ABSTRACT
The RecA protein of Escherichia coli performs a number of ATP-dependent, in vitro reactions and is a DNA-dependent ATPase. Small oligodeoxyribonucleotides were used as DNA cofactors in a kinetic analysis of the ATPase reaction. Polymers of deoxythymidilic acid as well as oligonucleotides of mixed base composition stimulated the RecA ATPase activity in a length-dependent fashion. Both the initial rate and the extent of the reaction were affected by chain length. Full activity was seen with chain lengths >= 30 nt. Partial activity was seen with chain lengths of 15-30 nt. The lower activity of shorter oligonucleotides was not simply due to a reduced affinity for DNA, since effects of chain length on KmATP and the Hill coefficient for ATP hydrolysis were also observed. The results also suggested that single-stranded DNA secondary structure frequently affects the ATPase activity of RecA protein with oligodeoxyribonucleotides.
DNA strand exchange catalyzed by Escherichia coli RecA protein in vitro has been divided into three distinct stages: presynapsis, synapsis and branch migration (1 ,2 ). ATP binding is required during all three stages of the reaction, but the role of hydrolysis is less clear. The ATPase activity of RecA protein is strongly stimulated by single-stranded (ss)DNA (3 -6 ). Early experiments suggested that ATP hydrolysis was necessary to form stable heteroduplexes (7 -8 ) and was required only for the branch migration stage of strand exchange (8 -9 ). Menetski et al. reported that all three phases of the strand exchange reaction can occur in the absence of ATP hydrolysis, since the non-hydrolyzable analog ATP[[gamma]]S can replace ATP (10 ). In addition, they found that plectonemic joint molecules up to 3.4 kb in length were formed. These studies utilized completely homologous DNA molecules as substrates and, under these reaction conditions, it was shown that the ATPase activity of RecA protein was required to dissociate the protein from the heteroduplex products once the reaction was complete (10 -12 ). Recently Kim et al. (13 ) and Rosselli and Stasiak (14 ) showed that in addition to recycling RecA, ATP hydrolysis is required to facilitate the bypassing of structural barriers, such as heterologous sequences, that may exist in either one or both strands during strand exchange. In addition, Kim et al. showed that ATP hydrolysis is required for strand exchange involving four strands (15 ). Finally, Konforti and Davis showed that the displaced strand and ATP hydrolysis are factors that determine the polarity of strand exchange in vitro (16 ).
The majority of studies referenced above used genome length DNA, such as M13 and [Phi]X174, as the ssDNA cofactor and the replicative forms of these phages as homologous duplex DNA. If the DNA cofactor is a short oligonucleotide, simpler complexes of RecA, ATP and ssDNA can be formed. A number of investigators used single-stranded oligodeoxyribonucleotides of defined length and sequence as ssDNA cofactors to investigate the repressor cleavage activity of RecA (4 ,5 ,17 ,18 ), the ssDNA binding activity of the protein (19 ), the sequence dependence of the ATPase activity (20 ), the importance of ends in the ATPase activity of RecA protein (21 ), the ability of RecA to form D loops (22 ,23 ), the role of the ATPase activity in strand exchange (14 ), the synapsis event in the homologous pairing of DNA molecules by RecA (23 ), the role of RecA in blocking transcription (24 ), the base pairing and interactions involved in homologous recognition promoted by RecA protein (25 -27 ), the interaction between RecA and fluorescent dye-labeled oligonucleotides (28 ) and the interactions between strands in RecA-DNA complexes (29 ).
Analysis of the binding of RecA to oligonucleotides was performed using non-stoichiometric amounts of RecA and the non-hydrolyzable ATP analog ATP[[gamma]]S (19 ). Under these conditions, RecA protein can bind to oligonucleotides 16-20 nt in length, suggesting a complex of 5-7 RecA monomers. Furthermore, this binding required a 10- to 100-fold molar excess of RecA over nucleotide concentration and required ATP[[gamma]]S. RecA was able to bind weakly to oligonucleotides 9-15 nt in length but was unable to bind to oligonucleotides 8 nt in length.
Amaratunga and Benight demonstrated that (dA)16,(dT)16 and(dT[middot]dC)8 were able to stimulate the ATPase activity above that of the DNA-independent reaction, while oligo(dT[middot]dC[middot]dA)8, (dC[middot]dG)16 and (dA[middot]dG)8 could not (20 ). Although RecA can bind to oligonucleotides 9-15 nt in length, polymers of (dT) or (dA) less than 20 nt in length were inefficient as cofactors in either the ATPase reaction or in the repressor cleavage reaction (17 ). For polymers of (dT) >25 nt in length, KD(app) in the ATPase reaction was a function of DNA chain length (21 ). Oligo(dT)50 was shown to be as effective as poly(dT) in stimulating the ATPase activity of RecA, oligo(dT)40 and oligo(dT)30 were found to be less active than poly(dT) and oligo(dT)25 was marginally active. It was concluded from this study that efficient binding to oligonucleotides, with ATP as cofactor, required an oligonucleotide of 30-50 nt in length, suggesting 10-17 RecA monomers in the complex if the binding site is 3 nt/RecA. Thus it is apparent that although RecA can bind to oligonucleotides 20 nt in length, efficient binding (as determined by maximum stimulation of ATPase activity) occurs when the chain length is >= 25 nt; below this length, binding appears to be less efficient and ATPase activity is not stimulated significantly above background.
Analysis of the D loop reaction using a mixture of ADP and ATP[[gamma]]S (1.1 and 0.3 mM) showed that oligonucleotides with 26 bases of homology are able form D loops, whereas 20 bases are not sufficient (23 ). Maximum efficiency was seen with 38-56 bases of homology, in agreement with another study showing that oligonucleotides 30-50 nt in length (with ATP[[gamma]]S) were required to form D loops (22 ). It was later shown that RecA protein was able to pair homologous oligonucleotides 15 bp long, with as few as 8 bp of homology being required to form synaptic complexes that were unstable in the absence of RecA (23 ). Using fluorescent dye-labeled oligonucleotides, DNA substrates to which RecA protein has an enhanced affinity, it was shown that using this artificial substrate in the presence of ATP[[gamma]]S, RecA could bind an 18mer and use this chain length in strand exchange reactions (28 ). Thus, from these studies using oligodeoxyribonucleotides as ssDNA cofactor, it is clear that although RecA can bind to oligonucleotides 9-20 nt in length under non-stoichiometric conditions, longer chain lengths (i.e. >25 nt) are required for the repressor cleavage activity, the ability to form D loops and for maximum stimulation of ATPase activity of the enzyme.
To further investigate this length dependence of RecA protein, a steady-state kinetic analysis of the ATPase activity with oligodeoxyribonucleotides as ssDNA cofactor was performed. These experiments were designed to provide insight into the inability of short oligonucleotides (i.e. <25 nt) to function as cofactors for RecA and into the ability of longer chain lengths ( >= 30 nt) to function efficiently as ssDNA cofactor. Our results show that the ability of an oligodeoxyribonucleotide to function as an efficient cofactor in the ATPase reaction correlates with its ability to promote efficient ATP binding, as evidenced by a low KmATP and a Hill coefficient for ATP binding >= 3. In contrast, oligonucleotides which are not efficient cofactors for ATPase do not promote efficient ATP binding, i.e. they demonstrate an increase in KmATP and a Hill coefficient for ATP binding <3. The ability of an oligodeoxyribonucleotide to function as a cofactor for ATPase is further complicated by the effects of single-stranded DNA secondary structure.
Phosphocellulose P11 was obtained from Whatman, ATP and ADP were purchased from Boehringer Mannheim and Polymin P was from BRL. [2,8-3H]ATP (sp. act. 25 Ci/mmol) was from NEN DuPont, plastic backed polyethyleneimine (PEI)-cellulose sheets were from Brinkman and Ecolume was from ICN.
The sequences of the oligonucleotides used in this study are shown in Table 1 . The nomenclature for the oligonucleotides is based on an index number in a laboratory collection. Oligo(dT)s were purchased from either Pharmacia [poly(dT) and oligo(dT)10] or from Bio-Synthesis Inc. (Denton, TX). Oligodeoxyribonucleotides of mixed base composition were either purchased from Bio-Synthesis Inc. or synthesized in-house on an Applied Biosystems DNA Synthesizer (Applied Biosystems Inc., Foster City, CA). The oligonucleotides were stored in 1* TE buffer (Tris-HCl-EDTA buffer, pH 8.0) at -20oC and initial concentrations determined by measuring the absorbance at 260 nm. For oligo(dT)s an [epsilon] value of 8520 M-1cm-1 was used to determine concentration (30 ) and for the oligonucleotides of mixed base composition, nucleotide concentrations were determined by taking an A260 of 1 to be 20 [mu]g/ml (31 ). All concentrations are reported as total nucleotides. In all experiments, unless otherwise stated, oligonucleotides were always used in a 200-fold excess so that concentration differences due to variations in extinction coefficients due to base composition were not significant
Single-stranded M13 phage DNA (M13 ssDNA) was prepared by infecting E.coli strain JM101 with wild-type M13, at a multiplicity of infection of 10. The procedure used to purify single-stranded phage DNA was that of Neuendorf et al. (32 ). The concentration of phage DNA was determined by taking an A260 of 1 to be 40 [mu]g/ml (31 ). The purified DNA was aliquoted and stored in 1* TE buffer, pH 8.0, at -20oC. All concentrations are reported as total nucleotides.
RecA protein was purified from E.coli strain GE1171 [F- [Delta](lac pro)XIII ara argE(Am) gyrA(Nalr) rpoB(Rifr) sfiA11 lexA::Tn5(Kanr) [Delta]recA1398 srl::Tn10(Tetr)], which contains the plasmid pGE226, a pBR327-based plasmid with the tetr gene removed and replaced by a 3 kb fragment containing the E.coli recA+ gene. The plasmid confers ampicillin resistance and is ~5 kb in size. The procedure used to purify RecA is based on the procedures of Weinstock et al. (33 ) and Griffith and Shores (34 ). The procedure involves a selective extraction from Polymin P followed by phosphocellulose chromatography. This is followed by spermidine acetate precipitation of RecA protein (pH 7.5, final concentration of spermidine acetate 7 mM). The precipitate is pelleted at 10 000 r.p.m. for 15 min at 4oC. The resulting pellet is resuspended in a buffer containing 20 mM Tris-HCl, pH 7.5, 1 mM DTT, 0.1 mM EDTA and 20% glycerol and stored aliquoted at -80oC. The concentration of purified RecA protein was determined at 280 nm using [epsilon] = 27 000 M-1cm-1 (2 ). Using an assay that followed the conversion of native [3H]DNA to acid-soluble nucleotides, no detectable nuclease contaminant was found.
The hydrolysis of ATP by RecA protein was monitored using a thin layer chromatography assay employing [3H]ATP. The concentrations of ATP and ADP were determined at 259 nm (pH 7.0) using [epsilon] = 15 400 M-1cm-1 (35 ). Unless otherwise specified, reaction conditions were as follows: reaction mixtures (30 [mu]l) contained 20 mM buffer (Tris-HCl, pH 8.0), 10 mM MgCl2, 30 mM NaCl, 1 mM DTT, [2,8-3H]ATP (67 [mu]Ci/ml), DNA at a final concentration of 200 [mu]M nucleotides and RecA protein as indicated. Reactions were conducted in 500 [mu]l plastic Eppendorf tubes at 37oC. Aliquots (0.5 [mu]l) were spotted onto PEI-cellulose strips (7 * 50 mm) containing ATP and ADP markers. The chromatography plates were developed in 1 M formic acid and 0.5 M LiCl. The ATP and ADP spots were identified with a UV lamp, cut out and counted in scintillation fluid (Ecolume, ICN).
Table 1
All assays used to measure ATP hydrolysis were performed as described above. At least eight ATP concentrations were used to determine KmATP and at each concentration, the velocity of the reaction was determined from time points taken before 30% hydrolysis had occurred during the linear portion of the time course. At 40% hydrolysis, the time course became non-linear due to the accumulation of ADP, a competitive inhibitor of the ATPase activity of RecA (33 ), and as a result these data points were not included in rate calculations. The values for KmATP and Vmax were determined from Eadie-Hofstee plots (36 ). The Hill coefficients were calculated from the slope of Hill plots, where data were replotted by analogy with the Hill plot for ligand binding (37 ).
To ensure that ssDNA chain length was the principal variable, ATPase activity was measured under conditions where the concentration of ATP (900 [mu]M) and ssDNA (200 [mu]M) were in large excess over RecA protein (1 [mu]M). The kinetics of the hydrolysis of ATP when either M13 ssDNA or other genome size DNAs are the DNA cofactor are well characterized (6 ,33 ,36 ,38 ). Consequently, M13 ssDNA was the standard with which we compared other polynucleotides. Under the conditions of our assay with M13 ssDNA, RecA protein was able to efficiently hydrolyze ATP with a kcat of between 16 and 24 min-1 (Fig. 1 and Table 2 ). These numbers agree with the published turnover numbers for RecA protein (36 ,39 ). For the reaction with M13 ssDNA, the maximum extent of hydrolysis varied between 60 and 80% in different experiments (Fig. 2 and Table 2 ).
The ATPase activity of RecA protein is affected by ssDNA in a length-dependent fashion. Our primary conclusion is that the steady-state kinetic parameters, KmATP, Vmax, Hill coefficient for ATP binding, the extent of the ATPase reaction and kcat, are all a function of ssDNA chain length. These effects are likely to be interrelated. In addition to chain length, ssDNA secondary structure affects the efficiency of ssDNA in stimulating the ATPase activity of RecA protein.
The effects we have observed of ssDNA chain length on ATPase activity are not inconsistent with previous reports (19 -21 ). Brenner et al., however, concluded that RecA had different affinities for the various oligonucleotides and Vmax for ATP hydrolysis remained the same (21 ). Their experiments used DNA concentrations of 100 [mu]M or lower, while our experiments used DNA concentrations up to 1000 [mu]M and, in some cases, up to 2000 [mu]M. At these high concentrations, the binding of RecA to ssDNA appears not to be the predominant factor.
The extent of the ATP hydrolysis reaction was also affected by ssDNA chain length. RecA protein is unable to hydrolyze 100% of the ATP in an in vitro reaction that does not contain an ATP regenerating system due to competitive inhibition of ATP binding by ADP (6 ,33 ). As chain length decreased, the maximum extent of ATP hydrolysis decreased from 60-70% for poly(dT) to 1.4% for oligo(dT)10. Since the KmATP also increased with shorter chain lengths, it is possible that this results in greater sensitivity to inhibition by ADP, resulting in a lower extent of hydrolysis.
It was previously demonstrated (33 ,41 ), that cooperative interactions between RecA monomers along the DNA backbone are important for ATPase activity. It was further suggested (39 ) that ATPase activity requires the formation of clusters of 15-20 ATP-RecA molecules bound contiguously along the ssDNA backbone and that an individual RecA molecule becomes a fully active ATPase only when it is part of a cluster that has exceeded this size. Our data demonstrate that oligonucleotides <30 nt in length are unable to fully stimulate the ATPase activity of RecA protein. This suggests that short DNA molecules are unable to provide sufficient lattice length along which a sufficient number of RecA monomers can be contiguously bound to give full ATPase activity. The minimum length of 30 nt, assuming a stoichiometry of 3 nt/monomer (42 ), produces a minimum cluster size of 10 RecA monomers. Although the agreement between the two studies in terms of the size of the active complex is only approximate, it is clear that a large complex of RecA monomers is required for full ATPase activity.
What can we infer about the structure of the clusters? The Hill coefficient of >3 may mean that a RecA monomer contacts three or more other monomers, which facilitates ATP binding to the monomer. Structural studies (12 ,43 ,44 ) indicate that RecA forms a filament with six monomers per turn and each monomer contacts the two adjacent neighboring monomers. How can a structure where each monomer contacts only two other subunits give rise to a Hill coefficient of >3? One possibility is that filaments are also paired side by side, so that each monomer makes both inter- and intra-filament contacts. Such a structure has been observed (44 ). Alternatively, RecA monomers may not only contact their immediate neighbors in a filament, but also subunits located one turn up- or downstream, positioned on top of and below the monomer. Although such contacts were not observed in the RecA crystal structure (44 ), this structure did not include DNA and thus the monomers are likely to be in a different conformation than in the ATPase reaction.
These models also account for the increase in KmATP and decrease in the Hill coefficient as chain lengths get shorter. Although there will always be interaction with immediate neighbors, there will be fewer inter-filament or up- or downstream interactions as the chain length decreases. Hence, the Hill coefficient will decrease and affinity for ATP will be poorer due to this loss of cooperativity. Note that there need be no change in the affinity of RecA for DNA as chain length decreases in these models.
The fact that chain length affects ATP binding does not necessarily explain the effect of chain length on kcat. What is the trigger for hydrolysis of ATP and why should it be affected by oligonucleotide chain length? A possible explanation is that hydrolysis requires, and perhaps is triggered by, the formation of the multisubunit complex. Thus, as chain length decreases the complexes formed are smaller, resulting in a reduced maximum velocity for hydrolysis. The idea that a multisubunit complex must form for hydrolysis is consistent with the model that hydrolysis of ATP may propagate longitudinally through the nucleoprotein filament in the form of coordinated waves (45 ). This requires that one monomer affect the activity of its neighbors separated by five or six RecA monomers (46 ), as proposed here.
The biological significance of the 30 nt length may be related to the minimum size required for homologous pairing. It has been demonstrated in vivo that there is a minimal DNA length (23-40 bp) below which RecA-dependent homologous recombination becomes inefficient (47 ,48 ). There is also a sharp cut-off between 30 and 151 bp of homology for the pairing reaction promoted by RecA protein in vitro (1 ) and it has also been reported that oligonucleotides with 26 bases of homology are able to form D loops, whereas 20 bases are not sufficient (23 ). Maximum efficiency required 38-56 bases of homology, in agreement with another report that oligonucleotides 30-50 nt in length (with ATP[[gamma]]S) were required to efficiently form D loops (22 ). These observations suggest that the requirement for a multisubunit complex including 30 nt of ssDNA ensures that the region involved in homologous pairing is long enough to ensure unique recognition of sequences.
This work was supported in part by PHS grant GM3524 to G.M.W. This work was submitted in partial fulfillment of the PhD degree by P.R.B.
*To whom correspondence should be addressed. Tel: +1 713 792 5266; Fax: +1 713 794 4150; Email: georgew@utmmg.med.uth.tmc.edu
+Present address: Sections of Microbiology and of Molecular and Cellular Biology, University of California at Davis, Davis, CA 95616, USA
Oligonucleotide no.
Length (nt)
Oligonucleotide sequence
44
25
GGGCCTCTAGATTTAAAGCTAGCGC
51
25
GATCCATAGTGAGGGCAACTAAACC
28
27
TGGGTCCAGCGCCTATTCAGCATCGAT
29
27
cgcaaggcccatctaagagtcgccgat
30
27
cgtcagcgtggtctaaccggaagattc
31
27
ttaaaacaataagcttaaaaataaata
32
27
tgtcaagactgtaagcttccatttttg
50
27
aagatcatTcctttgatggttagcttc
41
32
gAtcctaagtaagtaaggagaaaaaaatggct
42
32
gatcagccatttttttctccttacttacttag
43
32
ggccgcgctagctttaaatctagaggcccccc
45
33
aattagcccgcctaatgagcgggcttttttttg
46
33
aattcaaaaaaaagcccgctcattaggcgggct
49
33
gcaaaacgggaaaggatccgtccaggacgctca
156
40
TCTGGACTTGTTTCATAAGGGCGGCCTTCATCGACATAAG
157
40
GAAGTAGCTGTATTCGAACCGTCTGAAGCTGAGATCTCTC
158
40
TTCGACTCTAGAGAGTGTGAGTCTTTCGAGACCTAGGACC
149
41
gatcggtaccgaacatattgactatccggtattacccggca
150
41
AGCTTTCCGGGTAATACCGGATAGTCAATATGTTCGGTACC
REFERENCES
Return