Mutational Analysis of the PapB Transcriptional Regulator in Escherichia coli REGIONS IMPORTANT FOR DNA BINDING AND OLIGOMERIZATION*

PapB is a transcriptional regulator in the control of pap operon expression in Escherichia coli . There are PapB homologous proteins encoded by many fimbrial gene systems that are involved in the regulation of fim-briae-adhesin production, and previous studies suggested that PapB binds DNA through minor groove contact. Both deletion and alanine-scanning mutagenesis were used to identify functionally important regions of the PapB protein. Mutations altering Arg 61 or Cys 65 caused deficiency in DNA binding, indicating that these residues are critical for PapB binding to DNA. Alanine substitutions at positions 35–36, 53–56, and 74–76 resulted in mutants that were impaired in oligomerization. All these amino acid residues are conserved among the PapB homologous proteins, suggesting their importance in the whole family of regulatory proteins. The transcriptional efficiency of all the mutants was clearly reduced as compared with that of wild-type PapB. Taken together, we have localized regions in the PapB protein that are involved in DNA binding and oligomerization, and our results show that both functions are required for its activity as a transcriptional regulator. The PapB protein is a transcriptional regulator of the pap promoter; at certain levels it stimulates pap operon expression, but at higher levels it represses (1). There are several PapB homologous proteins that are involved in the regulation of the production of different fimbrial adhesins. SfaB is a positive regulator for the production of S fimbriae (2). The products of the fanA and fanB genes both show resemblance to PapB and are suggested to act as transcriptional antiterminators involved in the control obtain truncated used truncated used obtain truncated repressor produce a aa fused

The PapB protein is a transcriptional regulator of the pap promoter; at certain levels it stimulates pap operon expression, but at higher levels it represses (1). There are several PapB homologous proteins that are involved in the regulation of the production of different fimbrial adhesins. SfaB is a positive regulator for the production of S fimbriae (2). The products of the fanA and fanB genes both show resemblance to PapB and are suggested to act as transcriptional antiterminators involved in the control of K99 fimbriae production (3). Biosynthesis of the Escherichia coli CS31A surface antigen is negatively controlled by the PapB homolog, ClpB, together with LRP (4). In the pef operon, located on the 90-kilobase virulence plasmid in Salmonella typhimurium, the gene pefB is postulated to encode a PapB-like protein (5). Among all of these PapB-like proteins, little is known about their mode of action or the actual molecular mechanism by which fimbrial expression is affected.
The PapB protein is composed of 104 amino acids (aa), 1 and its molecular size according to SDS-polyacrylamide gel electro-phoresis analysis is approximately 11 kDa (6). The aa sequence of the PapB protein does not contain any obvious recognition motifs that are common in other DNA binding proteins. It was recently shown that the PapB protein recognizes a DNA structure, which includes T/A triplets in terms of 9-base pair repeats, and the evidence suggests that it binds to DNA through minor groove contact in an oligomeric fashion (7). This novel DNA-binding mode could be important for PapB homologous proteins to function as transcriptional regulators of different fimbrial operons.
In the present study, both truncated derivatives and variants altered by alanine-scanning mutagenesis were used to identify residues important for DNA binding and oligomerization of PapB protein. The efficiency of such mutants in transcriptional activation of the pap operon was also tested.
DNA Techniques-Plasmid isolation, gel electrophoresis, transformation, amplification of DNA by PCR, and DNA labeling were performed by standard procedures (15). Restriction endonuclease digestions and DNA ligation reactions were performed under conditions recommended by the manufacturers (Roche Molecular Biochemicals, New England Biolabs Inc.).
Expression and Purification of Wild Type and Mutant PapB Proteins with the 6ϫHis Affinity Tag-The plasmids containing wt and mutant papB genes (based on pQE30) were introduced into JM109, and protein expression was induced by addition of IPTG (final concentration 1 mM) at the logarithmic phase. The His-tagged wt and mutant PapB proteins were purified as described previously (7).
Anti-PapB Antisera and Immunoblotting-Polyclonal antisera with antibodies recognizing PapB were produced by immunization of rabbits with purified PapB protein. The PapB antibody was purified as described by Taraseviciene et al. (16). For detection of PapB by immunoblotting, we used a procedure essentially as described before (17). ECL Western blot was performed according to the manufacturer's protocol.
Analysis of Protein-DNA Interactions-Gel mobility shift assays to detect protein-DNA interaction were performed as described previously (1,18). DNA fragments containing the PapB binding site 1 of the pap operon (1, 7) were obtained by end-labeling purified PCR products with [␥-32 P]ATP and T4 polynucleotide kinase. Oligonucleotides 740 (5Ј-CC-CAAGCTTAATCCGTTACCGCCAGCGCCT-3Ј) and 1140 (5Ј-CCCAAG-CTTTAAACGATCTTTTAACCCACAAAAC-3Ј) were used as the two primers, while plasmid pHMG1 (19) was used as the template in the PCR amplification. The purified wt or mutant PapB proteins with 6ϫHis tag were mixed with 32 P-end-labeled DNA fragments (5000 -10,000 cpm) in the presence of 0.5 g of poly(dI-dC) and 50 mM KCl in buffer B (25 mM Hepes, pH 7.5, 0.1 mM EDTA, 5 mM dithiothreitol, 10% glycerol) in a final volume of 10 l. The reaction mixtures were incubated at 25°C for 15 min and then immediately loaded onto 8% polyacrylamide-bisacrylamide (37.5:1) for electrophoresis.
␤-Galactosidase Assay-To measure the ␤-galactosidase specific activity, we used the method described by Miller (20). All data represent the average values obtained from at least three separate experiments.
Chloramphenicol Sensitivity Test-Cm sensitivity was tested by plating dilutions from a fresh overnight culture on LB plates containing Cm at 0, 10, 25, 50, 100, 150, and 200 g/ml (21). Sensitivity was expressed as the Cm concentration at which the plating efficiency of the cells fell below 50%.
Superinfection Immunity Test-E. coli 71-18 lacI q cells (8), transformed with the constructs described in Table I (those cloned into pBF21), were grown at 37°C to an A 620 nm of 0.6 in LB medium containing Cb. The chimeric proteins, consisting of the N-terminal (DNA binding) domain of cI repressor fused to a desired protein, were induced by addition of 250 l of 100 mM IPTG to 0.4-ml portions of the culture. Following this induction, the cells were infected by addition of 3 l of phage 146 (5 ϫ 10 10 plaque-forming units/ml) (22) diluted 1/10,000 just before infection in 10 mM Tris-HCl, pH 8.0, buffer containing 10 mM MgSO 4 . After a 5-min incubation at 20°C, 5 ml of LB containing Cb (100 g/ml) and 10 mM MgSO 4 were added to the infected cultures, which were then transferred to a 37°C shaker bath. At the indicated times, 0.4-ml aliquots were removed, centrifuged for 5 min to pellet the cells, and 150 l of supernatant were mixed with 30 l of chloroform. The number of phages in each supernatant was determined by the quantitative titer method (23) using E. coli 71-18 lacI q as recipient strain.
In Vitro Cross-linking-In vitro cross-linking was carried out as described by Ueguchi et al. (24). Strains carrying plasmids expressing His-tagged wt or mutant PapB were grown at 37°C to the mid-logarithmic phase in LB in the presence of 1 mM IPTG. Cells were harvested from 5 ml of the culture, washed with 10 mM Tris-HCl, pH 8.0. After freeze-thawing in the same buffer, cells were collected and resuspended in 500 l of a cross-linking buffer (1 M triethanolamine-HCl, pH 8.5, 0.25 M NaCl, 5 mM dithiothreitol). Cell extracts were then prepared by sonication, followed by ultracentrifugation (100,000 ϫ g, 30 min). These samples were incubated at room temperature in 200 l of the crosslinking buffer with or without dimethyl suberimidate (1 mg/ml). From the reaction mixtures, proteins were precipitated with trichloroacetic acid, and then analyzed by SDS-polyacrylamide gel electrophoresis. Protein bands were detected by immunoblotting with purified polyclonal anti-PapB antiserum.
Test of PapB Efficiency as Transcriptional Activator-Plasmids expressing wt or mutant PapB were introduced into strain MC1029 together with plasmid pHMG15 pap (I ϩ , B1) A: lacZ. All these strains were grown at 37°C to the mid-logarithmic phase in LB with or without IPTG induction (final concentration 1 mM). To test how efficient PapB protein derivatives could activate transcription, we monitored the expression of ␤-galactosidase from pHMG15, and the expression level obtained with wt-PapB in absence of IPTG was set to 1.0.

Regions Involved in DNA Binding-
The amino acid sequence of the PapB protein does not contain any obvious recognition motifs that are common in other DNA-binding proteins. The computer prediction of the secondary structure of PapB suggested a hydrophilic region (aa 61-70) (Fig. 1A), which, in a comparison with PapB-like proteins from other fimbrial gene systems (Fig. 2), seemed to contain several conserved residues and could be important for DNA binding. To test this prediction, different PCR fragments containing wt and truncated papB gene(s) were cloned into the QIAexpress vector PQE 30 ( Table I). The wt and truncated PapB proteins with a 6ϫHis affinity tag attached to the N-terminal were purified through a Ni-NTA column (Fig. 3A), and were analyzed for specific binding to site 1 in the J96 pap-pili regulatory region (1). As shown in Fig Arg 61 and Cys 65 Are Critical for PapB Binding to DNA-Alignments of the amino acid sequence of proteins in the PapB family showed some regions of strong conservation (Fig. 2). In order to further identify the critical residues, alanine-scanning mutagenesis was applied to these regions. Alanine was chosen as the replacement residue, since it is tolerated in both hydrophobic and hydrophilic environments. Importantly, this change should function to remove a small region of charge or a large side chain without affecting its potential for being correctly folded and assembled (25,26). The alanine codons utilized in the mutagenesis were GCA and GCT. The PCR fragments containing different mutant papB genes were cloned into pQE30 (Table I). The His-tagged mutant PapB proteins were purified through a Ni-NTA column (Fig. 3B) and analyzed for DNA binding. The in vitro binding studies indicated that most of the mutants could still bind to DNA. In several cases the DNA binding was clearly weaker, but in some cases DNA binding occurred to the same extent as with wt-PapB. Examples of the binding patterns in gel mobility shift assays are shown in Fig. 4 for selected mutants with different DNA binding affinities. These results are summarized below (Fig. 7) for all the mutants tested. The two mutants harboring alanine substitutions at positions 61 or 65 clearly showed a defective DNA binding (Fig. 4B). To test if the -SH group of Cys 65 contributed to the DNA binding, the PapB protein was modified by incubation together with iodoacetamide, which could block the -SH group of Cys 65 . The modified PapB completely lost the DNA binding ability (data not shown), suggesting the importance of the sulfhydryl-group of Cys 65 in DNA binding. Taken together, we propose that Arg 61 and Cys 65 are most essential for PapB binding to DNA.
Oligomerization State of Wild-type (wt) PapB-The wt-PapB protein has a tendency to form large, insoluble aggregates, and previous studies have indicated that PapB could act as an oligomer and could protect regions as long as 30 -70 nucleotides (7). However, both the even numbers of 9-base pair repeats in the PapB binding site(s) and the results of binding saturation experiments suggested that PapB might act as a dimer or a tetramer in vivo (7).
To investigate the active form of PapB as a transcriptional regulator, a test to determine the actual capability of the PapB molecules to oligomerize in vivo was devised. Based on plasmid pBF21, which carries the bacteriophage cI repressor gene under the control of a tandemly repeated lacUV5 promoter, and pBF22 in which the DNA encoding the oligomerization domain of the repressor (C-terminal domain) was deleted (14), we constructed plasmid pYN51 encoding the chimeric protein consisting of the N-terminal DNA binding domain of the cI repressor and the wt-PapB molecule ( Table I). The test relies on the capability of an oligomerization-proficient protein domain to functionally replace the natural C-terminal domain of phage cI repressor conferring biological activity to the N-terminal DNA binding domain of the same repressor (21) (Fig. 5A). The oligomerization proficiency of a protein can be quantified by determining the number of lytic plaques formed following infection (with phage ) of E. coli cells harboring the pBF expression vectors and producing, upon induction with IPTG, the various types of chimeric cI repressor molecules (14). In cells expressing the entire wt-cI repressor or the chimeric protein consisting of the N-terminal DNA binding domain of cI repressor and wt-PapB, the number of plaques formed was much lower (superinfection immunity) as compared with those expressing only the N-terminal domain of the repressor (superinfection sensitivity) (Fig. 5 and Table II). The constructed plasmids were also introduced by transformation of strain JH607 (ϭ 112O s P s ), which was originally described by Beckett et al. (10) as a reporter for co-operative binding by repressor to adjacent operators. The weak operator site O s 2 overlaps the promoter P s , while the strong operator site O s 1 is positioned upstream of O s 2 and does not affect transcription by P s . Thus, co-operative binding to both operator sites increases the efficiency of repression for the downstream reporter genes cat and lacZ. Different repressor activities can be distinguished by chloramphenicol sensitivity and ␤-galactosidase activity. In cells expressing the entire wt-cI repressor, both of the reporters were significantly repressed as compared with those expressing only the N-terminal domain of the repressor. wt-PapB, on the other hand, could replace, to a large extent, the function of the C-terminal domain of the cI repressor and substantially repressed the expression of the two reporter genes. This gave rise to low ␤-galactosidase activity and Cm sensitivity (Table  II). This effect can only be due to a PapB-induced oligomerization of the DNA binding domain of the repressor, since a nonspecific interference of PapB with the reporter system was  ruled out by the finding that wt-PapB fused to a defective N-terminal domain of the cI repressor, as in the construct pYN69, was completely unable to repress the expression of the reporter genes (Table II) and prevent the lytic development of phage (Fig. 5). [53][54][55][56][74][75][76] Are Impaired in Oligomerization-To test the oligomerization states of different PapB mutants, we prepared a series of constructs (e.g. pYN52-pYN103) encoding the chimeric proteins consisting of the N-terminal DNA binding domain of cI repressor and mutant PapB molecules (Table I). By using the same oligomerization test system, it was found that PapB mutants with alanine substitutions at positions 35-36, 53-56, and 74 -76 did not function like wt-PapB to replace the C-terminal domain of the cI repressor and repressed the expression of the reporter genes (Table II). In cells expressing the chimeric proteins consisting of the N-terminal DNA binding domain of cI repressor and these PapB mutants, the number of plaques was as high as those expressing only the N-terminal domain of the repressor (Fig. 5). All the truncated PapB mutants showed the same phenotype (Table  II), suggesting that the tertiary structure is required for oligomerization. In contrast, the DNA binding-defective PapB mutants in which Arg 61 or Cys 65 were mutated displayed similar oligomerization capability as that of wt-PapB ( Fig. 5 and Table II).

PapB Mutants with Alanine Substitutions at
To further assess the different oligomerization states of wt and mutant PapB proteins, we performed in vitro cross-linking studies in the presence of cross-linker dimethyl suberimidate. Such cross-linking experiments showed that PapB indeed could form a kind of dimer in solution, together with a strong band on the top of the gel which might be the aggregates (Fig. 6). Similarly, a clear band about the dimer size was also observed for DNA binding-defective mutants PapB(R61A) and PapB-(C65A). However, the oligomerization defective mutants (according to the in vivo test) with alanine substitutions at positions 35-36, 53-56, and 74 -76 did not form dimers in the presence of dimethyl suberimidate, although most of the proteins aggregated (Fig. 6). The results of the in vivo co-operative binding and the in vitro cross-linking tests were consistent with each other, pointed to the importance of these aa residues in the oligomerization of PapB.
The DNA binding ability of these oligomerization-defective mutants was also tested. As shown in Fig. 4C, His-PapB(L35A,L36A) could still weakly bind to DNA, but no distinct shifted band was formed (indicated by "(ϩ)" in Fig. 7). The three other oligomerization-defective mutants, PapB(D53A, Y54A), PapB(L55A,V56A), and PapB(Y74A,F75A,S76A), showed similar DNA binding patterns (Fig. 7). The oligomerization capability might therefore be of importance for PapB to bind in a co-operative fashion to the DNA. Taken together, we were able to locate discrete regions in the PapB protein involved in DNA binding and others involved in oligomerization.
Both the DNA Binding and Oligomerization Capabilities Are Required for PapB to Function as a Transcriptional Activator-To test the in vivo activity of the different PapB mutants, the DNA fragments containing all the mutant papB genes were cloned into plasmid pMMB66EH in which expression is controlled by the IPTG-inducible tac promoter (12). The constructs were then introduced into the strain MC1029 together with the plasmid pHMG15 containing the pap (I, B1) A: lacZ operon fusion (6). The relative efficiencies as transcriptional activators of the different PapB mutants were monitored by their effects on expression of ␤-galactosidase activity. The expression level of PapB from the tac promoter vector in absence of inducer complemented the deficiency due to the papB1 mutation, and activated production of ␤-galactosidase from the papA: lacZ fusion. The ␤-galactosidase activity thereby obtained in the case of wt-PapB was set to 1.0 (Fig. 7). Induction of higher levels of PapB expression gave a reduced activation due to its autoregulatory properties (1). However, it was of interest in the present study to test if any of the mutant PapB proteins could cause activation if present at higher cellular levels. The test therefore was performed both in absence and in presence of the inducer IPTG. The results indicated that the transcriptional activities of PapB mutants defective in either the DNA binding or the oligomerization ability were clearly reduced as compared with wt-PapB (Fig. 7). In some cases (e.g. H43A,S44A, V47A, I48A, and K62A), overproduction of mutant protein by IPTG induction led to partially or fully restored capability to activate papA: lacZ expression. The E66A mutant, which was only partially defective in DNA binding, retained activating and repressing properties that were rather similar to that of wt-PapB. Taken together, we conclude from these studies that both the DNA binding and the oligomerization capabilities are required for PapB to fully function as an active transcriptional regulator. DISCUSSION In the present study, we investigated the regions essential for DNA binding and oligomerization of PapB by using both deletion and alanine-scanning mutagenesis. The gel mobility shift assays with different truncated PapB mutants showed that the predicted hydrophilic region (aa 61-70) is important for DNA binding (Figs. 1B and 4A). The alanine-scanning mutagenesis indicated that the conserved Arg 61 and Cys 65 are the two most critical residues required for DNA binding (Figs. 4B  and 7), and this is consistent with the fact that the iodoacet-amide-modified PapB showed defective DNA binding.
The in vivo oligomerization studies ( Fig. 5 and Table II) suggested that wt-PapB could function as an oligomer in vivo. This is also in keeping with our previous findings that the PapB protein interacts as an oligomer of 8 -10 subunits at J96 site 1 (7). There are other examples of DNA binding proteins that interact as oligomers at repeated DNA sequence motifs. For example, the E. coli oxidized OxyR protein recognizes the two-fold dyad symmetry sequence ATAGXtXXXaXCTATXXX-XXXXATAGXtXXXaXCTAT, and binds to four adjacent major grooves of the DNA helix as a tetramer (27,28). In the case of PapB, we propose that it binds to DNA by recognizing 9-base pair repeats with T/A triplets at conserved positions and by contacting two adjacent minor grooves in a tetrameric fashion.
As reviewed by Werner et al. (29), the interaction of minor groove intercalating proteins with DNA is primarily hydrophobic at the site of insertion. For TATA box-binding protein TBP (30 -32), the high mobility group domain proteins SRY (33) and LEF-1 (34), and for the E. coli purine repressor protein PurR (35,36), these nonpolar interactions extend over most of the interaction surface between the protein and the DNA. Base recognition may occur through van der Waals interactions between ␤or ␥-branched amino acids (Val, Ile, Leu) or the faces of aromatic amino acids (Phe or Tyr) and the minor groove edges and faces of the DNA bases. Formation of salt bridges or hydrogen bonds occurs between DNA phosphates and Lys or Arg residues of the proteins. In the above mentioned cases, relatively few hydrogen bonds between protein side chains and the DNA bases are present at the interface, with Ser, Thr, Asn, or Tyr forming base-specific hydrogen bonds (29). The integration host factor of E. coli contacts the DNA exclusively via the phosphodiester backbone and the minor groove (37). The crystal structure of an integration host factor-DNA complex indicates that several arginines are involved in minor groove con- FIG. 5. In vivo protein oligomerization assay. A, schematic illustration of the principle of the assay as described under "Results." Cells harboring the different constructs were induced to express wt or chimeric cI repressor molecules by addition of IPTG. Following infection of these cells with phage , lytic development of the phage is prevented in cells producing wt-cI repressor or chimeric cI repressor molecules composed of N-terminal DNA binding domain and wt-PapB, which can bind cooperatively to the DNA operator sites by virtue of protein-protein interactions. Cells producing a truncated repressor molecule lacking the C-terminal domain or chimeric cI repressor molecules composed of N-terminal DNA binding domain and oligomerization-defective mutant PapB, undergo a complete lytic cycle, producing a rapid burst of phages. B, number of plaques (plaque-forming units/ml) of phages released at the indicated times after infection of E. coli cells harboring different constructs. tact by making salt bridges or hydrogen bonds to conserved bases (38). However, the interaction of viral histone-like protein p6 with DNA is proposed to be mainly electrostatic, and Arg 6 is essential for the activity (39). Considering our present results, we suggest that Arg 61 and Cys 65 are involved in the minor groove contact of PapB to DNA. However, we can not determine what kind of interaction it is at the present time.
To identify the regions important for oligomerization, the conserved amino acid residues were mutated by alanine substitutions. The DNA binding-defective mutants R61A and C65A showed similar oligomerization ability as that of the wt-PapB (Figs. 5 and 6 and Table II). Alanine substitutions at positions 35-36, 53-56, and 74 -76 resulted in mutants that were impaired in oligomerization (Figs. 5 and 6 and Table II). All these oligomerization-defective mutants only retained weak DNA binding ability (Figs. 4C and 7), which suggested that the protein interaction is important for PapB to bind cooperatively to the DNA. The transcriptional activation efficiency of either the DNA-binding or the oligomerization-defective mutants was much weaker than that of wt-PapB (Fig. 7).
The E. coli H-NS is another relatively small protein (136 amino acids) known to bind DNA in an oligomeric fashion. It may act as a negative regulator of transcription (40,41), functioning by "transcriptional silencing" (42) and "repression via DNA topology" (41, 43) through oligomeric binding to DNA. Three putative functional domains have been identified in a Proteins that were tested included wild-type (wt) and the N-terminal DNA binding domain of cI repressor, chimeric proteins composed of the DNA binding domain of cI and wt or mutant PapB.
b Superinfection immunity (Imm.) means the number of plaques formed after infection was low (less than 10 5 pfu/ml), while superinfection sensitivity (Sens.) means the number of plaques was high (up to 10 10 pfu/ml).
c Chloramphenicol sensitivity was tested as described. Numbers indicate the Cm concentration at which the plating efficiency of the cells fell below 50%. d ␤-Galactosidase activity was tested as described under "Experimental Procedures." The value of strain JH607/pBF22 which produced the N-terminal DNA binding domain of the cI repressor was set as 1.0 for comparison. H-NS (44): the N-terminal domain involved in transcriptional repression, the central region involved in oligomerization, and the C-terminal domain involved in DNA binding. The possibility that H-NS oligomerization and DNA binding functions can be coupled has been highlighted recently (14,44). For a small protein like PapB (104 amino acids), it may be difficult to identify certain separate domains. Based on the results we have obtained, we conclude that we have identified the primary regions in the PapB protein involved in DNA binding and oligomerization. Functionally the two properties may be related to each other, and they are both required for it to function as an active transcriptional regulator. These functionally defined amino acid residues are conserved among all other PapB homologous proteins (Fig. 2), suggesting their importance in the structure and function of the whole PapB-family of proteins.