Biochemical Characterization of Bacteriophage T4 Mre11-Rad50 Complex*

The Mre11-Rad50 complex (MR) from bacteriophage T4 (gp46/47) is involved in the processing of DNA double-strand breaks. Here, we describe the activities of the T4 MR complex and its modulation by proteins involved in homologous recombination. T4 Mre11 is a Rad50- and Mn2+-dependent dsDNA exonuclease and ssDNA endonuclease. ATP hydrolysis is required for the removal of multiple nucleotides via dsDNA exonuclease activity but not for the removal of the first nucleotide or for ssDNA endonuclease activity, indicating ATP hydrolysis is only required for repetitive nucleotide removal. By itself, Rad50 is a relatively inefficient ATPase, but the presence of Mre11 and dsDNA increases ATP hydrolysis by 20-fold. The ATP hydrolysis reaction exhibits positive cooperativity with Hill coefficients ranging from 1.4 for Rad50 alone to 2.4 for the Rad50-Mre11-DNA complex. Kinetic assays suggest that approximately four nucleotides are removed per ATP hydrolyzed. Directionality assays indicate that the prevailing activity is a 3′ to 5′ dsDNA exonuclease, which is incompatible with the proposed role of MR in the production of 3′ ssDNA ends. Interestingly, we found that in the presence of a recombination mediator protein (UvsY) and ssDNA-binding protein (gp32), Mre11 is capable of using Mg2+ as a cofactor for its nuclease activity. Additionally, the Mg2+-dependent nuclease activity, activated by UvsY and gp32, results in the formation of endonuclease reaction products. These results suggest that gp32 and UvsY may alter divalent cation preference and facilitate the formation of a 3′ ssDNA overhang, which is a necessary intermediate for recombination-mediated double-strand break repair.

Replication in T4 3 phage is initiated through origin-dependent and -independent pathways (1). The origin-dependent pathway relies on the host cell RNA polymerase to synthesize an mRNA transcript, which is then processed and remains stably bound at the origin in what is known as an R-loop (2). The origin-independent pathway relies on homologous recombination (HR) within or between phage genome(s) (3). HR is essential to the lifecycle of T4 phage because the origindependent mode of initiation ends after only a few rounds of replication (4). A strong link between recombination and DNA repair was recognized early on using T4 phage as a model system (for review, see Ref. 5). In the late 1960s it was reported that certain genes when mutated severely affected replication, recombination, and DNA repair (6). Indeed, the current models of DNA double-strand break (DSB) repair share many of the same steps as recombination-mediated initiation of replication (7,8).
DSBs are one of the most lethal forms of DNA damage, and unless repaired they may lead to gross chromosomal rearrangements, deletions, duplications, or cell death (9,10). DSBs are caused by exogenous agents such as ionizing and ultraviolet radiation and chemical mutagens or endogenous agents such as reactive oxygen species. In eukaryotic cells, DSBs are repaired through two pathways, non-homologous end joining (NHEJ) and HR; however, in T4 phage it appears that all DSBs are repaired through HR (4,11).
The biochemical activities necessary for HR have been identified, and in most cases, the proteins responsible for these activities are known (8,12,13). HR begins with the processing of a dsDNA end to produce a 3Ј ssDNA overhang. In Escherichia coli, this step is performed by the well characterized helicase-nuclease complex, RecBCD (14). However, the genomes of eukaryotes, Archaea, and T4 phage do not contain any obvious RecBCD orthologs, and it appears that the Mre11-Rad50 (MR) complex together with other proteins specific to each system perform the analogous function (15). After end resection, the recombinase (RecA or RecA ortholog) binds to the ssDNA and forms a nucleoprotein filament that is capable of undergoing strand invasion and exchange with homologous DNA templates (16). This strand invasion event produces a D-loop structure that acts as a scaffold and primer for DNA synthesis (17). The HR pathway ends with either the resolution of the Holliday junction by a nuclease or the dissociation and strand annealing of the newly synthesized DNA strands (18).
In T4 phage, all of the proteins implicated in HR are well characterized except for Mre11 and Rad50 (gp47 and gp46, respectively) (19 -24). In eukaryotic systems, the MR complex has long been associated with the repair of DSBs (25). In Saccharomyces cerevisiae, null mutants of Mre11, Rad50, or Xrs2 (a third protein found in the S. cerevisiae MR complex) are unable to initiate and process meiotic DSBs and are sensitive to DNA-damaging agents that produce DSBs such as ionizing radiation or methylmethane sulfonate (18). This led to a proposal that the MR complex is responsible for resection (5Ј to 3Ј dsDNA exonuclease activity) of DSBs (26). However, in vitro, the eukaryotic MR complex exhibits a 3Ј to 5Ј dsDNA exonuclease and ssDNA endonuclease activity (27,28). The products of these activities are not 3Ј ssDNA overhangs, and therefore, the true function of the MR complex has been elusive.
Rad50 is a member of the SMC (structural maintenance of chromosome family of proteins), which are in turn members of the ATP binding cassette (ABC) superfamily. All known ABC protein family members are dimeric and bind ATP at the interface between two monomers. It is thought that the energy of ATP hydrolysis is used to drive conformational changes within the ABC protein/domain, which are then propagated to other associated proteins or domains (29). Most ABC protein family members are membrane transporters, such as the Cystic Fibrosis Transmembrane Regulator transporter and p-glycoprotein, which, when mutated, cause cystic fibrosis and multidrug resistance, respectively (30). In addition to transporters, the DNA repair enzymes MutS and RecF are ABC proteins. The structure of the nucleotide binding domain of Pyrococcus furiosus (Pfu) Rad50 has been solved in the apo-and ATP-bound forms (31). The structures reveal significant conformational rearrangements that occur upon ATP binding and/or hydrolysis, which are thought to control the activity of Mre11 (32). Rad50 and related SMC proteins have an unusual architecture where the ABC protein motifs are split by a long coiled-coil region. In Rad50, the apex of the coiled-coil contains a conserved CXXC motif, which dimerizes with a second Rad50 monomer to create a zinc binding site similar to a zinc finger (33). The purpose of the coiled-coil motif is not entirely clear, but it is thought to be involved in tethering the DSB to a homologous template or tethering the two ends of the DSB to each other (15). In eukaryotes the coiled-coil domain contains ϳ900 amino acids, whereas in T4 phage the coiled-coil contains around 200 residues. Presumably, the long coiled-coil region in eukaryotes is responsible for the expression difficulties encountered with the eukaryotic proteins. To date, only microgram quantities of eukaryotic Rad50 have been purified (27,28,34,35).
Mre11 is a member of the Ser/Thr protein phosphatase family, which contains four conserved motifs and requires divalent cations for activity (36). Although the mutation of Mre11 has only a mild effect in S. cerevisiae (37), in human cells the nuclease activity of Mre11 is essential, and its inactivation results in embryonic lethality and extreme sensitivity to ionizing radiation (38). In humans, inherited Mre11 mutations cause ataxia telangiectasia-like disorder, the symptoms of which include hypersensitivity to ionizing radiation and cerebellar degeneration (39). The Pfu Mre11 protein is dimeric and binds dsDNA near the dimeric interface (40). As mentioned above, in vitro assays reveal that Mre11 possesses a 3Ј to 5Ј dsDNA exonuclease activity along with ssDNA endonuclease activity (27).
Recently, several papers have clarified the role of the MR complex in the processing of DSBs. The Mimitou and Symington (41) and Ira and co-workers (42) laboratories simultaneously demonstrated that DSBs are processed through a two-step mechanism involving the MR complex, Sae2, ExoI, Sgs1, and Dna2. In the initial step, the MRX complex, in conjunction with Sae2, removes ϳ100 -200 bases from the 5Ј end of the DNA. In the second step, ExoI or Sgs1 with Dna2 catalyze rapid and processive removal of thousands of bases from the 5Ј end of the DNA. The resulting 3Ј ssDNA overhang can undergo Rad51-dependent HR or Rad52-dependent singlestrand annealing.
In this report, we express, purify, and biochemically characterize the MR DNA repair complex from T4 phage. We find that the T4 MR complex shares all the same activities as its eukaryotic counterparts, but expression and purification of the Rad50 and Mre11 proteins is rapid and yields milligram quantities of purified proteins, which allows for the mechanistic examination of the complex. Our experiments indicate that the T4 MR complex is a DNA-activated ATPase, a Mn 2ϩ -dependent ssDNA endonuclease. and a Mn 2ϩ -and ATP-dependent 3Ј to 5Ј dsDNA exonuclease. ATP hydrolysis is positively cooperative with respect to ATP concentration, and during the exonuclease reaction four nucleotides are removed per ATP hydrolyzed. Although the observed 3Ј to 5Ј dsDNA exonuclease activity is incompatible with its proposed in vivo function, we find that the presence of two T4 recombination proteins, UvsY and gp32, alter the nuclease activity of the MR complex, which may promote the formation of a 3Ј ssDNA recombinogenic end.

EXPERIMENTAL PROCEDURES
Cloning, Expression, and Purification of T4 Mre11 and wt-Rad50 and K42M-Rad50-The open reading frames for gp46 and 47 were PCR-amplified from bacteriophage T4 genomic DNA (Sigma) and cloned into the pTYB1 expression vector (New England Biolabs) using the NdeI and SapI restriction sites using the following primers: T4Rad50-NdeF, 5Ј-GGTG-GTCATATGAAGAATTTTAAACTTAATAG; T4Rad50-SapR, 5Ј-GGTGGTTGCTCTTCCGCATTAAACCATTACA-GTAAATCG; T4Mre11-NdeF, 5Ј-GGTGGTCATATGAAA-ATTTTAAATTTAGGTG; T4Mre11-SapR, 5Ј-GGTGGTTG-CTCTTCCGCATCATTGTGTTGCCTCTACATATAG. The K42M-Rad50 mutant was generated using the QuikChange TM protocol (Stratagene) using the mutagenic forward primer, K42M-Rad50F (5Ј-GGACGAAATGGCGGTGGTATGTCT-ACTATGCTAGAAGCC-3Ј, where the mutant codon is highlighted in bold). The reverse mutagenic primer is the reverse complement of the forward. The pTYB1 vector creates a fusion between the gene of interest and a self-cleavable intein/ chitin binding domain purification tag. The gp46-pTYB1, K42M-gp46-pTYB1, and gp47-pTYB1 expression vectors were separately transformed into BL21(DE3) cells and plated on LB-agar plates containing 150 g/ml ampicillin. A single colony from each plate was used to separately inoculate 50-ml flasks of LB-ampicillin that were shaken for 16 h at 37°C. 10 ml of starter culture was used to inoculate four 1-liter flasks of LB-ampicillin per protein, which were shaken at 225 rpm at 37°C to an A 600 of 0.8. The flasks were then cooled to 18°C, and expression was induced by the addition of 0.2 mM isopropyl 1-thio-␤-D-galactopyranoside. After 16 h the cells were collected by centrifugation at 8000 ϫ g for 8 min, and pellets were frozen at Ϫ20°C.
The purification scheme for T4 Rad50 (including K42M-Rad50) and Mre11 was identical through the chitin column elution step. A frozen cell pellet (ϳ15 g) was resuspended in 75 ml of ice-cold lysis buffer (20 mM Tris-HCl, 500 mM NaCl, 1 mM EDTA, pH 8.0) and lysed by sonication. All purification steps were carried out at 4°C. The lysed cells were then centrifuged at 18,000 ϫ g for 60 min. The cell-free extract was loaded onto a column packed with 5 ml of chitin beads (New England Biolabs). The column was then washed with 100 ml of lysis buffer followed by an overnight wash with 1. DNA Substrates-The M13 phage ssDNA was purified using PEG precipitation and phenol/chloroform extraction (43). The 50-bp dsDNA substrate was prepared by annealing equal amounts of two complementary oligonucleotides with the following sequences: ds50-mer_top (5Ј-CTC TTG GTG ATT ATG ATG GTT GCA ATA CAT TTA ATT TCA TTA TCA ATT AG-3Ј) and ds50-mer_bottom (5Ј-CTA ATT GAT AAT GAA ATT AAA TGT ATT GCA ACC ATC ATA ATC ACC AAG AG-3Ј). The 17 position 2-aminopurine (2-AP)-containing substrate was of identical sequence to ds50-mer_top except that the adenine at position 17 (relative to the 3Ј end of the DNA) was replaced with 2-AP. The DNA substrate with the 2-AP at the 1 position (3Ј end) had an identical sequence to ds50-mer_top except that the 3Ј G was replaced with 2-AP. To prepare the 5Ј-labeled 50-bp DNA, the annealed duplex was diluted to a concentration of 5 M and treated with polynucleotide kinase (New England Biolabs) and 10 Ci of [␥-32 P]ATP in a volume of 20 l (total ATP concentration of 0.165 M) for 30 min. After 30 min cold ATP was added to a final concentration of 100 M and allowed to react for 30 additional minutes. The reaction was then quenched at 65°C for 20 min followed by removal of free nucleotides by passing the reaction through a G-25 Sephadex spin column. To prepare the 3Ј-labeled 50-bp DNA, the annealed duplex was diluted to a concentration of 10 M and treated with 5 units of T4 DNA polymerase (Promega) in the presence of 1 mM dCTP, dTTP, and dATP in a final volume of 20 l for 5 min. The reaction was then heated at 65°C for 20 min to inactivate the T4 polymerase. After heat inactivation, 5 units of T4 exonucleasedeficient polymerase was added to the reaction along with 20 Ci of [␣-32 P]dGTP in a final volume of 38 l. The reaction was allowed to proceed at 37°C for 30 min before the addition of 2 l of 0.1 mM dGTP. The reaction was continued for 30 additional minutes, then heat-inactivated for 20 min at 65°C. Finally, the reaction was passed through a G-25 Sephadex spin column to remove free nucleotides and precipitated protein. The uniformly labeled 1.68-kb dsDNA was prepared by performing a 100-l PCR reaction in Taq polymerase buffer (Promega) using 5 units of Taq polymerase, 40 Ci of [␣-32 P]dGTP, 0.1 mM cold dNTP mix, 1 M gp46-Nde-F, 1 M gp46-Sap-R, and 100 ng of the gp46-pTYB1 vector. The labeled PCR product was purified using the Qiagen PCR purification kit following the manufacturer's protocol and quantitated using liquid scintillation counting.
Nuclease and ATP Hydrolysis Assays-The standard conditions for the nuclease reactions were 20 mM Tris-HCl, 50 mM KCl, 2 mM ATP, pH 7.6, and a divalent metal concentration indicated in the figure legends. The concentrations of ssDNA, dsDNA, Rad50, and Mre11 varied for each experiment and are given in the individual figure legends. The gp32 and UvsY proteins were purified by intein/chitin-based affinity chromatography as described (44). Reactions that were analyzed using 1% Tris acetate-EDTA-agarose gels ( Fig. 1) were quenched in 5 mM Tris-HCl, pH 7.5, 200 mM EDTA, 0.2% orange G, 0.015% bromphenol blue, 0.015% xylene cyanol FF, and 7.5% Ficoll 400. The agarose gels were stained with 10 g/ml ethidium bromide for 30 min before visualization. Reactions that were analyzed using Tris borate-EDTA-urea-PAGE (15% polyacrylamide) (Figs. 2 and 7) were quenched with 50 mM EDTA, 95% formamide, 0.015% xylene cyanol FF, and 0.015% bromphenol blue. The PAGE gels ran for 2 h at 35 mA followed by overnight exposure to a PhosphorImager plate and subsequent analysis using a Storm PhosphorImager and ImageQuant software (GE Healthcare). The reactions that were analyzed using PEI-TLC (Figs. 4, C and D, 5A, and 6) were quenched in an equal volume of 200 mM EDTA. The quenched samples were then spotted onto the PEI plates (1-2 l volume) and developed in 300 mM KP i , pH 7.0. The TLC plate was then dried and exposed for 2 h to the PhosphorImager plate followed by analysis using the Storm PhosphorImager and ImageQuant software.
To examine the steady-state exonuclease activity of the MR complex, we employed a real-time continuous assay based on the enhancement of fluorescence when the nucleotide analog, 2-AP, is excised from DNA (45). The assay was carried out at 30°C using excitation and emission wavelengths of 310 and 375 nm, respectively, on a Cary Eclipse (Varian) spectrofluorometer. The reaction buffer contained 50 mM Tris-HCl, pH 7.6, 50 mM KCl, 5 mM MgCl 2 , 0.3 mM MnCl 2 , 0.1 M MR complex, and 1.3 M 2-AP labeled dsDNA. The reactions were monitored for ϳ5-10 min, and velocities were determined using at least three min of the time course.
To determine the steady-state ATP hydrolysis activity of Rad50 and the MR complex, we employed the standard coupled fluorometric ATP hydrolysis assay (46) carried out at 30°C using excitation and emission wavelengths of 340 and 460 nm, respectively, on either a SLM-Aminco (SLM Instruments, Inc.) or a Cary Eclipse spectrofluorometer (Varian). The reaction buffer contained 50 mM Tris-HCl, pH 7.6, 50 mM KCl, 5 mM MgCl 2 , 50 M NADH, 150 M phosphoenolpyruvate, 6.67 units/ml pyruvate kinase, and 10 units/ml lactate dehydrogenase. In reactions where Mre11 was included, its concentration was always in slight excess over the Rad50 concentration. In reactions containing DNA, its concentration was held in excess over that of Rad50 or the MR complex and well above the K activation for DNA. 4 Steady-state rate constants were determined using Sigmaplot 10.0/Enzyme Kinetics Module 1.3 (Systat Software, Inc.).

Expression and Purification of T4 Rad50 and Mre11-
Overexpression of eukaryotic Mre11 and Rad50 requires the use of either S. cerevisiae or insect cells and yields only microgram amounts of purified protein (27,28,34). Moreover, eukaryotic Rad50 is difficult to overexpress separately and requires co-expression of Mre11 to form significant amounts of soluble protein (47). We have achieved overexpression and purification of T4 Rad50 by constructing a C-terminal fusion with the intein/chitin binding domain provided by the pTYB1 vector. The chitin binding domain facilitates capture by chitin beads, and the intein domain allows for reductant-mediated self-cleavage (48). After self-cleavage, the Rad50 protein that elutes from the chitin beads contains a strong absorbance at 260 nm, indicating the presence of bound nucleotides or nucleic acids. Dialysis of the protein did not reduce 260-nm absorbance; therefore, we applied the eluted protein to phosphocellulose resin. The material that strongly absorbed at 260 nm was not retained by the column, but nearly 100% of the protein was efficiently bound. After a buffer wash, the NaCl concentration was increased from 200 to 400 mM, which eluted the Rad50 protein. This protein had a 280/260 nm ratio of 1.98, indicating that there is little to no nucleotide or nucleic acid contamination in our preparation of Rad50. Additionally, no appreciable enhancement of DAPI dye fluorescence was observed upon the addition of protease-K-digested Rad50, indicating the absence of contaminating DNA (data not shown). The overall yield for T4 Rad50 was ϳ3 mg of protein per liter of LB (Fig. 1A, third lane). The Rad50 Walker A motif mutant, K42M-Rad50, behaved essentially identical to the wild-type protein.
T4 Mre11 was also C-terminal-fused to the intein/chitin binding domain provided by the pTYB1 vector. Mre11 was expressed in an identical fashion as Rad50 but was ϳ10-fold more abundant in the cell-free extracts as compared with Rad50. Additionally, the 280/260-nm absorbance ratio indicated that no nucleotides or nucleic acids were present in the protein that eluted from the chitin column. Therefore, only a single column was necessary to achieve the desired level of purity. The overall yield for T4 Mre11 was ϳ25 mg of protein/liter of LB (Fig. 1A, second lane).
Requirements for Mre11 Nuclease Activity-To determine the requirements for the predicted Mre11 nuclease activity, we performed agarose gel electrophoresis-based nuclease assays. The metal, protein, and ATP dependence was tested using circular ssDNA (M13mp18) and linear dsDNA (Fig. 1B). The ssDNA nuclease reaction required the presence of Mre11, MnCl 2 , and Rad50. Under these conditions, replacement of MnCl 2 with MgCl 2 or ZnCl 2 resulted in no observable nuclease activity (data not shown). The dsDNA nuclease reaction required the same components; however, ATP was also necessary for nucleolytic degradation of the dsDNA. In addition to defining the components required for nuclease activity, these results strongly argue against the presence of a contaminating nuclease in our preparations of either Mre11 or Rad50 (as both proteins are required). We next tested the endo/exo activity using linear and circular ssDNA or dsDNA (Fig. 1C). These reactions indicate that the T4 MR complex contains ssDNA endonuclease activity (i.e. both linear and circular ssDNA are degraded) and dsDNA exonuclease activ-  ity (i.e. only linear dsDNA is degraded). We also confirmed that the nuclease activity resides in the Mre11 subunit (Fig.  1D). This assignment is predicted based on sequence analysis; however, the nuclease subunit of the T4 MR complex had been previously assigned to Rad50 rather than Mre11 (49). We separately assayed the ability of Mre11 and Rad50 to degrade dsDNA at elevated concentrations of protein in the presence of MnCl 2 and ATP (Fig. 1D). Although the activity is Ͼ100-fold less than the MR complex, as expected, the nuclease activity resides within the Mre11 subunit.
Polarity of the dsDNA Exonuclease Activity of the MR Complex-To determine the nuclease polarity of the MR complex, we compared the product pattern resulting from nuclease activity using either 5Ј-or 3Ј-labeled 50-bp dsDNA (Fig. 2). The presence of intermediate-length products in reactions using the 5Ј-labeled DNA substrate and their absence in reactions with the 3Ј-labeled DNA substrate is consistent with a 3Ј to 5Ј exonuclease.
Steady-state ATP Hydrolysis Activity of Rad50 and the MR Complex-Rad50 is part of the ABC protein superfamily, and it is known that Rad50 catalyzes the hydrolysis of ATP (50). Additionally, the data shown in Fig. 1 indicate that ATP is necessary for the dsDNA exonuclease activity of the MR complex. For these reasons we examined the Mg-ATP hydrolysis activity of the MR under conditions where the nuclease activity is negligible (i.e. in the absence of Mn 2ϩ ). We determined the steady-state Mg-ATP hydrolysis rate constants for Rad50 and the MR complex with and without DNA (Table 1). Because these reactions are performed without Mn 2ϩ , DNA is not consumed and can be considered an activator rather than a substrate. We found that by itself, Rad50 is a relatively weak ATPase, and the rate of ATP hydrolysis is sigmoidally dependent on the concentration of ATP. The data fit best to the Hill equation, where v is velocity, V max is the maximal velocity at saturating substrate concentration, K m is the Michaelis constant (substrate concentration where v ϭ 0.5⅐V max ), and n is the Hill coefficient. An . The large activation of ATPase activity is specific to linear dsDNA as circular plasmid DNA or a 50-base ssDNA oligonucleotide had no effect (data not shown). Mutation of the predicted Walker A lysine to methionine (K42M-Rad50) reduced the observed ATPase activity ϳ200-fold compared with Rad50 alone and ϳ4000-fold compared with the MR-DNA complex. This large decrease in activity is expected given the role the Walker A lysine plays in stabilizing negative charge that forms during the transition state for ATP hydrolysis. The residual activity that is observed represents the upper limit for the level of ATPase contamination in our preparation of Rad50 (less than 0.025%). Steady-state Exonuclease Activity of the MR Complex-To continuously monitor the exonuclease activity of the MR complex, we employed the fluorescent nucleotide analog, 2-AP, which has low fluorescence when incorporated into DNA and high fluorescence as a free nucleotide. We performed the assay under steady-state conditions ([DNA] Ͼ [enzyme]) with the 2-AP probe located at either the 1 or 17 position relative to the 3Ј end of a 50-bp DNA substrate. As seen in Fig. 3A, when the 2-AP probe is located at the 1 position, the steady-state exonuclease activity of both the wild-  type and K42M-Rad50 MR complex is essentially unaffected by the presence of either ATP or the non-hydrolyzable ATP analog, AMP-PNP (k cat values ranging between 2.3 and 3.2 min Ϫ1 ). When the 2-AP probe is located at the 17 position (Fig. 3B), the steady-state exonuclease rate of both wild-type and the K42M-Rad50 MR complex is much reduced compared with the 1 position substrate with k cat values of 0.03 and 0.015 min Ϫ1 , respectively. The addition of ATP to the wildtype MR complex increases the exonuclease rate (0.24 min Ϫ1 ) but decreases the rate of the K41M-Rad50 MR complex (0.008 min Ϫ1 ). The addition of AMP-PNP decreases the rate of both the wild-type and K42M-Rad50 MR complex with k cat values of 0.001 and 0.006 min Ϫ1 , respectively. Metal Dependence of ATP Hydrolysis and Exonuclease Activities-Even though the nuclease activity strongly prefers MnCl 2 over MgCl 2 , the ATP hydrolysis activity of the MR-DNA complex prefers MgCl 2 (Fig. 4A). Under steady-state conditions, the apparent k cat for ATP hydrolysis in the presence of 5 mM MgCl 2 or 5 mM MnCl 2 is 2.8 Ϯ 0.3 and 0.23 Ϯ 0.03 s Ϫ1 , respectively. Because the nuclease activity requires MnCl 2 , yet under physiological conditions the concentrations of Mg 2ϩ far exceeds that of Mn 2ϩ (51), we measured the rate of ATP hydrolysis at fixed MgCl 2 (5 mM) while varying the concentration of MnCl 2 from 0.1 to 5 mM (Fig. 4B). The ATP hydrolysis rate is inhibited by MnCl 2 with an IC 50 of 1.3 Ϯ 0.1 mM. We next explored the metal dependence of the nuclease reaction in more detail. We employed the uniformly [␣-32 P]dGTP-labeled 1.6-kb PCR product as the exonuclease substrate and again varied the concentration of MnCl 2 in the presence of 5 mM MgCl 2 (Fig. 4C). The nuclease activity sharply increased between MnCl 2 concentrations of 0 and 0.1 mM followed by a slow decrease in nuclease activity between 0.4 and 5 mM. This data can be fit using the following uncompetitive substrate inhibition equation, where v is velocity, V max is the maximal velocity at saturating activator concentration, K a is the Michaelis constant for the activator (concentration of A where v ϭ 0.5⅐V max ), and K i is the inhibition constant. The data-fitting yielded a K a Mn 2ϩ of 0.011 Ϯ 0.002 mM and K i Mn 2ϩ of 4.1 Ϯ 0.8 mM. Finally, we compared the MR complex nuclease activity with 5 mM MnCl 2 alone or with 5 mM MgCl 2 and 0.3 mM MnCl 2 in combination (Fig. 4D). The results of this assay indicate that the nuclease activity is essentially identical under both of these conditions. Rad50 ATP Hydrolysis and Utilization during the Exonuclease Reaction-A TLC-based assay was employed to simultaneously monitor the ATP hydrolysis and exonuclease activity of Rad50 and Mre11, respectively, using MgCl 2 and MnCl 2 concentrations of 5 and 0.3 mM, respectively (Fig. 5A). To visualize both DNA and ATP, we used a uniformly [␣-32 P]dGTP-labeled 1.6-kb PCR product as the DNA substrate and ATP that was labeled at the ␣-phosphate with 32 P. The concentration of ATP was held at 0.2 mM, which is ϳ4-fold above the K m ATP for the MR complex ( Table 1). The TLC assay enables visualization of both ATP/ADP and DNA/ dGMP, making the determination of ATP utilization highly accurate. The MR complex was held in excess over the DNA substrate so that the rate of exonuclease activity reflects the rate of nucleotide excision and translocation and not productive DNA binding. As seen in Fig. 5B, the rate of exonuclease and ATPase activity is 0.1664 and 0.0408 M/s Ϫ1 , respectively. This suggests that approximately four bases are removed for every ATP hydrolyzed. The concentration of dsDNA ends in this assay is 0.0217 M, which represents the upper limit for the concentration of DNA-bound MR complex. Assuming a fully saturated dsDNA substrate during the exonuclease reaction, the k cat for ATP hydrolysis is then 1.9 s Ϫ1 and the maximum rate of nucleotide removal is 7.67 s Ϫ1 .
The Effect of T4 UvsY and gp32 Proteins on the Exonuclease Activity of the MR Complex-Consistent with MR complexes from human, P. furiosus, and S. cerevisiae, Mn 2ϩ is the preferred metal for the exonuclease reaction in the T4 system. However, because physiological concentrations of Mg 2ϩ far exceed those of Mn 2ϩ (51), we further explored the ability of Mg 2ϩ to support exonuclease activity. To quantify the exonuclease activity of the MR complex with Mg 2ϩ and Mn 2ϩ , we employed the uniformly [␣-32 P] dGTP-labeled 1.6-kb PCR product and the TLC assay (Fig. 6). Protein was held in excess over the dsDNA substrate; therefore, the loss of product can be fitted to a single-exponential decay with rate constants of 0.105 Ϯ 0.001 and 0.009 Ϯ 0.0005 min Ϫ1 for Mn 2ϩ and Mg 2ϩ , respectively (an 11-fold difference between Mn 2ϩ and Mg 2ϩ ). We then tested the effect of UvsY and gp32 on the exonuclease rates in the presence of either Mg 2ϩ or Mn 2ϩ . UvsY and gp32 were chosen as possible candidates for the modulation of MR complex activity because these two proteins bind to the product of the MR reaction (i.e. ssDNA) and participate in the subsequent step of DSB repair (strand invasion). We found that the presence of UvsY and gp32 increased the Mg 2ϩ -dependent reaction by nearly 5-fold (rate constant of 0.0410 Ϯ 0.0037 min Ϫ1 ) while having no effect on the rate of the Mn 2ϩ -dependent reaction (rate constant of 0.1031 Ϯ 0.0104 min Ϫ1 ). Separately, UvsY and gp32 had no measurable effect on the Mg 2ϩ -dependent nuclease activity (data not shown). Several control experiments were performed to rule out the possibility of nuclease contamination in our preparations of UvsY and gp32 (Fig. 6). The Mg 2ϩ -dependent exonuclease reaction required Rad50, Mre11, UvsY, gp32, and ATP. Omission of Rad50, Mre11, or ATP resulted in a complete  loss of the exonuclease activity, strongly arguing against the introduction of a contaminating nuclease present in the preparations of UvsY or gp32. We also examined the effect of UvsY and gp32 on the steady-state rate of ATP hydrolysis by the MR complex in the presence of a DNA substrate (Table  2). Separately, UvsY and gp32 did not significantly alter the rate, whereas the combination reduced the rate by 1.9-fold. The polarity of the UvsY/gp32-enhanced, Mg 2ϩ -dependent exonuclease reaction was examined under steady-state conditions using 3Ј-labeled 50-bp dsDNA. This substrate was chosen because the appearance of any DNA products of intermediate size represents a deviation from the strict 3Ј to 5Ј exonuclease reaction that was observed in Fig. 2. Under steady-state conditions (which are required when short DNA substrates are used), the Mg 2ϩ -dependent reaction in the absence of UvsY/gp32 is nearly undetectable (i.e. the UvsY/gp32 enhancement of the steady-state Mg 2ϩ -dependent reaction appears greater than the 5-fold observed in Fig. 6). Therefore, we compared the Mn 2ϩ -dependent reaction to the Mg 2ϩ -dependent reaction, both, in the presence of UvsY/gp32. As shown in Fig. 7, in contrast to the reaction in the presence of Mn 2ϩ , the Mg 2ϩ -dependent reaction results in products with a length ranging between 15 and 25 bases, suggesting that an additional form of nuclease activity has occurred. Quantitation of this reaction reveals that ϳ32% of the products were between 15 and 25 bases with the remaining 68% as dGMP.

DISCUSSION
Bacteriophage T4 is a well established model system for investigating the biochemistry of proteins involved in DNA replication, recombination, and repair (3,52,53). Studies using the T4 phage model system led to the first proposed model for recombination-dependent DNA replication (RDR) (54). The RDR model begins with the formation of a singlestranded 3Ј DNA end that subsequently undergoes strand invasion and pairing with a homologous dsDNA template. The single-stranded 3Ј DNA can be formed as a result of incomplete replication of the lagging strand or through the action of a nuclease. Over the past three decades since the RDR model was proposed, all of the enzymes and proteins that are required to carry out RDR in T4 have been isolated and extensively characterized biochemically, with the notable exception of the nuclease complex thought to be responsible for the creation of the 3Ј ssDNA overhangs (Rad50 (gp46) and Mre11 (gp47)). Here we have described for the first time the expression, purification, and biochemical characterization of the Rad50 and Mre11 proteins from bacteriophage T4.
On a qualitative level it appears that the various activities of the MR complex are well conserved between all three kingdoms of life. The T4 MR complex is an ATPase, an ssDNA  endonuclease, and a 3Ј to 5Ј dsDNA exonuclease. ATP hydrolysis is positively cooperative and is activated by the presence of Mre11 and DNA. We have found that during repetitive dsDNA exonuclease activity, ϳ4 nucleotides are removed for every ATP that is hydrolyzed, and the maximum exonuclease rate is 7.7 nucleotides/s. ATP hydrolysis increases the rate of multiple nucleotide excisions but does not alter the rate of the first nucleotide excision. Interestingly, we have found that two T4 phage proteins that are necessary for the subsequent steps of RDR/DSB repair activate the Mg 2ϩdependent nuclease reaction, which produces an altered product profile as compared with the Mn 2ϩ -dependent exonuclease reaction.
To our knowledge, this is the first instance that the steadystate ATP hydrolysis rate constants have been reported for a Rad50 homolog in the presence and absence of Mre11 and a DNA substrate. We found that by itself, Rad50 is a relatively inefficient ATPase, with a k cat of 0.15 s Ϫ1 . The reaction is mildly positively cooperative with a Hill coefficient of 1.4, indicating Rad50 is a least a dimer, consistent with the structural analysis of Pfu Rad50 and other ABC proteins (31,32). Separately, the addition of DNA or Mre11 to Rad50 resulted in very minor perturbations in the kinetic parameters of Rad50. In contrast, the addition of both Mre11 and DNA resulted in a 20-fold increase in k cat and an increase in the Hill coefficient to 2.4. These reactions are performed with Mg 2ϩ as the only divalent cation so that there is essentially no nuclease activity under these conditions. This indicates that ATP hydrolysis can be uncoupled from nuclease activity and suggests that Mre11 and dsDNA act as allosteric activators of Rad50. The increase in the Hill coefficient from 1.4 to 2.4 suggests that in the presence of Mre11 and dsDNA, there are more than two ATP sites in communication with each other. It is not clear how this might occur, but it is tempting to speculate that binding of dsDNA promotes hetero-octamer (Mre11 2 /Rad50 2 ) 2 formation through zinc-hook-mediated tethering of the Rad50 coiled-coil domains (15). Hetero-octamers of the Pfu MR complex have been previously observed in electron microscopy (33), and atomic force microscopy experiments (55) have revealed that DNA binding results in a straightening of Rad50 coiled-coils, which prevents an intracomplex tether (hetero-tetramer) and promotes an intercomplex tether (hetero-octamer).
The exact function of Rad50 ATP hydrolysis activity has been unclear (56). Based on structural changes between the ATP-free and bound forms of Pfu Rad50, it has been proposed that ATP hydrolysis induces a conformational change that may promote product release (dNMP) or drive MR complex translocation (31). Our data are inconsistent with the product release model, as the rate of excision of the first nucleotide at the 3Ј end of the DNA is not affected by ATP. Additionally, the stoichiometry of ATP hydrolysis is one ATP per four nucleotides removed. If ATP hydrolysis was directly linked to product release, then a 1:1 correspondence between ATP hydrolysis and nucleotide removal would be expected. The data are more consistent with ATP being involved in the translocation of the MR complex after nucleotide excision and release. Because ATP and AMP-PNP do not affect the rate of the first nucleotide excision for either the wild-type or K42M-Rad50 MR complexes, it is unlikely that the MR complex must unwind or open the DNA duplex before nucleotide excision. In contrast, ATP hydrolysis increased the exonuclease rate when the probe is moved 17 bases away from the 3Ј end, indicating that ATP hydrolysis is necessary for multiple, processive nucleotide excisions. This suggests that the energy of ATP hydrolysis likely drives translocation of the MR complex. Consistent with this, when the probe is in the 17 position, AMP-PNP potently inhibits the exonuclease activity of the wild-type enzyme, and both ATP and AMP-PNP inhibit the K42M-Rad50 MR complex. The observed inhibition is likely due to the stalling of the MR complex after nucleotide excision and product release (i.e. the MR complex is unable to translocate or dissociate from the DNA in the ATP-bound form). In the absence of ATP, the MR complex likely dissociates from the DNA substrate and rebinds at the new ss/ dsDNA junction (non-processively). If this interpretation is correct, then the steady-state exonuclease rate constant in the presence of AMP-PNP is equal to the rate of dissociation of the MR complex from DNA. The 1:4 ATP/base stoichiometry suggests that the Mre11 is able to remove four nucleotides before it is required to translocate to the new ss/dsDNA junction. This would require "scrunching" of the DNA, similar to that which has been proposed for the NS3 helicase (57).
The data shown in Fig. 2 are consistent with a 3Ј to 5Ј dsDNA exonuclease. This activity is at odds with the proposed physiological function of the MR complex, a paradox that has been long recognized (15,58,59). It is conceivable that the observed 3Ј to 5Ј exonuclease activity is the relevant physiological activity and the MR complex does not directly participate in the production of a 3Ј ssDNA overhangs. However, this possibility seems unlikely for the following reasons; 1) it is well established that in eukaryotes the nuclease activity of the MR complex is required for the removal of Spo11/ Rec12, which is covalently linked to the 5Ј end of the DSB during meiosis (60), 2) the nuclease activity of Mre11 is required for the removal of topoisomerase-5Ј DNA covalent adducts that are stabilized by anti-cancer etoposide derivatives (10), 3) the nuclease activity of Mre11 is required for repair of ionizing radiation-induced DSBs (38), and 4) in T4 phage, Mre11 and Rad50 are necessary for the initiation of recombination-dependent replication and recombination-dependent DSB repair, which begins with the 5Ј to 3Ј resection of a DNA end (7,61). Even with the above data, the precise role of the MR complex in DSB repair is still an open question, and the polarity paradox remains. Therefore, we tested several T4 recombination proteins for a possible change in the nuclease mechanism of the T4 MR complex. In the presence of MnCl 2 , we did not observe any significant differences in the rate or product profile of MR nuclease assays in the presence of UvsX, gp32, and UvsY either alone or in combination. However, UvsY and gp32 were found to enhance the ability of the T4 MR complex to utilize MgCl 2 as a cofactor for the nuclease activity. We examined the product profile of the MgCl 2 -dependent reaction in the presence of UvsY and gp32, and in addition to the dGMP that is produced by 3Ј to 5Ј exonuclease activity, we also observed a distribution of products in the size range of 15-25 bases. These products could be the result of a 5Ј to 3Ј exonuclease or an endonuclease activity. We favor an endonuclease mechanism due to the absence of products between 50 and 25 bases, which would be produced if 5Ј to 3Ј exonuclease activity had occurred. Experiments are currently under way to determine the exact nature of the alteration in nuclease mechanism. A metal-dependent change nuclease mechanism has also been observed in the MR complex from the thermophilic organism P. furiosus when assayed at elevated (physiologically relevant) temperatures (59). In the case of the Pfu MR complex, the MgCl 2 -dependent reaction produces a 3Ј ssDNA overhang through endonucleolytic cleavage of sites close to the 5Ј end of the dsDNA, similar to what we are proposing here for the T4 MR complex.
The mechanism behind the increased ability to use Mg 2ϩ as a cofactor is not clear but likely involves physical interaction(s) between the MR complex and UvsY/gp32. Based on our observation that 5 mM MgCl 2 does not inhibit the nuclease activity of the complex even when the concentration of MnCl 2 is 50-fold lower (0.1 mM), it appears that the affinity of Mg 2ϩ for the Mre11 active site is low compared with Mn 2ϩ . This is consistent with the fact that Mre11 crystals grown in the presence of MgCl 2 only have one of the two required divalent cation sites occupied (62). If this is correct and Mre11 normally has weak affinity for MgCl 2 , then it is plausible to assume that UvsY and gp32 act in an allosteric manner to increase the affinity of Mg 2ϩ for the active site of Mre11. This hypothesis requires that gp32 and/or UvsY directly interact with the MR complex, which has been previously observed in T4 phage lysate (63). Additionally, in vivo experiments have strongly suggested a physical interaction between gp32 and the MR complex (64). To enhance the affinity of Mg 2ϩ for Mre11, the interaction of gp32 and/or UvsY with the MR complex would have to reorient the active site in some manner. The nature of this putative active site perturbation is unclear and may require crystallographic analysis of the T4 MR complex.
The mechanism behind the Mg 2ϩ -dependent change in product profile is also not clear and will require further investigation. One hypothesis is that the ssDNA endonuclease activity of the MR complex is enhanced in the presence of Mg 2ϩ and UvsY/gp32. This requires at least transient melting of the DNA duplex, which is an activity that has been previously observed in gp32 (65,66). If gp32 and/or UvsY were also to protect the 3Ј but not the 5Ј strand of the DNA from endonucleolytic cleavage, then this would result in the production of a 3Ј ssDNA overhang that is pre-bound with UvsY and gp32 and primed for the next step of recombination, filament formation by UvsX. It is clear from the data shown in Fig. 7 that even in the presence of UvsY and gp32, there is still a high level of 3Ј to 5Ј dsDNA exonuclease activity (nearly 70% of the products are produced as a result of exonuclease activity). This may indicate that additional proteins are required for the efficient production of a 3Ј ssDNA overhang. Based on recent results from S. cerevisiae and P. furiosus, which demonstrated the involvement of helicases in MR-dependent DSB resection (42, 59), a possible candidate is UvsW helicase. UvsW helicase is known to be necessary for DSB repair (67,68) and is a functional analog of Sgs1, which is the S. cerevisiae helicase involved in DSB resection. We are currently in the process of testing the above hypothesis.
With the biochemical characterization of the bacteriophage T4 MR complex reported here, the entire complement of T4 phage proteins involved in recombination and DSB repair have been purified and assayed in vitro. The bacteriophage T4 system provides milligram quantities of purified Mre11 and Rad50, which is not currently possible in the eukaryotic systems. Given the apparent conservation in DNA repair mechanisms between all kingdoms of life, it is expected that the T4 phage model system will provide valuable mechanistic insights into what has been termed the "keystone complex" of DNA repair (69).