|Home | About | Journals | Submit | Contact Us | Français|
Recombinases of the highly conserved RecA/Rad51 family play central roles in homologous recombination and DNA double-stranded break repair. RecA/Rad51 enzymes form presynaptic filaments on single-stranded DNA (ssDNA) that are allosterically activated to catalyze ATPase and DNA strand-exchange reactions. Information is conveyed between DNA- and ATP-binding sites, in part, by a highly conserved glutamine residue (Gln194 in Escherichia coli RecA) that acts as an allosteric switch. The T4 UvsX protein is a divergent RecA ortholog and contains histidine (His195) in place of glutamine at the allosteric switch position. UvsX and RecA catalyze similar strand-exchange reactions, but differ in other properties. UvsX produces both ADP and AMP as products of its ssDNA-dependent ATPase activity—a property that is unique among characterized recombinases. Details of the kinetics of ssDNA-dependent ATP hydrolysis reactions indicate that UvsX-ssDNA presynaptic filaments are asymmetric and contain two classes of ATPase active sites: one that generates ADP, and another that generates AMP. Active-site asymmetry is reduced by mutations at the His195 position, since UvsX-H195Q and UvsX-H195A mutants both exhibit stronger ssDNA-dependent ATPase activity, with lower cooperativity and markedly higher ADP/AMP product ratios, than wild-type UvsX. Reduced active-site asymmetry correlates strongly with reduced ssDNA-binding affinity and DNA strand-exchange activity in both H195Q and H195A mutants. These and other results support a model in which allosteric switch residue His195 controls the formation of an asymmetric conformation of UvsX-ssDNA filaments that is active in DNA strand exchange. The implications of our findings for UvsX recombination functions, and for RecA functions in general, are discussed.
Homologous recombination is an essential part of life and is important in heredity and evolution, as well as in the maintenance of genomic stability.1,2 Enzymes of the highly conserved RecA recombinase family catalyze DNA strand-exchange reactions that are central to the mechanisms of homologous recombination and recombinational DNA repair. Valuable information on recombination mechanisms has come through studies of the bacteriophage T4 recombination system and UvsX recombinase, the T4 RecA ortholog. UvsX plays central roles in phage recombination, recombination-dependent DNA replication, and double-stranded break repair pathways.3–5
Although relatively divergent in sequence, UvsX shares many properties in common with other well-characterized recombinases such as Escherichia coli RecA (EcRecA) and Saccharomyces cerevisiae Rad51 (ScRad51) proteins, including the ability to catalyze DNA strand exchanges between homologous single-stranded DNA (ssDNA) and double-stranded DNA (dsDNA) molecules, and the ability to catalyze ssDNA-dependent ATP hydrolysis.2,3 These activities require the formation of a presynaptic filament consisting of recombinase bound cooperatively to ssDNA. Despite these overall similarities, UvsX exhibits interesting differences in activity compared to other recombinases. Most notably, UvsX generates both ADP and AMP as products of its nucleotidase activity—a property that is unique among characterized recombinases.6 UvsX also exhibits distinctive sequence characteristics within the highly conserved RecA catalytic core domain, including the substitution of histidine for glutamine at a key allosteric switch position.7
The original crystal structure of EcRecA8 inferred the importance of the side chain of residue Gln194 in detecting the γ-phosphate of ATP bound to the P-loop region of the enzyme active site. Subsequent structures of EcRecA and orthologs from Mycobacterium smegmatis, Mycobacterium tuberculosis, Deinococcus radiodurans, Methanococcus voltae, and Thermococcus kodakaraensis confirm that this highly conserved glutamine residue plays an important role in sensing the γ-phosphate of bound nucleoside triphosphate.9–16 It is also clear from these and other studies that Gln194 is positioned at the base of the DNA-binding loop L2, which is the primary site involved in protein–ssDNA interactions.17–19 In the structure of an EcRecA derivative bound to ssDNA, the side chain of Gln194 forms part of a hydrogen-bonding network linking γ-phosphate to L2 and ssDNA.15 Thus, Gln194 is well-positioned to coordinate allosteric communications between nucleotide- and polynucleotide-binding sites. Data indicate that EcRecA is intolerant of all substitutions at the Gln194 position, since all mutants are defective in recombination and LexA repressor cleavage in vivo.17 Biochemical analyses of EcRecA-Q194N, EcRecA-Q194E, and EcRecA-Q194A mutants indicate that all retain basal ssDNA- and ATP-binding activities; however, ssDNA fails to stimulate the basal ATPase activity of EcRecA, and the enzyme fails to enter a high-affinity ssDNA-binding state in response to ATP binding.17–19 Available data indicate that Gln194 controls allosteric switching between the ATP-induced high-affinity ssDNA-binding state associated with homologous pairing activity and an ADP-induced low-affinity ssDNA-binding state associated with presynaptic filament dissociation.2,17–20
Sequence alignments and comparisons to known RecA structures indicate that, in UvsX protein, His195 occupies the same structural niche as Gln194 in EcRecA, so that this histidine residue in UvsX potentially represents a conservatively substituted allosteric switch residue.7 To study the importance of this residue in controlling the catalytic properties of UvsX protein, we carried out studies of the DNA strand-exchange and ATPase activities of wild-type UvsX in comparison with conservative H195Q and nonconservative H195A mutants. Both mutants exhibit defects in DNA strand-exchange reactions consistent with decreased ssDNA-binding affinity. Nevertheless, UvsX tolerates conservative substitution at the His195 locus, since the H195Q mutant retains weak DNA strand-exchange activity that is stimulated by the phage ssDNA-binding protein Gp32. Kinetic properties of H195Q and H195A mutants exhibit unexpected and striking differences from wild-type UvsX, including higher ssDNA-dependent ATPase activity, increased ADP/AMP product ratio, and lower cooperativity for substrates. The data indicate that His195 plays an important role in establishing asymmetry within the presynaptic filament between active sites that generate ADP as product and those that generate AMP. Mutations at this locus reduce active-site asymmetry, which correlates strongly with a reduced affinity for ssDNA and with reduced DNA strand-exchange activity. Our studies reveal new details of the allosteric mechanism of UvsX recombinase and have implications for the functioning of other RecA/Rad51 orthologs as well.
Figure 1 displays amino acid sequence alignments of UvsX enzymes from T-even bacteriophages, in comparison with other RecA family enzymes from diverse bacterial, eukaryotic, and archaeal organisms. The sequences shown span a segment of the strongly conserved catalytic core region of the RecA family surrounding DNA-binding loops L1 and L2. The T4 UvsX His195 residue corresponds to the allosteric switch residue Gln194 in EcRecA.7,17 This residue is widely conserved as glutamine throughout the bacterial, eukaryotic, and archaeal kingdoms, but is replaced by histidine in the UvsX subfamily (Fig. 1). The UvsX sequences shown in Fig. 1 are from representative T-even bacteriophages of diverse host ranges, including cyanophages. Interestingly, histidine also replaces glutamine at the corresponding position in the Dmc1 recombinase of the parasitic protist Theileria parva. Note that, in all sequences, the Gln/His switch residue immediately precedes the sequence-divergent DNA-binding loop L2, which contains the primary DNA-binding site.17–19
UvsX residue His195 was mutated to glutamine as a potentially conservative change mimicking the active-site configuration of cellular RecA orthologs. A histidine-to-alanine mutation was also introduced at this locus as a potential functional knockout. Figure 2 compares the DNA strand-exchange activities of wild-type UvsX to those of the H195Q and H195A mutants. Previous studies showed that, at relatively high concentrations, wild-type UvsX protein catalyzes DNA strand exchange between homologous ssDNA and dsDNA molecules in the absence of other protein factors such as Gp32 (T4 ssDNA-binding protein) and UvsY (T4 recombination mediator protein).21–23 We tested the intrinsic DNA strand-exchange activities of UvsX wild-type, H195A, and H195Q enzymes under similar conditions, with the results shown in Fig. 2a. At a concentration of 6 μM (a 2.4-fold excess over ssDNA-binding sites, assuming that n=4 nucleotide residues/protomer24), wild-type UvsX readily promotes DNA strand exchange in the absence of other protein factors, as seen by the time-dependent decrease in labeled linear dsDNA substrate and the corresponding appearance of labeled joint molecules and aggregates (Fig. 2a). These aggregates are a characteristic product of UvsX-catalyzed strand exchange, which generates branched networks of DNA caused by multiple synapsis events per DNA molecule.21,25,26 Otherwise identical reactions containing UvsX-H195A showed no DNA strand-exchange activity, whereas reactions containing UvsX-H195Q showed weak DNA strand-exchange activity, as evidenced by the appearance of DNA aggregates at late time points (Fig. 2a).
At lower enzyme concentrations, the DNA strand-exchange activity of wild-type UvsX is stimulated strongly by Gp32, the T4 ssDNA-binding protein.21–23 Figure 2b shows strand-exchange reactions performed with either wild-type, H195A, or H195Q UvsX at a concentration of 3 μM (a 1.2-fold excess over ssDNA-binding sites), all in the presence of 1.43 μM Gp32 (exactly saturating with respect to ssDNA-binding sites, assuming that n=7 nucleotide residues/protomer27,28). Under these conditions, the reaction with wild-type UvsX was very robust, with nearly all replicative fragment III (RFIII; linearized M13mp18 dsDNA) substrates converted into high-molecular-weight branched DNA products by the 2-min time point (Fig. 2b). UvsX-H195A again showed no activity, demonstrating that Gp32 apparently cannot rescue the strand-exchange defect of this mutant. In contrast, UvsX-H195Q exhibited a moderate level of strand-exchange activity in the presence of Gp32, with intermediates appearing as early as the 2-min time point (Fig. 2b).
The basal ATP-independent ssDNA-binding activities of wild-type and mutant UvsX proteins were compared at equilibrium using a fluorescent Oli-Green dye-displacement assay (Fig. 3). Wild-type UvsX binds tightly to ssDNA at this salt concentration (50 mM NaCl), yielding a binding curve with a sharp end point corresponding to a binding site size of four nucleotide residues per UvsX monomer (Fig. 3), consistent with previous results.24 The basal ssDNA-binding activities of H195Q and H195A mutants are similar to each other, but much weaker than that of wild type, since large excesses of either mutant protein over ssDNA are required to approach complete displacement of the OliGreen dye from ssDNA (Fig. 3). The weaker basal ssDNA-binding activities of these mutants may be at least partially responsible for their defects in DNA strand exchange.
The kinetics of the ssDNA-dependent ATPase activity of wild-type UvsX protein were investigated previously.6 Here, we present the steady-state kinetic parameters of ATP hydrolysis by the UvsX-H195Q and UvsX-H195A mutants, in direct comparison with those of wild-type UvsX. Figure 4 shows typical time courses of ATP hydrolysis with nucleotide substrates/products resolved by thin-layer chromatography (TLC). The figure clearly illustrates the production of both ADP and AMP products by wild-type UvsX. In comparison, UvsX-H195Q and UvsX-H195A mutants produce ADP at a rate exceeding that of wild type under the conditions tested; however, their AMP production is strongly suppressed compared to that of wild type (Fig. 4).
To quantify differences in the ATPase properties of UvsX wild type and mutants, steady-state kinetic experiments were performed with ATP as variable substrate and at constant saturating ssDNA concentrations. Results are shown in Fig. 5 and Table 1. Reaction velocities were measured independently for ADP and AMP production using the TLC assay described under Materials and Methods. Velocity curves are sigmoidal in nature (Fig. 5); therefore, data were analyzed by fitting either to the Hill equation or to the logistic sigmoid function.29–32 Under apparent maximum velocity conditions (saturating ATP and ssDNA), wild-type UvsX exhibited a kcat of 8.6 s−1 for ADP production versus a kcat of 4.6 s−1 for AMP production (Table 1), for an ADP/AMP product ratio of ~1.9. These parameters are similar to those previously published.6 In both reactions (ADP and AMP generation), wild-type UvsX exhibits strong positive homotropic cooperativity for the ATP substrate (Fig. 5 and Table 1). The apparent Km values of wild-type UvsX for ATP are similar to previously published values.6 With wild-type UvsX, the apparent Km for ATP is ~4-fold lower, and kcat/Km is ~8-fold higher, when measured for the ADP-producing reaction versus the AMP-producing reaction (Table 1).
The two UvsX missense mutants H195A and H195Q are remarkably similar to each other in terms of their ATPase kinetics, but exhibit surprising differences from wild-type UvsX. Under apparent maximum velocity conditions (saturating ATP and ssDNA), both mutants exhibit ~4-fold increases in kcat value for ADP production compared to wild type (Fig. 5 and Table 1). At the same time, both mutants show a 2- to 3-fold reduced kcat for AMP production compared to wild-type UvsX (Table 1). Therefore, the ADP/AMP production ratio of the mutants is on the order of 15–19, or approximately 10-fold greater than the ratio observed for the wild-type enzyme. The cooperativity of ATP binding is maintained in both mutants; however, apparent Km values for ATP are significantly higher for the two mutants compared to wild-type UvsX (Table 1). For both mutants, kcat/Km values for the ADP-producing reaction exceed those for the AMP-producing reaction by a factor of approximately 30 (Table 1). The data therefore indicate that UvsX-H195A and UvsX-H195Q mutants both have a markedly higher specificity than wild-type UvsX for generating ADP, instead of AMP, during ssDNA-dependent ATP hydrolysis.
Results of experiments performed with ssDNA as variable component and at constant saturating ATP concentration are shown in Fig. 6 and Table 2. As expected, kcat values are similar to those obtained in the ATP-variable experiments (Fig. 5 and Table 1). Wild-type UvsX exhibits interesting differences in the kinetics of ADP versus AMP production, as ssDNA is varied. The apparent Km for ssDNA is 8-fold higher in the AMP-generating reaction than in the ADP-generating reaction (Table 2). Also, ADP production occurs with very high cooperativity towards ssDNA (Hill coefficient h ≈ 4), whereas cooperativity towards ssDNA appears greatly reduced (h ≈ 1.3) in the AMP-generating reaction. These results suggest that the ADP- and AMP-producing ATPase reactions may differentially modulate the ssDNA-binding affinity/cooperativity of UvsX.
UvsX-H195A and UvsX-H195Q mutants both exhibit reduced cooperativity for ssDNA compared to wild-type UvsX (Fig. 6 and Table 2). In fact, with both mutants, the Hill coefficient is reduced to approximately 1 in the AMP-generating reaction, implying little or no cooperativity for ssDNA in this reaction component. UvsX-H195A and UvsX-H195Q exhibit significant differences in ssDNA-binding affinity compared to wild type and to each other (Table 2). H195Q’s apparent Km for ssDNA is 6-fold higher than that of wild type in the ADP-generating reaction, but its Km in the AMP-generating reaction actually appears lower than that of wild type. In contrast, H195A’s apparent Km for ssDNA is 7-fold higher than that of wild type in the ADP-generating reaction, and its Km in the AMP-generating reaction is 2-fold higher than that of wild type. Put in perspective, it takes 42.3 μM ssDNA to yield a half-maximal velocity of AMP production by H195A, versus only 2.7 μM ssDNA to yield a half-maximal velocity of ADP production by wild-type UvsX (Table 2).
The data in Table 2 indicate that UvsX-H195A is, overall, the weakest binder to ssDNA under steady-state conditions, whereas H195Q appears to retain intermediate ssDNA-binding affinity compared to H195A and wild type. Thus, steady-state ssDNA-binding affinity correlates closely with DNA strand-exchange activity. Based on studies at variable ssDNA concentrations, UvsX wild type, H195Q, and H195A exhibit a small range of kcat/Km values for ADP production, but a larger range of kcat/Km values for AMP production (Table 2). As a result, the kcat/Km ratios for the ADP-producing reactions versus the AMP-producing reactions increase from 15 (wild type) to 24 (H195Q) to 71 (H195A), indicating an increasing specificity for ADP over AMP production with loss of hydrogen-bonding capability at the allosteric switch position.
UvsX could produce AMP by directly hydrolyzing ATP to AMP plus pyrophosphate, or it could carry out a stepwise reaction, generating ADP plus phosphate in step 1, followed by AMP plus phosphate in step 2. To begin resolving between these models, we examined the abilities of UvsX wild type and mutants to carry out an ssDNA-dependent ADPase reaction. The reactions were slow and so were carried out over a 4-day time period to ensure that linearity was established. The TLC time courses in Fig. 7 show that significant hydrolysis of an exogenous ADP substrate to an AMP product occurs with wild-type UvsX; however, the kcat of the reaction is only 6.3×10−5 s−1 (Table 3), which is 70,000-fold slower than the rate of AMP generation observed with ATP as starting substrate (Table 1). The apparent Km of wild-type UvsX for exogenous ADP is ~14.5 mM (Table 3), which is 16-fold higher than the apparent Km for ATP in the AMP-generating reaction (Table 1). These results indicate that the AMP produced in reactions starting with ATP as substrate does not arise through the rebinding of free ADP product to the enzyme, since free ADP is such a poor substrate. The apparent Km values of wild-type UvsX for ssDNA are similar between the ADPase reaction and the AMP-generating component of the ATPase reaction (Tables 2 and and33).
Figure 7 also shows that the H195A and H195Q mutations effectively knock out UvsX’s ADPase activity. The kcat values of both are estimated to be approximately 4×10−8 s−1 (Table 3), which is barely distinguishable from background. This rate is 7–8 orders of magnitude slower than the rates of AMP production observed with UvsX-H195A and UvsX-H195Q in reactions starting with ATP as substrate (Table 1).
Rates of inorganic pyrophosphate (PPi) production by UvsX wild-type, H195A, and H195Q were measured during ssDNA-dependent ATPase reactions using the spectrophotometric coupled assay system described in Materials and Methods. Briefly, this assay measures PPi production by coupling its consumption by PPi:fructose-6-phosphate phosphotransferase to NADH oxidation catalyzed by glycer-aldehyde-3-phosphate dehydrogenase, via linking reactions catalyzed by aldolase and triosephosphate isomerase. All three forms of UvsX exhibited slow rates of PPi production (kcat on the order of 0.03–0.05 s−1; Table 3) compared to their rates of AMP production as measured in TLC assays (Tables 1 and and2).2). This amounted to an AMP/PPi production ratio of ~150 for wild type and ~30 for both mutant forms of UvsX. Control experiments in which different amounts of exogenous PPi were spiked into reaction mixtures confirmed that the coupled assay system had sufficient range to measure PPi production rates equivalent to the fastest rates of ATP consumption observed in our experiments, plus sufficient sensitivity to measure much lower rates (data not shown). Our failure to detect significant PPi production by UvsX conflicts with the stated findings of a previous study.6 Our data suggest that AMP production via direct hydrolysis of ATP to AMP+PPi is not an important component of the UvsX ATPase mechanism. Instead, the data in Table 3 argue that both wild-type and His195 missense mutant forms of UvsX prefer a processive stepwise pathway of AMP production in which: (1) ATP is first hydrolyzed to ADP+inorganic phosphate (Pi), and (2) ADP is retained in the active site and hydrolyzed to AMP+ Pi before it can be released.
Results of this study demonstrate that residue His195 plays major roles in regulating ssDNA binding, nucleotidase, and DNA strand-exchange properties of UvsX recombinase, consistent with its predicted roles in sensing the γ-phosphate of bound ATP and as an allosteric switch. The data also highlight striking differences in kinetic properties between UvsX and other RecA orthologs, which may be useful for understanding the evolution of catalytic and allosteric mechanisms in the RecA/Rad51 family.
UvsX exhibits much greater tolerance for mutations at His195 than does E. coli RecA at its equivalent position, Gln194. UvsX-H195Q represents a conservative mutation that, in principle, should make the active site more RecA-like. In fact, this mutant retains weak DNA strand-exchange activity that is stimulated further by Gp32. In addition, both UvsX-H195Q and UvsX-H195A mutants are stronger ssDNA-dependent ATPases than wild-type UvsX. Both observations stand in contrast with EcRecA, in which the reciprocal mutation Q194H and also Q194A both knock out RecA recombination/repair and LexA coprotease functions in vivo, while in vitro studies of RecA-Q194A demonstrated loss of both strand-exchange and ssDNA-dependent ATPase activities.17 Thus, UvsX appears to have a more flexible active site that can accommodate structural changes while still maintaining ssDNA-activated enzymatic activities.
Based on kinetic parameters of ssDNA-dependent ATP hydrolysis and on equilibrium binding experiments, both UvsX-H195A and UvsX-H195Q mutants exhibit lower apparent binding affinity and cooperativity for ssDNA than does wild-type UvsX. This is consistent with the notion that His195 plays a key role in inducing a high-affinity ssDNA-binding conformation of UvsX in response to ATP binding. The steady-state data indicate that UvsX-H195A has the weakest ssDNA-binding affinity, while that of H195Q is intermediate between H195A and wild type. The dramatic reduction in the apparent cooperativity of ssDNA binding seen with H195A and H195Q (Table 2) strongly implicates the His195 residue as a transmitter of allosteric effects between UvsX protomers in the presynaptic filament.
Details of the kinetics of ADP and AMP production via the ssDNA-dependent ATPase reaction indicate that UvsX-ssDNA filaments contain two different asymmetric classes of active sites—one that generates ADP, and another that generates AMP (possible models shown in Fig. 8). Recall that the ADP- and AMP-producing reactions occur simultaneously in reaction mixtures containing UvsX, MgATP, and ssDNA. The steady-state kinetic parameters of the ADP- and AMP-producing reactions can therefore be measured in the same experiment under identical conditions (Figs. 4–6). The results demonstrate that, for wild-type UvsX, the apparent Km and kcat/Km values for the ATP substrate are markedly different in ADP-producing versus AMP-producing reactions (Table 1). Even more dramatic differences are seen in the Km and kcat/Km values for the ssDNA substrate; in addition, the Hill coefficient for ssDNA is reduced from ~4 in the ADP-producing reaction to only 1.3 in the AMP-producing reaction (Table 2). Thus, the simultaneous reactions exhibit distinct kinetic properties with respect to both ATP and ssDNA binding. This argues strongly for two classes of active sites within UvsX-ssDNA presynaptic filaments. Evidence also exists for active-site asymmetry in other members of the RecA family, including ScRad51 and EcRecA: The X-ray crystallographic structure of ScRad51 filaments formed in the presence of ssDNA and nucleoside triphosphate revealed that alternating protomers are in different conformations, creating two classes of nucleotide-binding sites at alternating protomer–protomer interfaces within the filament.33 A related “dimeric protomer” model was proposed to explain the existence of two classes of NTP-binding sites within presynaptic filaments of EcRecA protein.34 Therefore, conformations containing asymmetric active sites may be a general feature of the dynamics of presynaptic filaments from many organisms, with T4 UvsX representing an extreme example in which the asymmetry is reflected not only in substrate binding but also in catalysis.
How might the two different classes of active sites be distributed within filaments? UvsX-ssDNA filaments have head, tail, and interior regions, with most association/dissociation events occurring at the head or the tail. Conceivably, UvsX could assume different conformations at the head or tail region versus the filament interior,35 forming different classes of active sites that regionally produce either ADP or AMP as products of the ATPase reaction (Fig. 8a). Another possible arrangement is the alternating-protomers configuration observed crystallographically for ScRad51.33 If this model also holds for UvsX, then conceivably it could produce ADP and AMP at alternating sites along the filament (Fig. 8b). A third possibility that cannot be strictly ruled out yet is that each UvsX protomer contains two ATPase active sites (not shown), as evidence was presented for two nucleotide-binding sites in each protomer of M. smegmatis RecA protein.14
Results of this study suggest that within the asymmetric UvsX-ssDNA filament, sites that generate AMP do so by a stepwise kinetic mechanism, wherein ATP is first hydrolyzed to ADP+Pi then ADP is retained at the active site and further hydrolyzed to AMP+Pi. Thus, while all active sites generate ADP, only one class of sites releases ADP as product, while the second class retains it and performs additional chemistry. Evidence for the stepwise mechanism includes the extremely low observed rates of hydrolysis of exogenous ADP by UvsX, as well as the very low observed ratio of PPi/AMP product formation (Table 3). Our data conflict with a previous report stating that UvsX produces AMP and PPi in approximately equal quantities (unpublished results mentioned in Formosa and Alberts6). We considered the possibility that our purified UvsX fraction contained a contaminating pyrophosphatase activity. Control reactions argued against this possibility, since identical rates of PPi consumption were observed in the presence and in the absence of UvsX when known amounts of exogenous PPi were added to the coupled assay system (data not shown). We are unaware of any other enzyme that generates AMP from ATP by a processive stepwise mechanism. Further experimentation will be necessary to confirm this mechanism for UvsX.
It is clear that His195 plays an important role in maintaining active-site asymmetry and thereby promoting the AMP-producing component of UvsX’s ATPase reaction. The H195A and H195Q mutations decrease active-site asymmetry, as indicated by a ~10-fold increase in ADP/AMP product ratio compared to wild type, based on the kcat values in Tables 1 and and2.2. Simultaneously, the kcat of ADP production increases by ~4-fold over wild type in H195Q/A mutants (Tables 1 and and2).2). The combined data suggest that both mutations increase the rate of ADP product release. We hypothesize that UvsX-H195Q/A mutants release ADP too quickly to undergo a subsequent conformational change that locks the filament into its asymmetric form. Rapid ADP release by these mutants may reset filaments to a “ground state” having predominantly one class of active sites, as shown in the model in Fig. 8c. Such filaments would primarily generate ADP as product, which is what we observe experimentally. The relationship between ADP retention/release and conformational changes within presynaptic filaments is an attractive target for further biochemical and mutagenesis studies.
Diminished AMP production by the UvsX-H195Q mutant correlates strongly with diminished DNA strand-exchange activity, suggesting that active-site asymmetry maintained by His195 is important for UvsX recombination function. (UvsX-H195A lacks any strand-exchange activity, indicating that the reaction has been completely uncoupled from ATPase activity in this mutant.) AMP production by UvsX is associated with relatively low ssDNA-binding affinity and cooperativity, as inferred from kinetic parameters in Table 2. This suggests a potential biological function for AMP production in the turnover and dynamics of UvsX-ssDNA presynaptic filaments. Previous work demonstrated that T4 presynaptic filaments undergo assembly and collapse, which are closely linked to the ATPase cycle.37 AMP production by UvsX may help to ensure that filaments eventually dissociate from DNA, clearing the way for replication enzymes that are needed for recombination-dependent replication/repair. Similarly, by lowering UvsX cooperativity, AMP production may promote the breakup of long filaments by increasing the probability of interior subunit dissociation. Recent studies have demonstrated that presynaptic filament turnover is an important component of DNA strand-exchange reactions promoted by RecA family recombinases.35–37 AMP production by UvsX may be one mechanism by which the T4 recombination system ensures efficient turnover of recombination intermediates.
Chemicals, biochemicals, and commercial enzymes were purchased from Sigma, unless specifically noted. [α-32P]ATP was purchased from Amersham Biosciences. Polyethyleneimine cellulose chromatography plates were purchased from EMD Chemicals. Restriction endonucleases were purchased from New England Biolabs. Oligonucleotide primers were purchased from Qiagen. Circular ssDNA from the bacteriophage M13mp18 was isolated by extraction from purified phage particles.38 Supercoiled M13mp18 dsDNA (RFI) was purchased from New England Biolabs. Linearized RFIII M13mp18 dsDNA was generated by digestion with XbaI, treated with calf intestinal phosphatase to dephosphorylate the 5′ ends, and radiolabeled using T4 polynucleotide kinase in accordance with the manufacturer’s instructions. The radiolabeled RFIII was further purified by Qiagen’s gel extraction protocol. All DNA concentrations were determined by absorbance at 260 nm and are reported in units of nucleotides.
Gp32 (34 kDa) and wild-type UvsX (43 kDa) proteins were purified and stored as described.24,39 All T4 protein stock solutions used in this study were judged to be greater than 98% pure, based on analyses of Coomassie-blue-stained SDS polyacrylamide gels. In addition, all protein solutions were found to be nuclease-free according to published criteria.40 All protein concentrations were determined by absorbance at 280 nm, using molar extinction coefficients of 69,760 (M cm)−1 for UvsX and 41,360 (M cm)−1 for Gp32 (unpublished results).41
Plasmid pET27b(+)-encoding UvsX was a generous gift from Dr. S. J. Benkovic (Pennsylvania State University). The H195A and H195Q mutants were created using the QuikChange site-directed mutagenesis kit (Stratagene). Primers were designed by creating a missense mutation at positions 583–585 in the UvsX gene, changing the codon sequence of CAT to GCA for UvsX-H195A, and to CAA for UvsX-H195Q. To construct UvsX-H195A, primer 1 (5′-CCATGTATTGCTATTAACGCAACATACGAAACACAAGAAATGTTTAG-3′) and reverse complement primer 2 (5′-CTAAACATTTCTTGTGTTTCGTATGTTGCGTTAATAGCAATACATGG-3′) were used. For UvsX-H195Q, primer 3 (5′-CCATGTATTGCTATTAACCAAACATACGAAACACAAGAAATGTTTAG-3′) and reverse complement primer 4 (5′-CTAAACATTTCTTGTGTTTCGTATGTTTGGTTAATAGCAATACATGG-3′) were used. Plasmids pET27b-UvsX-H195A and pET27b-UvsX-H195Q were sequenced at the Vermont Cancer Center to verify successful mutagenesis. These plasmids were transformed into E. coli BL21(DE3) cells. Growth and induction were carried out at 37 °C in LB media containing 75 μg ml−1 kanamycin. Recombinant proteins were induced by the addition of 1 mM IPTG for 4 h. Subsequent purification, storage, and quantification of UvsX-H195A and UvsX-H195Q proteins were carried out as described for wild-type UvsX.24
DNA strand-exchange reactions were carried out at 37 °C as described,6 with the following adjustments. The reaction mixtures contained 20 mM Tris-acetate (pH 7.4), 90 mM potassium acetate, 10 mM magnesium acetate, 1 mM DTT, 100 μg ml−1 BSA, 10 U ml−1 pyruvate kinase, 3.3 mM phosphoenolpyruvate, 10 μM M13mp18 ssDNA, 10 μM linear M13mp18 dsDNA (RFIII), 0 or 1 μM Gp32, and 0.5–6 μM recombinase, depending on the experiment. Reaction mixtures were incubated at 37 °C for 5 min and initiated by the addition of ATP to a final concentration of 3 mM. The reaction was allowed to continue for the time stated and stopped by the addition of strand-exchange Stop solution [final concentrations: 1× Promega loading dye, 5% SDS, and 40 mM ethylenediaminetetraacetic acid (EDTA)]. The samples (10 μl) were separated by electrophoresis at 130 V for 2 h through 1% agarose gels in the presence of a Tris-acetate–EDTA buffer system. The gels were dried onto Millipore Immobilon-Ny+ using a gel drier overnight at 45 °C and exposed to a K-screen (Kodak) for 1 day. The phosphorimage was then scanned using a Bio-Rad Personal Molecular Imager-FX and analyzed using Quantity One V4.5.1 (Bio-Rad) software.
Protein–ssDNA interactions of wild-type and mutant UvsX proteins were quantified by OliGreen dye displacement. OliGreen dye (Molecular Probes/Invitrogen) undergoes great fluorescence enhancement upon saturation with ssDNA. Protein binding to ssDNA reduces OliGreen fluorescence by displacing the dye from the polynucleotide. Fluorescence measurements were made on a PTI Quanta-master QM-6 fluorometer, with excitation and emission wavelengths set at 480 and 520 nm, respectively. Dye-displacement assays were carried out at 25 °C in buffer containing 20 mM Tris-acetate (pH 7.4), 1 mM DTT, 10 mM MgCl2, and 50 mM NaCl. The concentration of M13mp8 ssDNA was 1 μM nucleotides. The amount of OliGreen dye added was stoichiometric with respect to ssDNA. This amount was determined empirically from the break point of calibration curves in which dye was titrated into a fixed concentration of ssDNA while dye fluorescence was being measured. Experimental dye/ssDNA solutions were titrated with UvsX, UvsX-H195Q, or UvsX-H195A protein while the decrease in dye fluorescence was being monitored. Parallel control experiments included: (1) titration of identical dye/ssDNA solutions with equivalent volumes of UvsX storage buffer, and (2) titration of dye solution lacking ssDNA with protein. Control data were used to correct experimental data for small effects of dilution, photobleaching, and dye–protein interactions.
Rates of ssDNA-dependent ATP hydrolysis by wild-type and mutant UvsX proteins were determined by TLC assays. Reactions were performed at 37 °C as described,6 except for the following changes. Unless otherwise stated, the reactions contained 20 mM Tris-acetate (pH 7.4), 10 mM magnesium acetate, 90 mM potassium acetate, 1 mM DTT, 100 μg ml−1 BSA, 0.45 μM wild-type or mutant UvsX recombinase, 0–120 μM M13mp18 ssDNA, and 0–10 mM ATP, in a final reaction volume of 100 μl. In experiments with ssDNA as variable component, reactions were initiated by the addition of ATP to a final concentration of 3 mM ([α-32P] ATP final specific activity, 10 μCi ml−1). In experiments with ATP as variable component, enzyme concentrations were lowered to 0.1 μM to maximize linear reaction time points, and ssDNA concentration was fixed at 20 μM (UvsX wild-type reactions) or 100 μM (UvsX mutant reactions). At various time points, 5-μl samples were removed and quenched into 2 μl of 280 mM EDTA to give a final concentration of 80 mM EDTA. Quenched samples were spotted out at 0.5-μl volumes onto polyethyleneimine cellulose TLC plates (20 cm×20 cm) at 1-cm intervals. After all the samples had been spotted and dried, the TLC plates were developed with 0.75 M KH2PO4 and air-dried. The TLC plates were exposed for 1 h to a K-screen (Kodak) and scanned by a Bio-Rad Personal Molecular Imager-FX. Quantification of the phosphorimage was performed by Quantity One v. 4.5.1 (Bio-Rad) software and subsequently fitted using KaleidaGraph v. 3.6.2 (Synergy Software) software. Except as noted below, kinetics data were fitted to the Hill equation:
where ν is the initial velocity at any given substrate concentration S, Vmax is the maximum velocity, K0.5 is the apparent Michaelis constant (equal to the substrate concentration at which ν =0.5Vmax), and h is the Hill coefficient. For UvsX-H195A and UvsX-H195Q mutants, rates of AMP production at variable ssDNA concentrations were best described by the Michaelis–Menten equation:
where Km is the Michaelis constant. For UvsX wild type, rates of ADP production at variable ssDNA concentrations were best described by the logistic sigmoid function:
where Vmin is the minimum velocity associated with basal levels of hydrolysis, k is proportional to the Hill coefficient [k=hln(10)], x is the logarithmic cofactor concentration (ssDNA), and X50 is the negative logarithm of apparent Km. The logistic sigmoid function is mathematically equivalent to the Hill equation, but incorporates an additional minimal asymptote parameter to explain basal levels of activity.29–32 A minimum of three reactions were run at each substrate concentration, and velocities were corrected for the amount of recombinase in each reaction.
ADPase reaction rates were determined using the standard TLC conditions described above, except that ADP, instead of ATP, was used as substrate. [α-32P]ADP was generated by reacting 25 μM [α-32P]ATP with 0.75 U ml−1 hexokinase in 5 mM Hepes (pH 7.6), 30 mM MgCl2, and 300 μM D-glucose, as described previously.42 The reaction was allowed to proceed for 4 h at 30 °C and immediately stopped by boiling for 3 min. The radiolabeled ADP was determined to contain 1.66% unreacted [α-32P]ATP based on TLC separation and densitometry using QuantityOne Software (Bio-Rad). The final reactions mixtures contained 20 mM ADP ([α-32P]ADP final specific activity, 10 μCi ml−1) and were carried out for 4 days at 20 °C. Reaction rates were determined as described above for ATPase TLC assays.
Pyrophosphate production rates by UvsX wild type, UvsX-H195A, and UvsX-H195Q were determined by modification of a published coupled assay system described in Van Schaftingen et al.43 Reactions contained 0.45 μM recombinase, 4.5 μM ssDNA (for UvsX wild type) or 40 μM ssDNA (for UvsX mutants), 17 mM glucose 6-phosphate, 2.5 mM fructose 6-phosphate, 1 μM fructose-2,6-bis-phosphate, 0.46 mM NADH, 2.86 U ml−1 aldolase, 2.86 U ml−1 glyceraldehyde-3-phosphate dehydrogenase, 2.86 U ml−1 triosephosphate isomerase, and 0.043 U ml−1 pyrophosphate:fructose-6-phosphate phosphotransferase. Reactions were carried out in 20 mM Tris-acetate (pH 7.4), 10 mM magnesium acetate, and 90 mM potassium acetate buffer containing 6 U ml−1 PK and 3 mM PEP for regeneration of ATP from ADP. Reactions were carried out at 37 °C in a volume of 350 μl. Reactions were started by the addition of 5 mM ATP. Reactions were monitored using a Hitachi U-2000 spectrophotometer, recording change in A380 over time. Rates were calculated from the linear portions of time courses. All reactions were repeated a minimum of five times. Pyrophosphate-positive controls were performed as described above by adding an initial concentration of 2.5 mM pyrophosphate. Spiking of the recombinase reactions with 2.5 μM and 2.5 mM pyrophosphate during the reaction course was performed for additional positive controls.
Multiple amino acid sequence alignments were performed using Tcoffee software,44 using full-length protein sequences obtained from the National Center for Biotechnology Information (NCBI) and Protein Data Bank (PDB) protein databases. Sequences used in the alignment included the following: enterobacteria phage T4 UvsX (NCBI AAD42669), D. radiodurans RecA (NCBI BAA21330), M. tuberculosis RecA (PDB 1MO6), Mycobacterium segmatis RecA (PDB 1UBG), E. coli RecA (NCBI AAA75588), Prochlorococcus phage P-SSM2 (cyanophage) UvsX (NCBI AAX44563), Prochlorococcus phage P-SSM4 (cyanophage) UvsX (NCBI AAX46948), enterobacteria phage RB49 UvsX (NCBI AAL87837), enterobacteria phage RB69 UvsX (NCBI AAP75946), Pyrococcus furiosus RadA (NCBI AAL82050), Thermoplasma acidophilum RadA (NCBI NP_394563), Mus musculus Rad51 (NCBI BAA02718), Homo sapiens Rad51 (NCBI AAD49705), S. cerevisiae Rad51 (NCBI CAA45563), T. parva strain Muguga Rad51 (NCBI XP_763865), H. sapiens Dmc1 (NCBI CAG30372), and T. parva strain Muguga Dmc1 (NCBI XP_76380).
The authors thank Dr. Hang Xu, Dr. Jie Liu, and Jennifer Tomczak for helpful comments, suggestions, and technical assistance.
This work was supported by National Institutes of Health program project grant P01 CA098993.