A Single Mutation, RecBD1080A, Eliminates RecA Protein Loading but Not Chi Recognition by RecBCD Enzyme
1999; Elsevier BV; Volume: 274; Issue: 38 Linguagem: Inglês
10.1074/jbc.274.38.27139
ISSN1083-351X
AutoresDaniel G. Anderson, Jason Churchill, Stephen C. Kowalczykowski,
Tópico(s)Nuclear Structure and Function
ResumoHomologous recombination and double-stranded DNA break repair in Escherichia coli are initiated by the multifunctional RecBCD enzyme. After binding to a double-stranded DNA end, the RecBCD enzyme unwinds and degrades the DNA processively. This processing is regulated by the recombination hot spot, Chi (χ: 5′-GCTGGTGG-3′), which induces a switch in the polarity of DNA degradation and activates RecBCD enzyme to coordinate the loading of the DNA strand exchange protein, RecA, onto the single-stranded DNA products of unwinding. Recently, a single mutation in RecB, Asp-1080 → Ala, was shown to create an enzyme (RecBD1080ACD) that is a processive helicase but not a nuclease. Here we show that the RecBD1080ACD enzyme is also unable to coordinate the loading of the RecA protein, regardless of whether χ sites are present in the DNA. However, the RecBD1080ACD enzyme does respond to χ sites by inactivating in a χ-dependent manner. These data define a locus of the RecBCD enzyme that is essential not only for nuclease function but also for the coordination of RecA protein loading. Homologous recombination and double-stranded DNA break repair in Escherichia coli are initiated by the multifunctional RecBCD enzyme. After binding to a double-stranded DNA end, the RecBCD enzyme unwinds and degrades the DNA processively. This processing is regulated by the recombination hot spot, Chi (χ: 5′-GCTGGTGG-3′), which induces a switch in the polarity of DNA degradation and activates RecBCD enzyme to coordinate the loading of the DNA strand exchange protein, RecA, onto the single-stranded DNA products of unwinding. Recently, a single mutation in RecB, Asp-1080 → Ala, was shown to create an enzyme (RecBD1080ACD) that is a processive helicase but not a nuclease. Here we show that the RecBD1080ACD enzyme is also unable to coordinate the loading of the RecA protein, regardless of whether χ sites are present in the DNA. However, the RecBD1080ACD enzyme does respond to χ sites by inactivating in a χ-dependent manner. These data define a locus of the RecBCD enzyme that is essential not only for nuclease function but also for the coordination of RecA protein loading. double-stranded DNA single-stranded DNA E. coli single-stranded DNA-binding protein non-χ-containing χ-containing adenosine 5′-O-(thiotriphosphate) In Escherichia coli, the early steps of homologous recombination and dsDNA1break repair are catalyzed by the RecBCD enzyme and RecA protein (1Eggleston A.K. West S.C. Curr. Biol. 1997; 7: 745-749Abstract Full Text Full Text PDF PubMed Google Scholar). The RecBCD enzyme is heterotrimeric, being composed of the products from the recB, recC, and recD genes (for review see Ref. 2Kowalczykowski S.C. Dixon D.A. Eggleston A.K. Lauder S.D. Rehrauer W.M. Microbiol. Rev. 1994; 58: 401-465Crossref PubMed Google Scholar). The importance of these two enzymes is clearly illustrated by the behavior of their mutants; deletion of either the recB or recC gene reduces conjugal recombination proficiency by 100- to 1000-fold (3Howard-Flanders P. Theriot L. Genetics. 1966; 53: 1137-1150Crossref PubMed Google Scholar, 4Emmerson P.T. Genetics. 1968; 60: 19-30Crossref PubMed Google Scholar), whereas a recA null mutation reduces it approximately 100,000-fold (5Clark A.J. Margulies A.D. Proc. Natl. Acad. Sci. U. S. A. 1965; 53: 451-459Crossref PubMed Scopus (462) Google Scholar). The central step in homologous recombination is catalyzed by RecA protein, which binds cooperatively to DNA and then promotes pairing and exchange between homologous DNA molecules (for review see Ref. 6Kowalczykowski S.C. Eggleston A.K. Annu. Rev. Biochem. 1994; 63: 991-1043Crossref PubMed Scopus (279) Google Scholar). One requirement for this step, however, is that the donor DNA must possess a region of single-stranded character, because RecA protein binds poorly to dsDNA at physiological conditions. Initial binding of RecA protein is random, but subsequent cooperative binding extends the RecA nucleoprotein filament in the 5′ → 3′ direction (7Register III, J.C. Griffith J. J. Biol. Chem. 1985; 260: 12308-12312Abstract Full Text PDF PubMed Google Scholar). A result of this polar extension of the RecA nucleoprotein filament is that 3′-ssDNA ends are more likely to be coated with RecA protein than 5′-ssDNA ends and, therefore, are more recombinagenic (8Konforti B.B. Davis R.W. J. Biol. Chem. 1990; 265: 6916-6920Abstract Full Text PDF PubMed Google Scholar). The processing of a dsDNA break into a form suitable for RecA protein binding is catalyzed by the RecBCD enzyme. The RecBCD enzyme is both a helicase and a nuclease, and it initiates recombinational repair from dsDNA ends. After binding to an end, RecBCD enzyme processively unwinds and degrades the DNA, unwinding up to 30 kilobase pairs per binding event (9Roman L.J. Eggleston A.K. Kowalczykowski S.C. J. Biol. Chem. 1992; 267: 4207-4214Abstract Full Text PDF PubMed Google Scholar). Degradation of the DNA is asymmetric, with the 3′-terminal strand being degraded much more vigorously than the 5′-terminal strand (10Dixon D.A. Kowalczykowski S.C. Cell. 1991; 66: 361-371Abstract Full Text PDF PubMed Scopus (121) Google Scholar, 11Dixon D.A. Kowalczykowski S.C. Cell. 1993; 73: 87-96Abstract Full Text PDF PubMed Scopus (185) Google Scholar). DNA processing by the RecBCD enzyme is regulated by an eight-base DNA element, Chi (χ; 5′-GCTGGTGG-3′) (12Lam S.T. Stahl M.M. McMilin K.D. Stahl F.W. Genetics. 1974; 77: 425-433Crossref PubMed Google Scholar, 13Stahl F.W. Crasemann J.M. Stahl M.M. J. Mol. Biol. 1975; 94: 203-212Crossref PubMed Scopus (77) Google Scholar, 14Smith G.R. Kunes S.M. Schultz D.W. Taylor A. Triman K.L. Cell. 1981; 24: 429-436Abstract Full Text PDF PubMed Scopus (212) Google Scholar, 15Bianco P.R. Kowalczykowski S.C. Proc. Natl. Acad. Sci. U. S. A. 1997; 94: 6706-6711Crossref PubMed Scopus (69) Google Scholar). χ was originally identified in vivo as a recombination hot spot that stimulates homologous recombination about 10-fold in its vicinity.In vitro analysis revealed that χ elicits this response by regulating a number of RecBCD enzyme activities. When a translocating RecBCD molecule recognizes a χ site, the enzyme pauses, and DNA degradation on the 3′-terminal strand is attenuated (10Dixon D.A. Kowalczykowski S.C. Cell. 1991; 66: 361-371Abstract Full Text PDF PubMed Scopus (121) Google Scholar). The enzyme then continues unwinding the DNA but now degrades the 5′-terminal strand preferentially (16Anderson D.G. Churchill J.J. Kowalczykowski S.C. Genes Cells. 1997; 2: 117-128Crossref PubMed Scopus (50) Google Scholar, 17Anderson D.G. Kowalczykowski S.C. Genes Dev. 1997; 11: 571-581Crossref PubMed Scopus (151) Google Scholar) (Fig. 1). On DNA molecules longer than the processive translocation length, processing by RecBCD enzyme creates a long, 3′ ssDNA overhang, which is both a common early intermediate in homologous recombination and an ideal substrate for RecA protein. In addition to regulating the nuclease properties of the RecBCD enzyme, χ also induces the RecBCD enzyme to coordinate the loading of RecA protein onto the ssDNA downstream of χ (18Anderson D.G. Kowalczykowski S.C. Cell. 1997; 90: 77-86Abstract Full Text Full Text PDF PubMed Scopus (248) Google Scholar). The initial binding of RecA protein to ssDNA is slow (19Chabbert M. Cazenave C. Helene C. Biochemistry. 1987; 26: 2218-2225Crossref PubMed Scopus (54) Google Scholar), and RecA protein cannot compete efficiently with other ssDNA-binding proteins (20Kowalczykowski S.C. Krupp R.A. J. Mol. Biol. 1987; 193: 97-113Crossref PubMed Scopus (198) Google Scholar, 21Kowalczykowski S.C. Clow J.C. Somani R. Varghese A. J. Mol. Biol. 1987; 193: 81-95Crossref PubMed Scopus (110) Google Scholar). Facilitation of RecA protein binding to ssDNA by RecBCD enzyme overcomes this inhibition (18Anderson D.G. Kowalczykowski S.C. Cell. 1997; 90: 77-86Abstract Full Text Full Text PDF PubMed Scopus (248) Google Scholar). Reconstitution of in vitro recombination reactions using RecA protein, RecBCD enzyme and single-stranded DNA binding (SSB) protein established that these proteins catalyze efficient pairing between χ-containing dsDNA and homologous supercoiled DNA, forming a recombination intermediate known as a D-loop (10Dixon D.A. Kowalczykowski S.C. Cell. 1991; 66: 361-371Abstract Full Text PDF PubMed Scopus (121) Google Scholar). The domains responsible for the individual activities of the RecBCD enzyme are just starting to be elucidated. The RecBC enzyme (without the RecD subunit) is a very processive helicase but has no significant nuclease activity (16Anderson D.G. Churchill J.J. Kowalczykowski S.C. Genes Cells. 1997; 2: 117-128Crossref PubMed Scopus (50) Google Scholar, 22Masterson C. Boehmer P.E. McDonald F. Chaudhuri S. Hickson I.D. Emmerson P.T. J. Biol. Chem. 1992; 267: 13564-13572Abstract Full Text PDF PubMed Google Scholar, 23Korangy F. Julin D.A. Biochemistry. 1993; 32: 4873-4880Crossref PubMed Scopus (53) Google Scholar). Recently, it was shown that the RecBC enzyme also coordinates the loading of RecA protein and that this loading is independent of χ. Instead, RecBC enzyme constitutively loads RecA protein onto the DNA strand that terminates 3′ at the dsDNA end at which RecBC enzyme entered (24Churchill J.C. Anderson D.G. Kowalczykowski S.C. Genes Dev. 1999; 13: 901-911Crossref PubMed Scopus (108) Google Scholar). The fact that RecBC enzyme has little nuclease activity indicates that the RecD subunit plays an essential role in DNA degradation. However, analysis of mutations in the RecB subunit establishes that the C-terminal domain of this enzyme is absolutely essential for all nuclease activities of the RecBCD enzyme (25Yu M. Souaya J. Julin D.A. Proc. Natl. Acad. Sci. U. S. A. 1998; 95: 981-986Crossref PubMed Scopus (95) Google Scholar, 26Yu M. Souaya J. Julin D.A. J. Mol. Biol. 1998; 283: 797-808Crossref PubMed Scopus (79) Google Scholar). Furthermore, fusion of the C-terminal domain of the RecB subunit with the DNA binding domain of T4 phage gene 32 protein creates a nonspecific nuclease, showing that the C terminus of RecB subunit is indeed a nuclease. Finally, a single point mutation in the putative Mg2+binding site of the RecB subunit, Asp-1080 → Ala, creates a holoenzyme (RecBD1080ACD) that behaves much like the RecBC enzyme; it is a processive helicase with no measurable nuclease activity. Here we show that despite being an efficient helicase, the RecBD1080ACD enzyme is unable to load RecA protein. This inability to load RecA protein is independent of whether the processed DNA contains χ sites. Although the RecBD1080ACD enzyme does not load RecA protein in response to χ, it can still recognize χ: χ-containing DNA is processed at a slower rate than DNA without χ. These data provide the first insight into the domain that is responsible for either 1) transmitting the χ recognition event into the enzymatic alterations that are necessary for proper RecBCD enzyme function or 2) RecA protein loading. RecBC enzyme was purified as described (24Churchill J.C. Anderson D.G. Kowalczykowski S.C. Genes Dev. 1999; 13: 901-911Crossref PubMed Scopus (108) Google Scholar). RecBC enzyme concentration was determined using an extinction coefficient of 3.6 × 105m−1cm−1 at 280 nm, derived by adding the respective extinction coefficients for the individual subunits (23Korangy F. Julin D.A. Biochemistry. 1993; 32: 4873-4880Crossref PubMed Scopus (53) Google Scholar). No contaminating protein bands were detected when 1 μg of protein was loaded on an SDS-polyacrylamide gel and stained with Coomassie Brilliant Blue. RecBD1080ACD enzyme was a gift from D. A. Julin and M. Yu (University of Maryland) (26Yu M. Souaya J. Julin D.A. J. Mol. Biol. 1998; 283: 797-808Crossref PubMed Scopus (79) Google Scholar). E. coli SSB protein was isolated from strain RLM727 and purified according to LeBowitz (27LeBowitz J. Biochemical Mechanism of Strand Initiation in Bacteriophage Lambda DNA Replication.Ph.D. thesis. The Johns Hopkins University, 1985Google Scholar). Protein concentration was determined using an extinction coefficient of 3.0 × 104 at 280 nm. RecA protein was purified using a procedure based on spermidine precipitation 2S. C. Kowalczykowski, unpublished data. (28Griffith J. Shores C.G. Biochemistry. 1985; 24: 158-162Crossref PubMed Scopus (80) Google Scholar). Protein concentration was determined using an extinction coefficient of 2.7 × 104m−1cm−1 at 280 nm. All restriction endonucleases and DNA-modifying enzymes were purchased from New England BioLabs. The enzymes were used according to Sambrook et al. (29Sambrook J. Fritsch E.F. Maniatis T. Molecular Cloning: A Laboratory Manual, Second Edition. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York1989Google Scholar) or as indicated by the vendor. The plasmids pBR322 χo (wild type) and pBR322 χ+F225 (14Smith G.R. Kunes S.M. Schultz D.W. Taylor A. Triman K.L. Cell. 1981; 24: 429-436Abstract Full Text PDF PubMed Scopus (212) Google Scholar) were prepared from strains S819 and S818, respectively, provided by G. R. Smith and A. F. Taylor. Plasmid pBR322 χ3F,3H was created by ligation of the oligonucleotide linker 5′-ATCTAGA CCACCAGC CAGCGCGTGT CCACCAGC TCAGCATCGA CCACCAGC TCGAGTGCA-3′ into the Pst I-Ase I site of pBR322 χo (Chi sites are shown in bold), followed by insertion of a linker 5′-GTCATTAAT GCTGGTGG GCGCAGACTC GCTGGTGG TCACATGGCG GCTGGTGG CTGCAG-3′ into the second (position 1480) Ppu MI site of the plasmid. All plasmid DNAs were purified by cesium chloride density gradient centrifugation (29Sambrook J. Fritsch E.F. Maniatis T. Molecular Cloning: A Laboratory Manual, Second Edition. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York1989Google Scholar). The molar concentration of the dsDNA in nucleotides was determined using an extinction coefficient of 6290m−1 cm−1 at 260 nm. Plasmid DNA was linearized with Nde I and radioactively labeled at the 5′ end by sequential reactions with shrimp alkaline phosphatase followed by T4 polynucleotide kinase and [γ-32P]ATP (NEN Life Science Products) using methods given by the vendor or by Sambrook et al. (29Sambrook J. Fritsch E.F. Maniatis T. Molecular Cloning: A Laboratory Manual, Second Edition. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York1989Google Scholar). The coupled RecA-RecBCD reactions were conducted in the presence of 25 mm Tris acetate (pH 7.5), 8 mm magnesium acetate, 5 mm ATP, 1 mm dithiothreitol, 1 mm phosphoenolpyruvate, 4 units/ml pyruvate kinase, 40 μm (nucleotides) 5′ end-labeled Nde I-linearized dsDNA, 80 μm(nucleotides) supercoiled DNA, 20 μm RecA protein, 8 μm SSB protein, 2.25 nm total RecBD1080ACD enzyme (corresponding to 0.25 RecBD1080ACD enzyme molecules/linear dsDNA end), or 24.6 nm total RecBC enzyme (20% active, corresponding to 0.53 functional RecBC enzyme molecules/linear dsDNA end) (10Dixon D.A. Kowalczykowski S.C. Cell. 1991; 66: 361-371Abstract Full Text PDF PubMed Scopus (121) Google Scholar, 18Anderson D.G. Kowalczykowski S.C. Cell. 1997; 90: 77-86Abstract Full Text Full Text PDF PubMed Scopus (248) Google Scholar). RecBD1080ACD and RecBC enzyme concentrations were chosen to provide approximately equal amounts of helicase units. Assays were performed at 37 °C. Exonuclease protection assays were initiated by the addition of RecBC or RecBD1080ACD enzyme after preincubation of all standard components except supercoiled DNA for 2 min. After 3 min, a mixture of poly(dT) and ATPγS was added to a final concentration of 200 μm (nucleotides) and 5 mm, respectively. After 1 min of incubation, exonuclease I was added to a concentration of 100 units/ml (18Anderson D.G. Kowalczykowski S.C. Cell. 1997; 90: 77-86Abstract Full Text Full Text PDF PubMed Scopus (248) Google Scholar, 24Churchill J.C. Anderson D.G. Kowalczykowski S.C. Genes Dev. 1999; 13: 901-911Crossref PubMed Scopus (108) Google Scholar). Unwinding reactions were performed in the presence of 25 mmTris acetate (pH 7.5), 1 or 8 mm magnesium acetate, 5 mm ATP, 1 mm dithiothreitol, 1 mmphosphoenolpyruvate, 10 μm (nucleotides) linear (2.3 nm dsDNA ends) 5′ end-labeled dsDNA, and 2 μmSSB protein. Reactions were preincubated for 2 min at 37 °C followed by addition of RecBCD (0.046 nm) or RecBD1080ACD (0.23 nm) enzyme. Assays were performed at 37 °C. Aliquots of the reaction mixture (20 μl) were taken at the indicated time points and added to 20 μl of stop buffer (0.125 m EDTA, 2.5% SDS, 10% Ficoll, 0.125% bromphenol blue, and 0.125% xylene cylanol) to halt the reaction and to deproteinize the sample. This was followed by the addition of 1.5 μl of 600 units/ml proteinase K and incubation at 37 °C for 5 min. Samples were electrophoresed in 1% agarose gels for approximately 15 h at 1.4 V/cm in TAE (40 mm Tris acetate (pH 8.0), 2 mm EDTA). The gels were dried onto DE-81 paper (Whatman) and analyzed on a Molecular Dynamics Storm 840 PhosphorImager using Image-QuaNT software. Since RecBC enzyme is a helicase with no significant nuclease activity, processing of linear dsDNA by the RecBC enzyme produces two full-length ssDNA molecules. For convenience, we refer to the strand of DNA terminating 3′ at the entry point of RecBC or RecBCD enzyme as the "top-strand" and the complementary strand of DNA as the "bottom-strand" (Fig.1) (17Anderson D.G. Kowalczykowski S.C. Genes Dev. 1997; 11: 571-581Crossref PubMed Scopus (151) Google Scholar). Recently it was shown that the RecBC enzyme coordinates the loading of RecA protein onto the top-strand of DNA during unwinding (24Churchill J.C. Anderson D.G. Kowalczykowski S.C. Genes Dev. 1999; 13: 901-911Crossref PubMed Scopus (108) Google Scholar). This facilitated loading results in the efficient production of the RecA nucleoprotein complex, which mediates pairing between the top-strand DNA and a homologous supercoiled DNA to form joint molecules (24Churchill J.C. Anderson D.G. Kowalczykowski S.C. Genes Dev. 1999; 13: 901-911Crossref PubMed Scopus (108) Google Scholar). Like RecBC enzyme, RecBD1080ACD enzyme possesses helicase activity but no significant nuclease activity (26Yu M. Souaya J. Julin D.A. J. Mol. Biol. 1998; 283: 797-808Crossref PubMed Scopus (79) Google Scholar). Furthermore, χ does not induce nuclease activity in either enzyme 3J. C. Churchill and S. C. Kowalczykowski, unpublished observations. (26Yu M. Souaya J. Julin D.A. J. Mol. Biol. 1998; 283: 797-808Crossref PubMed Scopus (79) Google Scholar). Thus, processing of linear DNA by either RecBD1080ACD or RecBC enzyme produces only full-length ssDNA. In Fig. 2, we tested the ability of RecBD1080ACD enzyme to process linear dsDNA into substrates suitable for joint molecule formation by RecA protein. Linear pBR322 χ+F (which contains a χ site) and homologous supercoiled DNA were incubated with RecA protein and SSB protein, and then either RecBD1080ACD enzyme or RecBC enzyme was added to start the reaction. After 2 min, the dsDNA was almost completely unwound by both RecBD1080ACD enzyme and RecBC enzyme (Fig.2). However, only 1% of the ssDNA produced by RecBD1080ACD enzyme was incorporated into a joint molecule. In contrast, 36% of the ssDNA produced by RecBC enzyme was incorporated into joint molecules after 2 min. In the next section, we show that this dramatic difference reflects an inability of RecBD1080ACD enzyme to load RecA protein. To test whether RecBD1080ACD enzyme can load RecA protein onto the ssDNA products of unwinding, we examined the sensitivity of these unwinding products to exonuclease I degradation (18Anderson D.G. Kowalczykowski S.C. Cell. 1997; 90: 77-86Abstract Full Text Full Text PDF PubMed Scopus (248) Google Scholar, 24Churchill J.C. Anderson D.G. Kowalczykowski S.C. Genes Dev. 1999; 13: 901-911Crossref PubMed Scopus (108) Google Scholar). If the ssDNA products were coated with SSB protein, then they were sensitive to the 3′ → 5′ exonuclease activity of exonuclease I (30Molineux I.J. Gefter M.L. J. Mol. Biol. 1975; 98: 811-825Crossref PubMed Scopus (73) Google Scholar). However, if instead RecA protein was loaded onto the ssDNA, then they were resistant to exonuclease I after stabilization with ATPγS (18Anderson D.G. Kowalczykowski S.C. Cell. 1997; 90: 77-86Abstract Full Text Full Text PDF PubMed Scopus (248) Google Scholar). In this assay, the linear dsDNA was 5′ end-labeled and incubated with RecA protein and SSB protein. Either RecBD1080ACD enzyme or RecBC enzyme was added and incubated for 3 min. Last, a mixture of excess poly(dT) and ATPγS was added. The addition of the nonhydrolyzable ATP analog, ATPγS, induced a high affinity DNA-binding state in RecA protein (31Menetski J.P. Varghese A. Kowalczykowski S.C. Biochemistry. 1988; 27: 1205-1212Crossref PubMed Scopus (69) Google Scholar). This stabilized any bound RecA protein, whereas the addition of excess poly(dT) sequestered the free RecA and SSB protein. After another minute of incubation, the presence of RecA or SSB protein on the 3′ ends of the ssDNA was detected by either protection or enhancement of exonuclease I degradation, respectively. Because wild-type RecBCD enzyme requires activation by χ to load RecA protein, we compared the RecA-loading properties of RecBD1080ACD enzyme with DNA either devoid of or containing χ (Fig. 3). RecA loading was examined using Nde I-linearized pBR322 χo (which has no χ sites; Fig. 3 A) and Nde I-linearized pBR322 χ3F,3H (which has 6 χ sites; Fig. 3 B). When χo DNA is unwound by RecBD1080ACD enzyme in the presence of RecA and SSB proteins, only 2% of the resultant ssDNA was resistant to exonuclease I degradation (Fig. 3 A, 10 min time point). In contrast, 48% of the ssDNA produced by RecBC enzyme is protected from exonuclease I. This pattern is unaffected by the presence of χ in the dsDNA; 6% of the ssDNA was protected when unwound by RecBD1080ACD enzyme, and 44% of the ssDNA was protected when unwound by RecBC enzyme. These data show that, as expected, RecBC enzyme loads RecA protein only onto the top-strand, since approximately one-half of the ssDNA was protected from exonuclease I digestion (24Churchill J.C. Anderson D.G. Kowalczykowski S.C. Genes Dev. 1999; 13: 901-911Crossref PubMed Scopus (108) Google Scholar). In contrast, the mutant RecBD1080ACD enzyme was completely defective in RecA loading, as defined by this 3′ end protection assay. Since the χ sites were not at the ends of the DNA tested, it was possible that RecBD1080ACD enzyme might have started loading RecA protein only after it reached a χ site. Had this been the case, cooperative binding would have rapidly extended the RecA nucleoprotein filament to the 3′ end of the top strand, resulting in exonuclease protection of the entire top-strand ssDNA molecule. Instead, no protection of either strand was observed. Alternatively, if RecA protein was loaded at χ, but extension of the nucleoprotein filament was inhibited for some unknown reason, exonuclease I degradation of the region of ssDNA upstream (3′) of χ would have produced a ssDNA fragment of the same size as the top-strand, downstream χ-specific fragment (see Fig. 1) (17Anderson D.G. Kowalczykowski S.C. Genes Dev. 1997; 11: 571-581Crossref PubMed Scopus (151) Google Scholar). No fragments of this size were observed (Fig. 3 B). Thus, we conclude that the RecBD1080ACD enzyme cannot load RecA protein regardless of the presence of χ. In Fig. 3, we showed that the RecBD1080ACD enzyme is incapable of loading RecA protein, even when the DNA contains χ sites. One possible explanation for this failing is that χ recognition is required for activation of RecA loading, and the RecBD1080ACD enzyme cannot recognize χ. The traditional way to examine χ recognition in vitro is to examine the production of χ-specific ssDNA fragments. However, since the RecBD1080ACD enzyme lacks nuclease activity, it does not produce χ-specific fragments (26Yu M. Souaya J. Julin D.A. J. Mol. Biol. 1998; 283: 797-808Crossref PubMed Scopus (79) Google Scholar), (Fig. 3 B). Nevertheless, in the absence of χ-specific fragment formation, it is still possible to detect χ recognition using the phenomenon of χ-dependent inactivation. Under conditions of low free Mg2+, wild-type RecBCD enzyme is inactivated after encountering a χ site while translocating through dsDNA (32Dixon D.A. Churchill J.J. Kowalczykowski S.C. Proc. Natl. Acad. Sci. U. S. A. 1994; 91: 2980-2984Crossref PubMed Scopus (60) Google Scholar). Therefore, the rate and extent for unwinding of χ-containing DNA by the RecBD1080ACD enzyme was examined at low free Mg2+ concentrations. The concentration of free Mg2+ was controlled by varying the relative concentrations of Mg2+ and ATP (33Eggleston A.K. Kowalczykowski S.C. J. Mol. Biol. 1993; 231: 605-620Crossref PubMed Scopus (31) Google Scholar). The unwinding of 5′ end-labeled Nde I-linearized pBR322 χo (which has no χ sites) was compared with unwinding of 5′ end-labeled Nde I-linearized pBR322 χ3F,3H (which has 6 χ sites). We observed that at all protein concentrations, the initial rates were approximately equal regardless of the presence of χ (Fig.4 A). However, the rate of unwinding of χ-containing DNA slowed later in the reaction. This difference was not as dramatic as that seen for wild-type RecBCD enzyme (Fig. 4 B), which was completely inactivated after encountering χ sequences during unwinding. Since the RecA-loading properties of the RecBD1080ACD enzyme were tested at a "high" (8 mm) concentration of magnesium ion, the inactivation reactions were repeated in the presence of 8 mm magnesium (Fig. 5). As expected from previous work (32Dixon D.A. Churchill J.J. Kowalczykowski S.C. Proc. Natl. Acad. Sci. U. S. A. 1994; 91: 2980-2984Crossref PubMed Scopus (60) Google Scholar), wild-type RecBCD enzyme unwound both χ and non-χ-containing DNA completely, although the χ-containing DNA was unwound slightly slower. Surprisingly, the RecBD1080ACD enzyme was completely inactivated by the processing of χ-containing DNA. This reaction was repeated at several different protein concentrations with the same results; unwinding of linear dsDNA containing χ inactivates the RecBD1080ACD enzyme at high magnesium acetate concentrations (data not shown). These data show that the RecBD1080ACD enzyme can recognize χ, although this recognition leads to inactivation at high magnesium ion concentrations. The mechanism by which RecA protein is loaded onto ssDNA by RecBCD enzyme remains unclear. However, our analysis of the RecBD1080ACD enzyme reveals an important role for the RecB subunit in the mediation of RecA protein loading. We show that unlike the RecBC enzyme, which is also a processive helicase with no significant nuclease activity, the RecBD1080ACD enzyme cannot promote efficient D-loop formation in reconstituted recombination reactions (Fig. 2). Despite the absence of nuclease activity and the presence of helicase function, the RecBD1080ACD enzyme does not stimulate joint molecule formation even if χ is present in the DNA. This inability to promote D-loop formation results from an inability of RecBD1080ACD enzyme to load RecA protein onto the unwound ssDNA (Fig. 3). In contrast, RecBC protein promotes the loading of RecA protein onto about 50% of the resultant ssDNA, which is consistent with its ability to load RecA protein asymmetrically onto the top-strand of DNA during unwinding. Since χ recognition is required for RecA loading by wild-type RecBCD enzyme, one possible explanation for the inability of the RecBD1080ACD enzyme to load RecA protein is that it simply cannot recognize χ. To test this possibility, we examined whether RecBD1080ACD enzyme, like RecBCD enzyme, inactivates in response to χ. Using the χ-dependent inactivation assay, we established that unwinding by RecBD1080ACD enzyme is slowed in the presence of χ (Fig. 4). This effect is not as dramatic as that seen with wild-type RecBCD enzyme, which is completely inactivated by χ. Unexpectedly, the RecBD1080ACD enzyme is inactivated by χ in the presence of high Mg2+ (Fig.5). In contrast, processing of DNA by wild-type RecBCD enzyme was only slightly slower when the DNA contained χ sites. These results establish that RecBD1080ACD enzyme can indeed recognize χ, although its response to χ is somewhat different than for the wild-type enzyme. Inactivation of RecBCD enzyme in response to χ at low concentrations of Mg2+ does not cause the enzyme to dissociate from the DNA at χ (32Dixon D.A. Churchill J.J. Kowalczykowski S.C. Proc. Natl. Acad. Sci. U. S. A. 1994; 91: 2980-2984Crossref PubMed Scopus (60) Google Scholar). Rather, RecBCD enzyme continues to process the DNA until it reaches the end of the DNA molecule; however, this χ-modified enzyme is unable to re-initiate unwinding on another DNA molecule. Examination of RecBD1080ACD enzyme products after exonuclease treatment (Fig. 3) suggests that RecBD1080ACD enzyme does not dissociate at χ either. If RecBD1080ACD enzyme had dissociated at χ, a Y-shaped, partially unwound dsDNA structure would have formed. The subsequent degradation by exonuclease I would have resected the 3′-terminating ssDNA strand, thereby producing dsDNA with a 5′ overhanging end. DNA of this type would be resistant to both RecBCD enzyme and exonuclease I (17Anderson D.G. Kowalczykowski S.C. Genes Dev. 1997; 11: 571-581Crossref PubMed Scopus (151) Google Scholar), and hence, stable, but no intermediates of this type were observed. Thus, in this sense, the inactivation of RecBD1080ACD enzyme observed at high Mg2+ is similar to that observed by RecBCD enzyme at low Mg2+; recognition of χ does not appear to cause dissociation of the RecBD1080ACD enzyme from the DNA. Rather, χ-induced inactivation prevents the enzyme from starting a new round of DNA processing. As previously mentioned, Asp-1080 in RecB protein was originally characterized as an essential component of the nuclease domain (26Yu M. Souaya J. Julin D.A. J. Mol. Biol. 1998; 283: 797-808Crossref PubMed Scopus (79) Google Scholar). Comparison with other nucleases suggested that it is required for binding of Mg2+ in the active site. Weaker binding of Mg2+ by RecBD1080ACD enzyme can explain the χ-dependent inactivation observed at an unexpectedly high concentration of Mg2+ (Fig. 5). More importantly, however, our data illuminate a previously unknown additional role for this domain by establishing that it is also required for RecA protein loading. Even though RecBD1080ACD enzyme can recognize χ by at least one measure and can continue unwinding past the χ site, it is unable to load RecA protein onto the resultant ssDNA. We believe there are two molecular interpretations of this data. 1) Asp-1080 in RecB is required to transmit the χ-recognition signal to whatever element is responsible for RecA protein loading, or 2) Asp-1080 is an essential component of the RecA protein-loading machinery. Consistent with our observation that the C-terminal domain of the RecB subunit is important for RecA protein loading, analysis of the truncated RecB1–929C enzyme, which has 30 kDa deleted from the C terminus, shows that it is also an efficient helicase that cannot load RecA protein. 4J. J. Churchill and S. C. Kowalczykowski, submitted for publication. In addition, recent characterization of the RecB(Thr807Ile)CD enzyme (also known as RecB2109CD enzyme) shows that it too is unable to load RecA protein. 5D. A. Arnold and S. C. Kowalczykowski, submitted for publication. Thus, it is clear that the RecB subunit plays a central role in the loading of RecA protein onto ssDNA. It remains unclear what role RecD subunit plays in RecA protein loading. RecBC enzyme loads RecA protein constitutively, whereas RecBD1080ACD enzyme cannot load RecA protein at all. One interpretation of these results is that the RecD subunit represses the RecA protein-loading activities of the RecBCD enzyme, and only following its inactivation is RecA protein-loading manifest. However, this simple interpretation is complicated by the observation that RecB1–929C enzyme cannot load RecA protein either. This shows that mutations in the RecB subunit can directly affect the RecA protein-loading properties of the enzyme. It will be interesting to test the RecA protein-loading properties of RecBD1080AC enzyme to further clarify the controlling role of the RecD subunit. The exact nature of the facilitated loading of RecA protein and its relationship to the nuclease activity of the RecB subunit remains to be determined. We are especially grateful to Doug Julin and Misook Yu for sharing their results before publication and particularly for their generous gift of RecBD1080ACD enzyme. We also thank the many members of the Kowalczykowski laboratory for their critical reading of this material: Richard Ando, Deana Arnold, Carole Bornarth, Piero Bianco, Frédéric Chédin, Frank Harmon, Julie Kleiman, Jim New, Erica Seitz, and Susan Shetterley.
Referência(s)