Trans-aconitate methyltransferase regulator (TamR) is a member of the ligand-responsive multiple antibiotic resistance regulator (MarR) family of transcription factors. In Streptomyces coelicolor, TamR regulates transcription of tamR (encoding TamR), tam (encoding trans-aconitate methyltransferase) and sacA (encoding aconitase); up-regulation of these genes promotes metabolic flux through the citric acid cycle. DNA binding by TamR is attenuated and transcriptional derepression is achieved on binding of ligands such as citrate and trans-aconitate to TamR. In the present study, we show that three additional genes are regulated by S. coelicolor TamR. Genes encoding malate synthase (aceB1; SCO6243), malate dehydrogenase (mdh; SCO4827) and isocitrate dehydrogenase (idh; SCO7000) are up-regulated in vivo when citrate and trans-aconitate accumulate, and TamR binds the corresponding gene promoters in vitro, a DNA binding that is attenuated by cognate ligands. Mutations to the TamR binding site attenuate DNA binding in vitro and result in constitutive promoter activity in vivo. The predicted TamR binding sites are highly conserved in the promoters of these genes in Streptomyces species that encode divergent tam–tamR gene pairs, suggesting evolutionary conservation. Like aconitase and trans-aconitate methyltransferase, malate dehydrogenase, isocitrate dehydrogenase and malate synthase are closely related to the citric acid cycle, either catalysing individual reaction steps or, in the case of malate synthase, participating in the glyoxylate cycle to produce malate that enters the citric acid cycle to replenish the intermediate pool. Taken together, our data suggest that TamR plays an important and conserved role in promoting metabolic flux through the citric acid cycle.
Members of the multiple antibiotic resistance regulator (MarR) family of transcriptional regulators are involved in the regulation of a variety of important biological processes. As part of its regulatory mechanism, a MarR family member often responds to specific ligands [1–6]. For instance, HucR from Deinococcus radiodurans can regulate the expression of the adjacent uricase-encoding gene by responding to the ligand urate, the substrate for uricase [7,8]. In the absence of ligand, gene expression is repressed, while binding of urate to HucR attenuates DNA binding, resulting in increased gene activity. HucR is a founding member of a subfamily of MarR proteins (named UrtR for urate-responsive transcriptional regulator). Compared with canonical MarR homologues, these proteins are characterized by an N-terminal extension containing a conserved tryptophan as well as three other residues inferred from biochemical analyses to be involved in ligand binding or in stabilizing the protein fold [8–11].
Trans-aconitate methyltransferase regulator (TamR), a MarR homologue from Streptomyces coelicolor, was previously shown to belong to this subfamily. Streptomycetes are best known for their complex secondary metabolism and the production of a large number of clinically significant antibiotics. Hundreds of Streptomyces species have been characterized, of which only a few are either plant or human pathogens . TamR regulates the transcription of tamR, tam (encoding trans-aconitate methyltransferase; Figure 1A) and sacA (encoding aconitase) . During the aconitase-catalysed citrate isomerization step of the citric acid cycle (Figure 1B), the reaction intermediate cis-aconitate may be released from the enzyme–substrate complex; if this occurs, it will be spontaneously converted to the more stable isomer trans-aconitate . Unlike cis-aconitate, trans-aconitate is an efficient inhibitor of aconitase, thus blocking the conversion of citrate to isocitrate [15–18]. Trans-aconitate methyltransferase can use trans-aconitate as a substrate and catalyse a methyl group transfer from S-adenosyl methionine (SAM) to form the trans-aconitate methylester, whose ability to function as an inhibitor is significantly attenuated. Therefore, TamR-mediated up-regulation of trans-aconitate methyltransferase prevents the accumulation of trans-aconitate and ensures metabolic flux through the citric acid cycle [17,18]. This is further reinforced by the simultaneous regulation of an aconitase-encoding gene by TamR . Up-regulation of both genes is accomplished by binding of the structurally similar ligands citrate, trans-aconitate, cis-aconitate and isocitrate to TamR, resulting in attenuation of DNA binding; these ligands are all closely related to the citrate isomerization step catalysed by aconitase . As a circumstance under which citrate and trans-aconitate may accumulate involves the inactivation of aconitase by oxidative disassembly of its iron–sulfur cluster, one important function of TamR may be to restore metabolic flux through the citric acid cycle under conditions of oxidative stress. That the tam–tamR locus organization and the TamR binding sites in the tam–tamR intergenic region and in the promoter of sacA are conserved in different Streptomyces species emphasizes the importance of this regulatory mechanism .
Summary of metabolic processes related to the target genes of TamR
We show here that S. coelicolor TamR regulates additional genes whose products are involved in maintaining flux through the citric acid cycle. These genes encode malate synthase, malate dehydrogenase and isocitrate dehydrogenase. Malate dehydrogenase and isocitrate dehydrogenase function in the citric acid cycle. Malate synthase, which participates in the glyoxylate cycle, can provide malate to the citric acid cycle by synthesizing malate from glyoxylate and acetyl-CoA, thus replenishing the intermediate pool (Figure 1B). We also find that predicted TamR binding sites are highly conserved in the promoters of these genes from different Streptomyces species, indicating that the contribution of TamR to regulation of enzymes involved in the citric acid and glyoxylate cycles is conserved among streptomycetes.
TamR binding site identification and logos
The sequence logos in Figure 2, which show the consensus sequence of the TamR binding sites, were generated using WebLogo (http://weblogo.berkeley.edu/) . The relative frequency of base pairs at each position is represented by the height of each nucleotide. To identify other potential target genes, a weight matrix corresponding to the 18 bp consensus TamR binding site in the tam–tamR intergenic region was used in genome-scale screens of Streptomyces genomes using PATSER. The sequences of S. coelicolor TamR (SCO3133), trans-aconitate methyltransferase (SCO3132), malate synthase (SCO6243), malate dehydrogenase (SCO4827) and isocitrate dehydrogenase (SCO7000) were used as queries using BlastP on the NCBI website for identification of homologues in other streptomycetes. TamR binding sites were identified by inspection of the respective promoters.
Cloning and purification of TamR
Cloning and purification of TamR was carried out as described previously . Briefly, tamR (SCO3133) was amplified from genomic DNA of S. coelicolor A3(2) M145 strain and cloned into NdeI–EcoRI sites of pET28b (Novagen). Recombinant plasmid was transformed into Escherichia coli BL21(DE3) for overexpression of protein. TamR with an N-terminal His6-tag was purified using HIS-Select Nickel Affinity column (Sigma). Protein concentration was determined from its absorbance at 280 nm using the molar absorption coefficient calculated based on the amino acid sequence.
DNA binding assays
A 193-bp DNA segment, which contains the promoter region of the aceB1 (SCO6243) gene was amplified from S. coelicolor genomic DNA using primers aceB1-Fw (5′-TTGAGTGAGCGAGGTGG-3′) and aceB1-Rv (5′-TGAGCTGTCACTTCCTTCA-3′). DNA concentration was determined by its absorbance at 260 nm. This DNA segment was 32P-labelled at the 5′-ends using T4-polynucleotide kinase (T4-PNK). For electrophoretic mobility shift assay, 32P-labelled pACEB1 (0.07 nM; stoichiometric conditions) was incubated with TamR in binding buffer (0.5 M Tris (pH 8.0), 50 mM NaCl, 0.06% detergent Brij58, 20 μg/ml BSA and 2% glycerol) at 25°C for 30 min before reactions were loaded on to 6% non-denaturing polyacrylamide gels (39:1 (w/w) acrylamide:bisacrylamide). The unusually high buffer concentration was used for ease of comparison with previous experiments reporting binding of TamR to other gene promoters and was originally selected to be able to analyse the response of TamR to the ligand urate; urate was dissolved in 0.4 M NaOH, necessitating the high buffer concentration to avoid changes in pH on addition of ligand. The gels were pre-run for 20 min in 0.5× Tris/borate/EDTA (TBE) buffer at 4°C before the samples were loaded and run at 10 V/cm for 2 h in 0.5× TBE buffer to separate free DNA and complex. After the gel was dried, it was exposed to phosphor screens. Data were obtained using a Storm 840 phosphorimager (GE Healthcare). The densitometric result was obtained using ImageQuant 5.1 (Molecular Dynamics). The quantitative data were fitted to the Hill equation: f=fmax·[TamR]n/(K + [TamR]n), where [TamR] is the protein concentration, f is fractional saturation, K is a constant and n is the Hill coefficient. Tangents were drawn to the initial rise and the final plateau and their intercept used to calculate the stoichiometry of binding. KaleidaGraph 4.0 (Synergy Software) was used for this analysis.
To determine the effect of ligands on the formation of TamR–DNA complex, reactions with different concentration of ligand were prepared. Ligands were dissolved in distilled water and added before addition of TamR (0.16 nM) to DNA (pACEB1). After 30 min of incubation at 25°C, samples were analysed using electrophoretic mobility shift assays (EMSA) under the conditions described above. Data were analysed by fitting to exponential decay equation: f=Ae−kL, where f represents fractional saturation, L represents the ligand concentration, A represents the saturation plateau and k represents the exponential decay constant. Quantification results derive from at least three independent experiments.
Primers mdh-Fw (5′-CCTTCTTTTGCGTGCCC-3′) and mdh-Rv (5′-GCGTTCTCCGTATGACAGC-3′) were used to amplify 184 bp DNA containing the promoter region of mdh. Primers idh-Fw (5′-TCCCGCTCCCCTGA-3′) and idh-Rv (5′-TGGTCGAGTCAGTCACCG-3′) were used to amplify 85 bp DNA covering the promoter region of idh. For EMSA, 0.13 nM 32P-labelled pMDH or 0.46 nM 32P-labelled pIDH was used in each reaction. All other conditions were as described above. For analysis of specificity of binding, 30 nt synthetic oligodeoxyribonucleotides were purchased and annealed by heating to 95°C followed by slow cooling to room temperature. Duplex DNA was labelled and used in EMSA as described above, using 0.3 nM DNA per reaction and a buffer composed of 0.25 M Tris (pH 8.0), 25 mM NaCl, 0.06% detergent Brij58 and 2% glycerol.
In vivo regulation of gene activity
S. coelicolor cultures, which were germinated from spores in tryptone yeast extract broth (International Streptomyces Project ISP medium 1) were grown for 36 h at 30°C (280 rpm) before they were treated with either citrate (100 mM) combined with Ca2+ (10 mM) or H2O2 (10 mM) for 2 h. Reference cultures were grown in unsupplemented medium. Then, the S. coelicolor cells were harvested by centrifugation and immediately washed twice using 50 mM sodium phosphate buffer (pH 6.4). The total RNA was then immediately isolated using Illustra RNAspin Mini Isolation Kit (GE Healthcare). AMV reverse transcriptase (New England BioLabs) was used to generate cDNA for quantitative PCR using designed primers and total RNA.
Quantitative PCR was performed on an Applied Biosystems 7500 Real-Time PCR system with gene rpoA (house-keeping gene encoding RNA polymerase α subunit) as internal control using SYBR Green I as fluorescent dye. After data were validated, comparative CT (2−ΔΔCT) method was used for data analysis . The primers used in this experiment were: ACEB1-RT5: 5′-TCCGCCTACACCCGCAGCAT-3′, ACEB1-RT3: 5′-AAGACGAAGACCTCGTTCCAGAG-3′, MDH-RT5: 5′-AGTACCCGGACATCTTCCAC-3′, MDH-RT3: 5′-GACG-GTCGGGATGAACTC-3′, IDH-RT5: 5′-GCCACGATGATG-AAGGTCT-3′, IDH-RT3: 5′-ACTCCAGGCCCTTGTAGATG-3′, RPOA-RT5: 5′-AAGCTGGAGATGGAGCTGAC-3′, RPOA-RT3: 5′-TTGAGAACCGGCGAGTAGAT-3′.
Reporter constructs with wild-type and mutant promoter sequences
The gene encoding d1EGFP, a destabilized version of enhanced green fluorescent protein (EGFP) encoded by pd1EGFP-N1, was cloned between BamHI and NotI sites in the integrative plasmid pSET152 that carries an apramycin resistance gene for selection in E. coli and Streptomyces . Synthetic oligodeoxyribonucleotides were designed that contain the mdh promoter (93 bp upstream of the start codon) flanked by XbaI and BamHI sites. A construct in which the central six base pairs of the TamR site were changed from GATACC to CTATAG was also obtained based on EMSA results with 30 bp DNA representing the TamR site in the mdh promoter. Oligodeoxyribonucleotides were purified by denaturing gel electrophoresis and annealed by heating to 95°C followed by slow cooling to room temperature. Duplex DNA was digested with XbaI and BamHI and cloned into the corresponding sites in pSET152 carrying d1EGFP to generate pSET–EGFP–mdh and pSET–EGFP–mdhm1, respectively. Plasmid sequences were confirmed by sequencing. The resulting constructs were conjugated from E. coli ET12567(pUZ8002)  to S. coelicolor M145, and apramycinR transconjugants were selected by overlaying with apramycin (25 μg/ml) and nalidixic acid (25 μg/ml) . The colonies were screened by sensitivity to kanamycin and growth in ISP-1 supplemented with 50 μg/ml apramycin. The integration of the plasmid to the genomic DNA was confirmed by performing PCR using primers pSET152F1 5′-GAAAGGGGGATGTGCTGC-3′ and pSET152R1 5′-CTGCGTTATCCCCTGATTCT-3′ that anneal upstream of the inserted mdh promoter sequence and downstream of the egfp gene in the pSET152 plasmid. The confirmed strains were used for fluorescence microscopy experiments.
For visualization of EGFP expression, S. coelicolor cultures were germinated from spores and grown in ISP medium 1 for 24 h before they were treated with citrate (100 mM) and Ca2+ (10 mM). After 2 h of incubation, the fluorescence was measured. Samples were mounted for microscopy by spotting liquid cultures directly on microscope slides and covering with a coverslip. Fluorescence microscopy was performed using the Leica DM RXA2 fluorescence microscope with a 100× objective. The fluorescence microscope was equipped with a SensiCam QE cooled digital 12-bit CCD camera system and Slidebook software (version 5.5, Intelligent Imaging Innovations). Fluorescence images were subjected to deconvolution to improve the signal to noise ratio. At least two independent experiments were performed for each construct.
Predicted TamR binding sites in the promoters of three target genes
S. coelicolor TamR regulates transcription of its own gene (tamR, SCO3133) and that of a divergent gene encoding trans-aconitate methyltransferase (tam, SCO3132). A consensus binding site was identified based on a comparison of the intergenic region between these two genes in select Streptomyces species (Figure 1A) . TamR also binds the promoter of a gene encoding aconitase (sacA, SCO5999) and controls its expression . In order to find additional genes potentially regulated by TamR, a weight matrix based on the frequencies of individual bases occurring at each position within the 18 bp consensus binding motif was applied in a genome-scale screen of the S. coelicolor genome using the program PATSER (http://rsat.ulb.ac.be/genome-scale-patser_form.cgi). Features with a score ≥10, corresponding to sites in gene promoters, were examined and found to include the expected sites in the tam–tamR intergenic region and in the sacA promoter. Equivalent sites were also found on examination of other Streptomyces genomes (e.g. S. avermitilis and S. griseus). This is consistent with the previously reported conservation of TamR in various Streptomyces species . In addition, three genes were found to contain predicted TamR sites in both S. coelicolor and most other Streptomyces genomes available on the PATSER website; these genes encode malate synthase (aceB1, SCO6243), malate dehydrogenase (mdh, SCO4827) and isocitrate dehydrogenase (idh, SCO7000). No candidate sites were found in promoters for genes encoding other enzymes in the citric acid cycle.
To examine more rigorously whether TamR binding sites are generally conserved in the promoters of these genes among species that encode TamR, the sequences of S. coelicolor TamR and trans-aconitate methyltransferase were used as queries with BlastP on the NCBI website. Examination of all available Streptomyces sequences revealed a subset of species that encode a homologue of S. coelicolor trans-aconitate methyltransferase (shown in Supplementary Table S1); notably, all species that encode this homologue also encode TamR. For all species but S. ambofaciens, the tamR gene is encoded divergently from the tam gene. A BlastP of species that encode tam–tamR gene pairs with S. coelicolor malate synthase, malate dehydrogenase and isocitrate dehydrogenase revealed significant conservation of these enzymes as well as conservation of predicted TamR binding sites in their gene promoters (Supplementary Table S1).
In the promoter region of aceB1, encoding malate synthase (Figure 2A), two predicted TamR binding sites are located near each other with 3 bp overlap (Figure 2B; overlapping base pairs in positions 16–18). These two binding sites and their organization are conserved in several Streptomyces species, which encode the tam–tamR gene pairs (Supplementary Table S1). The sequence of these sites is highly conserved, particularly site 1 (positions 16–33; Figure 2B). Malate synthase uses acetyl-CoA and glyoxylate as substrates to produce malate (Figure 1B); activation of aceB1 gene expression would therefore be expected to promote formation of malate for entry into the citric acid cycle. The presence of two potential TamR sites in the gene promoter predicts association of two TamR proteins.
Potential TamR target genes with predicted TamR binding sites in the promoter region
One predicted TamR binding site is found in the promoters of genes encoding S. coelicolor malate dehydrogenase (mdh) (Figures 2C and 2D) and isocitrate dehydrogenase (idh) (Figures 2E and 2F). Isocitrate dehydrogenase catalyses the conversion of isocitrate to α-ketoglutarate, and malate dehydrogenase converts malate to oxaloacetate, both key steps in the citric acid cycle. The predicted TamR binding sites are almost invariably present in the corresponding genes in Streptomyces species that encode tam–tamR gene pairs, with the site in the idh promoter being particularly well conserved (Figure 2F and Supplementary Table S1). The significant conservation of predicted TamR binding sites in the promoter regions of mdh and idh suggests that regulation of these genes by TamR is evolutionarily conserved. In addition, alignment of the predicted TamR binding sites from different S. coelicolor target genes reveals significant conservation (Supplementary Figure S1).
TamR binds to the promoter region of predicted target genes
To determine whether TamR can bind to the promoter regions of aceB1, mdh and idh, labelled DNA fragments representing the target gene promoters were used in EMSA. The TamR used in EMSA was overexpressed in E. coli and purified to apparent homogeneity, as described previously . As expected for MarR homologues, TamR exists as a dimer under physiological conditions, as determined by gel filtration chromatography .
Titration of 0.07 nM pACEB1 DNA, which contains the promoter region of aceB1, with TamR revealed formation of one predominant complex, with a second complex of an electrophoretic mobility intermediate between that of free DNA and the predominant complex that was barely detectable (Figure 3A). The formation of two distinct complexes is consistent with the predicted existence of two TamR sites. The data also suggest that formation of a complex with a protein:DNA stoichiometry of 2:1 is favoured over a 1:1 complex. Considering that TamR binds the tam–tamR intergenic region with a Kd ~17 pM  and the significant conservation of its predicted binding sites (Supplementary Figure S1), we surmised that reaction conditions used were likely stoichiometric ([DNA] > Kd). Quantification of EMSA data indicated saturation of pACEB1 DNA at a protein:DNA ratio of 1.8:1, consistent with the presence of two TamR sites (Figure 3D).
TamR binding to aceB1, mdh and idh promoters
A titration of TamR with pMDH DNA, which contains the promoter region of mdh showed that TamR forms a single complex with this DNA (Figure 3B). Using 0.13 nM DNA, quantification of EMSA data indicates saturation at a protein:DNA ratio of 0.8:1, consistent with the presence of a single TamR site (Figure 3E). Similarly, titration of pIDH DNA, which contains the idh promoter region yields a single complex (Figure 3C). TamR saturates this DNA (0.46 nM) at a protein:DNA ratio of 1:1 (Figure 3F). That the latter gene promoters with a single predicted TamR site are saturated at a protein:DNA ratio of 1:1 also indicates that the TamR preparation is fully active, and that saturation of the pACEB1 DNA at a ratio of ~2:1 indeed reflects the accommodation of two TamR molecules on this DNA and not a partially inactive protein preparation.
Specificity of TamR binding to the tam–tamR intergenic region was previously demonstrated using competition assays . To verify binding specificity further and to evaluate which sequence elements are most important for TamR binding, we created a 30-bp duplex representing the cognate site in the mdh promoter and three duplexes in which sets of three base pairs within each half-site were simultaneously mutated (Figure 4A). Because sequence flanking the identified palindrome was completely symmetrical, we substituted one base pair in the cognate site (changing the first three base pairs of the second half site from ACC to ATC) to render the entire sequence palindromic. TamR binds well to the completely palindromic MDH sequence (Figure 4A). As shown in Figures 4B–4E, TamR binding is compromised by substitutions in either of the three sets of three base pairs, with substitutions of the central base pairs (MDH-mut1, Figure 4C) particularly deleterious to binding. This not only confirms that TamR binding is sequence-specific, but that interactions at the centre of the palindrome are particularly important for binding.
Binding of TamR to 30 bp DNA containing TamR binding site
The binding of TamR to aceB1 promoter DNA is attenuated by trans-aconitate, cis-aconitate, citrate and isocitrate
Since binding of TamR to tam–tamR intergenic DNA is attenuated by citrate, trans-aconitate, cis-aconitate and isocitrate , the effect of these ligands on the binding of TamR to pACEB1 DNA was also investigated. All four ligands significantly attenuated the binding of TamR to this DNA (Figures 5A–5D). The ability of these ligands to attenuate TamR–pACEB1 complex formation was reflected by IC50 values; citrate, trans-aconitate, and cis-aconitate showed comparable efficiency with IC50 values of 50.6±3.5, 48.0±1.9 and 32.1±0.7 mM, respectively (Figures 5E and 5F) whereas isocitrate was somewhat less efficient at attenuating DNA binding, with an IC50 of 87.0±2.0 mM (Figure 5E). By comparison, succinate and α-ketoglutarate, which are also intermediates in the citric acid cycle, do not affect DNA binding by TamR (data not shown).
Effect of ligands on the binding of TamR to the promoter region of aceB1
In vivo effect of citrate and H2O2 on transcription
To investigate the transcriptional regulation of the three target genes (aceB1, mdh and idh) by TamR, transcript levels were measured in S. coelicolor under conditions in which the TamR ligands accumulate. In S. coelicolor, the transport of exogenous citrate across cell membranes is mainly accomplished by the CitMHS family of transporters [24,25]. Citrate needs to form metal complex with specific metals (with Ca2+ or Fe3+) for this transport to occur, otherwise exogenous citrate cannot be transported by this family of transporters. Therefore, S. coelicolor was cultured in medium with 100 mM citrate and 10 mM CaCl2. As reported previously, supplementing the culture medium with citrate only has no effect on tam, tamR or sacA transcript levels, whereas the addition of both citrate and Ca2+ leads to significantly increased transcript levels . Quantifying transcripts using qRT-PCR (quantitative reverse transcription-PCR) showed that the transcript level of aceB1, mdh and idh increased 3.7±0.8-, 1.4±0.1- and 2.8±0.9-fold, respectively, when citrate and Ca2+ was supplemented in the medium (Figure 6).
In vivo gene regulation of TamR target genes
Aconitase contains an iron–sulfur cluster that is susceptible to oxidation, leading to inactivation of enzymatic activity [26–29]. Under such conditions, intracellular levels of the substrate citrate increase and the intermediate cis-aconitate may be released and converted to trans-aconitate. Although accumulation of trans-aconitate would be expected to mirror the number of inactivated copies of aconitase, citrate was reported to accumulate to more than 14 mM on inactivation of the enzyme . Addition of H2O2, which inactivates aconitase and therefore causes an increase in intracellular citrate concentrations, elevated the transcript levels of aceB1, mdh and idh to a different extent, but comparably to the addition of citrate, except for the mdh transcript, which was produced more efficiently on addition of H2O2 (Figure 6). The transcript levels of aceB1, mdh and idh increased 2.7±0.3-, 5.7±1.2- and 4.4±2.0-fold, respectively. By comparison, transcript levels of tam, tamR and sacA were also differentially affected by H2O2, but increased somewhat more efficiently on treatment of cultures with H2O2 compared with uptake of citrate, perhaps due to a combined effect of greater intracellular citrate levels on inactivation of aconitate with H2O2 and the associated accumulation of trans-aconitate .
To verify association of TamR with its cognate sites in vivo, reporter constructs were created in which the gene encoding EGFP under the control of the wild-type mdh promoter or the mutant mdh promoter in which the central base pairs of the TamR sites were mutated was integrated into the S. coelicolor genome. Imaging of cells carrying egfp under the control of the wild-type mdh promoter showed no fluorescence (Figure 7A), consistent with the interpretation that expression is repressed. However, addition of citrate resulted in cells consistently expressing EGFP (Figure 7B). By contrast, when egfp was expressed under the control of the mutant mdh promoter, bright fluorescence was observed in the absence of citrate (Figure 7C); addition of citrate did not change the expression (not shown). These experiments support the conclusion that TamR functions to repress the mdh promoter, and that repression is relieved in the presence of citrate.
In vivo activity of mdh promoter
The TamR regulon is conserved among Streptomyces species
Enzymes involved in the citric acid cycle are expected to be present at some level regardless of growth conditions. However, their levels have been shown to fluctuate significantly depending on circumstances such as carbon source and oxygen level [31–35]. Our data suggest that S. coelicolor TamR regulates aceB1, mdh and idh in addition to three target genes described previously (tam, tamR and sacA). These genes either encode enzymes in the citric acid cycle or encode proteins with the capacity to affect flux through this metabolic pathway.
The gene locus consisting of divergently oriented tamR and tam genes is highly conserved among the Streptomyces species in which this gene pair is encoded. Exceptions include a few Streptomyces species in which the tam gene is not immediately adjacent to and divergent from tamR but preceded by another gene and S. ambofaciens in which the tam gene is not divergent from tamR (Supplementary Table S1). The frequent occurrence of divergent tam–tamR gene pairs suggests a shared evolutionary origin, whereas the absence of these genes in many other Streptomyces species may reflect gene loss. That TamR binding sites are highly conserved in the intergenic region between these two genes as well as in the promoters of genes encoding aconitase, malate synthase, malate dehydrogenase and isocitrate dehydrogenase further indicates that the regulatory role of TamR is evolutionarily conserved.
The tam–tamR gene locus often includes a gene encoding a predicted glyoxalase (Figure 1A; the predicted glyoxalase gene overlaps the tam gene by 3 bp). Although this particular homologue remains uncharacterized, we note that glyoxalases are generally involved in detoxification of the metabolite methylglyoxal, which is formed when dihydroxyacetone phosphate and glyceraldehyde 3-phosphate accumulate during glycolysis . This situation may pertain if flux through the citric acid cycle is compromised. It is conceivable that TamR-mediated up-regulation of glyoxalase may alleviate the toxicity associated with methylglyoxal production.
The promoters for aconitase, malate dehydrogenase and isocitrate dehydrogenase contain a single TamR site, whereas the malate synthase gene promoter features two sites that overlap by three base pairs. The latter is reminiscent of the tam–tamR intergenic region, which contains six sites that all overlap each other by three base pairs . This organization of cognate sites places the centres of each site on opposite faces of the double helix, predicting that two TamR dimers bind on opposite faces of the duplex. A similar organization of two overlapping cognate sites is seen in the promoter for Agrobacterium tumefaciens PecS, where the extent of protection from DNase I digestion is much greater than the protection of a single site, consistent with occupancy by two PecS molecules . The stoichiometry of TamR binding to the malate synthase gene promoter and the formation of a single predominant complex suggests that two TamR dimers effectively bind simultaneously to this promoter. Despite the difference in stoichiometry, changes in gene expression on accumulation of ligand are comparable (Figure 6), and the concentrations of ligand associated with 50% reduction in complex formation in vitro is comparable to that seen for TamR binding to the six sites in the tam–tamR intergenic region. Since the number of TamR sites does not appear to correlate with either levels of gene expression or sensitivity to ligand in vitro, it is conceivable that the existence of accessory sites may contribute to the exclusion of other regulatory factors by occluding the promoter.
TamR promotes flux through the citric acid cycle
All TamR target genes are related to the citric acid cycle (Figure 1). Genes sacA, mdh and idh encode enzymes (aconitase, malate dehydrogenase and isocitrate dehydrogenase, respectively) in the citric acid cycle. Gene tam encodes trans-aconitate methyltransferase, which is important for preventing the accumulation of a toxic byproduct of the citrate isomerization step, trans-aconitate [17,18]. Gene aceB1 encodes a malate synthase, which can provide malate for the citric acid cycle from glyoxylate and acetyl-CoA. This indicates that TamR plays an important role in restoring metabolic flux through the citric acid cycle.
The roles of proteins encoded by TamR target genes reflect an interesting relationship and provide a rationale for concerted regulation (Figure 1). In the citric acid cycle, one acetyl-CoA is degraded per cycle to produce NADH (or NADPH) and ATP. Replenishing the intermediate pool is important to maintain the efficiency of the cycle as many intermediates function as precursors for other biosynthetic reactions. If aconitase is inactivated, cis-aconitate may be released. In this circumstance, trans-aconitate methyltransferase becomes important for preventing accumulation of the inhibitory metabolite trans-aconitate (Figure 1B). Secondly, inactivation of aconitase can result in depletion of other citric acid cycle intermediates. Up-regulation of the gene encoding aconitase will alleviate this situation as well as consume accumulated citrate (which is also inhibitory), and malate synthase will synthesize malate from glyoxylate and acetyl-CoA, thereby providing malate to the citric acid cycle. It is therefore reasonable for TamR to regulate genes encoding trans-aconitate methyltransferase, aconitase and malate synthase to restore cycle flux. Malate dehydrogenase and isocitrate dehydrogenase both catalyse a dehydrogenation step that produces energy, therefore the elevated expression of these two enzymes would be beneficial to the efficiency of the citric acid cycle and its energy production.
If inactivation of aconitase is the primary circumstance in which TamR-mediated gene regulation becomes important, then this scenario corresponds to a situation in which the metabolites citrate and trans-aconitate accumulate, as discussed above. What would then be the rationale for TamR-mediated up-regulation of the gene encoding isocitrate dehydrogenase, considering that isocitrate would likely be depleted on inactivation of aconitase? Converting isocitrate to succinate and glyoxylate by isocitrate lyase via the glyoxylate bypass circumvents energy-producing steps of the citric acid cycle (Figure 1B); up-regulation of the gene encoding isocitrate dehydrogenase would therefore favour the citric acid cycle and prevent glyoxylate bypass once isocitrate becomes available.
The glyoxylate bypass allows cells to utilize simple carbon compounds when carbohydrates such as glucose are scarce by synthesizing malate from glyoxylate and acetyl-CoA. In E. coli, there are two types of malate synthase, named malate synthase A and malate synthase G [37–39]. Both can synthesize malate from glyoxylate and acetyl-CoA. In E. coli, the aceB–aceA–aceK operon is very important to the glyxoylate cycle, encoding malate synthase A (aceB), isocitrate lyase (aceA) and isocitrate dehydrogenase kinase/phosphatase (aceK) [40–42]. Isocitrate lyase and malate synthase A carry out the glyoxylate bypass, and isocitrate dehydrogenase kinase/phosphatase is a bifunctional enzyme which can phosphorylate or dephosphorylate isocitrate dehydrogenase to regulate its activity [43–47]. Malate synthase G, which is encoded by the glcB gene, can be induced by glycolate and synthesizes malate from glyoxylate produced by the glycolate pathway [37–39].
S. coelicolor also harbours two genes encoding malate synthase, aceB1 and aceB2 (SCO0983). A gene encoding isocitrate lyase (SCO0982) is adjacent to aceB2, encoding malate synthase A. The aceB1 gene encodes a homologue of malate synthase G, and surrounding genes encode allantoinase (SCO6247), allantoicase (SCO6248) and an IclR homologue (SCO6246) (Figure 2A). Allantoinase (EC 22.214.171.124) and allantoicase (EC 126.96.36.199) catalyse consecutive reactions that convert allantoin to ureidoglycolate and urea as part of the purine degradation pathway. Ureidoglycolate is then converted to urea and glyoxylate, the latter a substrate for malate synthase. The gene organization suggests that malate synthase G may replenish the citric acid cycle intermediate pool with malate, which is converted from the glyoxylate produced by purine degradation or other metabolic processes.
The glyoxylate cycle bypasses two iron–sulfur proteins (fumarase and succinate dehydrogenase) in the citric acid cycle, both of which contain iron–sulfur clusters that are vulnerable to oxidative stress. If the citric acid cycle is compromised due to oxidative stress, the glyoxylate cycle may be able to sustain the cell until damaged enzymes in the citric acid cycle are regenerated. TamR-mediated up-regulation of malate synthase gene expression would also be beneficial from this point of view.
Hao Huang and Smitha Sivapragasam performed experiments. All authors contributed to experimental design and data analysis. Hao Huang and Anne Grove wrote the paper.
We thank Gregg Pettis for providing E. coli ET12567(pUZ8002) and pSET152.
This work was supported in part by the National Science Foundation [grant number MCB-1051610 (to A.G.)].
Present address: Pediatric Oncology, Dana-Farber Cancer Institute, Boston, MA 02215, U.S.A.