Functional Analysis of the Bacteriophage T4 Rad50 Homolog (gp46) Coiled-coil Domain*

Background: The Mre11-Rad50 (MR) complex coordinates DNA double strand break repair. Results: Mutations of the Rad50 coiled-coil disrupt the enzymatic functions of the MR complex. Conclusion: The Rad50 coiled-coil regulates and mediates communication between the enzymatic domains of Rad50 and Mre11. Significance: The Rad50 coiled-coil plays a more direct role in the enzymatic function of Rad50 and Mre11 than was previously assumed. Rad50 and Mre11 form a complex involved in the detection and processing of DNA double strand breaks. Rad50 contains an anti-parallel coiled-coil with two absolutely conserved cysteine residues at its apex. These cysteine residues serve as a dimerization domain and bind a Zn2+ cation in a tetrathiolate coordination complex known as the zinc-hook. Mutation of the zinc-hook in bacteriophage T4 is lethal, indicating the ability to bind Zn2+ is critical for the functioning of the MR complex. In vitro, we found that complex formation between Rad50 and a peptide corresponding to the C-terminal domain of Mre11 enhances the ATPase activity of Rad50, supporting the hypothesis that the coiled-coil is a major conduit for communication between Mre11 and Rad50. We constructed mutations to perturb this domain in the bacteriophage T4 Rad50 homolog. Deletion of the Rad50 coiled-coil and zinc-hook eliminates Mre11 binding and ATPase activation but does not affect its basal activity. Mutation of the zinc-hook or disruption of the coiled-coil does not affect Mre11 or DNA binding, but their activation of Rad50 ATPase activity is abolished. Although these mutants excise a single nucleotide at a normal rate, they lack processivity and have reduced repetitive exonuclease rates. Restricting the mobility of the coiled-coil eliminates ATPase activation and repetitive exonuclease activity, but the ability to support single nucleotide excision is retained. These results suggest that the coiled-coiled domain adopts at least two conformations throughout the ATPase/nuclease cycle, with one conformation supporting enhanced ATPase activity and processivity and the other supporting nucleotide excision.

Failure to properly detect and repair DNA double strand breaks (DSB) 4 in the genome can lead to chromosomal rearrangement, genome degradation, oncogenesis, and cell death (1). The necessity of a robust response to DSBs is underscored by the conservation of the Mre11-Rad50 (MR) complex across all domains of life (2). The MR complex localizes to DSBs shortly after their formation (3) and serves a number of enzymatic and architectural functions. The MR complex initiates the 5Ј to 3Ј resection that is required for homologous recombination (4) and microhomology-mediated end joining (5). Deletion of Mre11 or Rad50 causes embryonic lethality in mammals (4), and mutations in Mre11 are implicated in genetic disorders characterized by neural degeneration, oncogenesis, and hypersensitivity to radiation (6,7). In bacteriophage T4, the MR complex is required for obligate recombination-dependent DNA replication after a short period of origin-mediated replication of the genome (8).
Mre11 is a Mn 2ϩ -dependent nuclease that is composed of three structural domains (9). The capping and the nuclease domains are considered to make up the core structure of Mre11 (10). X-ray crystal structures of the MR complex reveal that the capping and nuclease domains interact with the N-and C-terminal lobes of Rad50, respectively (11)(12)(13). The C-terminal domain (CTD) of Mre11 is the smallest of the three domains but contains the largest Rad50-binding interface (12). Connecting the CTD to the capping domain is a flexible linker, which in the T4 system acts as an autoinhibitory element in the absence of Rad50 (14). In vitro, Mre11 is capable of single-stranded DNA endonuclease and dsDNA exonuclease activity in the 3Ј to 5Ј direction (11)(12)(13). The processive exonuclease activity of Mre11 requires the ATPase activity of Rad50 (15).
Rad50 is a member of the Structural Maintenance of Chromosomes family, which is a group of soluble ATP-binding cassette (ABC) enzymes that associate DNA in all domains of life (16). Structural Maintenance of Chromosomes proteins have a common structure consisting of two halves of an ABC domain that are separated in sequence by an antiparallel coiled-coil (17,18). The two halves of the ABC domain dimerize to form the nucleotide binding domain (NBD) that contains two ATP-active sites that are shared across the dimeric interface (19). The length of the coiled-coil in Rad50 varies from 250 amino acids in T4 phage to 900 amino acids in humans (2). A hinge domain is located at the apex of the coiled-coil, distal from the ABC domain, which serves as a second site of dimerization (20,21). In Rad50, dimerization in this hinge domain occurs by tetrahedral coordination of a Zn 2ϩ cation by absolutely conserved cysteine residues in the zinc-hook motif (CXXC) (20).
The precise role(s) of the Rad50 coiled-coil in MR complex function is unclear. In yeast, the phenotype of a zinc-hook deletion mutant is identical to that of complete Rad50 deficiency (22). Similarly, deletion or shortening of the coiled-coil eliminates both Spo11-induced DSB formation and telomere maintenance and greatly reduces the efficiency of homologous recombination (22). Most models for coiled-coil function involve its ability to physically link DNA strands to each other (20,(23)(24)(25). In eukaryotes, it has been proposed that the coiledcoil/zinc-hook domain of the MR complex tethers the ends of a DSB to the homologous sister chromatid that will be used as a template for homologous recombination-dependent DNA repair (26). Some models have multiple MR complexes bridging the two ends of the DSB to each other, as well as tethering the broken strands to the undamaged template (24). Genetic studies using the T4 phage system are consistent with this tethering arrangement. In T4, the two ends of the DSB are repaired in a coordinated manner so that a single homologous template is used to repair both ends of the DNA, even in the presence of a large number of available templates (27,28). It is presumed that this end coordination is achieved through the physical tethering of the DSB ends to each other, likely through the coiledcoils of the MR complex. Depletion of T4 Rad50 results in reduced end coordination so that each end of the DSB invades different homologous templates (28,29).
In addition to the role of the coiled-coil in bridging DNA strands, structural studies have suggested that the interaction between the CTD of Mre11 and the base of the Rad50 coiledcoil is a major route of communication between Rad50 and Mre11 (11,12). The Rad50 coiled-coil is connected to the ATP active site through what has been termed the "signature coupling helices." ATP binding at the dimer interface of Rad50 causes the N-terminal domain to rotate, which causes the p-helix wedge to move between the signature coupling helices and spread them apart (11,30). This movement is linked to the repositioning of the coiled-coil, which in turns alters the position of the Mre11 CTD by nearly 30 Å (11). As the CTD is connected to the main body of Mre11, it is reasonable to propose that the large movement of the CTD is transmitted through the flexible linker and cap domain to the Mre11 active site (30,31).
To further clarify the function of the T4 Rad50 coiled-coil, we have investigated the in vivo and in vitro effects of its disruption by various means. Our purified Rad50 contains one tightly bound Zn 2ϩ cation per dimer, and its removal via mutation of the zinc-hook has drastic effects. In vivo, the loss of Zn 2ϩ binding is lethal and in vitro it broadly affects the enzymatic properties of the MR complex. Similarly, disruption of the coiled-coil via proline insertion or the introduction of a disulfide cross-link affects both ATPase and nuclease activity. Combining our results with an analysis of the Rad50 x-ray crystal structures in the apo and ATPbound forms suggests the position of the coiled-coils may differentially regulate the ATPase and nuclease activities of Rad50 and Mre11, respectively.

Experimental Procedures
Analysis of CXXC Mutants in Vivo-We tested whether mutations in the CXXC motif of T4 Rad50 cause lethality as follows. Derivatives of plasmid pBR322 were constructed containing an EcoRI fragment of the T4 46 gene (Rad50) using an overlap PCR strategy with appropriate primers. The PCR fragments were cleaved with EcoRI and ligated to EcoRI-linearized, phosphatase-treated pBR322 vector, generating the following protein/DNA sequences at the zinc-hook motif: 1) wild-type CPTC (TGT CCA ACC TGT); 2) SPTC (TCA CCA ACC TGT); 3) CPTS (TGT CCA ACC AGT); and 4) SPTS (TCA CCA ACC AGT). The identities of the plasmids were verified by DNA sequencing. Next, DH5␣ carrying each of the four plasmids was infected with T4 tdSG2-46 am2 , a phage derivative carrying dual amber mutations in gene 46 at a multiplicity of infection of 3 (29). After a 90-min incubation at 37°C, the phage lysates were harvested. Marker rescue from the plasmid can replace the dual-amber mutation region of gene 46 with the corresponding region of the plasmid, which contains the coding sequence of the CPTC region. The titer of gene 46am ϩ recombinants was determined by plating appropriate dilutions of each lysate on nonsuppressing Escherichia coli strain AB1 and the total titer by plating on supD-containing strain CR63.
Plasmid Construction and Mutagenesis-The Rad50 ⌬coil construct was generated using a standard overlap PCR method to delete the coiled-coil and insert the codons for the "GGSEGGGSGEGGGTGS" linker between the N-and C-terminal domains of Rad50. The final PCR product was then subcloned into the pET28b expression vector. The C-terminal domain of T4 Mre11 (residues 290 -339, Mre11 CTD , where Mre11 CTD is a 49-amino acid peptide containing residues 290-339 of Mre11) was amplified using PCR and subcloned into the pTYB1 vector (15). The mutants Rad50 2CS (C288S/C291S), Rad50 2KP (K210P/K351P), and Rad50 xlink (C312S/S183C) were generated using the Stratagene QuikChange mutagenesis strategy. Rad50 C312S was also generated and subjected to complete biochemical characterization. All constructs were verified by DNA sequencing.
Expression and Purification of Mre11 and Rad50 Proteins-T4 Mre11 and Rad50 were expressed and purified as described previously (15,33). Purified proteins were concentrated to 60 -200 M. The extinction coefficients for determining the concentration of T4 Mre11, Mre11 CTD , Rad50, and Rad50 ⌬coil are 69,130, 3,840, 33,140, and 20,340 cm Ϫ1 M Ϫ1 , respectively. Aliquots were flash-frozen with liquid nitrogen and stored at Ϫ80°C. SDS-PAGE analysis was carried out using standard protocols (34) with 8% polyacrylamide/bisacrylamide at a 37.5:1 ratio. The gels were scanned T4 Rad50 Homolog Coiled-coil Domain using a Bio-Rad ChemiDoc TM XRSϩ system and analyzed using the ImageJ software (35).
Atomic Absorption Spectroscopy-The molar ratio of Zn 2ϩ to Rad50 was measured by atomic absorption spectroscopy. Rad50 samples were dialyzed against a buffer solution of 400 mM sodium chloride and 20 mM Tris-HCl (pH 8.0) and diluted to 5 and 10 M or 6 and 12 M. Assuming a molar ratio of one Zn 2ϩ ion per Rad50 dimer, these concentrations were predicted to fall within the range of standard solutions of 0.1-1 g/ml zinc (1.5-15.3 M) used to calibrate the instrument. The zinc concentration of each sample was measured on a 55B AA atomic absorption spectrometer (Agilent Technologies) with a 213.9 nm lamp. Measurements were made in triplicate, and the molar ratio of Rad50 molecules to zinc atoms was calculated for each sample.
Size Exclusion Chromatography-Multiangle Light Scattering (SEC-MALS)-Protein samples subjected to SEC-MALS analysis were injected into a Superdex TM 200 10/300 GL column in tandem with a Dawn Helios II multiangle light scattering detector (Wyatt Technology, Santa Barbara, CA) at a flow rate of 0.5 ml/min. All buffers contained 20 mM Tris-HCl (pH 7.9) and 400 mM sodium chloride. Buffers containing 1 mM ATP also contained 5 mM MgCl 2 . Samples containing Rad50 S183C/C312S in its reduced state were run using the standard buffer with 10 mM dithiothreitol (DTT). Data were analyzed using ASTRA 6 software (Wyatt Technology).
ATP Hydrolysis (ATPase) Kinetics-Steady-state ATPase activity was measured at 30°C using a coupled assay linking ADP production to NADH oxidation (36). Each 300-l reaction contained 50 mM Tris-HCl (pH 7.6), 50 mM KCl, 5 mM MgCl 2 , 0.1 mg/ml BSA, 0.17 mM NADH, 1 mM phosphoenolpyruvate, pyruvate kinase (1.8 units), and lactose dehydrogenase (3 units) (Sigma). When present, Mre11 was kept at a slight excess over Rad50 (1.2-fold), and the DNA concentration used was at least 4-fold over the enzyme complex concentration. Reactions were started with the addition of Rad50 or MR complex, and the rate of change in NADH fluorescence was monitored at an emission wavelength of 460 nm (excitation wavelength of 340 nm) on a Cary Eclipse spectrofluorometer (Varian). Estimates of kinetic constants (k cat , K m , and Hill coefficient) were determined by fitting reaction velocities at various ATP concentrations to the Hill equation using SigmaPlot 10.0/Enzyme Kinetics Module 1.3 (Systat Software, Inc.).
DNA K d Determination-The equilibrium binding constant of Rad50 for DNA was determined by the fluorescence anisotropy of the 50-bp DNA substrate 5Ј-labeled with hexachlorofluorescein (DNA HEX ). The dsDNA concentration was held constant at 10 nM, and the concentration of added Rad50 protein was varied from 0 to 12 M in 50 mM Tris-HCl (pH 7.6), 50 mM KCl, and 5 mM MgCl 2 . Anisotropy was measured on a Biotek Synergy Two 96-well plate reader with an excitation wavelength of 535 nm and an emission wavelength of 560 nm. Data were fit to a bimolecular equilibrium model using Dynafit (37).
Nuclease Assays-The nuclease products of Mre11-Rad50 complex were analyzed by denaturing PAGE. Each reaction contained 50 mM Tris-HCl (pH 7.6), 50 mM KCl, 5 mM MgCl 2 , 0.3 mM MnCl 2 , 0.1 mg/ml BSA, and 1.3 M DNA HEX and was started with the addition of 400 nM MR complex. When included, the ATP concentration was 1 mM. At specified time points, aliquots of the reaction were quenched in an equal volume of 100 mM EDTA, 50% formamide solution. Two microliters of each quenched solution were analyzed using 16% urea-PAGE (19:1 acrylamide/bisacrylamide). The gels were imaged using a Typhoon PhosphorImager with an excitation wavelength of 532 nm and a 555-nm bandpass filter for emission detection.
Position-specific nuclease activity was monitored by the release of the fluorescent adenine analog, 2-aminopurine (2AP) (38). A 50-bp dsDNA was constructed from 50 base oligonucleotides labeled at the 1-or 17-position from the 3Ј end with 2AP. The intrinsic fluorescence of this nucleotide is quenched by base stacking, which is then relieved upon cleavage of the nucleotide from the DNA. Each 300-l reaction contained 50 mM Tris-HCl (pH 7.6), 50 mM KCl, 5 mM MgCl 2 , 0.3 mM MnCl 2 , 0.1 mg/ml BSA, 1.3 M 2AP of labeled DNA, and 0 or 1 mM ATP and was initiated by the addition of MR complex. Emission was monitored at 375 nm with an excitation wavelength of 310 nm in a Cary Eclipse spectrofluorometer (Varian). When 2AP was located at the 17-position relative to the 3Ј end, the activity was measured over the initial 5 min of the reaction.

Analysis of CXXC Mutants in Vivo-
We constructed a series of four plasmids containing a 1-kb segment of T4 DNA with the CXXC-encoding region near the middle. In addition to the WT control, which encodes CPTC, we constructed otherwise identical plasmids with SPTC, CPTS, and SPTS. We then used a plasmid marker rescue approach to determine whether the single or double substitutions created a viable phage. In this approach, a conditional lethal gene 46 amber double-mutant phage infected the plasmid-bearing strains, and viable recombinants were selected on a nonsuppressing host. The dual gene 46 amber mutations are very near the CPTC-encoding region. Therefore, when plasmid-phage recombination replaces the amber mutations with plasmid-borne sequence to generate non-amber phage, the plasmid-borne CPTC region is usually also incorporated into the phage genome. With the WT plasmid, viable phage were generated at a frequency of about 2 ϫ 10 Ϫ4 , but each of the three mutant plasmids generated many fewer recombinants. The low frequency of viable phage from the mutant plasmids (ϳ10 Ϫ5 ) can be explained by multiple crossovers generating 46 ϩ phage. We sequenced four plaques arising from each of the three mutant plasmids, and all 12 were WT for the amber mutations and for the DNA encoding CPTC. We conclude that the fairly conservative cysteine to serine substitution at either location of the CXXC motif of T4 Rad50 results in a lethal mutation, i.e. completely or severely compromises the function of the T4 MR protein complex.
Rationale for Mutants Generated for Biochemical Characterization-The relative positions of the mutants that were generated and purified for this study are shown in Fig. 1, A and B. The Rad50 2CS mutation was created to eliminate Zn 2ϩ binding by the zinc-hook and corresponds to the SPTS construct used in the phage infections described above. The two cysteine residues that are mutated to serine in the Rad50 2CS protein are T4 Rad50 Homolog Coiled-coil Domain SEPTEMBER 25, 2015 • VOLUME 290 • NUMBER 39 absolutely conserved in all Rad50 proteins and were previously known to affect the in vivo properties of the MR complex in other systems (22,39,40).
Rad50 2KP was generated to disrupt the structure of the coiled-coil at a position that is approximately midway between the NBD and the zinc-hook. Fig. 1B shows a coiled-coil probability calculation for Rad50 WT and Rad50 2KP (41). This analysis predicts that the replacement of nonconserved lysine residues on the ascending and descending arms of the coiled-coil with proline will completely interrupt the structure.
The Rad50 S183C protein was originally generated to measure the distance and/or movement of the coiled-coil using bifunctional cysteine-reactive cross-linkers. The serine at position 183 is poorly conserved and is positioned between the Rad50 NBD and the midway point of the ascending arm of the coiledcoil (Fig. 1). To specifically label Rad50 S183C , it would be necessary to mutate the single reactive cysteine that is present in T4 Rad50 WT , generating the Rad50 C312S mutant. Assays using Ellman's reagent confirmed the loss of the reactive cysteine in Rad50 C312S , and a complete biochemical characterization indicated no significant differences relative to Rad50 WT (data not shown). Although not explicitly stated hereafter, the Rad50 S183C and Rad50 xlink proteins contain the C312S mutation. Following the purification of Rad50 S183C , it was discovered that the mutant was unsuitable for cysteine labeling because a spontaneous disulfide bond forms during the purification process (see below).
Protein Purification-All proteins expressed to approximately the same degree and displayed similar affinity for the phosphocellulose column (binding at 200 mM NaCl and eluting at 400 mM). Each protein was purified a minimum of two times, and the biochemical properties of repeated protein preparations were not significantly different (i.e. less than ϳ25% change in all kinetic and binding constants). SDS-PAGE analysis indicated that the proteins were at least 90% pure; however, Rad50 S183C appeared as two distinct bands when DTT was omitted from the loading buffer, indicating the formation of a disulfide bond (Fig. 2A). The mobility of the cross-linked protein in relation to the standards suggests that the molecular mass of the protein is near 200 kDa, which is most consistent  (20) and the Thermotoga maritima MR complex (12) (Protein Data Bank codes 1L8D and 3THO, respectively). The remainder of the coiled-coil is represented by the thin lines separating the head domain from the zinc-hook. The length of this section is proportional to the number of amino acids predicted to be in the coiled-coil of T4 Rad50. B, plot of coiled-coil probability versus the amino acid sequence of T4 Rad50 WT (green) and Rad50 2KP (blue). The probabilities were determined using the COILS server with a window width of 28 amino acids (41).

T4 Rad50 Homolog Coiled-coil Domain
with a Rad50 trimer; however, both MALS (see below) and MALDI-TOF mass spectrometry (data not shown) indicate that the cross-linked species is dimeric. The cause of the aberrant mobility is unclear. Densitometric analysis of the protein indicates that ϳ78 Ϯ 9% of the protein is in the cross-linked form (Fig. 2B). The nonreduced version of Rad50 S183C will be referred to hereafter as Rad50 xlink to indicate that it is predominantly in the cross-linked state. Although the analysis indicates that ϳ20% of the Rad50 xlink is in the reduced state, the biochemical parameters for both ATPase and nuclease activity indicate that the observed activity is dominated by the crosslinked fraction (e.g. K m -ATP for Rad50 xlink is 10-fold elevated compared with Rad50 WT , see Table 4).
The WT and mutant proteins were then subjected to SEC-MALS analysis to determine their oligomeric state. Rad50 WT , Rad50 2CS , Rad50 xlink , and Rad50 S186C eluted as a single peak (Fig. 3A), and the MALS analysis indicated that the proteins were in the dimeric form (Table 1). This indicates that the disulfide cross-link in Rad50 xlink is intradimer, rather than between two or more Rad50 dimers, which would elute as a tetramer or higher oligomeric state. The elution profiles of Rad50 2KP and Rad50 ⌬CC are significantly different from the other Rad50 proteins (Fig. 3A). MALS analysis of the major peak for both of these proteins indicates that they are monomeric. Inclusion of ATP in the running buffer decreases their elution time and reproducibly increases their apparent molecular weight as determined by MALS (Table 1). This suggests that ATP binding promotes the dimerization of these mutants, but the monomerdimer interconversion occurs in fast exchange relative to the time scale of the experiment (42). Because the apparent molecular weight is closer to that of the monomer, the equilibrium must lie in that direction.
Zn 2ϩ Analysis-All Rad50 proteins were extensively dialyzed against three 2-liter changes of buffer prior to analysis using atomic absorption spectroscopy (Table 2). Rad50 WT and Rad50 xlink were found to contain approximately one Zn 2ϩ cation per Rad50 dimer, whereas Rad50 2KP and Rad50 2CS contained negligible concentrations of Zn 2ϩ when compared with a buffer control.
DNA and Mre11 Binding-The equilibrium dissociation constant of the Rad50 proteins for dsDNA (K d -DNA) was measured using fluorescence anisotropy (Table 3). Similar to the Pyrococcus furiosus (Pfu) system, removal of the coiled-coil from T4 Rad50 results in a 100-fold increase in K d -DNA, indicating a severe reduction in dsDNA binding affinity (43). Similarly, the Rad50 2KP and Rad50 2CS mutants also display reduced   Table 1. With the exception of Rad50 2KP and MR 2KP , the major peak of the elution profile was used in the analysis. In the case of Rad50 2KP and MR 2KP , the molecular masses from the three peaks indicated were used. The numbers shown above the peak correspond to the numbers given in column 2 of Table 1. SEPTEMBER 25, 2015 • VOLUME 290 • NUMBER 39 affinity, with 20-and 4.5-fold increases in K d -DNA, respectively. In contrast, the dsDNA affinity for Rad50 S183C and Rad50 xlink is nearly unchanged compared with Rad50 WT .

T4 Rad50 Homolog Coiled-coil Domain
The ability of the Rad50 proteins to bind to Mre11 was assessed by SEC-MALS. An excess of Rad50 was incubated with Mre11 for several minutes at room temperature prior to separation via SEC and analysis by MALS (Fig. 3B and Table 1). Although T4 Mre11 alone is monomeric, two subunits bind to dimeric Rad50 to form a heterotetrameric complex (38,44). With the exception of Rad50 ⌬coil , all Rad50 mutants bind Mre11, as measured by the disappearance of the Mre11 peak and the formation of a peak that elutes earlier than Rad50 alone. The MR complexes formed using Rad50 WT , Rad50 S183C , Rad50 xlink , and Rad50 2CS elute at the expected time relative to the elution of standard proteins (data not shown), and MALS analysis indicates that the molecular masses of the eluted complexes are consistent with a heterotetramer (190 -200 kDa, Table 1). Interestingly, with Rad50 2KP , the Mre11 peak is absent, indicating MR complex formation. However, three peaks are observed, and the MALS analysis indicates that both heterotetramers and heterodimers are formed along with monomeric Rad50 2KP . This indicates that Rad50 dimer formation is not a prerequisite to Mre11 binding. Similar to Rad50 2KP alone, inclusion of ATP in the running buffer causes a slight decrease in the elution times of heterodimeric MR 2KP and monomeric Rad50 2KP , but it does not affect the heterotetrameric MR 2KP.
ATPase Kinetics-The kinetic constants for the ATPase reaction catalyzed by the isolated Rad50 mutants are relatively similar to Rad50 WT (Table 4). This indicates that none of the mutants generated here significantly perturb the overall fold of the Rad50 NBD. In contrast to Rad50 in isolation, MR 2CS , MR 2KP , and MR xlink show large differences compared with Rad50 WT when Mre11 and DNA are present. Mre11-Rad50 ⌬coil was not tested because the MALS analysis indicates that it does not interact with Mre11. Rad50 WT shows a 30-fold increase in ATPase rate upon complex formation with Mre11 and DNA, whereas Rad50 2CS , Rad50 2KP , and Rad50 xlink are unaffected by their presence ( Table 4). The mutant complexes also display large decreases in cooperativity, with Hill coefficients of 1.2, 1.4, and 1.3 for MR 2CS , MR 2KP , and MR xlink , respectively, compared with a Hill coefficient of 2.5 for MR WT . Because the ATPase activity of MR xlink is significantly reduced compared with MR WT , we measured the effect of DTT addition to an ongoing reaction. As seen in Fig. 4A, the addition of DTT to the ongoing the reaction results in an immediate increase in the ATPase rate. To determine the kinetics of disulfide formation following reduction, MR xlink was reduced with DTT   and then diluted into the ATPase assay buffer that lacked ATP but contained oxidized glutathione in excess of final DTT concentration. ATP was then added to the reaction at various time points, and the initial rate of ATPase activity served as a reporter of the oxidation state of Rad50 S183C . As seen in Fig. 4B, the rate of ATPase inactivation followed an exponential curve with a rate constant of 1.3 Ϯ 0.38 min Ϫ1 . The progress curves were linear for up to 15 min following ATP addition, suggesting the presence of ATP inhibits further disulfide formation (Fig. 4C). It is clear that alterations in the Rad50 coiled-coil directly affect the ATP active site of the NBD. Because the primary linkage between Mre11 and Rad50 is through the CTD of Mre11 and the base of the coiled-coil of Rad50, we examined the effect of the isolated Mre11 CTD (Mre11 CTD ) on the ATP activity of Rad50. As seen in Fig. 5, the ATPase activity of Rad50 in complex with DNA and Mre11 CTD resembles that of the complete MR WT complex, with an 8-fold increase in k cat -ATP and a Hill coefficient of 1.8 ( Table 4). The addition of DNA to Rad50 does not result in activation of ATPase activity (15) indicating that the activation requires a minimal addition of DNA and the CTD of Mre11 (Table 4).
Nuclease Activity-To quantitatively determine the impact of the Rad50 mutations on the nuclease activity of the MR complex, we measured the rate of 2AP excision from the 1st and 17th positions relative to the 3Ј end of the DNA (Table 5) (38). In the absence of ATP and with 2AP located at the 1st position, MR WT and all mutants exhibit similar steady-state kinetics with apparent k cat values ranging from 9.2 to 4.1 min Ϫ1 . When the probe is moved to the 17th position, ATP is now required for activity, and only MR S183C behaves like MR WT (apparent k cat of 0.2 min Ϫ1 ). The nuclease activities of MR 2CS , MR 2KP , and MR xlink are 26-, 9-, and 13-fold reduced compared with MR WT (Table 5).
To visually characterize the nuclease activity of Mre11 in complex with the Rad50 mutants, we analyzed the products of the nuclease assay using denaturing PAGE. Fluorescently labeled dsDNA was incubated with the MR complex, and samples of the reaction were quenched at time points spanning the course of 60 min (Fig. 6). In the absence of ATP, MR WT , MR 2CS , and MR 183C have very similar activities, whereas MR 2KP and  MR xlink are significantly reduced in their nuclease activity. A close inspection of the gel indicates that for both MR 2KP and MR xlink , a single excision occurs relatively efficiently, but the rate of subsequent excisions is extremely reduced (especially so with MR xlink ). The addition of ATP to the MR WT complex reaction strongly activates its nuclease activity, with the appearance of short DNA products appearing at the earliest time points (Fig. 6). This is indicative of a highly processive reaction that includes a single rate-determining slow step prior to the onset of nuclease activity (expanded discussion can be found in Ref. 38). As expected, the ATP-dependent nuclease activity of MR S183C is identical to MR WT . In contrast, ATP reduces the nuclease activity of MR 2CS , MR 2KP , MR xlink as compared with reactions in the absence of ATP. This behavior is often observed in mutant MR complexes with low ATPase activity or in the presence of nonhydrolyzable ATP (15,44,45). The inhibition may be due to the inability to dissociate from or translocate forward on the DNA substrate (i.e. the complex becomes stuck), or alternatively, slow ATP hydrolysis may prevent the binding of the complex to recessed ends.

Discussion
The function of the Rad50 coiled-coil has been the subject of intense research (22,24,25,40,43). Previous in vivo experiments using the yeast model system indicate that the presence of the zinc-hook and a full-length coiled-coil are critical to the proper functioning of the MR complex (22,40,46). Here, we show that this is also true in the T4 system, as mutation of the zinc-hook cysteine residues is lethal, indicating that the ability to form a complex capable of DNA tethering is necessary for phage survival. In addition to these in vivo studies, electron microscopy (EM) and atomic force microscopy experiments have strongly supported a model where the coiled-coil and zinc-hook play an architectural role by physically tethering DNA strands to promote either homologous recombination or DNA end-joining (20,(23)(24)(25).
The effects of altering the zinc-hook and coiled-coil in vivo have most often been interpreted under the assumption that the enzymatic properties of the MR complex are unaffected, although recent work has suggested communication between the coiled-coil/zinc-hook and the globular domain (30,40). The data presented here directly demonstrate this functional crosstalk. The coiled-coil is physically linked to the Rad50 active site through the signature coupling helices (11,30). The mutants generated here, including the complete coiled-coil deletion, are largely unaffected in ATPase activity when measured in the absence of Mre11 and DNA, suggesting that there is minimal coupling between the coiled-coil and the ATP active site in the uncomplexed Rad50 dimer. In contrast, when Mre11 and dsDNA are bound, the disruption of the coiled-coil results in a loss of ATPase activation and a large decrease in cooperativity. These results suggest that the combined binding of Mre11 and DNA alter the position or reduce the flexibility of the coiledcoil, and these changes are transmitted allosterically to the ATP active site. As suggested by Petrini and co-workers (40), the coupling of the coiled-coil to the ATPase active site likely enables tethering-dependent regulation of ATP-dependent nuclease activity.
Eliminating Zn 2ϩ binding by the zinc-hook through mutation (Rad50 2CS ) appears to have no effect on the stability of the Rad50 dimer (Fig. 3A). This was somewhat unexpected, as other Rad50 homologs appear to have a relatively weak dimer interface in the absence of ATP (11,12,30). X-ray crystallography of an MR complex with the zinc-hook and most of the coiled-coil removed revealed an open form of the complex where the Rad50 subunits are separate and the Mre11 dimer interface maintains the heterotetrameric state of the complex (12). SAXS and disulfide cross-linking experiments using P. furiosus Rad50 suggest an equilibrium exists between open and closed forms of the MR complex in the absence of ATP (11,30). This type of equilibrium may be unlikely in the T4 system because Rad50 2CS is dimeric without ATP, indicating stable dimer formation does not require the zinc-hook. Additionally, unlike eukaryotic and archaeal Mre11, T4 Mre11 is monomeric in the absence of Rad50; therefore, it is difficult to envision the Mre11 dimer interface holding the MR complex in the open state (38).  Table 4.  The removal of the entire coiled-coil results in a monomeric protein (Fig. 3A), indicating that the coiled-coil itself and not solely the zinc-hook is an important factor for dimer stabilization. The presence or absence of the coiled-coil is likely communicated to the dimer interface via the signature coupling helices (11). This communication can also be seen in the Rad50 2KP mutant. The coiled-coil perturbing proline residues appear to alter the conformation of Rad50 in the direction of the zinc-hook (because Zn 2ϩ is poorly bound) and in the direction of the NBD (because ATPase activity is affected and dimerization is disrupted). The effect of the coiled-coil on Rad50 dimerization may explain the possible differences between the T4 MR complex and other homologs with regard to potential large scale changes in quaternary structure. T4 Rad50 has a relatively short coiled-coil (ϳ250 amino acids) when compared with prokaryotes (ϳ500), archaea (ϳ600), and eukaryotes (ϳ900). The longer coiled-coil in these other systems may allow Rad50 to adopt additional conformational states, including the open state seen in crystal structures and solution (12,47,48).

17-Position 2AP (؉)ATP
Perhaps the most unexpected result from this study was the spontaneous formation of a disulfide bond between the coiledcoils of the Rad50 dimer. This indicates that the coiled-coils come into close proximity, likely by adopting a parallel conformation. This parallel conformation can be seen in an unliganded Rad50 crystal structure (Protein Data Bank code 1II8 (45)) and in several EM and atomic force microscopy imaging studies (20,24,25,32). These imaging experiments indicate that the coiled-coil is very dynamic, but it appears that the MR complex primarily adopts three conformations (20,25). In the first conformation, the zinc-hooks from each Rad50 subunit come together in an intradimer fashion so that the coiled-coils form a ring-like structure. In the second conformation, the two coiledcoils are parallel and extended out from the NBD so that they are unresolvable and appear as a single rod. In the third conformation, the coiled-coils are again parallel, but the zinc-hooks participate in an inter-Rad50 dimer interaction so that the MR complex is hetero-octomeric. The latter two conformations would allow for the intra-dimer cross-linking of the coiled-coils that is seen in Rad50 xlink .
The reformation of the disulfide bond following reduction of Rad50 xlink is relatively rapid, suggesting that the parallel conformation of the coiled-coils is a highly populated, or at a minimum the interconversion between the parallel and ring-like states occurs at a fast rate (regardless of where the equilibrium lies). We also observed that ATP binding and/or hydrolysis prevents further disulfide formation, as the ATPase time courses under oxidative conditions are linear over a course of several minutes. This suggests that bound ATP shifts the population of Rad50 conformers away from the parallel state, likely by stabilizing the ring-like state. The severely perturbed ATPase activity of Rad50 xlink is consistent with this notion, as Rad50 xlink has a low affinity for ATP and is a poor ATPase. The nuclease activity of Rad50 xlink is consistent with defects in ATP binding and hydrolysis, as it is only able to perform nonprocessive nucleotide excisions. The 2AP-containing substrates confirm this, as the nuclease activity with the probe at the 1-position is nearly normal, whereas the 17-position is severely reduced.
Based on the enzymatic properties of the Rad50 xlink , we propose that the coiled-coils of the T4 MR complex alternate between parallel and ring-like states during its catalytic cycle. In this model, ATP binding and hydrolysis by Rad50 requires the ring-like state, whereas nucleotide excision by Mre11 occurs when the coiled-coils are in the parallel conformation (Fig. 7). Because Mre11 is connected to the coiled-coils via its CTD, the movement of the coiled-coils into the parallel conformation may drive the opening of the Mre11 dimer interface. This opening would then allow access to the otherwise protected nuclease active site (12,31). This is consistent with our previous finding that disruption of the Mre11 dimer interface enhances the nuclease activity of the MR complex (38). The two states proposed here correspond well to two of the three conformational states proposed by Deshpande et al. (30) for the globular domain of the P. furiosus MR complex. In that model, ATP stabilizes a closed state that is capable of binding to dsDNA ends but has poor nuclease activity. ATP hydrolysis then promotes the formation of an intermediate state that is nucleasecompetent. The third state proposed by Deshpande et al. (30) is an ATP-free open state that, as discussed above, may be unlikely in the T4 MR complex.
To alternate between the ring-like and parallel states, the zinchook region must also adopt different conformations. The structure of the zinc-hook from P. furiosus Rad50 shows the coiled-coils on either side of the Zn 2ϩ at an angle of about 140°relative to each other (20). This orientation corresponds well to the ring-like state; however, it is difficult to envision this angle being retained in the parallel state. It is likely that the short unstructured regions between the end of the coiled-coil and the invariant cysteine residues of the zinc-hook act as a hinge that allows the coiled-coil to adopt either conformation.
The experiments described here have led us to suggest that the T4 MR complex may function somewhat differently from other MR complex homologs. Specifically, the properties of the T4 MR complex are not consistent with the open state that involves noninteracting Rad50 subunits. This difference may be the result of the shorter coiled-coil that presumably reduces the number of possible conformational states that the MR complex can adopt. It may, however, be due to differences in the preparation of purified proteins. We have found that purified T4 Rad50 contains a tightly bound Zn 2ϩ cation that cannot be removed with extensive dialysis. EM imaging of human and P. furiosus MR complexes suggests that the affinity for Zn 2ϩ is lower in these Rad50 homologs as compared with T4 and some of the conformational states that are observed may be due to the lack of Zn 2ϩ bound to the zinc-hook (32). The biochemical and biophysical characterization of full-length MR complexes with bound and unbound Zn 2ϩ from a wide variety of homologs will likely be necessary to determine whether there is a unified mechanism for all MR complexes.