JOURNAL OF BACTERIOLOGY, May 1996, p. 2580–2585 0021-9193/96/$04.0010 Copyright q 1996, American Society for Microbiology
Vol. 178, No. 9
Specific In Vivo Protein-Protein Interactions between Escherichia coli SOS Mutagenesis Proteins PIOTR JONCZYK*
AND
ADRIANNA NOWICKA
Institute of Biochemistry and Biophysics, Polish Academy of Sciences, 02-106 Warsaw, Poland Received 10 November 1995/Accepted 20 February 1996
One of the components of the RecA-LexA-controlled SOS response in Escherichia coli cells is an inducible error-prone DNA replication pathway that results in a substantial increase in the mutation rate. It is believed that error-prone DNA synthesis is performed by a multiprotein complex that is formed by UmuC, UmuD*, RecA, and probably DNA polymerase III holoenzyme. It is postulated that the formation of such a complex requires specific interactions between these proteins. We have analyzed the specific protein-protein interactions between UmuC, UmuD, and UmuD* fusion proteins, using a Saccharomyces cerevisiae two-hybrid system. In agreement with previous in vitro data, we have shown that UmuD and UmuD* are able to form both homodimers (UmuD-UmuD and UmuD*-UmuD*) and a heterodimer (UmuD-UmuD*). Our data show that UmuC fusion protein is capable of interacting exclusively with UmuD* and not with UmuD. Thus, posttranslational processing of UmuD into UmuD* is a critical step in SOS mutagenesis, enabling only the latter protein to interact with UmuC. Our data seem to indicate that the integrity of the entire UmuC sequence is essential for UmuC-UmuD* heterotypic interaction. Finally, in our studies, we used three different UmuC mutant proteins: UmuC25, UmuC36, and UmuC104. We have found that UmuC25 and UmuC36 are not capable of associating with UmuD*. In contrast, UmuC104 protein interacts with UmuD* protein with an efficiency identical to that of the wild-type protein. We postulate that UmuC104 protein might be defective in interaction with another, unknown protein essential for the SOS mutagenesis pathway. Experimental data available to date have suggested the involvement of DNA pol III in the SOS mutagenic pathway (4, 17, 23, 40, 41). Recent studies have shown that DNA pol III holoenzyme, in concert with UmuD9, UmuC, and RecA proteins, promotes in vitro replication across the site of the DNA lesion (21, 40, 41). The postulated function of RecA in this process is to target the UmuD9-UmuC complex to the DNA replication-blocking lesion (1, 11, 21, 41, 48). The postulated complex, called the mutasome, containing RecA, UmuD9, and UmuC, presumably allows the stalled DNA pol III to resume DNA replication (14, 21, 41, 54). Currently, however, the molecular mechanism by which these proteins enable DNA pol III to copy the damaged template is not understood. It has been speculated that during error-prone translesion DNA synthesis, a specific modification of DNA pol III may occur (5, 14, 49). This postulated modification may allow DNA pol III to extend the de novo-synthesized DNA strand beyond the blocking lesion. Data from in vitro studies indicate that formation of the active mutasome requires specific protein interactions (21, 40, 41). Presumably, appropriate interactions between UmuD or UmuD9 and UmuC are crucial for functioning of this complex. However, it is unclear how this complex functions at the molecular level in translesion DNA synthesis. We have been particularly interested in elucidating the nature of protein-protein interactions between the SOS mutagenesis proteins in vivo. In an attempt to achieve this goal, we have used a Saccharomyces cerevisiae two-hybrid system (8, 19, 20) designed to examine specific protein-protein interactions. In the work reported here, we investigated in vivo interactions between UmuD9, UmuD, and UmuC proteins. Our data clearly indicate that UmuC is capable of interacting with only UmuD9 and not with UmuD. We propose that posttranslational processing of UmuD into a shorter, UmuD9 protein enables the latter protein to interact with UmuC efficiently so that the complex of the two proteins can carry out its biological function.
Exposure of Escherichia coli cells to agents that damage DNA and introduce replication-blocking lesions results in the induction of the SOS regulon (13, 22, 33, 51, 52). Approximately 20 cellular genes are derepressed when activated RecA promotes the proteolytic cleavage of LexA protein, the repressor of SOS genes (22, 25, 32, 33, 44, 51, 52). One of the components of the RecA-LexA-controlled SOS response is an inducible error-prone DNA replication pathway that causes a significant increase in the mutation rate (13, 22, 33, 51, 52). Genetic and physiological experiments indicate that the products of the umuDC operon and the recA gene are essential components of the SOS mutation pathway (3, 12, 16, 17, 26, 44, 47, 48). This is supported by the fact that umuD and umuC mutants of E. coli are virtually nonmutable with UV light and many chemical agents (26, 47). The recA and umuDC genes are members of the SOS regulon and are expressed at elevated levels after DNA damage (15, 31, 46, 53, 55). In SOS-competent cells, activated RecA also mediates the cleavage of UmuD at its Cys-24–Gly-25 bond, yielding a shorter UmuD9 product, which has been proposed to be an active form of UmuD for mutagenesis (6, 37, 45, 54). Several data suggest that UmuC, UmuD9, and RecA proteins facilitate the proceeding of the DNA replication complex beyond lesions in the template (5, 21, 40, 41). In addition, it has been shown that overexpression of the umuDC operon results in cold sensitivity of DNA replication (35). Also, it has been observed that a high cellular level of UmuC is exceptionally harmful to the defective DNA polymerase III (pol III) of the dnaQ49 mutant (38). These data suggest that UmuD and UmuC proteins may directly interact with the DNA replication apparatus. * Corresponding author. Mailing address: Institute of Biochemistry and Biophysics, Polish Academy of Sciences, 02-106 Warsaw, Pawin ´skiego 5A, Poland. Fax: (48) 39 12 16 23. Electronic mail address:
[email protected]. 2580
VOL. 178, 1996
UmuD, UmuD9, AND UmuC PROTEIN-PROTEIN INTERACTIONS
TABLE 1. Interaction of UmuD and UmuD9 fusion proteins: quantitative assay of b-galactosidase activity Plasmid(s)a
b-Galactosidase activityb
pCL1(GAL4) ................................................................. 3,243 6 357 pGBT9 1 pGAD424umuD.......................................... ,2 pGBT9 1 pGAD424umuD9 ........................................ ,2 pGBT9umuD 1 pGAD424.......................................... ,2 pGBT9umuD9 1 pGAD424 ........................................ ,2 pGBT9umuD 1 pGAD424umuD ............................... 83 6 17 pGBT9umuD9 1 pGAD424umuD9 ............................ 255 6 70 pGBT9umuD9 1 pGAD424umuD.............................. 923 6 280 pGBT9umuD 1 pGAD424umuD9.............................. 1,423 6 193 a
The Y187 reporter strain was transformed with the indicated plasmids. b-Galactosidase specific activities are calculated as nanomoles of O-nitrophenyl galactoside hydrolyzed per minute per milligram of protein (42). The values are averages for the four transformants, each assayed in triplicate (6 standard deviations). b
MATERIALS AND METHODS Bacterial and yeast strains and media. E. coli DH5a [supE44 DlacU169 (f80 lacZDM15) hsdR17 recA1 endA1 gyrA96 thi-1 relA1] was the transformation recipient for all plasmid constructions. All two-hybrid system experiments were done with S. cerevisiae Y187 (MATa gal4D gal80D his3 trp1-901 ade2-101 ura3-52 leu2-3,112 met URA3::GAL1-lacZ). Strain Y187 was kindly provided by S. Elledge (Baylor College of Medicine, Houston, Tex.). Yeast YEPD medium and SMM were prepared as described previously (42). For drug selection, Luria broth plates were supplemented with ampicillin (100 mg/ml). Methods. Manipulations and sequencing of DNA were carried out by standard procedures (43). The S. cerevisiae Y187 strain was transformed simultaneously with a pGBT9-derived plasmid (e.g., pGBTumuD) and a pGAD424-derived plasmid (e.g., pGAD424umuD9) by the method of Chen et al. (7). b-Galactosidase assays. For quantitative studies, yeast strains were grown at 288C to stationary phase in synthetic medium (SMM plus 3% glucose) lacking leucine and tryptophan, diluted 1:10 in SMM plus 2% ethanol (lacking leucine and tryptophan), and then incubated at 288C for 48 h. The b-galactosidase activity was determined as described elsewhere (42). For qualitative studies, a filter assay with 5-bromo-4-chloro-3-indolyl-b-D-galactopyranoside (X-Gal) as a substrate was performed as described previously (50). Construction of GAL4 protein fusion plasmids. Plasmids for the GAL4 twohybrid fusion assay were prepared by cloning PCR-amplified fragments into pGBT9 (Clontech) (containing amino acids 1 to 147 of the DNA-binding domain of GAL4) and pGAD424 (Clontech) (containing amino acids 768 to 881 from the trans-activation domain of GAL4). The umuD, umuD9, and umuC coding sequences were obtained from pSE117 (39), kindly provided by G. Walker (Department of Biology, Massachusetts Institute of Technology, Cambridge), by PCR amplification with the following forward and reverse primers, respectively: umuD, 59-ATGTTGTTTATCAAGCCTGCG-39 and 59-GGCAAACATCAG CGCATC-39; umuD9, 59-GGCTTTCCTTCACCGGCA-39 and 59-GGCAAACA TCAGCGCATC-39; and umuC, 59ATGTTTGCCCTCTGTGATG-39 and 59-TT TTCCTGCCGCTATATTT-39. Taq polymerase (Promega), which has a terminal transferase activity and preferentially adds adenine to the 39 ends of PCR amplification products (9), was used for the PCR. The PCR conditions, for the Perkin-Elmer thermal cycler, model 2400, consisted of 25 cycles of denaturation at 948C for 90 s, annealing at 558C for 90 s, and extension at 728C for 120 s. The PCR products were purified by agarose gel electrophoresis and cloned into the SmaI site with T overhangs of pGBT9 and pGAD424. pGBT9 and pGAD424 were prepared by SmaI digestion and T tailing as described previously (34). To confirm the presence of the in-frame junction, recombinant plasmids pGBT9umuD, pGBT9umuD9, and pGBT9umuC were sequenced with GAL4bd primer (59-GAAGAGAGTAGTAACAAAGG-39) for the pGBT9 constructs or GAL4ad primer (59-GCGTTTGGAATCACTACAGG-39) for the pGAD424 constructs. The entire DNA sequences of the PCR-derived umuD, umuD9, and umuC inserted DNA fragments have been verified by dideoxy sequencing. Construction of umuC deletion plasmids. The 1.3-kb EcoRI-PstI fragment containing the entire umuC sequence was cut out from pGBT9umuC (the 59 EcoRI site and the 39 PstI site were from the multiple cloning site of pGBT9). This fragment was digested with different restriction enzymes, and appropriate fragments were isolated and cloned into pGBT9. Carboxy- and amino-terminal deletions of umuC were produced as follows: D1umuC [EcoRI-NruI(756) 0.35-kb DNA fragment], D2umuC [EcoRI-EcoRV(1045) 0.7-kb DNA fragment], and D3umuC [EcoRI-SmaI(1232) 0.83-kb DNA fragment] were cloned into pGBT9 digested with EcoRI and SmaI; D4umuC [EcoRI-BamHI(1601) 1.2-kb DNA fragment] was cloned into pGBT9 digested with EcoRI and BamHI; and D5umuC [(756)NruI-PstI 0.93-kb DNA fragment] and D6umuC [(453)PvuII-PstI 1.2-kb DNA fragment] were cloned into pGBT9 digested with SmaI and PstI. The numbers in parentheses indicate the start positions of sites based on the
2581
umuDC operon numbering system (39). The presence of in-frame junctions in all recombinant plasmids has been confirmed by sequencing. Construction of pGBT9-3umuC25, pGBT9-3umuC36, and pGBT9-3umuC104 plasmids. pGBT9-3 was constructing by removing BamHI, SalI, and PvuII unique sites from pGBT9 as follows. First, pGBT9 was SalI digested, filled in, BamHI digested, filled in, and religated. The resulting pGBT9-2 plasmid has no BamHI or SalI sites. As confirmed by sequencing, the DNA cloning site in pGBT9-2 is 59-GGATTCCCGGGGATTCGACCTGCAGCC-39. This plasmid was PvuII digested, treated with mung bean nuclease (29), and religated. In the resulting plasmid, pGBT9-3, the entire umuC coding sequence was cloned as described for pGBT9umuC. The umuC coding sequence has unique PvuII(455) and BamHI(1603) sites (39). The mutants umuC25, umuC36, and umuC104 have single-base-pair substitutions at nucleotides 1288, 642, and 720, respectively (28). The mutations in these sites allowed replacement of the PvuII-BamHI DNA fragment in the wild-type umuC DNA sequence in plasmid pGBT9-3umuC with appropriate PvuII-BamHI DNA fragments isolated from plasmids carrying umuC mutations. As a source of umuC mutant DNA sequence, we used plasmids pRW124-C25, pRW124-C36, and pRW124-C104 (56), kindly provided by R. Woodgate (National Institute of Child Health and Human Development, Bethesda, Md.). After the wild-type umuC sequence was replaced with the umuC25, umuC36, and umuC104 mutant sequences, the presence of the mutant nucleotides was confirmed by sequencing.
RESULTS Considerable efforts have been made to purify UmuC, UmuD, and UmuD9 proteins and to examine their specific interactions under in vitro conditions (54). However, the results of these in vitro studies are hard to evaluate because of difficulties in obtaining purified UmuC protein in a soluble and active form. Consequently, it is not clear whether UmuC is able to interact specifically with only UmuD9 or UmuD or with both of them. To broaden our understanding of the role of UmuD9, UmuD, and UmuC in SOS mutagenesis, we decided to employ an in vivo strategy to examine the specific physical association between these proteins. We have chosen the yeast two-hybrid system devised by Fields and Song (19). Interactions between UmuD and UmuD* proteins. To evaluate the usefulness of the two-hybrid system for studying E. coli protein-protein interactions, we started our study by examining interactions between UmuD and UmuD9 proteins. It has been shown that UmuD (139 amino acids) and UmuD9 (115 amino acids) are able to form homodimers and a heterodimer in vitro (2, 30, 54). We cloned the complete DNA coding sequence of UmuD (starting from the first ATG codon) and that of UmuD9 (starting from the 25th GGC codon of the umuD gene, which encodes the first amino acid of UmuD9) into both pGBT9 and pGAD424. All pairwise combinations of pGBT9, pGBT9umuD, or pGBT9umuD9 and pGAD424, pGAD424umuD, or pGAD424umuD9 were introduced into yeast reporter strain Y187. After selection, cotransformants were screened for their ability to produce b-galactosidase by a filter assay (not shown) and by measuring b-galactosidase activity quantitatively in solution (Table 1). As a reference, we used plasmid pCL1 (19), which contains the entire GAL4 coding region (881 amino acids). The results shown in Table 1 indicate that UmuD fusion protein is able to interact specifically with itself (homodimer formation) and with UmuD9 fusion protein (heterodimer formation). UmuD9 fusion protein is also able to form a homodimer. The observed efficiency of dimer formation, measured by the activity of the reporter lacZ gene, was as follows: UmuD-UmuD9 . UmuD9-UmuD9 . UmuD-UmuD. This could reflect different abilities of these proteins to interact. The same specificity and similar differences in efficiency of dimerization between UmuD and UmuD9 have been shown in in vitro experiments using purified proteins (2). The above data indicate that the yeast two-hybrid system can be used for investigating protein-protein interactions of E. coli proteins.
2582
JONCZYK AND NOWICKA
J. BACTERIOL.
TABLE 2. Interactions between UmuC, UmuD, and UmuD9 fusion proteins: quantitative assay of b-galactosidase activity Plasmids
b-Galactosidase activitya
pGBT9umuC 1 pGAD424 ............................................ ,2 pGBT9umuC 1 pGAD424umuD.................................. ,2 pGBT9umuC 1 pGAD424umuD9 ................................516 6 134 pGBT9umuD 1 pGAD424umuC.................................. ,2 pGBT9umuD9 1 pGAD424umuC ................................ ,2 a
strong as the interaction found between wild-type UmuC protein and UmuD9 protein.
Determined as described in Table 1, footnote b.
Interaction of UmuC and UmuD* fusion proteins. In the next series of experiments, we examined whether we could detect interaction between UmuC fusion protein and either UmuD or UmuD9 fusion proteins. The results presented in Table 2 show that UmuC is able to associate exclusively with UmuD9. There is no detectable interaction between UmuC fusion protein and UmuD fusion protein. It is interesting that the formation of UmuC-UmuD9 protein complex occurs only when UmuC is cloned on the pGBT9 plasmid. No interaction occurs when UmuC is fused to the trans-activation domain of GAL4 protein (pGAD424). It seems that these differences resulted from a need of the appropriate UmuC tertiary structure for specific interaction with UmuD9. The presence of the additional 113 amino acids (GAL4 trans-activation domain) fused at the N terminus of UmuC might interfere with the proper folding of UmuC and as a consequence prevent interaction with UmuD9 fusion protein. In contrast, the presence of the GAL4 binding domain fused to UmuC allows proper folding to occur. Deletion mapping of the UmuC region engaged in interaction with UmuD*. The UmuC protein is 422 amino acids long. In an attempt to define a specific domain of UmuC which is responsible for interaction with UmuD9, a series of amino- and carboxy-terminal deletions of the UmuC was constructed and their ability to interact with UmuD9 was analyzed (Table 3). Unexpectedly, we have found that none of the GAL4-DUmuC fusion proteins were able to interact with UmuD9 fusion protein. Even removal of 13 amino acids (3% of the protein) from the amino-terminal end of UmuC [pGBT9D6umuC(14-422)] or 26 amino acids (6% of the protein) from the carboxyterminal end of UmuC [pGBT9D4umuC(1-396)] completely eliminated the ability of UmuC to interact with UmuD9. Our finding suggests that the integrity of the entire UmuC sequence is critical for UmuC-UmuD9 complex formation. Interaction of UmuD* with mutant UmuC proteins. To characterize further the UmuC-UmuD9 interaction, we examined the ability of the UmuD9 fusion protein to interact with several UmuC mutant proteins. E. coli umuC25, umuC36, and umuC104 mutants were originally isolated as nonmutable by UV light (26, 47). Single-base-pair substitutions that result in missense mutations in each mutant allele of umuC have been identified (28). If the interaction of UmuC with UmuD9 has biological significance, we might expect that at least some mutations in umuC might affect the ability of UmuC to interact with UmuD9. We therefore cloned all three umuC mutant DNA sequences on pGBT9-3 (see Materials and Methods), and interactions with UmuD and UmuD9 fusion proteins were assessed by the ability to transactivate the lacZ reporter construct (Table 4). We did not observe an interaction between UmuC25 and UmuC36 proteins with either UmuD or UmuD9 protein. In contrast, a specific interaction between UmuC104 protein and UmuD9 protein was found. As a matter of fact, it was as
DISCUSSION Umu-dependent, error-prone translesion DNA synthesis occurs as a part of the RecA-LexA-controlled SOS response (13, 22, 33, 51, 52). It is believed that error-prone DNA synthesis is performed by a multiprotein complex that is formed by UmuC, UmuD9, RecA, and probably pol III holoenzyme (21, 40, 41). On the basis of in vitro studies, it is postulated that formation of such a complex requires specific interactions between these proteins. However, these in vitro data have not been verified under in vivo conditions. We have used a yeast two-hybrid system to identify specific associations between UmuC, UmuD, and UmuD9 proteins in vivo. We have shown that, in this system, UmuD and UmuD9 fusion proteins are able to form both homodimers (UmuDUmuD and UmuD9-UmuD9) and a heterodimer (UmuDUmuD9) (Table 1). Our data also suggest different efficiencies of interactions between the tested proteins, as measured by transcriptional activation of the reporter lacZ gene. The b-galactosidase activities indicate very strong and/or stable UmuDUmuD9 interaction, weaker UmuD9-UmuD9 interaction, and still weaker association in UmuD-UmuD fusion proteins. Previous in vitro studies with UmuD and UmuD9 proteins have shown the same specificity of interactions and the same efficiency of dimerization for the two proteins (2, 30). In the experiment whose results are shown in Table 1, we observed that the UmuD-UmuD9 interaction was sensitive to the direction in which it was tested. UmuD fused to the DNA-binding domain (GBD) of GAL4 (pGBT9umuD) was more active in binding UmuD9 fused to the trans-activation domain (GAD) of GAL4 (pGAD424umuD9) than was GAD-UmuD (pGAD424 umuD) in binding to GBD-UmuD9 (pGBT9umuD9). Similar disparities are often seen in studies such as this (for example, see references 18, 20, and 24). These differences probably reflect alterations in folding or binding when tested proteins are fused to either GAD or GBD of GAL4. We are aware that all of the gene fusions used in this study are made at the N termini of the tested proteins. Thus, the conformation of the N termini of these hybrid proteins and, in turn, the physical properties of particular fusion proteins may be affected. However, in spite of the limitations of the yeast two-hybrid system, our results seem to indicate that in yeast cells both bacterial fusion proteins tested are able to form specific protein-protein associations. Thus, the heterologous system we are using is adequate for the study of interactions between bacterial proteins. It has been postulated that the formation and stability of the different types of UmuD and UmuD9 dimers might have pro-
TABLE 3. Interaction of UmuD9 fusion protein with different fragments of UmuC protein: quantitative assay of b-galactosidase activity Plasmids
b-Galactosidase activitya
pGBT9umuC(1-422) 1 pGAD424umuD9.................. 516 6 134 pGBT9D1umuC(1-113) 1 pGAD424umuD9 .............. ,2 pGBT9D2umuC(1-210) 1 pGAD424umuD9 .............. ,2 pGBT9D3umuC(1-272) 1 pGAD424umuD9 .............. ,2 pGBT9D4umuC(1-396) 1 pGAD424umuD9 .............. ,2 pGBT9D5umuC(115-422) 1 pGAD424umuD9 .......... ,2 pGBT9D6umuC(14-422) 1 pGAD424umuD9 ............ ,2 b
a
Determined as described in Table 1, footnote b. Numbers in parentheses indicate the amino acids of the UmuC protein present in the GAL4-UmuC fusion protein. b
VOL. 178, 1996
UmuD, UmuD9, AND UmuC PROTEIN-PROTEIN INTERACTIONS
TABLE 4. Interaction of UmuD and UmuD9 fusion proteins with wild-type or mutant UmuC proteins: quantitative assay of b-galactosidase activity Plasmids
b-Galactosidase activitya
pGBT9umuC 1 pGAD424umuD9 ............................... 516 6 134 pGBT9umuC 1 pGAD424umuD................................. ,2 pGBT9-3umuC25 1 pGAD424umuD9 ........................ ,2 pGBT9-3umuC25 1 pGAD424umuD.......................... ,2 pGBT9-3umuC36 1 pGAD424umuD9 ........................ ,2 pGBT9-3umuC36 1 pGAD424umuD.......................... ,2 pGBT9-3umuC104 1 pGAD424umuD9 ...................... 535 6 88 pGBT9-3umuC104 1 pGAD424umuD........................ ,2 a
Determined as described in Table 1, footnote b.
found biological consequences. It has been hypothesized that intact UmuD is not simply an inactive form of UmuD9 but might serve as a dominant inhibitor of UmuD9-dependent mutagenesis (2, 30). Both the in vitro experiments (2, 30) and our results indicate that the UmuD-UmuD9 complex is stronger and/or more stable than either of the two homodimers. If it is assumed that the observed differences in the strength of dimerization reflect the in vivo situation in E. coli cells, whether these differences have biological consequences can be questioned. We take into consideration several possibilities. (i) It has been shown that both LexA and l repressor appear to be in their monomeric form while they are undergoing RecAmediated cleavage (10). Our data indicate that the interaction within UmuD-UmuD is the weakest one (Table 1). Thus, it is likely that the frequent, spontaneous monomerization of the UmuD-UmuD complex occurs and consequently may result in a more efficient RecA-dependent processing of UmuD at the early stages of SOS induction. (ii) At later steps of SOS induction, the preferential generation of a strong and/or stable UmuD-UmuD9 heterodimer could prevent excessive processing of UmuD (2). (iii) The formation of the UmuD-UmuD9 heterodimer could also play an important role in turning off the capacity for SOS mutagenesis by sequestering UmuD9 (2). (iv) UmuD-UmuD9 heterotypic interaction could influence the formation of the UmuC-UmuD9 complex (see below). The results of previous in vitro studies (2, 54) suggest that UmuC interacts with both UmuD and UmuD9 proteins. Our data (Table 2) clearly indicate that, in the two-hybrid system, under in vivo conditions, UmuC fusion protein is capable of interacting only with UmuD9 and not with UmuD. Thus, this finding is consistent with the concept that posttranslational processing of UmuD to UmuD9 is probably a critical step in SOS mutagenesis, enabling the latter protein to interact with UmuC. However, the fact that we could not detect an interaction between UmuD and UmuC fusion proteins in a heterologous system, such as the yeast two-hybrid system, does not rule out the possibility that a UmuC-UmuD complex may form in E. coli cells. It is known that the umuDC operon is tightly controlled by LexA, as would be expected of one of the last operons to be induced by DNA damage (27). Processing of UmuD to mutagenically active UmuD9 is also an inefficient reaction (6, 53). The low efficiency of this reaction might provide E. coli cells additional time to repair DNA damage via an error-free pathway before being committed to the error-prone translesion DNA synthesis (6, 53). We postulate that the formation of a strong but probably mutagenically inactive UmuD-UmuD9 complex would provide an additional mechanism modulating the ability of E. coli to carry out SOS mutagenesis. We propose that this mechanism is mediated by depletion of a pool of the
2583
free UmuD9 which is available to interact with UmuC protein. In consequence, the probability of formation of UmuC-UmuD9 complex is decreased. We have attempted to use the two-hybrid system to identify the region of UmuC protein responsible for interaction with UmuD9 fusion protein. A series of GAL4-DUmuC fusions has been constructed. We have tested their ability to interact with UmuD9 fusion protein. Our results show that none of the analyzed GAL4-DUmuC fusion proteins is able to associate with UmuD9 protein (Table 3). These results suggest that the integrity of the entire UmuC sequence is essential for UmuCUmuD9 heterotypic interaction. This conclusion is in agreement with our finding that formation of the UmuC-UmuD9 protein complex occurs only when UmuC has been fused to the DNA-binding domain of GAL4 protein. No interaction occurs when UmuC has been fused to the trans-activation domain (GAD) of GAL4 protein (Table 2). The mechanism underlying such directionality is unclear, but this phenomenon has been observed for numerous protein pairs in the two-hybrid system (18, 36). Directionality of transcription activation of the reporter gene might be explained by positing that fused GAL4 amino acids cause some conformational perturbation or a steric hindrance within the binding sites of tested proteins. Our data seem to indicate that proper folding of UmuC protein is a critical factor for the ability of UmuC protein to associate with UmuD9. The presence of additional GAD amino acids at the N terminus of the UmuC protein might negatively interfere with the folding process and with the ability of UmuC protein to acquire an appropriate tertiary structure. Genetic and physiological experiments indicate that the products of umuDC genes are essential components of the SOS mutation pathway (26, 47). All of the chromosomal umuDC mutants isolated to date are nonmutable by UV light and many chemical agents. If physical association between UmuC and UmuD9 proteins is required in an SOS mutagenic process, it is reasonable to assume that at least some umuC mutations might affect interaction between UmuC and UmuD9. In our studies, we have used three UmuC mutant proteins, UmuC25, UmuC36, and UmuC104, each resulting from the presence of a missense mutation (28). Our analysis of the mutant UmuC proteins revealed that they have different abilities to interact with UmuD9 fusion protein. We have found that UmuC25 and UmuC36 proteins are not capable of associating with UmuD9 (Table 4). The umuC36 mutation (Glu753Lys) and the umuC25 mutation (Thr-2903Lys) lie in the two distinct and highly conserved regions of the UmuC-like proteins. One subdomain is in the amino-terminal portion between amino acid residues 70 and 110, and the other is in the carboxy terminus between residues 271 and 326 (28, 56). Our results suggest that amino acid residues Glu-75 and Thr-290, which are located in these subdomains, might be crucial for the ability of UmuC to interact with UmuD9. Our results seem to explain the nonmutable phenotype of the E. coli cells that carry either a umuC25 or a umuC36 mutation. However, we cannot exclude the possibility that the presence of a umuC25 or umuC36 mutation affects the stability of the mutant UmuC protein by changing its interaction with chaperone protein complexes Hsp60 and Hsp70 (40). An intriguing observation is that UmuC104 mutant protein interacts with UmuD9 protein with an efficiency identical to that of the wild-type UmuC protein (Table 4). This suggests that the nonmutability of the umuC104 mutant is due to a deficiency in a function of UmuC other than its interaction with UmuD9. This function may involve interaction with a specific protein or DNA. The umuC104 mutation (Asp-
2584
JONCZYK AND NOWICKA
1013Asn) lies in a block of five amino acids (YSIDE) that is 100% conserved in all known UmuC homologs (28, 55). An attractive possibility is that these five conserved amino acids of UmuC-like proteins are involved in a specific interaction with another protein which is distinct from UmuD9 and which is essential for the SOS mutagenesis pathway (e.g., DNA polymerase). A mutation (e.g., umuC104) that abolishes interaction with this hypothetical protein leads to the nonmutable phenotype, as do the mutations which make UmuC protein unable to interact with UmuD9 (e.g., umuC25 or umuC36). We are designing a protocol to isolate the extragenic suppressors of umuC104. This approach is likely to enable identification of a protein or proteins which interact with UmuC in addition to UmuD9. We believe that the usefulness of the yeast two-hybrid system for testing E. coli proteins should substantially complement genetic and in vitro studies of proteins involved in SOS mutagenesis. ACKNOWLEDGMENTS We are indebted to Z. Cies´la for his support and for critical reading of the manuscript; we thank I. Fijalkowska, M. Hryniewicz, M. Skoneczny, and the members of our research group for many helpful discussions and R. Woodgate, G. C. Walker, and S. Elledge for plasmids and the yeast strain. This work was supported by KBN grant 6 PO4A 043 09. REFERENCES 1. Bailone, A., S. Sommer, J. Knezevic, M. Dutreix, and R. Devoret. 1991. A RecA protein mutant deficient in its interaction with the UmuDC complex. Biochimie 73:479–484. 2. Battista, J. R., T. Ohta, T. Nohmi, W. Sun, and G. C. Walker. 1990. Dominant negative umuD mutations decreasing RecA-mediated cleavage suggest roles for intact UmuD in modulation of SOS mutagenesis. Proc. Natl. Acad. Sci. USA 87:7190–7194. 3. Blanco, M., G. Herrera, P. Collado, J. E. Rebollo, and L. M. Botella. 1982. Influence of RecA protein on induced mutagenesis. Biochimie 64:633–636. 4. Bridges, B. A., R. P. Motershead, and S. G. Sedgwick. 1976. Mutagenic DNA repair in Escherichia coli. Mol. Gen. Genet. 144:53–58. 5. Bridges, B. A., and R. Woodgate. 1985. Mutagenic repair in Escherichia coli: products of the recA gene and of the umuD and umuC genes act at different steps in UV-induced mutagenesis. Proc. Natl. Acad. Sci. USA 82:4193–4197. 6. Burckhard, S. E., R. Woodgate, R. H. Scheuerman, and H. Echols. 1988. UmuD mutagenesis protein of Escherichia coli: overproduction, purification and cleavage by RecA. Proc. Natl. Acad. Sci. USA 85:1811–1815. 7. Chen, D., B. Yang, and T. Kuo. 1992. One-step transformation of yeast in stationary phase. Curr. Genet. 21:83–84. 8. Chien, C., P. L. Bartel, R. Sternglanz, and S. Fields. 1991. The two-hybrid system: a method to identify and clone genes for proteins that interact with a protein of interest. Proc. Natl. Acad. Sci. USA 88:9578–9582. 9. Clark, J. M. 1988. Novel non-template nucleotide addition reactions catalysed by prokaryotic and eucaryotic DNA polymerases. Nucleic Acids Res. 16:9677–9686. 10. Cohen, S., J. Knoll, J. W. Little, and D. W. Mount. 1981. Preferential cleavage of phage l repressor monomers by recA protease. Nature (London) 294:182–184. 11. Dutreix, M., B. Burnett, A. Bailone, C. M. Radding, and R. Devoret. 1992. A partially deficient mutant, recA1730, that fails to form normal nucleoprotein filaments. Mol. Gen. Genet. 232:489–497. 12. Dutreix, M., P. L. Moreau, A. Bailone, F. Galibert, J. R. Battista, G. C. Walker, and R. Devoret. 1989. New recA mutations that dissociate the various RecA protein activities in Escherichia coli provide evidence for the additional role for RecA protein in mutagenesis. J. Bacteriol. 171:2415–2423. 13. Echols, H., and M. F. Goodman. 1990. Mutation induced by DNA damage: a many protein affair. Mutat. Res. 236:301–311. 14. Echols, H., and M. F. Goodman. 1991. Fidelity mechanisms in DNA replication. Annu. Rev. Biochem. 60:477–511. 15. Elledge, S. J., and G. C. Walker. 1983. Proteins required for UV and chemical mutagenesis: identification of the products of the umuC locus of Escherichia coli. J. Mol. Biol. 164:175–192. 16. Ennis, D. G., B. Fisher, S. Edmiston, and D. W. Mount. 1985. Dual role for Escherichia coli RecA protein in SOS mutagenesis. Proc. Natl. Acad. Sci. USA 82:3325–3329. 17. Ennis, D. G., N. Ossana, and D. W. Mount. 1989. Genetic separation of Escherichia coli recA functions for SOS mutagenesis and repressor cleavage.
J. BACTERIOL. J. Bacteriol. 171:2533–2541. 18. Estojak, J., R. Brent, and E. A. Golemis. 1995. Correlation of two-hybrid affinity data with in vitro measurements. Mol. Cell. Biol. 15:5820–5829. 19. Fields, S., and O. Song. 1989. A novel genetic system to detect proteinprotein interactions. Nature (London) 340:245–246. 20. Fields, S., and R. Sternglanz. 1994. The two-hybrid system: an assay for protein-protein interactions. Trends Genet. 10:286–292. 21. Frank, E. G., J. Hauser, A. S. Levine, and R. Woodgate. 1993. Targeting of the UmuD, UmuD9 and MucA9 mutagenesis proteins to DNA by RecA protein. Proc. Natl. Acad. Sci. USA 90:8169–8173. 22. Friedberg, E. C., G. C. Walker, and W. Siede. 1995. DNA repair and mutagenesis, p. 407–522. ASM Press, Washington, D.C. 23. Hagenesee, M. E., T. Timme, S. K. Bryan, and R. E. Moses. 1987. DNA polymerase III of Escherichia coli is required for UV and ethyl methanesulfonate mutagenesis. Proc. Natl. Acad. Sci. USA 84:4195–4199. 24. Hong, Y., K. Levay, J. F. Murphy, P. G. Klein, J. G. Shaw, and A. G. Hunt. 1995. A potyvirus polymerase interacts with the viral coat protein and Vpg in yeast cells. Virology 214:159–166. 25. Horii, T., T. Ogawa, T. Nakatani, T. Hase, H. Matsubara, and H. Ogawa. 1981. Regulation of SOS functions: purification of E. coli LexA protein and determination of its specific site cleaved by the RecA protein. Cell 27:515–522. 26. Kato, T., and Y. Shinoura. 1977. Isolation and characterization of mutants of Escherichia coli deficient in induction of mutations by ultraviolet light. Mol. Gen. Genet. 156:121–131. 27. Kitagawa, Y., E. Akaboshi, H. Shinagawa, T. Horii, H. Ogawa, and T. Kato. 1985. Structural analysis of the umu operon required for inducible mutagenesis in Escherichia coli. Proc. Natl. Acad. Sci. USA 82:4336–4340. 28. Koch, W. H., D. G. Ennis, A. S. Levine, and R. Woodgate. 1992. Escherichia coli umuDC mutants: DNA sequence alternations and UmuD cleavage. Mol. Gen. Genet. 233:443–448. 29. Kroeker, W. D., D. Kowalski, and M. Laskowski. 1976. Mung bean nuclease. I. Terminally directed hydrolysis of native DNA. Biochemistry 15:4463–4467. 30. Lee, M. H., T. Ohta, and G. C. Walker. 1994. A monocysteine approach for probing the structure and interactions of the UmuD protein. J. Bacteriol. 176:4825–4837. 31. Little, J. W. 1983. The SOS regulatory system: control of its state by the level of the RecA protease. J. Mol. Biol. 167:791–808. 32. Little, J. W., S. H. Edmiston, Z. Pacelli, and D. W. Mount. 1980. Cleavage of the Escherichia coli lexA protein by the recA protease. Proc. Natl. Acad. Sci. USA 77:3225–3229. 33. Little, J. W., and D. W. Mount. 1982. The SOS regulatory system of Escherichia coli. Cell 29:11–22. 34. Marchuk, D., M. Drumm, A. Saulino, and F. S. Collins. 1991. Construction of T-vectors, a rapid and general system for direct cloning of unmodified PCR products. Nucleic Acids Res. 19:1154. 35. Marsh, L., and G. C. Walker. 1985. Cold sensitivity induced by overproduction of umuDC in Escherichia coli. J. Bacteriol. 162:155–161. 36. Mosteller, R. D., W. Park, and D. Broek. 1995. Analysis of interaction between Ras and CDC25 guanine nucleotide exchange factor using yeasts GAL4 two-hybrid system. Methods Enzymol. 255:135–148. 37. Nohmi, T., J. R. Battista, L. A. Dodson, and G. C. Walker. 1988. RecAmediated cleavage activates UmuD for mutagenesis: mechanistic relationship between transcriptional derepression and the posttranslational activation. Proc. Natl. Acad. Sci. USA 85:1816–1820. ´ ledziewska-Go´jska, and Z. Cies´la. 1994. 38. Nowicka, A., M. Kanabus, E. S Different UmuC requirements for generation of different kinds of UVinduced mutations in Escherichia coli. Mol. Gen. Genet. 243:584–592. 39. Perry, K. L., S. J. Elledge, B. B. Mitchell, L. Marsh, and G. C. Walker. 1985. umuDC and mucAB operons whose products are required for UV light and chemical induced mutagenesis: UmuD, MucA, and LexA proteins share homology. Proc. Natl. Acad. Sci. USA 82:4331–4335. 40. Petit, M.-A., W. Bedale, J. Osipiuk, C. Lu, M. Rajagopalan, P. McInerney, M. F. Goodman, and H. Echols. 1994. Sequential folding of UmuC by the Hsp70 and Hsp60 chaperone complexes of Escherichia coli. J. Biol. Chem. 269:23824–23829. 41. Rajagopalan, M., C. Lu, R. Woodgate, M. O’Donnell, M. F. Goodman, and H. Echols. 1992. Activity of the purified mutagenesis proteins UmuC, UmuD9 and RecA in replicative bypass of an abasic DNA lesion by DNA polymerase III. Proc. Natl. Acad. Sci. USA 89:10777–10781. 42. Rose, M. D., F. Winston, and P. Hieter. 1990. Methods in yeast genetics. A laboratory course manual. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y. 43. Sambrook, J., E. F. Fritsch, and T. Maniatis. 1989. Molecular cloning: a laboratory manual, 2nd ed. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y. 44. Sassanfar, M., and J. W. Roberts. 1990. Nature of SOS-inducing signal in Escherichia coli: the involvement of DNA replication. J. Mol. Biol. 212:79–96. 45. Shinagawa, H., H. Iwasaki, T. Kato, and A. Nakata. 1988. RecA proteindependent cleavage of UmuD protein and SOS mutagenesis. Proc. Natl. Acad. Sci. USA 85:1806–1810. 46. Shinagawa, H., T. Kato, T. Ise, K. Makino, and A. Nakata. 1983. Cloning and characterization of the umu operon responsible for inducible mutagenesis in
VOL. 178, 1996
UmuD, UmuD9, AND UmuC PROTEIN-PROTEIN INTERACTIONS
Escherichia coli. Gene 23:167–174. 47. Steinborn, G. 1978. Uvm mutants of Escherichia coli K12 deficient in UV mutagenesis. I. Isolation of uvm mutants and their phenotypical characterization in DNA repair and mutagenesis. Mol. Gen. Genet. 165:87–93. 48. Sweasy, J. B., E. M. Witkin, N. Sinha, and V. Roegner-Maniscalco. 1990. RecA protein of Escherichia coli has a third essential role in SOS mutator activity. J. Bacteriol. 172:3030–3036. 49. Villani, G., S. Bouteux, and M. Radman. 1978. Mechanism of ultravioletinduced mutagenesis: extent and fidelity of in vitro DNA synthesis on irradiated templates. Proc. Natl. Acad. Sci. USA 75:3037–3041. 50. Vojtek, A. B., M. Hollenberg, and J. A. Cooper. 1993. Mammalian Ras interacts directly with the serine/threonine kinase Raf. Cell 74:205–214. 51. Walker, G. C. 1984. Mutagenesis and inducible responses to deoxyribonucleic acid damage in Escherichia coli. Microbiol. Rev. 48:60–93.
2585
52. Witkin, E. M. 1976. Ultraviolet mutagenesis and inducible DNA repair in Escherichia coli. Bacteriol. Rev. 40:869–907. 53. Woodgate, R., and D. G. Ennis. 1991. Levels of chromosomally encoded Umu proteins and requirements for in vivo UmuD cleavage. Mol. Gen. Genet. 229:10–16. 54. Woodgate, R., M. Rajagopalan, C. Lu, and H. Echols. 1989. UmuC mutagenesis protein of Escherichia coli: purification and interaction with UmuD and UmuD9. Proc. Natl. Acad. Sci. USA 86:7301–7305. 55. Woodgate, R., and S. G. Sedgwick. 1992. Mutagenesis induced by bacterial UmuDC proteins and their plasmid homologues. Mol. Microbiol. 6:2213– 2218. 56. Woodgate, R., M. Singh, O. I. Kulaeva, E. G. Frank, A. S. Levine, and W. H. Koch. 1994. Isolation and characterization of novel plasmid-encoded umuC mutants. J. Bacteriol. 176:5011–5021.
