-stranded DNA structure is likely to generate a regional conformational distortion at the boundaries to double-stranded DNA molecules which is recognized by, and susceptible to, nuclease S1. A report that sequences at or near both boundaries of a triple-stranded DNA structure through Hoogsteen base pairing are reactive to certain chemical reagents (24,25) supports this. It seems that nuclease S1 first recognizes and cleaves (nicks) one of the double strands at the boundary and immediately after the nicking, the same site at the opposite strand is cleaved as predicted from the known mode of action of nuclease S1. It is interesting to note that the cleavage by nuclease S1 occurs at only one of the two boundaries of the triple-strand structure, at the 5' termini of deoxyoligonucleotides (not the 3' termini). It is probable that either the structure of distorted DNA or the mode of association of RecA protein with that region, which should affect the susceptibility to nuclease S1, differs between the two boundaries. In any event, our results suggest that triple-stranded DNA structure can be used as a target for various specific modifications or recognition through enzymatic, chemical or physical means, and one may even create unique sites inside the triple-strand structure for modification by employing deoxyribonucleotides specifically designed for that purpose.
ACKNOWLEDGEMENTS
We thank O. Ohara for their valuable suggestions, and K. Okumura and M. Inada for their help and encouragement. We are particularly indebted to K. Kitamoto and T. Minetoki of Ozeki Shuzo for providing recombinant nuclease S1.
REFERENCES
Roberts,R.J. (1988) Restriction enzymes and their isoschizomers. Nucleic Acids Res., 16, 271–313.
Ando,T. (1966) A nuclease specific for heat-denatured DNA is isolated from a product of Aspergillus oryzae. Biochim. Biophys. Acta, 114, 158–168.
Wei,C.F., Alianell,G.A., Bencen,G.H. and Gray,H.B.,Jr (1983) Isolation and comparison of two molecular species of the BAL 31 nuclease from Alteromonas espejiana with distinct kinetic properties. J. Biol. Chem., 258, 13506–13512.
West,S.C., Cassuto,E. and Howard-Flanders,P. (1981) RecA protein promotes homologous-pairing and strand-exchange reactions between duplex DNA molecules. Proc. Natl Acad. Sci. USA, 78, 2100–2104.
Lindsley,J.E. and Cox,M.M. (1989) Dissociation pathway for RecA nucleoprotein filaments formed on linear duplex DNA. J. Mol. Biol., 205, 695–711.
Umlauf,S.W., Cox,M.M. and Inman,R.B. (1990) Triple-helical DNA pairing intermediates formed by RecA protein. J. Biol. Chem., 265, 16898–16912.
Hsieh,P., Camerini-Otero,C.S. and Camerini-Otero,R.D. (1990) Pairing of homologous DNA sequences by proteins: evidence for three-stranded DNA. Genes Dev., 4, 1951–1963.
Rao,B.J., Dutreix,M. and Radding,C.M. (1991) Stable three-stranded DNA made by RecA protein. Proc. Natl Acad. Sci. USA, 88, 2984–2988.
Muller,B., Burdett,I. and West,S.C. (1992) Unusual stability of recombination intermediates made by Escherichia coli RecA protein. EMBO J., 7, 2685–2693.
Vogt,V.M. (1973) Purification and further properties of single-strand-specific nuclease from Aspergillus oryzae. Eur. J. Biochem., 33, 192–200.
Kroeker,W.D. and Kowalski,D. (1978) Gene-sized pieces produced by digestion of linear duplex DNA with mung bean nuclease. Biochemistry, 17, 3236–3243.
Godson,G.N. (1973) Action of the single-stranded DNA specific nuclease S1 on double-stranded DNA. Biochim. Biophys. Acta, 308, 59–67.
Zhou,X.G. and Gray,H.B.,Jr (1990) Mechanism of exonuclease action of BAL 31 nuclease. Biochim. Biophys. Acta, 1049, 83–91.
Cheng,S., Van Houten,B., Gamper,H.B., Sancar,A. and Hearst,J.E. (1988) Use of psoralen-modified oligonucleotides to trap three-stranded RecA-DNA complexes and repair of these cross-linked complexes by ABC excinuclease. J. Biol. Chem., 263, 15110–1517.
Bianco,P.R. and Weinstock,G.M. (1996) Interaction of the RecA protein of Escherichia coli with single-stranded oligodeoxyribonucleotides. Nucleic Acids Res., 24, 4933–4939.
Golub,E.I., Glazer,P.M., Ward,D.C. and Radding,C.M. (1996) RecA protein-mediated irreversible fixation of an oligodeoxyribonucleotide to specific site in DNA. Mutat. Res., 351, 117–124.
Lee,B.R., Kitamoto,K., Yamada,O. and Kumagai,C. (1995) Cloning, characterization and overproduction of nuclease S1 gene (nucS) from Aspergillus oryzae. Appl. Microbiol. Biotechnol., 44, 425–431.
Griffin,L.C. and Dervan,P.B. (1989) Recognition of thymine adenine base pairs by guanine in a pyrimidine triple helix motif. Science, 245, 967–971.
Beal,P.A. and Dervan,P.B. (1991) Second structural motif for recognition of DNA by oligonucleotide-directed triple-helix formation. Science, 251, 1360–1363.
Strobel,S.A. and Dervan,P.B. (1990) Site-specific cleavage of a yeast chromosome by oligonucleotide-directed triple-helix formation. Science, 249, 73–75.
Koob,M. and Szybalski,W. (1990) Cleaving yeast and Escherichia coli genomes at a single site. Science, 250, 271–273.
Garrison,P.N., Koob,M. and Barnes,LD. (1992) Physical mapping of the Saccharomyces cerevisiae Ap4A phosphorylaseI-encoding gene by the Achilles’ cleavage method. Gene, 111, 131–134.
Strobel,S.A., Doucette-Stamm,L.A., Riba L, Housman,D.E. and Dervan,P.B. (1991) Site-specific cleavage of human chromosome 4 mediated by triple-helix formation. Science, 254, 1639–1642.
Ferrin,L.J. and Camerini-Otero,R.D. (1991) Selective cleavage of human DNA: RecA-assisted restriction endonuclease (RARE) cleavage. Science, 254, 1494–1497.
Hartman,D.A., Kuo,S.R., Broker,T.R., Chow,L.T. and Wells,R.D. (1992) Intermolecular triplex formation distorts the DNA duplex in the regulatory region of human papillomavirus type-11. J. Biol. Chem., 267, 5488–5494.(Yasushi Shigemori and Michio Oishi*)
*To whom correspondence should be addressed. Tel: +81 438 52 3945; Fax: +81 438 52 3946; Email: oishi@kazusa.or.jp
Present address:
Yasushi Shigemori, Biotechnology Division, Aisin Cosmos R & D Co. Ltd, 2-3-9 Kazusa-kamatari, Kisarazu, Chiba 292-0818, Japan
ABSTRACT
A novel procedure to cleave DNA molecules at any desired base sequence is presented. This procedure is based upon our finding that double-stranded DNA molecules at a site where RecA-mediated triple-stranded DNA structure with a complimentary deoxyoligonucleotide is located can be cleaved by a single-strand specific nuclease, such as nuclease S1 or BAL31, between the first base at the 5' termini of the deoxyoligonucleotides and the nearest base proximal to the 5' termini. Accordingly, the sequence as well as the number of the cleavage sites to be cleaved can be custom designed by selecting deoxyoligonucleotides with specific base sequences for triple-stranded DNA formation. The basic characteristics of the cleavage reaction and typical applications of the procedure are presented with actual results, including those which involve cleavage of complex genomic DNA at the very sites one desires.
INTRODUCTION
Although restriction enzymes have been widely used in molecular biological studies, the DNA cleavage sites are limited to specific base sequences, base sequences with palindromic structures (1). Cleavage at any site in DNA molecules, if established, would greatly expand current DNA technologies beyond the boundaries imposed by the limitations in recognizing sequences associated with restriction enzymes. While studying susceptibility of triple-stranded DNA molecules to various nucleases we found that nuclease S1 (2) and BAL31 (3) cleave DNA molecules specifically at the site where a triple-stranded DNA structure (4–9) is located. Nuclease S1 is known to cleave single-stranded DNA (or RNA) endonuclolytically to yield mono- and oligonucleotides with 5' phosphate as the ultimate product (2,10–12). The enzyme also attacks double-stranded DNA molecules at the sites where nicks are present producing blunt-ended double-stranded DNA with 5' phosphate (11,12). BAL31 has similar endonucleolytic activities towards single-stranded DNA but it also attacks double-stranded DNA in a non-discriminate fashion to a much lesser extent than nuclease S1 does (3,13).
Since triple-stranded DNA can be formed readily at any location in double-stranded DNA with a homologous deoxyoligonucleotide in the presence of RecA protein (4–16), we have explored the possibility of applying this finding to specific cleavage (and subsequent cloning) of DNA molecules (fragment between) at the very sites one desires.
MATERIALS AND METHODS
Deoxyoligonucleotides
Deoxyoligonucleotides (HPLC purified grade) were custom synthesized by Sawady Technology, Tokyo, Japan. Unless otherwise specified, no phosphate moiety was present at the 5' termini.
Triple-strand cleavage
The following protocols were employed for triple-strand cleavage except for the cleavage of human genomic DNA (see below). Target DNA (600 ng), homologous 60mer deoxyoligonucleotides and RecA protein (6 μg; Epicenter Technologies, Madison, WI, USA) were mixed and incubated in a reaction mixture (20 μl), which contained ATP-S (2.4 mM, HPLC purified grade; Roche Diagnostics, Mannheim, Germany), magnesium acetate (14 mM) and Tris-acetate (30 mM, pH 7.2), for 30 min at 37°C. For nuclease S1 digestion, the mixture was combined with a reaction mixture (80 μl) containing nuclease S1 (200 units; Takara Shuzo, Shiga, Japan), NaCl (340 mM) and ZnSO4 (1.35 mM), glycerol (6.8%, v/v) and sodium acetate (40 mM, pH 4.6) for 30 min at 55°C. For BAL31 digestion, the mixture was combined with a reaction mixture (80 μl) containing nuclease BAL31 (20 units; Takara Shuzo), NaCl (600 mM), CaCl2 (12 mM), MgCl2 (12 mM), EDTA (1 mM) and Tris–HCl (20 mM, pH 8.0) for 15 min at 37°C. S1 nuclease (or BAL31) and RecA protein were then inactivated by treating the reaction mixture with SDS (0.5%, w/v), Proteinase K (0.1 mg/ml; Roche Diagnostics), EDTA (5 mM) in Tris–HCl buffer (150 mM, pH 8.8) for 30 min at 54°C, followed by phenol–chloroform treatment. After the addition of a one-tenth volume of sodium acetate (3 M), DNA was ethanol precipitated, dissolved in 20 μl of TE buffer (10 mM Tris–HCl, pH 8.0 and 1 mM EDTA) and subjected to agarose (1%) gel electrophoresis or other analyses.
For the cleavage of human genomic DNA, human DNA (50 μg) (Promega, Madison, WI, USA), which had been treated with NotI (New England Biolabs, Beverly, MA, USA), was incubated in a reaction mixture (100 μl) containing deoxyoligonucleotides , RecA protein (50 μg; Epicenter Technologies), ATP-S (4.8 mM; Roche Diagnostics), magnesium acetate (14 mM) and Tris-acetate (30 mM, pH 7.2) for 3 h at 37°C. The sample was diluted with a reaction mixture (300 μl) containing recombinant nuclease S1 (17) (see text; 200 units, kindly provided by Ozeki Shuzo Co., Nishinomiya, Japan), NaCl (340 mM), ZnSO4 (1.35 mM), glycerol (6.8%, v/v) and sodium acetate (40 mM, pH 4.6) incubated for 30 min at 40°C. The reaction was terminated by incubating (54°C for 30 min) in a solution with EDTA (5 mM), SDS (0.5%), Proteinase K (0.1 mg/ml; Roche Diagnostics) and Tris–HCl (150 mM, pH 8.8). After phenol–chloroform treatment, DNA was precipitated by ethanol and dissolved in TE buffer (40 μl). DNA was then electrophoresed on agarose gel (1%).
Determination of the nuclease S1 cleavage site
pBR322 DNA, which had been linearized by NdeI, was incubated with one of the following deoxyoligonucleotides (60mer, HPLC purified grade) terminating their 5' termini with cytosine (oligo C, nucleotides 5–64 of pBR322 DNA), adenine (oligo D, nucleotides 6–65), thymine (oligo E, nucleotides 7–66) or guanine (oligo F, nucleotides 8–67) in the presence of RecA protein. Each sample was subjected to nuclease S1 treatment. After removal of RecA protein and nuclease S1, the samples were treated with either restriction enzyme SspI (New England Biolabs) or EcoRV (New England Biolabs). The 5' terminus of each fragment was labeled with 32P through the exchange reaction by T4 polynucleotide kinase (MEGALABELTM Labeling Kit; Takara Shuzo) using ATP (3000 Ci/mmol; Amersham Pharmacia Biotech, Buckinghamshire, UK), electrophoresed on denaturing polyacrylamide sequence gel (4.5%) and autoradiographed. For determination of the size of each fragment (number of bases), labeled control DNA (M13mp18 using nucleotides 6310–6326 deoxyoligonucleotide as a primer for the dideoxy-mediated chain termination method) was elecrophoresed in the same gel.
RESULTS
Triple-strand cleavage by nuclease S1 and BAL31
As shown in Figure 1A, when pBR322 DNA, which had previously been subjected to triple-stranded DNA formation with a 60mer homologous deoxyoligonucleotide (oligo A, nucleotides 3787–3846 of pBR322 DNA), was incubated with nuclease S1 and electrophoresed, two additional shorter DNA bands with 2.9 and 1.5 kb, which were expected from the cleavage of DNA at or near the triple-strand structure, appeared (lane 1). The shorter DNA bands did not appear without RecA protein for triple-stranded DNA formation (lane 2), without nuclease S1 treatment (lane 3), without both treatments (control, lane 4), or with a non-homologous deoxyoligonucleotide (oligo B, homologous to nucleotides 1210–1269 of M13mp18 DNA) (lane 5). The cleavage efficiency was generally 60–80% and was not increased further with higher nuclease S1 concentrations suggesting that the efficiency of triple-stranded DNA formation (with RecA protein) is a limiting factor for the cleavage efficiency under the condition we employed. Essentially the same results were obtained with BAL31 treatment, but with considerably lesser cleavage efficiency (Fig. 1B, lanes 3 and 4). Because of better efficiency of cleavage, further characterization was carried out with nuclease S1.
Figure 1. Cleavage of DNA molecules carrying a RecA-mediated triple-stranded DNA structure by nuclease S1 and BAL31. (A) Linearized (NdeI) pBR322 was incubated with a 60mer deoxyoligonucleotide (oligo A or oligo B) in the presence of RecA protein, treated with nuclease S1, subjected to agarose (1%) gel electrophoresis and stained with ethidium bromide. (B) The same as (A), but includes the results with BAL31 (lane 4) and control (lane 3). We show molecular size markers (kb) in lane M.
Effect of the chain length of deoxyoligonucleotides
As seen in Figure 2, the cleavage of DNA molecules by nuclease S1 at sites where triple-strand structures are present was observed most prominently for DNA that had been subjected to triple-strand formation using deoxyoligonucleotides with more than 60 nt (lanes 1, 2 and 3 for 100, 80 and 60mers, respectively) and no cleavage was observed with an oligomer of 20 nt (lane 6). It is likely that either a portion of the triple-strand structures with <60mer deoxyoligonucleotides becomes dissociated during incubation with nuclease S1, or triple-strand formation itself is affected by the chain length of the deoxyoligonucleotides, or both. For the experiments described below, we used 60mer deoxyoligonucleotides throughout for the cleavage of triple-strand DNA.
Figure 2. Effect of the chain length of deoxyoligonucleotides. pBR322 DNA, which had been incubated with deoxyoligonucleotides whose chain length ranged from 100 to 20 nt in the presence of RecA protein, was treated with nuclease S1, subjected to agarose gel electrophoresis and stained with ethidium bromide. Lane 1, pBR322 DNA incubated with a 100mer deoxyoligonucleotide (nucleotides 5–104); lane 2, the same as lane 1, but with a 80mer deoxyoligonucleotide (nucleotides 5–84); lane 3, the same as lane 1, but with a 60mer deoxyoligonucleotide (oligo C, nucleotides 5–64); lane 4, the same as lane 1, but with a 40mer deoxyoligonucleotide (nucleotides 5–44); lane 5, the same as lane 1, but with a 30mer deoxyoligonucleotide (nucleotides 5–34); and lane 6, the same as lane 1, but with a 20mer deoxyoligonucleotide (nucleotides 5–24). We show molecular size markers (kb) in lane M.
Determination of cleavage sites
First, pBR322 DNA linearized by the digestion with StyI was subjected to triple-strand formation with one of the four following deoxyoligonucleotides (60mer): oligo C (nucleotides 5–64 with cytosine at nucleotide 5), oligo D (nucleotides 6–65 with adenine at nucleotide 6), oligo E (nucleotides 7–66 with thymine at nucleotide 7) and oligo F (nucleotides 8–67 with guanine at nucleotide 8) (see Fig. 3A) and subsequently to cleavage by nuclease S1. As seen in Figure 3B, in addition to the original DNA (4.4 kb), two shorter DNA fragments (3 and 1.4 kb, respectively), each apparently corresponding to the fragments produced by the cleavage at or near the triple-stranded DNA structures, were produced with almost equal cleavage efficiency with all of these four DNA samples. To determine the precise site of cleavage, pBR322 DNA linearized by NdeI and carrying triple-strand structures with these deoxyoligonucleotides were subjected to similar nuclease S1 cleavage. After removal of RecA protein and nuclease S1, DNA was treated with a restriction enzyme (SspI) whose cutting site (with a blunt end) exists between nucleotides 4170 and 4171 of pBR322 DNA (see Fig. 3A). The products, after labeling the 5' termini of all the products with 32P through the exchange reaction by T4 polynucleotide kinase, were electrophoresed on sequencing gel and autoradiographed. As seen in Figure 3C, in addition to the signals at the top which were derived from labeled larger DNA fragments (presumably the NdeI–SspI fragment, the fragments between the cleaved sites at the triple-strand structures and NdeI, and the uncleaved SspI–NdeI fragment, see Fig. 3A), signals derived from small DNA fragments with 195 (lane 1 for oligo C), 196 (lane 2 for oligo D), 197 (lane 3 for oligo E) and 198 nt (lane 4 for oligo F) in their chain length were observed. Since those chain lengths correspond exactly to those of the fragments between the SspI site and the 5' termini of deoxyoligonucleotides in the triple-strand structure (see Fig. 3A), we have concluded that nuclease S1 cleaves both of the target DNA strands between the first base at the 5' termini of deoxyoligonucleotides in the triple-strand structure and the nearest base proximal to the 5' termini, producing blunt-ended DNA fragments with 5' phosphates. Since the intensity of the signals was not significantly varied among these bands (see also Fig. 3B), the cleavage apparently proceeds regardless of the type of base located at the 5' termini of deoxyoligonucleotides in the triple-strand structure. The presence of a phosphate moiety at the 5' terminus deoxyoligonucleotides had no effect on the cleavage pattern (data not shown). When we electrophoresed the same DNA samples but treated them with EcoRV (instead of SspI), which cleaves the DNA between nucleotides 186 and 187 (see Fig. 3A), we detected minor DNA bands (Fig. 3D) in addition to the major band, suggesting that nuclease S1 gives additional, probably secondary, cuts inside the triple-stranded DNA structure, producing one or a few base shorter cleavage products (see below). A tentative model for the overall triple-strand cleavage reaction by nuclease S1 is presented in Figure 4.
Figure 3. Determination of the nuclease S1 cleavage site. Linearized pBR322 DNA was subjected to triple-strand cleavage with a deoxyoligonucleotide terminating the 5' termini with either cytosine (oligo C), adenine (oligo D), thymine (oligo E) or guanine (oligo F). A portion of the DNA from each sample was electrophoresed on agarose gel and another portion of the DNA, after treating with SspI or EcoRV and labeling the 5' termini with 32P, was electrophoresed on a sequence gel and autoradiographed. (A) Diagrammatic representation of the positions and directions of deoxyoligonucleotides used for triplex cleavage. Nucleotide numbers (nt. numbers) are those registered in GenBank. The distances in the diagram do not necessarily reflect actual distances. (B) Agarose gel electrophoresis patterns of the products. Lane 1, products with oligo C; lane 2, products with oligo D; lane 3, products with oligo E; and lane 4, products with oligo F. In lane M, we show molecular size markers. (C) Autoradigraphic patterns of DNA after digestion with SspI and terminal labeling. (D) Auto radigraphic patterns of DNA after digestion with EcoRV and terminal labeling. (C and D) Lane 1, products with oligo C; lane 2, products with oligo D; lane 3, products with oligo E; and lane 4, products with oligo F. We also show sequence patterns of the control M13mp18 DNA on the left-hand side (lane M) and size of the DNA fragments (nt) on the right-hand side.
Figure 4. A tentative model for the overall triple-strand cleavage reaction by nuclease S1. The triple-strand structure shown in IV is highly speculative and may exist only very transiently.
Use of triple-strand cleavage for isolation (cloning) and generation of deletion of specific DNA sequences
In order to demonstrate the usefulness of our finding (to examine whether this procedure, in fact, can be used for excision and subsequent cloning of a specific DNA fragment between the very sites one desires), we attempted to actually excise and isolate the sequence between nucleotides 1 and 1276 in pBR322 DNA which corresponds to the entire tetracycline resistance gene. pBR322 DNA was subjected to simultaneous cleavages at two triple-strand structures in a single DNA molecule using a pair of 60mer deoxyoligonucleotides with opposite directions (their 5' termini facing each other) to avoid possible minor cleavage in the triple-strand structures (see Fig. 3D). One of the deoxyoligonucleotides (oligo G in Fig. 5A, complimentary to nucleotides 4302–4361) had a 5' terminus corresponding to the nearest nucleotide (nucleotide 4361), just proximal to the first nucleotide (nucleotide 1) of the gene, and the other (oligo H in Fig. 5A, nucleotides 1277–1336) had a 5' terminus corresponding to the nucleotide 1277, just distal to the last base (nucleotide 1276) of the gene (Fig. 5A). Figure 5B shows the cleavage pattern; no deoxyoligonucleotides (lane 1), oligo H alone (lane 2), oligo G alone (lane 3) and oligo G plus H (lane 4). When oligo G and H were simultaneously employed (lane 4), an extra band of 1.3 kb (indicated by an arrow) which corresponds to the entire tetracycline gene (1276 bp) was detected. The DNA band was eluted from the gel and blunt-end-ligated to a vector (pUC19), which is tetracycline sensitive, at a dephosphorylated HincII site. Colonies which presumably carried inserts were first examined for their phenotype (tetracycline resistance). All of the clones with inserts (total 22) were tetracycline resistant. Sequence analysis showed four out of five inserts represented complete sequences of the tetracycline resistance gene (nucleotides 1–1276). In one insert, one nucleotide at nucleotide 1 was missing in the 1276 bp insert. These results demonstrate that by employing a pair of deoxyoligonucleotides in which 5' termini of the deoxyoligonucleotides are facing each other, excision and isolation of DNA sequences from the sites one wants has become possible.
Figure 5. Simultaneous double triple-strand cleavage and cloning of tetracycline resistant gene. Linearized (NdeI) pBR322 DNA was subjected to triple-strand cleavage using oligo G or oligo H, or both. (A) Diagrammatic representation of double triple-strand cleavage in which positions and directions of the deoxyoligonucleotides used are shown. P1, P2, A(TG) and (TG)A indicate promoter 1, promoter 2, the initiation codon and termination codon of the gene, respectively. Distances in the diagram do not necessarily reflect actual distances. (B) Products stained with ethidium bromide after simultaneous double triple-strand cleavage. Arrow indicates the DNA fragment (1.3 kb) corresponding to the excised tetracycline resistant gene.
We were also able to generate deletions (the entire sequence of the tetracycline resistance gene) in open circular pBR322 DNA by employing a pair of deoxyoligonucleotides just inside the 5' and 3' termini of the target sequence (data not shown). In this case, the 5' termini of the deoxyoligonicleotides are directed outwardly and in the opposite direction. Generation of small deletions, shorter than the sum of the deoxynucleotides, was produced by employing pairs of deoxyoligonucleotides in which the 5' termini were facing each other, although the success rate of obtaining deletions exactly as originally intended was slightly <30% among the clones examined (data not shown).
Triple-strand cleavage of complex genomic DNA
We examined whether triple-strand cleavage can be performed directly with highly complex genomic DNA. Human genomic DNA pre-treated with NotI was subjected to simultaneous triple-strand cleavage at a specific gene (the p53 gene). One pair with oligo I (complimentary to nucleotides 553–612) and oligo J (nucleotides 965–1024) was used to excise a block of sequence (351 bp from nucleotides 843 to 949) which includes the exon 1 (107 bp) of the p53 gene, and another pair with oligo K (complimentary to nucleotides 18539–18598) and oligo L (nucleotides 19877–19936) was employed to excise the entire sequence of exon 11 (1278 bp from nucleotides 18599 to 19876) exactly (see Fig. 6A). Since treatment of complex genomic DNA such as human DNA with the nuclease S1 preparation used for the above experiments gave a low level non-specific cut, probably by a small amount of nucleases contaminated in the preparation, we employed nuclease S1 which was derived from a cloned nuclease S1 gene (17) for this experiment. After triple-strand cleavage, the DNA was electrophoresed and Southern hybridized using a radiolabeled p53 probe. As seen in Figure 6B, when the oligo I and J pair was employed, a band (0.3 kb) corresponding to the expected excised sequence (352 bp) was detected (lane 2 indicated by an arrow). No such band was detected in the control experiment in which non-homologous deoxyoligonucleotides (oligo G and H) were employed for the triple-strand cleavage (lane 1). Similar results were obtained with DNA products with another pair of deoxyoligonucleotides (oligo K and L), in which a DNA fragment (1.3 kb) close to the expected size (1278 bp) was observed (lane 4). The results suggest that triple-strand cleavage can be applied to highly complex DNA such as human genomic DNA.
Figure 6. Triple-strand cleavage of sequences specific to human p53 gene. (A) Diagrammatic representation of triple-strand cleavage of the sequences in p53 gene in which positions and directions of the deoxyoligonucleotides used are shown (top, oligo I and J; bottom, oligo K and L). In the diagram, exons and introns are shown as shaded (exons) and bold (introns) lines. Distances in the diagram do not necessarily reflect actual distances. (B) Patterns of Southern blot hybridization. NotI-treated human DNA (50 μg) was incubated with RecA protein and subjected to nuclease S1 treatment as described in Materials and Methods. DNA was electrophoresed on agarose gel (1%) and, after staining with ethidium bromide, subjected to Southern blot hybridization using radiolabeled p53 probes which were randomly primed by dCTP (6000 Ci/mmol; Amersham Pharmacia Biotech) using a kit (BcaBESTTM Labeling Kit; Takara Shuzo) and autoadiographed. Arrows indicate the positions of signals detected by hybridization. (C) Patterns of DNA stained with ethidium bromide after electrophoresis.
DISCUSSION
In this paper, we have presented our findings, which makes it possible to cleave DNA molecules at any known sequence one desires by a relatively simple process. The sequences as well as the number of cleavage sites can be custom designed by selecting specific base sequences. This may be particularly useful when one wants to make a single cut to introduce a specific sequence or to generate a deletion at a desired position in DNA molecules. There have been previous attempts to cleave DNA molecules at targeted sites. DNA molecules can be specifically cleaved by chemical treatment of triple-strand structures formed through Hoogsteen base pairing, although the target base sequences to be cleaved are limited only to polypurines and polypyrimidines (18–20). Another approach was to first protect the sites to be cleaved against methylation either by forming a complex with specific proteins (21,22) or Hoogsteen base pairing (23), and protected (unmethylated) sites were subsequently cleaved by methylation-sensitive restriction enzymes. In these cases, however, the sequences to be cleaved are still limited to those recognized by (methylation-sensitive) restriction enzymes.
Besides the introduction of a single cut or generation of a deletion at desired sites, one of the possible applications of the finding would be for isolation of any sequence of interest from even complex genomic DNA. Nowadays, isolation of a specific sequence can be relatively easily achieved by amplifying the sequence by PCR. On the other hand, there are several cases in which PCR is not satisfactorily applicable. For example, amplification of a long stretch of sequence over several kilobases is still very difficult, though not impossible, by PCR. Introduction of mismatched bases during PCR in these long sequences often poses a problem. Our procedure may offer an alternative way of isolation of specific sequences in such cases.
We have shown that nuclease S1 cleaves both the target DNA strands between the first base at the 5' termini of deoxyoligonucleotides in the triple-strand structure and the nearest base proximal to the 5' termini, thus producing blunt-ended DNA fragments with 5' phosphate. Triple-stranded DNA structure is likely to generate a regional conformational distortion at the boundaries to double-stranded DNA molecules which is recognized by, and susceptible to, nuclease S1. A report that sequences at or near both boundaries of a triple-stranded DNA structure through Hoogsteen base pairing are reactive to certain chemical reagents (24,25) supports this. It seems that nuclease S1 first recognizes and cleaves (nicks) one of the double strands at the boundary and immediately after the nicking, the same site at the opposite strand is cleaved as predicted from the known mode of action of nuclease S1. It is interesting to note that the cleavage by nuclease S1 occurs at only one of the two boundaries of the triple-strand structure, at the 5' termini of deoxyoligonucleotides (not the 3' termini). It is probable that either the structure of distorted DNA or the mode of association of RecA protein with that region, which should affect the susceptibility to nuclease S1, differs between the two boundaries. In any event, our results suggest that triple-stranded DNA structure can be used as a target for various specific modifications or recognition through enzymatic, chemical or physical means, and one may even create unique sites inside the triple-strand structure for modification by employing deoxyribonucleotides specifically designed for that purpose.
ACKNOWLEDGEMENTS
We thank O. Ohara for their valuable suggestions, and K. Okumura and M. Inada for their help and encouragement. We are particularly indebted to K. Kitamoto and T. Minetoki of Ozeki Shuzo for providing recombinant nuclease S1.
REFERENCES
Roberts,R.J. (1988) Restriction enzymes and their isoschizomers. Nucleic Acids Res., 16, 271–313.
Ando,T. (1966) A nuclease specific for heat-denatured DNA is isolated from a product of Aspergillus oryzae. Biochim. Biophys. Acta, 114, 158–168.
Wei,C.F., Alianell,G.A., Bencen,G.H. and Gray,H.B.,Jr (1983) Isolation and comparison of two molecular species of the BAL 31 nuclease from Alteromonas espejiana with distinct kinetic properties. J. Biol. Chem., 258, 13506–13512.
West,S.C., Cassuto,E. and Howard-Flanders,P. (1981) RecA protein promotes homologous-pairing and strand-exchange reactions between duplex DNA molecules. Proc. Natl Acad. Sci. USA, 78, 2100–2104.
Lindsley,J.E. and Cox,M.M. (1989) Dissociation pathway for RecA nucleoprotein filaments formed on linear duplex DNA. J. Mol. Biol., 205, 695–711.
Umlauf,S.W., Cox,M.M. and Inman,R.B. (1990) Triple-helical DNA pairing intermediates formed by RecA protein. J. Biol. Chem., 265, 16898–16912.
Hsieh,P., Camerini-Otero,C.S. and Camerini-Otero,R.D. (1990) Pairing of homologous DNA sequences by proteins: evidence for three-stranded DNA. Genes Dev., 4, 1951–1963.
Rao,B.J., Dutreix,M. and Radding,C.M. (1991) Stable three-stranded DNA made by RecA protein. Proc. Natl Acad. Sci. USA, 88, 2984–2988.
Muller,B., Burdett,I. and West,S.C. (1992) Unusual stability of recombination intermediates made by Escherichia coli RecA protein. EMBO J., 7, 2685–2693.
Vogt,V.M. (1973) Purification and further properties of single-strand-specific nuclease from Aspergillus oryzae. Eur. J. Biochem., 33, 192–200.
Kroeker,W.D. and Kowalski,D. (1978) Gene-sized pieces produced by digestion of linear duplex DNA with mung bean nuclease. Biochemistry, 17, 3236–3243.
Godson,G.N. (1973) Action of the single-stranded DNA specific nuclease S1 on double-stranded DNA. Biochim. Biophys. Acta, 308, 59–67.
Zhou,X.G. and Gray,H.B.,Jr (1990) Mechanism of exonuclease action of BAL 31 nuclease. Biochim. Biophys. Acta, 1049, 83–91.
Cheng,S., Van Houten,B., Gamper,H.B., Sancar,A. and Hearst,J.E. (1988) Use of psoralen-modified oligonucleotides to trap three-stranded RecA-DNA complexes and repair of these cross-linked complexes by ABC excinuclease. J. Biol. Chem., 263, 15110–1517.
Bianco,P.R. and Weinstock,G.M. (1996) Interaction of the RecA protein of Escherichia coli with single-stranded oligodeoxyribonucleotides. Nucleic Acids Res., 24, 4933–4939.
Golub,E.I., Glazer,P.M., Ward,D.C. and Radding,C.M. (1996) RecA protein-mediated irreversible fixation of an oligodeoxyribonucleotide to specific site in DNA. Mutat. Res., 351, 117–124.
Lee,B.R., Kitamoto,K., Yamada,O. and Kumagai,C. (1995) Cloning, characterization and overproduction of nuclease S1 gene (nucS) from Aspergillus oryzae. Appl. Microbiol. Biotechnol., 44, 425–431.
Griffin,L.C. and Dervan,P.B. (1989) Recognition of thymine adenine base pairs by guanine in a pyrimidine triple helix motif. Science, 245, 967–971.
Beal,P.A. and Dervan,P.B. (1991) Second structural motif for recognition of DNA by oligonucleotide-directed triple-helix formation. Science, 251, 1360–1363.
Strobel,S.A. and Dervan,P.B. (1990) Site-specific cleavage of a yeast chromosome by oligonucleotide-directed triple-helix formation. Science, 249, 73–75.
Koob,M. and Szybalski,W. (1990) Cleaving yeast and Escherichia coli genomes at a single site. Science, 250, 271–273.
Garrison,P.N., Koob,M. and Barnes,LD. (1992) Physical mapping of the Saccharomyces cerevisiae Ap4A phosphorylaseI-encoding gene by the Achilles’ cleavage method. Gene, 111, 131–134.
Strobel,S.A., Doucette-Stamm,L.A., Riba L, Housman,D.E. and Dervan,P.B. (1991) Site-specific cleavage of human chromosome 4 mediated by triple-helix formation. Science, 254, 1639–1642.
Ferrin,L.J. and Camerini-Otero,R.D. (1991) Selective cleavage of human DNA: RecA-assisted restriction endonuclease (RARE) cleavage. Science, 254, 1494–1497.
Hartman,D.A., Kuo,S.R., Broker,T.R., Chow,L.T. and Wells,R.D. (1992) Intermolecular triplex formation distorts the DNA duplex in the regulatory region of human papillomavirus type-11. J. Biol. Chem., 267, 5488–5494.(Yasushi Shigemori and Michio Oishi*)