Single Molecule Analysis of a Red Fluorescent RecA Protein Reveals a Defect in Nucleoprotein Filament Nucleation That Relates to Its Reduced Biological Functions*

Fluorescent fusion proteins are exceedingly useful for monitoring protein localization in situ or visualizing protein behavior at the single molecule level. Unfortunately, some proteins are rendered inactive by the fusion. To circumvent this problem, we fused a hyperactive RecA protein (RecA803 protein) to monomeric red fluorescent protein (mRFP1) to produce a functional protein (RecA-RFP) that is suitable for in vivo and in vitro analysis. In vivo, the RecA-RFP partially restores UV resistance, conjugational recombination, and SOS induction to recA− cells. In vitro, the purified RecA-RFP protein forms a nucleoprotein filament whose kcat for single-stranded DNA-dependent ATPase activity is reduced ∼3-fold relative to wild-type protein, and which is largely inhibited by single-stranded DNA-binding protein. However, RecA protein is also a dATPase; dATP supports RecA-RFP nucleoprotein filament formation in the presence of single-stranded DNA-binding protein. Furthermore, as for the wild-type protein, the activities of RecA-RFP are further enhanced by shifting the pH to 6.2. As a consequence, RecA-RFP is proficient for DNA strand exchange with dATP or at lower pH. Finally, using single molecule visualization, RecA-RFP was seen to assemble into a continuous filament on duplex DNA, and to extend the DNA ∼1.7-fold. Consistent with its attenuated activities, RecA-RFP nucleates onto double-stranded DNA ∼3-fold more slowly than the wild-type protein, but still requires ∼3 monomers to form the rate-limited nucleus needed for filament assembly. Thus, RecA-RFP reveals that its attenuated biological functions correlate with a reduced frequency of nucleoprotein filament nucleation at the single molecule level.

The fusion of native proteins to various fluorescent proteins has found widespread use in biology. If the fusion protein retains proper function, then the behavior and localization of the protein can be followed in living cells (1). Complementing the single-cell analysis, it is now possible to image the behavior of a fluorescent protein at the single molecule level (2)(3)(4)(5)(6)(7)(8). However, despite the growing popularity of fusion protein studies, a detailed biochemical analysis of the fusion protein is much less common, even though such examination is crucial for molecular interpretations. Thus, an in vivo and in vitro analysis of the function of a fusion protein relative to the wild-type protein is an essential prerequisite.
Homologous recombination is an important process not only for generating genetic variation, but also for maintaining genomic integrity through the repair of DNA breaks. In Escherichia coli, recombinational repair of double-stranded DNA (dsDNA) 3 breaks is mediated by the RecBCD pathway, whereas the repair of ssDNA gaps is mediated by the RecF pathway (9). Both of these recombination pathways require the functions of RecA protein.
RecA protein is essential to recombinational DNA repair (9 -11). RecA-like proteins are ubiquitous and highly conserved (12,13). The ATP-bound form of the protein binds to ssDNA and polymerizes along the DNA to form an extended nucleoprotein filament (14 -16). This is the functional form of the protein that interacts with dsDNA to search for a homologous sequence. Upon finding homology, RecA protein promotes the exchange of identical DNA strands to produce the heteroduplex joint molecules. The joint molecules can be converted into Holliday junctions and resolved by the RuvABC proteins to produce recombinant DNA products (17).
The binding of RecA protein to ssDNA is competitive with the ssDNA binding (SSB) protein (18,19). The assembly of RecA protein onto ssDNA that is complexed with SSB protein is a kinetically slow process, which is catalyzed by so-called mediator or loading proteins (20). RecBCD enzyme is one such RecA-loading protein (21,22), but an additional set of loading proteins are the RecF, RecO, and RecR proteins that can form various subassemblies to facilitate the RecA-mediated displacement of SSB from ssDNA (23)(24)(25)(26). In addition, a class of mutations that map to recA itself were isolated as suppressors of RecF function (srf) that produced mutant RecA proteins with an enhanced intrinsic ability to displace SSB from ssDNA (27). One such mutant is the RecA803 protein, in which valine 37 is mutated to methionine (28,29). This mutant RecA protein displays a higher intrinsic rate of nucleoprotein filament assembly on ssDNA, which is responsible for its enhanced capacity to displace DNA-bound SSB protein.
RecA protein was successfully fused to green fluorescent protein (GFP) and was visualized in living bacteria (30). The RecA-GFP protein foci were seen to appear after UV irradiation and to be dependent on the recB and recF gene products. Although this protein is clearly functional in vivo, it was unfortunately, largely insoluble in vitro, thereby limiting large scale purification. 4 Therefore, to facilitate biochemical use, an alternative fusion protein was constructed. In the present study, the monomeric red fluorescent protein (mRFP1 (31)) was fused to the carboxyl terminus of the RecA803 protein (referred to as RecA-RFP). The hyperactive RecA803 was used because it assembles on ssDNA more rapidly and competes better with SSB than the wild-type proteins and, as will be shown below, fusion to mRFP1 resulted in attenuated activity; thus, fusion to a hyperactive RecA protein permitted retention of at least partial function. The purified fluorescent protein binds to DNA but shows attenuated ATP and dATP hydrolysis activities. Although nucleoprotein filament assembly is inhibited by SSB protein under typical reaction conditions, we found that nucleoprotein filament formation and enzymatic activities are restored when dATP is substituted for ATP, or when the pH is lowered to 6.2. These characteristics are similar to those of the partially defective RecA142 mutant protein (32,33), thereby showing that the RFP fusion converted a hypermorphic protein to a hypomorphic RecA fusion protein. Fortunately, because the behavior of this RecA-RFP protein closely fits the biochemical profile of a previously characterized mutant RecA protein, we could understand its behavior. By observing assembly on single molecules of dsDNA, we could see that nucleation of a RecA-RFP filament was ϳ3-fold slower than for the wild-type protein. Importantly, these findings lend direct single molecule support to conclusions from ensemble studies where it was shown that biological function of the RecA protein correlates with its ability to displace SSB protein that, in turn, is related to the rate of RecA protein nucleation onto DNA (34).
RecA-RFP Protein Purification-RecA-RFP protein was purified as follows. SS2081 strain was grown overnight at 37°C in L broth containing amp, kan, and tet. The cells were harvested and stored in buffer (250 mM Tris-HCl (pH 7.5) and 25% sucrose) at 1 g/ml. Until use, the harvested cells were kept frozen at Ϫ80°C. Cells were lysed using 1.3 mg/ml lysozyme and 0.4% Brij-35 in the presence of 0.5 mM phenylmethylsulfonyl fluoride, and then centrifuged in a Beckman Ti45 at 41,000 rpm for 45 min. To the cleared lysate, ammonium sulfate (Am 2 SO 4 ) was added to 35% saturation. After centrifugation, the supernatant was recovered and the fluorescent protein was precipitated by addition of Am 2 SO 4 to 45% saturation. After centrifugation, the pellet was dissolved in R buffer (20 mM Tris-HCl (pH 7.5), 10% glycerol, 0.1 mM DTT, and 0.1 mM EDTA), and the sample was dialyzed against R buffer overnight. The solution was loaded onto a Q-Sepharose column (GE) equilibrated with R buffer. The RecA-RFP fractions were eluted by running a linear salt gradient to 500 mM NaCl in R buffer. The RecA-RFP eluted at ϳ300 mM NaCl. The pooled fractions were precipitated with 55% saturated Am 2 SO 4 , centrifuged, and then dialyzed against P buffer (20 mM potassium phosphate (pH 6.5), 10% glycerol, 0.1 mM DTT, and 0.1 mM EDTA) overnight. The dialyzed sample was loaded onto an ssDNA-cellulose column that was equilibrated with P buffer, and the fluorescent protein was eluted by washing with P buffer containing 200 mM NaCl and 1 mM ATP. The pooled protein fractions were dialyzed overnight against TEDS buffer (20 mM Tris-HCl (pH 7.5), 0.1 mM EDTA, 0.1 mM DTT, and 100 mM NaCl). The dialyzed sample was loaded onto Mono Q HR10/10 column (GE) equilibrated with TEDS buffer and was eluted with a salt gradient (100 -1000 mM NaCl). The pool after the Mono Q column was concentrated by dialysis against storage buffer (20 mM Tris-HCl (pH 7.5), 1 mM DTT, 0.1 mM EDTA, and 10% glycerol). The concentration of the fluorescent RecA protein was determined using the Bradford assay (Bio-Rad), using wild-type RecA protein as the standard. The yield of RecA-RFP was ϳ4.4 mg from 4 liters of culture, and its stock concentration was 21.5 M. A Hewlett-Packard HP8453 spectrophotometer was used.
Other Proteins and Reagents-SSB protein (37) and wild-type RecA protein (38) were purified as described previously. All chemicals were reagent grade and solutions were prepared using NanoPure water.
DNA Substrates Used for Biochemical Analysis-Poly(dT) (ϳ220 nucleotides in length) was purchased from GE Healthcare. M13 mp7 ssDNA and dsDNA were isolated as follows. At an A 600 of ϳ0.2, a fresh culture (400 ml) of XL-1 blue was infected with 2.4 ϫ 10 11 M13 mp7 phage. After 3 h of shaking at 37°C, cells were pelleted by centrifugation. The supernatant was used for preparation of the ssDNA (below). The pellet was washed with 20 ml of 50 mM Tris-HCl (pH 7.5) containing 10% sucrose, and M13 dsDNA was purified using QIAtip100 columns (Qiagen) according to the manufacturer's instructions. The M13 dsDNA was precipitated in ethanol and suspend in 100 l of TE buffer (10 mM Tris-HCl (pH 7.5) and 1 mM EDTA).
M13 mp7 ssDNA was prepared from the supernatant described above. The filtered (0.2 m) supernatant (50 ml) was ultracentrifuged in a Beckman 70Ti at 25,000 rpm for 60 min at 4°C. The resulting clear pellet was suspended in 500 l of 10 mM Tris-HCl (pH 7.5), 10 mM MgSO 4 , and 50 mM NaCl; then, 20 l of 0.5 M EDTA (pH 8.0), 25 l of 10% SDS, and 2.5 l of Proteinase K (1 mg/ml) were added. After mixing, the solution was incubated at 65°C for 30 min. The cooled solution was then extracted with an equal volume of phenol, followed by phenol/ chloroform, and then ether. The DNA was precipitated in ethanol and suspended in 200 l of TE buffer. Oligonucleotides and EcoRI-linearized M13 dsDNA were labeled at the 5Ј-end by T4 polynucleotide kinase using [␥-32 P]ATP; unincorporated [␥-32 P]ATP was removed using a MicroSpin S-200 HR column (GE).
UV Sensitivity Measurement-Cultures of exponentially growing cells (in L broth with amp, kan, and tet for selection of the plasmid and the host drug marker linked to the recA deletion) were diluted in L broth and spread on L agar plates. The plates were irradiated with UV light (254 nm) for various times. Colonies were scored after growth at 37°C for 20 h in the dark.
SOS Induction-Strains were grown in L broth at 37°C, washed in phosphate buffer, and UV irradiated for 10 s at 0.5 J/m 2 . The cells were then diluted into L broth and assayed for ␤-galactosidase activity (in duplicate) at the indicated times. The isogenic strains are derivatives of DM4000 and the different alleles were introduced by P1 transduction (39).
DNA Strand Exchange Assay-The procedure previously described was used (32), except for the buffer. Standard buffer (25 mM TrisOAc (pH 7.5), 10 mM Mg(OAc) 2 , and 0.1 mM DTT) containing 3 mM phosphoenolpyruvate, 80 units/ml pyruvate kinase, 1 mM ATP, 5 M (nucleotides) M13 ssDNA, 0.45 M SSB, 10 M (nucleotides) linear M13 dsDNA (linearized using EcoRI restriction endonuclease, and 5Ј-end labeled with [␥-32 P]ATP), and 3 M of either wild-type or fluorescent RecA protein were used. The reaction was initiated by addition of the linear dsDNA. Samples were withdrawn at the indicated times, and analyzed by agarose gel (0.8%) electrophoresis in TAE buffer at 25 volts for 16 -17 h.
Single Molecule Visualization of RecA-RFP Nucleoprotein Filaments-The experimental procedure was similar to that reported, with slight modifications (8). A three-channel flow cell was used to generate separate laminar flow channels of ϳ1.5 mm width each. Bacteriophage DNA, ligated to a 3Ј-biotinylated oligonucleotide complementary to cosR, was attached to streptavidin-coated 1-m polystyrene beads (Bangs Laboratories, Inc.). The DNA-bead complex was trapped in the first channel in 40 mM TrisOAc (pH 8.2), 15% sucrose, and 30 mM DTT. The trapped DNA-bead complex was then moved to the third channel containing the indicated concentration of RecA-RFP protein in 40 mM MES (pH 6.2), 15% sucrose, 30 mM DTT, 2 mM Mg(OAc) 2 , and 1 mM ATP␥S. The DNA-bead complex was incubated in the third channel for 5 min to form the nucleoprotein filament; the resultant fluorescent RecA-DNA-bead complex was then moved back to the second (observation) channel that contained the same solution, but lacked the fluorescent RecA protein. The flow rate was 120 m/s; the temperature was 37 Ϯ 1°C. For nucleation studies, the buffer composition in each channel was changed to permit comparison to previously published nucleation frequencies (8)

RecA-RFP Partially Complements the Phenotypic Deficiencies of a recA Null
Strain-A fusion of RecA protein to green fluorescent protein (GFP) resulted in a protein that partially restored recombination and survival to cells after UV irradiation, and was used to visualize RecA focus formation at sites of DNA damage in individual living E. coli cells (30). To define the phenotype of the RecA-RFP protein, and to permit comparison to the RecA-GFP protein previously described, these functions were also examined. The recA803-mrfp1 (hereafter referred to as RecA-RFP) construct partially complements the UV-repair deficiency of a recA null strain to a level comparable with that of the wild-type recA-gfp construct (Fig. 1A).
The partial complementation of UV survival suggests that RecA-RFP is at least partially active for recombinational DNA repair. To test recombination function directly, conjugational recombination was examined. Table 1 shows that recombina-tion by RecA-RFP is reduced only about 2-fold relative to wildtype, and is the same, within error, as the previously characterized RecA-GFP fusion.
A third biological function of the RecA protein is induction of the SOS-response via proteolytic cleavage of the LexA repressor. This activity was measured by using a reporter gene comprising a fusion of ␤-galactosidase to the LexA-regulated sulA gene (39). The results show that RecA-RFP induces the expression of this reporter gene in response to UV irradiation to almost the same extent as wild-type (Fig. 1B), indicating that RecA-RFP is proficient in LexA repressor cleavage in vivo. These collective results indicated that RecA-RFP was at least partially functional in vivo; as a consequence, the protein was purified and characterized in vitro.
Purification of RecA-RFP Protein-First, we attempted to purify the RecA-RFP fusion protein using a procedure that is standard for the wild-type RecA (38); however, that procedure was unsuccessful. Consequently, we modified the protocol (see "Experimental Procedures") and succeeded in purifying the fluorescent protein. As presented in Fig. 2, the excitation and emission peaks of RecA-RFP are 582 and 608 nm, respectively. We also attempted purification of the fusion of wild-type RecA protein with green fluorescent protein, using both the same procedure and a standard RecA purification protocol (38); however, the RecA-GFP protein was less soluble, and it showed little activity in vitro. 5 Therefore, we decided to focus on only the RecA-RFP protein.
RecA-RFP Protein Has ssDNA-dependent ATPase and dAT-Pase Activities-RecA protein has an ATPase activity that is stimulated by DNA. We first examined the ssDNA-dependent ATPase activity using an ssDNA, poly(dT), which is devoid of DNA secondary structure. Fig. 3A shows that the rate of ATP hydrolysis by RecA-RFP increases with increasing protein concentration, until it saturates at 1.6 M RecA-RFP. This concentration corresponds to saturation occurring at a molar ratio of 1 RecA-RFP per ϳ3.1 nucleotides, which is the canonical value for nucleoprotein filament formation by wild-type RecA protein. However, in the plateau region, the rate of ATP hydrolysis is only 13 M ATP/min, which corresponds to a k cat value of 8.1 min Ϫ1 . In comparison, the ATPase activity of the native RecA protein saturates at approximately the same protein concentration (1.5 M), but exhibits a 3-fold greater rate of ATP hydrolysis (39 M ATP/min), which corresponds to the expected k cat of 26 min Ϫ1 .
RecA protein is also a dATPase (40), and similar results were obtained when dATP hydrolysis by RecA-RFP was examined   The excitation spectrum was obtained by measuring the fluorescence at 608 nm, and emission spectrum was as obtained by exciting at 582 nm. Each spectrum was normalized to the peak value in each scan.  (19). Furthermore, k cat is also known to be unaffected by SSB protein (19). Therefore, a change in the observed rate of ssDNA-dependent ATP hydrolysis can be interpreted in terms of a proportional change in the amount of RecA-ssDNA complex formed. When ssDNA containing DNA secondary structure (e.g. M13 ssDNA) is used, ATP hydrolysis by wild-type RecA protein is stimulated by SSB protein due to disruption of the secondary structure by SSB protein. Subsequently, the SSB is displaced by RecA protein resulting in formation of more RecA-ssDNA complex, which is manifest in the increased rate of ATP hydrolysis (18,19). Consequently, we next examined the ssDNA-dependent ATP hydrolysis activity of RecA-RFP in the presence of SSB protein (Fig. 4). When M13 ssDNA is used, wild-type RecA protein shows the well documented stimulation of activity by SSB protein (Fig. 4A) (18,19). In contrast, the ATPase activity of RecA-RFP protein is almost completely inhibited by the SSB protein (Fig. 4A). These results show that the fluorescent fusion protein cannot displace SSB protein from ssDNA under these conditions (19).
Because the steady-state affinity of RecA protein for ssDNA is greater in the presence of dATP than ATP, all activities of RecA proteins are enhanced when dATP is used as the nucleotide cofactor instead of, or in addition to, ATP (40,41). In the case of RecA-RFP, its dATPase clearly does not recover to levels approaching the wild-type protein; nonetheless, activity is seen to increase (to about 12 M/min) with increasing RecA-RFP protein concentrations even when SSB protein is present (Fig.  4B). Saturation should result in a dATP hydrolysis rate in the vicinity of 16 M/min (based on poly(dT), Fig. 3), suggesting nucleoprotein filament formation to about 75% of maximal saturation; however, due to limitations imposed by solubility of  the RecA-RFP stock solution, higher concentrations of RecA-RFP could not be tested. These findings show that RecA-RFP nucleoprotein filament formation is occurring in the presence of dATP and SSB protein.
Previously, we showed that mutant RecA proteins (RecA142 and RecA430), which displayed reduced functionality in the presence of ATP, could be partially rescued in vitro by dATP (32). We also previously reported that defect of the RecA142 protein is largely suppressed both in vitro and in vivo by decreasing the pH (33). Consequently, we next examined both ATP and dATP hydrolysis by RecA-RFP at pH 6.2 (Fig. 5). As previously reported, at this pH, the ATPase activity of wild-type RecA in the absence of SSB protein is sufficiently high that the stimulatory effect of SSB protein is modest when compared with pH 7.5 (Fig.  5A). In contrast, RecA-RFP is almost completely inhibited by SSB protein, except at the highest protein concentrations tested, when ATP is present. However, when dATP is substituted for ATP, the dATPase activity increases at all RecA-RFP concentrations (Fig. 5B). This latter finding suggests that even though dATP turnover remains reduced, dATP-RecA-RFP nucleoprotein filaments can form when SSB protein is present.
To permit a more direct comparison with the DNA strand exchange experiments reported below, activity in the presence of a reduced concentration of SSB protein was also examined (Fig. 5, C and D). At this decreased SSB concentration (0.45 M), the competitive inhibitory effect of SSB is reduced. As expected, the ATP hydrolysis activity of RecA-RFP is less inhibited (Fig. 5C), and also the dATP hydrolysis activity of RecA-RFP is greater in the presence of SSB protein (Fig.  5D). Thus, it is clear that RecA-RFP protein is forming nucleoprotein filaments at the lower pH in both the presence of ATP and dATP.
RecA-RFP Can Promote DNA Strand Exchange at pH 6.2-The RecA-ssDNA filament will catalyze DNA strand exchange with homologous dsDNA in vitro. To test the functionality of RecA-RFP, DNA strand exchange was examined between M13 phage circular ssDNA and linearized dsDNA. In the presence of ATP, wild-type RecA forms joint molecule intermediates and nicked circular DNA products over time (Fig. 6A, left panel); however, at pH 7.5, the RecA-RFP protein forms neither intermediates nor products (Fig. 6A).
As mentioned above, the fluorescent RecA protein shares many characteristics with the RecA142 mutant protein.
Although RecA142 protein does not catalyze DNA strand exchange at pH 7.5 (32), it promotes DNA strand exchange at slightly acidic conditions (33). Consequently, DNA strand exchange was also assayed at pH 6.2 (Fig. 6A, right panels). As predicted, RecA-RFP protein produces joint molecule intermediates at the lower pH, although it is unable to produce nicked circular heteroduplex DNA product. These finding suggest  that, although functional, the RecA-RFP nucleoprotein filaments are incomplete, in agreement with the ATP hydrolysis data.
DNA strand exchange was also conducted using dATP instead of ATP (Fig. 6B). Now, the fluorescent RecA produces both joint molecule intermediates and some nicked circular DNA products, even at pH 7.5 (left panels). At the lower pH, RecA-RFP forms even more joint molecule intermediates (right panel). However, the yield of DNA strand exchange product is decreased for wild-type RecA with dATP at pH 6.2 (Fig. 6B). This seemingly unexpected result is likely due to enhanced binding of wild-type RecA protein to the dsDNA substrate under these conditions (33), causing an inhibition of DNA strand exchange (42); this inhibitory phenomenon is characteristic of, and well documented for, the eukaryotic Rad51 protein that readily binds duplex DNA to reduce DNA strand exchange (43).
RecA-RFP Can Be Used to Visualize Nucleoprotein Filament Formation on Single dsDNA Molecules-We next examined RecA-RFP nucleoprotein filament formation at the single molecule level using the protocol described (8). After incubating a DNA-bead complex in the RecA-RFP channel for 5 min in the presence of ATP␥S, a fluorescent RecA nucleoprotein filament forms on the dsDNA (Fig. 7A). The length of the ATP␥S-RecA-RFP nucleoprotein filament (L RecA ) is 22 Ϯ 2 m (n ϭ 21) (Fig.  7B). The length of the naked DNA (L naked ) was measured to be 12.8 Ϯ 0.6 m in a separate single molecule assay wherein a fluorescent tag was attached to the opposite cohesive end of DNA. Therefore, the extension of the nucleoprotein filament (R ϭ L RecA /L naked ) relative to naked dsDNA is 1.7 Ϯ 0.2. This extension is consistent with the extension (1.65 Ϯ 0.05) measured previously for wild-type RecA protein (8), suggesting that in the presence of ATP␥S, RecA-RFP can form complete and fully extended nucleoprotein filaments on dsDNA.
Nucleation of RecA-RFP Nucleoprotein Filaments-As established previously, nucleoprotein filament assembly is regulated at the nucleation step (8). To examine nucleation by RecA-RFP, the kinetics of nucleoprotein filament formation were examined. A trapped dsDNA molecule was incubated for the time indicated in the channel containing the RecA-RFP protein, and was then moved to the observation channel for analysis. Repetition of these steps permitted visualization of the time course of nucleation on the same DNA molecule (Fig. 8A). The appearance of RecA-RFP clusters was determined from an intensity profile of molecules such as those shown in Fig. 8A; the average total number of clusters on the DNA is plotted as a function of time in Fig. 8B. It is evident that the rate of nucleation increases with RecA-RFP concentration. However, compared with the wild-type protein, the rate of nucleation was only about onethird (8). 5 Given that the fluorescein-RecA protein used previously displayed normal levels of enzymatic activity, and that RecA-RFP displays attenuated activity, our finding that nucleation frequencies of RecA-RFP are reduced is consistent. Previously (34), from ensemble studies, we established that the rates of RecA protein assembly onto DNA, which are nucleation-limited, parallel the biochemical and biological behavior of mutant RecA proteins, for those with either reduced or enhanced functions. Thus, our findings with RecA-RFP refine that conclusion, and are the first to reveal a correlation between a single molecule nucleation frequency and biological activity.
Finally, when the observed rates of nucleation from Fig. 8B were plotted as a function of RecA-RFP protein concentration, the resultant plot (Fig. 8C) shows the expected power dependence on protein concentration (8). The exponent was 2.7 Ϯ 0.2, and this value provides an estimate of the number of monomers required to form the nucleus of a RecA filament.

DISCUSSION
Our goal was to produce a fluorescent RecA fusion protein that could be used for visualization both in vivo and in vitro, which would permit comparisons of physiological function and biochemical behavior. Knowing that the creation of such a fusion with RecA protein often created inactive RecA protein (30), 6,7 we fused monomeric red fluorescent protein to a hyperactive RecA protein, RecA803, hoping to produce a fusion protein that retained at least partial RecA function. Although RecA803 assembles on DNA more quickly and competes better with SSB than wild-type RecA protein (28,29), we show here that the fusion with RFP results in a protein that has reduced functionality. In vivo, when overproduced, RecA-RFP displays the same phenotypic behavior as the wild-type RecA-GFP and RecA803-GFP (30). RecA-RFP confers partial restoration of UV sensitivity to a recA null strain, partial restoration of conjugal recombination, and nearly complete recovery of the SOS response. In terms of RecA focus formation upon generation of DNA damage, the RecA803-GFP protein showed a greater number of foci per cell on average than RecA-GFP (30), consistent with the enhanced assembly capacity of RecA803 protein (29). Thus, although not identical, these three fusion proteins are suitable for in vivo investigations of RecA function. The RecA-RFP protein, however, has the advantage of being more soluble and, thus, more useful for biochemical and single molecule analyses in vitro.
Biochemically, we showed that the RecA-RFP protein: 1) displays lower levels for ATP and dATP hydrolysis compared with wild-type RecA protein (k cat is reduced 2.4 -3.2-fold); 2) forms nucleoprotein filaments that are sensitive to inhibition by SSB protein; 3) formed nucleoprotein filaments with a greater steady-state stability when using dATP, decreasing the pH, or lowering SSB protein concentrations; 4) is proficient in DNA strand exchange provided that dATP is present or, with ATP, that a lower pH of 6.2 is used; 5) can be used to image the assembly of individual RecA-RFP filaments on dsDNA; and finally, 6) nucleates on dsDNA with only about one-third the frequency of wild-type RecA protein.
Although the enhanced biochemical activities observed when using dATP or shifting the pH to 6.2 might seem unexpected, these are well known characteristics of RecA protein (33, 44 -52). The binding of dATP to RecA protein results in a protein that has a higher affinity for DNA, assembles more quickly, and dissociates more slowly than the ATP complex (26,40,41,51,(53)(54)(55). Likewise, reducing the pH to 6.2 results in similar effects (56). Both of these enhancing effects are additive; consequently, wild-type and mutant RecA proteins are most active in the presence of dATP at pH 6.2. Furthermore, we had previously established that dATP enhanced the activity of the wild-type protein, even in the presence of ATP (40,41). Thus, all of these biochemical variables are physiologically meaningful, because both intracellular pH and nucleotide pools change with growth conditions.
The biochemical behavior of RecA-RFP protein is somewhat comparable, although not identical, to that of the mutant RecA142 protein (32,33). The recA142 allele, in which isoleucine 225 is substituted with valine (57), is partially deficient for both recombination and SOS induction in vivo (58,59). Although the purified RecA142 protein has ssDNA-dependent ATPase activity, it has no DNA strand exchange activity with ATP at pH 7.5 (32). Resembling the RecA-RFP protein, the ATPase activity of the RecA142 protein is inhibited by SSB, showing that nucleoprotein filament formation is disrupted by SSB protein. This failure to successfully compete with SSB protein both in vitro and in vivo is sufficient to explain all of failings of the RecA142 protein (32). Surprisingly, it was subsequently discovered that the recombination deficiency was restored at lower pH values (6.2-6.5) both in vitro and in vivo (32,33); however, the RecA-RFP protein did not recover recombinational DNA repair function at lower pH growth conditions (data not shown). The only other major difference between RecA142 and RecA-RFP proteins is that the k cat for ATP hydrolysis by RecA142 is similar to that of the wild-type protein (32), whereas the k cat of RecA-RFP protein is reduced severalfold.
Because many characteristics of RecA-RFP protein resemble those of RecA142 protein, its defect may have a similar underlying cause. The mutation in the RecA142 protein (I225V) is located in a part of the hydrophobic core of the RecA protein that impinges upon the NH 2 -terminal region, which is one side of the monomer-monomer interface (33). The NH 2 terminus contains a large ␣-helix and short ␤-strand involved in RecA self-assembly (60 -62). The fluorescent protein used here was produced by fusing RFP to the COOH-terminal end of the hyperactive RecA803 protein, using a short amino acid linker. Although the COOH terminus does not directly contribute to the monomer-monomer interface (60,63), nonetheless, it may occlude this interface and thereby impede the assembly process. The fact that the rate of nucleation onto DNA is reduced supports this conclusion. The RecA-RFP protein was hoped to have activity comparable with the wild-type RecA protein. Instead, the fluorescent fusion protein displays a reduced functionality. Nevertheless, the fusion protein can still be useful for in vivo experiments, provided that one bear in mind that it is a hypomorph.
We were able to visualize individual filaments of RecA-RFP forming on single DNA molecules. The length measurements showed that the RecA-RFP nucleoprotein filament is extended 1.7-fold relative to naked DNA, which is in precise agreement with our previous measurements using wild-type RecA protein that was modified with carboxyfluorescein (8), indicating that a complete nucleoprotein filament is formed. Furthermore, we measured directly the rates of filament nucleation by RecA-RFP. Compared with the wild-type protein, the nucleation frequencies are reduced. The nucleation of RecA-RFP on dsDNA is slower than the wild-type protein by a factor of ϳ3. This finding is the first report of the nucleation frequency of a RecA protein with altered functional characteristics. Extensive biochemical analysis of mutant RecA proteins, and the correlation with their phenotypic behavior, led to the conclusion that the ability of a RecA protein to compete with SSB protein for DNA binding is the most predictive biochemical trait that correlates completely with its biological function (34). This ability to compete with SSB protein, in turn, was related directly to the rate at which a RecA protein assembles onto ssDNA. Using dsDNA, we previously showed that assembly of RecA protein is controlled by a rate-limiting nucleation step involving 4.5 (Ϯ 0.5) monomers. This value is close, but not identical, to the value obtained with the RecA-RFP fusion protein; although this apparent difference may reflect real differences between the two proteins, we believe that it largely reflects the difficulty of measuring the rates of nucleation at the highest concentrations of RecA, which disproportionately bias the fit to the power function. Finally, although the absolute rate of nucleation onto ssDNA and onto the SSB-ssDNA complex will unquestionably be different from the rate of nucleation onto dsDNA, if we assume that the rates will parallel one another, then our results here show that a 3-fold lower rate of nucleation is sufficient to manifest reduced function in vivo and in vitro.
Finally, even though chemical modification of the RecA protein yielded a protein that fully retained biochemical activities (8), we note that the fluorescent fusion protein described has the advantage that it can be used to analyze quantitatively both in vivo cellular function and in vitro biochemical activity for the same protein. Thus, exact correlations of in vivo and in vitro function are now possible. Given that the functions of RecA protein can be altered by physiological factors such as ATP, dATP, pH, concentration of RecA and SSB proteins, volume-excluding conditions (64), and polyvalent ions (65), comparative individual cell analysis in vivo and individual molecule analysis in vitro will be very revealing.