Residues 137 and 153 at the N terminus of the XylS protein influence the effector profile of this transcriptional regulator and the sigma factor used by RNA polymerase to stimulate transcription from its cognate promoter.

The 321-residue XylS and XylS1 proteins, encoded by the pWW0 and pWW53 plasmids respectively, differ in only 5 residues at positions 4, 53, 90, 137, and 153. As a result, the effector profile of XylS is wider than that of XylS1, and XylS mediates higher levels of transcription from its cognate-regulatable promoter than does XylS1. We generated a series of XylS-pWW0 mutants and found that the single mutants Asp-137-->Glu and His-153-->Asn exhibited an activation pattern different from that of the wild-type regulator. In the double-mutant XylSD137E,H153N the effector profile for benzoates was similar to that of XylS1. This suggests that these two residues are crucial for effector recognition and regulator activation to stimulate transcription. XylS-dependent transcription from its cognate promoter is mediated by RNA polymerase with sigma(32) or sigma(38), whereas XylS1 uses RNA polymerase with sigma(32) or sigma(70). We also found that point mutations at positions 137 and 153 of XylS led RNA polymerase to mediate transcription with sigma(70) rather than with sigma(38), as demonstrated by primer extension analysis in a sigma(70)-thermosensitive background proficient and deficient in sigma(38). This suggests that a positive transcriptional regulator can choose the RNA polymerase complex that mediates transcription from a given promoter.

Gene regulators respond to specific environmental, cellular, and other signals by stimulating or inhibiting transcription, translation, or some other event in gene expression so that the rate of synthesis of the gene product is appropriately modified (1,2). Research efforts in our laboratory have focused on the regulation of the TOL plasmid-encoded catabolic pathway for the metabolism of benzoate and alkylbenzoates. Genes encoding the TOL meta-cleavage pathway in pWW0 are grouped into a single operon whose expression is positively regulated at the level of transcription by the xylS gene product, which is activated by benzoate effectors (for review, see Ref. 3). Another well studied TOL plasmid is pWW53, which bears two functional homologous meta-pathway operons, together with two functional copies of the xylS regulatory genes (xylS1 and xylS3) (4 -7). XylS-pWW0 and XylS1-pWW53 differ in five amino acids, whereas XylS3 shows conserved sequence similarity with these two proteins only at their C-terminal end (6,7), where there is a bipartite DNA binding domain made of two ␣-helixturn-␣-helix motifs (as in the MarA, Rob, and AraC proteins) (8 -11). All three XylS isoforms are able to stimulate transcription from the Pm 1 promoter (7).
Both XylS and XylS1 mediate transcription with RpoH ( 32 ) in the early log phase (7,16); however, XylS-pWW0-dependent transcription from Pm is subsequently mediated by RNA polymerase with 38 (16), whereas in the case of XylS1 the factor used by RNA polymerase is 70 (7). Therefore, the minor differences at the protein sequence level between the two XylS proteins also impose dependence on the used by RNA polymerase. In light of these effects we decided to introduce stepwise mutations in XylS-pWW0 to determine their effects on transcriptional activity of the mutants.  (19), and E. coli UQ285 (thi lacZ4 argG75 rpoD285). This last strain has a short deletion at the rpoD gene so that 70 is nonfunctional at 42°C (19). We also used E. coli EEZ286, an rpoS mutant of UQ285 constructed after P1 transduction (20) of rpoS59::Tn10 into the UQ285 background (this study). Bacteria were grown at 30°C in Luria-Bertani medium supplemented, when required, with 100 g/ml ampicillin, 25 g/ml kanamycin, 50 g/ml streptomycin, 30 g/ml chloramphenicol, or 10 g/ml tetracy-cline. Growth was determined turbidometrically at 660 nm.
The following previously constructed plasmids were used. pCMX2 is a tetracycline resistance derivative of pSELECT containing the entire xylS gene inserted into the BamHI site (21). pERD100 is an IncQ group plasmid that carries a fusion of Pm to a promoterless ЈlacZ gene and encodes resistance to tetracycline (22). Plasmid pJLR107 has a 401base pair PstI fragment of the TOL plasmid containing the Pm promoter fused to a promoterless ЈlacZ gene in pMD1405 (pBR replicon) and encodes resistance to ampicillin (23). pLOW2 is a pACYC177 derivative, low copy number cloning vector that encodes resistance to kanamycin (24).
DNA Techniques-DNA preparation, digestion with restriction enzymes, and analysis by agarose gel electrophoresis, isolation of DNA fragments, ligations, transformations, transduction with P1 phage, and sequencing reactions were done according to standard procedures (25) or to the manufacturer's recommendations.
Construction of xylS Mutants by PCR-The xylS mutants were generated by overlap extension PCR mutagenesis (26) with internal oligonucleotide primers that exhibited one mismatch with respect to the wild-type sequence. The external primers were 5Ј-GGCACTGG-GATCGTTCAAGC-3Ј and 5Ј-GGATTTTTGCTTATTGAACG-3Ј. After DNA amplification, the resulting DNA was digested with XhoI and BglII, and the 761-base pair XhoI-BglII xylS mutant fragments were inserted between the XhoI-BglII sites of pCMX2 to yield plasmids pCMX2::xylS* (the asterisk indicates that one or more of the amino acids in the wild-type protein has been changed). In a first round of mutagenesis we generated single mutants at positions 4, 53, 90, 137, and 153 as indicated in Table I. These single mutants were used to construct the double mutants shown in Table I. All the xylS mutant alleles generated in this study were verified by DNA sequencing. Plasmids pCMX2::xylS* bearing the xylS* mutant alleles were digested with EcoRI and XbaI, and the 1609-base pair EcoRI-XbaI fragments, which contained the entire set of xylS* mutant alleles, were subcloned between the EcoRI and XbaI sites of pLOW2 to generate plasmids pLAR1 through 18, which encoded the mutant XylS proteins shown in Table I.
␤-Galactosidase Assays-E. coli bearing the wild-type xylS allele or xylS mutant alleles in pLARx (the x indicates a number between 1 and 18) plus pERD100 or pJLR107 were grown overnight in Luria-Bertani medium containing the appropriate antibiotics. Cultures were diluted 100-fold in the same medium supplemented or not with 1 mM benzoate analog. After 5 h of incubation, ␤-galactosidase activity was assayed in permeabilized whole cells as described previously (26).
Western Blot Analysis-Rabbit polyclonal antibodies against the XylS protein were produced and used to detect XylS proteins fractionated in SDS-PAGE gels with goat anti-rabbit antibodies conjugated to peroxidase according to the protocol described by Jones and Gregory (27).
RNA Preparation, Analysis, and Primer Extension-RNA was extracted with a modification of the guanidinium isothiocyanate-phenol method (16). The RNA concentration was determined by measuring A 260 . Hybridization of the 32 P-5Ј-end-labeled single-stranded DNA primer (10 5 cpm) complementary to the mRNA transcript produced from the Pm promoter and primer extension with avian myeloblastosis virus reverse transcriptase were done as described previously (16). In all assays 20 g of total RNA was used as the template. The oligonucleotide used in the reverse primer extension reactions was 5Ј-GATGT-GCTGCAAGGCGATTAAGTTG-3Ј. cDNA products were analyzed in a urea-polyacrylamide sequencing gel, and the intensities of the bands in the autoradiography were densitometrically determined. All assays were done at least three times, and nonsignificant differences were detected in the levels of the different extended products.

Response of the XylS Mutant Regulators to Benzoates Substituted at Position 3 on the Aromatic Ring-
The wild-type XylS regulator and the series of single point mutant regulators constructed by overlapping PCR mutagenesis were cloned in the low copy number plasmid pLOW2 as described above. The resulting plasmids were transformed in E. coli MC4100 (pERD100), and ␤-galactosidase activity in response to the addition of 3MB was determined (Table I). This benzoate analog was chosen in this first series of assays because it is the best effector for both XylS-pWW0 and XylS1-pWW53. The level of transcription mediated by the wild-type XylS-pWW0 increased about 40-fold with respect to the basal level, whereas the XylS1-pWW53 protein mediated only a 4-fold induction ( Table  I). Each of the single point mutants of XylS-pWW0 were also assayed, and the results showed little effect for the changes Cys-4 3 Arg, Cys-53 3 Gly, and Gly-90 3 Asp and a modest effect for the point mutants Asp-137 3 Glu and His-153 3 Asn (the activity decreased to about 50% that measured for the wild-type XylS-pWWO protein) ( Table I).
Given that the induction level mediated by XylS1 with 3MB was significantly lower than in the single mutants, we generated all possible double mutations. The level of induction of the double mutant XylSD137E,H153N was similar to that of XylS1 (Table I). The combination of either the Asp-137 3 Glu or His-153 3 Asn mutation with Cys-4 3 Arg resulted in levels of activity lower than that found with any of the single mutants. However, when Asp-137 3 Glu or His-153 3 Asn was combined with a mutation in position 53 or 90, the activity was similar to that seen with the Asp-137 3 Glu or His-153 3 Asn mutation alone (Table I).
This prompted us to analyze the combination of double mutant XylSD137E,H153N with the rest of the mutations. The result was that these mutants behaved like the double mutant and exhibited an induction level similar to that determined with XylS1-pWW53 (Table I). These results, therefore, suggest that residues 137 and 153 in XylS-pWW0 play a key role in effector recognition and activation of this regulator.
To determine whether the effect of the mutations was specific for 3MB, we tested the level of induction with two other substituted benzoate analogs that are effectors of XylS-pWW0, 3CB and 3-bromobenzoate. In general, 3CB and 3-bromobenzoate induced less ␤-galactosidase activity than 3MB with any of the mutants, and the reduction in the induction level for each of the mutants with these effectors followed a pattern similar to the decreases observed with 3MB as the effector (Table I).
In Vivo Stability of the Mutants-To test whether the decrease in activation reflected worse stability of the mutant  shown in the table) were grown on Luria-Bertani medium with tetracycline and kanamycin. ␤-Galactosidase activity was assayed in permeabilized cells as described under "Experimental Procedures." Induction ratio is defined as the enzyme level in cells grown in the presence of the inducer to the level determined in cells grown without an effector. Basal levels of ␤-galactosidase were between 30 H153N. We tested the effector profile of the mutant regulators with all possible mono-and disubstituted methylbenzoates. Fig. 1 shows the results for the wild-type XylS-pWW0. The order of strength of inducibility was 3MB Ͼ 2,3DMB Ͼ B Ͼ 3,4DMB Ͼ 2MB ϭ 4MB ϭ 2,5DMB Ͼ 3,5DMB ϭ 2,6DMB, with the last two dimethylbenzoates being considered noneffectors for XylS. Fig. 1 shows the induction profile of the single mutants XylSD137E and XylSH153N and the double mutant XylSD137E,H153N. The two single mutants recognized most of the effectors, but as with the benzoates substituted at position 3, all mutants exhibited lower levels of induction than the wild type. In contrast, the double mutant XylSD137E,H135N seemed not to respond to 3,4DMB, 2MB, or 4MB and exhibited an effector profile similar, if not identical, to that of the XylS1-pWW53 regulator (Fig. 1).
Characterization of a Collection of XylS Mutants at Positions 137 and 153-To further confirm the crucial role of residues 137 and 153 of XylS-pWW0 in effector recognition and regulator activation, we introduced as many mutations as possible by random overlapping mutagenesis at positions 137 and 153 in the XylS protein. We sequenced about 20 putative mutants for each position and found 5 new mutants at position 137 (aspartic acid was replaced by glutamic acid, proline, serine, arginine, and leucine) and two new mutants at position 153 (histidine was replaced by glycine and aspartic acid) (the mutant proteins were shown to be stable as deduced from the level of protein made by the cells and revealed by Western blot immunodetection of XylS proteins). We determined ␤-galactosidase expression from Pm mediated by each mutant in response to a series of substituted benzoates. Replacement of aspartic 137 with arginine or proline resulted in a mutant regulator unable to stimulate transcription with any benzoate effector (Table II), whereas replacement of aspartic acid by glutamic acid or serine reduced activation capacity to about 50% that observed for the wild type. Replacement with apolar leucine resulted in a mutant with very low activity. At position 153 replacement of histidine with glycine or aspartic acid resulted in mutants that showed no activity (Table II). These results suggest that residues 137 and 153 influence the ability of XylS to activate transcription from Pm.

Residues at the N Terminus of XylS Influence the Factor Used by RNA Polymerase to Stimulate Transcription from Pm; in Vivo Evidence for Transcription from Pm Mediated by RNA Polymerase with 70 by XylS-pWW0 Mutant Proteins-
Marqués et al. (16) showed that 3MB induces the heat shock response upon addition to cell cultures and that RNA polymerase uses 32 to mediate the XylS1-and XylS-dependent expression from Pm. Once the cultures reached the mid-log growth phase, RNA polymerase uses 38 to mediate transcription from Pm in the case of XylS and 70 in the case of XylS1. We tested whether transcription mediated by XylS mutants from a Pm:ЈlacZ fusion followed the pattern of induction of the wild-type regulator when induction was assayed in a 38 -deficient background. The results obtained confirmed that expression from Pm in the mid-log growth phase mediated by XylS1 is 38 -independent, whereas transcription stimulation from Pm with the XylS regulator is dependent on 38 (not shown). In contrast, in all three mutants we tested, the dependence on 38 was lost. To further confirm that the level of ␤-galactosidase activity reflected the transcriptional activity from Pm, we determined the level of expression from Pm in different backgrounds by using 70ts mutants proficient and deficient in the synthesis of 38 . The parental strain of these isogenic strains is E. coli P90A5c. This strain bearing the xylS or xylS1 alleles is able to transcribe the Pm promoter in response to the addition of 3MB regardless of the incubation temperature, as expected ( Fig. 2A). E. coli UQ285 has a 70ts that is thermosensitive at  H153N) (q) or pLAR32 (XylS1) (OE) were grown overnight on Luria-Bertani medium with kanamycin and tetracycline. Cultures were diluted 100fold in the same medium but with 1 mM indicated benzoate, and after 5 h, ␤-galactosidase activity was assayed in permeabilized cells. 2MB, 3MB, 4MB, 2-, 3-, and 4-methylbenzoate, respectively; 2,3DMB and 3,4DMB, 2,3-and 3,4-dimethylbenzoate, respectively; B, benzoate.
42°C. Because expression from Pm with XylS seems to be dependent on 38 whereas with XylS1 it seems to depend on 70 , one would expect transcription from Pm to occur at 30°C in UQ285 and at 42°C with XylS and at 30°but not at 42°C with XylS1. We tested transcription from Pm in XylS-and XylS1-proficient backgrounds in the absence and in the presence of 3MB. In these genetic backgrounds, no transcription from Pm took place in the absence of the effector (not shown).
In the presence of 3MB, transcription occurred at 30°C with either XylS or XylS1 (Fig. 2B), but at 42°C transcription from Pm was seen only in the XylS-proficient background (Fig. 2B) (as a control of the correct functioning of the UQ285 strain, note that transcription from the 70 tandem Pr1 and Pr2 promoters occurred at 30°C but not at 42°C (not shown)). We then tested transcription stimulation from Pm at 30°C and 42°C with XylSD137E and XylSH153N. The results obtained were similar with both mutants and are shown for XylSD137E in Fig. 2B. XylSD137E behaved like XylS1 rather than like XylS; we interpret this result to indicate that XylSD137E facilitated transcription from Pm with 70 rather than with 38 in the logarithmic growth phase.
To further confirm the switch in factors in RNA polymerase to transcribe Pm depending on the regulator used, we constructed E. coli EEZ286, which in addition to the 70ts mutation has an inactive 38 allele. In this mutant background, transcription from Pm did not occur in the absence of 3MB; however, in the presence of the effector and in cultures incubated at 30°C, transcription from Pm took place in the XylS1 background but not in the XylS background (Fig. 2C). Mutants XylSD137E and XylSH153N, in contrast with XylS, were able to stimulate transcription from Pm, as shown for XylSD137E in Fig. 2C. DISCUSSION XylS and XylS1 are 98.44% identical and differ in only 5 of 321 residues, all of which are located in the N-terminal end. The availability of these two different alleles of the regulator is of great interest, because these "minor" differences are clearly reflected in the regulator ability to induce transcription from Pm with different effectors and in the interactions with the transcriptional machinery. With regard to the kinetic properties, Gallegos et al. (7) first reported that the induction rate from Pm in the XylS-pWW0 regulator was about 6 -10-fold as high as that in XylS1 and that XylS-pWW0 regulator had a wider effector profile than XylS1-pWW53. The introduction of stepwise mutations in XylS-pWW0 revealed that the single mutations with the greatest effect on the degree of transcriptional activation and the effector profile were those that involved residues 137 or 153. This was further confirmed when the effector profile for effectors in the double mutant XylSD137E,H153N were found to be similar to those of the XylS1-pWW53 regulator. This indicated that these residues may be involved in effector interactions.
The importance of the involvement of residues 137 and 153 in interactions with effectors was confirmed when random mutations at these positions resulted in point mutants (i.e. XylSD137R, XylSD137P, XylSH153G, and XylSH153D) that had lost the ability to recognize effectors. Recently the MarR regulator, which recognizes 2-hydroxybenzoate (salicylate) as an effector, has been crystallized with the aromatic carboxylic acid (28). In this regulator salicylate is bound in two sites, SalA and SalB; in both sites the salicylate carboxylate hydrogen bound to the positively charged lateral chain of arginine residues. The salicylate hydroxyl group is hydrogen-bound to the hydroxyl chain of a threonine at the SalA site and to the backbone carboxyl of an alanine at the SalB site. This indicates that substituents on the aromatic ring can establish interactions with different residues. On the basis of these results it is tempting to speculate that in XylS, histidine 153 might be involved in interactions with substituents on the benzoate aromatic ring, because mutations in this residue to uncharged or negatively charged amino acids resulted in loss of the ability to recognize benzoate derivatives. This in turn may be due to loss of interactions with the carboxyl group. The change His-153 3 Asn resulted in a less efficient regulator, but the amide lateral chain of glutamine may allow certain interactions with benzoate derivatives.
Given the negative character of aspartic acid 137, this residue cannot be involved in interactions with the carboxyl chain of benzoate. Instead, residue aspartic acid 137 may be involved in other interactions with substituents or in the intramolecular signal transmission chain that leads to activation of the regulator. In fact, altering the charge of this residue influenced the ability of the regulator to recognize benzoates.
Our previous study revealed cysteine 41 as another amino  1-3), pLAR18 (XylS1) (lanes 7-9), or pLAR5 (XylSD137E) (lanes 4 -6) were grown until the late exponential phase in the presence of 3MB at 30°C. When the turbidity was ϳ1.2, a sample was withdrawn for RNA extraction (lanes 1, 4, and  7), then each culture was divided into two halves. One was kept as a control at 30°C (lanes 2, 5, and 8), and the other was incubated at 42°C (lanes 3, 6, and 9), and after 45 min RNA was extracted. Primer extension and separation of cDNA was done as described under "Experimental Procedures." acid that may be part of the XylS effector pocket. The phenotype of mutants XylSD137R and XylSH153G regarding the lost ability to recognize benzoates is similar to the one we found when cysteine 41 was replaced with leucine (12). However, at position 41, certain mutations resulted in other phenotypic changes. For example, replacement of cysteine 41 with arginine resulted in a mutant that exhibited a semiconstitutive phenotype (i.e. it mediated a high level of transcription from Pm in the absence of effectors). Mutant XylSC41G showed altered effector specificity; it did not recognize 4MB but retained the ability to recognize 3MB (12). That different mutations in positions 41, 137, and 153 give rise to different phenotypes is consistent with the different roles of residues 41, 137, and 153 in interactions with effectors or signal transmission to achieve the active form of this transcriptional regulator. However, whether any of these residues interacts with the effectors is still unknown. Details of these interactions await resolution of the three-dimensional structure of this regulator with and without the effector.
The XylS protein has so far been difficult to purify because of the intrinsic insolubility of the protein, a characteristic shared by many proteins of the XylS/AraC family of regulators (29 -31). However, we recently solubilized the N-terminal end of XylS upon fusion to MalE and are now trying to obtain further insights into the structural characteristics of the N-terminal end of XylS. 2 Two members of the XylS family, SoxS and MarA, which are about 110 amino acid long, are equivalent to the C terminus of the XylS family of proteins. This coincidence suggests that the C-terminal end in members of the XylS/AraC family is involved in DNA binding and probably in interactions with the transcriptional machinery. A study by Michá n et al. (31) shows that although the N and C termini of XylS have specific functions, they are not independent, and they cross-talk. Their relationship is further supported by the present finding that mutations at the N terminus influence which factor the RNA polymerase chooses to stimulate transcription from Pm. In fact, the Pm promoter is unique in that XylS-dependent transcription from Pm requires RNA polymerase with either 32 or 38 , whereas XylS1 uses 32 or 70 with RNA polymerase. 3-Methylbenzoate induces the heat shock response (16), and upon initial transcription with 32 , this factor is replaced by 38 for XylS and 70 for XylS1. It is also interesting to note that minor changes in XylS influence the preferred factor that RNA polymerase chooses for transcription from Pm. Single point mutations at positions 137 and 153 of XylS are sufficient to allow RNA polymerase to function with 70 instead of 38 . As described under "Results," the transcription initiation point from Pm is the same regardless of the factor RNA polymerase used to transcribe the Pm promoter, and although 70 and 38 belong to the same class of factor (32)(33)(34)(35), it is likely that RNA polymerase with different factors recognizes different nucleotides at the Ϫ10 and Ϫ35 boxes in Pm.
In summary, XylS1 and XylS are almost identical, differing in only five residues. Two of the five amino acids, at positions 137 and 153, had major effects on effector recognition and significantly influenced the effector pattern. These residues in XylS also influence whether XylS mediates transcription from Pm with RNA polymerase using either 70 or 38 . Further studies with this system are expected to reveal the mechanisms by which subtle changes in XylS can influence transcription from Pm to such a significant degree.