Glyoxylate and Pyruvate Are Antagonistic Effectors of the Escherichia coli IclR Transcriptional Regulator*

The Escherichia coli isocitrate lyase regulator (IclR) regulates the expression of the glyoxylate bypass operon (aceBAK). Founding member of a large family of common fold transcriptional regulators, IclR comprises a DNA binding domain that interacts with the operator sequence and a C-terminal domain (C-IclR) that binds a hitherto unknown small molecule. We screened a chemical library of more than 150 metabolic scaffolds using a high-throughput protein stability assay to identify molecules that bind IclR and then tested the active compounds in in vitro assays of operator binding. Glyoxylate and pyruvate, identified by this method, bound the C-IclR domain with KD values of 0.9 ± 0.2 and 156.2 ± 7.9 μm, as defined by isothermal titration calorimetry. Both compounds altered IclR interactions with operator DNA in electrophoretic mobility shift assays but showed an antagonistic effect. Glyoxylate disrupted the formation of the IclR/operator complex in vitro by favoring the inactive dimeric state of the protein, whereas pyruvate increased the binding of IclR to the aceBAK promoter by stabilizing the active tetrameric form of the protein. Structures of the C-IclR domain alone and in complex with each effector were determined at 2.3 Å, confirming the binding of both molecules in the effector recognition site previously characterized for the other representative of the family, the E. coli AllR regulator. Site-directed mutagenesis demonstrated the importance of hydrophobic patch formed by Met-146, Leu-154, Leu-220, and Leu-143 in interactions with effector molecules. In general, our strategy of combining chemical screens with functional assays and structural studies has uncovered two small molecules with antagonistic effects that regulate the IclR-dependent transcription of the aceBAK operon.

The Escherichia coli isocitrate lyase regulator (IclR) regulates the expression of the glyoxylate bypass operon (aceBAK). Founding member of a large family of common fold transcriptional regulators, IclR comprises a DNA binding domain that interacts with the operator sequence and a C-terminal domain (C-IclR) that binds a hitherto unknown small molecule. We screened a chemical library of more than 150 metabolic scaffolds using a high-throughput protein stability assay to identify molecules that bind IclR and then tested the active compounds in in vitro assays of operator binding. Glyoxylate and pyruvate, identified by this method, bound the C-IclR domain with K D values of 0.9 ؎ 0.2 and 156.2 ؎ 7.9 M, as defined by isothermal titration calorimetry. Both compounds altered IclR interactions with operator DNA in electrophoretic mobility shift assays but showed an antagonistic effect. Glyoxylate disrupted the formation of the IclR/operator complex in vitro by favoring the inactive dimeric state of the protein, whereas pyruvate increased the binding of IclR to the aceBAK promoter by stabilizing the active tetrameric form of the protein. Structures of the C-IclR domain alone and in complex with each effector were determined at 2.3 Å , confirming the binding of both molecules in the effector recognition site previously characterized for the other representative of the family, the E. coli AllR regulator. Site-directed mutagenesis demonstrated the importance of hydrophobic patch formed by Met-146, Leu-154, Leu-220, and Leu-143 in interactions with effector molecules. In general, our strategy of combining chemical screens with functional assays and structural studies has uncovered two small molecules with antagonistic effects that regulate the IclRdependent transcription of the aceBAK operon.
A classical bacterial repressor, termed IclR, participates in the regulation of the aceBAK operon in Escherichia coli (1,2). The aceBAK operon encodes for the enzymes of the glyoxylate bypass (3) that are required during growth on acetate since it bypasses the two CO 2 -evolving steps of the Krebs cycle (4). The expression of these enzymes is induced during growth on minimal medium supplemented with acetate or fatty acids (3) as well as in rich medium as a result of the acetate accumulation during exponential phase. According to early genetic studies (1,2) IclR represses the expression of aceBAK operon as well as its own gene by binding to specific operator sequences in the promoter region. Two separate operator sequences, termed IclRboxes, have been identified in the promoter region upstream of the aceB gene (5). The primary operator sequence (IclR box II) has been mapped between Ϫ52 and Ϫ19 bases of the aceB promoter, and binding of IclR to this site prevents the RNA polymerase-promoter interaction (5). Binding to the second site, IclR box I, which is located between Ϫ125 and Ϫ99 bases of the aceB promoter, disassembles the open complex through a protein-protein interaction between IclR and the RNA polymerase ␣-subunit (5). A third IclR-box was located between ϩ14 and Ϫ21 bases of the iclR promoter (6). All three IclR-boxes identified in the promoters of aceB and iclR genes contain near-perfect palindromes. Although not identical, their sequences share a 15-mer consensus motif, 5Ј-TGGAAATNATTTCCA, which was identified by in vitro random oligonucleotide selection experiments (7).
In growth media containing acetate, IclR is believed to release the DNA as a result of binding a small molecule. Although this molecule is likely related to a metabolite within the glyoxylate bypass pathway, the precise nature of this IclR effector remains nevertheless in doubt. Phosphoenolpyruvate was reported to alter IclR binding to DNA in vitro (8) but recently was shown to have no effect on IclR activity by Yamamoto and Ishihama (5).
IclR is the founding member of a large family (Pfam 01614, COG1414) of sequence-related microbial transcriptional regulators with more than 500 members identified in bacterial and archaeal genomes. Numerous bacterial genomes such as E. coli K12 (9) and Pseudomonas aeruginosa PA01 (10) contain multiple members of this family (8 and 9, respectively) with the largest number (40 members) identified so far in Bordetella bronchiseptica RB50 (11). The characterized members of this family are involved in the regulation of diverse catabolic pathways ranging from the degradation of plant cell polysaccharides in the plant pathogen Erwinia chrysanthemi (12) to the metabolism of aromatic acids in E. coli, Acinetobacter (13,14), and Pseudomonas (15).
The characteristic structural features for members of this family, gained primarily from the crystal structure of TM0065 from Thermotoga maritime (PDB code 1MKM), are fully consistent with the role of transcription regulators responding to chemical stimuli (16). The TM0065 structure confirmed the presence of N-terminal classic-winged helix-turn-helix DNA binding domain as well as demonstrated that the effector binding C-terminal domain has significant structural homology to the PAS/GAF domain, a known small molecule binding motif (17,18). The N-and C-terminal domains in the structure are connected by a linker helix, which together with N-terminal domain participates in protein dimerization. In the crystal lattice TM0065 dimers further oligomerize into tetramers, which appear to be the functional unit common for a number of IclR regulators (16). The C-terminal domain retains its structure and small molecule binding function when expressed without the DNA binding domain and was structurally characterized for three E. coli representatives of IclR family, KdgR (PDB code 1YSP), YiaJ (PDB code 1YSQ), and AllR 3 (Ref. 20, PDB code 1TF1). Although sharing relatively low sequence identity between them and with previously characterized TM0065, all four structures were very similar, confirming a common structural framework for most of the IclR family members.
AllR is involved in transcriptional regulation of the allantoin catabolism (19) and shares 42% sequence identity with IclR. The AllR C-terminal domain structure was recently determined in the presence and absence of its chemical effector glyoxylate (20), providing the first detailed view of the molecular interactions of an IclR family representative with a small molecule. According to the AllR-glyoxylate complex structure, the location of the effector binding pocket in IclR regulators corresponded to the co-factor-accommodating cavity in PAS/ GAF domains; however, the binding mechanisms were not conserved between these structurally similar domains (17,18,20).
To address these issues we used IclR as a test case and screened it against a large number of intermediates associated with glyoxylate bypass including phosphoenolpyruvate as part of a library of more than 150 metabolites by a high-throughput protein stability assay (25,26). This methodology takes advantage of the well documented phenomenon of enhanced thermal stability of the protein in the presence of a specifically bound ligand (27). The molecules that had a stabilizing effect were further assessed for the effector role for IclR using biophysical and functional assays.

EXPERIMENTAL PROCEDURES
DNA Manipulations and Gene Cloning-Standard methods were used for site-directed mutagenesis, chromosomal DNA isolation, restriction enzyme digestion, agarose gel electrophoresis, ligation, and transformation (28). Plasmids were isolated using spin miniprep kits (Qiagen), and PCR products were purified using Qiaquick purification kits (Qiagen). Mutagenesis was performed using the QuikChange site-directed mutagenesis kit (Stratagene).
For protein expression and purification the iclR gene was amplified from wild type E. coli BW25113 chromosomal DNA by PCR. To identify the boundaries of the ligand binding domain of IclR, a multiple sequence alignment of IclR and its homologues was performed. The link between the DNA and ligand binding domains in E. coli IclR was predicted to lie between amino acids Ser 98 and Leu 101. Two fragments corresponding to the 176 and 173 C-terminal amino acids were amplified and tested for expression of the soluble domain as previously described (20). The expression of a fragment of iclR corresponding to amino acids 98 -274 yielded a soluble domain of IclR (C-IclR) that was used for further experiments. All amplified fragments were cloned into the NdeI and BamHI sites of a modified form of pET15b (EMD Biosciences) in which a TEV protease cleavage site replaced the thrombin cleavage site and a double stop codon was introduced downstream from the BamHI site. This construct provides for an N-terminal hexahistidine tag separated from the protein by a tobacco etch virus protease recognition site (ENLYFQ2GS).
Construction of Deletion Mutants-An IclR deletion mutant was generated by the methods described by Datsenko and Wanner (29). To prepare competent cells for transformation, BW25113 (lacI q rrnB T14 ⌬lacZ WJ16 hsdR514 ⌬araBAD AH33 ⌬rhaBAD LD78 ; 29) containing pKD46 was cultured at 30°C in SOB broth (28) containing 100 g of ampicillin/ml. When the A 600 reached 0.5, the culture was centrifuged at 4000 rpm for 5 min, and the cells were washed 3 times with cold water before being resuspended in 1% of the original culture volume of water. PCR methods were used to amplify the kanamycin resistance gene (km) from pKD4 by using the primers iclRFw (5Ј-GTTCAACATTAACTCATCGGATCAGTTC-AGTAACTATTGCATTAGCTAACGTGTAGGCTGGAG-CTGCTTC-3Ј) and iclRrv (5Ј-GCGATTAACAGACACCC-TTATTCTATTGCCACTCAGGTATGATCATATGAAT-ATCCTCCTTA-3Ј). The PCR products were purified with a Qiagen kit, treated with DpnI, and repurified by electrophoresis. The km gene was transformed into BW25113-competent cells by electroporation (Gene Pulser; pulse controller at 200 ohms, capacitance at 250 microfarads, and voltage at 25 kV). After electroporation, the cells were grown with shaking in 1 ml of SOC (28) medium at 37°C for 1 h, and the cultures were plated onto Luria-Bertani (LB) agar containing 25 g of kanamycin/ml. The Km r transformants were purified on new kanamycin-LB plates. The mutants in which the target genes were replaced by the km gene were verified by PCR using the primers iclRcfw (5Ј-CTCATCGGATCAGTTCAGTAAC-3Ј) and iclRcrv (5Ј-CTTATTCTATTGCCACTCAGG-3Ј). To delete the km gene from the chromosome, pKD46 was removed from the cells by growing the bacteria at 37°C, and then pCP20, expressing the Flip recombinase, was introduced by transformation. The transformants containing pCP20 were grown overnight with shaking at 42°C, and the cultures were plated on LB agar without antibiotics. Colonies were tested for sensitivity to kanamycin and ampicillin. crp mutants were constructed by P1 transduction (30) using the strain JW3320 as donor (National BioResource Project, National Institute of Genetics, Mishima, Japan (31)).
Real-time Quantitative PCR Studies-Bacterial cells were cultured in MOPS minimal media (32) with 10 mM pyruvate as carbon source, and when required, 10 mM glyoxylate was added. When the A 600 reached 0.5, cells were collected by centrifugation at 4°C. Total RNA was subsequently isolated with RiboPure TM -Bacteria (Ambion) in accordance with the manufacturer's protocol. cDNAs were synthesized with the superscript first-strand synthesis kit (Invitrogen) in accordance with the manufacturer's instructions and stored at Ϫ80°C before use. Real-time quantitative PCR was carried out on the Applied Biosystems 7300 apparatus (Applied Biosystems) using Platinum SYBR Green qPCR SuperMix UDG (Invitrogen) in accordance with the manufacturer's recommended protocol. Primers used for the real-time PCR were as follows: for rrsC, 5Ј-CAGCCACACTGG-AACTGAGA-3Ј and 5Ј-GTTAGCCGGTGCTTCTTCTG-3Ј; for aceB, 5Ј-CTGCGTGACCATATTGTTGG-3Ј and 5Ј-CAGGCG-TGAGTAAGCATTCA-3Ј; for aceK, 5Ј-TTCGTGCCTGCTAT-CAACTG-3Ј and 5Ј-AGATATTGAGCGGCACCATC-3Ј.
Protein Purification-The His-tagged fusion proteins were overexpressed in E. coli BL21-Star(DE3) cells (Stratagene) harboring an extra plasmid encoding three rare tRNAs (AGG and AGA for Arg, ATA for Ile). The cells were grown in LB at 37°C to an A 600 ϳ 0.6, and expression was induced with 0.4 mM isopropyl 1-thio-␤-D-galactopyranoside. After the addition of isopropyl 1-thio-␤-D-galactopyranoside, the cells were incubated with shaking at 15°C overnight. The cells were harvested, resuspended in binding buffer (500 mM NaCl, 5% glycerol, 50 mM HEPES, pH 7.5, 5 mM imidazole), flash-frozen in liquid N 2 , and stored at Ϫ70°C. The thawed cells were lysed by sonication after the addition of 0.5% Nonidet P-40 and 1 mM each of PMSF and benzamidine. The lysate was clarified by centrifugation (30 min at 17,000 rpm) and passed through a DE52 column preequilibrated in binding buffer. The flow-through fraction was then applied to a metal chelate affinity column charged with Ni 2ϩ . After the column was washed, the protein was eluted from the column in elution buffer (binding buffer with 500 mM imidazole). The hexahistidine tag was then cleaved from the protein by treatment with recombinant His-tagged tobacco etch virus protease. The cleaved protein was then resolved from the cleaved His tag and the His-tagged protease by flowing the mixture through a second Ni 2ϩ column.
The purified proteins were dialyzed against 10 mM HEPES, pH 7.5, 500 mM NaCl and concentrated using a BioMax concentrator (Millipore). Before crystallization, any particulate matter was removed from the sample by passage through a 0.2-m Ultrafree-MC centrifugal filtration device (Millipore). Selenomethionine-labeled proteins were expressed using the same vector and host strain but in supplemented M9 media (33). The sample was prepared under the same conditions as the native protein except for the addition of 5 mM ␤-mercaptoethanol to the purification buffers.
Size-exclusion Chromatography-100 l of protein sample contained 10 mM HEPES, pH 7.5, 500 mM NaCl, 25 M IclR wild type or its mutant derivatives, and when indicated, 1 mM glyoxylate or 10 mM pyruvate. After 20 min of incubation on ice, samples were injected onto a Superose 12 10/300 GL gel filtration column (Amersham Biosciences) installed on an Á kta system (Amersham Biosciences) equilibrated with 10 mM HEPES, pH 7.5, 500 mM NaCl. Filtration was performed at 4°C at a flow rate of 0.5 ml/min, and the protein concentration was monitored by measuring the absorbance at 280 nm. Blue dextran 2000 was used to determine the void volume. A mixture of protein molecular mass standards containing ␤-amylase (200 kDa), bovine serum albumin (66 kDa), carbonic anhydrase (29 kDa), and cytochrome c (12.4 kDa) was applied to the column under similar conditions. The elution volumes and molecular masses of the protein standards were used to generate a standard curve from which the apparent molecular mass was determined.
Crystallization and Data Collection-The primary crystallization condition was determined by a sparse crystallization matrix (Hampton research kits Crystal Screen 1 and PEG/Ion Screen) at room temperature using the sitting drop vapor diffusion technique in 96-well plates. This condition was optimized by varying the pH and the concentration of the solutes. For apoIclR the best crystals were obtained using hanging drops in 0.2 M potassium acetate, pH 7.0, 20% polyethylene glycol 3350 for 2-5 days at room temperature. For diffraction studies, the crystals were flash-frozen with the crystallization buffer plus 30% ethylene glycol as cryoprotectant. For the complex structures the crystals obtained in the crystallization condition described above were soaked in 10 mM solution of either glyoxylate or pyruvate for 1 h. For diffraction studies, the crystals were flash-frozen with the crystallization buffer plus 25% ethylene glycol as the cryoprotectant.
Diffraction data of crystals of C-IclR were collected at 100 K at the 19ID beamline of the Structural Biology Center at the Advanced Photon Source, Argonne National Laboratory. The absorption peak and rising inflection point were determined by calculating and plotting fЈ and f Љ values against energy (34) from the fluorescence spectrum. The three-wavelength anomalous diffraction data were collected from a Se-Met-substituted protein crystal using an inverse beam strategy. All crystallographic data were measured with the custom-built 3 ϫ 3-tiled CCD (charge-coupled device) detector (35). The experiment, data collection, visualization, and processing were controlled with the HKL2000 program package (36). Statistics of data collection and processing are provided in Table 3. Diffraction data for glyoxylate and pyruvate complexes were collected at home source (Rigaku FR-E generator with Raxis4ϩϩ detector), and the data were integrated and scaled with *TREK (37).
Structure Solution and Refinement-Data from each of the three wavelengths was processed with the HKL2000 (36) and then input into the program SOLVE (38). SOLVE located 24 of a possible 28 selenium sites and gave a mean figure of merit of 0.66 (0.51 in the highest resolution bin). Density modification and automatic model building using RESOLVE (39,40) resulted in excellent maps and partial models for each of the four subunits. The models were completed with the aid of 4-fold noncrystallographic symmetry and the graphics program O (41). Refinement of atomic positions and individual b-factors using the program CNS-1.1 resulted in an R factor of 23.0% (R free 29.9%) for data from 19.91 to 2.30 Å. The final model comprises 5499 protein atoms (4 C-IclR domains) and 335 water molecules. The model has excellent stereochemistry as judged by PROCHECK (42), with no Ramachandran violations. All residues of the ligand binding domain except for three C-terminal amino acids (two in the case of subunit D) were located in the experiment. Both glyoxylate and pyruvate complexes structures were solved by molecular replacement using apo-structure as starting model. Structures were refined using REF-MAC5 (43,44).
After incubation for 20 min at 37°C, samples were separated on 5% acrylamide-bisacrylamide nondenaturing gels in 0.5ϫ Tris borate-EDTA buffer, pH 8.3 (TBE). Electrophoresis was performed at 100 V using ice-cold 0.5ϫ TBE as a running buffer, and DNA was transferred from the polyacrylamide gel to the Biodyne B Positive Nylon Membrane (Pierce) by electroblotting at 380 mA for 40 min in 0.5ϫ TBE. Transferred DNA was crosslinked for 15 min using UV cross-linker equipped with 312-nm bulbs. Biotin-labeled DNA was detected by horseradish peroxidase/Super Signal detection system (Pierce). Membranes were exposed to Eastman Kodak Co. x-ray film. For quantitative EMSA, the distribution of bound and unbound DNA was quantified using ImageJ (National Institutes of Health).
In Vitro Transcription Runoff Assays-For in vitro transcription runoff the complete aceB promoter was amplified from position Ϫ297 to ϩ434 and cloned in a derivative of the pGF-Puv vector (Clontech) in which the gfp gene was replaced with the aceB promotor in the SapI-EcoRI position. The expected specific transcript was 434 bases long. The recombinant plasmids were amplified in E. coli DH5␣, cut with EcoRI, and gelpurified. The reaction mixtures (20 l) contained template DNA (5 nM), purified IclR or its mutant forms (100 nM), and glyoxylate (1-1000 M) in a buffer consisting of 40 mM Tris-HCl, pH 8.0, 50 mM KCl, 10 mM MgCl 2 , 0.01% Triton X-100, and 1 unit of RNase inhibitor. The reaction mix was preincubated for 10 min at 37°C, and 20 nM RNA polymerase (Epicenter) was added. The reaction mix was further incubated for 5 min at 37°C, and transcription was initiated by adding 2 l of a mixture containing 2 mM each ATP, GTP, and CTP, 1.5 mM UTP, and 0.5 mM biotin-11-UTP (PerkinElmer Life Sciences). After 20 min at 37°C, the reactions were terminated with 10 mM EDTA. The ethanol-precipitated transcripts were analyzed on 6% acrylamide, 7 M urea gels. Electrophoresis was performed at 100 V using ice-cold 0.5ϫ TBE as a running buffer, and RNA was transferred from the polyacrylamide gel to the Biodyne B positive nylon membrane (Pierce) by electroblotting at 380 mA for 40 min in 0.5ϫ TBE. Transferred RNA was cross-linked for 15 min using UV cross-linker equipped with 312-nm bulbs. Detection was performed as described above. Preheated biotinylated 2-log DNA ladder (BioLabs) was used as molecular weight marker.
Screening Using Static Light Scattering-Purified C-IclR protein was screened against a library of 160 compounds using static light scattering (StarGazer) (45). Protein samples were diluted to a final concentration of 0.4 mg/ml in 100 mM Hepes, pH 7, 150 mM NaCl (26). 50-l aliquots of protein solution containing the chemical compounds at 1 mM were placed in duplicate into clear-bottom 384-well plates (Nunc TM , Nalgene Nunc International) and heated from 27 to 80°C at the rate of 1°C/min. The formation of protein aggregates was monitored by static light scattering. Images of scattered light were captured every 30 s, and the light intensities were translated to arbitrary numbers using StarGazer TM proprietary software. Intensities were plotted against temperature for each sample well, and transition curves were fitted using the Boltzmann equation. The midpoint of each transition was calculated and compared with the one calculated for the reference sample. If the difference between them was greater than 1.5°C, the corresponding compound was considered to be a "hit," and the experiment was repeated to confirm the effect.
Circular Dichroism (CD) Spectroscopy-CD spectra were collected at 25°C on a AVIV 62D CD spectrophotometer from 200 to 260 nm using a 0.1-mm path length cell with a scan rate of 100 nm/min, a time constant of 1.0 s, a bandwidth of 1 nm, and a sensitivity of 100 millidegrees. Each spectrum was the average of 10 scans. After background subtraction, the spectrum was expressed in molar ellipticity. The proteins were analyzed at a final concentration of 1 mg/ml, and the assays were performed in 10 mM HEPES, pH 7.5, 0.5 M NaCl. When necessary, glyoxylate or pyruvate was added at 1 mM.
Isothermal Titration Calorimetry-Measurements were performed on a VP-Microcalorimeter (MicroCal, Northampton, MA) at 25°C. The protein was thoroughly dialyzed against 10 mM HEPES, pH 7.5, and 500 mM NaCl. Solutions of glyoxylate (0.5 mM) and pyruvate (1.5 mM) were directly prepared in dialysis buffer. Each titration involved a series of 4-l injections of effector molecules into the protein solution. The mean enthalpies measured from injection of the ligand in the buffer were subtracted from raw titration data before data analysis with ORIGIN software (MicroCal). Titration curves were fitted by a nonlinear least squares method to a function for the binding of a ligand to a macromolecule (46). From the curve thus fitted, the parameters ⌬H (reaction enthalpy), KA (binding constant, KA ϭ 1/K D ), and n (reaction stoichiometry), were determined. From the values of KA and ⌬H, the changes in free energy (⌬G) and in entropy (⌬S) were calculated with the equation ⌬G ϭ ϪRT ln KA ϭ ⌬H Ϫ T⌬S, where R is the universal molar gas constant, and T is the absolute temperature.

Screening of a Small Molecule Library for Binding the IclR-
Significant sequence similarity between IclR and previously characterized AllR regulators, particularly among amino acids involved in glyoxylate binding (20), indicated that the IclR effector may have chemical features similar to glyoxylate rather than phosphoenolpyruvate, which was proposed as the IclR effector by Cortay et al. (8). To clarify the nature of the IclR effector molecule, we decided to perform an unbiased screen of IclR against a variety of glyoxylate-associated compounds along with phosphoenolpyruvate as part of large set of metabolic scaffolds.

TABLE 1 Stabilization effect of ligand binding
The thermal stabilization of each protein by 1 mM ligand concentration was evaluated using differential static light scattering. 1 Delta temperature was calculated as the difference in the transition temperature between the protein in the absence and in the presence of a given ligand.
The expression of the IclR Ser-97 to Arg-287 fragment produced a soluble polypeptide (see "Experimental Procedures" for details) corresponding to the effector binding domain. This protein, termed C-IclR, was tested for binding against 158 metabolic scaffolds (47) by high-throughput protein stability assay using static light scattering technology (25,26). The library contained intermediates of the glyoxylate bypass and the three carboxylic acids cycle, including glyoxylate, malate, oxaloacetate, citrate, isocitrate, ␣-ketoglutarate, succinate, and fumarate as well as phosphoenolpyruvate. The glyoxylate binding domain of AllR regulator, C-AllR (20), was also tested against the same library as a control. The thermal melt conditions for C-IclR and C-AllR were established to generate interpretable unfolding data (described under "Experimental Procedures"). The compounds that induced a shift in the midpoint transition temperature (⌬Tm) of proteins by more than 1.5°C are listed in Table 1.
Glyoxylate was identified as the strongest thermostabilizing compound for C-IclR. The shift in C-IclR ⌬Tm in the presence of glyoxylate (11.1°C) was close to the effect of this compound (9.9°C) on C-AllR. Additional compounds inducing significant stabilization of C-IclR included allantoate, pyruvate, and phosphoenolpyruvate. Apart from glyoxylate, allantoate, DL-glyceraldehyde 3-phosphate, and phosphoenolpyruvate, C-AllR was stabilized by allantoin but not by pyruvate ( Table 1). The thermal stabilizing effect of glyoxylate on C-IclR was confirmed by circular dichroism spectroscopy (Fig. 1). In agreement with the differential light-scattering results, the circular dichroism analysis showed an increase in 12°C in the melting temperature of C-IclR in the presence of glyoxylate.
All five compounds identified for C-IclR contained a common chemical motif in the form of adjacent carbonyl or hydroxyl groups (Table 1) with the strongest thermal stabilizing agent, glyoxylate, being the simplest representation of this chemical scaffold. The intermediate role played by glyoxylate in the pathway regulated by IclR also strengthens its case as a candidate for the native IclR effector molecule. At the same time identification of several metabolites causing IclR thermal stabilization and having a common chemical signature indicates that this sensor protein might be reactive to more than one effector molecule.

Effect of Small Molecule Modulation of IclR Binding in Vitro-
The effect of glyoxylate, allantoate, DL-glyceraldehyde-3-phosphate, pyruvate, and phosphoenolpyruvate on IclR activity was   6, and 14), phosphoenolpyruvate (lanes 7 and 8), and glyceraldehyde 3-phosphate (lanes 9 and 10) were added at a concentration of 0.1 mM (3, 5, 7, 9, and 13) or 1 mM (4,  6, 8, 10, and 14). tested in vitro by EMSA. The binding of IclR to the DNA fragment corresponding to the promoter region of the aceBAK operon and containing IclR box II (5) was analyzed in the presence and the absence of 100 M or 1 mM concentrations of each compound. According to EMSA results, glyoxylate caused dissociation of the IclR regulator from DNA, thus increasing the fraction of unbound DNA (Fig. 2). On the contrary, pyruvate and, to a lesser extent, DL-glyceraldehyde 3-phosphate and phosphoenolpyruvate stabilized the IclR-DNA complex and resulted in a decrease of unbound DNA (Fig. 2). Allantoate did not affect the IclR/aceB complex (data not shown).
In vitro transcription run-off (Fig. 3) and EMSA (Fig. 4) experiments were performed at different concentrations of glyoxylate. In the in vitro transcription run-off experiment the level of the aceB transcript was normalized to the concentration of the bla transcript used as internal control. The presence of IclR reduced the level of the aceB transcript to 30% of the level measured in the absence of this protein (Fig. 3, lanes 1 and 2). The addition of glyoxylate resulted in gradual increase of the aceB transcription, reaching up to 74% of the original value at 50 M glyoxylate (Fig. 3, lane 6). In EMSA experiments more than half of the bound operator DNA was released by IclR in the presence of 5 M glyoxylate (Fig. 4A), whereas 50 M pyruvate was required to achieve 100% DNA shift (Fig. 4B). Although a substantial difference in the concentrations of glyoxylate needed to decrease IclR binding to DNA was observed between the two techniques tested, they both confirmed the dissociation of the IclR-operator complex in a concentration-dependent manner. This apparent discrepancy could be partially explained by the IclR interactions with the C-terminal domain of the RNA polymerase ␣ subunit, as described earlier (5).
Phosphoenolpyruvate and DL-glyceraldehyde 3-phosphate were able to stabilize the IclR-DNA complex (Fig. 2); however, the concentrations required to obtain this effect (over 10 mM, data not shown) were out of the physiological range for these compounds (47)(48)(49)(50). Thus, phosphoenolpyruvate and DL-glyceraldehyde 3-phosphate were not further considered for an IclR effector role.
Because glyoxylate and pyruvate had distinctive and antagonist effects on the formation of the IclR-DNA complex, in the next series of EMSA experiments we analyzed the combined effect of glyoxylate and pyruvate on the stability of the IclR-DNA complex. In this setting one of compounds was added in increasing amounts, whereas the concentration of the other was kept constant. The results obtained (Fig. 4A) demonstrated that the glyoxylate-induced dissociation of the IclR-operator complex was overturned by a 50-fold excess of pyruvate (500 M pyruvate with 10 M glyoxylate). These results were corrob- orated by doing the reciprocal experiment (Fig. 4B). The common chemical features of these compounds suggested that this effect might be due to the competition for the same binding site in IclR, with glyoxylate having a significantly higher binding affinity than pyruvate.
To further characterize IclR interactions with potential effector molecules, the thermodynamic properties of IclR interactions with glyoxylate and pyruvate were determined using isothermal titration calorimetry. The titration of C-IclR with each compound followed an exothermal heat change profile, giving rise to a sigmoidal binding curve with glyoxylate or hyperbolic with pyruvate (Fig. 5). The calculated thermodynamic parameters are summarized in Table  2. The stoichiometry of the reaction, 0.5, is consistent with the binding of one ligand molecule per IclR dimer. In accordance with EMSA results, the C-IclR dissociation constant (K D ) for glyoxylate (0.9 Ϯ 0.2 M) was significantly lower (ϳ150 times) than that for pyruvate (156.2 Ϯ 7.9 M). Although the K D values obtained for binding of glyoxylate and pyruvate to the fulllength IclR were higher (14.2 Ϯ 1.2 and 321.7 Ϯ 20 M, respec-tively), their relative ratio was within the same range as obtained for C-IclR.
In conclusion, our EMSA results show glyoxylate as the strongest candidate for the IclR effector molecule, which destabilized IclR interactions with DNA in vitro. By contrast, pyruvate affected IclR interactions by stabilizing the IclR-DNA complex.
Modulation of IclR Binding by Glyoxylate in Vivo-To assess the effect of glyoxylate on IclR activity in vivo, the expression level of aceBAK operon was measured during growth on minimal media in the presence and absence of this compound. Although early genetic studies identified IclR as the main regulator of aceBAK operon (51), recent E. coli transcriptome analysis indicated that this operon is also under control of the global catabolite regulation protein, Crp. The Crp potential binding sequence was found overlapping the IclR box II (52). In light of this new data, the IclR activity was monitored in the presence and the absence of a functional Crp regulator.
Single ⌬iclR and ⌬crp as well as double (⌬iclR⌬crp) deletion strains were prepared in the E. coli BW25113 strain background. Deletion strains were grown to exponential phase on MOPS minimal medium with pyruvate (10 mM) as the carbon source in the presence and in the absence of 5 mM glyoxylate. The expression levels of aceB and aceK genes were measured using real-time quantitative PCR. The expression profiles of both genes followed the same trend; thus, only the results for aceB gene will be discussed further on. The expression of aceB in the ⌬iclR and ⌬crp strains was increased by 3.3-and 24-fold, respectively, compared with the wild type strain. The double ⌬iclR⌬crp deletion caused a 31-fold increase in aceB expression corresponding to the combined effect of each gene deletion. The presence of glyoxylate in the media had no significant effect on the level of aceB expression in ⌬iclR and ⌬iclR⌬crp strains lacking the functional IclR regulator, whereas it triggered a 1.4fold increase when the IclR regula-  tor was present in the cells (⌬crp strain) (Fig. 6). The glyoxylatemediated induction was dose-dependent since in the presence of 10 mM glyoxylate a 2.7-fold induction in aceB expression was observed (data not shown). In the case of the wild type strain containing both IclR and Crp regulators, the presence of glyoxylate in the media did not result in a detectable increase of aceBAK, expression probably due to effector-independent tight repression by the functional Crp regulator.
The substantial derepression effect of the Crp deletion confirmed the major role played by this global regulator on the expression of the aceBAK operon. The effect of the IclR deletion was significantly smaller yet consistent with its role as a specific regulator of this operon. The derepression effect in the presence of glyoxylate observed in the ⌬crp strain appears to be specific to the presence of functional IclR, indicating that glyoxylate can affect the activity of the IclR regulator in a manner expected from the native effector molecule. Unfortunately we were unable to study the corepressor effect of pyruvate in vivo since the crp deletion strain is unable to grow on the variety of carbon sources (52) required for these kinds of studies.
Structure of the IclR Effector Binding Domain-To provide a structural framework for IclR interactions with the newly identified small molecules, the crystal structures of the IclR effector binding domain alone (apo structure) and in complex with glyoxylate and pyruvate were determined at 2.3, 1.8, and 2.3 Å, respectively. Final R/R free for the apo structure was 0.23/0.299 (PDB code 1TD5). Final R/R free for complex structures was 0.215/0.262 (PDB code 2O99) for the glyoxylate and 0.177/0.233 (PDB code 2O9A) for the pyruvate complexes. All three models possessed excellent stereochemistry as judged by PROCHECK (42), with no Ramachandran violations. Statistics for all three structures are shown in Table 3. To ease the description and comparison to other structures, the assigned numbers for ␤ sheets and ␣-helices are given in relation to the full-length homolog TM0065 (16).
The C-IclR domain displays a similar architecture to the two other IclR family members whose structures have been determined, TM0065 (16) and AllR (20). The C-IclR structure comprises a centrally located six-stranded anti-parallel ␤-sheet surrounded on one side by two long ␣-helices (␣5 and ␣9) and on the other by three shorter ␣-helices (␣6 -␣8). The ␤-sheet is strongly curved, with the shortest helix (␣6) fitting on the inside of the half-barrel (Fig. 7).
In complex structures both the glyoxylate and pyruvate molecules occupied the cavity corresponding to the effector binding pocket in the AllR structure (20). The map quality was assessed by constructing an omit F o Ϫ F c electron density map. The map was calculated after 10 cycles of refinement without the ligand in the model (Fig. 8). Similar to the C-AllR-glyoxylate complex structure, no significant conformational changes were observed in C-IclR complexes with pyruvate or glyoxylate when compared with C-IclR apo structure. The ligands were bound iclR crp iclR  in the same orientation with almost identical positions for atoms C1, C2, O1, O2, and O3 (Fig. 8). Oxygen atoms of glyoxylate and pyruvate were coordinated by an extensive network of hydrogen bonds with the side chains and backbone of the protein. For each ligand the O1 atom is hydrogen-bound to the Asp-212O␦2 and Gly-160N, the O2 atom to the Ser-239O␥ and Ser-241O␥, and the O3 to the Ala-161N. The C1-Sg distance varies from 1.86 to 2.09 Å for glyoxylate complex and from 1.90 to 1.97 Å for the pyruvate complex. Both complex models were refined with the same restraints with no covalent bond between ligand C1 and Cys-222S␥ in the dictionary. In agreement with previous sequence analysis, all but two amino acids, Ile-134 and Leu-143, in the effector binding area are conserved between IclR and AllR (20). The corresponding positions in AllR are represented by leucine (Leu-129) and methionine (Met-138), respectively (Fig. 9).
The presence of an isoleucine residue at position 134 in the IclR structure results in the shortening of the distances between Leu-154 and glyoxylate atoms by 2 Å in the IclRglyoxylate complex structure compared with the same region in C-AllR-glyoxylate structure (contact distance C␦1 to C2, 3.68 versus 4.44 Å; C␦2 to C1, 4.17 versus 6.16 Å). In addition, the average distances between protein atoms are shorter by an average of 1 Å in C-IclR-glyoxylate versus C-IclR structures (10.8 and 11.6 Å, respectively), whereas there is no such difference between C-AllR and C-AllR-glyoxylate structures (11.96 and 12.1 Å, respectively) for the same parameters. These structural dissimilarities between binding cavities in C-IclR and C-AllR may contribute toward close to a 10-fold difference in K D for glyoxylate as determined by isothermal titration calorimetry ( Table 2).
Taking into account the almost identical position of glyoxylate and pyruvate in the C-IclR structures, the interactions involving the methyl group of pyruvate, which is absent in glyoxylate, become critical in understanding the different effects of these compounds on IclR activity. The methyl group is well defined in the pyruvate electron density map (Fig. 8B) and is oriented toward the side chain of the Leu-143 residue belonging to the other molecule of the C-IclR dimer. The distances between C3 of the pyruvate and C␦2 atom of Leu-143 are on average 4.24 Ϯ 0.4 Å for the four molecules in the C-IclR asymmetric unit. Leu-143 makes part of the hydrophobic patch formed by side chains of Met-146, Leu-154, His-216, and Leu-220, which are in the range of van de Waals interactions with the pyruvate methyl group. The presence of methionine instead of leucine in the corresponding position in AllR would explain why pyruvate had no stabilization effect (Table 1).
In C-IclR the interdomain interactions between C-IclR monomers are similar to those observed in the C-AllR and TM0065 structures. The C-IclR and C-AllR dimers are superimposable to a root mean square deviation (r.m.s.d.) of 1.75 Å, compared with an r.m.s.d. of 1.01 Å for individual subunits. This common arrangement of the C-terminal domains of three different IclR family members in three different unit cells supports the physiological relevance of these interactions for tetramerization of IclR regulators (Fig. 7B). The additional intermolecular interactions between the pyruvate methyl group and hydrophobic patch observed in the C-IclR-pyruvate complex structure would, thus, stabilize the tetramerization state of IclR protein.
Effect of Glyoxylate and Pyruvate on the Oligomerization of IclR-To establish the effect of glyoxylate or pyruvate on the oligomeric state of IclR protein, this protein was preincubated with 1 mM glyoxylate or 10 mM pyruvate and analyzed by gel filtration. The untreated IclR protein was used as the control.
In the absence of effector molecules IclR was eluted in two peaks corresponding to the dimeric (60 Ϯ 5 kDa) or tetrameric (120 Ϯ 12 kDa) states. The abundance of both fractions (54/46% relative percentage, respectively) indicated that under our experimental conditions the IclR protein is in a dynamic equilibrium between these two oligomeric states (Fig. 10, Table 4). After preincubation with glyoxylate, IclR was eluted predominantly in the form of the lower (60 kDa) molecular mass fraction, indicating a dramatic shift of equilibrium toward the dimeric state. On the other hand, preincubation with 10 mM pyruvate resulted in elution of IclR in the form of the higher (120 kDa) molecular mass fraction corresponding to the tetramer. Similar results were obtained if glyoxylate was included in the elution buffer; however, the tetrameric state of IclR was more favorable if pyruvate was included in the elution buffer. These results can be explained by the difference in affinity between the ligands (Table  2). Overall, size exclusion experiments demonstrated that glyoxylate and pyruvate have a dramatic effect on the oligomeric state of the IclR regulator. This suggests that the glyoxylate effect on IclR activity is due to the destabilization of tetramers, which is required for efficient binding to the operator DNA, whereas the co-repressor effect observed for pyruvate might be due to the stabilization of the IclR tetramer on the DNA.

Functional Analysis of Key Amino Acids Involved in Interdomain Interactions and Effector
Binding-In light of the important effect of glyoxylate and pyruvate on the IclR oligomeric state, we investigated the specific role of amino acids Leu-143, Met-146, Leu-154, and Leu-220 in the interdomain interactions and effector binding. In the course of this analysis each of these amino acids was replaced by alanine or aspartic acid. The mutant proteins were purified and submitted to gel filtration analysis and EMSA with the IclR box II fragment in the presence and absence of glyoxylate or pyruvate (Table 4, Figs. 11 and 12).
The alanine substitution of Leu-143 did not significantly affect the gel filtration profile of IclR (Fig. 11). However, according to the EMSA experiments a higher concentration (100 nM) of the L143A mutant protein was required to achieve the DNA binding comparable with the wild type IclR ( Table 4). The addition of pyruvate stabilized complex of this mutant with operator DNA similarly to the wild type IclR, whereas the glyoxylate had a significantly lower effect on the L143A mutant (Fig. 12).
In the presence of the latter ligand an additional band was observed on the EMSA gels. This additional band migrated faster than the one corresponding to the wild type IclR-DNA complex but slower that the free DNA (Fig. 12). This band would correspond to an IclR-DNA complex with a lower molecular weight than in case of the wild type IclR. Taking into account that IclR might bind DNA as a tetramer (5), this lower molecular weight complex would correspond to the association of operator DNA with the dimers of L143A as an intermediate in the releasing of IclR from the protein-DNA complex.  A and C) or pyruvate (B and D). In A and B the side chains of amino acids within a distance of 8 Å are shown and labeled in the figure. Protein residues (yellow), glyoxylate (pink), and pyruvate (green) are represented as stick figures. The backbone of the protein is shown as a thin yellow line. Leu-143 from another monomer is represented in cyan. In C and D a 3 F o Ϫ F c omit map was built using the REFMAC5 program (43,44) for the glyoxylate and pyruvate complex structures, respectively. The map was calculated after 10 cycles of refinement without the ligand in the model. The omit F o Ϫ F c electron density map covering the ligands is shown in stick representation (blue for C or cyan for D). Neighboring residues are also shown in stick representation (cyan in C or yellow in D). The gel filtration profiles of the alanine substitutions of Met-146, Leu-154, and Leu-220 showed them predominantly as dimers in solution (70/30 relative percentage, respectively, Table 4); however, they were still able to tetramerize over DNA as evidenced by a high molecular weight migrating complex (Fig. 12). The mutant proteins showed a significant variation of responses to glyoxylate and pyruvate (Table 4). Although the M146A mutant response to these ligands was similar to the wild type IclR protein, the L220A mutant showed no response to glyoxylate but was stabilized by pyruvate. The L154A mutant, to the contrary, responded to glyoxylate but not to pyruvate. These results put in evidence the different involvement of these amino acids in ligand binding.
In the next series of mutagenesis the hydrophobic region was altered by single replacement of Leu-143, Met-146, Leu-154, or Leu-220 residues to the charged residue, aspartate. The mutations had a dramatic effect on IclR oligomerization, with all mutant proteins being dimers as judged by gel filtration analysis (Table 4). According to EMSA results, L143D and L220D bound the operator DNA at a higher protein concentration (100 -150 nM) if compared with the wild type regulator (50 nM) and formed only low molecular weight complexes with DNA. Both mutants were not able to respond to either glyoxylate or to pyruvate (Fig. 12).
EMSA experiments on M146D showed low and high molecular weight protein-DNA complexes (Fig. 12). In agreement with the mutations to alanine, M146D does not interfere in the binding of glyoxylate or pyruvate. Moreover, glyoxylate disrupted the formation of the high molecular weight protein-DNA complex, increasing the amount of the lower molecular weight complex (Fig. 12). The presence of pyruvate, on the contrary, increases the formation of the high molecular weight complex. These results show that although the Met-146 is important for the stabilization of tetramers in solution, it is not directly involved in the binding of the ligands, although it might contribute to the overall hydrophobicity of the area.
In the case of L154D, the EMSA profile was similar to that of the wild type IclR; however, the response to glyoxylate was weaker (Fig. 12). Pyruvate had no stabilization effect on this mutant protein (Fig. 12), confirming the Leu-154 role in the specific interactions with pyruvate.

DISCUSSION
The IclR family of transcriptional regulators shares common structural features with their C-terminal domains representing small molecule binding modules adapted to recognize a wide variety of cellular metabolites as effector molecules. The identity of the effector molecules remains  a Protein necessary to obtain a 50% DNA shift; glyoxylate and pyruvate were added at 1 mM. b Gel filtration was performed using a Superose 12 column. Protein (100 g), glyoxylate and pyruvate were added at 1 or 10 mM, respectively. Relative percentages were calculated from the high of the peaks. c Concentration of protein necessary to obtain a 50% DNA shift on EMSA. d The effect of the ligand on the protein/DNA complex was classified as ϪϪϪ ϭ 95%, ϪϪ ϭ 50%, Ϫϭ 25% decrease or ϩϩϩ ϭ 95%, ϩϩ ϭ 50%, ϩ ϭ 25% increase in binding.
NE ϭ no effect on the stability of the protein/DNA complex. e ND, not determined. unknown for a vast majority of these proteins. This reflects a general situation within the large transcriptional factor families such as LysR (53), LuxR (54), TetR (55), and GntR (56), account-ing for thousands of transcriptional regulators that respond to unidentified chemical stimuli which need to be addressed.
Using the founding member of the IclR family as a test case, we explored the possibility of screening a small molecule library using a high throughput protein stability assay as a general strategy to identify putative ligands for such families' representatives. Although the phenomenon of protein thermostabilization by a bound small molecule was known for decades (57)(58)(59), only recently it became applicable in screening chemical libraries (25,26,60). The breakthrough came from the development of technologies enabling the thermo-denaturation of a large number of protein samples in parallel monitoring by static light scattering or fluorimetry adapted to microplate format (25,26). This kind of technology enables the high-throughput screening of protein-ligand interactions independent of the knowledge of the protein function. A similar technology named Ther-moFluor was previously used to identify and elaborate upon the hypothetical function of a pyridoxal phosphate-dependent aminotransferase enzyme (61). In this study we screened a small molecule library containing representatives of the main chemical scaffolds identified in the E. coli metabolome (47) as well as a large number of chemicals associated with the three carboxylic acids cycle and the glyoxylate bypass. Glyoxylate was identified as the strongest thermo-stabilizing compound for IclR. In EMSA and in vitro transcription runoff experiments the presence of glyoxylate disrupted IclR complex formation with operator DNA, an effect that would be expected from a physiologically relevant effector. Allantoate, DL-glyceraldehyde 3-phosphate, pyruvate, and phosphoenolpyruvate also had a stabilizing effect on IclR. Surprisingly, three of these compounds stabilize the binding of IclR to DNA. All five compounds demonstrated a common chemical moiety in the form of vicinal carboxyl groups. Structural analysis of the C-IclR complexes with glyoxylate and pyruvate clearly showed that this chemical substructure is the key element in interactions with IclR. This observation indicates that the chemical screen-based approach can provide the necessary information for identification of chemical moieties specifically recognized by IclR family members as well as for other effector-binding proteins in general.
Only two compounds, glyoxylate and pyruvate, altered the binding of IclR in vitro at a physiologically relevant concentration. The role of glyoxylate as IclR effector is supported by its role as an intermediate in the glyoxylate bypass, which is regulated by IclR. The importance of this pathway for cellular metabolism is reflected in the complex regulation of this bypass at transcriptional and post-transcriptional levels. At the transcriptional level it  The glyoxylate and pyruvate binding on mutant IclR proteins was tested by EMSA. The name of the corresponding mutant protein and its concentration used for binding is indicated above each image. Glyoxylate and pyruvate were tested at 1 mM. The full binding conditions are described under "Experimental Procedures." is tightly regulated by the binding of at least six different transcription factors, and it is extremely dependent on the metabolic state of the cell. The aceBAK operon is under negative regulation by IclR, FadR (indirectly by regulation of the iclR expression (62), Crp (52), and ArcA (63). Additionally, it is under positive regulation by integration host factor (IHF) (64) and FruR (65,66). Once induced at transcriptional level, the flow through the pathway is regulated in part by the inhibitory phosphorylation of isocitrate dehydrogenase (67,68). Gluconeogenesis also plays an essential role during growth on acetate by synthesizing glucose from nonhexose precursors. The genes pckA and ppsA belong to two parallel pathways for the gluconeogenic conversions of the three carboxylic acid cycle intermediates to phosphoenolpyruvate with the malic enzymes (sfcA and maeB) and ppsA forming one path for the conversion of malate to phosphoenolpyruvate and pckA forming the other for the conversion of oxaloacetate to pyruvate (69 -71).
Taking into account the physiological data, we could hypothesize that the cell could maximize the efficiency of expression/ repression of the aceBAK operon by using two antagonistic effectors on the same transcription factor. By EMSA and biophysical tests we were able to establish that IclR binds glyoxylate and pyruvate. Pyruvate is the first non-phosphorylated intermediate in glycolysis and a key intermediate in catabolic and biosynthetic reaction pathways. There are several pathways that deliver this compound (48,72,73). Siddiquee et al. (48,49) have shown that the intracellular concentration of pyruvate can be from 0.15 to 0.4 mM depending if samples were taken from batch or continuous cultures. These values are within the affinity constants defined for IclR binding. Glyoxylate, on the other hand, is an important intermediate of the central microbial metabolism in the glyoxylate bypass. Glyoxylate is also generated from glycolate or purine degradation (74,75). The intracellular concentration for this metabolite in E. coli has not been reported; however, there are studies in the closely related bacterium Salmonella typhimurium (76) which suggest that the concentration of glyoxylate should be very low (ϳ10 M). Thus, the big difference in affinity between glyoxylate and pyruvate binding to IclR might be linked to their intracellular concentrations.
The posttranscriptional regulation of isocitrate dehydrogenase, termed the branch point effect, has been the focus of numerous studies (77,78). This effect is a consequence of the difference in the affinities of isocitrate dehydrogenase and isocitrate lyase for isocitrate (K m of 8 and 600 mM, respectively) (77,78). As a result, the flux of the glyoxylate bypass is strikingly sensitive to the phosphorylation state of isocitrate dehydrogenase. It has been proposed that the presence of pyruvate during the acetate metabolism alters the isocitrate dehydrogenase activity, resulting in inhibited growth due to reduced carbon flux through the glyoxylate shunt (79). Our results clearly show that the effect observed by El-Mansi et al. (79) could be explained by the binding of pyruvate to IclR, which would quickly repress the expression of aceK (encoding for isocitrate dehydrogenase kinase/phosphatase) and would be reflected as a decrease in isocitrate dehydrogenase activity.
According to the data on IclR presented in this study, a mechanism of dual regulation by two effectors can be proposed.
When cells grow on complex media or if glucose is added to the media, Crp is the primary transcription regulator (52). The DNA binding sequences for IclR (box II) and CRP in the ace-BAK operon have been predicted in silico to overlap (52). However, when the levels of cAMP increase and CRP is inactive (i.e. growth on acetate) the regulation due to IclR becomes predominant. Under these conditions the concentration of glyoxylate increases due to an increase in the flux through the shunt and is now able to compete with pyruvate for binding of IclR. We suggest that glyoxylate is the primary ligand for IclR, which will trigger IclR dissociation from the DNA. However, when the concentrations of pyruvate are increased due to gluconeogenesis, the complex aceB-IclR-pyruvate might be formed as a safe mechanism to quickly repress the glyoxylate shunt. This mechanistic hypothesis is supported by the fact that both ligands bind to the same site in IclR.
According to the structural analysis and mutagenesis studies presented in this work, the hydrophobic patch formed by Leu-143, Met-146, Leu-154, and Leu-220 plays a critical role in IclR interactions with pyruvate. Moreover, mutagenesis studies demonstrated that Leu-154 substitution by alanine would abolish the pyruvate binding by IclR while leaving the binding of glyoxylate unaffected.
Glyoxylate interactions with IclR strongly resemble those of glyoxylate with another E. coli member of this protein family, the allantoin utilization regulatory protein (AllR). The two proteins share 42% sequence identity, and all but two amino acids, Ile-134 and Leu-143, are conserved in the effector binding area (20). The binding of glyoxylate causes conformational changes in both IclR and AllR proteins, resulting in their dissociation from corresponding DNA binding sites. Nevertheless, in the case of IclR, the glyoxylate-bound conformation causes the shift from tetramer to dimer state, whereas in case of AllR the binding of glyoxylate appeared not to affect its dimer to tetramer ratio (20). In the AllR-glyoxylate complex structure Met-138 (corresponding to the Leu143 in IclR) is involved in interactions with the glyoxylate bound to the adjacent AllR molecule. This interaction would strengthen the tetramerization interface in the glyoxylatebound conformation of AllR; however, it is absent in the IclR-glyoxylate complex structure. Full understanding of the differences between IclR and AllR interactions with glyoxylate will have to wait for the structural information on the full-length IclR and AllR in complex with this ligand.
The presence of two similar regulators, AllR and IclR, responsive to the same effector in E. coli, might be conditional upon different physiological roles played by these regulators. The pathways regulated by these transcriptional factors share glyoxylate as a common intermediate; however, the physiological conditions required for their expression are quite different. AllR regulates the utilization of allantoin as the sole nitrogen source. The genes coding the enzymes involved in this metabolic process are organized in three transcriptional units (allA, gcl-hyi-glxR-o484-allB-o433-glxK, and allD-f411-f261) that constitute the allantoin regulon. These three units are induced by growth on allantoin or glyoxylate and only seem to be regulated by the action of AllR and AllS (80). On the other hand, IclR represses the expres-sion of the genes that code for the enzymes of the glyoxylate shunt, and the regulation of this operon is extremely complex. Isothermal titration calorimetry experiments indicated that AllR has a 10-fold lower affinity for glyoxylate than does IclR. It may be possible that the regulators are tuned to launch their transcription programs as graded responses to the presence of glyoxylate and that the intracellular concentrations of glyoxylate integrate different metabolic inputs, i.e. acetate metabolism for IclR and allantoin utilization for AllR. Oh et al. (81) characterized the transcript profile of E. coli in acetate cultures using microarrays. The authors observed that not only the genes involved in glyoxylate shunt pathway (ace and glc operons) but also those involved in other glyoxylate-related metabolic pathways such us glycolate and allantoin metabolism were all up-regulated. In a similar manner, Pellicer et al. (82) observed cross-induction of glcDEF and aceBAK operons attributable to pathway intersection at the glyoxylate level. However, in this case the signal molecule for each transcription factor is different.
In summary, we present the binding of two small molecule effectors to a single transcription factor (IclR) with antagonistic functional effects in E. coli. The two ligands recognized by the IclR transcriptional regulator act as corepressor (pyruvate) or activator (glyoxylate) on the transcription of the aceBAK operon, likely in response to intracellular concentrations of these compounds in the bacterial cell.