The RecD subunit of the RecBCD enzyme from Escherichia coli is a single-stranded DNA-dependent ATPase.

We have expressed the RecD subunit of the RecBCD enzyme from Escherichia coli as a fusion protein with a 31-amino acid NH2-terminal extension including 6 consecutive histidine residues (HisRecD). The overexpressed fusion protein can be purified in urea-denatured form by metal chelate affinity chromatography. The mixture of renatured HisRecD protein and the RecB and RecC proteins has a high level of ATP-dependent nuclease activity with either single- or double-stranded DNA, enhanced DNA unwinding activity, enhanced ATP hydrolysis activity in the presence of a small DNA oligomer cosubstrate, and chi-cutting activity. These are all characteristics of the RecBCD holoenzyme. The HisRecD protein by itself hydrolyzes ATP in the presence of high concentrations of single-stranded DNA (polydeoxythymidine). The activity is unstable at 37 degrees C, but is measurable at room temperature (about 23 degrees C). The HisRecD has very little ATPase activity in the presence of a much shorter single-stranded DNA (oligodeoxy(thymidine)12). HisRecD hydrolyzes ATP more efficiently than GTP and UTP, and has very little activity with CTP. We also purified a fusion protein containing a Lys to Gln mutation in the putative ATP-binding site of RecD. This mutant protein has no ATPase activity, indicating that the observed ATP hydrolysis activity is intrinsic to the RecD protein itself.

The RecBCD enzyme from Escherichia coli is an important enzyme in the DNA metabolism of the cell, acting in homologous recombination, resistance to UV irradiation and chemical DNA damaging agents, and degradation of foreign DNA (reviewed in Refs. [1][2][3]. The enzyme catalyzes several reactions in vitro. ATP hydrolysis enables it to unwind double-stranded DNA (4 -6), and the enzyme is a potent nuclease on double-stranded DNA in the presence of ATP and excess magnesium ion (7,8). The doublestrand nuclease activity of RecBCD is suppressed when it encounters a sequence (5Ј-GCTGGTGG) in the DNA (9 -12). The enzyme continues to unwind the DNA past the sequence (11) and the unwound DNA is a substrate for recombination catalyzed by the RecA protein (13)(14)(15). RecBCD enzyme is also an ATPstimulated nuclease with single-stranded DNA, and singlestranded DNA stimulates ATP hydrolysis (7,8).
The first preparations of the RecBCD enzyme were thought to contain only the proteins encoded by the recB and recC genes (8, 16 -19). However, one experiment indicated the possible existence of an additional component, essential for enzymatic activity, which could be separated from both the RecB and RecC proteins by treating the enzyme with high salt concentrations (20). A later study showed that this was a third protein subunit, called RecD, encoded by a gene (recD) adjacent to the recB gene (21).
The function of the RecD subunit in catalysis by RecBCD enzyme is not clear, although it is implicated in several activities of the enzyme. The identification of the RecD subunit was made possible by the fact that it is required for high levels of nuclease activity with single-or double-stranded DNA (21,22). Amino acid sequence analysis shows that RecD and RecB contain consensus ATP-binding motifs as well as other sequences similar to those in a family of DNA helicases (23). A RecBCD enzyme with a mutation in the ATP-binding site of RecB retains single-stranded DNA-dependent ATPase activity, which we concluded is catalyzed by the wild-type RecD subunit in that enzyme (24). The enzyme with a Lys to Gln mutation in the putative ATP-binding site of the RecD subunit (the RecBCD-K177Q enzyme) retained all the activities for which we tested (ATPase, nuclease, and helicase), although each was reduced in rate and other properties (25,26). The effect of sequences on the nuclease activity has led to the proposal that causes inactivation of, and perhaps ejection of, the RecD subunit (2,(27)(28)(29)(30). This is consistent with the fact that RecBC is a helicase but lacks significant nuclease activity (31)(32)(33), and that recombination in recD Ϫ cells is not affected by sequences (22).
Purification of the RecD protein has been difficult and has led to little insight into its function in the reactions catalyzed by RecBCD. No catalytic activity was found to co-purify with the RecD subunit isolated from the RecBCD holoenzyme (20,21). The RecD protein was overexpressed and purified from inclusion bodies, and the isolated subunit restored nuclease activity and -specific cleavage to RecBC, but no activity was found in RecD alone (32). The difficulty of obtaining RecD protein in native form led us to prepare RecD fused to an NH 2 -terminal peptide containing six consecutive histidine residues. The protein (HisRecD) can be purified by metal chelate affinity chromatography in either native or denatured form, and reconstituted with the RecB and RecC proteins to obtain high levels of ATP-dependent nuclease activity, and other activities characteristic of the RecBCD enzyme. The HisRecD protein alone catalyzes ATP hydrolysis in the presence of high concentrations of single-stranded DNA.

Materials
Isopropyl ␤-D-thiogalactopyranoside, guanidinium hydrochloride, and DTT 1 were from U. S. Biochemical Corp.; Triton X-100 was from J. * This research was supported by Grant GM39777 from the National Institutes of Health. 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.
RecBCD, RecB, and RecC Purification-RecBCD enzyme was purified as described (36), and the concentration was determined from the absorbance at 280 nm, using ⑀ 280 ϭ 4 ϫ 10 5 M Ϫ1 cm Ϫ1 (6). The RecB and RecC proteins were purified and quantitated, using ⑀ 280 ϭ 1.7 ϫ 10 5 M Ϫ1 cm Ϫ1 (RecB), and ⑀ 280 ϭ 2 ϫ 10 5 M Ϫ1 cm Ϫ1 (RecC) as described (32,33). The concentrations of each enzyme or subunit given under "Results" and in the figure legends are those based on the absorbance readings, unless stated otherwise. However, each enzyme preparation apparently contained some inactive enzyme. The nuclease activity of RecBCD was maximal with about 2 enzyme molecules per DNA end (measured as in Ref. 25 Construction of a Plasmid Expressing the His-tagged RecD Protein (pHisRecD)-The His-tagged RecD protein (HisRecD) was expressed using the vector pTrcHisB (Invitrogen Corp.). The recD gene was transferred to the vector as follows. A 712-base pair fragment from the plasmid pPvSm19, which contains the recD gene (36), was amplified by polymerase chain reaction. The downstream primer annealed across the BspEI site (bold) within the gene (primer number 1: 5Ј-GGCATCT-TCCGGAATG) and the upstream primer annealed at the 5Ј-end of the gene (primer number 2: 5Ј-GGAAGATCTGCAAAAGCAATTACTGG). Primer number 2 is partially complementary to the recD gene (underlined) and introduces a BglII site upstream of the gene (bold). The amplified 712-base pair fragment was digested with BspEI and BglII and ligated into pPvSm19 cleaved with the same enzymes (the BglII site in pPvSm19 is within a portion of the recB gene contained in this plasmid). The recD gene was removed from this plasmid by cleavage with BglII and BamHI (BamHI cleaves downstream of the 3Ј-end of the recD gene in pPvSm19), and ligated into pTrcHisB cleaved with the same enzymes, to produce pHisRecD. The recD fragment is in the correct orientation for expression when the BglII sticky ends are joined to the compatible BamHI ends.
The resulting plasmid (pHisRecD, 6.6 kilobase pairs) encodes the RecD protein fused to a 31-amino acid leader peptide (3.4 kDa) containing six consecutive histidine residues. The protein is expressed from a trc promoter and a ribosome-binding site within the vector sequence. The RecD amino acid sequence was altered slightly during the subcloning procedure, since the first two residues in RecD, Met-Lys, are changed to a single Asp in the fusion protein. The 5Ј terminus of the fusion gene was sequenced to confirm the structure of the junction between the vector and recD gene sequences.
The NH 2 -terminal leader peptide encoded by the vector also includes a sequence recognized by the protease enterokinase (-Asp-Asp-Asp-Asp-Lys-). However, we have been unable to cleave the NH 2 -terminal peptide with enterokinase (Invitrogen Corp.), under a variety of reaction conditions, enterokinase concentrations, incubation times, etc. Consequently we have worked with the fusion protein itself, since it gives high levels of activity despite the additional peptide.
Construction of a Plasmid Expressing His-tagged RecD-K177Q Protein (pHisRecD-K177Q)-The plasmid pPvSm19-DK177Q, encoding a Lys to Gln mutation in the putative ATP-binding site of RecD (36), was digested with BsmI, and the 442-base pair fragment containing the mutation was ligated into pHisRecD cleaved with the same enzyme. The presence of the mutation in the resulting plasmid, pHisRecD-K177Q, was also confirmed by DNA sequencing.

Methods
Purification and Renaturation of His-tagged RecD Protein-The His-RecD protein was expressed in E. coli strain JM109 (recA Ϫ recBCD ϩ ) or V186 (⌬recBCD) transformed with pHisRecD. Cells were grown in 25 ml of LB broth (38) at 37°C. Isopropyl ␤-D-thiogalactopyranoside (1 mM) was added when the A 650 ϭ 0.4, and growth was continued for 4 h. The cells were harvested and resuspended in either denaturing binding buffer (20 mM sodium phosphate, pH 7.8, 0.5 M NaCl, 6 M guanidinium hydrochloride) or native binding buffer (20 mM sodium phosphate, pH 7.8, 0.5 M NaCl). For the denatured preparation, the resuspended cells were rocked gently for 10 min at room temperature, and then the lysates were cleared by centrifugation for 10 min at 17,000 ϫ g. The lysate was applied to a 2-ml nickel-containing column (ProBond resin, Invitrogen Corp.) in denaturing binding buffer. The column was washed in 20 mM sodium phosphate, pH 6.0, 8 M urea, 0.5 M NaCl, and then the same mixture at pH 5.3. The HisRecD fusion protein was eluted in denaturing elution buffer (20 mM sodium phosphate, pH 4.0, 8 M urea, 0.5 M NaCl). Its concentration was determined from the absorbance at 280 nM, using ⑀ 280 ϭ 48,500 M Ϫ1 cm Ϫ1 , calculated for HisRecD (the NH 2 -terminal leader peptide contains a single tyrosine residue, and no tryptophans). The typical yield was about 0.5 mg from 25 ml of culture.
For preparation under native conditions, lysozyme (0.1 mg/ml) was added to the resuspended cells, followed by incubation for 10 min at 37°C, 5 min on ice, and then the lysate was cleared by centrifugation as above. The lysate was applied to a 2-ml ProBond column which was washed with 20 mM sodium phosphate, pH 7.8, 0.5 M NaCl, then with 20 mM sodium phosphate, pH 6.0, 0.5 M NaCl, and finally the bound protein was eluted in a gradient of 50 -500 mM imidazole in 20 mM sodium phosphate, pH 6.0, 0.5 M NaCl.
The denatured HisRecD protein was renatured by first diluting with denaturing elution buffer to less than 50 g/ml, and then dialyzing in a collodion membrane (25-kDa cut-off; Schleicher & Schuell) at 4°C against renaturation buffer (10 mM Tris-HCl, pH 8.0, 1 mM DTT, 0.1% Triton X-100, 0.5 M NaCl, 20% glycerol). The denatured HisRecD protein was stable when stored at 4°C. However, we have not found conditions under which the renatured HisRecD can be stored without substantial loss of reconstitution activity. Thus, HisRecD was renatured as needed.
Reconstitution of HisRecD with RecBC-The reconstitution method used for most experiments was as follows: the renatured HisRecD protein (in renaturation buffer) was mixed with RecC (in RecBCD dilution buffer: 10 mM potassium phosphate, pH 7.0, 0.1 mM EDTA, 0.1 mM DTT, 0.4 mg/ml bovine serum albumin, 20% glycerol), and placed at room temperature for 4 h. RecB (in RecBCD dilution buffer) was then added and incubation was continued overnight at room temperature. The final reconstitution mixture usually contained 80 nM RecB and RecC and 300 nM HisRecD, in a 1:1 mixture of renaturation buffer and RecBCD dilution buffer. A few reconstitutions were done by adding HisRecD to RecBC and incubating for 2-4 h at room temperature. The reconstituted RecBC(HisRecD) enzyme is relatively stable. The enzyme lost only 20% of its nuclease activity on double-stranded DNA after 2 months at Ϫ20°C, in this solution.
Native Polyacrylamide Gel Electrophoresis-Native gel electrophoresis was carried out on 5% polyacrylamide gels (30:1, acrylamide:bisacrylamide) prepared in imidazole/HEPES buffer (43 mM imidazole, 35 mM HEPES, pH 7.4) containing 10 mM MgCl 2 . The gels were pre-run at 5.5 V/cm for 30 min before protein samples were loaded, and then the samples were run at 3.5 V/cm. The gels were silver-stained with Rapid-Ag-Stain (ICN Biomedicals).

RecBCD Enzyme Assay Methods
The standard reaction conditions we used for these experiments were 43 mM imidazole, 35 mM HEPES buffer, pH 7.4, 10 mM MgCl 2 , 0.67 mM DTT, at 37°C, unless indicated otherwise. The reconstituted RecB-C(HisRecD) prepared as above was diluted 10 -50-fold with a 1:1 mixture of renaturation buffer and RecBCD dilution buffer, and then diluted 10 -100-fold further into the reaction mixtures. Thus, the components of the renaturation and reconstitution buffers were also present in the reaction mixtures. The other enzymes used in these experiments (RecBCD holoenzyme, RecBC, and HisRecD alone) were also diluted with the same buffer mixture, so that all enzymes were compared under the same reaction conditions. ribonucleoside triphosphate; nt, nucleotide; pd(T) 12 and pd(T) 25-30 , 5Јphosphorylated oligodeoxythymidine, 12 residues, and a mixture of 25-30 residues, in length. -Specific Cleavage Reactions--Specific cleavage reactions were done essentially as in Ref. 39 using pBR322-FH (11), containing two sequences in opposite orientations. The plasmid was cleaved with ClaI, treated with calf intestinal alkaline phosphate, and 5Ј-end-labeled with polynucleotide kinase, and [␥-32 P]ATP. Markers for -specific cleavage were prepared by cleaving the labeled, ClaI-cut DNA again with AvaI.

Purification of HisRecD and Reconstitution with RecB and
RecC-A protein of the expected size (slightly larger than RecD) was found in lysates of cells transformed with pHisRecD, after induction with isopropyl ␤-D-thiogalactopyranoside. We purified the histidine-tagged RecD protein (HisRecD) under both native and denaturing conditions, following the procedures recommended by Invitrogen Corp. The HisRecD could be highly purified in a single step ( Fig. 1; some preparations contained a small amount of an impurity, probably the 21-kDa protein seen previously in His-tagged proteins purified from E. coli (40)).
The assay for RecD activity was the ability to restore ATPdependent nuclease activity to RecBC, on linear single-and double-stranded DNA. The nuclease activities with both singleand double-stranded DNA (Fig. 2) are completely dependent on the presence of RecB, RecC, HisRecD protein, and ATP (not shown), as expected for RecBCD enzyme activity (8,21,32). The reconstituted enzyme had a high level of activity in this assay. The activity of the reconstituted enzyme depends on: 1) the efficiency with which the three protein subunits assemble to make RecBC(HisRecD), and 2) the fraction of renatured HisRecD protein which is active. We estimated both quantities by comparing the nuclease activity of the reconstituted enzyme to that of the RecBCD holoenzyme. The nuclease activity on double-stranded DNA with RecBC(HisRecD) prepared by mixing RecB and RecC (2 nM each) with a large excess of HisRecD (45 nM), followed by 10-fold dilution into the assay, was the same as that of about 0.06 -0.065 nM RecBCD in the assay (not shown). After accounting for the presence of inactive RecBCD and RecB/C protein (see "Experimental Procedures"), this suggests that about 60 -65% of the RecBC could be converted to RecBC(HisRecD): 65% ϭ (0.065/2 nM active RecBCD)/(0.2/4 nM active RecB/C) ϫ 100. The reconstituted enzyme used for the experiment in Fig. 2A was prepared by mixing HisRecD (290 nM) with an excess of RecB and RecC (400 nM each). The activity on double-stranded DNA, with 0.58 nM HisRecD, and 0.8 nM RecB and RecC in the assay mixture, was about the same as that of 0.17 nM RecBCD ( Fig. 2A). This suggests that at least (0.17/2)/0.58 ϫ100 ϭ 15% of the HisRecD is active, or 23%, if only 65% of RecB and RecC assemble to RecBC(HisRecD). (Both calculations assume that the specific activity of 100% reconstituted RecBC(HisRecD) would be the same as that of RecBCD.) The nuclease activity appeared when the proteins were re- constituted at room temperature (about 23°C), but little activity was found if the reconstitution was done on ice, under any conditions. The time required varied somewhat depending on the protein concentrations. Thus, the nuclease activity in a mixture of 3.5 nM RecBC and 45 nM HisRecD required about 2 h before reaching a maximal level. The rate of appearance of activity did not increase significantly if HisRecD was incubated overnight with either RecB or RecC and then the remaining subunit was added, as compared with adding HisRecD to RecBC. This suggests that the slow appearance of the nuclease activity is not due to slow assembly of HisRecD with only one of the other two proteins, since in that case overnight preincubation with that particular subunit would have allowed final assembly of the RecBC(HisRecD) complex to be rapid. (We did notice that slightly greater activity (about 2-fold) was obtained by adding the subunits sequentially (the method given under "Experimental Procedures") rather than incubating all three subunits together. Thus we used that procedure as the standard way to prepare RecBC(HisRecD) for further experiments.) The HisRecD purified under native conditions had lower activity than the renatured HisRecD, but greater activity was obtained when this HisRecD protein was first dialyzed against renaturation buffer. This protein was also less pure after the affinity chromatography step than was the denatured His-RecD. We also purified HisRecD from V186 cells (⌬recBCD) and obtained nuclease activity when the protein was added to RecB and RecC, indicating that the presence of the wild-type recBCD genes in JM109 did not affect the HisRecD preparation from those cells. Most further experiments were done with HisRecD purified under denaturing conditions from JM109.
We also tested several factors during renaturation for their effects on the yield of active HisRecD. Dialysis against renaturation buffer containing (NH 4 ) 2 SO 4 (0.5 M) worked about as well as NaCl, and both gave greater activity than dialysis against the same concentrations of KCl, NH 4 Cl, or potassium glutamate. Dialysis to lower the urea concentration in steps (4, 2, 1, 0 M) did not lead to significantly greater yield of activity. There was no effect of including ZnCl 2 (0.1 mM; see Ref. 32) in the dialysis buffer. Lower activity was observed when: 1) renaturation was done at room temperature rather than 4°C; 2) renaturation buffer contained 0.2 M NaCl rather than 0.5 M NaCl; 3) the renaturation buffer was MOPS, pH 6.5, rather than Tris-HCl, 8.0.
Native Polyacrylamide Gel Electrophoresis-The nuclease activities shown above strongly indicate that the HisRecD is bound to RecBC to make a RecBC(HisRecD) complex. We sought to detect the RecBC(HisRecD) complex directly by analyzing mixtures of the proteins on nondenaturing polyacrylamide gels (32). There was a band present in the mixture of RecB, RecC, and HisRecD (lane 1, Fig. 3) which co-migrates with RecBCD (lane 3) and is presumably RecBC(HisRecD). The renatured HisRecD does not form a clear band (lane 4), indicating that it may not be a single discrete species. The HisRecD protein would also have little charge and thus low mobility in the gel (pH 7.4), since its estimated pI is about 6.8, given its content of charged residues. The RecBC(HisRecD) band was not seen in gels prepared without MgCl 2 . We believe that Mg 2ϩ mainly stabilizes the reconstituted enzyme in the gel, since the nuclease activity of the RecBC(HisRecD) mixture was about the same whether or not MgCl 2 was present during the reconstitution (not shown).

Reconstitution of Other RecBCD Enzyme Activities
Unwinding Activity-The rate of DNA unwinding by the RecBCD enzyme, measured by the RecJ/ExoI coupled assay, is about 4-fold faster than that of the RecBC enzyme (37). Reconstitution of RecBC with HisRecD protein gave about a 2.5-fold increase in the rate of DNA unwinding in this assay (Fig. 4).
-Specific Cleavage-Reactions of RecBCD and RecBC(His-RecD) with a -containing DNA molecule are shown in Fig. 5. Each enzyme produces a full-length single-stranded product, as well as DNA fragments which comigrate with the markers from AvaI digestion. Since the AvaI sites are only 75 nt away from F, and 13 nt from H, these fragments serve as markers for -specific cleavage. The -specific bands were not produced in reactions with [5Ј-32 P]pBR322, which lacks sites, nor by the RecBC enzyme. This experiment shows that the reconstituted RecBC(HisRecD) enzyme is able to recognize and cleave double-stranded DNA near a sequence. -Specific cleavage has also been observed with RecBCD enzyme prepared by reconstituting RecB and RecC with native RecD protein (32).
ATP Hydrolysis Stimulated by pd(T) 12 -We found previously (41) that RecBCD has much greater ATPase activity than does RecBC in the presence of low concentrations of pd(T) 12 oligomers (K m for pd(T) 12 ϭ 4.5 M oligomers for RecBC versus 0.1 M for RecBCD). We believe that the ATP hydrolysis is catalyzed by the RecB subunit in both enzymes, since the RecBCD-K177Q mutant enzyme also has a low K m for pd(T) 12 , close to that of RecBCD. The reconstituted RecBC(HisRecD) enzyme also has much greater ATPase activity at 1.5 M pd(T) 12 compared with RecBC alone (Fig. 6). This observation provides further evidence for physical association of the proteins, since we believe that it is the presence of RecD (wild-type, mutant, or, in this case, HisRecD) in the RecBCD holoenzyme, that

ATP Hydrolysis by HisRecD Protein
In the course of studying ATP hydrolysis by the reconstituted enzymes, we also incubated identical samples of RecBC and HisRecD separately, to be used as controls. ATP hydrolysis was then measured at 37°C. Interestingly, the HisRecD protein alone catalyzed a small amount of ATP hydrolysis in these experiments with poly(dT) as the DNA co-substrate (Fig. 7,  closed circles), but the reaction stopped in about 2 min. A second burst of ATP hydrolysis occurred if more HisRecD was added, but it also stopped in about 2 min (Fig. 7, open circles). Little ATP hydrolysis was observed if the HisRecD was diluted and placed at 37°C for 2 min, and then ATP and other reaction components were added to initiate the reaction (Fig. 7, closed  squares). There was no detectable ATP hydrolysis by HisRecD in the absence of DNA (data not shown). These experiments suggested that HisRecD is a DNA-dependent ATPase but that it is not very stable and loses activity quickly at 37°C. Consistent with this conclusion, much greater ATP hydrolysis by HisRecD was observed when the reaction was done at room temperature (about 23°C; Fig. 7, closed triangles). The ATP hydrolysis activity of HisRecD was not stabilized significantly at 37°C by the presence of either bovine serum albumin or RecC (not shown; we did not add RecB, since it is itself an ATPase). The HisRecD protein must be more stable when it is assembled with RecBC, since the RecBC(HisRecD) enzyme is very active in the assays shown above, all carried out at 37°C, and it is also more stable in long-term storage than is HisRecD itself.
These experiments provide direct evidence that HisRecD, and presumably also RecD, is a DNA-dependent ATPase. There is ample reason to believe that RecD should be an ATPase (see "Introduction"), and it is unlikely that a contaminant would co-purify with the HisRecD prepared in this way. We nonetheless carried out the following control experiment to confirm that the activity is due to HisRecD itself. We prepared three cultures of JM109 cells, transformed with pHisRecD, pHis RecD-K177Q, or pTrcHisB (vector-only control). Cell growth, lysis, and metal-chelate affinity chromatography under denaturing conditions were carried out as described under "Experimental Procedures." The fractions eluting from the columns were analyzed by SDS-polyacrylamide gel electrophoresis. His-RecD was found only in fractions from pHisRecD or pHisRecD-K177Q-containing cells, and not in the pTrcHisB control (not shown). All three protein preparations contained a small amount of an impurity (see above). Fraction number 3 from each preparation was then renatured, and either reconstituted with RecB and RecC and assayed for nuclease activity, or assayed for ATPase activity directly. Neither activity was found with the vector-only control, but addition of either His-RecD or HisRecD-K177Q to RecB and RecC gave nuclease activity (Fig. 8A). The nuclease activity was greater with RecBC reconstituted with the wild-type HisRecD than with the HisRecD-K177Q mutant, as expected based on previous results  7-9 and 11-13). M, heat-denatured ClaI-cut [5Ј-32 P]pBR322-FH, cleaved again with AvaI. Numbers to the side indicate the sizes of the 5Ј-labeled fragments. ss, heat-denatured ClaI-cut [5Ј-32 P]pBR322-FH marker. Schematic below shows the structure of pBR322-FH. The arrows indicate the direction from which RecBCD must approach the sequence for recognition. The number of nucleotide residues from the sequence to the labeled 5Ј-end, corresponding to the approximate sizes of the fragments produced by the enzyme, are also indicated.  (25). However, we observed ATP hydrolysis activity only with the wild-type HisRecD, and not with HisRecD-K177Q, nor with the vector-only control (Fig. 8B). This result is consistent with the conclusion that the observed ATP hydrolysis activity is due to HisRecD, and also that the Lys to Gln mutation has greatly reduced the ATPase activity of RecD (as surmised previously but not shown directly (36,37)).
Masterson et al. (32) found no ATP hydrolysis activity in RecD protein purified under denaturing conditions and then renatured. However, RecD would be expected to have little activity in the reaction conditions apparently used in that work (37°C, 1.1 nM RecD, 30 M denatured DNA), if the native RecD protein behaves similarly to HisRecD. Lieberman and Oishi (20) also did not find ATPase activity with RecD (the "␣ subunit" in that work (21)), but the protein concentration was unknown, and probably was quite low.

Characterization of the ATP Hydrolysis Activity of HisRecD and Comparison to RecB
We then studied some of the properties of the HisRecD ATPase activity, and compared it to that of RecB, assayed under identical conditions (these conditions are not optimal for RecB, as that enzyme is somewhat more active at pH 7 than at pH 7.4, and it is active at 37°C). These experiments were done to obtain comparative information as to the specificity of the two proteins. These properties could of course change when the subunits are in the RecBCD holoenzyme. For example, the ATP hydrolysis activity of the isolated RecB subunit itself is much lower than that of RecB in RecBC or the RecBCD holoenzyme (32,33,41).
We initially used high concentrations of poly(dT) to test for ATP hydrolysis catalyzed by HisRecD, based on the results of previous experiments with the RecBCD enzyme (41). That enzyme exhibits ATP hydrolysis activity stimulated by poly(dT) binding at two sites. ATP hydrolysis by RecB is stimulated by a high affinity site, while DNA binding in a second site with much lower affinity appeared to stimulate ATP hydrolysis by RecD. Consistent with those observations, HisRecD has significant ATPase activity with 1 and 0.1 mM poly(dT) (average length: 221 nt), but much lower activity with 0.01 mM poly(dT) and almost none with pd(T) 12 (Fig. 9). There was also essentially no ATP hydrolysis by HisRecD with linear doublestranded DNA (XhoI-cut pTZpPB700, 0.01-1 mM nt residues; data not shown). The RecB reaction rate with 1 mM poly(dT) was about 30 -40-fold greater than that of HisRecD, under the reaction conditions of Fig. 9. The RecB-catalyzed reaction rate was about the same with pd(T) 12 and poly(dT), each at 1 mM nucleotide residues (not shown). Thus ATP hydrolysis by RecB was more than 1400-fold faster than that by HisRecD with pd(T) 12 as the DNA co-substrate.
The ATP hydrolysis activities of HisRecD and RecB depend rather differently on the Mg 2ϩ concentration (Fig. 10). HisRecD was maximal at about 2 mM Mg 2ϩ , and then decreased slightly at still higher concentrations (Fig. 10A). The activity of RecB increased up to 10 mM Mg 2ϩ , the highest concentration we tested (Fig. 10B). Thus, the ATP hydrolysis activity of the two proteins was about the same at low Mg 2ϩ (about 0.5 mM). Both proteins also hydrolyzed Ca⅐ATP, although less efficiently than Mg⅐ATP (not shown). There was no detectable ATP hydrolysis by HisRecD in the absence of DNA, with 2 mM Mg 2ϩ , under the conditions of Fig. 10 (data not shown). The Mg 2ϩ concentration has significant effects on the activity of RecBCD, particularly the relative levels of nuclease and helicase activity, and the levels of -specific versus nonspecific cutting (15,39,42). We note that RecBCD can unwind DNA when ATP is in excess over the Mg 2ϩ (5,43), conditions where neither subunit alone has much ATP hydrolysis activity.
HisRecD hydrolyzes ATP most efficiently, with lower activity with GTP and UTP, and almost none with CTP (Fig. 11A). ATP is also the best substrate for RecB under these conditions, but CTP, GTP, and UTP are hydrolyzed at a rate 35-50% that of ATP (Fig. 11B). The same relative rates of hydrolysis by RecB were seen at 1 mM NTP, 10 mM MgCl 2 , whereas we detected hydrolysis solely of ATP by HisRecD (24 nM) under those conditions (not shown).
We also tested the HisRecD protein for nuclease activity, using 5Ј-32 P-labeled single-stranded oligomers as the DNA substrate. We have observed no nuclease reaction with either HisRecD alone, or with HisRecD incubated with either RecC or RecB, at 200 or 500 M ATP (data not shown). We do detect nuclease activity with the purified RecBCD enzyme, the reconstituted RecBC(HisRecD) enzyme, and a low level of activity with the RecBC enzyme, using this assay. 2 These experiments thus provide no support for the possibility that the RecD subunit contains the nuclease active site of the RecBCD enzyme, despite the importance of RecD for this activity (20,21,32). DISCUSSION The reconstituted RecBC(HisRecD) enzyme has high levels of all the activities of RecBCD. The levels of activity show that the presence of the 31-residue amino-terminal peptide has little effect on the ability of the fusion protein to reconstitute with the other subunits and to influence the activity of the other subunits. We cannot be certain as to how the peptide might affect the activity and properties of RecD itself. In any case, the availability of this protein and the results we have so far will aid in further study of the function of RecD in RecBCD enzyme activity. HisRecD will also be a useful material for studying the RecD protein by itself, and this subunit reconstituted with the wild-type and with other mutant subunits.
The RecD subunit has been implicated to some degree in all activities of the RecBCD enzyme (ATP hydrolysis, helicase, nuclease, and the response to sequences). Indeed, it is difficult to assign separate functions to the subunits of the enzyme. Instead it appears that all three participate in the nuclease reaction on single-and double-stranded DNA, and in unwinding double-stranded DNA. An exception is that RecB alone has some helicase activity, but it is much slower than that of RecBCD (6,44).
The observation of ATP hydrolysis by HisRecD, and its properties, support the conclusion that RecD hydrolyzes ATP during the reaction catalyzed by RecBCD on double-stranded DNA (37). The specificity of HisRecD for single-stranded DNA is consistent with previous conclusions that RecD interacts with the partially unwound DNA produced by the helicase activity driven by ATP hydrolysis by RecB (24,45,46). The requirement of HisRecD for high poly(dT) concentrations may result from having isolated the subunit from its partners, since the local concentration of single-stranded DNA at the unwinding fork made by RecBCD would be relatively high, even at low total DNA concentration.
The nucleotide specificities of the two ATPase subunits are interesting in light of previous studies of the RecBCD enzyme. Wright et al. (7) observed double-strand nuclease activity by RecBCD with all eight ribo-and deoxyribonucleoside triphosphates. The activity showed sigmoidal dependence on the nucleotide concentration, and the minimum concentration required for activity went in the order ATP Ͻ GTP Ͻ UTP Ͻ CTP. We found later that the RecBCD-K177Q mutant enzyme required a slightly higher ATP concentration than did RecBCD before double-strand nuclease activity was observed (25), as did RecBC for detectable DNA unwinding (37). We interpreted these observations to indicate that ATP hydrolysis by both RecB and RecD, in RecBCD, contribute to the DNA unwinding and nuclease reactions, at low ATP concentrations. ATP hydrolysis by RecB alone is sufficient for these reactions (in either RecBC or RecBCD-K177Q), but a higher ATP concentration is required. The nucleotide specificity of HisRecD together with the observations of Wright et al. (7) are consistent with this earlier interpretation that ATP hydrolysis by both subunits contributes to the reaction (unwinding and/or nuclease) with double-stranded DNA.
ATP hydrolysis by RecD can have no essential role in RecBCD helicase activity, since RecBC is a helicase (31,32,33). The protein may function to accelerate translocation and enhance processivity during unwinding, since RecBC and RecBCD-K177Q are slower and less processive than RecBCD (26,33,37). The dependence of ATP hydrolysis rate on DNA length can be diagnostic of a mechanism where ATP hydrolysis is coupled to movement along the DNA (47). Although further study of the kinetics of ATP hydrolysis by HisRecD (or RecD itself) is clearly necessary, our observation that poly(dT) stimulates ATP hydrolysis much more effectively than pd(T) 12  consistent with this proposed function for RecD. Finally, although RecD is dispensable for helicase activity, it cannot be ruled out that ATP hydrolysis by RecD can be coupled to DNA unwinding. A possible mechanism is that RecB must begin to unwind the DNA, but ATP hydrolysis by RecD could also contribute to DNA unwinding once the process has begun.
Finally, RecD is required for the high nuclease activity of RecBCD with single-and double-stranded DNA (20,21,32). Several recent experiments indicate that inactivates RecD, at least as far as its function in the nuclease reaction (28 -30). The -inactivated enzyme does not appear to reactivate readily under some conditions, since a sequence in one DNA molecule can lead to protection of a non--containing DNA in vitro (9) and in vivo (12,29,30). Overexpression of RecD protein in vivo can suppress the effect of a sequence (29,30), an observation that supports the proposal that RecD is ejected from the RecBCD holoenzyme at a sequence (27,29,30), and that the RecD protein is unable to reassemble when it is present in normal (low) amounts in the cell. Our results indicate that dissociated RecD subunit would lose activity at 37°C, if its stability is similar to that of HisRecD. Thus the results are consistent with the proposal that sequences act to turn off RecBCD nuclease activity irreversibly in the cell (30).