Effect of mutations in the C-terminal domain of Mu B on DNA binding and interactions with Mu A transposase.

Bacteriophage Mu transposition requires two phage-encoded proteins, the transposase, Mu A, and an accessory protein, Mu B. Mu B is an ATP-dependent DNA-binding protein that is required for target capture and target immunity and is an allosteric activator of transpososome function. The recent NMR structure of the C-terminal domain of Mu B (Mu B223-312) revealed that there is a patch of positively charged residues on the solvent-exposed surface. This patch may be responsible for the nonspecific DNA binding activity displayed by the purified Mu B223-312 peptide. We show that mutations of three lysine residues within this patch completely abolish nonspecific DNA binding of the C-terminal peptide (Mu B223- 312). To determine how this DNA binding activity affects transposition we mutated these lysine residues in the full-length protein. The full-length protein carrying all three mutations was deficient in both strand transfer and allosteric activation of transpososome function but retained ATPase activity. Peptide binding studies also revealed that this patch of basic residues within the C-terminal domain of Mu B is within a region of the protein that interacts directly with Mu A. Thus, we conclude that this protein segment contributes to both DNA binding and protein-protein contacts with the Mu transposase.

Bacteriophage Mu is a temperate phage that undergoes two forms of transposition during its life cycle. The first form of transposition occurs when Mu DNA integrates into its host genome through a nonreplicative transposition event called conservative integration (1)(2)(3). The second form occurs after integration, when the Mu DNA amplifies itself through a process known as replicative transposition (4). The conservative integration reaction has never been reproduced in vitro, but the replicative transposition pathway has been characterized extensively using purified substrate and both host-encoded and phage-encoded proteins (5).
Mu replicative transposition can be performed in vitro using a supercoiled mini-Mu substrate, which contains the left end binding region (attL), the right end binding region (attR), and the enhancer element within a donor DNA molecule. The reaction occurs through a series of higher order nucleoprotein complexes called transpososomes. Transpososome formation up to and including strand transfer occurs in the presence of two phage-encoded proteins, Mu A and Mu B, and two host encoded proteins, HU and IHF (Fig. 1). In the presence of the mini-Mu substrate, Mu A, HU, IHF, and an appropriate divalent metal ion, the initial LER complex is formed. The LER complex is a transitory three-site complex that through numerous proteinprotein and protein-DNA interactions brings together the left end, the right end, and the enhancer elements (6). The first stable complex formed is the Type 0 (stable synaptic complex) (7,8). The Type 0 complex is characterized by the engagement of the Mu ends by the active site and accumulates in the presence of calcium, which does not support strand cleavage. In the presence of magnesium, the 3Ј-ends of the Mu DNA are nicked, and the Type 0 complex is quickly converted to Type 1 complex (cleaved donor complex) (9,10). The addition of Mu B, ATP, and target DNA results in the formation of the Type 2 complex (strand transfer complex) in which the 3Ј-ends of the Mu DNA are transferred to a target DNA molecule (9,10). The Type 2 complex is the most stable transpososome and must be destabilized by the host-encoded ClpX ATPase before replication can occur (11)(12)(13)(14). Target capture can also occur at the LER or Type 0 stage of the reaction (15). In the absence of target DNA, Mu B and ATP stimulate intramolecular strand transfer, whereby the 3Ј-ends of the Mu DNA are transferred into a new DNA site within the donor molecule (16,17).
The Mu A protein (transposase) is 663 amino acids in length (18) and is able to perform all of the chemical steps in the transposition reaction. Mu A can be divided into three distinct globular domains (19). Domain I is responsible for DNA binding of the enhancer element and the Mu ends (20,21). Domain II can be divided into two functionally distinct and complementary subdomains: domain II␣, which contains the conserved DDE motif believed to coordinate the divalent metal ion (22,23), and domain II␤, which is involved with transpososome assembly and may bind DNA (24,25). Domain III also contains two subdomains, domain III␣ and domain III␤. Domain III␣ has functional similarities to domain II␤, and they may both constitute a single functional domain (24 -26). Domain III␤ is a protein-protein interaction region that contains sequences bound by Mu B and ClpX (12,17,27,28).
The 312-residue phage-encoded protein Mu B is an ATP-dependent DNA-binding protein required for efficient Mu DNA transposition (16,29). Mu B is able to capture target DNA and interact with all of the transpososome complexes (15). Mu B is also an allosteric activator of the transposase and can stimulate transpososome formation on both mutant donor DNA substrates and with partially functional mutant forms of Mu A (17, 26, 30 -32). Along with target capture and transpososome formation, Mu B is also responsible for preventing Mu from transposing into itself, a process known as target immunity. During target immunity, Mu A stimulates the release of Mu B bound to DNA through the hydrolysis of ATP (33)(34)(35).
The Mu B protein has two globular domains (36). The 25-kDa N-terminal domain contains nonspecific DNA binding activity and has an ATPase motif (29,36). The 11-kDa C-terminal domain is also able to bind nonspecifically to DNA (40). Sitedirected mutagenesis of the N-terminal bipartite nucleotide binding motif (Walker A and Walker B boxes) results in the loss of ATPase activity (37). However, the isolated N-terminal domain by itself is unable to hydrolyze ATP (36). Mutations in the N-terminal domain of Mu B have been shown to affect target capture, but these mutants retain their ability to interact with the transpososome complex (37,38). A small truncation of the C-terminal domain of Mu B blocks replicative transposition but not integration, whereas longer truncations affect both replicative and integrative transposition in vivo (39).
The C-terminal domain of Mu B (Mu B 223-312 )is a four-helix bundle (Fig. 2) and has a similar fold to that of the N-terminal region of DnaB (see Ref. 40). Examination of the charged residues within the peptide revealed that there is a positively charged patch of amino acids on the surface of the structure. It was proposed that this patch may be involved in the DNA binding activity that has been associated with the C-terminal domain. In this study, we first confirmed that these residues are responsible for binding DNA, and then we investigated how these positively charged residues contribute to the in vitro The ribbon diagram, generated with the Swiss Pdb Viewer, represents the average of the 20 defined solution structures previously determined from NMR data with the initial 8 residues removed (40). The four helixes have been colored in green, blue, yellow, and purple and labeled as helix 1, 2, 3, and 4 respectively. Also shown in red are the three lysine side chains that that were changed to alanine residues.
FIG. 1. The in vitro Mu DNA strand transfer reaction. The earliest characterized complex to date is the LER, or the three-sited synaptic complex (6), in which the left (L) and right (R) ends are synapsed together with the enhancer (E), aided by the architectural proteins HU and IHF. The LER is rapidly converted to the Type 0 or stable synaptic complex (6) with weakened enhancer interactions (52) where the two Mu ends are engaged within the active site of the transposase. The Type 0 precedes the Type 1, or the cleaved donor complex, in which a nick has been generated at the 3Ј-ends of Mu, exposing 3Ј-OH groups and relaxing the vector domain (9,10). Strand transfer of the 3Ј-ends of the Mu DNA into a target molecule can lead to two types of products. In the presence of Mu B, ATP, and target DNA, intermolecular strand transfer occurs, forming a very stable Type 2, or strand transfer complex. Disruption of the Type 2 complex liberates -shaped molecules (not shown). In the absence of Mu B, intramolecular strand transfer results from integration into sites on the same mini-Mu donor plasmid. As shown, these insertions can be found in either the Mu or vector domains and can occur with the Mu DNA sequences in either of two orientations (not shown). Disruption of the intramolecular strand transfer complex results in the formation of either or dumbbell-like molecules depending on the orientation of the insertion. transposition reaction. We found that by changing three lysine residues in the C-terminal domain of Mu B to alanine residues, we generated a mutant protein that could no longer capture target or interact with the Mu transpososome. Peptide binding experiments further reveal that this region of Mu B can participate directly in Mu A-Mu B protein-protein contacts.

EXPERIMENTAL PROCEDURES
DNA, Reagents, and Enzymes-The 6.5-kb mini-Mu plasmid pBL08 has been previously described (6). The 7.2-kb mutant mini-Mu plasmid pMS9A1 has an A to G transition at the Mu left end terminal base pair (32). The target DNA used was the 5.2-kb plasmid pSD7 (32). All plasmids were purified using a cesium chloride gradient and dialyzed extensively.
Mutagenesis and Protein Purification-The HU (41) and IHF (42) proteins were purified as previously described. Mu A proteins were purified as described previously (43) through the P-11 step. No dithiothreitol was added to the wash step or the elution step of the P-11 column, and the protein was induced in the presence of 1% (w/v) glucose, which improved the induction and purity after the P-11 column.
To introduce mutations into Mu B 223-312 and the full-length Mu B genes, site-directed mutagenesis was performed using the QuikChange site-directed mutagenesis kit purchased from Stratagene, with a modified protocol. Two complementary oligonucleotides containing the desired mutation sequence were purchased (Sigma). Two separate 25-l PCRs were set up (as per standard QuikChange protocol) containing 250 pmol of each mutant primer in separate reactions with 50 ng of wild-type Mu B expression plasmid DNA. Each separate PCR was amplified for three cycles of 95°C for 50 s, 55°C for 60 s, and 68°C for 12 min with 0.5 units of Pfu Turbo DNA polymerase. The two amplification reactions were then combined to create a 50-l PCR containing both primers and their amplification products from the first three cycles. This new reaction was supplemented with an additional 0.5 units of Pfu Turbo DNA polymerase, and the reaction was run for 18 cycles of 95°C for 50 s, 55°C for 50 s, and 68°C for 12 min, followed by an additional extension for 7 min at 68°C. The amplification product was digested with DpnI for 90 min at 37°C and then transformed into electrocompetent DH5␣ cells. The mutant plasmids were then purified and sequenced.
Mutagenesis and purification of the Mu B 223-312 gene was performed using the Mu B 223-312 gene expression construct, pHH05 from the strain GC1876, described previously (40). One liter of cells was grown in LB in a 4-liter Fernbach flask in the presence of 50 g/ml ampicillin and 1% (w/v) glucose at 37°C, with a rotation of 225 rpm, to an A 595 of 0.65. The flask was then transferred to 18°C, with a rotation of 225 rpm and incubated for 45 min, and the cells were induced with 1 mM isopropyl-1-thio-␤-D-galactopyranoside for 18 h. The cells were harvested by spinning at 8500 ϫ g for 10 min, washed, and resuspended in 10 ml of 25 mM Tris-HCl, pH 7.5, plus 5 mM EDTA. Cells were incubated for 30 min at 4°C with lysozyme (0.3 mg/ml), followed by the addition of 1% Triton X-100 and five 30-s bursts using a Fisher model 300 sonifier at 50-watt power. The lysed cells were then pelleted at 27,200 ϫ g for 30 min. The supernatant was batch-absorbed at 4°C for 45 min to 500 l of glutathione-Sepharose 4B matrix (Amersham Biosciences) equilibrated with 25 mM Tris-HCl, pH 7.5, plus 5 mM EDTA. The matrix was then washed three times with 25 ml of 25 mM Tris-HCl, pH 7.5, plus 5 mM EDTA and three times with 25 mM Tris-HCl, pH 7.5, plus 5 mM EDTA and 1 M NaCl. The matrix was transferred into a 1 ϫ 2.5-cm column, washed three times more with 25 mM Tris-HCl, pH 7.5, plus 5 mM EDTA, and eluted with 50 mM Tris-HCl, pH 7.5, plus 5 mM EDTA and 20 mM glutathione. The eluate was digested with 1 unit of thrombin/mg of protein for 45 min at room temperature. The digested protein was then filtered (0.22-m filter) and loaded (1 ml/min) onto a 1-ml Resource prepacked Mono 15S column (Amersham Biosciences), equilibrated with 25 mM Tris-HCl, pH 7.5, using a high pressure liquid chromatograph (Waters). The Mu B 223-312 peptide was eluted from the Mono 15S column using a 0 -700 mM NaCl (Tris-HCl, pH 7.5) gradient for 40 min at a flow rate of 1 ml/min. Peak fractions were collected and dialyzed against 20 mM phosphate, pH 6.8, plus 1.5 M NaCl.
Mutagenesis and purification was performed on the full-length construct using the Mu B gene expression plasmid pML02 (38). Mu B was purified as described previously (38), with the following changes: the ethanol pellet was resuspended in 25 mM Tris-HCl, pH 8.8, plus 1 mM EDTA, 10 mM MSH, and 7.0 M urea; and the DEAE-Sepharose column was equilibrated with 25 mM Tris-HCl, pH 8.8, plus 1 mM EDTA, 10 mM MSH, and 6.2 M urea and eluted with 25 mM Tris-HCl, pH 6.8, plus 1 mM EDTA, 10 mM MSH, and 6.2 M urea. The CM-Sepharose column was equilibrated with 25 mM Tris-HCl, pH 6.8, plus 1 mM EDTA, 10 mM MSH, 6.8 M urea, and 25 mM NaCl, and the protein was loaded and washed with 4 ml of equilibrium buffer before being eluted with 7 ml of 25 mM Tris-HCl, pH 7.4, plus 1 mM EDTA, 10 mM MSH, 6.8 M urea, and 100 mM NaCl. The 3K protein was purified as above using SP-Sepharose instead of CM-Sepharose. All of the proteins were greater than 95% pure with the exception of the 3K protein, which was greater than 90% pure as judged by SDS-PAGE.
DNA Binding Assays-The Mu B 223-312 DNA binding assays were performed using various concentrations of peptide (described in the legends of Figs. 3 and 5) in the presence of 25 mM Hepes-NaOH, pH 7.6, 150 mM NaCl, and 15 g/ml DNA (pSD7). The reactions were incubated at 25°C for 10 min.
Affinity co-electrophoresis was performed essentially as described previously (44). Protein was embedded into 1% LMP agarose and 32 Pend-labeled 70-bp double-stranded DNA was run through the gel. The final protein concentration in the gel varied between 22.5 nM and 1.44 M as described throughout. The gels and electrophoresis buffer contained 25 mM Tris-HCl, pH 8.0, 0.1 mg/ml bovine serum albumin, 10 mM magnesium acetate, 50 mM potassium acetate, and 0.5 mM ATP. Gels were run at 4°C with circulating buffer at 4.2 V cm Ϫ1 for 5 h. The gels were dried and scanned using a PhosphorImager S and quantified using ImageQuant software from Amersham Biosciences. The fraction of DNA bound was taken as the percentage of radioactivity in the top half of each lane.
In Vitro Reactions-The Type 2 reactions were performed under standard reaction conditions (1ϫ): 15 g/ml supercoiled mini-Mu plasmid, 3 g/ml (44 nM) Mu A protein, 3.5 g/ml (315 nM) HU, 0.2 g/ml (18 nM) IHF, 5 g/ml (140 nM) Mu B, 2 mM ATP, 20 g/ml target DNA, 25 mM Hepes-NaOH, pH 7.6, plus 140 mM NaCl and 10 mM MgCl 2 . Reactions were incubated at 30°C for various amounts of time as described in the legends of Figs. 4 and 6. The intramolecular strand transfer reactions were performed at 2.5ϫ reaction conditions in the absence of target DNA at Mu B concentrations of 140 and 280 nM as indicated.
ATPase Assay-The ATPase assays were performed essentially as previously described (16). Reactions contained 35 g/ml (980 nM) Mu B in 150 mM NaCl, 10 mM MgCl 2 , 7.5% glycerol, 3 mM dithiothreitol, 0.3 mg/ml bovine serum albumin, 0.1 mM [␥-32 P]ATP (20 nCi), and 25 mM Tris, pH 7.8, and, where indicated, 40 g/ml (586 nM) Mu A and 50 g/ml pSD7. The reactions were incubated at 25°C for 60 min and stopped by the addition of 1 ml of activated charcoal (16% (v/v) Norit A in H 2 O). The reactions were incubated on ice for 10 min in a 1-ml Eppendorf tube and centrifuged for 10 min at V max in a table top centrifuge. Following centrifugation, the supernatant was filtered through a 0.22-m filter and was analyzed.
Electrophoresis and DNA Quantification-Electrophoresis for the in vitro reactions was in TAE buffer in 1% agarose horizontal slab gels (200 mm in length) run as described in the Fig. 4 legend. Loading dye consisted of 10 mM Tris-HCl, pH 7.8, plus 1 mM EDTA, 5% (w/v) sucrose, and 0.08% (w/v) bromphenol blue. DNA was visualized by staining gels with ethidium bromide (0.5 g/ml) for 30 min followed by a 30-min destaining in water. Gel documentation and quantification of fluorescent intensity made use of a camera and Alphaimager Software (version 4.03) from Alpha Innotech Corp. (San Leandro, CA).
Peptide-Mu A Binding-A cellulose filter containing synthetic peptides corresponding to the entire Mu B protein sequence was prepared by the Massachusetts Institute of Technology Biopolymers facility using an Abimed instrument. Each peptide consisted of 12 residues of Mu B sequence plus two additional ␤-alanines at the C-terminal end and was covalently attached to the filter via a C-terminal polyethylene glycol linker. The filter was probed essentially as described (45), except that the buffer was as follows: 10 mM Tris-Cl, pH 7.5, 150 mM NaCl, 1 mM dithiothreitol, 10 mM MgCl 2 , 5% glycerol, 0.1% Tween. For blocking, this solution also contained 0.1% nonfat powdered milk, whereas for Mu A binding, it contained ϳ1 M Mu A or Mu A 1-615 .

DNA Binding Activity of the C-terminal Domain Mu B 223-312
Peptide-The NMR solution structure of the Mu B 223-312 peptide revealed the presence of a large number of positively charged residues on the surface of the structure (40). Three lysine residues (positions 233, 235, and 236) located near the N terminus of helix one were predicted to be part of a potential DNA binding patch. To analyze whether these residues affect nonspecific DNA binding, lysines 233, 235, and 236, along with lysine 284 (not predicted to be part of the patch) were changed to alanine residues (K233A-Bc, K235A-Bc, K236A-Bc, and K284A-Bc). A triple alanine substitution was also made, incorporating the changes at positions 233, 235, and 236 (3K-Bc). These mutant Mu B 223-312 peptides were analyzed for their ability to bind DNA using a gel mobility shift assay (Fig. 3B). The K233A-Bc, K235A-Bc, and the K236A-Bc showed reduced DNA binding relative to wild-type peptide (Wt-Bc). The K284A-Bc peptide, which was not predicted to be part of the DNA binding pocket, showed DNA binding equivalent to wild type. The 3K-Bc peptide, however, did not show any detectable DNA binding at even the highest peptide concentration. To determine whether the 3K-Bc peptide was properly folded, far UV circular dichroism was performed on the Wt-Bc and the 3K-Bc peptides; these spectra contained similar absorbance peaks at 208 and 222 indicative of an ␣-helical protein (data not shown). The failure of the 3K-Bc peptide to bind DNA therefore did not appear to be the result of an overall change in the secondary structure of the peptide.
Type 2 Formation with the Full-length C-terminal Mu B Mutants-To further analyze the role of the DNA binding activity of the C-terminal domain of Mu B, identical mutations from the Mu B 223-312 peptides were made in the full-length protein. The C-terminal mutant proteins were analyzed for their ability to promote Type 2 complex formation (Fig. 4). The mutant proteins K233A, K235A, and K236A all displayed robust levels of Type 2 complex formation but at somewhat different efficiencies; K235A and K236A showed somewhat decreased efficiency of Type 1 to Type 2 conversion. The 3K mutant, however, did not form any Type 2 complex under standard reaction conditions. Interestingly, the single mutant proteins also formed varying amounts of intramolecular strand transfer products. In contrast, the 3K mutant did not stimulate any intramolecular strand transfer, suggesting that this pro-tein may be profoundly defective in interacting with the transposase tetramer.
DNA Binding Activity of the Full-length C-terminal Mu B Mutants-To analyze the effect that the C-terminal mutations had on DNA binding of the full-length Mu B protein, affinity co-electrophoresis assays were performed (see "Experimental Procedures"). Affinity co-electrophoresis assays involve embedding various amounts of protein in an agarose gel and electrophoretically passing radiolabeled DNA through the embedded protein (Fig. 5). This assay is particularly useful for studying interactions of Mu B with DNA. Full-length Mu B protein has a tendency to aggregate, and its DNA binding activity is not easily measured with other assays (37). DNA binding was determined by the ability of the embedded proteins to complex with the DNA and inhibit migration through the gel. The protein concentration required for half-maximal DNA binding was calculated as the amount of protein required to retain half of the DNA on the top 50% of the gel. Half-maximal binding was calculated in the presence of ATP and was 85 nM for wild-type protein, 90 nM for K233A, 136 nM for K235A, 100 nM for K236A, and 623 nM for the 3K full-length proteins. Single residue mutations on the full-length protein, therefore, displayed only a slight decrease in DNA binding affinity compared with the wild-type protein. In contrast, the 3K mutation had over a 7-fold decrease in affinity for DNA. These results were qualitatively similar with those from the gel shift assays with the mutant peptides (Fig. 3B) and suggest that each of these mutations have a modest affect on the ability of the full-length Mu B to interact with DNA, whereas the triple mutant has a much more profound affect. investigate the ability of full-length mutant Mu B proteins to stimulate the Mu A tetramer, we performed intramolecular strand transfer reactions with both substrate and transposase mutants (Fig. 6). The intramolecular assays allow for investigation of Mu B stimulation of strand transfer by Mu A, regardless of whether or not the Mu B protein can directly bind to FIG. 4. Kinetics of Type 2 formation using purified full-length mutant Mu B proteins. Strand transfer was assayed at 30°C under standard reaction conditions. The reactions were run on a 1% agarose gel at 2.75 V cm Ϫ1 for 12 h. The supercoiled target (ST), the supercoiled donor (SD), the relaxed target (RT), and the relaxed donor (RD) are indicated to the left of the gel. The positions of the Type 2, the Type 1, and the intramolecular products are shown to the right of the gel. The faint band observed with wild type Mu B, running between the Type 2 complex and the relaxed donor DNA, is an unstable form of the Type 2 complex where the target DNA is still supercoiled. The nomenclature for the mutant proteins is described in Fig. 3, except that all of the mutations are present in the full-length protein.

FIG. 5. Assay of DNA binding activity of full-length mutant Mu B proteins.
A, the DNA binding affinities for the wild-type and mutant full-length Mu B proteins were determined using affinity co-electrophoresis assays, essentially as described previously (44). Briefly, various concentrations of wild-type and mutant proteins were embedded into the lanes of a horizontal slab gel. Then 32 P-labeled 70-bp double-stranded DNA oligonucleotide was loaded into a single well spanning the width of the gel and electrophoretically run through the gel in the presence of 0.5 mM ATP. As protein-DNA complexes form during electrophoresis, migration of the radiolabeled DNA is retarded in the gel. B, quantification of affinity co-electrophoresis assay gels for wild-type, K233A, K235A, K236A, and 3K full-length Mu B proteins. DNA binding was determined as the percentage of the DNA retained at the top half of the gel (see "Experimental Procedures"). Data were fitted to the following equations, as described previously (37,44) target DNA. The mutant substrate pMS9A1 (pA1GL) contains a single nucleotide change from A to G at the left end terminal base pair that supports a 20% level of DNA cleavage under normal reaction conditions (31,46). Wild-type Mu B stimulated the cleavage reaction on this substrate over 3-fold. K233A stimulated DNA cleavage as well or better than wild-type Mu B. K235A and K236A stimulated the reaction of pMS9A1 over 2-fold. The triple mutant 3K, however, did not stimulate the reaction to any significant extent.
Mu B can also stimulate catalytically defective transposase mutants (25,26,30). The Mu A ban protein is a quadruple mutation in domain III␣ that performs DNA cleavage at less than 20% the efficiency observed for the wild-type Mu A under standard reaction conditions. Wild-type Mu B, K233A, K235A, and K236A all stimulated the reaction with this mutant transposase over 2-fold. The 3K mutant, however, did not significantly stimulate this reaction. The inability of the 3K mutant to allosterically activate the Mu donor cleavage reaction with a mutant substrate or a mutant transposase suggests that the 3K protein is compromised in its ability to functionally interact with the Mu A tetramer.
ATPase Activity of the Mu B Mutants-The N-terminal domain of Mu B has been shown to be responsible for the ATP hydrolysis activity of the full-length protein (37). To determine whether the C-terminal domain mutants caused any major conformational changes to the structure of Mu B, the mutant proteins were tested for their ability to hydrolyze ATP (Fig. 7). All of the full-length mutant Mu B proteins hydrolyzed ATP at a rate that was equal to or greater than that observed with the wild-type protein, suggesting no major conformational defects in the overall structure of the proteins. The reason for the apparent increase in ATPase activity displayed by the mutant proteins is not entirely clear, but it may result from variability in the amount of active Mu B present in each protein preparation. The final step of the Mu B purification procedure requires that the protein be refolded from its denatured state (see "Experimental Procedures"). It has been noted that up to 50% of the Mu B protein is inactive after purification (35). Since the ATPase activity of Mu B is highly cooperative, small changes in the amount of active protein could lead to large differences in ATPase activity The ATPase activity of the mutant Mu B proteins was stimulated by the addition of either Mu A or DNA (data not shown) but most significantly by the addition of both of these factors (Fig. 7). These results suggested that Mu A and DNA could still interact with Mu B to stimulate ATP hydrolysis. These data therefore suggest either that 1) the interactions between the mutant Mu B protein and Mu A and DNA are stabilized under ATPase assay conditions or 2) that the Mu A and DNA interactions required for stimulation of the Mu B ATPase activity are different or less demanding than the interactions needed to stimulate the transposase tetramer, since no allosteric activation of Mu A was observed with the 3K mutant (Figs. 4 and 5).
The ATPase activity of Mu B is also stimulated by the oligomerization of the protein and Mu B has been shown to oligomerize in the presence of ATP or ATP␥S 1 (35). To confirm the ability of the Mu B mutants to oligomerize, their ATPase activity was shown to be concentration-dependent (data not shown). To further demonstrate protein oligomerization, we incubated the Mu B variants with a chemical cross-linking reagent in the absence and presence of either ATP or ATP␥S. We found that the mutants formed oligomers in the presence of ATP or ATP␥S (data not shown). These results confirm that the mutant Mu B proteins retained the ability to interact with themselves and further suggest that the mutational changes did not cause any major defects in the overall structure of the proteins.
The Important Lysines Lie within a Region of Mu B That Interacts with Mu A-The results presented above indicated 1 The abbreviation used is: ATP␥S, adenosine 5Ј-(3-thiotriphosphate). that the positively charged region in the C-terminal domain of Mu B is involved in Mu B-DNA interactions but also suggest that this region may be involved in protein-protein interactions between Mu A and Mu B. To directly determine regions of the Mu B protein that have the capacity to bind to Mu A, we constructed a library of 12 amino acid peptides representing the entire Mu B protein sequence immobilized on a solid support. This peptide filter was then incubated with Mu A, and peptide-Mu A complexes where detected using anti-Mu A antiserum (Fig. 8). Although several regions of the Mu B sequence clearly bound Mu A in this experiment, the strongest interaction was observed with peptides initiating at positions Glu 217 -Lys 235 . This region corresponds well to the N-terminal portion of the C-terminal domain, the region containing the positively charged patch. Interestingly, the peptide initiating with Lys 236 was much poorer at Mu A binding than its neighbor, which carried both Lys 235 and Lys 236 , indicating either that the presence of both lysines, or Lys 236 in particular, is critical for Mu A binding. Control experiments in which Mu A 1-615 was used in place of wild-type Mu A revealed much less interaction with this family of peptides (data not shown), indicating that the observed binding was mediated by the C-terminal domain of Mu A. DISCUSSION The bacteriophage Mu B protein is an ATP-regulated DNA binding protein that makes protein-protein contacts with the Mu A transposase to stimulate transposition and control target site choice. Partial proteolysis studies have previously shown that Mu B has two domains, an N-terminal domain of 25 kDa and a C-terminal domain of 11 kDa (36). Analysis of the solution structure of the Mu B 223-312 peptide revealed a positively charged patch on the surface of the C-terminal domain. It was proposed that these residues may be responsible for DNA binding (40). We found that single residue alanine substitutions at Lys 233 , Lys 235 , and Lys 236 slightly decreased the DNA binding activity of the Mu B 223-312 peptide and that a triple mutation of these three residues completely inhibited DNA binding. Likewise, the DNA binding activity of the full-length 3K mutant protein was over 7 times weaker than that of the wild-type protein. Furthermore, functional assays of Mu B-stimulated transpososome activity and physical Mu A binding experiments revealed that this region of Mu B makes important contacts with the transposase. These results strongly suggest that this region in the C-terminal domain of Mu B plays an important role in both the protein-DNA and protein-protein contacts needed to regulate transposition.
Most of the previously characterized Mu B mutants appear to weaken the protein's affinity for DNA without strongly compromising its ability to interact with Mu A and allosterically activate the transpososome. Mutations in this class include those within or near the ATPase motif in the N-terminal domain (e.g. K106A, C99Y, and an insertion at residue 101) (37,38,46). It is not known whether these mutations are directly inhibiting DNA binding or if the reduced DNA binding activity of these mutant proteins is a result of other conformational changes related to the ATP-binding pocket. In addition, a Mu B mutant lacking residues 295-312 from the C-terminal region of the protein is also in this functional class (50). Interestingly, this protein is able to promote integrative but not replicative transposition in vivo (39,50).
In contrast, the 3K mutant described here is defective in both DNA binding and Mu A interaction. Initial analysis suggests that these defects may be a direct result of the amino acid changes, rather than an indirect effect on the folded state of the protein. In support of this conclusion, peptide binding experiments revealed that the major region of Mu B that binds to Mu A resides between residues 217 and 235, a segment that corresponds to the N-terminal region of domain II. Thus, this analysis has added a useful component to our understanding of the structure/function map of Mu B. Interestingly, the Mu B 223-312 peptide has structural similarities with the N-terminal domain of the helicase DnaB. Our findings suggest that these two domains may also have functional similarities, since the N-terminal domain of DnaB was also shown to be responsible for protein-protein interactions with the primase DnaG (48,49). It is also of interest that lysines 233, 235, and 236 are in a region of Mu B that is disordered but near the start of helix 1 in the solution structure (40). Interaction of this region with Mu A (and/or DNA) may help to stabilize an ordered structure in this region of the protein. The library of Mu B peptides are in sequential order from the N terminus to the C terminus of the protein. Each 12-residue peptide initiates with the residue that is one amino acid more C-terminal than that of the previous peptide. Thus, each peptide sequence differs from its neighbors by two amino acids. The N-terminal amino acids of the peptides in the first column are shown in one-letter code. Changes in exposure did not result in any changes in the observed pattern (data not shown).
An interesting feature of the ATPase activity of the mutant Mu B proteins is that they are stimulated by the combined presence of Mu A and DNA. These findings seem contradictory to our results showing that the mutant proteins had decreased DNA binding activity and that the 3K mutant could not interact with the transposase. This discrepancy is best explained in one of two ways. The first explanation is that Mu B interactions with DNA and Mu A are stabilized under ATPase assay conditions. Since the ATPase activity of Mu B is quite weak, ATPase assays (relative to transposition assays) are typically performed under higher Mu B concentrations for a longer incubation period and in the presence of glycerol and bovine serum albumin. These conditions are known to rescue a variety of Mu transposition defects. Another explanation may be that the Mu A-Mu B interactions that occur during the stimulation of the ATPase activity are different from the Mu A-Mu B interactions required for recruitment of target to the transpososome or for stimulation of strand transfer. Consistent with this hypothesis, we find several distinct regions of Mu B that interact specifically with Mu A, as assayed using the peptide library. It has been proposed that monomers of Mu A can stimulate ATP hydrolysis and DNA dissociation of Mu B from DNA (34,51); however, only a multimer of Mu A has been shown to form a stable target capture complex with Mu B. The stimulation of the ATPase activity by Mu A and DNA may therefore reflect differences between the Mu A-Mu B interactions required to cycle on and off of DNA during target immunity and the Mu A-Mu B interactions required during target capture by the transpososome. It is not clear from our studies whether the positive residue patch directly binds Mu A and DNA or whether mutation of these residues is causing other local conformational changes which inhibit these interactions. However, the peptide studies for both DNA binding and Mu A binding suggest that both activities may be direct.