The recombination mediator proteins RecFOR maintain RecA* levels for maximal DNA polymerase V Mut activity

DNA template damage can potentially block DNA replication. Cells have therefore developed different strategies to repair template lesions. Activation of the bacterial lesion bypass DNA polymerase V (Pol V) requires both the cleavage of the UmuD subunit to UmuD′ and the acquisition of a monomer of activated RecA recombinase, forming Pol V Mut. Both of these events are mediated by the generation of RecA* via the formation of a RecA–ssDNA filament during the SOS response. Formation of RecA* is itself modulated by competition with the ssDNA-binding protein (SSB) for binding to ssDNA. Previous observations have demonstrated that RecA filament formation on SSB-coated DNA can be favored in the presence of the recombination mediator proteins RecF, RecO, and RecR. We show here using purified proteins that in the presence of SSB and RecA, a stable RecA–ssDNA filament is not formed, although sufficient RecA* is generated to support some activation of Pol V. The presence of RecFOR increased RecA* generation and allowed Pol V to synthesize longer DNA products and to elongate from an unpaired primer terminus opposite template damage, also without the generation of a stable RecA–ssDNA filament.

DNA template damage can potentially block DNA replication. Cells have therefore developed different strategies to repair template lesions. Activation of the bacterial lesion bypass DNA polymerase V (Pol V) requires both the cleavage of the UmuD subunit to UmuD and the acquisition of a monomer of activated RecA recombinase, forming Pol V Mut. Both of these events are mediated by the generation of RecA* via the formation of a RecA-ssDNA filament during the SOS response. Formation of RecA* is itself modulated by competition with the ssDNA-binding protein (SSB) for binding to ssDNA. Previous observations have demonstrated that RecA filament formation on SSB-coated DNA can be favored in the presence of the recombination mediator proteins RecF, RecO, and RecR. We show here using purified proteins that in the presence of SSB and RecA, a stable RecA-ssDNA filament is not formed, although sufficient RecA* is generated to support some activation of Pol V. The presence of RecFOR increased RecA* generation and allowed Pol V to synthesize longer DNA products and to elongate from an unpaired primer terminus opposite template damage, also without the generation of a stable RecA-ssDNA filament.
DNA template damage represents a potential block to DNA replication. As such, the cell has developed many different strategies to maintain genomic integrity in the face of template lesions. Overall, these DNA damage repair strategies divide into those that are error-free and those that are error-prone. The latter pathways reflect the action of specialized DNA polymerases that can bypass directly the lesions in the DNA template. Escherichia coli has three translesion synthesis (TLS) 2 DNA polymerases: DNA polymerase (Pol) II, Pol IV (DinB (1)), and Pol V (UmuDЈ 2 UmuC (2, 3)). All three of these TLS polymerases are induced during the SOS response (4) and have been shown to be capable of lesion-induced or targeted mutagenesis (5). However, it is Pol V that is the prime agent of UV-induced mutagenesis discovered by Witkin (6).
UV-induced mutagenesis reflects an increase in spontaneous mutagenesis as a result of UV irradiation. Early models suggested that the replicative polymerase, the DNA polymerase III holoenzyme (7), might be modified by the action of the umu gene products to allow it to bypass template damage (8). Mutations in umuD and umuC eliminated the bulk of UV-induced mutagenesis (9,10). However, it was demonstrated subsequently that UmuC itself possessed DNA polymerase activity and that the UmuDЈ 2 UmuC complex could bypass template damage in the presence of activated RecA (RecA*) (2,3).
The TLS form of PolV contains a stoichiometric RecA monomer and has been termed a mutasome or Pol V Mut (11,12) that must bind an ATP moiety to be active (13). Also, whereas it is clear that the RecA is transferred from the 3Ј end of a RecAsingle-stranded (ss) DNA filament to Pol V (14), there is some debate about whether the RecA filament has to be directly adjacent to and downstream of the template damage (i.e. in cis) (15,16) or is in trans on a ssDNA not associated with the damaged template (17,18). RecA* is formed when ssDNA is generated as a result of replication forks encountering template lesions, as during the SOS response (19). The likely disposition of the replicated sister chromosomes is that of gaps in both nascent DNA strands (20) generated by the replisome skipping over the template lesion and continuing replication downstream (21,22). It has been accepted that RecA must compete with SSB for binding to the ssDNA in the gaps to form a RecA-ssDNA filament. Biochemically, the recombination mediator proteins RecF, RecO, and RecR are known to facilitate RecA nucleation on SSB-ssDNA (23) and in gaps (24). It is not surprising, then, that recFOR mutants display a delay in the induction of the SOS response (25,26).
These observations led Fujii et al. (27) to examine the roll of the RecFOR proteins in TLS by Pol V. They found, using a primer extension assay, that TLS by Pol V in the presence of SSB, RecA, and the ␤ sliding clamp required the RecFOR proteins. They suggested that the RecFOR requirement was a result of mediating RecA filament formation on the ssDNA. We have found a similar requirement for RecFOR in Pol V TLS and show that this requirement is an effect neither on access of Pol V to the 3Ј end of the primer nor on Pol V DNA polymerase activity per se but reflects a demand to maintain RecA* levels sufficient for maximal Pol V Mut activity during lesion bypass.

Reconstitution of TLS by Pol V
To examine Pol V TLS, we used the primer template (p/t) shown in Fig. 1A. Here the bottom, damage-containing template strand is 139 nt long. A 5Ј-[ 32 P] 27-nt-long primer is annealed such that its 3Ј-end is 5 nt upstream of the first thymidine residue in a cyclopyrimidine dimer (CPD) or 7 nt upstream of a tetrahydrofuran abasic site analog (THF). The ss template upstream of the 5Ј end of the primer prevents the ␤ processivity clamp from sliding off the primer template once it has been loaded.
DNA synthesis by Pol V was reconstituted in the presence of ␤, the DnaX clamp-loading complex (DnaX cx, 2 ␥␦␦Ј) (7), SSB, RecA, and RecFOR. The incubations were conducted in two steps. First, SSB, RecA, ␤, DnaX cx, RecA, and RecFOR were incubated with the p/t for 5 min at 37°C. Pol V was then added, and the incubation continued for 8 min. Reaction products were recovered by ethanol precipitation after phenol-CHCl 3 extraction, resuspended in denaturing loading dye, and electrophoresed through polyacrylamide gels (20%) containing 7 M urea. Product formation was quantified by phosphorimag-

PolV Mut and RecFOR
ing and is presented as a fraction of total primer. We define the extent of TLS as the primer extended up to and including the base opposite the lesion and extension as the primer extended past the lesion. Fig. 1B shows the results of omitting each of the protein components individually from the reaction using each of the three templates (undamaged, CPD, and THF). Only Pol V and RecA were absolutely required for DNA synthesis (Fig. 1B, (Fig. 1C), suggesting that Pol V processivity either had increased or had a greater level of RecA* than could be achieved in the presence of RecA and SSB was required for extensive Pol V-catalyzed DNA synthesis.
Note that the complete lack of DNA synthesis in the absence of RecA (Fig. 1B, left panel, lane 5) indicates that no other component of the reaction was contaminated to any substantive extent with a DNA polymerase activity. Similarly, the lack of DNA synthesis activity in the absence of Pol V (Fig. 1B, left panel, lane 2) demonstrates that the RecA itself was also not contaminated with a polymerase activity.
The results were different when damage-containing p/ts were used (Fig. 1B, center and right panels). Interestingly, ␤ and SSB were not required for TLS but were required for extension from the damage (Fig. 1B, center and right panels, compare lanes 3 and 4). The requirement of ␤ for bypass has been noted previously (2,15,30), although Pol V activity was not parsed in the same manner as we do here (i.e. TLS and extension). Indeed, Maor-Shoshani and Livneh (31) noted that ␤ did not stimulate initiation of DNA synthesis by Pol V but did stimulate bypass. Examination of their data suggests that the stimulation of bypass they observed was similar to what we observe here, i.e. stimulation of extension past the damage.
RecFOR stimulated extension past the damage on both the CPD and THF templates, whereas they had little effect on either TLS or overall DNA polymerase activity (Fig. 1B, center and right panels, compare lanes 1 and 9). We reasoned that this apparent requirement for RecFOR in extension from damage and the apparent increase in processivity observed in the pres-ence of RecFOR on the undamaged template might both reflect the action of RecFOR in generating RecA*. If this is the case, one would predict that the RecFOR effect would depend on the presence of SSB in the reaction.

Effect of SSB on RecFOR stimulation of Pol V activity
RecOR are sufficient to exchange RecA with SSB on SSBcoated ssDNA (23), and RecF is thought to provide nucleation for this exchange at the double-stranded edges of gaps (24). We therefore examined the effect of RecOR and RecFOR on Pol V activity in the presence and absence of SSB on the three templates (Fig. 2). Comparison of the extents of elongation on the various templates showed that RecFOR stimulated extension on the CPD and THF templates by roughly 5-and 6-fold, respectively, whereas there was little effect in the absence of SSB (Fig. 2B). RecOR had little effect in either the presence or absence of SSB with the CPD template but did show approximately half the stimulatory effect as RecFOR in the presence of SSB with the THF template (Fig. 2B). The requirement for RecF for the maximum effect is interesting and not obviously explained by current biochemistry. These templates do not model a gap in newly replicated DNA because there is no 5Ј-ended DNA downstream of the template damage where one might expect RecF to facilitate 5Ј 3 3Ј RecA filament growth

PolV Mut and RecFOR
toward the primer terminus. On the other hand, RecF, with RecR, has been shown to limit extension of the RecA filament from ssDNA onto dsDNA (32). Extension of the RecA filament at the primer terminus onto the dsDNA formed by the primer and the template would presumably inhibit DNA synthesis by Pol V; thus, preventing such extension could result in the stimulation we observe.
It is difficult to judge the dynamic between SSB and RecA with the Pol V system because of the absolute requirement for RecA to observe polymerase activity. We therefore turned to another Y family TLS polymerase from E. coli, DNA polymerase IV, and used it to replace Pol V in the primer extension assay on an undamaged template (Fig. 3). SSB stimulated extension by DNA Pol IV (Fig. 3, compare lanes 1 and 2), whereas RecA alone inhibited extension (Fig. 3, compare lanes 4 and 2), although it did not shut it down completely, suggesting that the region at the 3Ј end of the RecA-ssDNA filament is dynamic, allowing some access by the polymerase. SSB alone was sufficient to overcome the RecA inhibition (Fig. 3, compare lanes 3  and 4), arguing that under our conditions, even though RecA is in 20-fold molar excess of SSB tetramers, the preferred agent binding to the ssDNA is SSB and furthermore suggesting that mixed SSB-RecA-ssDNA filaments are fairly unstable. RecFOR alone had no significant effect (Fig. 3, compare lanes 2 and 7), suggesting that RecF was not competing with Pol IV for the primer terminus (it does bind both ends of a gap (24)) and that RecO, which is presumably bound to the ssDNA template, does not inhibit polymerization. Even in the presence of all five proteins, SSB, RecA, and RecFOR, Pol IV extended most of the primer to full length (Fig. 3, lane 6). Thus, the presence of Rec-FOR does not result in the generation of a stable RecA-ssDNA filament under these conditions, where we would expect to observe inhibition (as in Fig. 3, lane 4). If RecFOR are displacing SSB from the DNA to load RecA, SSB binding to ssDNA, RecA-ssDNA filament formation and dissociation, and SSB competition with RecA for binding to ssDNA are thus likely to be quite dynamic.
Another assay for examining whether a stable RecA-ssDNA filament was forming was developed by Arad et al. (33). These authors showed that accessibility of the 3Ј end of the primer, as measured by the ability of exonuclease III (Exo III) to degrade it, was inhibited by RecA but favored by SSB. Using this assay (Fig.  4), RecA inhibited Exo III digestion of the primer (Fig. 4, compare lanes 1 and 2), as expected. SSB had little effect on its own (Fig. 4, compare lanes 1 and 3), as did the loading of ␤ (Fig. 4,  compare lanes 1 and 4), suggesting that Exo III could push the ␤ clamp back toward the 5Ј end of the primer as it digested the DNA. Exo III digestion was partially recovered when RecA and SSB were present together (Fig. 4, compare lanes 1, 2, 3, and 7), consistent with our conclusions above that SSB was sufficient to displace (or exclude) RecA from the ssDNA template. In the presence of RecA and SSB, the addition of different combinations of RecF, O, and R did not result in any significant increased inhibition of digestion, compared with the combination of RecA and SSB (Fig. 4, compare lane 7 to lanes 8 -14), confirming that under the conditions of our assay, RecFOR was not establishing a stable RecA-ssDNA filament.

RecFOR stimulation of Pol V elongation past template damage is bypassed by RecA730
Our data suggest that in the presence of SSB and RecA, Rec-FOR are not required to form PolV Mut that is capable of sub- Pol IV-catalyzed primer extension with the indicated proteins present was as described under "Experimental procedures" using the undamaged p/t. Analysis was as described above in the legend to Fig. 1.   Figure 4. Primer accessibility assay. The indicated combinations of proteins were used in primer accessibility assays as described under "Experimental procedures." Primer degraded (%), the fraction of primer degraded by Exo III in each assay calculated as 1 minus the fraction of total radioactivity in the intact primer band after incubation converted to a percentage.

PolV Mut and RecFOR
stantial DNA synthesis on undamaged templates, although the length distribution of DNA products increases in the presence of RecFOR. However, stimulation by RecFOR becomes apparent when Pol V faces the difficult task of elongating from the unpaired primer terminus opposite a site of template damage. We propose that the RecFOR requirement for the latter reflects the need for an increased level of RecA* required for Pol V Mut to maintain activity. Such a model would also account for the increase in DNA synthesis patch length on the undamaged templates and is consistent with the requirement for the presence of SSB to observe any RecFOR effect. If the model is correct, we would expect that the RecFOR requirement would be bypassed with a variant RecA that was capable of constitutively generating RecA*. Such a variant is RecA730.
RecA730 was derived from RecA441 (34), a recA mutant initially known as tif-1, which was constitutively induced for prophage induction at elevated temperature (35). RecA730 was shown, unlike WT RecA, to be able to displace SSB from ssDNA at low (1 mM) concentrations of magnesium and at all temperatures tested (36), an ability that was attributed to a more rapid association with ssDNA than the WT. Constitutive induction of the SOS response was therefore attributed to the assembly of sufficient RecA730 patches on ssDNA regions that are typically available in an unstressed cell.
The activity of WT RecA and RecA730 was compared on the THF p/t in the presence of RecOR or RecFOR or in their absence (Fig. 5). RecA730 displayed the same effect of increasing the length distribution of the DNA synthesis products as we observed previously with RecFOR and WT RecA (Fig. 5, A,  compare lanes 6 and 1 with lane 5, and B). RecA730 stimulated elongation past the damage in the absence of RecFOR to a greater extent than observed with WT RecA in the presence of RecFOR (Fig. 5A, compare lanes 6 and 1). The nearly identical behavior of RecA730 compared with WT RecA plus RecFOR indicates that the observed effects of RecFOR reported above relate to the necessity of maintaining a sufficient pool of RecA* that will add the necessary activated RecA monomer to Pol V to form Pol V Mut.

Suppression of LexA cleavage in the presence of RecA and SSB indicates limited RecA filament formation
Whereas our Pol V assays require the transfer of an activated monomer of RecA* from the RecA-ssDNA filament to PolV, to support our argument that stable RecA-ssDNA filaments were not forming, we used another assay that more directly assesses formation of RecA-ssDNA filaments. Cleavage of the SOS repressor LexA requires binding directly to the deep helical groove of the RecA-ssDNA filament (37,38). To measure LexA cleavage, we purified a modified version of LexA that was developed by John Little (39) in which a portion of the LexA N terminus was deleted and replaced with His and protein kinase A tags (⌬N-His-PKA-LexA (NHP-LexA)). NHP-LexA was shown to be cleaved in the presence of RecA-ssDNA filaments at the same rate as full-length, WT LexA (39).
We examined [ 32 P]NHP-LexA cleavage under conditions identical to those of the Pol V primer extension assay using either RecA or RecA730. RecA730-mediated NHP-LexA cleavage in the absence of SSB was ϳ50% greater than that in the presence of RecA (Fig. 6, compare lanes 3 and 7), similar to what has been reported previously (36), and, as expected, based on our data presented above, RecA-mediated cleavage of NHP-LexA was almost completely inhibited in the presence of SSB (Fig. 6, compare lanes 3 and 4), supporting our conclusion that in the presence of SSB and RecA, a stable RecA-ssDNA filament is not formed. Surprisingly, SSB also inhibited RecA730mediated LexA cleavage significantly, although there was, unlike with RecA, detectable residual cleavage (Fig. 6, compare  lanes 7 and 8). The presence of either RecOR or RecFOR led to a partial recovery of LexA cleavage in the presence of SSB with both RecA and RecA730 (Fig. 6, compare lanes 5 and 6 with  lane 4 and lanes 9 and 10 with lane 8).
Because, unlike with RecA, RecFOR had little effect on RecA730-supported Pol V activity in the presence of SSB (Fig.  5), we suggest that the difference in LexA cleavage mediated by RecA730 compared with RecA in the presence of SSB reports on the extent of RecA* generation required for maintenance of maximal Pol V activity. Thus, using two different assays, one We note our surprising result that RecA730-ssDNA filament formation was inhibited by SSB similar to RecA-ssDNA filament formation as being inconsistent with a previous report (40). Whereas we cannot explain the different results for certain, one possible explanation might lie in the fact that in the previous report (36) M13 phage DNA was used as the DNA effector. Whereas this DNA is often used as a ssDNA, it is, in fact, highly structured (41) with extensive double-stranded regions. Therefore it is possible that Lavery and Kowalczykowski (36) were reporting on results where both RecA-ssDNA and RecA-dsDNA filaments were forming. The latter can also mediate LexA cleavage (42).

Conclusions
As judged by the observations that RecA in the absence of SSB strongly inhibits Pol V DNA synthesis activity (Fig. 1) and that this inhibition is obviated when SSB is present, we argue that even in the presence of RecFOR, a stable, exclusively RecA-ssDNA filament does not form. This conclusion is strongly supported by similar results measuring LexA cleavage under the identical conditions (Fig. 6), an assay that more directly assesses RecA-ssDNA filament formation. This view is somewhat counter to the established view that at elevated magnesium concentrations RecA can displace SSB (43). Nevertheless, RecA* is clearly being formed in our primer extension assays because Pol V DNA synthesis is activated. We suggest that the more likely scenario is that formation and dissociation of the RecA-ssDNA filament in the presence of SSB is quite dynamic, presumably mediated by mixed SSB-RecA-ssDNA filaments. This equilibrium results in the generation of sufficient Pol V Mut for simple elongation of a primer terminus. However, more difficult synthesis, for example, extensive synthesis of long DNA products with the undamaged template and extension from a primer terminus opposite a site of template damage, requires either higher concentrations of or continuous production of RecA*. It is therefore in these scenarios that we find a significant stimulation of Pol V Mut activity by the RecA mediator proteins RecFOR. The underlying reason for what we have observed is that Pol V Mut processivity is likely to be very low, and the polymerase may require repeated attempts before it is successful in elongating from an unpaired primer terminus.
Pol V Mut has been shown to possess an intrinsic ATPase activity that governs binding of the polymerase to the p/t (13). Binding to the p/t requires the nucleotide cofactor, whereas its hydrolysis favors dissociation of the polymerase. Turnover of ATP, compared with that of the nonhydrolyzable analog ATP␥S is very rapid and very little DNA synthesis or TLS activity was evident in the presence of ATP. In fact, bypass of a THF in the presence of transactivated Pol V Mut required the presence of RecA*. These observations are consistent with what we have reported herein.
We therefore speculate that the inherent activity of Pol V Mut is likely to be very low, possibly even only one nucleotide incorporation event per association with one RecA monomer. Such an activity cycle would act to keep the patch size synthesized by Pol V quite small and would provide a feedback loop; in the case of SOS induced by UV irradiation, for example, ss gaps that accumulate because of the replisome skipping over the lesions are the likely source of RecA-ssDNA filament formation (19,21,44). As those gaps are filled in during repair, the level of RecA* available will decline rapidly, and Pol V Mut activity, if stiochiometrically dependent on association with an activated RecA monomer and ATP, will therefore also do the same.

PolV Mut and RecFOR
amide gel containing 7 M urea using 100 mM Tris borate (pH 8.3), 2 mM EDTA as the electrophoresis buffer. The 139-mer DNA template band was located by UV shadowing and eluted by crush and soak in 10 mM Tris-HCl (pH 8.0 at 25°C), 1 mM EDTA for 16 h at 37°C. The slurry was centrifuged through a Spin-X centrifuge tube filter (Corning), and the eluted DNA was recovered by ethanol precipitation. The polyacrylamide gel-purified primer PR5 (5Ј-GTG CGT TAG ACT CCT CAA TAC GAA GTA-3Ј) was 5Ј end-labeled with [ 32 P] using T4 kinase and [␥-32 P]ATP as above. Excess ATP was removed by purification through a Sephadex G-50 spin column (GE Healthcare), and the labeled primer was annealed to the template as above.

Primer extension assay
Standard reaction mixtures (5 l) contained 20 mM Tris-HCl (pH 7.5), 4% (v/v) glycerol, 8 mM DTT, 80 g/ml BSA, 3 mM ATP, 8 mM MgCl 2 , and 0.5 mM dNTPs. The standard reaction was carried out as follows: 100 nM SSB (tetramer), 50 nM ␤ (dimer), 5 nM DnaX cx ( 2 ␥␦␦Ј), 2 M RecA, and when indicated, 50 nM RecF, 0.5 M RecO, and 0.5 M RecR were first incubated with 2 nM p/t for 5 min at 37°C. DNA synthesis was initiated by the addition of 500 nM DNA Pol V, and the reac-tions were incubated for 8 min at 37°C. The reactions were terminated by adding 30 mM EDTA. The products were then purified in the presence of 10 ng of tRNA as carrier by phenol/ chloroform extraction followed by ethanol precipitation and resuspended in formamide loading buffer. After heat denaturation at 95°C for 5 min, the products were separated by electrophoresis through a 20% polyacrylamide gel (19:1 acrylamide to bisacrylamide) containing 7 M urea using 100 mM Tris borate (pH 8.3), 2 mM EDTA as the electrophoresis buffer. The gels were fixed by soaking in 10% methanol, 7% HOAc, 5% glycerol and dried. The dried gels were autoradiographed, and DNA products were quantified using phosphorimaging and Image-Gauge software (Fuji). In reactions with Pol IV, the polymerase was present at 20 nM, and the incubation was for 4 min.

Primer accessibility assay
Reaction conditions were as for the primer extension assay except that dNTPs and Pol V were omitted and E. coli Exo III (0.005 unit, NEB) was included. All other proteins were at the same concentrations as in the primer extension assay. The reactions were carried out at 37°C as follows. RecA and the DNA substrate were first incubated for 2 min. Next SSB, ␤, DnaX cx, and RecFOR proteins were added as indicated, and the reactions were further incubated for 4 min. Exo III was then added, and the incubation was continued for 4 min. The reaction products were recovered and analyzed as for the primer extension assay.

LexA cleavage assay
NHP-LexA (100 M) was labeled in a 20-l reaction mixture containing 1ϫ PKA buffer (NEB), 200 M ATP, 30 Ci of [␥-32 P]ATP, and 2500 units of protein kinase A (NEB) for 45 min at 30°C. The reaction was terminated by the addition of EDTA to 50 mM, and [ 32 P]NHP-LexA was separated from residual ATP by gel filtration through a Biogel P10 column (1 ml) equilibrated and developed in NHP-LexA storage buffer. The [ 32 P]NHP-LexA pool was stored at Ϫ20°C. LexA cleavage reaction mixtures were identical to primer extension reaction mixtures for Pol V and contained ␤, DnaX cx, p/t, 2 M NHP-LexA, ϳ5,000 Cerenkov cpm of [ 32 P]NHP-LexA and RecA, RecA730, SSB, and either RecOR or Rec FOR as indicated. The reactions were incubated for 13 min, mixed with an equal volume of 2ϫ SDS loading dye, and analyzed by 20% SDS-PAGE. The gels were fixed in 10% methanol, 7% HOAc, 5% glycerol; dried; and imaged by phosphorimaging and autoradiography. The gels were quantified using Fuji ImageGauge software.