Escherichia coli DNA Polymerase V Subunit Exchange

DNA polymerase V consisting of a heterotrimer composed of one molecule of UmuC and two molecules of UmuD′ (UmuD′2C) is responsible for SOS damage-induced mutagenesis in Escherichia coli. Here we show that although the UmuD′2C complex remains intact through multiple chromatographic steps, excess UmuD, the precursor to UmuD′, displaces UmuD′ from UmuD′2C by forming a UmuDD′ heterodimer, while UmuC concomitantly aggregates as an insoluble precipitate. Although soluble UmuD′2C is readily detected when the two genes are co-transcribed and translated in vitro, soluble UmuD2C or UmuDD′C are not detected. The subunit exchange between UmuD′2C and UmuD offers a biological means to inactivate error-prone polymerase V following translesion synthesis, thus preventing mutations from occurring on undamaged DNA.

The Escherichia coli SOS response was first described in the mid-1970s (1,2) and is now known to involve the induction of Ͼ40 genes under control of the LexA repressor protein (3). Although many of the genes in the SOS regulon are involved in DNA damage repair and cell division (4), two "UV mutagenesis" genes, umuC and umuD (5)(6)(7), are required to observe either UV-or chemically induced mutations elevated typically ϳ100-fold above spontaneous background levels (4). SOS-induced mutations on the bacterial chromosome result primarily from the error-prone replication of damaged DNA templates by pol V, 1 a heterotrimer composed of one UmuC molecule bound to two molecules of UmuDЈ (UmuDЈ 2 C) (8,9).
Based upon the inherent error-prone nature of TLS, it is perhaps not too surprising that the activity of the Umu proteins is regulated at transcriptional and post-translational levels (10). The umu genes are arranged in an operon and are negatively regulated at the transcriptional level by the SOS repressor LexA (7). Despite the fact that the umu operon is one of the tightest regulated in the LexA regulon (11), transcriptional regulation is incomplete, and the cellular levels of UmuD and UmuC are additionally kept to a minimum through their rapid Lon-mediated proteolytic degradation (12,13). After cellular DNA damage, RecA nucleates on regions of single-stranded DNA and mediates the post-translational self-cleavage of LexA leading to its inactivation and derepression of genes in the LexA regulon, including umuD and umuC (4). UmuD undergoes a mechanistically similar cleavage reaction (14,15). In the case of UmuD, cleavage of its N-terminal 24 amino acids converts it to UmuDЈ and activates it for its role in SOS mutagenesis (16). UmuD and UmuDЈ both form homodimers but when mixed together associate preferentially as a UmuDDЈ heterodimer (17), and UmuDЈ becomes susceptible to proteolysis by the ClpXP serine protease (12,18). Degradation of the mutagenically active UmuDЈ subunit helps return cells to a resting state once cellular DNA damage has been repaired and the need for pol V has abated.
In vitro replication assays reveal that regulation of the catalytic activity of pol V is also complex and is modulated through multiple protein-protein interactions. For example, pol V does not catalyze TLS alone but is, instead, an essential component of a multiprotein "mutasome" complex composed of RecA protein, ␤ sliding-clamp, and single-strand binding protein (19,20). When taken in conjunction with the observation that pol V is, for all intents and purposes, catalytically inactive in the absence of UmuDЈ or RecA (21,22), previous studies have understandably focused almost exclusively on the biochemical properties of pol V when working as part of a mutasomal complex during TLS (21)(22)(23)(24). Here, we investigated the in vitro stability of the UmuDЈ 2 C complex and its ability to undergo subunit exchange with UmuD and UmuDЈ proteins. Although a soluble UmuDЈ 2 C complex can be purified over multiple chromatographic steps, UmuDЈ is readily displaced from UmuC by UmuD, concomitantly causing UmuC to aggregate as an inactive insoluble precipitate. Subunit exchange between UmuDЈ 2 C and UmuD provides a biologically plausible way to curtail the activity of error-prone pol V following TLS and thus reduces the chance that mutations will occur on undamaged DNA during SOS.
Interaction between pol V and in Vitro 35 S-Labeled UmuD Protein-Purified UmuD protein was labeled in vitro following a protocol modified from previous studies (29,30). In the labeling reaction, UmuD (5 mg/ml) and Trans 35 S label mix from ICN Biomedicals (0.5 mCi/ml) were incubated in 20 mM Tris⅐HCl (pH 8.5) and 10 mM EDTA at 37°C for 20 h. The mixture was loaded first to a Bio-spin P-6 column from Bio-Rad and then to a 20-ml Sephadex G-75 column to isolate the * This work was supported by National Institutes of Health Grants GM42554 and GM21422 (to M. F. G.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
¶ To whom correspondence should be addressed: University of Southern California, Dept. of Biological Sciences, SHS Rm. 172, University Park, Los Angeles, CA 90089-1340. Tel.: 213-740-5190; Fax: 213-821-1138; E-mail: mgoodman@usc.edu. 1 The abbreviations used are: pol V, polymerase V; TLS, translesion synthesis; ATP␥S, adenosine 5Ј-3-O-(thio)triphosphate. 35 S-labeled proteins. Fractions containing labeled UmuD 2 homodimer were collected. Unlabeled pol V (UmuDЈ 2 C complex) was mixed with 35 S-labeled UmuD 2 homodimer in a 1:3 ratio (0.7 M and 2.1 M, respectively) in R buffer (20 mM Tris⅐HCl, pH 7.5, 0.1 mM EDTA, 1 mM dithiothreitol, 20% glycerol) with 300 mM NaCl and incubated at 37°C for different lengths of time. Samples were chilled on ice before centrifugation at 14,000 ϫ g in a microcentrifuge for 10 min. The soluble supernatant was then separated on a 20-ml Superdex G-75 column at 4°C. The radioactivity from each gel filtration fraction was measured on a liquid scintillation counter, and SDS-PAGE was used to determine the distribution of 35 S-labeled UmuD and unlabeled pol V subunits. A control experiment was carried out with 35 S-labeled UmuD 2 homodimer incubated for 1 h at 37°C.
Polymerase V Catalyzed Primer Extension and Translesion Synthesis Assay-A synthetic 30-nucleotide primer (5Ј-ACT GAC CCC GTT AAA ACT TAT TAC CAG TAA-3Ј) was 5Ј-labeled with 32 P and annealed to a 39-nucleotide template (5Ј-TAC AGG XGT TTA CTG GTA ATA AGT TTT AAC GGG GTC AGT-3Ј) containing an abasic site (X), generating a 9-nucleotide overhang downstream of the primer 3Ј-end. The abasic site (X) on the template was 3 nucleotides from the primer 3Ј-end. The reaction buffer contained 20 mM Tris⅐HCl (pH7.5), 8 mM MgCl 2 , 5 mM dithiothreitol, 0.1 mM EDTA, 25 mM sodium glutamate, 40 g/ml bovine serum albumin and 4% (v/v) glycerol. Polymerase V (400 nM) was either pre-incubated at 37°C alone for 3 min or with purified UmuD, UmuD1, or UmuDЈ proteins before being added to a mixture of DNA (10 nM), RecA (250 nM), ATP␥S (1 mM), and 4 ϫ dNTPs (200 M each) to initiate the reaction. The reaction was incubated at 37°C for 10 min and quenched by adding an equal volume of 95% formamide, 40 mM EDTA. The products were heat-denatured and separated on denaturing polyacrylamide gels. Radioactive gel bands were analyzed on a Phospho-rImager using ImageQuant software (Amersham Biosciences).
Cross-linking Study of Interaction between pol V and UmuD-The method used for cross-linking experiments was adapted from Ref. 31. Polymerase V, UmuD, UmuDЈ, and bovine serum albumin were incubated in different combinations at 37°C for 10 min in a buffer containing 50 mM HEPES, pH 7.5, 100 mM potassium glutamate, 10 mM MgCl 2 , and 1 mM dithiothreitol before formaldehyde was added to 1% to initiate cross-linking. The mixture was incubated for 20 min at room temperature following the addition of formaldehyde and quenched by SDS-PAGE loading buffer. The samples were heated in a boiling water bath for 5 min and resolved on a 14% SDS-PAGE. Western blotting was conducted with antibodies that react with UmuDDЈ and UmuC to identify UmuD-, UmuDЈ-, and UmuC-containing species. In a set of control experiments, pol V was first incubated with formaldehyde for 20 min at room temperature to allow cross-linking between UmuC and UmuDЈ subunits before UmuD protein was added and further co-incubated for 10 min at 37°C.
In Vitro Transcription/Translation of UmuC, UmuDЈ, and UmuD-Plasmids pUmuC (pEC47), pUmuDЈ (pEC48), and pUmuD (pJM103) are pET11 derivatives that overproduce UmuC, UmuDЈ, and UmuD, respectively, from a T7 promoter (27,28). These plasmids were used as templates for in vitro expression of the various proteins. The in vitro transcription/translation reactions were carried out using the E. coli T7 S30 Extract System for Circular DNA (Promega) according to the manufacturer's protocol. L-[ 35 S]Methionine (ICN Biomedicals) was used in the reactions to radiolabel the synthesized protein products. The radiolabeled products were confirmed by SDS-PAGE. Formation of potential complexes among UmuC, UmuDЈ, and UmuD were analyzed by separating the products through a 4-ml Superdex G-200 gel filtration column as follows. Immediately following the in vitro transcription/translation reactions, the reaction mix was briefly treated with DNase and RNase; 40 l of each sample was applied, without centrifugation, to a Superdex G-200 column and separated at 4°C in the presence of R buffer containing 300 mM NaCl. 100-l fractions were collected and separated on SDS-PAGE, and the radioactivity in each band was determined by phosphorimaging.

RESULTS AND DISCUSSION
Subunit Swapping of pol V with 35 S-Labeled UmuD Protein-We have purified pol V as a native heterotrimeric complex (UmuDЈ 2 C) that remains intact and stable in aqueous solution (9). In contrast, UmuC containing the polymerase active site (21, 24) is essentially insoluble in aqueous solution (8,9,24) unless expressed as a recombinant maltose-binding fusion protein (24). Aside from the observation that pol V can be purified as a UmuDЈ 2 C heterotrimer (9), little is known about the physical properties of the heterotrimeric complex.
An important aspect regarding the regulation of pol V is to determine the ability of pol V to exchange the UmuDЈ subunits bound to UmuC. Using 35 S-labeled UmuD, we measured the exchange of the UmuDЈ subunit in pol V by the mutagenically inactive UmuD subunit (Fig. 1). Purified pol V and [ 35 S]UmuD (present primarily in the form of a dimer, i.e. [ 35 S]UmuD 2 ) were incubated for 1, 10, and 60 min, and the products were resolved by gel filtration with Superdex G75. Both peaks containing 35 S are present in the G75-included volume (Fig. 1a). There is a decreasing amount of 35 S label co-eluting with the UmuD 2 or UmuDDЈ dimer peak with increasing incubation time (Fig. 1a, fractions c and d) and an increasing amount of 35 S label with the UmuD monomer peak (Fig. 1a, fraction g) The identity of the product in each peak was determined using Coomassie Blue-stained gels (Fig. 1b). The "heavy" peaks (Fig. 1b, fractions c and d) corresponding to a 10-min incubation contain both UmuD and UmuDЈ. The heavy peaks elute with a molecular mass consistent with a UmuDDЈ heterodimer complex (Fig. 1a). The "light" peak (Fig. 1b, fraction g) is composed primarily of UmuD and contains a trace amount of UmuDЈ (Fig. 1b); this peak elutes with molecular mass markers corresponding to monomer species (Fig. 1b). As a control, [ 35 S]UmuD was incubated in the absence of pol V for 60 min and remained intact as a dimer (Fig. 1a, inset). Absent from any of the gel elution fractions, including the void volume, were UmuD 2 C or UmuDDЈC, whereas "free" UmuC was detected as an insoluble aggregate contained in the centrifuged pellet of the UmuDЈ 2 C/[ 35 S]UmuD 2 exchange reaction by Western blotting against the protein resolubilized in 8 M urea (data not shown).
Polymerase V Is Stabilized by UmuDЈ and Destabilized by UmuD-The gel filtration data demonstrate the important point that the UmuDЈ subunits of the pol V heterotrimer in solution are exchangeable with free UmuD or UmuD 2 , resulting in the formation of UmuDDЈ while releasing UmuC as an insoluble precipitate. The corollary to this observation is that incubation of UmuD with pol V should inactivate the polymerase, because the catalytic activity is present only in UmuC (21,24). The ability of pol V to copy undamaged DNA and to catalyze TLS is reduced in the presence of UmuD, the inhibition being much stronger at higher concentrations of UmuD (Fig. 2). A similar result occurred when pol V was incubated with UmuD1, a noncleavable UmuD mutant that cannot be converted to UmuDЈ in the presence of activated RecA (32) (Fig. 2).
In contrast, a small but significant (2-fold) stimulation of pol V activity occurs in the presence of excess UmuDЈ (Fig. 2). Thus, although the UmuDЈ 2 C complex is relatively stable and is able to maintain its integrity through multiple purification stages (9,21), the exchange of free UmuD or UmuDЈ proteins with a UmuDЈ monomeric component of pol V (UmuDЈ 2 C), nevertheless, takes place. Stimulation of pol V could occur if the inactive UmuDЈC is converted to active UmuDЈ 2 C in the presence of excess UmuDЈ. Our data, therefore, imply the existence of an equilibrium between UmuDЈ 2 C and UmuDЈC that strongly favors UmuDЈ 2 C.
Identification of pol V Subunit Exchange Products by Formaldehyde Cross-linking Analysis-Combined formaldehyde cross-linking and Western blot analyses were performed to identify the complexed and free forms of the Umu proteins following the addition of excess UmuD or UmuDЈ to pol V (Fig.  3). Incubation of UmuD alone followed by cross-linking with formaldehyde resulted in the formation of a UmuD 2 complex and free UmuD (Fig. 3, lane 4), and a similar experiment using UmuDЈ resulted in formation of UmuDЈ 2 and free UmuDЈ (Fig.  3, lane 5). Co-incubation of UmuD and UmuDЈ followed by cross-linking of the products resulted in the appearance of one additional complex of UmuDDЈ (Fig. 3, lane 6).
In a parallel experiment using formaldehyde cross-linking, when increasing amounts of UmuD were incubated with a fixed amount of pol V, the exchange product was UmuDDЈ (Fig. 3,  lanes 8 -10). An equally important observation concerns the putative exchange products that are not observed in solution; neither UmuD 2 C nor UmuDDЈC complexes were detected (Fig.   3, lanes 8 -10). That is, essentially all of the UmuDЈ 2 C (Fig. 3,  lane 7) was converted to a UmuDDЈ plus UmuC, and no soluble formaldehyde cross-linked UmuD 2 C or UmuDDЈC complexes were observed. The absence of detectable free UmuC by Western analysis (Fig. 3, lanes 8 -10) is attributable to its insolubility in aqueous solution (8,9,24,33) causing it to aggregate and form a visible precipitate. However, a clear UmuC signal was observed on a Western blot of pol V incubated in the absence of UmuD (Fig. 3, lane 7).
The addition of increasing amounts of UmuDЈ to pol V resulted in the recovery of only pol V (where both UmuC and UmuDЈ are detected) ϩ free UmuDЈ (Fig. 3, lanes 11-13). There was a roughly 1.5-2.0-fold increase in the intensity of the UmuDЈ 2 C band incubated in the presence of excess UmuDЈ (Fig. 3, lanes 11-13) compared with no incubation with UmuDЈ (Fig. 3, lane 7). These data are consistent with those showing that an excess UmuDЈ stimulates pol V activity by a similar 1.5-2-fold increase (Fig. 2). There was no UmuDDЈ exchange observed when proteins were cross-linked with formaldehyde prior to the protein co-incubation reactions (Fig. 3, lanes 17-19). Once again, our data imply that there is ongoing subunit exchange within pol V and that formaldehyde cross-linking helps minimize the exchange and the aggregation of insoluble UmuC.
Transcription/Translation in Vitro Synthesis of pol V to Identify Putative Umu Complexes-Although it is tempting to suggest that soluble complexes composed of UmuD 2 C do not occur based on the inability to detect them by swapping UmuD and UmuC subunits using highly purified proteins (Figs. 1 and 3), it is nevertheless possible that such complexes can still form in vivo. Perhaps the best chance of mimicking "life-like" conditions in vitro is to use a transcription/translation system to co-express 35 S-labeled UmuC and UmuD proteins (Fig. 4). Each   FIG. 2. Inhibition of pol V activity by UmuD and stimulation by UmuD. Polymerase V-catalyzed primer extension was carried out as described under "Experimental Procedures." Prior to initiating polymerization, pol V was pre-incubated at 37°C for 3 min either alone or with varying concentrations of wild type UmuD, a noncleavable mutant UmuD1, or UmuDЈ. The relative pol V activity was plotted as a function of molar ratios of UmuD (closed circles), UmuD1 (open squares), or UmuDЈ (triangles) to pol V. Polymerase V activity following preincubation in the absence of added protein is designated as 100%. of the overexpressed 35 S-labeled protein products, UmuC, UmuD, and UmuDЈ, were readily detected by SDS-PAGE (Fig.  4a) or by SDS-PAGE following gel filtration on Superdex G200 (Fig. 4, b-f). UmuDЈ and UmuC gave products of the expected mass. In vitro transcription/translation of UmuD gave two products, one that corresponds to full-length UmuD and one that migrates slightly faster than UmuDЈ. The expression of this smaller protein was variable and was even evident in a noncleavable mutant of UmuD, suggesting that its expression was driven from the UmuD plasmid itself and not from posttranslational modification of UmuD during the in vitro transcription/translation reactions.
When UmuC is expressed alone and chromatographed on G200, it is detected only as a high molecular weight aggregate in the excluded volume (Fig. 4b). Expression of either UmuDЈ (Fig. 4c) or UmuD (Fig. 4d) results primarily in the formation of homodimers of each protein.
The key results are that when UmuC and UmuDЈ are co-expressed, fractions corresponding to pol V are detected (Fig. 4e, fractions designated as "UmuDЈ 2 C"). In contrast, when UmuC and UmuD are co-expressed, no soluble complex appears to form between the two proteins (Fig.  4f). These data are consistent with the formaldehyde crosslinking data in which no soluble UmuD 2 C complex was observed (Fig. 3). Although UmuD alone migrates as a soluble dimer (Fig. 4d), when mixed with UmuC, a fraction of the UmuD elutes in the void volume along with insoluble UmuC. The data indicate that UmuD and UmuC are able to interact physically and that the interaction causes both UmuD and UmuC proteins to aggregate as insoluble precipitates. A likely explanation for the absence of UmuC in the gel filtration void volume for the subunit exchange reaction (Fig. 1) is that the insoluble protein aggregates and is removed by centrifugation prior to loading the sample on the column. A key difference in the two gel filtration experiments is that UmuC is present in g amounts when carrying out subunit exchange using purified proteins ( Fig. 1) but only in ng amounts using the transcription/translation system (Fig. 4).
We also carried out similar in vitro co-expression experiments with noncleavable UmuD1 in place of wild type UmuD, and the results obtained were similar to that seen with wild type UmuD (data not shown). In an experiment where UmuC, UmuD1, and UmuDЈ were all co-expressed together, most of the UmuD1 and UmuDЈ migrated as a soluble heterodimer, whereas UmuC and a small fraction of UmuD1DЈ eluted in the void volume as insoluble aggregates.
Biological Relevance of pol V Subunit Exchange-Based upon the known biochemical properties of the Umu proteins, it has been hypothesized that they form a variety of protein complexes. UmuD and UmuDЈ can exist as homodimers but prefer to heterodimerize whenever possible. UmuC may exist as a monomer as well as in a complex with UmuD 2 , UmuDDЈ, and UmuDЈ 2 . Although the UmuD and UmuDЈ proteins have been successfully isolated in both homodimeric (8,15) and heterodimeric forms (8), UmuC has only been isolated in significant quantities as a soluble UmuDЈ 2 C complex (9,21,23), thereby raising the question of whether UmuC, UmuD 2 C, or UmuDDЈC are present in vivo. And, if present in vivo, they may still be largely insoluble in the cell.
To date, monomeric UmuC has only been isolated as a recombinant maltose-binding protein fusion (24) or as a native untagged form but only after denaturation and renaturation of the overexpressed protein (8,33). A key step in the purification of denatured/renatured UmuC included affinity chromatography utilizing UmuDDЈ proteins as the ligand. UmuC bound avidly to this column and could only be eluted under denaturing conditions (8). These observations were originally interpreted as indicating tight complex formation between UmuDDЈ and UmuC. However, given our inability to detect any such complexes in solution, we now believe it much more likely that Once cellular DNA damage has been repaired, the SOS-inducing signal wanes, and there is a concomitant decrease in the rate at which UmuD is converted to UmuDЈ. Although UmuD can exist as a homodimer, it prefers to heterodimerize with UmuDЈ and, in doing so, displaces UmuC, causing it to aggregate as an insoluble and inactive precipitate. When complexed with UmuD, UmuDЈ becomes a substrate of the ClpXP protease and is rapidly degraded. This allows UmuD to homodimerize and leads to the Lon-mediated proteolytic degradation of UmuD. Thus, pol V subunit exchange combined with efficient proteolytic degradation of the Umu proteins helps to curtail error-prone translesion DNA synthesis and keep pol V-dependent mutagenesis on undamaged DNA to a minimum.
UmuDDЈ caused UmuC to aggregate and precipitate on the column.
In vivo, co-expression of UmuDDЈ and UmuC also led to a dramatic increase in the half-life of UmuC (34). This observation was also taken as an indication that UmuDDЈ interacts with UmuC so as to protect it from Lon-mediated degradation. Perhaps a more likely alternative hypothesis, based on our current data, is that the apparent increase in the stability of UmuC resulted from interactions between UmuDDЈ and UmuC that caused UmuC to aggregate within the cell, thereby making it inaccessible to Lon-proteolysis. The in vivo and in vitro data taken together offer convincing evidence that any interaction between UmuC and UmuD 2 or UmuDDЈ appears quantitatively and qualitatively different from that involving UmuDЈ 2 C. The latter is stable and resistant to cellular proteolysis and is readily purified as a complex (9,21,23). Even so, the pol V complex is susceptible to subunit exchange.
We propose that subunit exchange provides an internal "selfdestruct" mechanism to curtail the activity of error-prone pol V following TLS, and this reduces the chance that mutations will occur on undamaged DNA during SOS. Our data show that subunit exchange between the UmuDЈ 2 C complex and intact UmuD has two dramatic effects. First, it most likely leads to the inactivation of the catalytic function of UmuC as the protein aggregates as an insoluble precipitate (Fig. 5). Second, it results in the formation of a soluble UmuDDЈ complex in which UmuDЈ is susceptible to rapid proteolysis by ClpXP (Fig. 5). It is possible that monomeric UmuC and UmuD 2 C exist in a cell, but the physical properties of such Umu complexes appear to be entirely different from that of the soluble and catalytically active UmuDЈ 2 C (pol V). Thus, based on current biochemical data, we invoke Occam's razor to suggest that the sole soluble complex involving UmuC in the cell is a heterotrimer composed of UmuDЈ 2 C, which is very likely the biologically relevant form of pol V (20).