Overexpression, Purification, and Characterization of the SbcCD Protein from Escherichia coli *

The sbcC and sbcD genes mediate palindrome inviability in Escherichia coli. ThesbcCD operon has been cloned into the plasmid pTrc99A under the control of the strong trc promoter and introduced into a strain carrying a chromosomal deletion of sbcCD. The SbcC and SbcD polypeptides were overexpressed to 6% of total cell protein, and both polypeptides copurified in a four-step purification procedure. Purified SbcCD is a processive double-strand exonuclease that has an absolute requirement for Mn2+ and uses ATP as a preferred energy source. Gel filtration chromatography and sedimentation equilibrium analyses were used to show that the SbcC and SbcD polypeptides dissociate at some stage after purification and that this dissociation is reversed by the addition of Mn2+. We demonstrate that SbcD has the potential to form a secondary structural motif found in a number of protein phosphatases and suggest that it is a metalloprotein that contains the catalytic center of the SbcCD exonuclease.

The sbcC and sbcD genes mediate palindrome inviability in Escherichia coli. The sbcCD operon has been cloned into the plasmid pTrc99A under the control of the strong trc promoter and introduced into a strain carrying a chromosomal deletion of sbcCD. The SbcC and SbcD polypeptides were overexpressed to 6% of total cell protein, and both polypeptides copurified in a four-step purification procedure. Purified SbcCD is a processive double-strand exonuclease that has an absolute requirement for Mn 2؉ and uses ATP as a preferred energy source. Gel filtration chromatography and sedimentation equilibrium analyses were used to show that the SbcC and SbcD polypeptides dissociate at some stage after purification and that this dissociation is reversed by the addition of Mn 2؉ . We demonstrate that SbcD has the potential to form a secondary structural motif found in a number of protein phosphatases and suggest that it is a metalloprotein that contains the catalytic center of the SbcCD exonuclease.
The metabolism of palindromic DNA is best understood in Escherichia coli. Replicons that contain long DNA palindromes (greater than 150 -200 base pairs of total length) suffer two fates when introduced into wild-type cells (1). They are either not recovered (inviability) or recovered with evidence of deletion in and around the palindrome (instability). However, strains of E. coli exist in which replicons containing long DNA palindromes can be propagated. These strains carry mutations in either the sbcC or sbcD genes (12,13). Both genes have been mapped, cloned and sequenced, and shown to be transcribed from a common promoter (14). The sbcD gene encodes a 44.7-kDa polypeptide and contains the conserved sequence DXHX n GDXXDX n GNH(D/E) (n ϭ ϳ25) found in the serine/ threonine phosphatases and other phosphoesterases (15,16). The yeast recombination/repair proteins, RAD32 of Schizosaccharomyces pombe (17) and MRE11 of Saccharomyces cerevisiae (18), and the human equivalent of MRE11 (19) also contain this sequence. They are more closely related to SbcD than other members of the phosphoesterase family and, it has been suggested, are mechanistically similar (15,16). The sbcC gene encodes a large 118.7-kDa polypeptide that contains the ATP-A and ATP-B nucleotide binding motifs in globular regions at either end of the protein (14,20). These are separated by a large ␣-helical region that is predicted to form two coiled-coil domains interrupted by a short spacer (16). This arrangement is similar to that described for a number of proteins in the structural maintenance of chromosomes (SMC) 1 family, which are mainly involved in chromosome condensation and segregation (21). Examples include SMC1 and SMC2 of S. cerevisiae (22,23) and MukB of E. coli (24). The S. cerevisiae and human RAD50 proteins (25,26), both of which interact with MRE11 in vivo, and the mouse RAD50 homologue (27) also have the same overall organization (16,27).
Sequence analyses and genetic studies suggest that sbcC and sbcD encode a nuclease. The derived amino acid sequences of SbcC and SbcD are similar to those of the putative major exonucleases of bacteriophage T4 (gp46 and gp47) and T5 (gpD13 and gpD12) (14,20). In addition, the gam gene of bacteriophage allows palindrome-containing phage to plate on strains of E. coli that are sbcCD ϩ , suggesting that the Gam protein might actually inhibit two nucleases, RecBCD and SbcCD (28). We have shown that the SbcCD protein from E. coli possesses an ATP-dependent double-strand DNA exonuclease activity that requires both SbcC and SbcD (29). In this work we describe the construction of an sbcCD deletion strain and the overexpression of the SbcC and SbcD polypeptides, detail the purification of SbcCD, investigate the biochemical properties of the double-strand exonuclease activity, and characterize the general physical properties of the purified protein.

EXPERIMENTAL PROCEDURES
Reagents-Benzamidine was prepared as a 10 mM stock. Phenylmethylsulfonyl fluoride was prepared as a 50 mM stock in isopropanol. The metals used were manganese acetate, cobalt chloride, ferrous chloride, zinc chloride, magnesium sulfate, cupric sulfate, calcium chloride, and nickel sulfate. Ferrous chloride and manganese acetate were always prepared just before use. All of these reagents and ATP, dATP, dCTP, dGTP, dTTP, ADP, ATP␥S, GTP, and GDP were all purchased from Sigma. pUC4K and pUC18 were purchased from Pharmacia Biotech Inc.
Buffers-Buffer A contained 50 mM Tris, pH 7.5, 1 mM EDTA, 50 mM NaCl, 10 mM ␤-mercaptoethanol, 10% (v/v) glycerol. Buffer P contained 10 mM phosphate, pH 6.4, 10 mM ␤-mercaptoethanol, 10% (v/v) glycerol. Sucrose cell buffer contained 50 mM Tris, pH 7.5, 1 mM EDTA, 50 mM NaCl, 10 mM ␤-mercaptoethanol, 10% (w/v) sucrose. TAE buffer contained 40 mM Tris base, 1.1% (v/v) acetic acid, and 1 mM EDTA. DNA gel loading buffer contained 0.25% (w/v) bromphenol blue, 30% (v/v) glyc-* This work has been funded by grants from the Biotechnology and Biological Sciences Research Council and the Medical Research Councils of the United Kingdom. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Bacterial Strains and Plasmids-An E. coli sbcCD deletion-derivative was constructed. The 2.1-kb BamHI-NruI and 3.3-kb NruI-EcoRI fragments that flank sbcCD were isolated from pFG103 (a plasmid that contains the whole of sbcCD and flanking sequences) (13). The 1.25-kb SalI restriction fragment from pUC4K (which contains a gene encoding kanamycin resistance) was isolated and made blunt-ended by filling in using the Klenow fragment of DNA polymerase I (New England Biolabs). These three fragments were then joined to the 2.67-kb BamHI-EcoRI restriction fragment from pUC18 (which contains a gene encoding ampicillin resistance and an origin of DNA replication) in a four fragment ligation step. After transformation, colonies that were resistant to kanamycin and ampicillin and whose plasmids gave a 9.32-kb fragment after restriction at a unique EcoRI-site were isolated. Further restriction analysis was used to confirm the structure. Plasmid, linearized with BamHI and EcoRI, was then introduced into the nucleasefree, recD1009, strain of E. coli FS1585 (obtained from F. W. Stahl). Colonies resistant to kanamycin and sensitive to ampicillin were isolated and used to transduce the E. coli strain JM83 (30). The loss of the sbcCD operon was verified by observing the ability of this strain to plate bacteriophage containing a long DNA palindrome. The resulting strain was DL733.
The sbcC and sbcD genes from E. coli were cloned into the isopropyl-1-thio-␤-D-galactopyranoside (IPTG)-inducible, ampicillin-resistant, expression vector pTrc99A (Pharmacia) via a multistep cloning procedure that will be described in detail elsewhere. 2 This yielded plasmid pDL761 (see Fig. 1A), which was introduced into DL733. The outcome, E. coli DL776 (selected as a kanamycin-and ampicillin-resistant colony that was unable to propagate bacteriophage containing a long palindrome) was used for the overexpression of SbcC and SbcD.
Overexpression of SbcC and SbcD-To ensure the reproducible overexpression of SbcC and SbcD polypeptides DL776 cells stored at Ϫ80°C (in 50% glycerol w/v) were streaked onto Luria Broth plates (31) supplemented with 100 g/ml ampicillin (LBA medium) and incubated for 16 h at 37°C. Colonies were then stored at 4°C for a maximum period of 12 h. When colonies were stored for longer periods, the level of overexpression declined dramatically. A single colony was picked into 5 ml of LBA medium and grown once more for 16 h at 37°C. 100 l of this overnight culture was used to inoculate a further 10 ml of LBA medium. After incubation at 37°C for 2 h, IPTG was added to 1 mM. At various times 1 ml samples were removed and the cells were pelleted by centrifugation. Cells were resuspended in 100 l of SDS gel loading buffer, boiled for 4 min, and analyzed by SDS-polyacrylamide gel electrophoresis (SDS-PAGE).
Growth and Storage of Cells-All cells were grown at 37°C with aeration in LBA medium. A single colony of DL776 was picked into 5 ml of LBA medium and grown for 16 h. The next day 50 l of this culture was diluted into 5 ml of fresh LBA medium and growth continued for 8 h. This 5-ml inoculum was used to inoculate a 500-ml overnight culture of DL776, which in turn was used to seed a 15-liter fermenter at an A 650 of 0.1. Growth was continued until A 650 ϭ 0.4, at which point IPTG was added to 1 mM. After 6 h IPTG induction cells were harvested by low speed centrifugation, and aliquots of cell paste were frozen in liquid nitrogen before storing at Ϫ 80°C. Typically, a 15-liter culture of DL776 yielded 75-85 g of cells.
Amino-terminal Sequence Analysis of SbcC and SbcD-Polypeptides were resolved on a 10% SDS-polyacrylamide gel and then electroblotted onto polyvinylidene difluoride Immobilon-P transfer membrane (Millipore) by the method described by Matsudaira (32). Transferred polypeptides were visualized by Amido Black (Bio-Rad) staining, and excised bands were sequenced using an Applied Biosystems 477A protein sequencer (WELMET, University of Edinburgh).
Gel Filtration-Fast protein liquid chromatography (FPLC) gel filtration was performed at 20°C on a Superose 6 HR 10/30 FPLC column (Pharmacia) run at 24 ml/h. High pressure liquid chromatography (HPLC) gel filtration was also performed at 20°C on a Rainin Dynamax 4.6 250-mm Hydropore-5-SEC column (with a guard column) run at 30 ml/h. Both columns were equilibrated either with buffer A alone or with buffer A containing 5 mM Mn 2ϩ . The SbcCD protein used for gel filtration experiments was prepared in a fashion similar to that described under "Purification of SbcCD Protein" except that all purification steps were performed in buffer A with no added Mn 2ϩ . 100 g of SbcCD (Fraction IV prepared in the absence of Mn 2ϩ ) at 2 mg/ml was applied to the HPLC column, and 400 g of SbcCD was applied to the FPLC column. In experiments that tested the effect of Mn 2ϩ , Mn 2ϩ was added to the protein directly from a 500 mM stock to give a final concentration of 5 mM. All samples were incubated for 5 min on ice and passed through a 0.2-m membrane filter (Millipore) prior to chromatography. Protein elution was monitored by measuring the absorbance at either 280 (FPLC) or 254 nm (HPLC). Fractions were also collected and analyzed by SDS-PAGE. Relative molecular mass was calculated from the FPLC trace by comparing the elution volume (Ve) of SbcC and SbcD with that of standards. The molecular mass standards (Bio-Rad) used to calibrate the columns were: bovine thyroglobulin (670 kDa), bovine gamma globulin (158 kDa), chicken ovalbumin (44 kDa), horse myoglobulin (17 kDa), and vitamin B-12 (1.35 kDa). Void volume was determined using Blue 2000 Dextran (Pharmacia).
Analytical Centrifugation-Sedimentation equilibrium analysis was performed on SbcCD (concentrations ranging from 0.2 to 0.5 mg/ml) as a service by the United Kingdom National Center for Macromolecular Hydrodynamics, University of Leicester. Runs were performed at 15°C in buffer A in the presence and the absence of 5 mM Mn 2ϩ .
Gel Electrophoresis-SDS-PAGE was performed according to standard procedures (31), and protein was resolved on 12% discontinuous gels. SDS gel loading buffer was added to samples that were boiled for 4 min prior to loading and visualized with Coomassie Brilliant Blue (Sigma). Molecular weight standards (Sigma) were: rabbit muscle myosin (205,000), E. coli ␤-galactosidase (116,000), rabbit muscle phosphorylase b (97,400), bovine albumin (66,000), hen egg white albumin (45,000), and carbonic anhydrase (29,000) from bovine erythrocytes. DNAagarose gel electrophoresis was also performed according to standard procedures (31) using a tris-acetate-EDTA buffer system and 1% gels.
Protein Densitometry-Coomassie Blue-stained protein gels were dried between two sheets of cellophane membrane (Bio-Rad) and scanned using a Scan Jet IIcx (Hewlett Packard) desktop scanner. It was assumed that Coomassie Blue staining was proportional to molecular weight, and band intensities were quantified using ImageQuant software (Molecular Dynamics).
Protein Concentrations-Protein concentrations were estimated using a protein assay kit (Bio-Rad) with BSA as a standard. The molar concentration of SbcCD was calculated assuming a stoichiometry of SbcC 6 :SbcD 12 .
DNA Substrates-Double-strand pUC19 DNA (New England Biolabs) was linearized with EcoRI (Boehringer Mannheim) and end-labeled at the 3Ј end using [␣-32 P]dATP (Amersham Corp., 800 Ci/mmol) and Klenow enzyme (New England Biolabs). This resulted in a substrate with a specific activity of 3.3 ϫ 10 5 cpm/g of total nucleic acid. Uniformly 32 P-labeled double-strand DNA was synthesized using the polymerase chain reaction in a 100-l reaction containing: 10 mM Tris, pH 9.0, 1.5 mM MgCl 2 , 50 mM KCl, 200 M each of dATP, dCTP, dGTP, and dTTP, 20 Ci of [␣-32 P]dCTP (Amersham Corp., 3000 Ci/mmol), 100 pmol of each primer, 30 ng of template DNA, and 5 units of Taq DNA polymerase (Boehringer Mannheim). The primers were 19 bases long with 5Ј ends corresponding to nucleotides 398 and 413 of pUC19, and the template DNA was pUC19 (2686 base pairs) linearized with EcoRI. The reaction mix was incubated for 3 min at 94°C, 45 s at 58°C, and 3 min at 72°C followed by 25 cycles of 1 min at 94°C, 45 s at 58°C, and 3 min at 72°C and a final 10 min at 72°C. Amplification of template DNA yielded a product of 2669 base pairs. Reaction products were resolved on a 1% agarose gel to remove contaminating DNA, the band corresponding in size to the expected product was excised and purified with a QIAquick gel extraction kit (Qiagen Inc.). This yielded a substrate with a specific activity of 4 ϫ 10 7 cpm/g of total nucleic acid.
Standard Assay for Double-strand Exonuclease Activity-Protein was incubated with DNA (7.5 nM double-strand ends) for 30 min at 37°C in a reaction mix (20 l) containing 5 mM Mn 2ϩ , 1 mM ATP, 25 mM Tris, pH 7.5, 1.25 mM dithiothreitol, 2% glycerol, and 100 g/ml BSA. Reactions were terminated by adding an equal volume of DNA gel loading buffer containing 50 mM EDTA. Reaction products were resolved on agarose gels that were dried onto Whatman DE81 paper before exposure to a storage phosphor screen and quantification using a Molecular Dynamics PhosphorImager.
Determination of V m and K m(app) -0.5 nM SbcCD (Fraction V) was incubated for 30 min at 37°C in a 20-l standard SbcCD reaction mixture containing the concentration of 32 P end-labeled DNA indicated. The amount of limit product generated was quantified and the V m and K m (app) values were determined using Erithacus GRAPHIT software.
Assay for SbcCD Processivity-7.5 nM (double-strand ends) of uniformly labeled pUC19 DNA was incubated at 37°C with 1 nM SbcCD (Fraction V) in a 140-l standard SbcCD reaction mixture. 18-l samples were removed at various times (0, 2, 5, 10, 20, 40, and 60 min), and the amount of limit product generated was quantified. To assess whether SbcCD acts in a processive manner the same reaction was performed, except 7.5 nM (double-strand ends) of unlabeled pUC19 DNA was incubated at 37°C with 1 nM SbcCD (Fraction V) in a 140-l standard SbcCD reaction mixture. 18-l samples were removed at various times (0, 2, 5, 10, 20, 40, and 60 min), and the amount of limit product generated was quantified. However, at 3 min SbcCD was challenged with an equal concentration of uniformly labeled pUC19 DNA.
Prediction of SbcD Secondary Structure-A secondary structure prediction (PHDsec) was generated by submitting the SbcD amino acid sequence (SWISSPROT accession number, P13457) to the Predict-Protein service provided by the European Molecular Biology Laboratory, Heidelberg.

RESULTS
Overexpression of SbcC and SbcD-The sbcCD operon of E. coli was subcloned into the plasmid pTrc99A placing the sbcC and sbcD genes under the control of the IPTG-inducible trc promoter and the lacI q repressor (Fig. 1A). The resulting plasmid, pDL761, was then introduced into the E. coli strain DL733, which carries a chromosomal deletion of the sbcCD operon (see "Experimental Procedures"). The strain that re-sulted, DL776, was used for the overexpression of the SbcC and SbcD polypeptides and subsequent purification of SbcCD protein.
A time course of the appearance of the putative SbcC and SbcD polypeptides is shown in Fig. 1B. Following the addition of IPTG, samples were taken every 1.5 h. Overproduction of the 118.7-kDa SbcC polypeptide was obvious after 1.5 h of IPTG induction (Fig. 1B, lane c), and maximal expression was reached after 6 h (Fig. 1B, lane f) with SbcC expressed to 6% of total protein (judged by protein densitometry; data not shown). A band corresponding in size to the 44.7-kDa SbcD polypeptide was also apparent at 1.5 h (Fig. 1B, lane c), again with maximal levels being reached after 6 h of induction (Fig. 1B, lane f). This was most likely to be SbcD; however, the 38.6-kDa E. coli lacI q repressor co-migrates with SbcD on 12% polyacrylamide gels (determined by amino-terminal sequencing; data not shown), making it difficult to quantify levels of SbcD. This was most obvious when DL776 was grown for 7.5 h in the presence and the absence of IPTG; levels of induction were similar in both (Fig. 1B, lanes g and h). Levels of SbcD were assumed to be at least the same as SbcC because sbcD is transcribed before sbcC (14).
Purification of SbcCD Protein-SDS-PAGE and ATPdependent double-strand exonuclease activity (28) were used to monitor the purification of SbcCD protein. A summary of the purification procedure and the results obtained are shown in Table I and Fig. 2. 20 g of wet weight of IPTG-induced DL776 cells were thawed overnight at 4°C then suspended in 60 ml of ice-cold sucrose cell buffer containing 0.1 mM benzamidine and 1 mM phenylmethylsulfonyl fluoride. Cells were then placed in an ice bath and sonicated 20 ϫ 30 s (for a tip with a 1-cm diameter) at maximum output using an MSE sonicator. The lysate was centrifuged at 20,000 rpm for 2 h in a Sorval SS-34 rotor to remove cellular debris. This and all subsequent steps were performed at 4°C.
To remove nucleic acid, polyethyleneimine (10%, pH 7.5) was added dropwise to 74 ml of clarified cell extract to give a final concentration of 0.075%. After stirring for 15 min, the pellet was removed by centrifugation (15 min at 15,000 rpm). 70 ml (23 mg/ml) of the resulting supernatant, referred to as Fraction I (Fig. 2, lane b), was applied at 48 ml/h to a column (2.6 ϫ 30 cm, 160-ml bed volume) containing DEAE-Sepharose Fast Flow (Pharmacia) equilibrated in buffer A. Unbound material was removed from the column by washing with at least two column volumes of buffer A, and bound material was eluted with a 500-ml gradient of 50 -300 mM NaCl in buffer A. 5-ml fractions were collected and assayed by SDS-PAGE. Those containing the peak of SbcCD protein (eluting between 200 and 225 mM NaCl) were then pooled to constitute Fraction II (48 ml, 4.25 mg/ml; see Fig. 2, lane c).
Solid ammonium sulfate was then added to Fraction II, with constant stirring, to 30% saturation. After a further 30 min of stirring, the pellet collected after centrifugation (20 min at 20,000 rpm) was resuspended in 10 ml of buffer A, 5 mM Mn 2ϩ to become Fraction III (10 ml, 11.17 mg/ml) (see Fig. 2, lane d). This ammonium sulfate precipitate was then applied at 30 ml/h to a Sephacryl S500 (Pharmacia) gel filtration column (100 ϫ 2.6 cm, 530 ml of bed volume) equilibrated in buffer A, 5 mM Mn 2ϩ . Protein was eluted at 30 ml/h, and 5-ml fractions were collected. The peak of SbcCD protein was identified using SDS-PAGE, and the most active fractions were determined by assaying for ATP-dependent double-strand DNA exonuclease activity. In this way fractions were selected, pooled, and dialyzed thrice against 2 liters of buffer A. This step also acts to remove exogenous manganese, which results in precipitate formation when present in phosphate buffers. This dialysate was Fraction IV (45 ml, 0.82 mg/ml; see Fig. 2, lane e). Although it contained faint traces of a 87-kDa contaminating band, it was 98% homogeneous and considered to be suitable for physical analysis. Therefore 2-ml aliquots were frozen at Ϫ80°C until required.
To obtain SbcCD suitable for biochemical analysis (free of the 87-kDa contaminating protein), 3.28 mg (4 ml) of Fraction IV was dialyzed twice against 2 liters of buffer P and applied at 30 ml/h to a 5-ml prepacked hydroxyapatite column (Econo-Pac CHT-II Cartridge, Bio-Rad) equilibrated in the same buffer. Unbound protein was removed by washing the column with buffer P, and bound protein was eluted in a 90-ml phosphate buffer gradient (buffer P containing 10 -600 mM phosphate). 1.5-ml fractions were collected and assayed for ATP-dependent double-strand DNA exonuclease activity. SbcCD free of the 87-kDa contaminating band was identified by SDS-PAGE. SbcCD eluted in two peaks between 130 and 250 mM phosphate. The first peak (eluting between 130 and 160 mM phosphate) was free of contaminating protein and was therefore pooled and dialyzed twice against 2 liters of buffer A before concentrating with a Microsep 10-kDa centrifugal concentrator (Flowgen). This concentrate was called Fraction V (0.64 mg/ml) and contained 224 g of apparently homogeneous SbcCD protein in 350 l of buffer A (Fig. 2, lane f). 20-l aliquots of Fraction V were frozen in liquid nitrogen and stored at Ϫ80°C until required.
Amino-terminal microsequencing was used to verify that the polypeptides purified were SbcC and SbcD. The sequence of eight amino acids from the amino terminus of the putative SbcD polypeptide was MRILHTSD, which is in complete agreement with that predicted (14). The published amino acid sequence of the six residues at the amino terminus of SbcC is MKILSL (14), and the sequence determined for the putative polypeptide was MGILSL.
Biochemical Properties of Purified SbcCD-To determine the reaction requirements for the double-strand DNA exonuclease activity of SbcCD, purified protein (Fraction V) was assayed under various reaction conditions. Standard SbcCD reaction conditions (29) were used with any alterations indicated in the text. A time course showing the amount of limit product obtained when 0.5 nM SbcCD was incubated with 32 P end-labeled pUC19 DNA is shown in Fig. 3A. Approximately 25% of the 32 P-label appears as product at 30 min, and this was chosen as the standard reaction time (Fig. 3B).
SbcCD had an absolute requirement for Mn 2ϩ as a divalent cation (Table II). Activity was stimulated to a much lesser extent by Cu 2ϩ but not Co 2ϩ , Fe 2ϩ , Zn 2ϩ , Mg 2ϩ , Cu 2ϩ , or Ca 2ϩ (Table II). No activity was detected in the presence of Mg 2ϩ (up to 50 mM; data not shown). In the presence of 1 mM ATP, Mn 2ϩ stimulated activity at concentrations of 2 mM and above. However, in the presence of 2.5 mM ATP, activity was not seen until the Mn 2ϩ concentration was 3 mM or above (Fig. 4A).
SbcCD required a nucleoside triphosphate co-factor for activity (Table II). In the presence of 5 mM Mn 2ϩ , maximum activity was observed with 1-4 mM ATP. Higher concentrations of ATP, 5-10 mM, were inhibitory (Fig. 4B). When SbcCD activity was assayed with increasing concentrations of ATP in the presence of 10 mM Mn 2ϩ , no such inhibition was observed (Fig. 4B). This inhibition and the lag seen in the stimulation of SbcCD activity by increasing the concentration of ATP from 1 to 2.5 mM (Fig. 4A) most likely reflect the chelation of free Mn 2ϩ by ATP. ATP␥S and GTP exerted a slight stimulatory effect on SbcCD activity, whereas the other nucleotides tested had no effect (Table II).
SbcCD was optimally active in 50 mM NaCl; however, it became increasingly sensitive to NaCl at concentrations greater than 100 mM (Fig. 4C). SbcCD was functional from pH 7 to 9.5. Acidic pH was extremely inhibitory, whereas alkaline pH was less so (data not shown). The more physiological pH of 7.5 was chosen for use in assays.
Mechanism of SbcCD Action-In an effort to determine if SbcCD acts in a catalytic or stoichiometric fashion, increasing concentrations of 32 P end-labeled DNA were incubated with 0.25, 0.5, or 1 nM SbcCD (Fraction V), and the V m and apparent K m for double-strand DNA ends were calculated. Fig. 5A and Table III show that V m increases linearly with increasing SbcCD protein concentration. This indicates that SbcCD acts in a catalytic manner. The apparent K m of SbcCD for doublestrand ends also increases ( Fig. 5A and Table III). K m should remain constant with increasing protein concentration, suggesting that some factor in the standard reaction mixture is limiting.  In the course of these studies it was noticed that a number of substrate molecules always escaped degradation, except when incubated with high concentrations of SbcCD (data not shown).
This suggested that SbcCD acts processively. To determine whether this was so, 1 nM SbcCD (Fraction V) was incubated with uniformly labeled DNA, and reaction products were analyzed at various time points over a 60-min period. In parallel, SbcCD was incubated with unlabeled DNA and then challenged at 3 min with uniformly labeled DNA (Fig. 5B). When labeled DNA was added as a challenge, it was not degraded at the same rate seen when SbcCD was incubated with labeled DNA alone. This experiment suggests that very little free SbcCD is available 3 min into a 60-min reaction and that SbcCD acts processively.
Physical Analysis of SbcCD-An estimate of the relative molecular mass of SbcCD was made by applying 400 g of purified protein (Fraction IV, prepared in the absence of Mn 2ϩ )

FIG. 3. Time course of SbcCD digestion.
A, an autoradiograph illustrating the amount of SbcCD-limit product produced over a 60-min time period (lanes a-h). Lane i shows substrate incubated for 60 min in the absence of SbcCD. B, amount of limit product formed with time (obtained by quantification of the bands in A).

TABLE II
Effect of divalent cations and nucleotides on the double-strand DNA exonuclease activity of SbcCD 0.5 nM SbcCD (Fraction V) was incubated with 3Ј-32 P end-labeled linear DNA (7.5 nM double-strand ends), in a 20-l reaction volume, at 37°C for 30 min. The reactions that tested the effect of divalent metal ions were carried out in 1 mM ATP, 1.25 mM dithiothreitol, 2% glycerol, 100 g/ml BSA, 25 mM Tris, pH 7.5, and the metal ions indicated at 5 mM. The reactions that tested the effect of nucleotide were carried out in 5 mM Mn 2ϩ , 1.25 mM dithiothreitol, 2% glycerol, 100 g/ml BSA, 25 mM Tris, pH 7.5, and the nucleotide indicated at 1 mM. Following the reactions, products were resolved on a 1% agarose gel and quantified on a PhosphorImager. In each case the amount of product formed under standard SbcCD exonuclease conditions was defined as 100%. Note: protein was diluted 1 in 100 in buffer A containing 500 g/ml BSA. Because buffer A contains 1 mM EDTA, this results in a final concentration of 0.1 mM EDTA in the reaction mix.  4. Effect of buffer conditions on the SbcCD double-strand DNA exonuclease activity. Assays were performed as in the legend to Table II with the alterations indicated. A, titration of SbcCD with increasing concentrations of Mn 2ϩ in the presence of 1 (q) or 2.5 mM (ࠗ) ATP. B, titration of SbcCD with increasing concentrations of ATP in the presence of 5 (q) or 10 mM (E) Mn 2ϩ . C, effect of NaCl concentration. Assays were as described above except that NaCl was present at the concentration indicated. All reactions contained 0.5 nM SbcCD (Fraction V). In each case the amount of product formed under standard SbcCD assay conditions was defined as 100%.
to a FPLC gel filtration column, and elution was monitored by following the absorbance at 280 nm (Fig. 6A). SbcC and SbcD did not co-migrate on this column. The molecular mass of both polypeptides was estimated by comparing the elution volume of SbcC and SbcD with that of protein standards (Fig. 6B). SbcC eluted with a relative molecular mass of 1150 kDa, much higher than that predicted for a monomer. Whether SbcC migrated anomalously, eluted at 1150 kDa as a result of interaction between monomers, or both was not clear. SbcD eluted as a broad peak ranging from 40 to 310 kDa. The peak fraction of SbcD had a relative molecular mass of 85 kDa, suggesting the majority of SbcD is in the form of a dimer.
Because Mn 2ϩ stimulated the double-strand DNA exonuclease activity of SbcCD, its effect on the elution properties of SbcC and SbcD were investigated. 100 g of Fraction IV (purified in the absence of Mn 2ϩ ) was applied to an analytical HPLC column in the presence and the absence of 5 mM Mn 2ϩ . Fractions were analyzed by SDS-PAGE. In the presence of Mn 2ϩ , the peak of SbcD appeared in the same fractions as the peak of SbcC (Fig. 7A). In the absence of Mn 2ϩ the peak of SbcC and SbcD did not co-migrate (Fig. 7B). This suggested that Mn 2ϩ promotes interaction between SbcC and SbcD. When Mn 2ϩ was present the molar ratios of SbcC:SbcD in fractions 12, 13, and 14 ( Fig. 7A) were 1.3, 0.9, and 1.1, respectively (judged by densitometry), suggesting a ratio of 1:1 in these peak fractions. In the absence of Mn 2ϩ the molar ratios of SbcC:SbcD in fractions 12, 13, and 14 ( Fig. 7B) were 6.8, 4.3, and 2.9, respectively.
Sedimentation equilibrium analysis (provided as a service by the United Kingdom National Center for Molecular Hydrodynamics, University of Leicester) was used to gain an absolute estimate of the native molecular mass of SbcCD (Fraction IV, purified in the absence of Mn 2ϩ ). In the absence of Mn 2ϩ a species with a molecular mass of 720 kDa was detected. In the presence of Mn 2ϩ a species with a molecular mass of 1210 kDa was observed. Sedimentation equilibrium analysis gives an estimate of mass independent of protein shape; therefore association was taking place in both the presence and the absence of Mn 2ϩ .
Prediction of SbcD Secondary Structure-SbcD contains the conserved phosphoesterase signature sequence DXHX n GDXX-DX n GNH(D/E) (n ϭ ϳ25) found in a family of phosphoesterases (15,16). Several protein phosphatase members of this family have had their x-ray structures solved (33)(34)(35). At the active site of these enzymes two metal ions are co-ordinated by a ␤-␣-␤-␣-␤ (␤ ϭ ␤ sheet, ␣ ϭ ␣ helix) secondary structure motif. The conserved elements of the phosphoesterase signature sequence map to the carboxyl termini of the ␤ sheets within this structure. i.e. ␤-DXH-␣-␤-GDXD-␣-␤-GNH(D/E). A secondary structure prediction was generated to determine if SbcD had the potential to form a ␤-␣-␤-␣-␤ secondary structure in the  Table II with the concentration of DNA indicated. Data were fitted using GRAPHIT software. B, SbcCD processivity determination. SbcCD was incubated with uniformly 32 Plabeled DNA in the absence of challenge DNA (q) or incubated with unlabeled DNA and then challenged with uniformly 32 P-labeled DNA after 3 min (E). Samples were taken at the times indicated.  region of the conserved phosphoesterase signature sequence. The results are shown in Fig. 8. The conserved phosphoesterase motifs were found in regions predicted to form loops next to sequences that have the potential to form a ␤-␣-␤-␣-␤ secondary structure. In addition, the conserved phosphoesterase motifs lie adjacent to the carboxyl termini of the ␤-sheets within this ␤-␣-␤-␣-␤ structure. These data indicate that SbcD has the potential to form a metal co-ordinating unit like that which exists within the protein phosphatases. DISCUSSION An E. coli strain carrying a chromosomal deletion of the sbcCD operon was constructed and used to overexpress the SbcC and SbcD polypeptides from a plasmid containing the sbcC and sbcD genes under the control of the strong trc promoter. Following 6 h of IPTG induction, both polypeptides were co-expressed to approximately 6% of total cell protein. A purification scheme was devised that yielded 2.53 mg of soluble SbcCD protein from 20 g (wet weight) of cells. The SbcC and SbcD polypeptides copurified through three chromatographic steps and an ammonium sulfate precipitation step. A significant reduction in yield had to be accepted due to the presence of an 87-kDa contaminating protein (Table I and Fig. 2).
The double-strand exonuclease activity of SbcCD has an absolute requirement for Mn 2ϩ ions, other divalent metal ions tested failed to stimulate activity (Table II). SbcCD is most active under conditions of high free Mn 2ϩ concentration (Fig. 4,  A and B). SbcD contains the phosphoesterase signature sequence and has the potential to form a ␤-␣-␤-␣-␤ secondary structure in and around this conserved sequence (Fig. 8). Many members of this class of protein require transition metals (including Mn 2ϩ ) for activity (15). For example, the recombinant catalytic subunit of protein phosphatase 1 (from rabbit) is produced as an inactive enzyme in E. coli that can only be activated by Mn 2ϩ . However, it has recently been demon-strated that a combination of Fe 2ϩ and Zn 2ϩ (but not the individual metal ions) can significantly activate this enzyme (36). These observations suggest that SbcD may also be a metalloprotein capable of co-ordinating two metal ions via the ␤-␣-␤-␣-␤ secondary structure. A two-metal catalytic mechanism has previously been proposed for the 3Ј to 5Ј exonuclease activity of DNA polymerase I (37).
SbcCD has a requirement for a nucleoside triphosphate cofactor (Table II). Stimulation was only seen when the concentration of ATP was lower than that of Mn 2ϩ (Fig. 4, A and B). This is reminiscent of the E. coli RecBCD enzyme whose nonspecific nuclease activities are more active when the concentration of Mg 2ϩ ions is greater than that of ATP (38,39). SbcC possesses the ATP-A and ATP-B nucleotide binding motifs; therefore it is likely to be the component of the SbcCD protein that interacts with ATP. To test whether ATP hydrolysis was required for activity, nuclease assays were performed in the presence of the nonhydrolyzable analog ATP␥S. The activity in the presence of 1 mM ATP␥S was 30% of that in the presence of 1 mM ATP, suggesting that hydrolysis is not required for activity. The RAD50 protein only binds double-strand DNA in the presence of an adenosine triphosphate at concentrations of 2-5 mM. High level ATPase activity is not associated with this protein (40). The DNA-binding protein MukB only binds ATP and GTP at concentrations Ͼ 0.1 mM in the presence of Zn 2ϩ . No ATPase or GTPase activity has been detected for MukB (24).
SbcCD acts catalytically and processively (Fig. 5, A and B). In the experiment that demonstrates that SbcCD acts catalytically, K m (app) varies with protein concentration (Table III), suggesting that a factor in the reaction is limiting.
SbcC and SbcD co-eluted throughout purification. It was surprising to find that purified SbcC and SbcD, at a concentration of 2 mg/ml, did not co-migrate when applied to a gel filtration column (Figs. 6A and 7B). In the final stage of purification, protein was applied to a hydroxyapatite column at 0.82 mg/ml, yet both polypeptides still co-eluted (Fig. 2). Therefore the dissociation of SbcC and SbcD that was observed was not due to a concentration effect alone. Human protein phosphatase 1, when purified from native sources, does not require metal ions for activity yet is converted into a metal ion-dependent form after long term storage. This suggests that residues in the vicinity of the active site alter in conformation, and then metal ions are released (Ref. 35 and references therein). Perhaps similar changes result in the dissociation of the SbcCD complex.
FPLC gel filtration analysis revealed that SbcC and SbcD have relative molecular masses of 1150 and 85 kDa, respectively (Fig. 6B). This suggests that in the peak fraction SbcD exists as a dimer and SbcC exists as a higher order multimer. Mn 2ϩ ions, in addition to stimulating the double-strand exonuclease activity of SbcCD, also promoted an interaction between SbcC and SbcD. In the presence of Mn 2ϩ , SbcD elutes in a 1:1 molar ratio with SbcC at high molecular mass. A similar metalmediated interaction is observed when the RuvB protein of E. coli forms dodecamers in the presence of Mg 2ϩ ions (41). The Mn 2ϩ stimulation of exonuclease activity indicates that these ions are important catalytically. However, the observation that Mn 2ϩ ions promote an interaction between SbcC and SbcD suggests that Mn 2ϩ also plays a structural role, perhaps by inducing a conformational change in SbcD that favors interaction with SbcC.
Sedimentation equilibrium analysis detected two species with molecular masses of 720 (in the absence of Mn 2ϩ ) and 1210 kDa (in the presence of Mn 2ϩ ). These data indicate that an interaction is taking place in the absence of Mn 2ϩ and that the relative molecular mass of SbcC in the absence of Mn 2ϩ estimated by FPLC gel filtration was an overestimate. This anomalous migration can be explained by sequence analyses predicting that SbcC forms a filamentous protein with two globular domains linked by a central rod. Proteins with this shape do not migrate typically upon gel filtration. From the gel filtration data it is clear that under the conditions used SbcC and SbcD do not readily associate in the absence of Mn 2ϩ . It is also probable that under the sedimentation conditions used to measure the 720-and 1210-kDa species, the monomer masses of SbcD and SbcC (44.7 and 118.7 kDa, respectively) would be hidden. 3 The 720-kDa species might represent a hexamer of SbcC. This figure is very close to that expected for a hexamer of SbcC (712.2 kDa). The molecular mass of 1210 kDa for the SbcCD complex in the presence of Mn 2ϩ is close to that predicted for SbcC 6 SbcD 12 (1249 kDa). This would argue in favor of a 2:1 ratio of SbcD to SbcC as opposed to the 1:1 ratio suggested by densitometry. Further experiments are required to resolve the precise stoichiometry of the complex. This work is a first step in understanding the mechanism of SbcCD action. The SbcC and SbcD polypeptides interact to form a high molecular weight complex in the presence of manganese ions. Together, in the presence of ATP, both polypeptides function as a processive double-strand exonuclease. SbcC is an SMC family member, and a number of these proteins were isolated as or have been shown to be DNA-binding proteins. SbcD is a member of a group of proteins that have in common the ability to hydrolyze phosphoester bonds (15). We suggest that SbcC is the main DNA-binding subunit of the SbcCD protein (which is activated by ATP-binding or hydrolysis) and that SbcD is a metalloprotein that contains the catalytic center for the hydrolysis of DNA. Exactly how both polypeptides interact with each other and palindromic DNA remains to be investigated.