Protein phosphorylation affects binding of the Escherichia coli transcription activator UhpA to the uhpT promoter.

Expression of the Escherichia coli sugar phosphate transporter UhpT is induced by extracellular glucose 6-phosphate through a transmembrane signaling process dependent on the sensor kinase UhpB and the UhpT homolog, UhpC. These proteins are thought to regulate the phosphorylation of the transcription activator, UhpA. To examine the effect of protein phosphorylation on the binding of UhpA to target sequences in the uhpT promoter region, the UhpA protein was overexpressed and purified. Purified UhpA was phosphorylated by acetyl phosphate in a reaction that was dependent on Mg2+ and on the presence of aspartate 54, the site of phosphorylation in homologous response regulators. Gel electrophoretic mobility shift and DNase I and hydroxyl radical protection assays showed that UhpA bound specifically to the region of the uhpT promoter extending from −80 to −50 bp, relative to the transcription start site. At higher concentrations of UhpA, binding was extended to the −32 region. Binding to the −64 element exhibited positive cooperativity and was stimulated severalfold by phosphorylation of UhpA, whereas extension to the downstream region was more strongly affected by phosphorylation. The consensus sequences for the high affinity UhpA-binding sites in the −64 element and for the downstream, low affinity sites are proposed. The pattern of in vitro binding by UhpA agreed with the in vivo observations that phosphorylation-independent assembly of the transcription initiation complex can occur at elevated concentrations of UhpA.

Expression of the uhpT gene allows growth of Escherichia coli on various phosphorylated sugars. The UhpT anion-exchanging transporter mediates the uptake of many organophosphate compounds, but its synthesis is induced only by extracellular glucose 6-phosphate (1)(2)(3). Regulatory proteins encoded by the uhpABC genes couple the expression of UhpT with detection of exogenous Glu-6-P (reviewed in Refs. 4 and 5). Because UhpC is related to the transporter UhpT in both amino acid sequence and transmembrane structure, it is likely to serve as the receptor for external Glu-6-P. UhpA and UhpB contain sequence motifs characteristic of the response regulators and sensor kinases, respectively, of two-component bacterial signaling systems. Transcription at the uhpT promoter is absolutely dependent on UhpA and is also subject to catabolite repression through the action of the catabolite gene activator protein, CAP 1 (6).
Sequence comparisons group UhpA into a family of responseregulator proteins that includes NarL from E. coli, DegU from Bacillus subtilis, and FixJ from Rhizobium species (7). Members of this family share a common modular organization with an N-terminal phosphorylation module linked through a highly variable, flexible linker to a C-terminal output module (8). The N-terminal domain contains a highly conserved aspartyl residue (Asp-54 in UhpA), which is the site of phosphorylation in homologous proteins, CheY and NtrC (9,10). This aspartyl residue is part of an acidic pocket with bound Mg 2ϩ necessary for structure and catalytic activity (11,12). Response regulators are normally phosphorylated by their cognate sensor kinases, but many can be phosphorylated by acetyl phosphate or other low molecular weight phospho-donors (13). The C-terminal domains of the UhpA family of response regulators contain a highly conserved segment predicted to form a helix-turn-helix motif that is similar to the DNA-binding regions of some transcription factors that are not regulated by protein phosphorylation, such as MalT from E. coli and LuxR from Vibrio fischeri (14). The existence of these predicted structural motifs has been verified by the recent description of the structure of NarL at 2.4-Å resolution (15).
When the uhpA gene is overexpressed from multicopy plasmids, high level constitutive expression from the uhpT promoter occurs even in the absence of UhpB and UhpC function (16). The D54N variant of UhpA, in which aspartate 54 at the putative site of phosphorylation is replaced with asparagine, is totally inactive for Uhp expression when in single gene copy, but is as active as the wild-type protein when overexpressed (17). It appears, therefore, that both protein concentration or relative stoichiometry and phosphorylation are important parameters for UhpA activity, and that the requirement for phosphorylation of UhpA is lost when it is overexpressed.
Previous genetic analyses identified four regulatory elements in the uhpT promoter: a Ϫ10 region typical of 70 -dependent promoters, a 10-bp inverted repeat centered at Ϫ32, a 31-bp hyphenated inverted repeat centered at Ϫ64, and a CAPbinding sequence centered at Ϫ103.5 (all nucleotide coordinates are relative to the transcription start site) (18). Multicopy plasmids carrying portions of the uhpT promoter were used in in vivo titration experiments to identify a binding site for UhpA by their ability to compete with the chromosomal uhpT promoter for limiting amounts of UhpA. Titration of UhpA was seen only if the multicopy plasmid carried the Ϫ64 element. Either half of the Ϫ64 element conferred reduced in vivo titration activity (18), suggesting that each half of the Ϫ64 element can bind UhpA, but not as well as the intact element. Disruption of the Ϫ32 element with a 6-bp linker substitution eliminated promoter activity, but the Ϫ32 element did not compete for UhpA binding in vivo and its function is yet unknown. Although UhpA protein has been implicated genetically as an essential activator of uhpT expression, the biochemical demonstration of its role in transcription activation has been lacking. Here we report that phosphorylation of UhpA by acetyl phosphate occurs on Asp-54, that UhpA binds specifically to target sequences in the uhpT promoter, and that phosphorylation strongly affects the ability of UhpA to bind to the Ϫ64 element and a downstream region.

EXPERIMENTAL PROCEDURES
Strains and Plasmids-Strain BL21(DE3) was obtained from Novagen, Inc. and bears a lysogen with the phage T7 gene 1 (19). Two oligonucleotides (i, 5Ј-GGGTCTAGAGGAGGAGACTCATGATCACCG-TTGCCCTTATA-3Ј; and ii, 5Ј-GGGTAATTAAGCTCGAGAACAACGT-C-3Ј) were used as primers in a polymerase chain reaction (PCR) (20) with plasmids pRJK10 (21) or pAlter uhpA[D54N] (17) as DNA template. Primer i introduces a BspHI site at the initiation codon of uhpA, and primer ii generates an XhoI site downstream of the uhpA coding sequence. The 610-bp PCR products containing uhpA ϩ or uhpA [D54N] were ligated into NcoI ϩ XhoI-digested pET-15b (Novagen, Inc.) to form plasmids pBW6 and pJLD1, respectively, in which the uhpA genes are expressed under the control of the IPTG-inducible T7lac promoter. The inserts cloned in pBW6 and pJLD1 were sequenced using the Sequenase 2.0 reaction system (United States Biochemical Corp.) to confirm that no sequence changes were introduced during amplification and cloning.
Isolation and manipulation of recombinant DNA molecules employed standard techniques (20) or the product manufacturers' recommendations.
Purification of UhpA-Plasmids pBW6 and pJLD1 were introduced into strain BL21(DE3) by transformation. Isolates were grown in Luria broth at 37°C to a culture A 600 of 0.7-0.8. IPTG was added to a final concentration of 1 mM and incubation was continued for 3 h until A 600 reached 1.15. Cells were harvested and suspended in buffer A (50 mM Tris acetate, pH 6.9, 0.1 mM EDTA, 1 mM MgCl 2 , 1 mM dithiothreitol (DTT)) containing 0.15 mM phenylmethylsulfonyl fluoride to minimize proteolysis. The cell suspension was passed three times through a French pressure cell at 15,000 pounds/square inch, and the lysate was centrifuged at 14,000 ϫ g for 10 min and then at 100,000 ϫ g for 1 h. Nucleic acids and bound proteins were precipitated by addition of polyethyleneimine (final concentration, 0.2%) for 10 min on ice, followed by centrifugation at 5,000 ϫ g for 10 min. The pellet was washed with TEMB buffer (25 mM Tris-HCl, pH 8.0, 0.5 mM EDTA, 5 mM MgCl 2 , 1 mM DTT, 430 mM betaine). UhpA was released from the particulate material by extraction with TEMB buffer containing 1 M NaCl. After 15 min on ice, the mixture was centrifuged at 5,000 ϫ g for 30 min, and the supernatant solution was retained. Following addition of saturated ammonium sulfate, pH 7.5, for 60 min, the fraction that precipitated between 10% and 45% saturation with ammonium sulfate was collected by centrifugation at 7,000 ϫ g for 80 min, suspended in buffer A, and dialyzed overnight against 2,000 volumes of buffer A. The supernatant fraction after centrifugation at 10,000 ϫ g for 5 min was applied to a MacroPrep DEAE support column (Bio-Rad), which was equilibrated and washed with buffer A, and developed with a linear gradient from 0 to 250 mM NaCl in buffer A. Fractions containing active UhpA (eluting at 50 mM NaCl) were pooled, dialyzed against buffer A, concentrated using a Centricon-10 membrane ultrafiltration device (Amicon Corp., Beverly, MA), and stored at Ϫ70°C. Protein concentrations were determined by the dye-binding method of Bradford (22). Purification was estimated by Coomassie Brilliant Blue staining (23) and scanning with a Personal Densitometer (Molecular Dynamics, Sunnyvale, CA).
N-terminal Amino Acid Sequencing-UhpA protein was resolved by SDS-PAGE and transferred by electroblotting to a polyvinylidene difluoride membrane. This membrane was subjected to direct automated amino acid sequence analysis using a vertical cross-flow reaction vessel (24), connected to an Applied Biosystems model 470A Sequenator with on-line model 120A phenylthiohydantoin analyzer, using the manufacturer's suggested reaction cycles.
Phosphorylation of UhpA-UhpA was phosphorylated by incubation at 37°C in a mixture containing 13 M UhpA and 10 mM acetyl phosphate in buffer D (50 mM Tris-HCl, pH 7.5, 6 mM MgCl 2 , 1 mM DTT). Acetyl [ 32 P]phosphate was prepared as described (13). For the experi-ments shown in Fig. 3, proteins (2.5-10 M) were incubated at 37°C in a 27-l reaction volume containing buffer D and 20 mM acetyl [ 32 P]phosphate. The half-life of phospho-UhpA (P-UhpA) was determined by labeling UhpA with acetyl [ 32 P]phosphate for 60 min and then removing excess acetyl phosphate by passing the reaction mixture through a Sephadex G-50 QuickSpin column pre-equilibrated with buffer D. The filtrate was then incubated at 37°C, and portions were pipetted at various times onto nitrocellulose filters (0.05 m pore size, 25 mm diameter, presoaked in buffer D) on a porous plastic filter support. Each 24-l sample was filtered under gentle vacuum and washed immediately with 0.5 ml of buffer D. Filters were dried and exposed to a phosphorimager plate. Background binding by a labeling mixture without UhpA was less than 0.01% of filter-bound 32 P-UhpA at time 0.
Electrophoretic Mobility Shift Assay-DNA fragments used for gel mobility shift assays were generated with PCR reactions. PCR primers that annealed to the coding strand of the uhp locus were as follows: A, 5Ј-CCCTTTTTGAATTCCCAGACACC-3Ј; B, 5Ј-CCGGCAAAACTAAG-AAATTTTCCAGGTTTTGCCTGG-3Ј; and C, 5Ј-CGCTATCTCAGGCC-TGATTTGCTG-3Ј. PCR primers used to anneal to the noncoding strand of the uhp locus were as follows: 1, 5Ј-GGGTCGGATCCCGAACCTG-GTTTAA-3Ј; 2, 5Ј-GATAGCGTCCAGGCAAAA-3Ј; and 3, 5Ј-TAT-GAAGTGAAAAGGTGA-3Ј. The DNA templates were plasmid pRJK10 or plasmid pRS415:P T RsaI derivatives containing 6-bp NcoI linker substitutions at selected sites throughout the uhpT promoter (18). PCR-products were purified on Wizard spin columns (Promega, Madison, WI) and labeled at the 5Ј end by incubation with T4 polynucleotide kinase (Life Technologies, Inc.) and [␥-32 P]ATP (3,000 Ci/mmol; DuPont NEN). Nucleotide precursors were removed by gel filtration through G-50 QuickSpin columns, and the labeled DNA was suspended in buffer B (25 mM Tris-HCl, pH 7.5, 5 mM MgCl 2 , 1 mM DTT, 0.5 mM EDTA) and quantified by ethidium bromide fluorescence using the saran wrap method (20). DNA binding reactions (10 l) contained buffer C (50 mM Tris-HCl, pH 7.5, 6 mM MgCl 2 , 1 mM EDTA, 1 mM DTT, 5% (v/v) glycerol), 120 ng of poly(dI-dC) (Boehringer Mannheim), 5.4 -7 nM 32 Plabeled DNA fragment, and differing amounts of UhpA in buffer B. The reaction mixes were incubated at 25°C for 20 min. Unlabeled competitor DNA was added in H 2 O, dried by lyophilization in the reaction tube, and dissolved in binding assay components before addition of UhpA. Gels were pre-electrophoresed to constant current before samples were loaded with addition of 1.1 l of 50% glycerol with 0.25% xylene cyanol and 0.25% bromphenol blue in buffer B. Samples were subjected to electrophoresis in 1.5-mm-thick 10% polyacrylamide gels (acrylamide: bisacrylamide, 38:1) with gel and electrode buffer of 96 mM Tris, pH 8.6, 90 mM borate, 3 mM EDTA. Electrophoresis was performed at 560 V for 1 to 2 h. The gels were dried under vacuum, and the positions of radioactive fragments were visualized on a phosphor storage screen, which was analyzed on a Molecular Dynamics PhosphorImager, running the ImageQuant program.
DNase I Footprinting-PCR primers A and 1 were labeled with T4 polynucleotide kinase and [␥-32 P]ATP and purified with Sephadex G-25 QuickSpin columns. The labeled primers were used in separate PCR reactions with unlabeled primers 1 and A, respectively, and template pRJK10 to generate 210-bp DNA fragments that were 5Ј-labeled on either the noncoding or coding strand. End-labeled DNA fragments were purified on 10% acrylamide gels and suspended in buffer B. A 10-l solution containing end-labeled DNA (1.4 -2.2 nM) in buffer C was mixed with different amounts of UhpA or P-UhpA, as indicated. The DNA-protein mixtures were incubated for 20 min at 25°C. Limited digestion of the protein-bound DNA was initiated by the addition of RQ1 DNase solution (Promega) (used at 2.3 ϫ 10 Ϫ3 units/l after dilution in 10 mM Tris-HCl, pH 8.0, 10 mM MgCl 2 , 5 mM CaCl 2 , 0.1 mM DTT). Digestion was carried out for 1 min at 25°C and terminated by the addition of formamide to 50% and boiling for 5 min. DNase digestion patterns were analyzed on 6% polyacrylamide gels containing 8.3 M urea in TBE buffer. For alignment of DNase digestion patterns with DNA sequences, the 210-bp A1 PCR fragment was sequenced using primers A or 1 and run alongside the digestion reaction lanes.
Hydroxyl Radical Footprinting-Hydroxyl radical footprinting (25) was performed essentially as described by Craig et al. (26). 5Ј-32 P-Labeled primer 1 was used in a PCR reaction with primer D, 5Ј-CTCGATACCTGGCACTGGAGCGGA, and plasmid pRJK10 as template, to generate a 285-bp DNA fragment labeled on the bottom strand. Labeled DNA was suspended in buffer B (prepared with HPLC-grade water) and incubated for 20 min at room temperature in a 40-l volume with 480 ng of poly(dI-dC) and UhpA or P-UhpA at indicated concentrations. A volume of 4 l of 14 mM sodium ascorbate and 2 l of 3% hydrogen peroxide was pipetted onto the side of the microcentrifuge tube. A drop containing freshly prepared solution of 60 mM Fe(NH 4 ) 2 (SO 4 ) 2 and 120 mM EDTA was pipetted as a second drop onto the side of the tube. The solutions were mixed by gentle vortex action. Reactions were allowed to proceed for 2 min and quenched by addition of 4 l of a solution containing 143 mM thiourea and 26 mM EDTA. Following addition of 10 g of yeast tRNA and 200 mM sodium acetate, the DNA was precipitated with ethanol, denatured by dissolution in 50% formamide, and separated by electrophoresis as described in the previous section.
Mass Spectrometry-Protein molecular weight measurements were made by electrospray mass spectrometry using a Finnigan-MAT TSQ7000 mass spectrometer system, as described previously (27). Samples were introduced by either capillary column HPLC or infusion at 0.6 l/min. Electrospray ionization was carried out at 4.5 kV with a sheath liquid flow of 70% methanol, 30% water at 1.2 l/min and a co-axial nitrogen flow. Spectra were acquired by scanning the first quadrupole of the tandem quadrupole instrument and were deconvoluted using the algebraic method of Mann.

RESULTS
Overexpression and Purification of UhpA Proteins-UhpA and its D54N variant were overexpressed using the pET15b expression plasmid carrying inserts of PCR products that should yield the UhpA proteins with their natural termini. IPTG induction of synthesis of phage T7 RNA polymerase resulted in high level expression of a 24-kDa polypeptide, which was not seen in uninduced cells or in cells carrying the vector plasmid lacking an insert. Fig. 1 shows the SDS-PAGE display of the purification of the wild-type UhpA protein through the steps of polyethyleneimine precipitation, fractional precipitation between 10% and 45% saturation with ammonium sulfate, and salt gradient elution from a DEAE column. Similar results were obtained for the UhpA[D54N] variant (data not shown). The size of the 24-kDa polypeptide is in reasonable agreement with the molecular mass of 20,889 Da deduced from the uhpA nucleotide sequence, and this polypeptide migrates at the position for 21 kDa on Tricine gels. 2 Based on densitometric scanning of the electropherograms, the degree of purification was Ͼ90%.
Confirmation that the purified protein was UhpA was provided by several tests. The sequence of the first 20 N-terminal residues was determined by automated Edman degradation and matched exactly that of the deduced UhpA polypeptide (28), showing that the N-terminal methionine residue was retained in the product. Purified UhpA was subjected to analysis by electrospray mass spectrometry, and the major protein species had a molecular mass of 20,900 Ϯ 4 Da, in close agreement with its predicted molecular mass (data not shown). Finally, rabbit polyclonal antisera raised to this polypeptide reacted in Western blot analysis with a 21-kDa polypeptide that was present only in uhpA ϩ cells and was present in amplified amounts in cells with a uhpA-bearing plasmid (17).
Phosphorylation of UhpA by Acetyl Phosphate-Several response regulators can be phosphorylated and activated by incubation with low molecular weight phospho-donors, such as acetyl phosphate (13, 29 -31). The transfer of phosphate to purified UhpA from acetyl [ 32 P]phosphate and the properties of the phosphorylated protein were investigated. The requirement for magnesium and for aspartate 54 is consistent with the proposed role of the conserved acidic pocket for binding the Mg 2ϩ atom necessary for the phosphotransfer reaction (11,12,32,33). There was no detectable difference in the electrophoretic mobilities of the phosphorylated and unphosphorylated forms of UhpA during SDS-PAGE ( Fig. 2A).
The time course of phosphorylation was slow, and steadystate levels were reached only after incubation for 1 h (Fig. 3A). From the amount of radioactivity in the separated protein species, a rough estimate of the stoichiometry of labeling was 0.5-1 mol of phosphate/mol of protein. Labeling was competed by nonradioactive acetyl phosphate (Fig. 3B). Half-maximal inhibition was obtained at 20 mM acetyl phosphate, indicating that the phosphate donor was acetyl phosphate and not some other species that might have formed during the chemical synthesis of the labeled acetyl phosphate.
The stability of phospho-UhpA (P-UhpA) was determined by separating UhpA from acetyl [ 32 P]phosphate by rapid gel filtration through a QuickSpin column after a 120-min incubation period to allow maximal protein labeling. At intervals, UhpA was separated from released inorganic phosphate by collection on nitrocellulose filters. The rate of loss of filterbound radioactivity indicated a half-time for hydrolysis of about 60 min at pH 7.5 (Fig. 3C). The stability of P-UhpA is thus comparable to the t1 ⁄2 of ϳ1.5 h for P-OmpR (34) and P-PhoB (13), and is much higher than for P-CheY and P-NtrC, whose half-lives are 6 -15 s and 4 -8 min, respectively (35,36).
Properties of P-UhpA-The extent of incorporation of radioactivity, the kinetics of labeling, and the lack of phosphorylation of the D54N variant suggested that treatment with acetyl phosphate resulted in the phosphorylation of Asp-54 on the majority of the UhpA molecules. The stoichiometry of labeling was investigated in two ways. First, UhpA protein before and 2 C. Webber, manuscript in preparation. after reaction with acetyl phosphate was rapidly separated from the reaction components by gel filtration in a spin column equilibrated with water. The protein, now in the presence of reduced salt concentrations, was rapidly injected for electrospray mass spectrometry. As stated above, the molecular weight of the major species of UhpA in the unphosphorylated sample was 20,904. The molecular weight of the major species in the phosphorylated sample was 20,985, indicative of the covalent addition of a single phosphate moiety (data not shown). These spectra were complicated by the presence of a complex mixture of bound ions, which led to a number of peaks of slightly higher molecular weight that were not seen when the protein sample was infused into the spectrometer in the presence of trifluoroacetic acid. However, this standard sample infusion protocol did not allow recovery of P-UhpA, presumably owing to its rapid hydrolysis.
To test for changes in state of oligomerization following phosphorylation, samples of UhpA were incubated with acetyl [ 32 P]phosphate in the presence or absence of EDTA and were separated by PAGE under nondenaturing conditions. The mobility of the protein was determined by staining with Coomassie Blue, and the location of the radioactive species was determined by exposure to phosphor storage screens. The two images were aligned with the aid of radioactive dye marks. Fig. 4A shows that, after 2 h of electrophoresis in a watercooled electrophoresis chamber, the phosphorylated form of UhpA (lane 2) migrated more rapidly than the native form (lane 1). The pattern of radioactivity shown in Fig. 4B confirmed that the radioactivity co-migrated with the Coomassie Blue-stained material in the phosphorylated sample (lane 2) and that phosphorylation was completely blocked by EDTA (lane 1). The majority (Ͼ90%) of the protein molecules incubated with acetyl phosphate exhibited this increased mobility and hence must be phosphorylated. The increased mobility of P-UhpA may be due to a change of conformation into a more compact structure, or to an increase in the net negative charge of the protein, or both. There was no indication for the formation of higher oligomers of UhpA as a result of phosphorylation.
UhpA Binds to the Ϫ64 Element of the uhpT Promoter-Merkel et al. (18) identified four regions of the uhpT promoter that are important for its function. The binding of UhpA to these DNA elements was investigated by electrophoretic mobility shift assays using DNA fragments, which were prepared by PCR and whose locations are indicated in Fig. 5. Fragment A1 carries the entire promoter region, extending from Ϫ142 to ϩ68 bp. 32 P-End-labeled fragment A1 was incubated at 25°C for 20 min with increasing concentrations of UhpA, followed by electrophoresis in a 10% polyacrylamide gel. An unusual gel shift pattern was routinely observed (Fig. 6A), in which increasing concentrations of UhpA resulted in a progressive decrease in probe mobility until a protein-DNA complex of defined mobility was obtained at higher concentrations of UhpA. In the range of UhpA concentrations giving shifted complexes of intermediate mobility, these complexes migrated with a characteristic wavy pattern, whereas the unshifted and the maximally retarded species formed discrete, straight bands. Silver staining of the gels revealed that the bulk of the UhpA protein comigrated with the maximally shifted complex (data not shown). Immunoblot analysis of mobility shift gels using polyclonal anti-UhpA antisera (17) failed to detect any UhpA protein migrating with the intermediate shifted probe (data not shown).
The location of the UhpA-binding sites were investigated by examining the retardation of electrophoretic mobility of a series of 32 P-labeled DNA fragments that carry different portions of the uhpT promoter and were incubated in the absence or presence of 5.4 M UhpA. Fig. 6B shows that the mobility of fragments A1 (coordinates Ϫ142 to ϩ68), B1 (Ϫ81 to ϩ68), and A2 (Ϫ142 to Ϫ38) was substantially retarded by UhpA. The only region present on all three of these fragments contains the . Samples were incubated at 37°C for 1 h before the reactions were stopped by addition of 10 mM EDTA and were analyzed by SDS-PAGE, as above. C, UhpA (10 M) was incubated with 20 mM acetyl [ 32 P]phosphate at 37°C for 2 h. Excess acetyl [ 32 P]phosphate was removed by passage through a Sephadex G-50 QuickSpin column. The filtrate was incubated at 37°C for the indicated times, after which portions were removed and filtered through nitrocellulose filters (0.05 m, pore size). Filters were washed and then exposed to a phosphor storage screen. The half-life of P-UhpA was calculated with a semilogarithmic plot. Ϫ64 element, from coordinates Ϫ81 to Ϫ38. Fragment A3 (coordinates Ϫ142 to Ϫ83) contains the site for binding by CAP (6) but showed no binding by UhpA. Fragment C1 (coordinates Ϫ44 to ϩ68) contains sequences downstream of the Ϫ64 element. This fragment underwent a partial shift, characterized by lower affinity binding and a much lesser degree of retardation than was seen with the fragments possessing the intact Ϫ64 region. This behavior suggests that a low affinity UhpAbinding sequence resides downstream of the Ϫ64 element.
A second gel mobility shift assay to demonstrate UhpA binding specificity examined the ability of excess unlabeled DNA fragments to compete with full-length labeled fragment A1 (Ϫ142 to ϩ68) for binding to UhpA. As shown in Fig. 6C, fragment B1 (Ϫ81 to ϩ68), which carries the Ϫ64 element, competed with fragment A1. However, even a 250-fold molar excess of fragments A3 (Ϫ142 to Ϫ83) or C1 (Ϫ44 to ϩ68) did not block binding of fragment A1 to UhpA, showing that there were no high affinity UhpA-binding sequences outside of the Ϫ64 element.
Binding of UhpA to uhpT Promoter Variants-The genetic relevance of the in vitro UhpA binding to linear uhpT promoter fragments was tested with a series of mutant promoters. Fig.  7A indicates the sites of seven linker substitutions (designated 2-8 in this figure), in which the sequence of individual 6-bp intervals was changed to the NcoI recognition sequence, CCATGG. Promoter activities under inducing conditions were determined previously from the level of expression of uhpT-lacZ reporter constructs carrying each substitution (18). PCR primers A and 1 were used to generate 32 P-labeled DNA fragments carrying these changes, and binding of 4 M UhpA was assayed by gel mobility shift (Fig. 7B). Only the DNA fragment with a 6-bp substitution in the downstream half of the Ϫ64 element (mutant 4, at Ϫ56 to Ϫ51) showed a significant change from the band shift pattern of wild-type uhpT promoter DNA. Variant 7 (Ϫ35 to Ϫ30) contains a linker substitution in the Ϫ32 element and exhibits complete loss of promoter activity, but gave a normal pattern of UhpA binding in the in vivo titration assay (18) and in this gel shift assay (Fig. 7C).
Phosphorylation of UhpA Enhances Its DNA Binding Activity-The effect of phosphorylation by acetyl phosphate on the ability of UhpA to bind to the uhpT promoter region was examined in the electrophoretic mobility shift assay. As shown in Fig. 8A, addition of unmodified UhpA in the concentration range up to 3.4 M resulted in the characteristic, progressive decrease in mobility of the A1 DNA fragment (coordinates Ϫ142 to ϩ68) (lanes 2-6). The mobility of half of the DNA sample was shifted at 1.3 M UhpA, but only a very small (coordinates Ϫ142 to ϩ68) for 20 min at room temperature. Portions were loaded onto a 10% polyacrylamide gel (1 ϫ TBE gel and electrode buffer) and electrophoresed for 2 h in a water-cooled chamber. The position of radioactive material was determined by exposure to a phosphor storage screen. B, 32 P-labeled DNA fragments A1, B1, C1, A2, and A3 at 6 nM were incubated for 20 min at room temperature in the absence or presence of 5.4 M UhpA, and then electrophoresed as described above. Lane L contains a mixture of the five DNA fragments. C, unlabeled DNA fragments (A3, B1, and C1) at 25-, 100-, and 250-fold molar excess were tested for competition with 32 P-labeled DNA fragment A1 (7 nM) for binding to 1.3 M UhpA. portion of the DNA was present in the maximally retarded complex, even at 3.4 M UhpA. In contrast, P-UhpA displayed greatly increased DNA binding activity (lanes 8 -12). Essentially all of the DNA fragment was shifted at 0.7 M UhpA and was present in the maximally shifted complex at 2 M P-UhpA. The presence of acetyl phosphate alone did not affect the mobility of the DNA fragment (lane 7).
Acetyl phosphate had no apparent effect upon the DNA binding activity of the UhpA[D54N] protein. As shown in Fig. 8B, the variation of DNA probe mobility as a function of the concentration of UhpA[D54N] protein was roughly identical in the absence or presence of acetyl phosphate and was similar to that of unphosphorylated wild-type UhpA (compare to Fig. 8A, lanes  2-6). UhpA[D54N] was expected to bind the uhpT promoter, since overexpression of this protein results in high constitutive promoter activity (17). These data show that phosphorylation of UhpA strongly affects its ability to bind to its target DNA sequences, and are consistent with the conclusion that the requirement for phosphorylation can be circumvented by overexpression of UhpA.
DNase I Protection by UhpA and P-UhpA-The locations of the UhpA-binding sequences were shown by testing for the ability of UhpA protein species to protect against DNase I cleavage (37) of the labeled top (noncoding) (Fig. 9, A and C) or bottom (coding) strands (Fig. 9B) of the uhpT promoter. Unfortunately, few sites within the very AϩT-rich Ϫ64 element were cleaved by DNase I. On both strands, the few susceptible sites within the Ϫ64 element showed decreased or enhanced cleavage upon addition of P-UhpA. At higher concentrations of P-UhpA, the protected region extended downstream from Ϫ50 to Ϫ32. No changes in DNase I sensitivity were seen at sequences upstream of Ϫ80 or downstream of Ϫ32. These results confirm the conclusions from the gel mobility shift analysis that the Ϫ64 element (Ϫ50 to Ϫ80) is the high affinity binding site for P-UhpA, and show that the sequences immediately downstream and overlapping the putative RNA polymerase-binding region are the low affinity binding region.
Protection against DNase I digestion by UhpA and P-UhpA was compared in Fig. 9C. Although there were too few DNasesusceptible sites for definitive conclusions, both forms of UhpA protected the Ϫ64 element effectively. Extension of the footprint to the low affinity site (Ϫ50 to Ϫ32) occurred with P-UhpA at 140 nM, but required much higher concentrations of unmodified UhpA (1.3 M). Thus, phosphorylation of UhpA may enhance cooperativity between UhpA monomers at the uhpT promoter to enable formation of a maximally retarded gel complex and extension of the binding to the low affinity site.
Hydroxyl Radical Footprinting with UhpA and P-UhpA-High resolution analysis of the sites of contact of UhpA at the uhpT promoter was obtained by the technique of hydroxyl radical footprinting (25), using the DNA fragment D1 labeled at the 5Ј-end of the bottom strand. The results shown in Fig. 10 confirmed and extended the conclusions from the DNase I footprinting. P-UhpA protected five regions spaced at roughly 10-bp intervals. Three of the protected sites lie within the Ϫ64 element and share the consensus sequence RAAAYY, where R is either purine and Y is either pyrimidine base. The two protected sites downstream of the Ϫ64 element were centered at Ϫ43 and Ϫ33, and shared the consensus sequence ANGCY.
Unphosphorylated UhpA protected the three sites in the Ϫ64 element with only slightly lower affinity than did P-UhpA. However, its protection of the two downstream sites was markedly weaker, even at the highest concentration of UhpA tested. DISCUSSION The UhpA protein was overexpressed and purified to show its specific binding to DNA sequences in the uhpT promoter region and to compare its binding properties with those expected from the regulation of uhpT expression in the intact cell. Protein purification was necessary for this goal because we were unable to demonstrate a UhpA-specific electrophoretic sion. The work presented here confirms this hypothesis by demonstrating that UhpA binds preferentially and specifically to the Ϫ64 element. These results agreed with the results showing in vivo titration of UhpA activity by multicopy plasmids carrying portions of the uhpT promoter (18). No other high affinity binding sequence was present in the region from Ϫ142 to ϩ68, which is consistent with results from our study of nested deletion sets showing that no specific sequences upstream of the CAP-binding site at Ϫ120 or downstream of the transcription start site are required for normal uhpT expression and regulation. Gel-mobility shift assays indicated the low affinity binding of UhpA to sequences downstream of the Ϫ64 element. These sequences were localized by DNase I and hydroxyl radical protection assays to the region immediately downstream of the Ϫ64 element and extending to the Ϫ32 element.
The number of UhpA molecules that bind to the promoter is not known, but the extent of the activator-binding regions in the uhpT promoter appear to be simpler than for other response regulators. For example, OmpR binds to four sites in the ompF promoter and to three sites in the divergently transcribed micF-ompC promoter (38 -41). The best studied target of NarL action, the narG promoter, contains eight NarL consensus binding heptamers, although not all of them appear necessary for regulation (42).
Previous studies showed that both UhpBC function and UhpA aspartate-54 are required for uhpT expression (17,28). Overexpression of UhpA resulted in high level, constitutive uhpT expression and loss of the normal requirements for inducer, UhpBC function, and the site of phosphorylation. One interpretation of these findings is that phosphorylation of UhpA might enhance its binding to its target sequences, but that DNA binding or formation of the active structure on the DNA could be achieved by unphosphorylated UhpA at elevated concentrations. Another possibility was that overexpressed UhpA is activated by covalent modification by another sensor kinase in some form of cross-talk. The observations obtained here favor the former hypothesis. Although phosphorylation of UhpA by soluble forms of UhpB has not been obtained yet, phosphate transfer from acetyl phosphate, a characteristic of many response regulators (13,30,31,33,43), was demonstrated. As seen in the other systems, phosphate transfer to UhpA was completely blocked by EDTA chelation and by the D54N substitution. Mass spectrometric analysis of the phosphorylated species showed an increase in molecular mass of 81 Da, indicating the covalent attachment of a single phosphate group, as found for P-OmpR (44). The shift of electrophoretic mobility of P-UhpA under nondenaturing conditions indicated that the majority of UhpA molecules (Ͼ90%) were modified. Furthermore, comparison of the electrophoretic mobility of P-UhpA in native gels of various polyacrylamide concentrations, in comparison to the migration of protein standards, indicated that P-UhpA was monomeric in solution under those conditions. 3 The rates of phosphorylation and dephosphorylation of UhpA were comparable to those seen for OmpR and PhoB (13,34) and were much slower than the corresponding processes in CheY and NtrC. This slow rate of spontaneous dephosphorylation of UhpA (t1 ⁄2 around 1 h) implies that the putative cophosphatase activity of UhpB must play an important role in uhpT regulation (28).
Binding of UhpA to its DNA target displayed unusual behavior in the electrophoretic mobility shift assay. As the UhpA concentration was increased, there was a progressive decrease in the mobility of the protein-DNA complex, associated with a wavy electrophoretic pattern, until a maximally retarded complex of defined mobility and typical sharp appearance was eventually achieved. The basis for this behavior is unresolved. Each DNA "step" may represent an increasing extent of protein oligomerization on the promoter or formation of complexes that differ in protein-DNA binding stability. We currently favor the view that the shifted species of intermediate mobility represent 3  DNA molecules that have dissociated from their specific complex with UhpA during the course of electrophoresis. The progressive decrease in probe mobility with increasing UhpA concentration may result from the protein concentrationdependent rebinding of the probe to UhpA molecules. The maximally retarded complex could indicate the formation of a different type of UhpA-DNA complex that is more stable during electrophoresis, or that the UhpA concentration is high enough to trap any DNA molecules that dissociate. The latter view is favored by the co-migration of the maximally shifted complex with the bulk of the UhpA protein molecules. Their co-migration could be explained by the fortuitous balance between the retardation in mobility caused by the increased hydrodynamic volume of the UhpA-DNA complex and its increased mobility owing to the binding of the polyanionic DNA. Silver staining and immunoblot analysis did not reveal the presence of detectable amounts of protein migrating in association with the complexes of intermediate mobility, consistent with the release of the DNA during electrophoresis.
DNase I and hydroxyl radical protection footprinting extended the results of the gel shift experiment by confirming that the Ϫ64 element, from coordinates Ϫ80 to Ϫ50, was the site of high affinity binding by UhpA and showing that the extension of UhpA binding to downstream sequences occurred at higher concentrations. This extension to the vicinity of Ϫ32 could allow direct interaction with RNA polymerase. Binding of UhpA to the Ϫ64 element appears to be less strongly affected by its phosphorylation than was the ability to extend the binding complex to the downstream region. The 6-bp linker substitution at positions Ϫ51 to Ϫ56 in the downstream half of the Ϫ64 element eliminated uhpT promoter activity, increased the concentration of UhpA necessary to bind to DNA, and prevented the formation of the maximally retarded UhpA-DNA complex. The symmetrically related substitution at positions Ϫ72 to Ϫ77 in the upstream half of the Ϫ64 element only reduced promoter function by 40% and had no major effect on complex formation. The difference in the consequences of sequence changes in the two halves of the palindromic Ϫ64 element suggests that the downstream half is more important for promoter activity. Presumably, occupancy of the downstream half of the Ϫ64 element is a prerequisite for cooperative extension of complex formation. Alternatively, UhpA may not bind in a symmetrical manner to the Ϫ64 region, but segments in both halves of the Ϫ64 palindrome (Ϫ51 to Ϫ56 and Ϫ72 to Ϫ77) are contacted in an apparently symmetrical manner.
The five sites of P-UhpA binding are strongly conserved in E. coli and Salmonella typhimurium (28). The three high affinity sites are in the Ϫ64 element. They are centered at residues Ϫ74, Ϫ65, and Ϫ54 and share the consensus sequence RAAAYY. The low affinity sites centered at Ϫ43 and Ϫ33 have a different consensus sequence of ANGCY. This consensus could be a degenerate form of the high affinity site and occupancy of the low affinity sites may require cooperative extension of the protein complex from the upstream sites. There is evidence supporting the importance of the downstream region between the Ϫ64 and Ϫ32 elements for promoter activity. This region is much more strongly conserved in E. coli and S. typhimurium, in which 9 of the 11 nucleotides match, than is the region between the CAP-binding site and the Ϫ64 element, in which only 4 of the 13 nucleotides were conserved. In addition, the NcoI-linker substitution at residues Ϫ46 to Ϫ41 reduced promoter activity by 70%.
Studies are under way to identify the effect of phosphorylation on UhpA structure and activity. Phosphorylation does not appear to affect protein oligomerization in solution, as judged by PAGE and gel filtration chromatography, but it clearly affects the DNA-binding properties. Phosphorylation of other response regulators induces their formation of dimers or higher oligomers that is related to their transcriptional activity (45)(46)(47). The recent presentation of the structure of NarL, a close homolog of UhpA, reveals the presence of a DNA-binding helixturn-helix motif in the C-terminal domain that is occluded by interaction with the N-terminal phosphorylation module (15). It is likely that phosphorylation decreases the interdomain interaction to render the DNA-binding surface accessible. This model is difficult to reconcile with our finding that unphosphorylated UhpA binds to specific targets in DNA, although with somewhat lower affinity than does P-UhpA. In the case of UhpA, the primary effect of phosphorylation seems to affect its ability to extend cooperatively along the DNA helix to the downstream sites. Perhaps the interdomain interaction in UhpA does not occlude the DNA-binding site, but the conformational change induced by phosphorylation might enhance the interaction on the DNA surface of UhpA monomers. The major difference in sequence between UhpA and NarL is that the flexible hinge in UhpA is 10 residues shorter than in NarL, which might rotate the C-terminal region relative to the phosphorylation module and thereby allow continuous exposure of the DNA-binding surface. This question can only be resolved through determination of the structure of UhpA.
The Uhp system offers several technical advantages for the study of transmembrane control of transcription activation by a two-component response regulator. There is a simple inducing signal, extracellular Glu-6-P in the M concentration range. The output of this signal is the increase in transcription of a single gene, uhpT, from undetectable levels in the uninduced state. The entire transcription control region of the uhpT promoter is relatively small, at 120 bp, and the sites for UhpA binding are limited to the 50-bp region from Ϫ80 to Ϫ30. There are, however, multiple UhpA-binding sequences of different affinity within this region. Now that an active form of UhpA has been obtained, biochemical analysis of the interactions between UhpA, CAP, and RNA polymerase at the uhpT promoter are possible.