Purification of a soluble UmuD'C complex from Escherichia coli. Cooperative binding of UmuD'C to single-stranded DNA.

The Escherichia coli UmuD' and UmuC proteins play essential roles in SOS-induced mutagenesis. Previous studies investigating the molecular mechanisms of mutagenesis have been hindered by the lack of availability of a soluble UmuC protein. We report the extensive purification of a soluble UmuD'C complex and its interactions with DNA. The molecular mass of the complex is estimated to be 70 kDa, suggesting that the complex consists of one UmuC (46 kDa) and two UmuD' (12 kDa) molecules. In contrast to its inability to bind to double-stranded DNA, UmuD'C binds cooperatively to single-stranded DNA as measured by agarose gel electrophoresis and confirmed by steady-state fluorescence depolarization. A Hill coefficient, n = 3, characterizes the binding of UmuD'C to M13 DNA and to a 600 nucleotide DNA oligomer, suggesting that at least three protein complexes may interact cooperatively when binding to DNA. The apparent equilibrium binding constant of UmuD'C to single-stranded DNA is approximately 300 nM. Binding of the complex to a short, 80 nucleotide, DNA oligonucleotide was detectable by fluorescence depolarization, but it did not appear to be cooperative. Binding of UmuD'C to single-stranded M13 DNA causes an acceleration of the protein-DNA complex, suggesting that the longer DNA may undergo compaction. The UmuD'C complex associates with RecA-coated DNA, and the UmuD'C complex remains bound to DNA in the presence of RecA.

The Escherichia coli UmuD and UmuC proteins play essential roles in SOS-induced mutagenesis. Previous studies investigating the molecular mechanisms of mutagenesis have been hindered by the lack of availability of a soluble UmuC protein. We report the extensive purification of a soluble UmuDC complex and its interactions with DNA. The molecular mass of the complex is estimated to be 70 kDa, suggesting that the complex consists of one UmuC (46 kDa) and two UmuD (12 kDa) molecules. In contrast to its inability to bind to doublestranded DNA, UmuDC binds cooperatively to singlestranded DNA as measured by agarose gel electrophoresis and confirmed by steady-state fluorescence depolarization. A Hill coefficient, n ‫؍‬ 3, characterizes the binding of UmuDC to M13 DNA and to a 600 nucleotide DNA oligomer, suggesting that at least three protein complexes may interact cooperatively when binding to DNA. The apparent equilibrium binding constant of UmuDC to single-stranded DNA is approximately 300 nM. Binding of the complex to a short, 80 nucleotide, DNA oligonucleotide was detectable by fluorescence depolarization, but it did not appear to be cooperative. Binding of UmuDC to single-stranded M13 DNA causes an acceleration of the protein-DNA complex, suggesting that the longer DNA may undergo compaction. The UmuDC complex associates with RecA-coated DNA, and the UmuDC complex remains bound to DNA in the presence of RecA.
The SOS regulon present in Escherichia coli is triggered as part of the cell's response to exogenous DNA damage (1,2). As a consequence of attempting to replicate past DNA lesions, an inducing signal is generated that results in the activation of RecA protein (RecA*) 1 (3). In its activated state, RecA* enhances the self-cleavage of the LexA transcriptional repressor (4), thus leading to the expression of more than 20 proteins that help the cell to avoid the lethal effects of DNA damage. Recent reviews of the SOS response are contained in Refs. 5 and 6.
An important, yet poorly understood, feature of the SOS response is error-prone repair or "SOS-induced" mutagenesis.
The UmuD and UmuC proteins are essential participants in this process (7,8), and the absence of either protein reduces mutagenesis by more than 100-fold to spontaneous background levels (2,9). The UmuDC operon is induced following LexA cleavage, and the process is further regulated by the need for the UmuD protein to undergo a RecA*-mediated cleavage. This event is mechanistically similar to LexA autodigestion, but in the case of UmuD (and its homologs), cleavage leads to activation of its mutagenesis function(s) (10 -12). Activated UmuDЈ forms homodimers and associates with UmuC to form a UmuDЈC complex (13). Both UmuDЈ and UmuC are believed to interact with RecA* (14,15) in such a way as to target the relatively small number of mutagenically active UmuDЈC molecules to lesions within DNA (15)(16)(17).
Replicative bypass of the RecA*-UmuDЈC-coated lesion is most likely performed by pol III holoenzyme (18,19). Although pol II is induced in response to DNA damage as part of the SOS regulon (20,21), its role in either lesion bypass or DNA repair has yet to be firmly established. Recent data suggest that pol II may be required for the bypass of abasic lesions provided that heat shock proteins are not induced (22). A two-step model describing error-prone synthesis was proposed by Bridges and Woodgate (23,24) in which pol III incorporates a nucleotide opposite a template lesion but cannot continue synthesis. Generation of a UmuDЈC-RecA* "mutasome" would enable the stalled pol III molecule to continue synthesis past the lesion (5,13).
The ability of E. coli to facilitate error-prone translesion synthesis depends upon the cellular levels of the Umu proteins. Under normal conditions the proteins are expressed at low levels (25). Genetic experiments in which the in vivo levels of the UmuDЈC proteins have been artificially manipulated leads to a variety of phenotypes. Overproduction of the Umu proteins in fully SOS-induced cells can cause a cold-sensitive phenotype that is associated with the rapid cessation of DNA replication (26). Expression of excess UmuC results in a nonmutable phenotype similar to what is seen when there is a deficiency of UmuC (27). Modest overproduction of UmuDЈC appears to inhibit recombinational functions of RecA and promotes a switch from error-free recombinational repair pathways to those that are error prone (5,28).
Testing of these models has been severely hampered by the lack of reconstituted "error-prone" and "error-free" lesion bypass assays. While pol III holoenzyme, RecA, and UmuDЈ have been extensively purified and characterized, UmuC has been more difficult to study because of its apparent insolubility and its lack of defined enzymatic activity. Previous studies of UmuC and UmuDЈC complexes were performed with UmuC that had been purified from a denatured form and renatured in the presence of chaperone proteins (13,29). In this paper we describe the first purification of a soluble, intact UmuDЈC complex and characterize its interactions with DNA.
Cells were harvested and resuspended in equal weight/volume (gm/ ml) amount of storage buffer (50 mM Tris-HCl (pH 7.5), 10% sucrose, 10 mM EDTA). The cells were then quickly frozen by dropwise addition of cell suspension into liquid nitrogen and stored at Ϫ70°C. Cells were thawed at 4°C in 2.5 volumes of lysis buffer (50 mM Tris-HCl (pH 7.5), 10% sucrose, 0.1 M NaCl). Once thawed, lysozyme was added to a final concentration of 0.4 mg/ml. The cell slurry was incubated for 1 h at 4°C followed by a 4-min incubation at 37°C and centrifuged at 11,800 rpm for 1 h in a Sorvall GSA rotor.
A solution of 10% polyethyleneimine HCl (pH 7.6) was added to lysate supernatant in 10-ml portions to a final concentration of 1.1%. The solution was stirred for 15 min after each addition. At the end of the last addition, the suspension was kept on ice for 20 min, and the precipitate was collected by centrifugation for 10 min at 9,000 rpm in a Sorvall SS34 rotor. Proteins were extracted by stirring the pellet for 15 min in 50 ml of R buffer (20 mM Tris-HCl (pH 7.5), 0.1 mM EDTA, 1 mM dithiothreitol, 10% (v/v) glycerol) containing 1 M NaCl. Following centrifugation (10 min at 6,000 rpm) the supernatant was collected and the extraction repeated until all proteins were eluted from the pellet.
Supernatants were combined and dialyzed overnight against R buffer containing 50 mM NaCl (from this point on the concentration of glycerol was increased to 20% glycerol (v/v)). The dialyzed sample was loaded onto a 200-ml DEAE-Sephacel (Pharmacia) column (3 ϫ 15 cm) at a flow rate of 1 ml/min. The column was washed with 2 column volumes of R buffer containing 50 mM NaCl, and protein was eluted from the column with three column volumes of R buffer containing 500 mM NaCl.
Fractions containing high concentrations of UmuDЈC were combined, dialyzed overnight against R buffer containing 250 mM NaCl, loaded onto a 150 -170-ml phosphocellulose (Whatman) column (3 ϫ 24 cm), and extensively washed. UmuC was eluted from the column with 2 volumes of R buffer containing 1 M NaCl. Fractions containing protein were combined and concentrated to 6 ml using Centriplus-30 concentrators (Amicon), and the total protein was loaded onto a 600-ml Superdex-75 (Pharmacia) column (2 ϫ 160 cm) equilibrated with R buffer containing 1 M NaCl. The column was run at a flow rate of 0.25 ml/min, and 2.0-ml fractions were collected. Fractions highly enriched for UmuDЈC were combined and concentrated using Centriplus-30 concentrators to final concentration of at least 1 mg/ml. These fractions were divided into aliquots (100 l) and quickly frozen at Ϫ70°C. A final yield of 2-4 mg of UmuDЈC complex was obtained from 30 liters of cells.
Microsequencing-The purified UmuDЈC was loaded in duplicate on a 10% SDS-polyacrylamide gel and transferred onto a ProBlot membrane (Applied Biosystems).
Microsequencing was performed as suggested (31) on one sample, while the other was processed for Western blot analysis with UmuC antiserum. Amino-terminal analysis for putative UmuC was performed by Lynn Williams (USC Microchemical Facilities) and found to be: NH 2 -MFALXDVNAFYASXE (X represents an unidentified residue), in agreement with the reported DNA sequence for UmuC (32,33).
DNA Gel Mobility Shift Assay-The UmuDЈC complex was incubated with either ssM13mp18 (NEB) or ss600-mer (HinfI restriction digest product of multiprimed ssM13mp18) in binding buffer (30 l of 20 mM Tris-HCl (pH 7.5), 8 mM magnesium acetate, 4% glycerol) for 20 min at 30°C. In all of the titration experiments, the salt concentration of the reaction was held at around 50 mM NaCl. All protein dilutions were done in 20 mM Tris-HCl (pH 7.5), 40 g/ml bovine serum albumin, 5 mM dithiothreitol, 4% glycerol. An aliquot of each binding reaction (20 l) was loaded onto 0.9% agarose gel and separated in LTAE buffer (6.7 mMTris-HCl, 3.3 mM sodium acetate, 1 mM EDTA, (pH 7.5)) with buffer recirculation (29). Upon completion of electrophoresis the gel was soaked in transfer buffer (25 mM Tris base, 190 mM glycine, 0.04% SDS, (pH 8.5)) and electrotransferred to nitrocellulose membrane. Protein blots were incubated with rabbit polyclonal UmuC or UmuDЈ/D antiserum followed by incubation with goat anti-rabbit IgG peroxidase conjugate (Pierce) and visualized by the ECL detection system (Amersham Corp.) according to manufacturer's directions. Properly exposed film (Hyperfilm ECL, Amersham) was subjected to densitometric analysis using a Hoefer scanning densitometer GS300. Different exposures were used to ensure that band intensities were within the linear range of the film.
Molecular Weight Determination by Gel Filtration-Gel filtration analysis was performed using an analytical Sephadex G-100 column equilibrated in R buffer ϩ 1.0 M NaCl. Pharmacia LMW (low molecular weight) calibration kit was used for protein standards. K av was calcu- where V e is elution volume for the protein, V t is total bed volume, and V 0 is column void volume or elution volume for blue dextran 2000. V 0 for the 86-ml column was 35 ml. ECL immunodetection was used to locate UmuC in the UmuC-only column fractions.
Etheno-M13 DNA/Rhodamine-X 80-Mer-Single-stranded M13mp18 DNA was purified by the hexadecyltrimethylammonium bromide method (34), extracted with phenol/chloroform twice, and ethanol-precipitated. The protocol used to make etheno-M13 DNA was based on methods described previously (35,36). The reaction was as follows; 400 g of ssM13mp18 DNA was reacted in 2 ml of 20 mM potassium phosphate (pH 7.0) and 0.2 M chloroacetaldehyde (Aldrich) at 37°C for 48 h. The slightly yellow product was dialyzed against a buffer containing 20 mM Tris-HCl (pH 7.5), 1 mM EDTA overnight at 4°C followed by dialysis at 50°C for 7 h to bring derivatized bases to maturation. The final product had an absorption maximum at 264 nm (broad maximum 250 -267 nm) and an absorption minimum at 244 nm with A 270 /A 260 ϭ 1.03 (starting material absorption minimum was 230 nm and A 270 /A 260 ϭ 0.84 (37)). The concentration of etheno-M13 DNA was determined using an extinction coefficient A 260 ϭ 7000 M Ϫ1 cm Ϫ1 , as described previously (37,38). Rhodamine-X (Molecular Probes) labeling of the 80-mer oligonucleotide was carried out as described (39).
Fluorescence Intensity and Anisotropy Measurements-All fluorescence intensity and anisotropy experiments were performed with a PTI Quantamaster model QM-1 fluorimeter. Reactions were performed in an 180-l ultra-microcuvette (Hellma) at 21°C. Reaction buffer used in all titrations was 20 mM Tris-HCl (pH 7.9), 8 mM magnesium acetate, 4% glycerol. Proteins were allowed to bind for 1 min before measurements were taken. For both measurements, the excitation wavelength was selected to be 315 nm and emission wavelength was monitored at 410 nm with a 6 nm bandpass. The excitation wavelength was selected for lowest interference from intrinsic protein fluorescence and highest signal emission by etheno-M13 DNA. Intensity correction calculations were made by subtracting UmuDЈC complex titration on unmodified ssM13 DNA from intensity measurements of UmuDЈC complex titration on etheno-M13 DNA. Anisotropy was determined by using a vertically oriented polarizer for the incident beam and a vertical and 90°r otation of the emission polarizer (dichroic sheet polarizers, Oriel). Three anisotropy measurements of 10-s duration were averaged for each titration point. Contaminating fluorescence signal from buffer was subtracted from all measurements. The fluorescence anisotropy is r ϭ f f r f ϩ f b r b ; r f and r b refer to anisotropies of free and bound fluorophores, and f f and f b are the fraction of total fluorescence due to each form of the fluorophore. The fraction of etheno-DNA bound by UmuDЈC complex the ratio of the quantum yield of the bound and free fluorophore). This equation corrects for the decrease in quantum yield of ethenoadenine when UmuDЈC binds DNA (40).
SDS-Polyacrylamide Gel Electrophoresis-Proteins were separated in a SDS-15% polyacrylamide gel according to Laemmli (41). Proteins were visualized following a 20-min staining with a 10% solution of Coomassie Brilliant Blue R-250 in 20% methanol, 10% acetic acid and destaining in a solution of 10% methanol, 10% acetic acid. Fractionated proteins were transferred to nitrocellulose, probed with antiserum against UmuDЈ,D or UmuC and visualized using an ECL detection system as described above. Polyclonal antibodies were raised in rabbits to affinity purified UmuC protein (27) and UmuD/DЈ proteins (13) by Hazelton Laboratories (Washington, D. C.). In all immunodetection experiments, antiserum was used at dilution of 1:20,000.
Protein Determination-Protein concentrations were measured according to the Bradford assay (Bio-Rad protein assay kit) using bovine plasma ␥-globulin (protein assay standard I, Bio-Rad) as a standard.

RESULTS
Previous attempts to purify UmuC involved extracting insoluble UmuC protein from inclusion bodies, solubilizing in the presence of 8 M urea and subsequently refolding of the purified polypeptide using either S9 ribosomal protein (13) or by sequential incubation with Hsp70 and Hsp60 heat shock chaperones (29). The effects of denaturation and renaturation on enzymatic and biochemical properties of UmuC are not known. Therefore, a major goal of this study was to purify and characterize a complex containing UmuC and UmuDЈ (UmuDЈC complex) while maintaining the complex in a soluble form throughout purification.
Purification of a UmuDЈC Complex-The UmuDЈC proteins were overexpressed and purified from an E. coli K-12 strain (RW82FЈIQ) (30) carrying a chromosomal deletion of the umuDC operon (42). Overproduction of UmuDЈC in its native form was accomplished by placing the operon on a plasmid under ptac control. Typical yields of UmuDЈC from a 30-liter culture (150 g wet cell weight) were 2-4 mg. We estimate that the recovery of soluble UmuC is 5-10-fold greater in cell extracts made from strain RW82FЈIQ compared with the strain (MC1000 containing pRK248(cIts) and pSB13) originally described for UmuC overproduction (11). Most of the soluble material obtained from crude cell lysates that cross-reacts with anti-UmuC antiserum was comprised of a tightly bound protein complex containing UmuC and UmuDЈ (i.e. UmuDЈC complex, see Fig. 1). RecA was not present in the complex based on the absence of detectable cross-reactivity to anti-RecA antibody.
In the initial purification step, nucleic acids were precipitated by addition of polyethyleneimine. UmuC and UmuDЈ coprecipitated with the nucleic acid and were extracted in the presence of 1 M salt and dialyzed into 50 mM NaCl (see "Experimental Procedures"). The remaining purification steps involved separation on DEAE-Sephacel, phosphocellulose, and Superdex-75 columns (see "Experimental Procedures"). UmuDЈC was present in 1 M NaCl following fractionation on Superdex-75 and attempts to reduce the salt concentration subsequent to gel filtration, e.g. by dialysis, were hampered by the formation of an insoluble protein precipitate containing both UmuC and UmuDЈ.
The UmuDЈC fractions from each stage of purification were fractionated by SDS-polyacrylamide gel electrophoresis and analyzed by Western immunodetection (Fig. 1). The UmuDЈC complex appeared to be greater than 80% pure (based on integration of densitometric band intensities in the linear range) following fractionation with Superdex 75 (Fig. 1A, lane 5).
Microsequencing was used to confirm the true identity of the soluble UmuC protein (see "Experimental Procedures"): 15 residues identified from the NH 2 terminus coincided with the published UmuC amino acid sequence (32,33).
Antibodies to both UmuDЈ and UmuC showed that at least a fraction of the overproduced UmuDЈ protein copurified with UmuC (Fig. 1B). Gel filtration of the phosphocellulose pool on a Superdex-75 column in 1 M NaCl failed to separate UmuDЈ from UmuC and analytical gel filtration using a Sephadex G100 column in the presence of 1 M NaCl, suggested that the proteins associated in a complex with a molecular weight of approximately 70 kDa (Fig. 2). This molecular mass is consistent with a composition of one UmuC (46 kDa) and two UmuDЈ (12 kDa) molecules (13). The elution volumes of the protein standards (Fig. 2) were determined by running a mixture of the standards alone and in the presence of the Superdex 75 fraction. It was possible to purify a small amount of soluble UmuC from RW82FЈIQ harboring a plasmid that overproduced UmuC in the absence of UmuDЈ. Elution of UmuC on an analytical Sephadex G100 column gave a value of K av slightly larger than that of ovalbumin (43 kDa). The apparent molecular weight for UmuC in the range of 45-47 kDa is in agreement with a previous determination (13).
The properties of UmuC and the UmuDЈC complex were investigated using a low ionic strength native agarose gel. Proteins were resolved by gel electrophoresis, transferred to nitrocellulose, and incubated with UmuC antiserum. UmuC is a basic protein having a calculated isoelectric point of about 9.6 based on amino acid composition. Therefore, UmuC should migrate toward the cathode in an agarose gel. When UmuC and two highly acidic UmuDЈ molecules (pI ϳ 4.5) combine to form a UmuDЈC complex, migration should be toward the anode. A fraction of soluble UmuC, concentrated after purification through a phosphocellulose column (see "Experimental Procedures"), migrated toward the cathode, but only after treatment with RNase (DNase-free) (Fig. 3, lane 3). An untreated sample of UmuC or one treated with DNase (RNase-free) remained in the wells (lanes 1 and 2, respectively). It appears that in the absence of UmuDЈ, and especially when UmuC was extracted from a crude cell lysate at low salt concentration (ϳ50 mM NaCl), it was tightly bound to RNA. The UmuDЈC complex migrated toward the anode as expected (Fig. 3, lane 4), and this pattern was essentially unaffected by treatment with DNase or RNase (Fig. 3, lanes 5 and 6, respectively).
Cooperative Binding of UmuDЈC to ssDNA-Single-stranded M13 DNA and a 600-nucleotide ssDNA product, prepared from a restriction digest of partially double-stranded M13 DNA (see "Experimental Procedures"), were used to measure the binding of UmuDЈC. UmuDЈC was incubated at varying ratios with ssDNA, and reactions were analyzed by agarose gel electrophoresis followed by Western blotting with anti-UmuC and anti-UmuDЈ antiserum. In separate experiments staining of the gels with ethidium bromide prior to electroblotting was used to show that the position of the bound DNA corresponded to the position of the bound UmuDЈC, i.e. UmuDЈC incubated in the presence of ssDNA migrated as a protein-DNA complex.
The results indicate that UmuDЈC bound cooperatively to a single-stranded 600 mer (Fig. 4). No detectable protein was bound at low UmuDЈC/DNA ratios (Fig. 4A, lanes 1-3) while in the "steep binding" range, binding increased from about 1 UmuDЈC bound per 300 nucleotides to about 1 per 50 nucleotides. Saturation of the 600-mer occurred when approximately one UmuDЈC was bound per 20 nucleotides. Binding of UmuDЈC to the ss600-mer resulted in measurable retardation as a function of increasing UmuDЈC/DNA ratios. To quantitate the binding of UmuDЈC to the ss600-mer, gel band intensities corresponding to free UmuDЈC and UmuDЈC complexed to DNA were integrated in the linear region of the film (see "Experimental Procedures"). A logarithmic Hill plot was fit by a straight line with a slope corresponding to a Hill coefficient n ϭ 3 (Fig. 4B). We estimated a value of K D ϭ 260 nM for the apparent binding constant of the UmuDЈC complex to the 600-mer.
The qualitative binding profile of UmuDЈC to M13 DNA was similar to that for the 600 mer. There was no detectable binding of UmuDЈC to the DNA at low UmuDЈC/DNA ratios (Fig. 5,  lanes 1-5). Saturation of the M13 DNA (7250 nt) was difficult to quantitate in the gel binding study, caused, in part, by an accelerated migration of the UmuDЈC-M13 DNA complex; the maximum clear attainable binding was roughly 1 UmuDЈC per 100 nucleotides occurring at an input level of about 1 UmuDЈC per 40 nucleotides. Since the binding measurements were carried out using salt concentrations near 50 mM NaCl, the high salt concentration of the UmuDЈC sample coupled with the requirement to maintain a low reaction volume limited the amount of of UmuDЈC that could be used in attempting to saturate the M13 DNA.
A more sensitive measurement for the binding of UmuDЈC to M13 DNA was performed using fluorescence depolarization, see e.g. Refs. 40 and 43. In this assay, the protein was incubated in the presence of chemically modified M13 DNA containing a random distribution of the fluorescent base ethenoadenine and to a lesser extent ethenocytosine (see "Experimental Procedures"). Protein-DNA binding was observed as an increase in the steady-state anisotropy of the protein-DNA complex with increasing amounts of UmuDЈC. Much more protein can be used in the fluorescence assay because of the increased reaction volume (180 l) compared with the gel assay (30 l). We were able to observe a saturation in the fraction of UmuDЈC bound to M13 DNA (Fig. 6). The data fit by a Hill plot gives an apparent binding constant in the range of 300 -350 nM and a slope of 3, similar to the Hill coefficient determined for the 600-mer. A plot for the increase in fluorescence intensity with increasing UmuDЈC overlapped with the anisotropy plot offering independent evidence for the cooperative nature of the binding of UmuDЈC to DNA (data not shown). In contrast to the retardation observed when UmuDЈC was bound to a 600-mer, binding of UmuDЈC to M13 DNA appeared to accelerate migration of the protein-DNA complex on an agarose gel at high UmuDЈC/DNA nucleotide ratios (Fig.  7) suggesting that the longer DNA may undergo compaction.
We observed no detectable binding of UmuDЈC to ssDNA less than 600 nucleotides long in the agarose gel assay. However, using fluorescence depolarization we observed binding of UmuDЈC to an 80-mer, and in contrast to the binding of UmuDЈC to either the 600-mer or to M13 DNA, binding to the 80-mer did not appear to be cooperative (data not shown). The K D for binding the 80-mer was near 150 nM, and ATP was not required for protein-DNA binding. We were unable to detect any significant difference in the binding of UmuDЈC to either UV-irradiated or unirradiated M13 single-stranded DNA using the agarose gel assay, and binding of UmuDЈC to irradiated DNA did not protect against degradation by T4 endonuclease V (data not shown). No detectable binding of UmuDЈC was observed either to UV-irradiated or unirradiated double-stranded DNA (data not shown).
Binding of UmuDЈC to ssDNA in the Presence of RecA*-Two experiments were performed to visualize the binding of UmuDЈC to M13 ss DNA in the presence of activated RecA protein: preincubation of RecA with DNA followed by the addition of UmuDЈC (Fig. 8A) and preincubation of UmuDЈC with DNA followed by addition of varying levels of RecA* (Fig. 8B). An agarose gel was used to separate the UmuDЈC-ssDNA (Complex I), UmuDЈC-RecA*-ssDNA (Complex II) and free UmuDЈC. Complexes containing UmuDЈC were visualized using UmuC antiserum, and bound and free ssDNA were located by staining with ethidium bromide.
The UmuDЈC-M13 DNA complex formed in the absence of RecA* caused an acceleration of the DNA (Fig. 8A, lane 1,  Complex I). Under these conditions, the amount of UmuDЈC complex in the reaction was limiting as all of it appeared to be bound to the ssDNA. Addition of excess UmuDЈ to the reaction gave an essentially similar result, although there was a slight increase in the amount of detectable free UmuDЈC complex (Fig. 8A, lane 2). In contrast, however, when RecA was preincubated with DNA (in the presence of ATP␥S) to form a RecA*nucleoprotein filament and UmuDЈC subsequently added to the reaction, a retarded complex was observed (Complex II) that is consistent with it being a UmuDЈC-RecA*-DNA complex (Fig.  8A, lane 3). Further addition of excess UmuDЈ to the reaction had no measurable effect on the formation of complex II (Fig.  8A, lane 4). It is interesting to note that under these conditions, where a significant fraction of the ssDNA would be expected to be bound by RecA*, there was a significant increase in the amount of free UmuDЈC complex (Fig. 8A, lanes 3 and 4). This result suggests that UmuDЈC is unable to compete RecA from ssDNA once a RecA*-nucleoprotein filament has formed.
Preincubation of UmuDЈC with DNA followed by the addi-tion of three different levels of RecA produced a series of complexes that were consistent with UmuDЈC-DNA complexes associated with variable amounts of RecA protein (Fig. 8B, lanes  2-4). Since UmuDЈC remained bound to the DNA following the addition of RecA, and UmuDЈC was limiting in the reactions, it is likely that RecA does not compete off the preassembled UmuDЈC-DNA complex but simply binds regions free from UmuDЈC. The complex formed by preincubating DNA with RecA (Fig. 8A, lanes 3 and 4) migrated more slowly than the complex II formed by preincubation with UmuDЈC (Fig. 8B,  lanes 2-4), suggesting that prior binding of UmuDЈC to DNA limited the ability of RecA to form a nucleoprotein filament. DISCUSSION Previously, Echols and co-workers succeeded in purifying denatured UmuC (13); upon renaturation it was shown that UmuC bound ssDNA (29). Although these results represented an important initial step for reconstituting lesion bypass in vitro (44), it is necessary to isolate and purify UmuC alone or complexed with UmuDЈ in soluble form to investigate the full range of biological interactions of these proteins.
We coexpressed the UmuDЈC proteins from a ptac promoter and purified the UmuC protein from a strain carrying a deletion of the entire chromosomal umuDC operon. Under these conditions a sizable fraction of a 46-kDa protein was soluble, and this protein was shown to be UmuC by antibody binding and microsequencing.
UmuC remained soluble throughout each purification step and copurified as a tightly bound complex with UmuDЈ (Fig. 1). The UmuDЈC complex with an estimated molecular mass of 70 kDa remained intact during chromatography on Superdex 75 in the presence of 1 M NaCl. These results suggest that this complex is composed of a UmuC monomer (46 kDa) and UmuDЈ dimer (24 kDa), which are tightly associated.
It was also possible to purify a small amount of soluble UmuC by overproducing UmuC alone in the absence of UmuDЈ. Free UmuC has an apparent molecular mass of 46 kDa based on gel filtration (Fig. 2), a finding suggesting that it exists as a monomer when in solution (13). UmuC is a basic protein and should therefore migrate in an electric field toward the cathode in a neutral agarose gel. However, the UmuC fraction obtained from the phosphocellulose column remained in the wells during electrophoresis unless it was first treated with RNaseA (Fig. 3). Following a pretreatment of the phosphocellulose fraction with RNaseA, UmuC migrated toward the cathode (Fig. 3, lane 3), suggesting that free UmuC is bound to RNA in crude cell lysates. UmuC can be released from RNA in the presence of 1 M NaCl (Fig. 2), and this property can be exploited in the future to purify substantially larger quantities of soluble UmuC. Our data suggest that UmuC binds preferentially to UmuDЈ precluding formation of significant amounts of the UmuC-RNA complex. Pretreatment of the UmuDЈC complex with RNase had negligible effect on its migration toward the anode in a neutral agarose gel (Fig. 3, compare lanes 5 and 6).
The binding studies with ssDNA offer potentially important insights into the properties of the UmuDЈC complex. Binding of the protein complex, either to a 600-mer fragment of M13 DNA (Fig. 4), or to full-length M13 DNA (Fig. 5), occurred cooperatively with a Hill coefficient of 3. Such cooperativity was observed using two independent measurements: (i) agarose gel electrophoresis and immunodetection of free UmuDЈC or UmuDЈC complexed with DNA (Figs. 4 and 5) and (ii) steadystate fluorescence depolarization of ethenoadenine-labeled M13 DNA at increasing concentration of UmuDЈC (Fig. 6). The depolarization data were also confirmed by an independent measurement of an increase in the fluorescence of ethenoadenine as a function of increasing UmuDЈC concentration (data not shown). As reported previously (29), in the gel assay UmuDЈC did not bind irradiated or unirradiated doublestranded DNA.
Fluorescence properties of etheno-modified DNA have been used successfully to calculate DNA binding parameters of RecA protein (37,38,45) and bacteriophage T4-coded Gene 32 protein (46). Since parameters for most activities attributed to these proteins have been confirmed by other methods, it is believed that higher affinity of binding to etheno-modified DNA over natural ssDNA for RecA and gp32 is caused by a disruption of secondary structure of ssDNA. In our experiments, using two independent techniques, we found that UmuDЈC bound cooperatively to both ethenoadenine-modified ssDNA (Fig. 6) and to unmodified ssDNA ( Fig. 4) with similar values for the Hill coefficient.
In addition to its ability to interact with free ss DNA, UmuDЈC interacts with ssDNA that had been pre-coated with RecA (Fig. 8A). These two properties, direct cooperative binding of UmuDЈC to DNA, and binding to a RecA nucleoprotein filament, may help to explain some of the intriguing phenotypes exhibited when the Umu proteins are overexpressed in vivo (26 -28). Under physiological conditions, the UmuDЈC proteins are maintained at low basal levels (25). Since UmuDЈ is normally expressed at about 12-fold higher levels than UmuC (25), most, if not all, of the UmuC is likely to exist as a UmuDЈC complex. Thus, even in fully SOS-induced cells there may be only about 200 UmuDЈC molecules per cell (25).
In contrast, under similar conditions, there are likely to be 30,000 -50,000 molecules of RecA protein/cell (3,25). Thus, under normal cellular conditions, UmuDЈC may associate with a RecA nucleoprotein filament, perhaps through an interaction between UmuDЈ-RecA* (15) or UmuC-RecA* (14). When a RecA* nucleoprotein filament has formed at a lesion in doublestranded DNA (47), a UmuDЈC-RecA* interaction would provide an efficient way to target Umu proteins to locations within the cell where they are required to facilitate error-prone translesion DNA synthesis (15).
Under conditions where UmuDЈC is moderately overproduced, e.g. when expressed from an operator constitutive mutant on a low copy number plasmid, some UmuDЈC might begin to bind directly to regions of ssDNA and inhibit the extent of RecA*-nucleoprotein filament formation. This interaction might be sufficient to inhibit RecA's recombinatorial activities directly (28) while allowing for the formation of a lesion-localized nucleoprotein structure (a putative mutasome) containing RecA, UmuDЈC and pol III (5,13) or perhaps pol II (20,22).
The nonmutable phenotype of cells that slightly overexpress UmuC alone (27) may result from increased formation of UmuDЈC-DNA complexes at regions of ssDNA generated when a cell attempts to replicate damaged DNA, which may hinder continued replication rather than enhance it and ultimately lead to killing of cells with premutagenic lesions, thereby reducing the number of mutants. Significant overproduction of UmuDЈC, e.g. from a multicopy plasmid in fully SOS-induced cells, may result in levels of UmuDЈC that can bind directly and cooperatively to regions of ssDNA, even in undamaged cells. Encountering UmuDЈC coated DNA rather than E. coli singlestranded binding protein-coated DNA may impede replication and thus result in the rapid cessation of DNA synthesis that is associated with cold sensitivity (26).
The nature of the interactions between pol III or pol II and the SOS-induced proteins described above, that lead to either error-free or error-prone lesion bypass, remain entirely speculative. Our aim is to elucidate the precise interactions between polymerase holoenzymes, RecA*, UmuDЈC, and any additional proteins that facilitate bypass of DNA template lesions by reconstituting an SOS lesion bypass system in vitro. The availability of an extensively purified source of soluble UmuDЈC complex is therefore a critical step toward achieving this goal.