A network of allosterically coupled residues in the bacteriophage T4 Mre11–Rad50 complex

Abstract

The Mre11–Rad50 (MR) protein complex, made up of a nuclease and ATPase, respectively, is involved in the processing of double‐strand breaks as part of an intricate mechanism for their repair. Although it is clear that the MR complex is subject to allosteric regulation and that there is communication between the nuclease and ATPase active sites, the underlying mechanisms are poorly understood. We performed statistical coupling analysis on Mre11 and Rad50 to predict linked residues based on their evolutionary correlation. This analysis predicted a coevolving sector of six residues that may be allosterically coupled. The prediction was tested using double‐mutant cycle analysis of nuclease and ATPase activity. The results indicate that a tyrosine residue located near the active site of Mre11 is allosterically coupled to several Rad50 residues located over 40 Å away. This allosteric coupling may be the basis for the reciprocal regulation of the ATPase and nuclease activities of the complex.

Abbreviations and symbols

2AP

2‐aminopurine

ABC

ATP binding cassette

CTD

Mre11 C‐terminal domain

DSB

double‐strand break

dsDNA

double‐stranded DNA

MALS

multiangle light scattering

MR

Mre11–Rad50 complex

Pfu

Pyrococcus furiosus

SCA

statistical coupling analysis

SEC

size exclusion chromatography

Tm

Thermotoga maritima

Introduction

The Mre11 and Rad50 proteins form a heterotetrameric complex that is involved in the initial steps of double‐strand break (DSB) processing.1 DSBs are among the most harmful form of DNA damage, and their accurate repair helps to prevent genomic rearrangements and cellular dysfunction.2 Homologs of Mre11 and Rad50 are found in every domain of life, including many dsDNA bacteriophage.3

The Mre11 nuclease is a member of the protein phosphatase super‐family and utilizes two active site Mn²⁺ cations to carry out its enzymatic activity.4 Mre11 is made up of three domains, a core nuclease domain, a capping domain, and a C‐terminal domain (CTD) that binds to the coiled‐coil of Rad50.5 A negatively charged flexible linker that is involved in the regulation of nuclease activity connects the CTD to the capping domain.6 In addition to the CTD, Mre11 makes minimal contact with Rad50 at three other interfaces, although the CTD is necessary and sufficient for complex formation.7 The active form of Mre11 is a dimer and the interface between Mre11 subunits appears to be dynamic and capable of adopting several different conformations.4, 8, 9, 10, 11

Rad50 ATPase is a member of the ATP binding cassette (ABC) protein super‐family, which is defined by six conserved motifs that come together to form two shared ATP active sites across the dimeric interface.12 Found between the ABC protein motifs is a domain that forms an intramolecular antiparallel coiled‐coil containing a conserved CXXC motif at its apex (referred to as the zinc‐hook).13 The two cysteines of the CXXC motif dimerize with a second CXXC to bind a Zn²⁺ divalent cation in a tetrathiolate linkage.13 The zinc‐hook is thought to be involved in Rad50‐dependent DNA tethering, although recent results have demonstrated that some level of DNA tethering can occur even in the absence of the coiled‐coil/zinc‐hook domains.3, 14

In the absence of Rad50, T4 Mre11 is a poor nuclease and has a low affinity for DNA.15 The low affinity for DNA appears to arise from an autoinhibition mechanism that was recently uncovered.6 The linker that separates the cap domain from the CTD is highly negatively charged and may be acting as a DNA mimic. Removal of the linker and CTD dramatically lowers the Km‐DNA of the nuclease reaction and results in a nuclease k _cat similar to that of the MR complex in the absence of ATP. Upon binding, Rad50 sequesters the CTD of Mre11, thereby disengaging the autoinhibitory interaction between the acidic linker and Mre11.

Mre11 and Rad50 form a heterotetrameric complex with two subunits of Rad50 and two subunits of Mre11. In the bacteriophage T4 system, the ATPase activity of Rad50 is increased by approximately 20‐fold upon complex formation with Mre11 and binding of dsDNA.15 The structural basis for the increase in Rad50 ATPase activity is unclear. It is not directly related to the nuclease activity of Mre11, as the activation does not require the presence of Mn²⁺ and nuclease deficient Mre11 mutants are not defective in the activation.15 Some Rad50 homologs are known to possess a weak helicase activity,16, 17 but thus far, this activity has not been detected in the bacteriophage T4 system. In the absence of nuclease activity, dsDNA can be considered a V‐type allosteric activator of MR complex ATPase activity, as the Km‐ATP for the MR complex with and without DNA is very similar.15 In the presence of Mn²⁺, ATP hydrolysis by the MR complex is necessary for processive translocation along the DNA substrate and an average of four nucleotides are removed for every one ATP hydrolyzed.15 Nonhydrolyzable ATP (AMP–PNP) strongly inhibits nuclease activity, and gel analysis indicates that AMP–PNP allows a single nucleotide to be excised, but then the MR complex becomes “stuck” on the DNA product and is unable translocate to the next nucleotide or dissociate from the DNA.15 In the absence of ATP, the MR complex nonprocessively degrades the DNA, removing a limited number of nucleotides per binding event.8 It has been hypothesized that ATP binding and/or hydrolysis acts as an allosteric regulator of the nuclease activity of Mre11.5, 7, 18, 19 The primary interaction between Mre11 and Rad50 is between the CTD of Mre11 and the base of the Rad50 coiled‐coiled domain. It appears that this interaction is an important conduit between Mre11 and Rad50 subunits, as the isolated Mre11 CTD increases the ATPase activity of Rad50 by 8‐fold.20

To test the hypothesis that Rad50 allosterically regulates the nuclease activity of Mre11 and vice/versa, we sought to identify amino acid residues that participate in the communication between the distal ATPase and nuclease active sites. To assist in this process we employed the technique of statistical coupling analysis (SCA).21 SCA is a bioinformatic method that identifies pairs of residues that are evolutionarily coupled. The basic assumption is that residues that display an evolutionary covariance, outside of simple evolutionary history, are energetically coupled for function and/or structure. SCA has been applied to a wide variety of proteins and in many instances, a network of energetically coupled residues has been revealed.22, 23, 24, 25, 26 Often these residues are located on known allosteric pathways or the predictions of SCA have suggested residues for mutation either alone or as a part of a double‐mutant cycle. Application of SCA to the MR complex has revealed a sparse network (8.5% of all amino acids in alignment) of coupled residues that can be divided into six sectors using principal component analysis.27 One of the sectors contained residues near both the nuclease and ATPase active sites and this sector was chosen for double‐mutant cycle analysis.28 Four out of five of the Rad50 residues showed moderate amounts of coupling in either nuclease or ATP hydrolysis activity, with a ΔΔG of interaction values ranging from 0.5 to 1.8 kcal/mol. These data support the hypothesis that Mre11 and Rad50 are allosterically connected and that SCA successfully predicts allosterically coupled residues.

Results

Multiple sequence alignment and statistical coupling analysis

Mre11 and Rad50 represent excellent proteins on which to perform inter‐protein SCA. Mre11 and Rad50 are highly conserved and found in all domains of life, including some bacteriophage. To our knowledge, Mre11 and Rad50 are always found as a complex and their relevant biochemical and biological functions absolutely dependent on complex formation. The overall sequence alignment was generated using T‐Coffee,29 a progressive sequence alignment program, followed by 3D‐Coffee,30 a structure‐based alignment program, and then the alignment was manually inspected and refined based on biochemical knowledge of the MR complex. Finally, we obtained an alignment matrix with 772 sequences (172 from eukaryote, 437 from bacteria, 93 from archaea, and 68 from bacteriophage) and 787 amino acids positions (361 for Mre11 and 426 for Rad50).

Application of SCA to the alignment identified a small sub‐set of the total residues that appear to be coevolving (70 residues out of 787 positions). These 70 positions (23 from Mre11 and 47 from Rad50) were further divided to five sectors according to their distribution on a plane formed by the third and 5th eigenvectors (PC3–PC5).27 We then mapped the five sectors on to the structure of the T. maritima MR complex7 and the coiled‐coil domain of P. furiosus Rad5013 (Figs. 1 and 2). In general, the correlated residues do not show a contiguous network of coupled residues spanning the Mre11–Rad50 subunit interfaces. The sector with the fewest number of residues (sector 2, 6 residues) contained residues near the active sites of Mre11 and Rad50. For this reason, we choose sector 2 to perform double‐mutant cycle analysis28, 31, 32, 33 to test the predictions from SCA. We reasoned that because this sector contains residues near both active sites of the MR complex, it is possible that they are functionally coupled.

Figure 1.

Statistical coupling analysis of MR. (A) The cross correlation among 787 positions in MR complex was expressed color‐coded, with red indicating high level of correlation and blue indicating low level of correlation. The first 361 positions were from Mre11 and the following 426 positions came from Rad50. (B) Important correlations from the positional correlation matrix were extracted with principle component analysis. 70 residues in MR complex were found highly correlated with each other. (C) The above 70 residues were divided into five groups (so‐called sectors) according to their distribution on the plane formed by PC3–PC5. The sectors 1–5 were colored by green, blue, yellow, magenta, and cyan, respectively. (D) The correlation matrix of 70 residues was resorted according to sector definition. The sectors were indicated by corresponding color bars on the top of the matrix. (E) The conservation patterns of sector residues were mapped back to sequences of MR. The left side showed a sequence similarity matrix, with red color for high sequence similarity and blue for low similarity. The sequences came from bacteria, phage, archaea, and eukaryote, respectively. A position in a sequence was colored black only if the residue at that position was same as the most abundant amino acid at that position in all sequences. The sectors were indicated by corresponding color bars on the top of the distribution map.

Figure 2.

Distribution of sector residues on the structure of MR complex. (A) The sector residues and residues that are highly conserved were mapped to structure of T. maritime MR complex (bottom; PDB ID: 3THO) and coiled‐coil domain structure of P. furiosus Rad50 (top; PDB ID: 1L8D). Highly conserved residues are colored red, whereas residues from sectors 1–5 were colored green, blue, yellow, magenta, and cyan, respectively. (B) The residues of sector 2 mapped on to the structure of T. maritime MR complex (PDB ID: 3THO) with residues renumbered to match bacteriophage T4 amino acids. Highly conserved residues are colored red whereas residues from sector 2 are colored blue.

Protein expression, purification, circular dichroism, and MR complex formation

All of the mutant proteins expressed and purified normally and SDS‐PAGE analysis indicated they are at least 90% pure [Supporting Information Fig. S1(A)]. The Rad50 proteins were free from detectable nuclease contamination, and the Mre11 proteins were free from detectable ATP hydrolysis activity. The circular dichroism spectra of the mutant proteins were very similar to their wild‐type counterparts, suggesting that none of the mutations have disrupted the global structure of the protein (Supporting Information Fig. S2). Consistent with this, all mutant proteins were capable of forming the MR complex, as measured by size exclusion chromatography (SEC) elution times and multiangle light scattering (MALS) analysis (Supporting Information Table SII). In every case, the presence of a twofold excess of Mre11 resulted in the disappearance of the Rad50 peak and the appearance of a new peak that when analyzed with MALS was determined to have a molecular mass consistent with a heterotetrameric MR complex.

Steady state ATPase kinetics

Rad50 by itself is a relatively weak ATPase, with a k _cat of 0.1 s⁻¹ and a Km of 20 μM 15 (Table 1). The addition of Mre11_WT and dsDNA increases the ATPase rate and Km by approximately 25‐ and 2‐fold, respectively. We have previously shown that the enhancement in ATPase activity is not directly related to the nuclease activity of Mre11, as nuclease‐deficient mutants fully support activation.15 This suggests that the activation is of an allosteric nature, where the binding of Mre11 and dsDNA stabilizes a conformation of Rad50 that is more appropriately positioned for hydrolysis chemistry, which is the rate‐determining step of the reaction for Rad50 and the MR and MR‐DNA complexes.34 Replacement of Mre11_WT with Mre11_Y197A reduces the level of Rad50_WT ATPase activation by fivefold, indicating that this mutant does not fully retain the ability to alter the conformation of Rad50 in a productive manner (Table 1). The ATPase activities of the Rad50 mutants were examined by themselves and in the presence dsDNA and Mre11_WT. We found that in the absence of dsDNA and Mre11, the ATPase activities of the Rad50 mutants were largely unaffected except for Rad50_F132A, which had a 12‐fold decrease in activity (Table 1). The Km‐ATP values for the Rad50_A48C, Rad50_R154A, and Rad50_E474Q were elevated 22‐, 10‐, and 12‐fold compared to Rad50_WT. The addition of Mre11_WT and dsDNA caused WT‐like or greater increases in ATPase activity for Rad50_A48C, Rad50_Y89A, and Rad50_R154A with 29‐, 71, and 56‐fold activation, respectively (Table 1). The DNA activation for Rad50_F132A was somewhat reduced (14‐fold activation) and Rad50_E474Q showed a much‐reduced value of only fourfold. The changes in Km‐ATP values for the mutant MR complexes were very similar to those of the Rad50 protein alone.

Table 1.

Steady State ATPase Kinetic Constants for WT and Mutant Rad50 and MR‐DNA Complexes

Protein	k cat‐ATP (s−1)	K M‐ATP (μM)	Hill coeff.	k cat/K M (×103 M−1 s−1)	(k cat/K M)mut/(k cat/K M)WT
Rad50WT	0.118 ± 0.01	24 ± 2	1.5 ± 0.2	4.9 ± 0.6	1
RA48C	0.079 ± 0.01	517 ± 116	1.3 ± 0.2	0.2 ± 0.1	0.04
RY89A	0.032 ± 0.002	34 ± 7	0.9 ± 0.1	0.9 ± 0.2	0.18
RF132A	0.010 ± 0.001	41 ± 13	0.8 ± 0.2	0.2 ± 0.1	0.04
RR154A	0.032 ± 0.003	251 ± 46	1.6 ± 0.4	0.1 ± 0.02	0.02
RE474Q	0.125 ± 0.003	287 ± 15	1.6 ± 0.1	0.4 ± 0.02	0.08
MWTRWT‐DNA	2.9 ± 0.1	64 ± 5	2.3 ± 0.2	45.3 ± 3.9	1
MWTRA48C‐DNA	2.29 ± 0.09	192 ± 12	1.8 ± 0.2	11.9 ± 0.9	0.26
MWTRY89A‐DNA	2.28 ± 0.05	49 ± 3	1.9 ± 0.2	46.5 ± 3	1.03
MWTRF132A‐DNA	0.14 ± 0.01	108 ± 19	1.0 ± 0.2	1.3 ± 0.2	0.03
MWTRR154A‐DNA	1.77 ± 0.04	152 ± 9	1.5 ± 0.1	11.6 ± 0.7	0.26
MWTRE474Q‐DNA	0.49 ± 0.03	510 ± 40	2.7 ± 0.5	1 ± 0.1	0.02
MY197ARWT‐DNA	0.60 ± 0.06	189 ± 41	1.3 ± 0.3	3.2 ± 0.8	0.07
MY197ARA48C‐DNA	0.06 ± 0.01	337 ± 54	1.6 ± 0.3	0.2 ± 0.5	0.004
MY197ARY89A‐DNA	1.53 ± 0.06	119 ± 9	2.3 ± 0.4	12.9 ± 1.1	0.28
MY197ARF132A‐DNA	0.10 ± 0.01	112 ± 8	1.5 ± 0.2	0.9 ± 0.1	0.02
MY197ARR154A‐DNA	0.33 ± 0.02	256 ± 18	2.4 ± 0.4	1.3 ± 0.1	0.03
MY197ARE474Q‐DNA	0.13 ± .01	343 ± 30	1.6 ± 0.2	0.4 ± 0.05	0.01

The Rad50 mutants were then combined with Mre11_Y197A to form the double mutant cycle and determine the free energy of coupling values (ΔΔG _int) (Table 1; Fig. 3).

Figure 3.

Double mutant cycle analysis of predicted coupled residues from sector 2. (A) An example of a double mutant cycle for kcat‐ATP using the Mre11_Y197A and Rad50_A48C proteins. ΔΔG and ΔΔG _int are calculated as described in “Materials and Methods”. (B) The ΔΔG _interaction values for the Rad50 mutants and Mre11_Y197A. Positive values represent synergism and negative values are partially additive. Mutants with ΔΔG _interaction values between −0.5 and 0.5 are considered to be independent from each other.

The ΔΔG _int values calculated using k _cat‐ATP for Rad50_R154A and Rad50_E474Q were largely independent when combined with Mre11_Y197A, with values of 0.27 and −0.15 kcal/mol, respectively. The remaining Rad50 mutants showed nonadditive effects with ΔΔG _int values of −0.70, −0.73, and 1.2 kcal/mol for Rad50_Y89A, Rad50_F132A, and Rad50_A48C, respectively. These ΔΔG _int values are moderate but are typical or even considered high for allosterically coupled residues located between 45.6 (Rad50_Y89A), 47.2 (Rad50_F132A), and 49.8 (Rad50_A48C) Å away from Mre11_Y197A. The ΔΔG _int values calculated using Km‐ATP indicate a lack of coupling between Mre11_Y197A and Rad50_A48C, Rad50_Y89A, and Rad50_R154A and minor partially additive effects for Rad50_F132A and Rad50_E474Q. The ΔΔG _int values using k _cat /Km‐ATP for Rad50_A48C and Rad50_Y89A follow their k _cat‐ATP values, whereas Rad 50_E474Q follows its Km‐ATP value.

Steady state nuclease kinetics

The steady state exonuclease activity of Mre11 can be measured using the fluorescent base analog 2‐aminopurine (2AP), which is quenched via base‐stacking with the DNA duplex, but greatly increases when excised from the DNA substrate.35, 36 When the 2AP probe is located near the 3′ end, the nuclease activity of Mre11 does not require ATP but is activated by Rad50.8 The dependence on ATP increases as the probe is moved away from the 3′ end. Previous data have suggested that the rate‐limiting step of the steady state nuclease reaction is a conformational change that follows substrate binding but precedes the first nucleotide excision.8 The ATP‐independent nuclease activity of M_WTR_WT using the 2 position substrate is 1.5 min⁻¹, whereas with M_Y197AR_WT it is reduced to 0.04 min⁻¹ (Table 2). The Rad50 mutants were complexed with Mre11_WT, and their nuclease activities were measured on the 2‐position substrate. The nuclease activities of M_WTR_A48C, M_WTR_F132A, M_WTR_R154A, and M_WTR_E474Q with the 2‐position substrate are only mildly affected, with decreases in activity ranging from 4 to 1.5‐fold (Table 2). The M_WTR_Y89A complex showed a larger defect in activity with a 20‐fold reduction compared to the WT complex.

Table 2.

Steady State Nuclease Activity of WT and Mutant MR Complexesa

Protein	2 Position (−) ATP (min−1)	17 Position (+) ATP (min−1)
MWTRWT‐DNA	1.49 ± 0.35	0.215 ± 0.028
MWTRA48C‐DNA	0.63 ± 0.01	0.062 ± 0.003
MWTRY89A‐DNA	0.074 ± 0.006	0.032 ± 0.001
MWTRF132A‐DNA	0.366 ± 0.009	0.017 ± 0.002
MWTRR154A‐DNA	0.885 ± 0.085	0.231 ± 0.019
MWTRE474Q‐DNA	0.343 ± 0.029	0.061 ± 0.007
MY197ARWT‐DNA	0.040 ± 0.008	0.026 ± 0.001
MY197ARA48C‐DNA	0.075 ± 0.010	0.042 ± 0.002
MY197ARY89A‐DNA	0.041 ± 0.004	0.062 ± 0.030
MY197ARF132A‐DNA	0.018 ± 0.011	0.0020 ± 0.0005
MY197ARR154A‐DNA	0.022 ± 0.001	0.009 ± 0.003
MY197ARE474Q‐DNA	0.022 ± 0.007	0.003 ± 0.001

The reported rates are considered to be apparent‐k _cat values.

To form the double‐mutant cycle, the Rad50 single mutants were combined with Mre11_Y197A and their nuclease activity was measured again on the 2‐position DNA substrate (Table 2; Fig. 3). The mutations making up the MY_197AR_F132A and MY_197AR_R154A complexes are nearly independent of each other with ΔΔG _int values of 0.053 and 0.36 kcal/mol, respectively. MY_197AR_A48C and MY_197AR_E474Q displayed a moderate amount of partial additivity with ΔΔG _int values of −0.93 and −0.52, respectively. The mutations making up M_Y197AR_Y89A showed the largest amount of partial additivity with a ΔΔG _int value of −1.82 kcal/mol.

We next investigated the ATP‐dependent nuclease activity of the MR complexes using the 17‐position DNA substrate. The ATP‐dependent nuclease activity of M_WTR_WT and M_Y197AR_WT using this substrate is 0.22 and 0.03 min⁻¹, respectively (Table 2). Replacement of Rad50_WT with Rad50_R154A has no effect on the nuclease activity of Mre11_WT, with an apparent kcat of 0.23 min⁻¹. The other Rad50 mutants have larger effects on the ATP‐dependent nuclease activity of Mre11_WT, with M_WTR_E474Q and M_WTR_A48C having 3.5‐fold decreases in activity and M_WTR_Y89A and M_WTR_F132A having 7‐ and 13‐fold decreases, respectively.

Again, we combined the Mre11_Y197A mutant with the Rad50 mutants in order to form the double‐mutant cycle. Similar to the ATP‐independent nuclease activity on 2‐position substrate, the Mre11_Y197A and Rad50_F132A mutations were largely independent of each other, with a ΔΔG _int value of 0.053 kcal/mol. Rad50_R154A and Rad50_E474Q showed a moderate amount of synergism with Mre11_197A with ΔΔG _int values of 0.69 and 0.49 kcal/mol, respectively. In contrast, Rad50_A48C and Rad50_Y89A showed partially additive effects with the Mre11_Y197A mutation with ΔΔG _int values of −1.04 and −1.69 kcal/mol, respectively.

Discussion

The ATPase and nuclease activities of Rad50 and Mre11, respectively, are thought to be regulated in a coordinated manner.37 ATP binding and/or hydrolysis by Rad50 is known to increase the processivity of the MR complex and is likely the driving force behind its translocation along the DNA substrate.15 Previous work has demonstrated that the MR complex can adopt multiple states, with the ATPase‐competent state being distinct from the nuclease‐competent state.7, 8, 20, 38 The routes of communication between the nuclease and ATPase active sites are not fully clear. Structural work suggests that the coiled‐coil domain is linked to the ATPase active site through the signature coupling helices.18 ATP binding, hydrolysis, and/or product release alters the position of the signature coupling helices, which in turn drive movement in base of the coiled‐coils. The coiled‐coils are bound to the CTD of Mre11 and it is likely that movement in the coiled‐coil domain affects the positioning of the Mre11 cap and nuclease domains.7, 14, 18 SCA has the ability to identify residues that are structurally or functionally linked and its application has confirmed known allosteric pathways or identified previously unknown routes of communication.23, 39 Likewise, double mutant cycle analysis has the ability to reveal the strength of interaction between two distal residues and therefore can confirm the predictions of SCA.28 The interpretation of a double mutant cycle is clearer in cases where the rate‐limiting step of the reaction is unaltered by the mutations so that the kinetic constants report on the same step of the reaction. However, if the rate‐limiting step of the reaction is altered by the mutation, then the magnitude of coupling may be underestimated. Determining the rate‐limiting step for each mutant and double‐mutant complex for the Rad50 ATPase and Mre11 nuclease reactions is impractical due to the number and difficulty of the assays required,8, 34 therefore the ΔΔG _int values reported here may not fully capture the degree to which the residues are coupled.

With the exception of Rad50_E474, which is part of the Rad50 Signature motif and H‐bonds with the 2′ and 3′ hydroxyl groups of ATP40 [Fig. 4(F)], the Rad50 residues of sector 2 are distal to the ATP active site. However, although they do not directly interact with ATP, three of the remaining four Rad50 residues in the sector are connected to the Walker A lysine residue that interacts with the γ phosphate of ATP.15 Rad50_A48 is located within the Walker A helix and is at the center of a hydrophobic cluster made up of three residues on the beta strand 1 (Leu6, Tyr11, and Ile14) and three residues on beta strand 4 (Leu75, Val77, and Leu79) [Fig. 4(B)]. Mutation of this residue to the larger cysteine (chosen because it is the second most common residue at this position) likely alters the position of the Walker A helix through steric clashes. Consistent with this, the Km‐ATP is significantly elevated, suggesting that ATP binding has been negatively affected. The introduced cysteine is over 40 Å away from the nearest cysteine residue (its symmetry partner), essentially ruling out disulfide formation as a cause of the altered kinetic properties. Rad50_F132 is located on helix B and is part of a hydrophobic cluster made up Ile49, Leu53, and Phe54, which are found towards the distal end of the Walker A helix [Fig. 4(D)]. Removal of the phenylalanine side chain results in a ∼10‐fold decrease in k _cat‐ATP, again suggesting that the position of the Walker A helix has been altered by the mutation. Rad50_Y89 is located on beta strand 5 and forms another hydrophobic cluster with Leu53, Phe54, Leu124, Ile125, and Met127 [Fig. 4(C)]. Leu53 and Phe54 are at the farthest end of the Walker A helix and pack against the other hydrophobic residues. Loss of the tyrosine side chain has only minor effects on the ATPase activity of the Rad50, consistent with it being the most distal of the three Walker A helix mutations. The hydrophobic clusters these three sector 2 residues are well‐conserved throughout Rad50 homologs, although the identities of the amino acids vary. As the Walker A lysine is perhaps the most important residue for ATP hydrolysis, the stabilization of its helix through hydrophobic interactions is very likely the primary function of these residues.

Figure 4.

Close‐up view of sector 2 residues identified using SCA. The figures were generated using the structures of Pfu Rad50 (PDB ID: 3QKT) and Mre11 (PDB ID: 3DSC) but using the amino acid numbering of bacteriophage T4 (based on the SCA alignment). Details regarding each residue can be found in the Discussion. In all cases, the residue identified by SCA is shown in green. Hydrophobic interactions are shown with space‐filling spheres and hydrogen bonding/electrostatic interactions are shown with sticks.

The fifth Rad50 residue in sector 2 is Arg154, which on the basis of Pyrococcus furiosus (Pfu) Rad50 structures in the presence and absence of ATP, is part of the N‐terminal subdomain that undergoes a significant rigid‐body rotation upon binding of ATP18 [Fig. 4(E)]. This rotation causes an almost 7 Å movement of the arginine sidechain, although its hydrogen bonding interactions with neighboring residues are maintained as the entire subdomain rotates. As Tainer and colleagues have previously described, this rigid body rotation is a major connection between the coiled‐coil domain and the ATP active site via the Q‐motif glutamine residue.18 Mutation of the Arg154 residue and elimination of the H‐bonds it participates in clearly affects the ATP active site with both an increase in Km‐ATP and a decrease in k _cat.

In addition to affecting ATPase activity, several Rad50 mutants affect the ATP‐independent nuclease activity of Mre11, suggesting a structural perturbation has propagated from the site of the mutation in Rad50 to the Mre11 active site. The nuclease activity of M_wtR_Y89A is lowered by 20‐fold, despite being nearly 50 Å away from the nuclease active site. Also showing reduced ATP‐independent nuclease activity are the MR complexes containing the Rad50_E474Q and Rad50_F132A mutants, both of which are decreased by fourfold. It is likely that the pathways of communication between the mutant side chains and the Mre11 nuclease site (which results in a decrease in activity) also exist in the WT enzyme and are used to coordinate the nuclease and ATPase activities.

The side‐chain of Mre11_Y197 is packed against Phe186 (Ile in Pfu), which is positioned between two histidine residues that bind the catalytic Mn²⁺ cations [Fig. 4(A)]. In Pfu Mre11, an additional hydrophobic residue is present on the other side of the tyrosine side chain (not shown in Fig. 4), whereas in T4 phage this residue is a glycine. As expected based on its location, the Mre11_Y197A mutation decreases the nuclease activity of the complex. Additionally, similar to the Rad50 mutants that affect nuclease activity, the Mre11_Y197A mutation also reduces ATPase activity, suggesting that the structural perturbation caused by the mutation has propagated to Rad50 active site. On the basis of this result, we suggest that Mre11_Y197 acts as a sensor of the nuclease active site (sensing substrate binding and/or product formation or release) and communicating this information to the ATP active site.

The coiled‐coiled domain is thought to be the primary conduit of communication between Rad50 and Mre11.18, 37 The structure of Rad50 suggests that all of the residues in sector 2 are connected to the coiled‐coil domain, which is connected to the Mre11 CTD. Rad50_R154 is located near the base of the coiled‐coil and its extensive H‐bonding interactions likely aid its stabilization. Rad50_E474 is connected to the coiled‐coil through the signature coupling helices, which connect the Signature helix to the coiled‐coil.18 The route of communication between the remaining three Rad50 residues and Mre11_Y197 appears to be less direct and is likely transmitted across the Rad50 dimer interface. Rad50_A48, Rad50_Y89, and Rad50_F132 participate in stabilization of the Walker A helix and in turn the Walker A lysine sidechain, which interacts with γ phosphate of ATP. Also interacting with the γ phosphate of ATP is the Signature motif serine across the dimer interface. The Signature motif serine is at the end of the signature helix, which is coupled to the coiled‐coil through the same pathway as Rad50_E474. This implies that the Signature motif serine residue may also be conformationally coupled to Mre11_Y197; however, because Rad50_S471 is 100% conserved in all Rad50 homologs it cannot be identified using SCA.

An unanticipated outcome of our double‐mutant cycle analysis was that the residues making up sector 2 are not coupled to each other in an identical fashion. A sector‐based SCA analysis performed by the Ranganathan group on the S1A protease family identified two sectors that were biochemically independent from one another.27 Mutation of residues in one sector affected catalytic properties, whereas mutation of the other sector resulted in changes in thermostability. The double‐mutant cycle analysis between sector 2 residues in Rad50 and Mre11 revealed coupling in non‐overlapping subsets of the enzymatic properties tested [Fig. 3(B)]. For example, Rad50_A48 and Mre11_Y197 display strong synergism in their perturbation of k _cat‐ATP, but have a partially additive effect on nuclease activity. On the other hand, for Rad50_Y89 and Mre11_Y197 show partially additive effects in both nuclease and ATPase activity. The interpretation of these differential effects, beyond that of indicating they are coupled, is unclear and will likely require structural analysis of the mutant MR complexes.

While the MR complex is found in all domains of life, there is a great deal of sequence divergence between archaeal, bacterial, eukaryotic, and bacteriophage homologs. Additionally, the length of the coiled‐coil domain is quite variable, containing nearly 1000 amino acids in eukaryotes, 600 in archaea, 550 in bacteria, and 250 in bacteriophage. The eukaryotic MR complex associates with Xrs2/Nbs1, ATM kinase, and CDK2, whereas the archaeal complex interacts with HerA and the bacteriophage complex interacts with SSB (gp32). Despite all of this diversity in sequence and binding partners, the success of the SCA analysis, which uses sequence information from all these homologs, suggests that the core allosteric communication pathways found in the MR complex are well conserved amongst the various homologs. This finding supports the use of the bacteriophage T4 system as a model system to investigate the fundamental biochemical and biophysical properties of the MR complex.

Materials and Methods

Materials

Oligodeoxynucleotides used for mutagenesis were purchased from either Integrated DNA Technologies or the Iowa State University DNA Facility. Phusion High‐Fidelity DNA Polymerase was purchased from Thermo Scientific. The expression vector used for subcloning T4 Mre11 and Rad50 genes was a pET28 plasmid purchased from Novagen. Plasmid DNA was propagated in E. coli XL‐1 Blue cells (Agilent Technologies) or E. coli E. cloni cells (Lucigen). Kits for plasmid DNA purification were purchased from Qiagen. DNA sequencing was performed at the Iowa State University DNA Facility. E. coli BL21 (λDE3) cells were obtained from Novagen. Isopropyl‐β‐D‐thiogalactopyranoside (IPTG) and antibiotics (ampicillin or kanamycin) were purchased from Gold Biotechnology. Chitin Beads were purchased from New England Biolabs. Coupling enzymes (pyruvate kinase and lactate dehydrogenase), dNTPs, nicotinamide adenine dinucleotide (NADH) and Nickel‐agarose were purchased from Sigma‐Aldrich Chemical Company. Phosphoenolpyruvate was purchased from Alfa Aesar. Adenosine 5'‐triphosphate was purchased from USB Corporation. Media components were purchased from Boston Bioproducts. All other chemicals were purchased from Fisher Scientific.

Multiple sequence alignment

Mre11/Rad50 sequences were collected by PSI‐Blast41 queried with Bacteriophage T4, E. coli, P. furiosus, S. cerevisiae, and human Mre11/Rad50 sequences. The redundant sequences, sequences that were too short (less than 200 amino acids) and unpaired Mre11 or Rad50 sequences were removed. The sequences of Bacteriophage T4, E. coli, P. furiosus, T. maritime, M. jannaschii, S. cerevisiae, and human Mre11 and Rad50 were aligned with 3D‐Tcoffee.30 For the sequence alignment of Rad50, the major portion of coiled‐coil domain (except 53 residues around the signature CXXC motif) was removed to facilitate sequence alignment. Due to the high diversity of Mre11/Rad50 sequences, structure‐based alignment42 (program CE) and biochemical knowledge of MR complex were used to manually refine the sequence alignment. The resulting sequence alignment was extended to include all Mre11/Rad50 sequences from Blast queries with ClustalW.43 Mre11 and Rad50 from same species were merged into one sequence. Only one copy of the sequences who share greater than 98% sequence identity with each other was kept. Columns in the alignment with too many gaps (over 20%) were eliminated.

Statistical coupling analysis

SCA has been documented thoroughly in previous publications from the Raganathan laboratory.21, 23, 27, 39 The alignment was treated as a 772 by 787 matrix. For each column (jth column), the most abundant amino acid was identified (a_j). Each element (i, j) in the column was set to one if a_ij  = a_j, otherwise to zero. The weighted positional conservation was calculated according to (1)):

$$P_j=\left|\mathrm{l}\mathrm{n}\frac{f_j\left(1-q^a\right)}{q^a\left(1-f_j\right)}\right|$$ (1)

where j, f, a, and q are column number, frequency of 1 in the column, most abundant amino acid type in the column, and background frequency of the amino acid type. The cross correlation was calculated according to (2)):

$$C_{ij}={P_i{P}_j(f}_{ij}^{ab}-f_i^a{f}_j^b)$$ (2)

where i, j, are column numbers and a, b, are corresponding amino acid in column i, j; C donates for correlation. Principal component analysis was applied to the matrix C to extract important correlation information.27 70 residues were selected based on their contribution to the Cij matrix, which is proportional to the amplitude of the first eigenvector.27 To further divide the 70 residues into statistically independent sectors, the distributions of these 70 residues along eigenvector 2–6 were analyzed. The residues were best separated along the 3rd and the 5th eigenvectors and five sectors were defined accordingly.

Mutagenesis and protein expression and purification

Mutagenesis was performed using the Stratagene QuikChange™ site‐directed mutagenesis protocol as previously described.40 Oligonucleotide sequences are listed in Supporting Information Table SI. The presence of the mutation and the integrity of the remaining coding sequence were verified by DNA sequencing. Protein expression and purification were accomplished essentially as previously described.40 Briefly, Mre11 and Rad50 were overexpressed in E. coli BL21(DE3) cells. WT and mutant Mre11 proteins were purified using a chitin affinity column, followed by intein‐mediated self‐cleavage. WT and mutant Rad50 proteins were purified using a nickel‐NTA and phosphocellulose chromatography.

Circular dichroism (CD) spectroscopy

CD studies on wild‐type and mutant proteins were performed at room temperature (∼22°C) on a Jasco J710 CD spectrometer in a 1 cm cell using a protein concentration of 0.3–0.5 mg/mL. Spectra were collected from 200 to 260 nm in increments of 1 nm. Each spectrum was blank‐corrected and normalized to the ellipticity of the WT protein at 216 nm to correct for minor differences in protein concentration.

Steady state ATPase kinetics

Steady state ATPase kinetics were performed using a coupled fluorometric assay using NADH, PEP, pyruvate kinase and lactate dehydrogenase as previously described.15 The Rad50 concentrations used in each assay was 1 µM for Rad50_WT and Rad50_A48C, 2 µM for Rad50_R154A and Rad50_E474Q, and 5 µM for Rad50_F132A. When Mre11 and DNA (1.3 μM) were included, the Rad50 concentration used in each assay was 0.2 µM for Rad50_WT, Mre11_WT, Rad50_A48C, Rad50_Y89A, Rad50_R154A, and 0.2 µM for Rad50_E474Q, 0.5 µM for Rad50_R154A, and 1 µM for Rad50_F132A. The concentration of Mre11 in these assays was held in slight excess (1.2‐fold) over the Rad50 concentration. Each assay was performed in triplicate. For estimates of maximum velocity (V _max), Michaelis constant (K _M) and Hill coefficient, the reaction velocities at various ATP concentrations were fitted to the Hill equation:

$$(v=\hspace{0.17em}\frac{V\mathrm{m}\mathrm{a}\mathrm{x}\hspace{0.17em}{\left[S\right]}^n}{{K_{\mathrm{M}}}^n+\hspace{0.17em}{\left[S\right]}^n})$$ (3)

using Sigmaplot 10.0/Enzyme Kinetics Module 1.3 (Systat Software). Previously, an F test was employed for the WT enzyme to justify fitting the data to a more complicated Hill model.15 The sequence of the oligonucleotides used for the DNA substrate is reported in Supporting Information Materials.

Steady state nuclease kinetics

Steady state nuclease kinetics were accomplished using a 2AP fluorometric assay as previously described.8 The assays were performed using two 50 nucleotide substrates, one containing a 2‐aminopurine at the second position from the 3′ end and one containing a 2‐aminopurine at the 17th position from the 3′ end. Protein concentrations for 2‐position assays were 0.05 µM for M_WTR_WT and 0.25 µM for all other mutant MR complexes. The protein concentration for all 17‐position assays was 0.4 µM. ATP concentrations of at least 5*K _M were used for all nuclease assays. Each reported apparent‐k_cat is an average of at least three trials. Oligonucleotide sequences are reported in Supporting Information Table SI.

Size exclusion chromatography/multiangle light scattering

SEC‐MALS was accomplished as previously described.6, 20 Briefly, a Superdex 200 10/300 GL column (GE Healthcare) was used in conjunction with a Dawn Helios II multiangle light scattering detector (Wyatt Technology, Santa Barbra, CA) at a flow rate of 0.5 mL/min. The column was equilibrated with buffer containing 20 mM Tris, pH 8.0, and 400 mM NaCl. For each run, 100 µL of sample with Mre11, Rad50, or MR complex was injected. The protein concentrations are 100 µM for Mre11 and 50 µM for Rad50. Data were analyzed using ASTRA 6 software (Wyatt Technology).

Double‐mutant cycle analysis

For each ATPase and nuclease k _cat, a ΔΔG was calculated with the equation:28

$$\Delta \Delta G=-\mathrm{R}\mathrm{T}\ln \frac{k_{\mathrm{c}\mathrm{a}\mathrm{t}}\left(\mathrm{M}\mathrm{u}\mathrm{t}\mathrm{a}\mathrm{n}\mathrm{t}\right)}{k_{\mathrm{c}\mathrm{a}\mathrm{t}}\left(\mathrm{W}\mathrm{T}\right)}.$$ (4)

For each ATPase K _M, a ΔΔG was calculated with the equation:

$$\Delta \Delta G=-\mathrm{R}\mathrm{T}\ln \frac{K_{\mathrm{M}}\left(\mathrm{W}\mathrm{T}\right)}{K_{\mathrm{M}}\left(Mutant\right)}.$$ (5)

For each ATPase K _M, a ΔΔG was calculated with the equation:

$$\Delta \Delta G=-\mathrm{R}\mathrm{T}\ln \frac{{k_{\mathrm{c}\mathrm{a}\mathrm{t}}/K}_{\mathrm{M}}\left(\mathrm{M}\mathrm{u}\mathrm{t}\mathrm{a}\mathrm{n}\mathrm{t}\right)}{{k_{\mathrm{c}\mathrm{a}\mathrm{t}}/K}_{\mathrm{M}}\left(\mathrm{W}\mathrm{T}\right)}.$$ (6)

For each pair of mutants, a ΔΔG _interaction was calculated using the equation:

$$\Delta \Delta G_{\mathrm{i}\mathrm{n}\mathrm{t}\mathrm{e}\mathrm{r}\mathrm{a}\mathrm{c}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}}=\Delta \Delta G_{\mathrm{D}\mathrm{o}\mathrm{u}\mathrm{b}\mathrm{l}\mathrm{e}\hspace{0.17em}\mathrm{m}\mathrm{u}\mathrm{t}\mathrm{a}\mathrm{n}\mathrm{t}}\hspace{0.17em}-\Delta \Delta G_{\mathrm{S}\mathrm{i}\mathrm{n}\mathrm{g}\mathrm{l}\mathrm{e}\hspace{0.17em}\mathrm{m}\mathrm{u}\mathrm{t}\mathrm{a}\mathrm{n}\mathrm{t}\hspace{0.17em}1}\hspace{0.17em}-\Delta \Delta G_{\mathrm{S}\mathrm{i}\mathrm{n}\mathrm{g}\mathrm{l}\mathrm{e}\hspace{0.17em}\mathrm{m}\mathrm{u}\mathrm{t}\mathrm{a}\mathrm{n}\mathrm{t}\hspace{0.17em}2}.$$ (7)

Supporting information

Supporting Information Figures.

Supporting Information Tables.