Activation of the Pseudomonas TOL Plasmid Upper Pathway Operon IDENTIFICATION OF BINDING SITES FOR THE POSITIVE REGULATOR XylR AND FOR INTEGRATION HOST FACTOR PROTEIN*

Expression of the Pseudomonaa putida TOL plasmid upper pathway operon requires a promoter that be-longs to the -121-24 class. Stimulation of transcrip- tion from this promoter is positively controlled by the effector-activated XylR protein and requires a form of RNA-polymerase holoenzyme containing the RpoN-en- coded u factor, uS4. XylR-dependent stimulation of transcription from the Pseudomonas TOL upper pathway promoter was examined using deletions, inser- tions, and in vivo dimethyl sulfate footprinting. Two upstream activator sequences were identified in the -160 (UAS1) and -130 (UASP) regions. Deletion of these two regions abolished transcription activation, although conservation of the UASP element alone allowed limited transcription stimulation. Separation of UASl from UASP by half a turn or a full turn significantly reduced XylR stimulation of transcription from the upper pathway operon promoter. An inverted repeated ATTTGN2CAAAT (where N is any nucleoside), which most likely represented the XylR recognition sequence, was identified. Binding of XylR was ob- served in vivo in the absence of effector, but changes in the binding pattern were induced in the presence of rn-methylbenzyl alcohol, a XylR effector. In vivo footprinting analysis revealed that changes in to cool on ice. For primer extension, 0.5 mM of each dNTP and 0.5 units of Klenow enzyme were added, and the reaction was run for 20 min at 42 “C. Samples were then run on sequencing gels. The intensity of the bands in the autoradiograph was recorded densitometrically and quantified as the band peak area. Enzyme Assays-8-Galactosidase was determined in permeabilized whole cells as described previously (2, 5) and expressed in units according to Miller (34).

Expression of the Pseudomonaa putida TOL plasmid upper pathway operon requires a promoter that belongs to the -121-24 class. Stimulation of transcription from this promoter is positively controlled by the effector-activated XylR protein and requires a form of RNA-polymerase holoenzyme containing the RpoN-encoded u factor, uS4. XylR-dependent stimulation of transcription from the Pseudomonas TOL upper pathway promoter was examined using deletions, insertions, and in vivo dimethyl sulfate footprinting. Two upstream activator sequences were identified in the -160 (UAS1) and -130 (UASP) regions. Deletion of these two regions abolished transcription activation, although conservation of the UASP element alone allowed limited transcription stimulation. Separation of UASl from UASP by half a turn or a full turn significantly reduced XylR stimulation of transcription from the upper pathway operon promoter. An inverted repeated ATTTGN2CAAAT (where N is any nucleoside), which most likely represented the XylR recognition sequence, was identified. Binding of XylR was observed in vivo in the absence of effector, but changes in the binding pattern were induced in the presence of rn-methylbenzyl alcohol, a XylR effector. In vivo footprinting analysis revealed that changes in the methylation pattern of G and T also occurred in the -50 to -90 region, which is probably occupied by integration host factor (IHF) protein. IHF was required for maximal expression from the TOL upper pathway operon promoter in Escherichia coli. Separation of the IHF site from UAS2 by a full helix turn did not significantly affect stimulation of transcription, which is consistent with this region playing a conformational role, rather than a regulatory one, in promoter function.
The TOL plasmid pWW0 of Pseudomonas putida encodes the genetic information for the mineralization of toluene and xylenes (1). The genes of the catabolic pathway are organized in four transcriptional units. Two of these, the "upper" and the meta pathway operons, contain the xyl structural genes encoding the corresponding catabolic enzymes for the oxida-* The work in Spain was supported by Grant BT 87/023 from the Comisibn Interministerial de Ciencia y Tecnologia. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement'' in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Efficient transcription initiation from the -12/-24 promoters requires a specific additional activator protein, e.g. NtrC for glnAp2 in Escherichia coli, NifA for nif genes in Klebsiella, and DctD for dicarboxylic acid transport in Rhizobium (12). In a number of RpoN-dependent promoters, the binding sites for the specific regulators are located more than 100 bp from the transcription initiation point. These upstream activator sequences (UASs) are often inverted repeats that represent the target site for the activator protein and can be moved more than one kilobase away without losing their ability to mediate transcriptional activation (13)(14)(15). Recently, IHF protein was shown to stimulate transcriptional activation of nif promoters (16)(17)(18), and sequence alignments have suggested that a putative IHF-binding site may be located in several - 12/-24 promoters (18, 19). For the best studied RpoN-dependent promoters, namely the nif promoters in Klebsiella pneumoniue and ntr promoters in Enterobacteriaceae, models have been proposed in which RpoN, as well as the activator protein, behave as DNA-binding proteins. Studies with the glnAp2 promoter have demonstrated that RNA-polymerase/RpoN binds to the promoter in the absence of its corresponding activator protein NtrC to form a closed complex (20-22), but interaction with the regulatory protein is then required for transcription initiation. The molecular mechanism of this protein interaction is as yet unknown (23,24).
NifA, DctD, NtrC, and XylR activators for different -121 -24 promoters all exhibit a conserved central domain, but differ in their N-terminal region, which is involved in specific interactions with sensor proteins or effectors. domain has been proposed as the region that interacts with RpoN/RNA-polymerase. This region harbors a putative ATPbinding site (25); ATP is required for open complex formation by NtrC (22). The C-terminal region contains a putative helixturn-helix motif characteristic of DNA-binding proteins and is the DNA-binding domain.
The model for remote stimulation of transcription in the -12/-24 promoters proposes DNA loop formation to bring UAS-bound regulators close to the RpoN/RNA-polymerase promoter complex. The role of IHF-mediated DNA bending may be to facilitate DNA loop formation and, therefore, favor specific contacts between activator and polymerase (16)(17)(18)23). However, since the UAS in some -12/-24 promoters is not absolutely required for activation, the activator proteins may interact directly with the RpoN/RNA-polymerase-promoter complex (26)(27)(28) in the absence of loop formation and IHF. In this study, we report the activation of the Pu promoter by the XylR protein. Deletions, insertions, and in vivo footprinting studies revealed three upstream regions, located between -106 and -144, between -144 and -200, and between -50 and -90, to be required for transcription activation. The two former regions seem to represent the Pu-XylR recognition sequences, whereas the latter one probably represents a target for IHF protein.
Plasmids Used and Constructed in This Stwly-Plasmid isolation, transformation, and cleavage by restriction enzymes, agarose gel electrophoresis, and gene cloning were performed according to the methods described by Maniatis et al. (31) with minor modifications.
The following plasmids were constructed during the course of this study. pERD400 was constructed by cloning a 333-bp EcoRI-EamHI fragment carrying the Pu promoter from pRD579 between the EcoRI and BamHI sites of pTZ19. pERD401 contains the wild type Pu promoter fused to the lacZ gene in pMC1403. pERD402 contains a deleted Pu promoter fused to the lac2 gene in pMC1403. The end point of the deletion is located at -200 bp with respect to the main transcription initiation point in the Pu promoter. pERD403, pERD404, and pERD405 were similarly constructed, except that the end points of the deletions were located at -88, -40, and -25, respectively. The in-frame fusions were constructed by cloning the deletedpromoters as blunt end-BamHI fragments between the unique SmaI-BamHI sites of pMC1403 (see the Ea131 deletions below). Plasmids pERD410, pERD411, and pERD412 carry a deleted Pu promoter extending from -64, -106, and -144, respectively, fused to the promoterless lacZ gene in pMC1403 (see below for the detailed construction of these plasmids). Plasmid pERD414 carries a single point mutation (A + G at -63) in the Pu promoter.
Bal.31 Deletions-Plasmid pERD400 linearized with EcoRI was digested with exonuclease Ba131 as described by Maniatis et al. (31). Aliquots were withdrawn at intervals of 5 min during 30 min, and digestion was stopped by adding EGTA to a final concentration of 0.2 M. DNA was then phenol-extracted, and the recessive ends were filled in with dNTPs using Klenow enzyme. Deleted promoters were obtained as Bal31-blunt end-BamHI fragments and subcloned between the SmaI-EamHI sites of pMC1403. Transformants of E. coli 5K were isolated, and plasmids bearing deleted Pu promoters were selected as having an extra BstEII restriction site. The precise end points of the deletions were located by DNA sequencing.
Site-directed Mutagenesis-Site-directed mutagenesis of the Pu promoter was carried out on pERD406. This plasmid is a derivative of pGC2 that carries the Pu promoter. E. coli CJ236 (pERD406) cells were grown in 2 X TY. Uracyl-enriched single-stranded DNA was obtained in the presence of the helper phage, M13K07 (29). Phosphorylated oligonucleotides exhibiting a single mismatch with respect to the wild type single-stranded DNA were used. 1 pg of singlestranded DNA in sodium chloride, sodium citrate buffer was hybridized to 20 ng of oligonucleotide at 70 "C for 5 min, and the mixture was allowed to cool slowly to room temperature. Primer extension was done in buffer (20 mM HEPES, 2 mM dithiothreitol, 10 mM MgC12) containing 0.5 mM of each dNTP, 1 mM ATP, pH 7.0, 2.5 units of T4 polymerase, and 1 unit of T4 ligase. The reaction was allowed to proceed for 2 h at 42 "C. Half of the reaction was transformed in the ung+ E. coli 5K.
In Vivo Footprinting Analysis--100 ml of E. coli harboring pRD579 or pRD579 and pTS174 were grown at 30 "C for 3 h (A600 = 0.6-0.7) in the presence and in the absence of 1 mM n-methylbenzyl alcohol. To these cultures we added 100 pl of fresh dimethyl sulfate in 1% saline phosphate (w/v) (150 mM NaCl, 40 mM K2HP04, 33 mM KH2P04, pH 7.2). Two min later, cells were rapidly collected by filtration through 0.45-pm pore filters, and washed twice with 150 ml of saline phosphate. DNA was extracted by the alkaline lysis method previously described (31).
DNA was methylated in vitro by incubating 25 pg of plasmid DNA in 50 mM sodium cacodylate buffer with 0.5% dimethyl sulfate for 2 min at 20 "C. The reaction was stopped by adding 50 pl of 1.5 M sodium acetate containing 1 M P-mercaptoethanol, followed by ethanol precipitation.
DNA was cleaved at methylated guanines in a final volume of 10 p1 of 1 M piperidine by heating at 90 "C for 30 min, after which DNA was recovered by ethanol precipitation.
One pg of cleaved DNA was hybridized at 95 "C for 2 min with 0.1 pmol of 5'-32P-labeled primer (5"CAGAGTTGAGAAAATACAAC-3' for the top strand and 5'-CCAGCGTCACAGACTCCAG-3' for the bottom strand) in a final volume of 10 p1 of 10 mM Tris-HC1, pH 8.0, 10 mM MgCI2. The mixture was allowed to cool on ice. For primer extension, 0.5 mM of each dNTP and 0.5 units of Klenow enzyme were added, and the reaction was run for 20 min at 42 "C. Samples were then run on sequencing gels. The intensity of the bands in the autoradiograph was recorded densitometrically and quantified as the band peak area.
Enzyme Assays-8-Galactosidase was determined in permeabilized whole cells as described previously (2,5) and expressed in units according to Miller (34).

5"Deletions in the TOL Plasmid Upper Pathway Operon
Promoter"Bal31 deletions in the Pu promoter in plasmid pERD400 were generated as described under "Materials and Methods." The deleted promoters were obtained as Bal31blunt end-BamHI fragments that were cloned between the SmaI-BamHI sites of pMC1403, so that in-frame fusions to a promoterless lac2 gene were generated. Four groups of deletions that had lost between 50 and 200 bp were chosen, and one plasmid of each group from pERD402 through pERD405 was further characterized. To establish the exact end of the deletions in pERD402 through pERD405, the 100-300 bp EcoRI-BamHI fragments of pMC1403::APu were subcloned in pTZ19 and sequenced. The accurate determination of the deletion established the ends for the four deletions at -200, -88, -40, and -25 bp with respect to the main transcription initiation point.
To generate three additional deletions in the Pu promoter, an oligonucleotide site-directed mutagenesis approach was taken. Mutagenesis was carried out by using uracyl-enriched single-stranded DNA prepared from pERD406. This DNA was hybridized independently to three oligonucleotides that exhibited a single mismatch with respect to the wild type Pu sequence (5'-ACAAAGAAAATCGATAATTTAGATG-3', 5'-GCCAGCGTGCTGCAGATTTTCTCTC-3', and 5'-CAGGTGGTTATTCGCEATTGATG-3') to generate A + G, T -+ C, and G -+ T changes at -63, -108, and -147, respectively. These substitutions created single restriction sites for CluI, PstI, and NruI, respectively. After primer extension and transformation in E. coli 5K, mutants with a single nucleotide change at -63, -108, and -147 were identified by their acquisition of the new corresponding site. In order to generate deletions using the new sites, the pERD406 derivatives were cleaved with the restriction enzyme for the new site, and then the sticky ends of ChI and PstI were made blunt with Klenow enzyme and the four dNTPs and nuclease S1, respectively. After phenol extraction, DNA was cleaved with BarnHI, and the deleted Pu promoters (blunt end-BamHI) were cloned between the SrnaI-BarnHI sites of pMC1403. These three APu in pMC1403 were called pERD410 (deletion from -64), pERD411 (deletion from -106), and pERD412 (deletion from -144) (see Fig. 1).
,&Galactosidase levels arising from wild type Pu promoter and APu promoters fused to lac2 in pMC1403 were estimated in E. coli 5K with or without the xylR gene in trans on plasmid pTS174 and also in the presence and in the absence of an effector for XylR protein, i.e. m-methylbenzyl alcohol. As expected, in the absence of XylR protein, no induction from Pu or APu was observed, regardless of the presence of a XylR effector (Fig. 1). In the presence of XylR, rn-methylbenzyl alcohol-dependent induction was only observed with the wild type promoter and with the deletion extending up to -200 bp from the main transcription initiation point (Fig. 1). The induced levels were about 10-fold higher than the basal level. The deletion extending up to -144 exhibited a low but repetitive increase in /3-galactosidase with respect to its corresponding basal level, whereas no induction at all was observed with the other deletions (Fig. 1). These results suggest that two DNA stretches in Pu, located between -106 and -144 and between -144 and -200 contain important elements for the full stimulation of transcription from Pu.
In Vivo Footprinting Analysis of Pu-The reactivity of guanine residues in the Pu promoters toward dimethyl sulfate was assayed in plasmid pRD579, a low copy number vector carrying a transcriptional fusion of Pu to a promoterless lac2 gene. DNA protection analysis was performed in E. coli bearing only pRD579 or bearing pRD579 and pTS174, a plasmid carrying the xylR gene, and in the presence and in the absence of the XylR protein effector rn-methylbenzyl alcohol. Representative autoradiograms from the primer extension analysis are shown in Fig. 2. Fig. 3 shows the results of the densitometric analysis. The log ratio between the intensity of the band in the absence and in the presence of XylR are given, with negative values indicating hypermethylation and positive values indicating protection. There were three regions in which protection or hypermethylation of G was observed. In the top strand, the region furthest from the proposed RNApolymerase binding site shows that G at -160 was hypermethylated in the absence of effector, but protected in the presence of rn-methylbenzyl alcohol. In this region, the G located at -169 was hypermethylated both in the absence and in the presence of m-methylbenzyl alcohol (Figs. 2 and 3). In the -130 to -140 region, G located at -131 and -139 were protected and hypermethylated, regardless of the presence of effector. In conclusion, two regions, one around -160 and one around -130, clearly showed XylR-mediated protection. In the text below, we will refer to these regions as UASl and UAS2, respectively. The primer extension strategy permitted detection of alterations in the in vivo methylation of the guanines in the region from -50 to -90. It was observed that G located at -54 and -70 were protected both in the presence and in the absence of rn-methylbenzyl alcohol. An unexpected finding was that a T located at -84 apparently exhibited reactivity toward dimethyl sulfate attack and piperidine cleavage. This residue was protected in the presence of effector but hypermethylated in its absence (see Fig. 2). Other workers have also observed unexpected reactivity of AT base pairs toward dimethyl sulfate (21, 35, 36). This effect could be due to methylation of the N-3 position of thymine, which is usually hidden in the DNA double helix (37); this may reflect a conformational change in DNA.
In the bottom strand, alterations in the -50 to -90, -130, and -160 regions were also observed. G located at -122, -128, and -130 were hypermethylated, regardless of the presence of effector. It is also worth noting that G at -144 and -146 were protected in the presence of effector. G at -161 and -166 were protected both in the absence and in the presence of effector, but this was very marked in the presence of mmethylbenzyl alcohol (see Fig. 3). The G located at -176 and  pTS174 (xylR) (lanes b and d ) in the presence (lanes d and e ) and in the absence (lanes b and c ) of rn-methylbenzyl alcohol were exposed to dimethyl sulfate. -178 were protected in the presence and hypermethylated in the absence of effector. Primer extension analysis also revealed that G at -92 was protected in the absence of mmethylbenzyl alcohol but hypermethylated in its presence.
We also found that a T residue at -45 was hyperreactive in the presence of the effector. The T at -45 did not generate an extension termination product with DNA preparations from cells treated with dimethyl sulfate in the absence of the effector.
Expression of P u in E. coli IHF Mutants-Sequence com-parison analysis (19) showed the presence of a putative binding site for IHF in the Pu region lying between -54 and -89 bp, where we observed alterations in the pattern of dimethyl sulfate methylation of G and T. A 100% match with a recently derived IHF consensus sequence (18,37) was found by aligning the -55 to -67 region of Pu with the consensus one (Fig.  4). The IHF requirement for the expression of Pu was examined in E. coli strains proficient or deficient in IHF synthesis and carrying pERD401 and pTS174, both in the presence and in the absence of m-methylbenzyl alcohol, a XylR effector. In the absence of xylR, expression from Pu was low regardless of the presence or absence of the aromatic alcohol. In the presence of xylR and rn-methylbenzyl alcohol, the expression level from Pu in the IHF mutant was about 25% of the level measured in the IHF-proficient strain.
To further assess the possible requirement of the -50 to -90 region in the activation of Pu, a single point mutation was generated within the IHF consensus sequence by replacing A at -63 with G, which created a single ClaI site. In addition, two other mutants were generated by inserting 2 extra bp (GC) in the open ClaI site with Klenow enzyme and the four dNTPs and by inserting 8 extra bp in an HpaI linker cloned in the filled ClaI site (CGGTTAAC). Moreover, a deletion extending from -64 to -111 was generated by using the PstI site at -106 and the ClaI site at -64 in mutant promoters. The promoters exhibiting a single point mutation, the 2 extra bp, the 8 extra bp, and a deletion of 47 bp were fused to lac2 in pMC1403, and the resulting plasmids were called pERD414, pERD415, pERD416, and pERD417, respectively.
The single point mutation introduced at -63 in the proposed IHF-binding site did not significantly alter the pattern of stimulation of transcription in IHF-deficient and IHFproficient E. coli strains when compared with the wild type promoter in the same E. coli backgrounds (Table I), which is consistent with IHF still recognizing the mutated site. In contrast with the wild type promoter, no increase in @-galactosidase expression from the deleted promoter in pERD417 was observed regardless of the strain used, in the presence or absence of XylR and in the presence or absence of m-methylbenzyl alcohol (see Table I). With the mutant promoters bearing 2 and 8 extra bp inserted into the IHF site (pERD415 and pERD416), the levels of expression in the IHF-background were similar to those determined for the wild type promoter in pERD401. However, in the IHF' background, maximal level of expression was about 20% of that determined for the wild type promoter in the same background, indicating the inserts disrupt IHF binding. Separation of the Putative XylR Binding Sites-In order to separate the -130 UAS2 region from the -160 UASl region, a single NruI restriction site created at -144 was used to add 6 and 10 extra bps by introducing HpaI and NcoI linkers of 6 and 10 mers, respectively. The mutant promoters were cloned in front of lac2 in pMC1403 to yield plasmids pERD419 and pERD420, respectively. @-Galactosidase levels in the presence  promoter with a single nucleotide change at -63 in pERD414, a mutant promoter with a 2-bp insertion at -64 in pERD415, or a mutant promoter with a deletion from -111 to -64 in pERD417 with or without pTSl74, were grown on LB supplemented with ampicillin (pERD only) or ampicillin and chloramphenicol (pERD and pTS174). Overnight cultures were diluted in the same medium with (+) or without (-) 1 mM rn-methylbenzyl alcohol (mMBA), and P-galactosidase was determined as described previously in permeabilized whole cells. Results are the average of three independent determinations, with standard deviations in the range of 20%.  Influence of deletion intervals and insertions between the -130 and -160 regions and between the -90 and -130 regions Plasmids pERD401 or pERD419 through pERD423 were transformed in E. coli 5K (without XylR) or E. coli 5K (pTS174) (with XylR). Transformants were grown overnight on LB supplemented with ampicillin (pERD plasmids) or ampicillin and chloramphenicol (pERD + pTS174). These cultures were diluted 100-fold on the same medium with (+) or without (-) m-methylbenzyl alcohol (MBA) as indicated above. &Galactosidase was determined 5 h later. and in the absence of XylR and with and without m-methylbenzyl alcohol were determined. Like the wild type promoter, the two mutant promoters exhibited low basal levels of expression, but in contrast with the wild type, reduced increases in &galactosidase with xylR and m-methylbenzyl alcohol were found. This was more evident when these regions were separated by a full helix turn, as in pERD420, suggesting that the positioning of the -130 and -160 regions is important for transcriptional activation from Pu (Table 11).
To alter the distance between the UAS regions and the putative IHF site, we used the PstI site created at -106. The sticky ends of the PstI site were removed by T4 polymerase, and the resulting blunt ends were ligated in the presence and in the absence of a 10-mer NcoI linker. This generated a promoter with a 4-bp deletion (without linker) or with 6 extra bp (with the linker). The promoter exhibiting the NcoI site was digested with this enzyme, and the sticky ends were filled in with the Klenow enzyme and the four dNTPs so that the mutant promoter exhibited 10 bp between the -90 region and UAS1. The three mutant promoters exhibiting a 4-bp deletion or having 6 and 10 extra bp were fused to lac2 in pMC1403 to generate plasmids pERD421, pERD422, and pERD423.
(3-Galactosidase activity was determined with and without XylR in trans and in the presence and the absence of mmethylbenzyl alcohol (Table 11). Introduction of a full helix turn did not affect the level of &galactosidase expression, but in contrast, shortening the distance by 4 bp or introducing half a turn (6 extra bp) clearly diminished expression of 8galactosidase (Table 11), indicating that the UAS region must be correctly phased with downstream sequences to function efficiently.

DISCUSSION
In this study we investigated transcriptional activation from the TOL plasmid Pu promoter by generating deletions, point mutations, and small insertions, and by using in vivo dimethyl sulfate methylation.
The results of our Bal31 deletions suggest that full transcription activation of Pu is dependent on two DNA sequences located between -200 and -144 (UAS1) and -144 and -106 (UAS2) (Fig. 5). These regions have been previously suggested to be putatively involved in transcription activation, based on sequence homology between the Ps and Pu promoters (19). Comparison of these regions shows a highly conserved domain with a 5-bp inverted repeat in the -130 region 5"ATTTGCT-CAAAT-3' (top strand) and a less well conserved inverted repeat in the -160 region 5'-ATTTGATCAAGG-3' (bottom strand). The deletion of both binding motifs resulted in the complete loss of activation of transcription from Pu. Retention of UAS2 after deletion of UASl still allowed limited induction. Thus, as is the case for other -12/-24 promoters (e.g. glnAp2, nifH, andfdh), upstream activator sequences are required for the full and efficient activation of Pu.
Our in vivo footprinting experiments strongly support a

T A T G G G G a G G G A A A G A A A A A A T G T T T C T T
""""""~ -" " " " " " " _ " " " " " " " " " " " " "

A A T C A A T A A T T T A G A ' T G A A A T A A G G G G A
b " " " " " " " " " " " " " " "

T T A G T T A T T A A A T C T ! A C T T T A T T C C C C T
" " " " " " " " " " " " ".""""..."" ~

C A A C T C T G T T C A G A T T G G T T G C T T T C G C G T T G A G A C A A G T C T A A C C A A C G A A A G C G C A T G T A T A T C C T C A A A G C G G G G T A C A T A T A G G A G T T T C G C C C
FIG. 5 mechanism of remote activation of Pu promoter by XylR in which XylR binds the UASl and UAS2 regions but shows, in addition to the two above motifs, a third region of altered reactivity to dimethyl sulfate. This latter region, located from -50 to -90, showed no sequence homology with the other two (see Fig. 5). This could reflect changes in DNA conformation/ protein binding related to the productive interaction of XylR with RNA-polymerase/RpoN complex during activation. I n vitro footprinting with partially purified XylR protein has shown that the XylR protein binds only to the two motifs located beyond the -106 point of the Pu promoter (37b). The i n vitro studies show a complex pattern of interaction of XylR with the UASl and UASB sequences. Thus, based on our in vivo data and on in vitro findings, we propose that the XylR protein recognizes and binds to these two UAS sequences (Fig. 5). This binding is independent of the availability of XylR effectors, although in the presence of effectors, changes in the binding pattern in the UASl motif were observed at positions -160 and -176 in the top and lower strands, respectively. Furthermore, changes in the intensity of protection or hypermethylation were also observed in the UASl motif ( Figs. 2 and 3). In contrast with our observations in the UASl region, no obvious changes in the binding pattern or the intensity of protection or hypermethylation were observed in the UASB motif. It is, nonetheless, worth noting that G at -144 and -146 in the lower strand became protected in the presence of a XylR effector (Fig. 3). It is quite possible that the contacts in UASB were also altered; however, this region has few guanines, limiting our ability to detect changes. The significance of these changes needs to be further examined by in vitro footprinting experiments and in vivo footprinting with mutant promoters exhibiting point mutations in the bases putatively contacted by XylR in the presence and in the absence of effectors. The binding of effectors to the XylR protein may simply represent a small structural modification that slightly alters the interaction between XylR protein and DNA. The interactions of XylR at UASl and UASB may not be symmetrical, as deduced from the fact that the inverted repeated sequences are positioned on opposite DNA strands. This pattern of nonsymmetrical interactions has also been observed in vitro (41).
The introduction of 6 bp between UASl and UASB or the separation of the activation sequences by a helix turn diminished stimulation of transcription from Pu. This suggests that either the XylR proteins do not bind to activator sequences or the XylR proteins bound to activator sequences are probably positioned at a definite distance; hence, cooperative interactions may be involved in stimulation of transcription.
Holtel et al. (19) identified the -50 to -90 region of Pu as a putative binding site for IHF in the Pu promoter. In fact, the stretch between -60 and -70 exhibited a 100% match with a derived IHF consensus sequence (18, 38). I n vitro footprinting experiments with purified IHF and Pu promoter have confirmed that the -54 to -90 region is contacted by IHF (41). Our in vivo studies of the top strand demonstrated G protection for G at -54 and -70. Protection occurred if XylR was present, but was independent of the presence of a XylR effector. In the bottom strand, it was observed that G at -92 was protected in the absence of m-methylbenzyl alcohol, but hypermethylated in the presence of effector. An unexpected finding i n vivo was the reactivity of T located at -84 in the top strand and -45 in the bottom strand toward dimethyl sulfate and piperidine cleavage. The T at -84 was protected in the presence of effector, but was hypermethylated in its absence. In contrast, the T at -45 was hypermethylated only in the presence of effector. Since the DNA sequence in this stretch is indeed very different from that of UASl and UAS2, we suggest that this region is recognized by IHF. The different reactivities of T in Pu in the presence and the absence of effector suggest the existence of DNA conformational changes, which are probably induced by the transcriptional complex (XylR + IHF + RNA-polymerase/RpoN).
In studies with E. coli deficient in IHF synthesis, we observed decreased stimulation of transcription from Pu. Furthermore, a deletion of the putative IHF-contacted sequences in plasmid pERD417 abolished activation from Pu (see Table  I), thus confirming a stimulatory role for this factor in transcription from Pu. This has also been observed for other promoters in which IHF seems to be involved (17,39). Since separation of the IHF-binding site by a helix turn from UASl and UAS2 had no effect on transcription activation, we suggest that this piece of DNA plays a physical role rather than a regulatory one, which would require direct contact between IHF and XylR. The putative role of IHF may be to bend DNA and thus facilitate direct interaction between XylR and the RNA-polymerase/RpoN complex bound in the downstream promoter element. This type of interaction is in accordance with a model involving loop formation, as was recently found to be the case for the NtrC-activated RpoN-dependent glnAp2 promoter (23).
Acknowledgments-We gratefully acknowledge the communication of unpublished results and discussions with V. de Lorenzo and K. N.
Timmis. We thank W. Cannon and E. Morett for assistance with the i n uiuo footprinting experiments, Fernando Rojo for critical reading of the manuscript, and Nacho Olivares for help in designing the figures.